Detection and Diagnosis of Defect in GIS Based on X-ray Digital Imaging Technology

For better application of X-ray digital imaging technology in defect detection in Gas Insulated Switchgear (GIS), it is essential to investigate the typical defect and establish the defect database, which has not been adequately performed in previous work. Systematic experimental research is also needed to accumulate data and experience. In this research, an experimental platform, including Computed Radiography (CR) imaging system and a GIS model, is built, and extensive tests of different kinds of typical defects are studied. The influence X-ray irradiation on SF6 under different tube voltage levels is firstly examined, which proves that the withstand voltage of SF6 gas has not been affected and no dissociation has been found. Then, several kinds of defects are tested by X-ray digital imaging technology. The successful application examples of “visual” detection of defects further prove the practicability and validity of the X-ray digital imaging technique. Finally, the image database of typical defects inside of GIS is established and the defect risk is also analyzed in three levels, which would be useful for the defect severity diagnosis and risk assessment.


Introduction
Image processing technology has been extensively adopted in abnormal defect detection and diagnosis of electrical equipment because of the high accuracy and rich processing content [1,2]. Of varied methods, X-ray digital imaging technology can observe the internal structure abnormality of electrical equipment under live operation conditions by non-contact and nondestructive testing [3][4][5][6]. Gas insulated switchgear (GIS) is one of the important equipment in the power system. The national economy will be impacted once it breaks down. Therefore, rapid and accurate diagnosis of the internal defects takes great significance to the stable and reliable operation of the power grid [7][8][9][10][11].
The commonly used methods for GIS equipment detection, such as ultra-high frequency (UHF) PD detection, ultrasonic partial discharge (PD) detection, and gas decomposition detection of SF 6 , are proved to be effective and widely used to find the internal insulation defect [12][13][14][15][16][17][18]. However, these methods are all indirect detections that are not able to diagnose the physical defects accurately when no PD occurs [19][20][21][22]. This problem could be solved effectively by X-ray irradiation for the high reducibility [23,24]. X-ray digital imaging detection could visually discover the severity of defects, whether there is a partial discharge or not, which realizes the visual detection of GIS without power off and improves the diagnosis ability of the equipment operation status.
As the energy of the X-ray is greater than most disturbances in the substation [25], X-ray imaging detection also has the advantages of interference immunity. However, it may also ionize the SF6 inside of GIS. If SF6 is decomposed due to X-ray irradiation, the internal insulation of GIS will be seriously reduced, even resulting in a safety accident. Thus, the ionization influence of X-ray on SF6 needs to be further investigated, thus as to ensure the practicability of the method. As for the X-ray digital imaging detection, it is practical to identify and diagnose the internal defects in GIS. Since the defects on the actual site are variable, the establishment of a typical defect database becomes essential to guide severity diagnosis and risk assessment, which has not been considered adequately in previous work.
Based on the analysis above, an experimental platform including Computed Radiology (CR) imaging system and GIS model is built to test the irradiation on SF6 gas. The withstand voltage and composition of SF6 gas after irradiated by X-ray under the different tube voltage levels are studied firstly. Next, an experimental arrangement is built, and the typical defects are detected by the X-ray digital imaging method, and the image results are obtained. Then, the defects under the different defect grades and their risk are analyzed in three levels. Accordingly, the image database of typical defects is established and generalized, while the X-ray detecting criterion for different defects is also given. The experimental results and the image database with the risk classification are original and would be useful for the practical application of X-ray digital imaging detection. Figure 1a gives the schematic diagram of the X-ray imaging detecting system [26][27][28][29][30][31]. The imaging plate that contains photosensitive phosphor will record the image signal once the X-ray radiated from the emitter is perceived. Then, the image plate is sent to a digital converter in which it will be scanned by the laser beam. The image information released in the form of visible light would be captured and converted into a digital signal, and the image could be obtained subsequently through the calculation. Next, the CR reader scans the screen with a laser beam, and the laser energy can release captured electrons and cause the radiation of visible light. The released visible light will be captured and transferred into a digital bitstream and finally encoded into the digital image.  Actually, it is impossible for the imaging plate to stick to the surface of the tested object inside of the GIS. Therefore, the digital image obtained on the imaging plate is inevitably enlarged, as Figure  1b shows. The magnification depends on the distance from the X-ray source to the detected part surface L1 and the distance from the detected part surface to the imaging plate L2. Generally, L1 is determined by the arrangement space of the X-ray source in the field detection, while L2 is determined by the size of the detected GIS equipment. Actually, it is impossible for the imaging plate to stick to the surface of the tested object inside of the GIS. Therefore, the digital image obtained on the imaging plate is inevitably enlarged, as Figure 1b shows. The magnification depends on the distance from the X-ray source to the detected part surface L 1 and the distance from the detected part surface to the imaging plate L 2 . Generally, L 1 is determined by the arrangement space of the X-ray source in the field detection, while L 2 is determined by the size of the detected GIS equipment.

Experimental Platform Construction
Considering the situation of the field application, the X-ray digital imaging system used for GIS contains the X-ray machine with a field moving bracket, the imaging plate, a CR scanner, a moving workstation with image processing and analysis software, the inspected GIS equipment, etc., which are shown in Figure 2a. This system can realize X-ray irradiation of high-risk parts in GIS, e.g., circuit breaker, disconnector, busbar connection, etc. Figure 2b gives an example of testing the disconnector of 220 kV GIS in the field. The information of internal structure is formed by the photosensitivity of the imaging plate, which is scanned using the laser digital scanning technology. The information gets transformed into digital image, then the investigation of the internal structure of GIS can be achieved.

