Artificial Neural Networks in Supply Chain Management, A Review

Artificial Neural Networks (ANNs) are a type of machine learning algorithm inspired by the structure and function of the human brain. In the context of supply chain management, ANNs can be used for demand forecasting, inventory optimization, logistics planning


Introduction
Artificial Neural Networks (ANNs) can play a significant role in development of supply chain management by improving various aspects of the supply chain, such as demand forecasting, inventory management, logistics optimization, and risk analysis.ANNs are computational models inspired by the biological neural networks in the human brain, and they can learn from data to recognize patterns and make predictions (Zhu et al., 2023).It's important to note that while ANNs can provide valuable insights and re-Also, the availability of high-quality and relevant data is crucial for training accurate and reliable ANN models.To create an artificial neural network, the best-predicted network structure should be found through trial and error, along with selecting a set of input variables (Fanoodi et al., 2019).An artificial neural network structure employing the back-propagation learning rule is shown in the Fig. 1 (Menhaj, 2005).
In the fundamental realm of neural networks, "back-propagation" and "node" are crucial concepts (Buscema, 1998).The "backpropagation" refers to the primary training algorithm used to update the weights of the network by minimizing the difference between the predicted output and the actual output.It's a crucial component of training neural networks.It utilizes the chain rule of calculus to calculate the gradient of the loss function with respect to the weights, allowing for adjustments that reduce the error in subsequent iterations (Shukla et al., 2021).A "node" in a neural network represents a unit within a layer.It's where computations occur.In a neural network, nodes are organized into layers: an input layer, one or more hidden layers, and an output layer.Each node receives input from the nodes in the previous layer, performs a computation (often a weighted sum of inputs), applies an activation function, and passes the result to the nodes in the subsequent layer (Zhang et al., 2021).
In supply chain management, ANNs can be used for a range of applications, such as demand forecasting, inventory management, logistics optimization, and risk management.ANNs can analyze historical sales data, customer trends, and other relevant factors to predict future demand accurately (Sang, 2021;Toorajipour et al., 2021).This enables warehouse managers to optimize inventory levels, reduce stockouts, and ensure efficient order fulfillment.ANN-based approaches provide a data-driven and predictive modeling technique that can effectively handle the complexity and uncertainty present in supply chain systems.In inventory optimization, ANNs can be utilized to analyze historical data, identify patterns, and forecast demand accurately (Li, 2021;Nunes et al., 2020).ANNs can analyze real-time data from multiple sources, such as sales, procurement, and production, to optimize inventory levels (Leung, 1995).By considering factors like lead time, demand variability and production capacity, ANNs can generate accurate reorder points, safety stock levels, and economic order quantities (Aburto and Weber, 2007).Smart and connected supply chain management is shown in the Fig. 2 (Gupta et al., 2021).The figure shows how the Internet of Things (IoT) devices and technology could encourage an effective supply chain while simultaneously mitigating the spread of COVID-19 (Gupta et al., 2021).
By understanding the relationships between various factors such as sales data, lead times, seasonality, promotions, and other relevant variables, ANNs can provide valuable insights for inventory management decision-making.ANNs can be utilized in supply chain management for various tasks, including supplier evaluation and selection.The process of supplier selection is a critical decision-making process that involves assessing and choosing the most suitable suppliers based on specific criteria and requirements (Akbari and Do, 2021).ANNs can aid in this process by analyzing large volumes of data, identifying patterns, and making predictions.To route optimization in supply chain management, ANNs can be used to analyze historical data, real-time data, and various parameters related to the transportation network, such as traffic conditions, distances, delivery time windows, vehicle capacities, and customer demands (Lima-Junior and Carpinetti, 2019).The network can learn from this data and develop a model that predicts optimal routes based on the input parameters.By leveraging ANNs for risk analysis and mitigation in supply chain management, organizations can enhance their ability to identify, evaluate, and proactively address potential risks (Helo and Hao, 2022).These technologies enable more effective decision-making, improve supply chain resilience, and reduce the overall impact of disruptions on business operations (Seyedan and Mafakheri, 2020).ANNs can help optimize the order picking process by determining the most Fig. 1.An artificial neural network structure employing the back-propagation learning rule (Menhaj, 2005).
M. Soori, B. Arezoo and R. Dastres Journal of Economy and Technology 1 (2023) 179-196 efficient picking routes within the warehouse (Guanghui, 2012).By analyzing factors like item popularity, order frequency, and warehouse layout, ANNs can suggest optimized picking sequences in order to minimize travel time and increase productivity in part production (Silva et al., 2017;Schniederjans et al., 2020).Soori et al (Soori et al., 2017(Soori et al., , 2014(Soori et al., , 2013(Soori et al., , 2016)).suggested virtual machining techniques to evaluate and enhance CNC machining in virtual environments.Soori et al (Soori et al., 2023a).explored machine learning and artificial intelligence in CNC machine tools to boost productivity and improve profitability in production processes of component employing CNC machining operations.To improve the performance of machined components, Soori and Arezoo (Soori and Arezoo, 2022a) reviewed the topic of measuring and reducing residual stress in machining operations.To improve surface integrity and decrease residual stress during Inconel 718 grinding operations, Soori and Arezoo (Soori and Arezoo, 2022a) proposed the optimum machining parameters employing the Taguchi optimization method.In order to increase the life of cutting tools during machining operations, Soori and Arezoo (Soori and Arezoo, 2022b) examined different method of tool wear prediction algorithms.Dastres and Soori (Dastres and Soori, 2021a) addressed improvements in web-based decision support systems to give solutions for data warehouse management using decisionmaking assistance.Dastres and Soori (Dastres and Soori, 2021b) reviewed applications of artificial neural networks in different sections, such as analysis systems of risk, drone navigation, evaluation of welding, and evaluation of computer simulation quality, to explore the execution of artificial neural networks for improving the effectiveness of products.To enhance network and data online security, Dastres and Soori (Dastres and Soori, 2020) suggested the secure socket layer.Dastres and Soori (Dastres and Soori, 2021d) studied the developments in web-based decision support systems to develop the methodology of decision support systems by evaluating and suggesting the gaps between proposed approaches.Recent developments in published articles are examined by Soori et.al (Soori et al., 2023b). in order to assess and improve the impacts of artificial intelligence, machine learning, and deep learning in advanced robotics.Soori and Arezoo (Soori and Arezoo, 2023c) studied the impact of coolant on the cutting temperature, roughness of the surface, and tool wear during turning operations with Ti6Al4V alloy.Recent developments from published papers are reviewed by Soori (Soori, 2023a) in order to examine and alter composite materials and structures.Soori et al (Soori et al., 2023a).examined the Internet of things application for smart factories in industry 4.0 to increase quality control and optimize part manufacturing processes.To minimize cutting tool wear during drilling operations, Soori and Arezoo (Soori and Arezoo, 2023d) designed a virtual machining system.Soori and Arezoo (Soori and Arezoo, 2023e) decreased residual stress and surface roughness to improve the quality of items produced utilizing abrasive water jet machining.In order to analyze and optimize energy consumption in industrial robots, different methods of energy usage optimization were reviewed by Soori et al (Soori et al., 2023b).
In order to monitor and improve supply chain management performance throughout the part production process, artificial neural networks are studied in the research work.As a result, by reviewing and analyzing previous successes in the applications of artificial neural networks in supply chain management, new ideas and concepts can be generated to fill any gaps.As a consequence, by boosting supply chain management using the application of artificial neural networks, the productivity of part manufacture can be increased.(Gupta et al., 2021).

