Supply chain risk analysis method and system based on artificial intelligence big data

Abstract

The invention discloses a supply chain risk analysis method and system based on artificial intelligence big data, and relates to the technical field of supply chain risk analysis, and the method comprises the following steps: obtaining all data related to commodities and suppliers from an internal system and an external source of an enterprise, and providing comprehensive data support for subsequent analysis and evaluation; based on comprehensive data, feature engineering is carried out, key features capable of reflecting the dependency relationship between commodities and suppliers are extracted, and core features capable of reflecting the dependency relationship between commodities and suppliers are extracted from big data. According to the invention, by integrating an internal system and external data and utilizing feature engineering and a machine learning model, the dependency relationship between commodities and suppliers is quantified intelligently and changes are monitored in real time. Therefore, a purchaser can take effective measures before the potential risk outbreaks, the supply interruption risk is reduced, and the continuity of a production line is guaranteed. For a long time, countermeasures are optimized, a continuous risk management mechanism is established, and the toughness and stability of the supply chain are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of supply chain risk analysis, and in particular to a supply chain risk analysis method and system based on artificial intelligence big data. Background Art

[0002] Supply chain risk analysis based on AI and big data involves the use of advanced AI algorithms and massive data processing technologies to identify, assess, and predict various potential risk factors within the supply chain. By collecting and analyzing big data from diverse sources, such as market trends, supplier performance, logistics information, weather changes, and socioeconomic indicators, AI can identify complex risk patterns and correlations, providing early warning of potential supply chain disruptions, quality issues, or cost fluctuations. This approach not only improves the accuracy and efficiency of risk identification but also supports companies in developing more scientific and dynamic risk response strategies, enhancing supply chain resilience and sustainability, and ultimately improving overall operational stability and competitiveness.

[0003] The existing technology has the following deficiencies: Existing technologies often struggle to accurately assess a product's dependence on suppliers through intelligent means in supply chain risk analysis. When certain key products are highly dependent on specific suppliers, and existing technology is unable to intelligently analyze and proactively identify these dependencies, a supply disruption can directly stall a company's production lines and prevent order fulfillment. This situation is particularly severe in the manufacturing industry, potentially paralyzing the entire production chain.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0005] The purpose of the present invention is to provide a supply chain risk analysis method and system based on artificial intelligence big data. By integrating the internal enterprise system and external data sources, comprehensive data support is ensured, providing a reliable basis for accurately evaluating the dependency relationship between goods and suppliers. Through the application of feature engineering and machine learning models, the degree of dependence of goods on suppliers can be intelligently quantified, and changes in dependencies can be monitored in real time. This allows purchasers to take effective countermeasures before potential risks break out, reduce the risk of supply interruptions, and ensure the continuity of production lines. In the long run, it can not only optimize the effectiveness of countermeasures, but also establish a continuous risk management mechanism to prevent the recurrence of similar risks, improve the resilience and stability of the supply chain, and solve the problems in the above-mentioned background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a supply chain risk analysis method based on artificial intelligence big data, comprising the following steps: Acquire all data related to products and suppliers from internal enterprise systems and external sources to provide comprehensive data support for subsequent analysis and evaluation; Based on comprehensive data, feature engineering is performed to extract key features that reflect the dependency between products and suppliers. Core features that reflect the dependency between products and suppliers are extracted from big data to provide data dimension support for quantitative analysis. The extracted features are dynamically quantified according to the monitoring window to construct a time series of dependency relationships. The dynamically quantified features are used as input variables and fed into the trained machine learning model. The dependency relationship between products and suppliers is quantitatively evaluated based on the training results. Based on the assessment results and the set dependency risk threshold, it determines whether there is a potential risk of high dependency between the product and the supplier. Through intelligent analysis, it promptly identifies high dependency risks and provides early warning information to buyers so that they can take countermeasures to alleviate the high dependency of the product on a single supplier. After implementing countermeasures, continue to obtain supply chain data on products and suppliers, and analyze changes in real-time dependencies during multiple subsequent monitoring windows to verify the effectiveness of countermeasures. Continuously monitor the real-time dependencies between products and suppliers to ensure that dependency risks continue to decrease. Summarize and optimize the effectiveness of response measures and establish long-term mechanisms to prevent the recurrence of high-dependence risks.

[0007] Preferably, all data related to products and suppliers is obtained from the enterprise's internal systems and external sources. The specific steps are as follows: First, integrate internal enterprise data, including extracting key information from ERP, supplier management systems, and inventory and logistics management systems; Next, collect external data sources, including industry reports, market monitoring platforms, logistics tracking services, and publicly available supplier credit assessments and financial health data; Finally, through data cleaning and standardization, the integrity and consistency of internal enterprise data and external data sources are ensured, laying the foundation for subsequent feature extraction and analysis.