Experimental Platform Construction
Considering the situation of the field application, the X-ray digital imaging system used for GIS contains the X-ray machine with a field moving bracket, the imaging plate, a CR scanner, a moving workstation with image processing and analysis software, the inspected GIS equipment, etc., which are shown in Figure 2a. This system can realize X-ray irradiation of high-risk parts in GIS, e.g., circuit breaker, disconnector, busbar connection, etc. Figure 2b gives an example of testing the disconnector of 220 kV GIS in the field. The information of internal structure is formed by the photosensitivity of the imaging plate, which is scanned using the laser digital scanning technology. The information gets transformed into digital image, then the investigation of the internal structure of GIS can be achieved.

Influence of X-ray Irradiation on SF6
X-ray radiation can cause material ionization, which will reduce the insulation of the material to some degree. Thus, the withstand voltage test and the SF6 dissociation test are taken thus as to examine the influence of X-ray irradiation on SF6. Figure 3 indicates the design of the test tooling filled with 0.4 MPa of SF6 as well as the plate electrode with a diameter of 150 mm and the chamfer of 30 mm, which are designed to get the stable value of the withstand voltage test. The gap between the electrodes is about 5 mm. Figure 4 shows the complete photo of the test tooling. The bottom is equipped with a gas-pressure meter and a gas port. The breakdown test is carried out by applying a voltage at both ends of the test tooling.

Influence of X-ray Irradiation on SF 6
X-ray radiation can cause material ionization, which will reduce the insulation of the material to some degree. Thus, the withstand voltage test and the SF 6 dissociation test are taken thus as to examine the influence of X-ray irradiation on SF 6 . Figure 3 indicates the design of the test tooling filled with 0.4 MPa of SF 6 as well as the plate electrode with a diameter of 150 mm and the chamfer of 30 mm, which are designed to get the stable value of the withstand voltage test. The gap between the electrodes is about 5 mm. Figure 4 shows the complete photo of the test tooling. The bottom is equipped with a gas-pressure meter and a gas port. The breakdown test is carried out by applying a voltage at both ends of the test tooling.

The Withstand Voltage Test after X-ray Irradiation
Thus as to obtain systematic experimental comparison results, the breakdown test of the tooling was made firstly without any treatment, and the breakdown voltage before X-ray irradiation was obtained 5 times. Subsequently, the test tooling was applied with X-ray irradiation under conditions that the tube voltage was set, respectively, as 100 kV/200 kV/300 kV, while the tube current was set as 5 mA. The breakdown voltage test was made again after 5 minutes of irradiation, then the average breakdown voltage under each tube voltage level was obtained, which is shown in Figure 5.  Figure 3 indicates the design of the test tooling filled with 0.4 MPa of SF6 as well as the plate electrode with a diameter of 150 mm and the chamfer of 30 mm, which are designed to get the stable value of the withstand voltage test. The gap between the electrodes is about 5 mm. Figure 4 shows the complete photo of the test tooling. The bottom is equipped with a gas-pressure meter and a gas port. The breakdown test is carried out by applying a voltage at both ends of the test tooling.   Thus as to obtain systematic experimental comparison results, the breakdown test of the tooling was made firstly without any treatment, and the breakdown voltage before X-ray irradiation was obtained 5 times. Subsequently, the test tooling was applied with X-ray irradiation under conditions that the tube voltage was set, respectively, as 100 kV/200 kV/300 kV, while the tube current was set as 5 mA. The breakdown voltage test was made again after 5 minutes of irradiation, then the average breakdown voltage under each tube voltage level was obtained, which is shown in Figure 5. After a comprehensive analysis of the results, it was found that the fluctuation of breakdown voltage, which was basically maintained as 114 kV, was not so much after X-ray irradiation. In consideration of the contingency of breakdown voltage itself, it was inferred that the breakdown voltage of SF6 remained broadly unchanged after the X-ray irradiation. Actually, the X-ray irradiation was usually less than 3 min in the X-ray digital imaging detection of GIS. Thus, it can be regarded that there was no influence of X-ray irradiation on the withstand voltage of SF6.    Thus as to obtain systematic experimental comparison results, the breakdown test of the tooling was made firstly without any treatment, and the breakdown voltage before X-ray irradiation was obtained 5 times. Subsequently, the test tooling was applied with X-ray irradiation under conditions that the tube voltage was set, respectively, as 100 kV/200 kV/300 kV, while the tube current was set as 5 mA. The breakdown voltage test was made again after 5 minutes of irradiation, then the average breakdown voltage under each tube voltage level was obtained, which is shown in Figure 5. After a comprehensive analysis of the results, it was found that the fluctuation of breakdown voltage, which was basically maintained as 114 kV, was not so much after X-ray irradiation. In consideration of the contingency of breakdown voltage itself, it was inferred that the breakdown voltage of SF6 remained broadly unchanged after the X-ray irradiation. Actually, the X-ray irradiation was usually less than 3 min in the X-ray digital imaging detection of GIS. Thus, it can be regarded that there was no influence of X-ray irradiation on the withstand voltage of SF6.

SF6 Dissociation Test after X-ray Irradiation
To prevent the decomposition of SF6 in GIS caused by the X-ray imaging digital detection at the experiment site, the influence of the X-ray irradiation on SF6 under various tube voltage levels was researched. The SF6 decomposition product detector was utilized in the test, and the accuracy of the After a comprehensive analysis of the results, it was found that the fluctuation of breakdown voltage, which was basically maintained as 114 kV, was not so much after X-ray irradiation. In consideration of the contingency of breakdown voltage itself, it was inferred that the breakdown voltage of SF 6 remained broadly unchanged after the X-ray irradiation. Actually, the X-ray irradiation was usually less than 3 min in the X-ray digital imaging detection of GIS. Thus, it can be regarded that there was no influence of X-ray irradiation on the withstand voltage of SF 6 .