Demand forecasting
Artificial Neural Networks (ANNs) have gained significant popularity in development process of supply chain management, particularly in demand forecasting.ANNs are a class of machine learning algorithms inspired by the structure and function of biological neural networks in the human brain.They are well-suited for complex pattern recognition and can effectively model nonlinear relationships in data (Kochak and Sharma, 2015).ANNs can analyze historical sales data, market trends, and other relevant factors to predict future demand more accurately.This information helps optimize inventory levels, production planning, and resource allocation (Barlas and Gunduz, 2011;Thomassey, 2010).ANNs can be used to predict future demand patterns based on historical sales data, market trends, and other relevant factors (Guo et al., 2021).By training an ANN model with historical sales data and other input variables, the model can learn complex relationships and patterns, allowing it to generate accurate demand forecasts (Allaoui et al., 2019).These forecasts help organizations optimize inventory levels, production schedules, and procurement activities.
When it comes to demand forecasting in supply chain management, ANNs can provide several benefits: 1. Non-linearity: ANNs can capture non-linear relationships between demand and various factors such as historical sales data, promotional activities, seasonality, economic indicators, and more.The process allows companies for more accurate demand predictions compared to traditional statistical methods (Wen et al., 2020).2. Pattern recognition: ANNs can enhance the capabilities of recognizing complex patterns in different data.They can identify hidden trends and dependencies that might be difficult to capture using traditional forecasting techniques.So, the process can detect subtle demand patterns and improve forecast accuracy during supply chain process (Fanoodi et al., 2019;Zhang et al., 2019).3. Adaptability: ANNs are capable of adapting to changing market conditions and demand patterns.As new data becomes available, the network can be retrained to incorporate the latest information, making the forecasts more up-to-date and reliable (Huang et al., 2021).4. Handling large datasets: Supply chain data can be extensive and include a variety of variables.ANNs can effectively handle large datasets, including both structured and unstructured data, to uncover meaningful insights and generate accurate forecasts (Efendigil et al., 2009;Meidute-Kavaliauskiene et al., 2022). 5. Demand segmentation: ANNs can also be used for demand segmentation, where they group customers or products into segments based on their demand patterns.This helps in developing tailored forecasting models for different segments, resulting in more precise predictions and improved inventory management (Khaldi et al., 2017).
Distributed simulation of order promising protocols in supply chain networks is shown in the Fig.
3 (Kiralp and Venkatadri, 2010).Each supply chain network in the multi-enterprise is a meta-network that consists of several supply chains.Every supply chain network in the multi-enterprise which can be divided into networks for suppliers, customers, or both (Kiralp and Venkatadri, 2010).As a result, an advanced distributed network control centre is proposed in order to synchronize the planning cycle clock and enable data and inventory transfer via a central mailbox system (Kiralp and Venkatadri, 2010).(Kiralp and Venkatadri, 2010).
A machine learning approach for predicting hidden links for a given node pair in supply chain with graph neural networks is shown in the Fig. 4 (Kosasih and Brintrup, 2022).
In order to forecast whether a connection will exist for a specific pair of nodes, the subgraph G(x, y) that surrounds the two nodes (x, y) are extracted.Then, the two nodes' ego graph as the enclosing subgraph are employed (Kosasih and Brintrup, 2022).Consequently, a machine learning method using graph neural networks for supply chain prediction of hidden linkages is proposed (Kosasih and Brintrup, 2022).
Overall, ANNs offer significant potential for demand forecasting in supply chain management.By leveraging their capabilities, organizations can improve their forecasting accuracy, optimize inventory levels, enhance customer service, and ultimately achieve better supply chain efficiency (Abolghasemi et al., 2020).

