The use of neural networks for the prediction of wear loss and surface roughness of AA 6351 aluminium alloy

doi:10.1016/j.matdes.2004.09.011

Materials & Design

Volume 27, Issue 2, 2006, Pages 156-159

https://doi.org/10.1016/j.matdes.2004.09.011 Get rights and content

Abstract

Artificial neural networks (ANNs) are a new type of information processing system based on modeling the neural system of human brain. Effects of ageing conditions at various temperatures, load, sliding speed, abrasive grit diameter in 6351 aluminum alloy have been investigated by using artificial neural networks. The experimental results were trained in an ANNs program and the results were compared with experimental values. It is observed that the experimental results coincided with ANNs results.

Introduction

Precipitation hardening or age hardening, is produced by solution treating and quenching an alloy in which a second phase is in solid solution at the elevated temperature but precipitates upon quenching and ageing at a lower temperature [1]. Precipitation hardening is applied to the some aluminium alloys. This strengthening is one of the most important hardening methods used to increase strength in aluminium alloys [2]. With convenient alloying and heat treatment, hardness can be increased as 40 times compared to high purity aluminium [3], [4].

For precipitation hardening to occur, the second phase must be soluble at an elevated temperature but must exhibit decreasing solubility with decreasing temperature. By contrast, the second phase in dispersion-hardening systems has very little solubility in the matrix, even at elevated temperatures. Usually there is atomic matching, or coherency, between the lattices of the precipitate and the matrix, while in dispersion-hardened systems there generally is no coherency between the second-phase particles and the matrix. The requirement of a decreasing solubility with temperature places a limitation on the number of useful precipitation-hardening alloy systems [5], [6], [7].

A great deal of research on the two body abrasive wear behavior of aluminum alloys has been carried out experimentally [8], [9]. The most important reason for the damage and consequently failure of machine parts is wear. A lot of experiments must be conducted in order to find an appropriate technology that renders an aluminum alloy of the maximum strength [10]. This wastes both man power and money.

In recent years artificial neural networks (ANNs) have emerged as a new branch of computing, suitable for applications in a wide range of fields. Artificial neural networks have been recently introduced into tribology by Jones et al. [11]. In this study, experimental and ANNs results have been compared. A lot of studies have been published in which the prediction of various parameters on aluminum alloys were investigated systematically [10], [12]. Song, Zhang, Tseng and Zang investigated the application of artificial neural networks on ageing dynamics in AA 7175 aluminum alloys. Several authors used ANNs for comparing experimental results [8]. For this reason artificial neural networks have been used to search the optimum technology adapted to AA 6351 aluminum alloy.

The aim of the present work was to investigate the wear loss and surface roughness of AA 6351 aluminum alloy with the use of artificial neural networks.

Section snippets

Material and experimental procedure

In this study, AA 6351 wrought alloy (AlMgSi1) was used as test material. The alloy was supplied from ETİALUMINYUM (Turkey). The chemical composition of the used material is given in Table 1.

The AA 6351 aluminum alloy is used in ship manufacturing due to its strength, bearing capacity to sea atmosphere, ease of workability and weldability. It is also used in building boat, column, chimney, rod, mould, pipe, tube, vehicle, bridge, crane and roof. One of the most important properties of AA 6351

Modeling with neural networks

Computers are an integral part of day to day activities in engineering design and engineers have utilized various applications to assist them improve their design [14], [15]. ANN mimic some basic aspects of the brain functions [13]. ANNs are based on the neural structure of the human brain, which processes information by means of interaction between many neurons [9], [14]. In the past few years there has been a constant increase in interest of neural network modeling in different fields of

Results and discussion

In this study, AA 6351 alloy subject to wear test and surface roughness at various environmental conditions and ageing temperatures were investigated. The following results were obtained.

AA 6351 aluminum alloy was kept at 520 ± 2 °C for an hour then taken in solid solution and then quenched at room temperature in water. Some specimens were aged at room temperature naturally where as others were aged artificially at 150, 160, 170 and 180 °C. They were worn using a pin-on disc model wear test

References (15)

H. Kaçar et al.
The effect of precipitation-hardening conditions on wear behaviours at 2024 aluminium wrought alloy
J Mater Process Technol
(2003)
C. Meriç
Physical and mechanical properties of cast under vacuum aluminium alloy 2024 containing lithium additions
Mater Res Bull
(2000)
R. Song et al.
The application of artificial neural networks to the investigation of ageing dynamics in 7175 aluminium alloys
Mater Sci Engineer
(1995)
W.Q. Song et al.
The effect of thermal ageing on the abrasive wear behavior of age hardening 2014 Al/Sic and 6061 Al/Sic composites
Wear
(1995)
S. Kenig et al.
Control of properties in injection molding by neural networks
Eng Appl Artif Intel
(2001)
M. Perzyk et al.
Prediction of ductile cast iron quality by artificial neural networks
J Mater Process Technol
(2001)
M.Y. Rafiq et al.
Neural network design for engineering applications
Comput Struct
(2001)

There are more references available in the full text version of this article.