SF 6 Dissociation Test after X-ray Irradiation
To prevent the decomposition of SF 6 in GIS caused by the X-ray imaging digital detection at the experiment site, the influence of the X-ray irradiation on SF 6 under various tube voltage levels was researched. The SF 6 decomposition product detector was utilized in the test, and the accuracy of the Energies 2020, 13, 661 5 of 18 equipment was ± 0.1 µL /L. Thus as to conduct the comparison under several cases conveniently, the composition of the gas under each voltage level was initially measured and recorded by using a spectrometer. Then, the X-ray irradiation was conducted under the corresponding tube voltage for 1/3/5/7 min. Afterward, the composition of the gas was measured and recorded again. Next, the gas was replaced, and the irradiation was conducted under the other tube voltage levels. Table 1 and Figure 6 display the results.  equipment was ± 0.1 μL /L. Thus as to conduct the comparison under several cases conveniently, the composition of the gas under each voltage level was initially measured and recorded by using a spectrometer. Then, the X-ray irradiation was conducted under the corresponding tube voltage for 1/3/5/7 min. Afterward, the composition of the gas was measured and recorded again. Next, the gas was replaced, and the irradiation was conducted under the other tube voltage levels. Table 1 and Figure 6 display the results.   Through the comparison of the results, it demonstrated that the content CF4, which is an inherent ingredient of the new SF6 gas, was extremely low. And the concentration of CF4 did not increase along with the duration time and the tube voltage level of the X-ray irradiation. Moreover, H2S and SO2 were both not founded from first to last. It can be known from the test that no dissociation of SF6 was found after X-ray irradiation for 7 min under 300 kV tube voltage.

Experimental Test of X-ray Digital Imaging Detection for GIS
Since the above analyses proved that X-ray irradiation had no influence on the withstand voltage and the decomposition of SF6, several kinds of defects were tested by X-ray digital imaging detection. The experimental model filled with SF6 of 0.4 MPa was set according to a 110 kV GIS. Figure 7 shows a test example of detecting free particles inside from the bottom of the tank. This section gives successful application examples of "visual" detection, and an image database of typical defects inside of GIS is established.

Free Metal Particles
In order to simulate the X-ray digital imaging atlas of metal-free particles in GIS, the typical Xray atlas of columnar metal particles was tested. The copper particles with diameters of 1.3 mm and 0.11 mm were specially made. The lengths of metal particles L were made in three grades as 1 mm, 3 mm, and 5 mm. The X-ray digital imaging of the copper-free particles is shown below.
It can be seen that the metal particles with the two kinds of diameters in the GIS tank can be clearly imaged, and the particles in Figure 8 were proportional to their length. However, the metal particles with a diameter of 0.11 mm and a length of 1 mm were not clearly imaged. By comparing the length of columnar particles of 3 mm and 5 mm, it showed that the length of metal particles with Through the comparison of the results, it demonstrated that the content CF 4 , which is an inherent ingredient of the new SF 6 gas, was extremely low. And the concentration of CF 4 did not increase along with the duration time and the tube voltage level of the X-ray irradiation. Moreover, H 2 S and SO 2 were both not founded from first to last. It can be known from the test that no dissociation of SF 6 was found after X-ray irradiation for 7 min under 300 kV tube voltage.

Experimental Test of X-ray Digital Imaging Detection for GIS
Since the above analyses proved that X-ray irradiation had no influence on the withstand voltage and the decomposition of SF 6 , several kinds of defects were tested by X-ray digital imaging detection. The experimental model filled with SF 6 of 0.4 MPa was set according to a 110 kV GIS. Figure 7 shows a test example of detecting free particles inside from the bottom of the tank. This section gives successful application examples of "visual" detection, and an image database of typical defects inside of GIS is established. Through the comparison of the results, it demonstrated that the content CF4, which is an inherent ingredient of the new SF6 gas, was extremely low. And the concentration of CF4 did not increase along with the duration time and the tube voltage level of the X-ray irradiation. Moreover, H2S and SO2 were both not founded from first to last. It can be known from the test that no dissociation of SF6 was found after X-ray irradiation for 7 min under 300 kV tube voltage.

Experimental Test of X-ray Digital Imaging Detection for GIS
Since the above analyses proved that X-ray irradiation had no influence on the withstand voltage and the decomposition of SF6, several kinds of defects were tested by X-ray digital imaging detection. The experimental model filled with SF6 of 0.4 MPa was set according to a 110 kV GIS. Figure 7 shows a test example of detecting free particles inside from the bottom of the tank. This section gives successful application examples of "visual" detection, and an image database of typical defects inside of GIS is established.

Free Metal Particles
In order to simulate the X-ray digital imaging atlas of metal-free particles in GIS, the typical Xray atlas of columnar metal particles was tested. The copper particles with diameters of 1.3 mm and 0.11 mm were specially made. The lengths of metal particles L were made in three grades as 1 mm, 3 mm, and 5 mm. The X-ray digital imaging of the copper-free particles is shown below.
It can be seen that the metal particles with the two kinds of diameters in the GIS tank can be clearly imaged, and the particles in Figure 8 were proportional to their length. However, the metal particles with a diameter of 0.11 mm and a length of 1 mm were not clearly imaged. By comparing the length of columnar particles of 3 mm and 5 mm, it showed that the length of metal particles with

Free Metal Particles
In order to simulate the X-ray digital imaging atlas of metal-free particles in GIS, the typical X-ray atlas of columnar metal particles was tested. The copper particles with diameters of 1.3 mm and 0.11 mm were specially made. The lengths of metal particles L were made in three grades as 1 mm, 3 mm, and 5 mm. The X-ray digital imaging of the copper-free particles is shown below.
It can be seen that the metal particles with the two kinds of diameters in the GIS tank can be clearly imaged, and the particles in Figure 8 were proportional to their length. However, the metal particles with a diameter of 0.11 mm and a length of 1 mm were not clearly imaged. By comparing the length of Energies 2020, 13, 661 7 of 18 columnar particles of 3 mm and 5 mm, it showed that the length of metal particles with a diameter of 0.11 mm was still in the corresponding proportion. It is proved that the free metal particle can be well detected by X-ray digital imaging technology.
Energies 2019, 19, x FOR PEER REVIEW 7 of 18 a diameter of 0.11 mm was still in the corresponding proportion. It is proved that the free metal particle can be well detected by X-ray digital imaging technology.