Inventory optimization
ANNs can assist in optimizing inventory levels by analyzing historical data, lead times, seasonality, and external factors.By accurately predicting demand fluctuations, they can help reduce excess inventory and avoid stockouts, leading to improved operational efficiency and cost savings.ANNs can assist in optimizing inventory levels by predicting demand fluctuations and identifying the optimal reorder points.ANNs can help organizations to determine the right inventory levels in order to minimize costs while meeting customer demand and maintaining operational efficiency (Praveen et al., 2019).ANNs can be used in inventory optimization within supply chain management to address various challenges, such as demand forecasting, inventory control, and replenishment strategies (Sustrova, 2016).
ANNs can be trained on historical sales data, customer behavior, economic indicators, and other relevant factors in order to predict future demand patterns.By analyzing complex relationships and nonlinearities in the data, ANNs can provide accurate demand forecasts, which help in optimizing inventory levels and reducing stock-outs or overstocking (He, 2013).ANNs can assist in determining optimal inventory control policies, such as reorder points and order quantities.By considering factors like lead time, variability in demand, and desired service levels, ANNs can learn the relationships between these variables and recommend inventory control decisions that minimize costs while meeting customer demand (de Paula Vidal et al., 2022).ANNs can be employed to optimize the timing and frequency of replenishment orders (Fan et al., 2013).By analyzing historical data and factors such as supplier lead times, transportation constraints, and order costs, ANNs can identify the most efficient replenishment strategies that minimize inventory costs and ensure timely availability of products (Nezamoddini et al., 2020).
ANNs can help optimize inventory control by determining the appropriate reorder points, safety stock levels, and order quantities (Ahmadimanesh et al., 2020).They can learn from historical data and patterns in order to predict optimal inventory levels that balance the trade-off between stockouts and excess inventory.By continuously adapting to changing demand patterns, ANNs improve inventory management efficiency.Hybrid process for smart inventory management is shown in the Fig. 5 (Mohamed, 2019).
To lower the possibility of mistakes during the data extraction and acquisition process, an intelligent inventory management system is developed (Mohamed, 2019).Therefore, smart warehouse management system which combines a network architecture neuron with a bar code scanner has been proposed to obtain real-time information about every item that is stocked (Mohamed, 2019).

Fig. 4.
A machine learning approach for predicting hidden links for a given node pair in supply chain with graph neural networks (Kosasih and Brintrup, 2022).
M. Soori, B. Arezoo and R. Dastres Journal of Economy and Technology 1 (2023) 179-196 Lead time is the amount of time a supplier needs to deliver products after receiving an order.Accurate lead time prediction is crucial for inventory optimization.ANNs analyze historical lead time data, along with variables like supplier performance, transportation delays, and production schedules in order to estimate lead times more accurately (Guillermo-Muñoz et al., 2020).This information can provide better planning and reduce the risk of stockouts or excess inventory in advanced supply chain process.
Effective utilization of ANN inventory Optimization in of supply chain management requires careful data preparation, model training, and validation to ensure accurate and reliable results (Bodendorf et al., 2022).Additionally, combining ANNs with other optimization techniques, such as mathematical models or expert knowledge, can further enhance the inventory optimization process in supply chain management (Zhou et al., 2019).

Supplier evaluation and selection
Artificial Neural Networks (ANNs) have been widely utilized in various fields, including supply chain management (SCM), for tasks such as supplier selection.Supplier selection is a critical aspect of SCM, as it involves identifying and evaluating potential suppliers based on multiple criteria to ensure the efficient functioning of the supply chain.Supplier evaluation and selection is a critical decision-making process that involves evaluating and choosing the most suitable suppliers based on multiple criteria and objectives (Liu and Ran, 2020).ANNs can be used to evaluate and select suppliers based on various criteria, including quality, delivery performance, price, and reliability.By analyzing past supplier data and performance metrics, ANNs can identify the most suitable suppliers for specific requirements (Azadnia et al., 2012).ANNs can be used to analyze supplier performance and assist in supplier selection and evaluation processes (Setak et al., 2012).By training an ANN model with historical supplier data, quality metrics, and other relevant variables, organizations can identify patterns and assess supplier performance in terms of delivery reliability, quality, pricing, and other criteria (Kuo et al., 2010a).This information can help in making accurate and optimized decisions about supplier selection and ongoing supplier management.
ANNs offer several advantages for supplier evaluation and selection in SCM due to their ability to handle complex patterns and relationships within the data.Here are some ways ANNs can be applied to supplier selection: 1. Data analysis and pattern recognition: ANNs can analyze large volumes of historical supplier data, including performance records, quality metrics, delivery times, and pricing information.By recognizing patterns and relationships within this data, ANNs can identify suppliers that exhibit desirable characteristics and predict future supplier performance (Tsai and Hung, 2016).2. Multi-criteria decision-making: Supplier selection involves considering multiple criteria simultaneously, such as cost, quality, delivery time, and reliability.ANNs handle such multi-criteria decision-making by incorporating various input variables and assigning weights to them based on their importance.The network can then compute a supplier's overall suitability score, allowing for effective supplier ranking and selection.3. Risk assessment and prediction: ANNs can assess and predict supplier-related risks by incorporating both quantitative and qualitative data.This can include analyzing factors like financial stability, market reputation, supplier location, and past performance.By considering these variables, ANNs can help identify high-risk suppliers and minimize potential disruptions in the supply chain (Nezamoddini et al., 2020).4. Real-time decision support: ANNs can be integrated into real-time decision support systems for supplier evaluation and selection.
By continuously monitoring and analyzing data streams, such as market conditions, supplier performance, and customer demands, ANNs can provide timely insights and recommendations for supplier selection decisions.5. Supplier relationship management: ANNs can also be used to manage and optimize supplier relationships.By analyzing data on past interactions, collaborations, and customer feedback, ANNs can identify patterns and suggest strategies for improving relationships with suppliers, leading to better supply chain performance (Ghorbani et al., 2012).6. Supplier Evaluation: Once the ANN is trained and validated, it can be used to evaluate new suppliers.By inputting supplier data into the trained ANN, it can provide predictions or rankings based on the supplier's suitability and performance (Kuo et al., 2010b).7. Green supplier evaluation and selection: Green supplier selection refers to the process of identifying and evaluating suppliers based on their environmental performance and sustainability practices.The efficiency and effectiveness of green supplier selection processes can be enhanced due to complex data analysis using the ANNs (Kuo et al., 2010b).8. Decision making: The output of the ANN can be combined with other decision-making criteria to make a final decision regarding supplier selection such as cost considerations, risk analysis, and strategic alignment in supply chain management (Fashoto et al., 2016).
The green supplier selection structure in supply chain management is shown in the Fig. 6 (Kuo et al., 2010b).
To develop green supplier selection criteria, a Delphi expert questionnaire is created and distributed among 10 supply chain and environmental experts (Kuo et al., 2010b).The final green supplier selection has six dimensions including "Quality," "Cost," "Delivery," "Service," "Environment" and "Corporate social responsibility" (Kuo et al., 2010b).As a result, a green supplier selection using integration of artificial neural network and MADA methods is proposed (Kuo et al., 2010b).
However, ANNs offer significant potential for supplier selection in SCM, their effectiveness relies on the availability of highquality data, appropriate model design, and careful validation.Moreover, ANNs should be used in conjunction with other decisionmaking methods and expert knowledge based systems in order to provide comprehensive and optimized supplier selection processes.