Cited by (77)

A review of surface quality control technology for robotic abrasive belt grinding of aero-engine blades
2023, Measurement: Journal of the International Measurement Confederation
Higher requirements are being imposed on blade surface quality due to the continuous improvement of thrust-to-weight ratio and endurance in advanced aero-engines. Robot abrasive belt grinding offers the advantages of excellent flexibility, convenient scheduling, strong adaptability, and low cost. It aligns well with the development trend of multi-specification and small-batch blade processing, serving as an effective method to enhance blade contour accuracy and surface integrity. Nonetheless, the surface quality and contour accuracy of robot belt grinding are significantly influenced by factors such as the robot’s low repeated positioning accuracy, weak structural stiffness, and elastic deformation during belt grinding. Consequently, the design and development of an efficient, flexible, and high-precision robotic belt grinding system hold significant importance in improving machining accuracy and controlling blade surface quality. Currently, there is extensive research on robotic belt grinding of aviation blades. However, there is a lack of comprehensive analysis of these technologies. To address this issue, we conducted an extensive literature review and utilized an inductive classification research approach to assess and analyze multiple key technologies associated with the processing technology. The study reveals the utilization of optimization algorithms and AI intelligent algorithms in various key technologies. Application of these algorithms results in a 70% improvement in the calibration accuracy of the robot system and a 20.5% enhancement in processing efficiency after trajectory planning. Furthermore, the integration of multiple sensors and analysis algorithms enables the prediction error of belt wear to be below 10% and achieves a prediction accuracy higher than 85% for surface roughness and residual stress. This paper aims to provide a comprehensive summary of the technology development related to surface quality control in robot belt grinding for aerial blades. It offers a range of technical options for building an intelligent and high-precision robot belt grinding system and contributes to the further advancement of surface quality control in blade grinding. Lastly, the paper presents a future outlook on the development of related technologies.
A machine learning approach to predict the wear behaviour of steels
2023, Tribology International
One or both surfaces, which come in contact with a relative motion between them, can experience material loss due to wear. This is a complex phenomenon involving several parameters consisting of both the material and experimental conditions. It is thus very much difficult to predict the volume loss under a specific condition as a consequence of wear. In this study, an effort was given to develop a machine learning approach involving several parameters such as composition, microstructure, hardness, load, sliding distance, temperature etc to quantify and predict the material loss. The outcomes obtained from the model were found to be logical with existing knowledge. The model predictions were validated with experimental results not used to build up the model.
A novel ensemble deep learning model for cutting tool wear monitoring using audio sensors
2022, Journal of Manufacturing Processes
Tool wear is an important parameter in the machining because the production, cost and performance is highly depend upon its performance. Therefore, the monitoring of cutting tool wear plays an important role in mechanical machining processes. With this aim, the present work deals with the application of novel ensemble deep learning model for cutting tool wear monitoring using audio sensors. The tool wear data during machining was extracted with an audio denoising technique combined with Fast Fourier Transform (FFT) and bandpass filters and dependent component analysis (DCA). Then, the ensemble convolutional neural networks (CNN) detection model was trained and audio signals were converted into audio images with different algorithms. Finally, the results confirm that this novel method is very accurate to predict the tool wear values under different cutting conditions.
Surface modification of multi-directional forged biodegradable Mg-Zn alloy by ball burnishing process: Modeling and analysis using deep neural network
2021, Journal of Manufacturing Processes
Citation Excerpt :
Author reported that R-value of 0.99965 with a maximum error of 2.675% is achieved to verify the prediction performance of the ANN model and the reliability and accuracy of the ANN model have been further verified by the test sets. Durmus et al. [36] investigated use of neural networks for the prediction of wear loss and surface roughness of AA6351 aluminum alloy. Author reported that the results obtained in ANN application are close to test results.