Flaky Free Particles
Metal and rubber gaskets are usually used during site installation, which may be missing in the GIS tank. In order to survey the X-ray digital imaging of flaky free particles caused by gaskets, a typical X-ray atlas of both metallic and non-metallic flaky particles are tested. The thickness of metal (iron) flaky particle in Figure 9 was 0.8 mm, and the area was not less than 1 cm 2 , and the non-metallic flaky particles in Figure 10 were made by flaky rubber.

Flaky Free Particles
Metal and rubber gaskets are usually used during site installation, which may be missing in the GIS tank. In order to survey the X-ray digital imaging of flaky free particles caused by gaskets, a typical X-ray atlas of both metallic and non-metallic flaky particles are tested. The thickness of metal (iron) flaky particle in Figure 9 was 0.8 mm, and the area was not less than 1 cm 2 , and the non-metallic flaky particles in Figure 10 were made by flaky rubber. a diameter of 0.11 mm was still in the corresponding proportion. It is proved that the free metal particle can be well detected by X-ray digital imaging technology.

Flaky Free Particles
Metal and rubber gaskets are usually used during site installation, which may be missing in the GIS tank. In order to survey the X-ray digital imaging of flaky free particles caused by gaskets, a typical X-ray atlas of both metallic and non-metallic flaky particles are tested. The thickness of metal (iron) flaky particle in Figure 9 was 0.8 mm, and the area was not less than 1 cm 2 , and the non-metallic flaky particles in Figure 10 were made by flaky rubber.  a diameter of 0.11 mm was still in the corresponding proportion. It is proved that the free metal particle can be well detected by X-ray digital imaging technology.

Flaky Free Particles
Metal and rubber gaskets are usually used during site installation, which may be missing in the GIS tank. In order to survey the X-ray digital imaging of flaky free particles caused by gaskets, a typical X-ray atlas of both metallic and non-metallic flaky particles are tested. The thickness of metal (iron) flaky particle in Figure 9 was 0.8 mm, and the area was not less than 1 cm 2 , and the non-metallic flaky particles in Figure 10 were made by flaky rubber.  Figure 9 shows that the image of iron-free particles is clear, and the non-metallic flaky particle could also be distinguished in Figure 10. Considering the thickness of the gaskets used in site installation process of GIS is generally over 1 mm, the X-ray digital imaging technology can be used to detect the flaky particle omitted inside the tank clearly.

Adsorbents and Cover
In order to absorb moisture in SF 6 gas, adsorbents were installed in the adsorbent cover inside GIS. The adsorbent cover was generally made of stainless steel. Some manufacturers use plastic covers to reduce costs. However, the thermal expansion coefficient of plastic material is much larger than that of metal material. When the temperature changes greatly, the degree of thermal expansion and contraction of GIS metal tank and plastic adsorbent cover varies greatly, which makes the cover age and damage easily. Facts have proved that if the adsorbent cover breaks down and the adsorbent drops off, it will cause the internal discharge fault in the GIS. It is necessary to detect and distinguish the cover material, then it could be possible to eliminate defects of the plastic cover. Figure 11 gives the covers with two different materials.
Energies 2019, 19, x FOR PEER REVIEW 8 of 18 Figure 10. Non-metallic flaky particle. Figure 9 shows that the image of iron-free particles is clear, and the non-metallic flaky particle could also be distinguished in Figure 10. Considering the thickness of the gaskets used in site installation process of GIS is generally over 1 mm, the X-ray digital imaging technology can be used to detect the flaky particle omitted inside the tank clearly.

Adsorbents and Cover
In order to absorb moisture in SF6 gas, adsorbents were installed in the adsorbent cover inside GIS. The adsorbent cover was generally made of stainless steel. Some manufacturers use plastic covers to reduce costs. However, the thermal expansion coefficient of plastic material is much larger than that of metal material. When the temperature changes greatly, the degree of thermal expansion and contraction of GIS metal tank and plastic adsorbent cover varies greatly, which makes the cover age and damage easily. Facts have proved that if the adsorbent cover breaks down and the adsorbent drops off, it will cause the internal discharge fault in the GIS. It is necessary to detect and distinguish the cover material, then it could be possible to eliminate defects of the plastic cover. Figure 11 gives the covers with two different materials.  Because there is no partial discharge before the cover is damaged, it could not be effectively detected by live detection methods such as UHF and ultrasound. Nevertheless, X-ray digital imaging technology could be used to detect the adsorbent cover in GIS in order to avoid equipment failure. The perspective images of the adsorbent covers made of different materials are given in Figure 12. It can be seen that the outline and mesh of the plastic adsorbent cover are not clear, while the adsorbent (molecular sieve particles) and the corners of the cover can be seen clearly. This is mainly Because there is no partial discharge before the cover is damaged, it could not be effectively detected by live detection methods such as UHF and ultrasound. Nevertheless, X-ray digital imaging technology could be used to detect the adsorbent cover in GIS in order to avoid equipment failure. The perspective images of the adsorbent covers made of different materials are given in Figure 12.  Figure 9 shows that the image of iron-free particles is clear, and the non-metallic flaky particle could also be distinguished in Figure 10. Considering the thickness of the gaskets used in site installation process of GIS is generally over 1 mm, the X-ray digital imaging technology can be used to detect the flaky particle omitted inside the tank clearly.