Route optimization
In the context of supply chain management route optimization, ANNs can be used to model complex relationships between various factors such as transportation costs, delivery times, vehicle capacities, traffic conditions, and customer demands.ANNs can optimize logistics and transportation routes by considering factors like traffic conditions, delivery schedules, and vehicle capacities (Teng, 2021).By analyzing real-time data and historical patterns, ANNs can suggest the most efficient routes, reducing transportation costs and delivery times (Ren et al., 2022;Carter and Ferrin, 1995).Here's an overview of how ANNs can be applied: 1. Data collection: Relevant data points are collected, including historical transportation data, geographical information, customer demand patterns, and any other factors that may influence the routing decisions (Chan et al., 2021).2. Data pre-processing: The collected data is pre-processed to remove outliers, handle missing values, normalize numerical features, and encode categorical variables into a suitable format for training the neural network (Biswas et al., 2019).3. Network architecture design: The structure of the neural network needs to be defined, including the layers number, the neurons number in each layer, and the functions of activation.Various architectures like feedforward networks, recurrent neural networks (RNNs), or convolutional neural networks (CNNs) can be considered based on the problem requirements during supply chain management (Jafarzadeh-Ghoushchi and Rahman, 2016).4. Training the network: The prepared dataset is used to train the neural network.The network learns in order to optimize routes by adjusting its internal parameters through an iterative process known as backpropagation.The objective is to minimize a defined cost function that quantifies the quality of the routes (Bhattacharya et al., 2014). 5. Route prediction: Once the neural network is trained, it can be used to predict optimized routes for new input scenarios.For example, given a set of customer locations, delivery deadlines, and other relevant factors, the network can generate an optimized route plan that minimizes transportation costs or reduces delivery times (Zhou et al., 2020).6. Evaluation and refinement: The predicted routes can be evaluated against predefined metrics such as cost, time, or resource utilization.This evaluation provides feedback to improve the network's performance.If necessary, the network can be fine-tuned or retrained with additional data to enhance its route optimization capabilities (Liu, 2021).
Swarm-neural network for a smart transportation system using logistic agents is shown in the Fig. 7 (Alkinani et al., 2022).
The group of dispersed edge devices gathers the sensory vehicular data, which is then briefly stored in the resource-constrained edge devices during the data-gathering phase before being forwarded to the data processing phase (Alkinani et al., 2022).This approach involves sampling data, which is then sent into the feature selection algorithm, which uses the sampled data to compute features.Now that the feature has been chosen, it may be sent to the SWNN for logistic type classification (Alkinani et al., 2022).As a result, logistic agentbased swarm-neural network is applied in order to present an intelligent transportation system (Alkinani et al., 2022).Model for optimizing logistics distribution routes based on recursive fuzzy neural networks is shown in the Fig. 8 (Liu, 2021).As a result, in order to enhance efficiency in path optimization system of e-commerce, an advanced model for optimizing logistics distribution routes using a recursive fuzzy neural network method is proposed (Liu, 2021).
Applying ANNs in supply chain route optimization holds promise for enhancing logistics efficiency, reducing costs, and improving customer satisfaction by ensuring timely and optimized deliveries (Meidute-Kavaliauskiene et al., 2022).Overall, ANNs offer the advantage of learning complex patterns and relationships from data, allowing them to provide optimized routing solutions in dynamic supply chain environments (Attaran, 2020).