[0008] Preferably, key features that can reflect the dependency relationship between goods and suppliers are extracted, wherein the extracted features include the degree of concentration of the proportion of goods supplied by a single supplier in the total supply and the degree to which the historical delivery success rate of the goods depends on a single supplier. The extracted features are dynamically quantified according to the monitoring window, and a supply concentration reference value and a fulfillment dependence reference value are generated respectively. The time series of the dependency relationship is constructed through the supply concentration reference value and the fulfillment dependence reference value, and the dependency relationship between the goods and suppliers is dynamically quantified.

[0009] Preferably, the supply concentration reference value and fulfillment dependence reference value generated by dynamically quantifying the concentration degree of the proportion of goods supplied by a single supplier in the total supply and the degree to which the historical delivery success rate of goods depends on a single supplier are used as input variables and input into the trained machine learning model to generate a dependency index. The dependency relationship between goods and suppliers is quantitatively evaluated based on the dependency index output by the machine learning model.

[0010] Preferably, the specific steps of extracting the concentration of the proportion of goods supplied by a single supplier in the total supply volume and dynamically quantifying the extracted features according to the monitoring window to generate a supply concentration reference value are as follows: By calculating the supply weight of a single supplier in the total supply volume, we provide basic data for the supply concentration reference value and quantify the supplier's contribution to the supply of goods. The calculation expression for the supply weight is: in: For suppliers k The supply weight of For suppliers k The number of products offered, For suppliers k The supply distance is used to reflect the potential impact of the supplier's geographical distribution on supply. It is a distance sensitive factor, which is used to adjust the influence of distance on the supply weight. v The index of the supplier. n Indicates the total number of suppliers; Based on the supply weight of each supplier, the supply concentration factor is calculated to quantify the dominant position of a single supplier in the total supply. The calculation expression of the supply concentration factor is: in: is the supply concentration factor, is a weight-sensitive factor used to nonlinearly amplify the impact of high-weight suppliers. For suppliers k The supply frequency indicates the proportion of times the supplier supplies goods within the monitoring window; By dynamically updating the supply concentration reference value within the monitoring window, a time series is generated to capture the dynamic change trend of the dependency relationship. The time series expression of the dynamic supply concentration reference value is: in: For the current monitoring window The generated supply concentration reference value, The previous monitoring window The generated supply concentration reference value, For suppliers k In the current monitoring window The supply weight of For suppliers k In the current monitoring window The supply frequency, is the time smoothing factor, which is used to control the weight distribution of the historical supply concentration factor and the current supply concentration reference value. The range is .

[0011] Preferably, the specific steps for extracting the degree to which the historical delivery success rate of a product depends on a single supplier and dynamically quantifying the extracted features according to the monitoring window to generate a fulfillment dependency reference value are as follows: Extract the supplier's fulfillment contribution factor to measure each supplier's relative contribution to the fulfillment of product demand and identify the supplier's performance in meeting product order demand. The calculation expression of the fulfillment contribution factor is: in: For suppliers k In the monitoring window The contribution factor of the internal performance of the product reflects its direct impact on performance. For suppliers k In the monitoring window The number of orders successfully delivered within a certain period indicates its actual fulfillment capability. For products in the monitoring window The total order demand within is used to normalize the supplier's performance. For suppliers k The performance weight factor of Calculate the fulfillment concentration factor of a single supplier to quantify the concentration of fulfillment demand on the supplier and identify the potential risks of high dependence. The calculation expression of the fulfillment concentration factor is: in: For suppliers kThe performance concentration factor reflects its dominant position in the overall performance demand. For suppliers k The square of the performance contribution factor emphasizes the impact of large performance contribution on concentration. is the sum of the squares of all suppliers' performance contribution factors, indicating the overall performance distribution and used to normalize the concentration. For suppliers k In the monitoring window The cumulative supply volume within the period is used to reflect the impact of actual supply capacity on concentration. is the performance distribution bias correction term, where Adjust the parameter for sensitivity, controlling the rate of change of the exponential function; Combining the fulfillment concentration factor and the fulfillment contribution factor, we generate a fulfillment dependence reference value that reflects the dynamic degree of dependence of a product on a single supplier, and construct a time series for trend analysis. The time series expression of the fulfillment dependence reference value is: in: For monitoring window The fulfillment dependency reference value generated within the system comprehensively reflects the dynamic dependency relationship between the product and all suppliers. is the performance intensity adjustment item, which amplifies the impact of high-intensity performance suppliers on overall dependence, where is the adjustment coefficient, which controls the adjustment range. For monitoring window The total number of successful deliveries of the products in the contract, used to normalize the fulfillment contribution.