Adsorbents and Cover
In order to absorb moisture in SF6 gas, adsorbents were installed in the adsorbent cover inside GIS. The adsorbent cover was generally made of stainless steel. Some manufacturers use plastic covers to reduce costs. However, the thermal expansion coefficient of plastic material is much larger than that of metal material. When the temperature changes greatly, the degree of thermal expansion and contraction of GIS metal tank and plastic adsorbent cover varies greatly, which makes the cover age and damage easily. Facts have proved that if the adsorbent cover breaks down and the adsorbent drops off, it will cause the internal discharge fault in the GIS. It is necessary to detect and distinguish the cover material, then it could be possible to eliminate defects of the plastic cover. Figure 11 gives the covers with two different materials.  Because there is no partial discharge before the cover is damaged, it could not be effectively detected by live detection methods such as UHF and ultrasound. Nevertheless, X-ray digital imaging technology could be used to detect the adsorbent cover in GIS in order to avoid equipment failure. The perspective images of the adsorbent covers made of different materials are given in Figure 12.  It can be seen that the outline and mesh of the plastic adsorbent cover are not clear, while the adsorbent (molecular sieve particles) and the corners of the cover can be seen clearly. This is mainly It can be seen that the outline and mesh of the plastic adsorbent cover are not clear, while the adsorbent (molecular sieve particles) and the corners of the cover can be seen clearly. This is mainly Energies 2020, 13, 661 9 of 18 due to the thin material and small density of the plastic adsorbent cover. And the outline of the cover is not clear compared with the negative.
For the metal adsorbent cover, the outline and mesh are clear, but the adsorbent is not obvious. It is because the adsorbent cover is thicker, and the stainless-steel density is larger, which increases the difficulty of X-ray penetration and the contrast of the adsorbent cover. As a result of casting technology, the cover has no corners and edges, which also reduces the electric field strength of the adsorbent cover.
According to the features above, the material of the adsorbent cover in GIS can be clearly distinguished and hidden troubles can be investigated in the visualization detection by X-ray digital imaging technology. The major problem that the adsorbent cover defect is difficult to identify and judge effectively could be solved, and this detection method provides strong technical support for condition-based maintenance of GIS equipment. Moreover, the equipment power outage and blindly disassembly inspection is also avoided.

The Loosened Metal Screw
The bolts are fixed in every structural part of the GIS equipment, and the loosening of bolts has become one of the important defects. The loosening of bolts can lead to bad alignment and contact, resulting in contact surface overheating and floating metal discharge and so on.
The normal and loosened metal screws are shown in Figure 13. The bolt loosening defects in GIS are visible under X-ray digital imaging. The slightly loosened bolt can continue live operation in a short time, but the gasket has loosened and may lead to floating discharge. After long-term operation, it will endanger the safety of the equipment. due to the thin material and small density of the plastic adsorbent cover. And the outline of the cover is not clear compared with the negative. For the metal adsorbent cover, the outline and mesh are clear, but the adsorbent is not obvious. It is because the adsorbent cover is thicker, and the stainless-steel density is larger, which increases the difficulty of X-ray penetration and the contrast of the adsorbent cover. As a result of casting technology, the cover has no corners and edges, which also reduces the electric field strength of the adsorbent cover.
According to the features above, the material of the adsorbent cover in GIS can be clearly distinguished and hidden troubles can be investigated in the visualization detection by X-ray digital imaging technology. The major problem that the adsorbent cover defect is difficult to identify and judge effectively could be solved, and this detection method provides strong technical support for condition-based maintenance of GIS equipment. Moreover, the equipment power outage and blindly disassembly inspection is also avoided.

The Loosened Metal Screw
The bolts are fixed in every structural part of the GIS equipment, and the loosening of bolts has become one of the important defects. The loosening of bolts can lead to bad alignment and contact, resulting in contact surface overheating and floating metal discharge and so on.
The normal and loosened metal screws are shown in Figure 13. The bolt loosening defects in GIS are visible under X-ray digital imaging. The slightly loosened bolt can continue live operation in a short time, but the gasket has loosened and may lead to floating discharge. After long-term operation, it will endanger the safety of the equipment.

Defect Database Establishment and Defect Risk Assessment
Taking the defects founded in actual X-ray digital imaging detection as examples, this section established atlases of different defect grades, etc., the general, serious, and critical defect. The results further prove the practicability and validity of the X-ray digital imaging technique and would be useful for the defect severity diagnosis.

False Welding without Gas Leaking
The compressed tank needs to be welded at the joint. Because the GIS tank is thick, it is easy to produce false welding. The tank may not leak gas after welding for a short time. However, after long time operation, especially when welding parts are corroded and exposed to the air directly, it will cause gas leakage and pose a threat to the safety of the equipment. Figure 14 gives an example.
Loosening Normal Figure 13. The normal and loosened metal screws.

Defect Database Establishment and Defect Risk Assessment
Taking the defects founded in actual X-ray digital imaging detection as examples, this section established atlases of different defect grades, etc., the general, serious, and critical defect. The results further prove the practicability and validity of the X-ray digital imaging technique and would be useful for the defect severity diagnosis.

False Welding without Gas Leaking
The compressed tank needs to be welded at the joint. Because the GIS tank is thick, it is easy to produce false welding. The tank may not leak gas after welding for a short time. However, after long time operation, especially when welding parts are corroded and exposed to the air directly, it will cause gas leakage and pose a threat to the safety of the equipment. Figure 14 gives an example. For GIS with false welding joints or lines in X-ray digital imaging atlas, if welding parts are found by on-site appearance inspection to be exposed in air due to the tank corrosion, it is recommended to evaluate them as an attentive state. And for non-corroded wedding joints, it can be regarded as a normal state.