Risk analysis and mitigation
Artificial Neural Networks (ANNs) have been increasingly used in supply chain management for risk analysis and mitigation.ANNs are a form of machine learning method that was motivated by the design and operation of biological neural networks.They are made up of networked artificial neurons that can learn from information and form hypotheses or judgments.ANNs can assess and predict supply chain risks, such as disruptions, delays, or quality issues (Lorenc and Kuźnar, 2021).By analyzing data from multiple sources, including social media, news feeds, and sensor data, ANNs can provide early warnings and recommend proactive measures to mitigate potential risks.ANNs can be used to identify and mitigate supply chain risks by analyzing historical data and external factors (Liu, 2022).By training an ANN model with historical risk data and relevant variables, it can identify patterns and correlations which can be used in assessing the likelihood and impacts of potential risks, such as supplier disruptions, natural disasters and market volatility (Kosasih et al., 2022).This information allows organizations to take proactive measures to mitigate risks and improve supply chain resilience.
ANNs can be used to analyze and mitigate risks at various stages of the supply chain, including procurement, production, transportation, and distribution during advanced supply chain management (Jianying et al., 2021).Here are some ways ANNs are used in supply chain risk analysis and mitigation: 1. Demand Forecasting: ANNs can be used to predict future demand patterns based on historical sales data, market trends, and other relevant factors.Accurate demand forecasting helps in inventory management and reduces the risk of stockouts or excess inventory (Aamer et al., 2020).2. Supplier Evaluation: ANNs can assess supplier performance by analyzing various factors such as quality, delivery reliability, and pricing.By considering historical data and other relevant variables, ANNs can identify high-risk suppliers and help in making informed decisions about supplier selection and management (Hui and Choi, 2016).3. Quality Control: ANNs can analyze quality-related data to identify patterns and anomalies that may indicate potential quality issues.By monitoring and analyzing data from production processes and quality inspections, ANNs can help in early detection and mitigation of quality-related risks (Cai et al., 2020;Minis, 2007).4. Transportation Optimization: ANNs can optimize transportation routes and schedules by considering factors such as distance, traffic conditions, and delivery deadlines.By minimizing transportation costs and ensuring timely delivery, ANNs contribute to reducing supply chain disruptions and risks (Liu et al., 2016). 5. Inventory Optimization: ANNs can analyze inventory data, including stock levels, lead times, and demand patterns in order to maintain optimal amounts of inventory and reduce the possibility of stockouts or surplus inventory.By considering various variables and predicting future demand, ANNs help in maintaining an optimal inventory level (Radhakrishnan et al., 2009).6. Risk Identification and Mitigation: ANNs can be trained to identify patterns and indicators of potential risks in the supply chain.
By examining information from multiple sources, such as historical performance data, market trends, and external factors (e.g., natural disasters), ANNs can be applied in early risk detection and support decision-making for risk mitigation strategies (Liu, 2022;Ali et al., 2019).
Risks of sustainable supply chain based on optimized BP neural networks in fresh grape industry is evaluated in order to enhance sustainability in fresh grape supply chain process (Jianying et al., 2021).Neural network analysis of risk in sustainable supply chain is shown in the Fig. 9 (Jianying et al., 2021).
ANNs should be used in conjunction with other analytical methods, expert knowledge, and appropriate risk management strategies to effectively address and mitigate supply chain risks (Wang, 2021).

Warehouse management
Artificial Neural Networks (ANNs) have been increasingly applied in various areas of supply chain management, including advanced warehouse management systems.ANNs are computer models that take their cues from the structure and operation of the human brain.They are made up of networked artificial neurons which communicate and analyze information.ANN-based approaches have shown promise in improving various aspects of warehouse management, such as demand forecasting, inventory management, and order picking optimization (Xiao and Hu, 2017).ANNs can enhance warehouse operations by optimizing storage locations, pick paths, and resource allocation.By learning from historical data and real-time inputs, ANNs can improve order picking efficiency, reduce labor costs, and minimize errors (Boone et al., 2019).
Here are some specific applications of ANNs in warehouse management: 1. Demand forecasting: ANNs can be used to predict future demand based on historical sales data, market trends, and other relevant factors.By analyzing patterns and relationships within the data, ANNs can provide more accurate demand forecasts, which can help optimize inventory levels, reduce stockouts, and minimize holding costs (Vairagade et al., 2019).2. Inventory management: ANNs can assist in optimizing inventory levels by considering various factors such as demand patterns, lead times, and supply constraints.By continuously analyzing and learning from data, ANNs can provide recommendations for order quantities, reorder points, and safety stock levels to ensure efficient inventory management (Praveen et al., 2019).3. Order picking optimization: Order picking is a critical operation in warehouse management that involves selecting items from their storage locations to fulfill customer orders.ANNs can optimize order picking processes by identifying the most efficient picking routes, minimizing travel distances, and improving the overall productivity of warehouse workers (Dumitrascu et al., 2020).
4. Quality control: ANNs can be used to detect and classify defects or anomalies in products or packaging materials.By analyzing sensor data or visual inputs, ANNs can learn to identify patterns associated with faulty items, enabling proactive quality control measures and reducing the risk of shipping defective products (Wang et al., 2017). 5. Route optimization: For larger warehouse operations with multiple distribution centers or complex logistics networks, ANNs can assist in optimizing transport routes.By considering factors such as distance, traffic conditions, delivery schedules, and vehicle capacities, ANNs can provide recommendations for efficient route planning, reducing transportation costs and improving delivery performance (Lin et al., 2022).
The successful implementation of ANNs in warehouse management requires sufficient and accurate data for training and ongoing monitoring.Additionally, it's essential to consider the specific characteristics and requirements of each warehouse operation to tailor the ANN models accordingly (Slimani et al., 2015).Overall, ANNs have the potential to enhance warehouse management processes by providing more accurate predictions, optimizing decision-making, and improving operational efficiency in supply chain management.