[0012] Preferably, the dependency index generated by the trained machine learning model when quantitatively evaluating the dependency relationship between products and suppliers is compared with a pre-set dependency index reference threshold, and intelligent analysis is performed to promptly identify high dependency risks. The specific steps are as follows: If the dependency index is greater than the dependency index reference threshold, it indicates that the dependency relationship between the product and the supplier is high. The system will generate an early warning message so that the purchaser can take countermeasures to alleviate the product's high dependency on a single supplier and reduce the risk of supply disruption. If the dependency index is less than or equal to the dependency index reference threshold, it will indicate that the dependency relationship between the product and the supplier is low, and no warning information will be generated through the system.

[0013] Preferably, after the countermeasures are implemented, the supply chain data of the products and suppliers is continuously obtained, and the changes in the real-time dependency relationships are analyzed during multiple subsequent monitoring windows. The specific steps are as follows: After the response measures are implemented, continue to obtain supply chain data on commodities and suppliers to analyze changes in dependencies over multiple subsequent monitoring windows; After obtaining the dependency index for multiple monitoring windows, a comprehensive analysis of the changing trends of the obtained dependency index is conducted to preliminarily assess whether the response measures have effectively alleviated the product's dependence on a single supplier. By calculating the dependency index change rate for each supplier, the degree of improvement or deterioration in the degree of dependence is quantified. The calculation expression for the dependency index change rate is: in: Indicates supplier k The rate of change of dependency index during multiple monitoring windows, T is the total number of monitoring windows obtained, indicating the time span of the evaluation, is a supplier k In the window The dependency index within It is the reference threshold of the dependency index, which is used to assess the risk of supplier dependency. is the time weight factor, which is used to give higher weight to the data in the recent time window, reflecting the impact of the current data on the evaluation; if , indicating that the dependency has been alleviated; if , it means that the dependency has not been alleviated.

[0014] Preferably, after completing the dependency index change analysis of a single supplier, further evaluate the dependency changes of all suppliers to measure the effectiveness of the overall response measures. By combining the importance weights of each supplier, calculate the effectiveness evaluation value of the overall response to provide guidance for the optimization strategy. The calculation expression of the effectiveness evaluation value is: in, It is the effectiveness evaluation value of the overall response measures, reflecting the comprehensive effect of changes in all supplier dependencies. is a supplier k Importance weight of Calculate the changes in all supplier dependencies by weighted average to obtain an evaluation of the overall response effect ; if , indicating that the dependency has been alleviated and the response measures are effective; if If it is positive, it means that the response measures are not effective and the supply chain strategy needs to be further optimized.

[0015] A supply chain risk analysis system based on artificial intelligence and big data, including a data collection and integration module, a feature engineering and quantification module, a machine learning assessment module, a risk assessment and early warning module, a response measure effectiveness verification module, and a long-term mechanism establishment and optimization module; Data collection and integration module, which obtains all data related to products and suppliers from internal enterprise systems and external sources, providing comprehensive data support for subsequent analysis and evaluation; The feature engineering and quantification module performs feature engineering based on comprehensive data to extract key features that reflect the dependency between products and suppliers. This module extracts core features that reflect the dependency between products and suppliers from big data, providing data dimension support for quantitative analysis. The extracted features are dynamically quantified according to the monitoring window to construct a time series of dependency relationships. The machine learning evaluation module uses the dynamically quantified features as input variables into the trained machine learning model and quantitatively evaluates the dependency between products and suppliers based on the training results. The risk assessment and early warning module determines whether there is a potential risk of high dependence between products and suppliers based on the assessment results and the set dependency risk threshold. It promptly identifies high dependency risks through intelligent analysis and provides early warning information to buyers so that they can take countermeasures and alleviate the high dependence of products on a single supplier. The countermeasure effectiveness verification module continues to obtain supply chain data of goods and suppliers after the countermeasures are implemented, and analyzes changes in real-time dependencies during multiple subsequent monitoring windows to verify the effectiveness of the countermeasures. It continuously monitors the real-time dependencies between goods and suppliers to ensure that dependency risks continue to decrease. The long-term mechanism establishment and optimization module summarizes and optimizes the effectiveness of response measures and establishes long-term mechanisms to prevent the recurrence of high-dependence risks.