Foreign Bodies in the Particle Trap
The particle trap inside the GIS is used to capture foreign bodies in metal or non-metallic impurities. Because the foreign bodies in the particle trap does not pose a threat to the electric field, the equipment can still be operated. However, if the foreign bodies are too large or accumulated for a long time, it may pose a security threat as shown in Figure 15. Generally, metal or non-metallic foreign bodies in the particle trap can be identified as general defects. If there are more than three foreign bodies or the height is higher than one-third of the particle trap, it is recommended to evaluate the state of attention.

Slight Loosening of Metal Screw in the Grading Shield
As the examples show in Figure 13, slight loosening of the screw (less than two threads) can be judged as a general defect. If the bolt or screw is only slightly loosened within two threads and still in the hood with no PD, it would be evaluated as a state of attention. For GIS with false welding joints or lines in X-ray digital imaging atlas, if welding parts are found by on-site appearance inspection to be exposed in air due to the tank corrosion, it is recommended to evaluate them as an attentive state. And for non-corroded wedding joints, it can be regarded as a normal state.

Foreign Bodies in the Particle Trap
The particle trap inside the GIS is used to capture foreign bodies in metal or non-metallic impurities. Because the foreign bodies in the particle trap does not pose a threat to the electric field, the equipment can still be operated. However, if the foreign bodies are too large or accumulated for a long time, it may pose a security threat as shown in Figure 15. For GIS with false welding joints or lines in X-ray digital imaging atlas, if welding parts are found by on-site appearance inspection to be exposed in air due to the tank corrosion, it is recommended to evaluate them as an attentive state. And for non-corroded wedding joints, it can be regarded as a normal state.

Foreign Bodies in the Particle Trap
The particle trap inside the GIS is used to capture foreign bodies in metal or non-metallic impurities. Because the foreign bodies in the particle trap does not pose a threat to the electric field, the equipment can still be operated. However, if the foreign bodies are too large or accumulated for a long time, it may pose a security threat as shown in Figure 15. Generally, metal or non-metallic foreign bodies in the particle trap can be identified as general defects. If there are more than three foreign bodies or the height is higher than one-third of the particle trap, it is recommended to evaluate the state of attention.

Slight Loosening of Metal Screw in the Grading Shield
As the examples show in Figure 13, slight loosening of the screw (less than two threads) can be judged as a general defect. If the bolt or screw is only slightly loosened within two threads and still in the hood with no PD, it would be evaluated as a state of attention. Generally, metal or non-metallic foreign bodies in the particle trap can be identified as general defects. If there are more than three foreign bodies or the height is higher than one-third of the particle trap, it is recommended to evaluate the state of attention.

Slight Loosening of Metal Screw in the Grading Shield
As the examples show in Figure 13, slight loosening of the screw (less than two threads) can be judged as a general defect. If the bolt or screw is only slightly loosened within two threads and still in the hood with no PD, it would be evaluated as a state of attention.

Metal Tip in the Grading Shield
The grading shield is a part of GIS equipment to balance voltage and reduce the electric field. Generally, the metal tip in the grading shield will not pose a threat to the operation of the equipment. However, the protruding tip, especially the tip burr, which is easy to fall off, may cause partial discharge. And it can lead to breakdown failure of GIS in serious cases. Figure 16 gives an example. The grading shield is a part of GIS equipment to balance voltage and reduce the electric field. Generally, the metal tip in the grading shield will not pose a threat to the operation of the equipment. However, the protruding tip, especially the tip burr, which is easy to fall off, may cause partial discharge. And it can lead to breakdown failure of GIS in serious cases. Figure 16 gives an example.
Thus, the defect identification basis in X-ray digital imaging is firstly determined. The sharp burrs, which are easy to fall off, can be judged as a general defect, and it is suggested to be evaluated as a state of attention. If there is no burr that is easy to fall off and only the inner corner bolt in the grading shield, it should not be judged as a general defect.

Adsorbent Cover Made of High Strength Rigid Plastics
If the hard-plastic adsorbent cover (need to be verified with the manufacturer) can be clearly identified in the X-ray digital imaging atlas, it can be regarded as a general defect. For those covers used less than 5 years, it can be evaluated as a normal state. Otherwise, it should be evaluated as an attention state.

Oblivious Loosening of Metal Screw in the Grading Shield and without PD
If the bolt or screw is loosened more than two threads, but no partial discharge occurs, it is recommended to be evaluated as a serious defect and abnormal state.

A Long Tip without PD
Long metal tips (such as contact fingers) in the grading shield will change and crack the electric field distribution. Serious cases will result in partial discharge and even equipment failure.
Thus, if a long metal tip in the grading shield (which has exceeded or will exceed the grading shield) can be clearly seen, it can be judged as a serious defect and evaluated as an abnormal state. Figure 17 shows a typical atlas. Thus, the defect identification basis in X-ray digital imaging is firstly determined. The sharp burrs, which are easy to fall off, can be judged as a general defect, and it is suggested to be evaluated as a state of attention. If there is no burr that is easy to fall off and only the inner corner bolt in the grading shield, it should not be judged as a general defect.

Adsorbent Cover Made of High Strength Rigid Plastics
If the hard-plastic adsorbent cover (need to be verified with the manufacturer) can be clearly identified in the X-ray digital imaging atlas, it can be regarded as a general defect. For those covers used less than 5 years, it can be evaluated as a normal state. Otherwise, it should be evaluated as an attention state.

Oblivious Loosening of Metal Screw in the Grading Shield and without PD
If the bolt or screw is loosened more than two threads, but no partial discharge occurs, it is recommended to be evaluated as a serious defect and abnormal state.