Customer sentiment analysis
Customer sentiment analysis is another trendy area where ANNs can be harnessed to improve SCM efficiency and formulate strategic SCM strategies (Swain and Cao, 2019).Analyzing unstructured data from social media and downstream customer feedback is an essential facet of understanding customer demands (Swain and Cao, 2019).By analyzing customer sentiments, businesses can gain valuable insights into their preferences, satisfaction levels, and concerns (Singh et al., 2018).Here's how sentiment analysis contributes to improving supply chain intelligence: 1. Demand forecasting: Understanding customer sentiment helps predict demand more accurately.Positive sentiments might indicate an increase in demand for certain products or services, while negative sentiments might signal a potential decline (Seyedan and Mafakheri, 2020).2. Inventory management: Sentiment analysis can assist in optimizing inventory levels.Positive sentiment around certain products might suggest the need for increased stock, while negative sentiment might prompt inventory adjustments or promotions to clear surplus.3. Supplier management: Knowing customer sentiment can also be crucial in managing suppliers.If certain suppliers consistently provide products that receive positive sentiment, it might be wise to strengthen those relationships.Conversely, negative sentiment could prompt revaluation or re-negotiation (Sathyan et al., 2021).4. Risk management: Sentiment analysis can help identify potential risks.If negative sentiment arises around a specific component or supplier, it might indicate a risk of disruption, allowing for proactive measures to mitigate these risks (Ganesh and Kalpana, 2022). 5. Product development: Analyzing sentiments related to current products or services can provide insights for future developments.
Understanding what customers like or dislike can guide product improvements or new launches (Purnama and Masruroh, 2023).6. Supply chain efficiency: Sentiment analysis can reveal patterns that impact the supply chain's efficiency.For instance, positive sentiment might be linked to faster delivery times or ease of returns, whereas negative sentiment might point towards inefficiencies in logistics (Ahmed et al., 2022).
By applying the sentiment analysis tools, natural language processing, and machine learning in supply chain, businesses can extract actionable insights from a vast amount of customer data.As a result, improving supply chain intelligence based on sentiment analysis leads to better service, faster deliveries, and enhanced overall customer experience, fostering loyalty and positive word-of-mouth.

Dynamic pricing
Dynamic pricing in supply chain intelligence involves the use of real-time data, analytics, and algorithms to adjust prices for goods or services based on various factors such as demand, supply, competitor pricing, and market conditions (Yang et al., 2022).Dynamic pricing (or revenue management in some industries) has consistently been a vital topic within SCM.ANNs can be effectively deployed to help supply chain firms in devising dynamic pricing strategies (Mayer, 2023).Using a combination of past purchasing and selling trends, compiled market data, and customized price rules, technology automates pricing procedures inside a business.The process can establish pricing ranges for logistics services based on variables such as demand, competition, lanes, equipment and types of transportation (Mayer, 2023).Supply chain intelligence is pivotal in dynamic pricing as it provides insights into various aspects like: 1. Demand Forecasting: Predicting consumer demand by analyzing historical data, market trends, and external factors.This information helps in adjusting prices according to anticipated demand fluctuations (Guizzardi et al., 2021).2. Real-time data: Constantly monitoring market changes, competitor pricing, and customer behavior allows for immediate adjustments to prices (Kayikci et al., 2022).3. Optimizing inventory: Understanding inventory levels and logistics help in determining the right pricing strategies to manage stock levels efficiently.4. Customer segmentation: Tailoring prices for different customer segments based on their purchasing behavior and willingness to pay (Victor et al., 2018). 5. Elasticity of demand: Analyzing how sensitive customers are to price changes helps in setting optimal prices that maximize revenue (Chen et al., 2018).
Dynamic pricing strategy allows companies to respond to changes in demand, supply, or other market variables swiftly.By integrating supply chain intelligence, businesses can leverage data analytics to optimize pricing strategies based on factors such as inventory levels, consumer behavior, competitor pricing, and even external factors like weather or geopolitical events.As a result, the method can help to maximize profits and maintain competitiveness in a fluctuating conditions of marketing.

Supply chain optimization
Artificial neural networks (ANNs) are a powerful tool used in supply chain management to optimize various aspects of the supply chain, including inventory management, demand forecasting, supply chain planning, and logistics optimization.ANNs are a type of machine learning model inspired by the structure and function of biological neural networks, and they excel at processing complex, nonlinear patterns and relationships in data (Reynolds et al., 2019).ANNs can integrate multiple aspects of the supply chain, such as demand forecasting, inventory management, and production planning, to optimize the overall supply chain performance (Senthil and Muthukannan, 2022).By considering the interdependencies between different processes, ANNs can identify opportunities for improvement and provide recommendations for decision-making (Abdallah et al., 2022).Here are some specific areas where artificial neural networks which can be employed in order to optimize supply chain management: 1. Demand Forecasting: In order to accurately estimate future demand, ANNs can evaluate previous sales data, industry trends, and other pertinent information.These forecasts can help in inventory planning, production scheduling, and overall supply chain optimization (Benkachcha et al., 2014).2. Inventory Management: ANN models can optimize inventory levels by considering factors such as demand variability, lead times, and service level targets.By dynamically adjusting reorder points and order quantities, ANNs can reduce stockouts while minimizing inventory holding costs (Bansal et al., 1998).3. Supplier Selection and Evaluation: ANNs can assist in supplier selection by analyzing various supplier attributes and performance indicators.By considering factors such as delivery times, quality metrics, and pricing data, ANNs can provide recommendations for the most suitable suppliers (Sharifnia et al., 2021).
4. Warehouse Optimization: ANNs can optimize warehouse operations by predicting order picking times, identifying optimal storage locations for different products, and improving inventory placement strategies.These optimizations can lead to reduced order processing times and improved order fulfillment rates (Du et al., 2020). 5. Transportation Routing and Scheduling: ANNs can analyze historical transportation data, traffic patterns, and delivery constraints to optimize routing and scheduling decisions.By considering factors such as delivery windows, vehicle capacities, and road conditions, ANNs can optimize transportation routes, leading to reduced costs and improved delivery efficiency.6. Risk Management: ANNs can help identify and mitigate supply chain risks by analyzing historical data, external factors, and market conditions.By identifying patterns and correlations, ANNs can assist in proactive risk management and enable more resilient supply chains (Xu et al., 2019).
ANNs can provide powerful tools for supply chain optimization process, however, effectiveness of ANNs implementation depends on the availability and quality of data.Proper data pre-processing, feature engineering, and model training are crucial for achieving accurate and reliable results.Additionally, ANNs should be used in conjunction with other optimization techniques and domain expertise to make informed supply chain decisions.