This research investigates the feasibility of improving surface integrity of the Mg-2%Zn alloy via a novel combined process of multidirectional forging (MDF) with ball burnishing technique. Mg-2%Zn alloy was subjected to multidirectional forging up to 5 passes at 280 °C and same sample is ball burnished with depth of press of 0.3 mm, feed of 300 mm/min and force of 250 N for 1 pass. Microstructure (optical and transmission electron microscope), phases (X ray diffraction), roughness (atomic force microscope) and mechanical properties (microhardness and tensile) were analyzed for cast, MDF processed and MDF + ball burnished samples. The study also defines a new reliability simulation framework to predict the roughness and hardness of the MDF + ball burnished samples using Deep Neural Network (DNN) approach. For the analysis, depth of press, feed and number of passes served as inputs for the DNN model. After configuring and training procedure of the model, the predicted roughness and microhardness of the MDF + ball burnished were compared with the results of experimental data. A large increase of microhardness was observed for MDF-5 + BB-1 (5 pass MDF processed Mg-2%Zn alloy + Ball burnished sample), found to be 98 Hv and surface roughness of 306 nm was achieved. According to the results, the obtained R-squared value is very close to the 99.5 percentage from the model for predicting roughness and hardness which is a promising result. Corrosion behavior of the AS-1, MDF-5 and MDF-5 + BB-1 was evaluated in Hank's balanced salt solution. Corrosion rate of the MDF-5 + BB-1 samples were found to be 0.87 mm/y which is 7 times better than that of AS-1 samples (6.10 mm/y). Grain refinement and residual stresses were higher in the MDF-5 + BB-1 sample, which helped significantly to enhance corrosion resistance.
A beetle antennae search improved BP neural network model for predicting multi-factor-based gas explosion pressures
2020, Journal of Loss Prevention in the Process Industries
A gas explosion in an underground structure may cause serious damage to the human body and ground buildings and may result in huge economic losses. The pressure of the gas explosion is an important parameter in determining its severity and designating an emergency plan. However, existing empirical and computational fluid dynamics (CFD) methods for pressure prediction are either inaccurate or inefficient when considering multiple influencing factors and their interrelationships. Therefore, for a more efficient and reliable prediction, the present study developed a multifactorial prediction model based on a beetle antennae search (BAS) algorithm improved back propagation (BP) neural network. A total of 317 sets of data which considered factors of geometry, gas, obstacle, vent, and ignition were collected from previous studies. The results showed that the established model can predict pressures accurately by low RMSE (43.4542 and 50.7176) and MAPE (3.9666% and 4.9605%) values and high R² (0.7696 and 0.7388) values for training and testing datasets, respectively. Meanwhile, the BAS algorithm was applied to improve both the calculation efficiency and the accuracy of the proposed model by enabling a more intelligent hyperparameter tuning method. Furthermore, the permutation importance of input variables was investigated, and the length (L) and the ratio of length and diameter (L/D) of geometry were found to be the most critical factors that affect the explosion pressure level.
Artificial neural network technique to predict the properties of multiwall carbon nanotube-fly ash reinforced aluminium composite
2019, Journal of Materials Research and Technology
Citation Excerpt :
Artificial neural network, an artificial intelligence modelling technique is a supervised learning algorithm. Artificial neural network is a reliable prediction technique and can be successfully employed in prediction of mechanical properties composites [22–29]. The final properties of the bulk composite depend on the content (wt.
In this study, prediction of density and hardness properties using artificial neural network (ANN) and micro structural evolution of multi walled carbon nano tubes (MWCNT) and fly ashes (FA)/Al composites produced by powder metallurgy were investigated. The influence of content (wt.%) of reinforcements(MWCNTs and FA), ball milling time and sintering time on the mechanical properties were experimentally determined by measuring density and hardness values which are the outputs obtained from the artificial neural network. It was found that amount of reinforcements, ball milling time and sintering time play a major role in dispersion and enhancement of the properties. It was also demonstrated that ANN model is a powerful prediction technique to predict the mechanical properties of the composites. Blend powder morphology and sintered composite structure were investigated by scanning electron microscope (SEM). It was found that reinforcements were well dispersed for prolonged ball milling time and sintering time.

View all citing articles on Scopus

View full text

Technical reportThe use of neural networks for the prediction of wear loss and surface roughness of AA 6351 aluminium alloy

Abstract

Introduction

Section snippets

Material and experimental procedure

Modeling with neural networks

Results and discussion

J Mater Process Technol

Mater Res Bull

Mater Sci Engineer

Wear

Eng Appl Artif Intel

J Mater Process Technol

Comput Struct

Technical report
The use of neural networks for the prediction of wear loss and surface roughness of AA 6351 aluminium alloy