A Long Tip without PD
Long metal tips (such as contact fingers) in the grading shield will change and crack the electric field distribution. Serious cases will result in partial discharge and even equipment failure.
Thus, if a long metal tip in the grading shield (which has exceeded or will exceed the grading shield) can be clearly seen, it can be judged as a serious defect and evaluated as an abnormal state. Figure 17 shows a typical atlas.

Adsorbent Cover Made of General Plastics
If the general-plastic adsorbent cover (need to be verified with the manufacturer) can be determined as Figure 18 indicated, it is recommended to be evaluated as a serious defect. For the post-commissioning equipment, it can be evaluated as an abnormal state.

No absorbent and Normal Moisture
After the equipment is put into operation, the moisture content of SF6 in the tank will naturally increase due to the difference of water pressure inside and outside the tank and the accumulation of water on the surface of the components inside the tank. The excessive moisture content will reduce the insulation performance of internal insulation materials to a certain extent, especially for solid insulation materials. In order to avoid the impact of excessive moisture on the safe and stable operation of the equipment, the adsorbent will be placed inside the GIS, which has the effect of absorbing water and reducing the moisture in SF6. However, sometimes the factory or site installation personnel do not strictly implement the process flow, which would cause the forgotten of adsorbents installation, like Figure 19 shows.

Adsorbent Cover Made of General Plastics
If the general-plastic adsorbent cover (need to be verified with the manufacturer) can be determined as Figure 18 indicated, it is recommended to be evaluated as a serious defect. For the post-commissioning equipment, it can be evaluated as an abnormal state.

Adsorbent Cover Made of General Plastics
If the general-plastic adsorbent cover (need to be verified with the manufacturer) can be determined as Figure 18 indicated, it is recommended to be evaluated as a serious defect. For the post-commissioning equipment, it can be evaluated as an abnormal state.

No absorbent and Normal Moisture
After the equipment is put into operation, the moisture content of SF6 in the tank will naturally increase due to the difference of water pressure inside and outside the tank and the accumulation of water on the surface of the components inside the tank. The excessive moisture content will reduce the insulation performance of internal insulation materials to a certain extent, especially for solid insulation materials. In order to avoid the impact of excessive moisture on the safe and stable operation of the equipment, the adsorbent will be placed inside the GIS, which has the effect of absorbing water and reducing the moisture in SF6. However, sometimes the factory or site installation personnel do not strictly implement the process flow, which would cause the forgotten of adsorbents installation, like Figure 19 shows.

No Absorbent and Normal Moisture
After the equipment is put into operation, the moisture content of SF 6 in the tank will naturally increase due to the difference of water pressure inside and outside the tank and the accumulation of water on the surface of the components inside the tank. The excessive moisture content will reduce the insulation performance of internal insulation materials to a certain extent, especially for solid insulation materials. In order to avoid the impact of excessive moisture on the safe and stable operation of the equipment, the adsorbent will be placed inside the GIS, which has the effect of absorbing water and reducing the moisture in SF 6 . However, sometimes the factory or site installation personnel do not strictly implement the process flow, which would cause the forgotten of adsorbents installation, like Figure 19 shows.
If the missing of absorbent is found in the X-ray digital imaging atlas, while the moisture has been inspected to be not exceeding the standard, it can be judged as a serious defect and abnormal state. If the missing of absorbent is found in the X-ray digital imaging atlas, while the moisture has been inspected to be not exceeding the standard, it can be judged as a serious defect and abnormal state.

Partial Spring Missing
Due to poor manufacturing and installation of GIS, the spring of the disconnector plug-in contact head is missing. When the spring is missing, the contact pressure will be reduced and the contact resistance will be increased, which can lead to bad contact alignment and bad contact, resulting in contact overheating, suspension discharge, and other defects. In serious cases, it can lead to the failure of GIS equipment.
Missing of partial spring in disconnector but without partial discharge detected, it could be evaluated as an abnormal state. Even if the measured circuit resistance is qualified, it should still be regarded as a serious defect.

Foreign Bodies in the Tank
The presence of metal or non-metal foreign bodies in the tank is the main reason for the failure of GIS. The metal-free particles in the tank, which is the main cause of the breakdown failure, are the most harmful to the equipment.
Thus, if foreign bodies are found in the tank (not in the particle trap or grading shield), it is recommended to be evaluated as a critical defect. For those found in the particle trap or grading shield, it can be judged according to the relevant content of general and serious defects.

The Metal Screw Loosened Obviously and Have PD
If the screw is loosened apparently (more than two threads) and partial discharge characteristic is detected, it consequently can be judged as a crisis defect and recommended to be evaluated as a serious state. Figure 20 shows the X-ray detection atlas of examples. Figure 19. Abnormal and normal absorbent.

Partial Spring Missing
Due to poor manufacturing and installation of GIS, the spring of the disconnector plug-in contact head is missing. When the spring is missing, the contact pressure will be reduced and the contact resistance will be increased, which can lead to bad contact alignment and bad contact, resulting in contact overheating, suspension discharge, and other defects. In serious cases, it can lead to the failure of GIS equipment.
Missing of partial spring in disconnector but without partial discharge detected, it could be evaluated as an abnormal state. Even if the measured circuit resistance is qualified, it should still be regarded as a serious defect.

Foreign Bodies in the Tank
The presence of metal or non-metal foreign bodies in the tank is the main reason for the failure of GIS. The metal-free particles in the tank, which is the main cause of the breakdown failure, are the most harmful to the equipment.
Thus, if foreign bodies are found in the tank (not in the particle trap or grading shield), it is recommended to be evaluated as a critical defect. For those found in the particle trap or grading shield, it can be judged according to the relevant content of general and serious defects.