Transportation and logistics optimization
Artificial Neural Networks (ANNs) have been increasingly used in supply chain management for transportation and logistics optimization.ANNs are a subset of machine learning algorithms inspired by the structure and functioning of biological neural networks in the human brain.They excel at pattern recognition, nonlinear relationships, and complex data analysis, making them well-suited for solving optimization problems in supply chain and logistics (Chan et al., 2021).ANNs can assist in optimizing transportation and logistics operations by predicting transit times, optimizing routing, and improving resource allocation (Chen, 2022).By analyzing historical transportation data, ANNs can identify patterns and factors that affect delivery times and optimize logistics processes to minimize costs and improve service levels (Tsolaki et al., 2022).Here are some key areas where ANNs can be applied to improve transportation and logistics optimization in supply chain management: 1. Demand forecasting: ANNs can be used to predict future demand patterns based on historical data, market trends, and other relevant variables.Accurate demand forecasting enables companies to optimize their inventory levels, production schedules, and transportation planning (Zavvar Sabegh et al., 2017).2. Route optimization: By considering variables including distance, traffic, weather, and delivery window, ANNs can assist in determining the most effective transportation routes.Businesses may save transportation costs, speed up deliveries, and increase overall logistics efficiency by optimizing their routes (Noorul Haq and Kannan, 2006).3. Vehicle routing and scheduling: ANNs can assist in optimizing vehicle routing and scheduling decisions by considering various factors such as delivery locations, vehicle capacity, time constraints, and customer preferences.This helps in minimizing empty miles, maximizing resource utilization, and reducing fuel consumption (Sharifnia et al., 2021).4. Warehouse management: ANNs can optimize warehouse operations by analyzing historical data on inventory levels, order patterns, and storage capacities.They can assist in determining optimal storage locations, inventory replenishment strategies, and order picking sequences, leading to improved warehouse efficiency (Fradinata et al., 2019). 5. Last-Mile delivery optimization: ANNs can aid in optimizing last-mile delivery operations, which is often the most expensive and challenging part of the supply chain.By considering variables like delivery locations, traffic conditions, and customer preferences, ANNs can help determine the most efficient delivery routes and schedules, reducing costs and improving customer satisfaction.6. Risk management: ANNs can analyze various risk factors, such as supply disruptions, weather events, and market volatility in order to provide insights for proactive risk management in transportation and logistics.By predicting and mitigating potential risks, companies can better protect their supply chain operations and minimize disruptions (Wu et al., 2023;Choi et al., 2019).
Management of supply chains and logistics for businesses engaged in international trade is shown in the Fig. 10 ( Xie and Chen, 2022).
In order to transmit positioning data and control signals, the data transmitting device is connected to the cloud service processor, positioning means, and temperature alarm device.The data collectors are linked via the network to get control signals and location data (Xie and Chen, 2022).To store location data and manage signal transmission, the cloud storage is coupled to one or more cloud service processors or data collectors (Xie and Chen, 2022).Thus, positional data, control signals, and their presentation are all possible with the display terminal.Consequently, it is possible to improve logistics transportation supply in order to support the growth of the small and micro machinery businesses (Xie and Chen, 2022).
ANNs offer significant potential for optimizing supply chain management in transportation and logistics systems (Baghizadeh et al., 2021).However, the successful implementation of the ANNs in supply chain management requires high-quality data, appropriate model training, and ongoing refinement (Świderski et al., 2018).Also, other machine learning techniques, such as genetic algorithms or reinforcement learning, can enhance the ANNs in order to address specific optimization challenges in supply chain management.So, the supply chain management for transportation and logistics optimization can be developed using the ANNs.