The Metal Screw Loosened Obviously and Have PD
If the screw is loosened apparently (more than two threads) and partial discharge characteristic is detected, it consequently can be judged as a crisis defect and recommended to be evaluated as a serious state. Figure 20 shows the X-ray detection atlas of examples. The broken adsorbent cover or the overflowing adsorbent caused by seasonal variation or sudden temperature changing will crack the electric field distribution, which will result in equipment failure in serious cases. Thus, this kind of defect can be evaluated as a critical defect.

Damage of Plastic Adsorbent Cover
The broken adsorbent cover or the overflowing adsorbent caused by seasonal variation or sudden temperature changing will crack the electric field distribution, which will result in equipment failure in serious cases. Thus, this kind of defect can be evaluated as a critical defect.

Missing Key Components
In extreme cases, some key parts of the equipment, such as a conductive rod, grading shield, are missing or partially missing because of the poor quality of the installation process. Figure 21 shows two examples of a conductive rod missing and partial contactor missing, which have been successfully found in the substation field by X-ray digital imaging detection. Figure 20. The obviously loosened metal screws in the grading shield with partial discharge.

Damage of Plastic Adsorbent Cover
The broken adsorbent cover or the overflowing adsorbent caused by seasonal variation or sudden temperature changing will crack the electric field distribution, which will result in equipment failure in serious cases. Thus, this kind of defect can be evaluated as a critical defect.

Missing Key Components
In extreme cases, some key parts of the equipment, such as a conductive rod, grading shield, are missing or partially missing because of the poor quality of the installation process. Figure 21 shows two examples of a conductive rod missing and partial contactor missing, which have been successfully found in the substation field by X-ray digital imaging detection. This kind of defect will pose a great threat to the safe and stable operation of the equipment, and even directly lead to failure. As a result, the missing key components found in X-ray digital imaging can be judged as critical defects, and it is recommended to be evaluated as a serious state.

No Absorbent and Excessive Moisture
As previous content mentioned, the excessive moisture which will reduce the insulation performance of the internal insulation materials is a serious threat to equipment safety. If the missing of absorbent is found while the moisture exceeded the standard, it can be judged as a critical defect and evaluated as a serious state.

Typical Defect Database of X-ray Visualization and Risk Assessment
According to the analysis above, the different types of defects are delimited into three grades in Table 2, thus as to provide a helpful reference for the judgment and identification of the typical defects in GIS. (A: general grade, B: serious grade, C: critical grade) Table 2. Risk assessment of typical defects in gas insulated switchgear (GIS). This kind of defect will pose a great threat to the safe and stable operation of the equipment, and even directly lead to failure. As a result, the missing key components found in X-ray digital imaging can be judged as critical defects, and it is recommended to be evaluated as a serious state.

No Absorbent and Excessive Moisture
As previous content mentioned, the excessive moisture which will reduce the insulation performance of the internal insulation materials is a serious threat to equipment safety. If the missing of absorbent is found while the moisture exceeded the standard, it can be judged as a critical defect and evaluated as a serious state.

Typical Defect Database of X-ray Visualization and Risk Assessment
According to the analysis above, the different types of defects are delimited into three grades in Table 2, thus as to provide a helpful reference for the judgment and identification of the typical defects in GIS. (A: general grade, B: serious grade, C: critical grade). 6. Discussion

Personnel Protection
What needs to be noticed is that the X-ray is a kind of ionizing radiation with very short wavelengths and harmful for living things. When X-ray irradiates an organism for a long time, cells can be inhibited, destroyed, or even died. The operation environment of GIS is in the substation and there are no animals or plants, only operators. Therefore, it is particularly important to take effective protection measures in X-ray field detection to ensure the safety of the operators.
Different countries have regulations on the protection area. For example, the area with a dose equivalent rate above 15 µSv/h is defined as the control area according to the standard GBZ 117-2015 in China [32]. In previous research, the radiation field, as well as the relationship between radiation dose rate and the protection distance, have been studied through the comparison and analysis of theoretical calculation and experimental detection [23,33]. The results show that: For the maximum tube voltage and tube current under 300 kV and 5 mA, respectively, in the actual detection, the air protection distance for the operator is less than 57 m at the angle between 135 • and 225 • in the direction of the X-ray beam. Moreover, operating in a building and setting a lead sheet are both tested to be effective for the radiation protection of the operator and public.

The Radiation Time and the Effect
As mentioned before, the X-ray irradiation in the actual detection of GIS is generally in a short time of less than 3 min. Thus, the tests in Section 3 were carried out up to 7 min, which is already exceeded the general irradiation time and long enough to investigate the influence in engineering application. In fact, if the effect of X-ray digital imaging detection is not so well within a short time, expanding the time could also be invalid commonly. Some other measures have to be taken in these cases, such as increasing tube voltage. For the long term of X-ray irradiations, the influence to SF 6 and the metal tank of GIS may be different. This scientific question is not covered in this paper and will be considered in future research.

Limitations
The detection based on the X-ray also has limitations. This method is highly dependent on the detection position and angle to obtain the best detection imaging. However, the ideal detecting position may not be achieved as a result of the limitation of the narrow space of GIS, especially for the equipment part of phase B, which is in the middle between phases A and C. The miniaturized testing equipment is useful in this case.
Testing personnel experience is also an important factor, which can reduce the number of repetitive tests. The image database of typical defects established in this paper synthesizes a great deal of experimental experience, including the common defect location, and it will be helpful for the practicing engineers.

Conclusions
The detection and diagnosis of defects in GIS based on X-ray digital imaging technology are profoundly investigated in this research, helping to improve and guide the actual application. The conclusions are summarized below.

1.
The experimental platform, including CR digital imaging system and the GIS model, is set up and the test tooling, which can be filled with SF 6 in a certain pressure, is designed. The withstand voltage test and dissociation test of SF 6 after X-ray irradiation under different tube voltages are launched, which indicates that the decomposition characteristics and withstand voltage of SF 6 gas are unaffected. 2.