Conclusion and future research work directions
The structure and operation of biological neural networks constitute the basis for ANNs, a subset of machine learning techniques.They are made up of a network of artificial neurons that can learn from information and anticipate or decide.Biological neural networks in the human brain served as the basis for the development of computing models known as ANNs.They are composed of linked nodes and synthetic neurons which transmit and process information.Artificial Neural Networks (ANNs) have proven to be effective tools in various fields, including supply chain management.ANNs have shown great potential in various areas of supply chain management (SCM), such as demand forecasting, inventory management and optimization, supplier selection, route optimization, risk analysis and mitigation, warehouse management, supply chain optimization, logistics optimization, and risk analysis.Artificial Neural Networks (ANNs) can indeed be utilized for route optimization in supply chain management.ANNs offer the advantage of learning complex patterns from data and can provide optimized routes for supply chain management by leveraging historical and real-time information.ANNs can optimize the routing of incoming and outgoing shipments within the warehouse by considering factors like vehicle capacity, traffic conditions, and delivery time windows.This helps reduce transportation costs, increase delivery efficiency, and improve customer satisfaction.ANNs can optimize logistics operations by analyzing factors such as transportation costs, route optimization, and delivery schedules.They can assist in determining the most efficient routes, optimal transportation modes, and scheduling deliveries to minimize costs and maximize customer satisfaction.ANNs can aid in optimizing production planning by analyzing historical production data, order backlog, and other constraints.They can predict production outcomes and recommend production schedules that maximize efficiency, minimize bottlenecks, and reduce costs.
Artificial Neural Networks (ANNs) have shown great potential in addressing various challenges in supply chain management (SCM).As for future research directions, here are a few areas where ANNs could be further explored and applied in SCM: 1. Demand forecasting: By examining previous sales data, industry trends, and outside influences, ANNs can be utilized to increase demand forecasting precision.Future research could focus on developing advanced ANN models that can handle large-scale datasets, incorporate unstructured data (such as social media sentiment), and adapt to dynamic market conditions.2. Data integration and collaboration: Supply chain management involves multiple stakeholders and vast amounts of data.Future research can explore the integration of ANNs with emerging technologies like blockchain or distributed ledger technology to enable secure and efficient data sharing, collaboration, and coordination among supply chain partners.3. Advanced forecasting models: ANNs have been widely used for demand forecasting in SCM.Future research can focus on developing advanced ANN architectures that can capture complex patterns and dependencies in demand data.This can involve the integration of deep learning techniques, such as convolutional neural networks (CNNs) or recurrent neural networks (RNNs), to improve accuracy and handle time series data more effectively.4. Quality Control: ANNs can analyze sensor data from various quality control systems within the warehouse, such as barcode scanners and image recognition systems.By learning from historical data, ANNs can detect patterns and anomalies, enabling early identification of quality issues and reducing the risk of defective products reaching customers.5. Multi-objective optimization: Supply chain decisions often involve multiple conflicting objectives, such as cost minimization and service level maximization.Future research can explore the use of ANNs for multi-objective optimization in SCM, where the network can learn to balance different objectives and provide decision-makers with a range of Pareto-optimal solutions.6. Inventory optimization: ANNs can assist in optimizing inventory management by predicting optimal inventory levels based on factors like demand patterns, lead times, and supply constraints.Future research could explore ANNs' ability to handle complex multi-echelon supply chain networks, incorporate demand uncertainties, and account for factors like perishability and seasonality.7. Supply chain risk management: By examining numerous data sources, such as supplier performance, weather patterns, geopolitical variables, and economic indicators, ANNs can assist in identifying and mitigating supply chain risks.Future research could focus on developing ANNs that can provide real-time risk assessment, recommend proactive risk mitigation strategies, and support decision-making under uncertainty.8. Sustainable supply chain management: With the growing emphasis on sustainability, ANNs can be leveraged to optimize supply chain operations while minimizing environmental impact (Liu et al., 2021).Future research can explore the use of ANNs for sustainable supply chain management, including carbon footprint estimation, green logistics optimization, and eco-design of products (Lim et al., 2022).9. Real-time decision-making: Supply chain operations often require real-time decision-making to respond to dynamic market conditions and disruptions.Future research can investigate the use of ANNs for real-time decision support systems in SCM.This can involve developing online learning algorithms that can continuously update network parameters based on incoming data, allowing for adaptive decision-making in dynamic environments.10.Supply chain risk management: ANNs can play a crucial role in assessing and managing supply chain risks.Future research can focus on developing ANN-based models to identify, predict, and mitigate risks in supply chains.This can involve integrating external data sources, such as social media or news feeds, to enhance the accuracy of risk prediction models.11.Explainability and interpretability: It might be difficult to comprehend the logic behind ANN models' predictions due to the fact that they are sometimes viewed as "black boxes."Future research can focus on developing methods to enhance the explainability and interpretability of ANN-based models in SCM.This can involve techniques such as attention mechanisms, feature importance analysis, or rule extraction to provide insights into decision-making processes.12. Supplier selection and evaluation: ANNs can assist in supplier selection and evaluation processes by considering multiple criteria, such as cost, quality, lead time, and reliability.Future research could explore the application of ANNs in automatically identifying relevant supplier evaluation criteria, analyzing large amounts of supplier-related data, and supporting more dynamic and collaborative supplier management approaches.13.Transportation optimization: ANNs can be utilized to optimize transportation operations, such as route planning, carrier selection, and mode choice.Future research could focus on developing ANNs that can incorporate real-time data, adapt to dynamic traffic conditions, and support sustainable transportation decisions by considering factors like carbon emissions and energy consumption.14.Blockchain-enabled supply chain analytics: ANNs can be combined with blockchain technology to enhance transparency, traceability, and trust in supply chain operations.Future research could explore the integration of ANNs with blockchain to develop secure and privacy-preserving analytics frameworks for supply chain data, enabling improved visibility and accountability across the supply chain.15.Human-AI collaboration in SCM: Future research could also focus on investigating how ANNs can facilitate effective collaboration between humans and AI systems in supply chain decision-making.This could involve developing ANN-based decision support systems that provide interpretable recommendations, enable interactive and intuitive interfaces, and support human learning and adaptation.
These are just a few potential areas for future research where ANNs can be applied to further enhance supply chain management.As technology advances and new challenges emerge, ANNs will continue to play a crucial role in shaping the future of SCM.

Fig. 10 .
Fig. 10.Management of supply chains and logistics for businesses engaged in international trade (Xie and Chen, 2022).