[0016] In the above technical solution, the technical effects and advantages provided by the present invention are: This invention ensures comprehensive data support by integrating the company's internal systems and external data sources, providing a reliable basis for accurately evaluating the dependency relationship between goods and suppliers. Through the application of feature engineering and machine learning models, it can intelligently quantify the degree of dependence of goods on suppliers and monitor changes in dependency relationships in real time. This allows purchasers to take effective countermeasures before potential risks break out, reduce the risk of supply disruptions, and ensure the continuity of production lines. In the long run, it can not only optimize the effectiveness of countermeasures, but also establish a continuous risk management mechanism to prevent the recurrence of similar risks and improve the resilience and stability of the supply chain. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction to the drawings required for use in the embodiments will be given below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0018] Figure 1 This is a flow chart of a supply chain risk analysis method based on artificial intelligence big data in the present invention.

[0019] Figure 2 This is a module schematic diagram of a supply chain risk analysis system based on artificial intelligence big data in the present invention. DETAILED DESCRIPTION

[0020] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of this disclosure will be thorough and complete and will fully convey the concepts of the example embodiments to those skilled in the art.

[0021] The present invention provides Figure 1 The supply chain risk analysis method based on artificial intelligence big data shown in the figure includes the following steps: Obtain all data related to products and suppliers from internal enterprise systems (such as ERP, supplier management systems) and external sources (such as market reports, logistics tracking systems) to provide comprehensive data support for subsequent analysis and evaluation, ensuring the completeness and accuracy of the analysis of product and supplier relationships; Acquiring all product and supplier-related data from internal enterprise systems and external sources can be divided into the following steps: First, integrate internal enterprise data, including extracting key information such as procurement records, delivery history, inventory status, and supplier performance metrics from ERP (Enterprise Resource Planning) systems, supplier management systems (SRM), and warehouse and logistics management systems (WMS / TMS). Next, collect external data sources, including industry reports, market monitoring platforms, logistics tracking services, and publicly available supplier credit assessments and financial health data. Finally, through data cleansing and standardization, ensure the integrity and consistency of internal enterprise data and external data sources, laying a solid foundation for subsequent feature extraction and analysis.

[0022] All data related to products and suppliers includes, but is not limited to, supplier supply ratios, on-time delivery rates, procurement history, contract terms, supplier location, financial status, and logistics methods. Furthermore, upstream and downstream data (such as raw material sources and changes in end-user demand) must be integrated to fully reflect the dynamic nature of the supply chain. This step requires ensuring data accuracy, completeness, and real-time availability, laying the foundation for subsequent feature extraction and model analysis.

[0023] Based on comprehensive data, feature engineering is performed to extract key features that reflect the dependency between products and suppliers. Core features that reflect the dependency between products and suppliers are extracted from big data to provide data dimension support for quantitative analysis. The extracted features are dynamically quantified according to monitoring windows (such as weeks, months, or quarters) to construct a time series of dependency relationships. Key features that can reflect the dependency relationship between goods and suppliers are extracted, including the degree of concentration of the proportion of goods supplied by a single supplier in the total supply and the degree to which the historical delivery success rate of goods depends on a single supplier. The extracted features are dynamically quantified according to the monitoring window to generate supply concentration reference values ​​and fulfillment dependence reference values ​​respectively. The time series of the dependency relationship is constructed through the supply concentration reference values ​​and the fulfillment dependence reference values, and the dependency relationship between goods and suppliers is dynamically quantified.

[0024] When a single supplier accounts for a high proportion of total supply, it generally indicates a high degree of dependence on that supplier. This dependence signifies that the supplier holds a critical position in the supply chain, and its supply capacity and stability have a direct impact on the production and delivery of the product. If the supplier interrupts supply for any reason (such as insufficient production capacity, natural disasters, or operational issues), the company faces the risk of supply chain disruption, potentially leading to production stoppages, order delays, and even reduced market competitiveness. Therefore, a high proportion of supply from a single supplier not only reflects supply chain fragility but also suggests that companies need to evaluate the feasibility of alternative suppliers and plan contingency supply chains. Dynamic monitoring of this indicator helps companies proactively identify and mitigate supply chain risks.

[0025] The specific steps for extracting the concentration of the proportion of goods supplied by a single supplier in the total supply and dynamically quantifying the extracted features according to the monitoring window to generate a reference value for supply concentration are as follows: By calculating the supply weight of a single supplier in the total supply volume, we provide basic data for the supply concentration reference value and quantify the supplier's contribution to the supply of goods. The calculation expression for the supply weight is: in: For suppliers k The supply weight of For suppliersk The number of products offered, For suppliers k The supply distance is used to reflect the potential impact of the supplier's geographical distribution on supply. It is a distance sensitivity factor, which is used to adjust the influence of distance on the supply weight. , v The index of the supplier. n Indicates the total number of suppliers; Supply weight refers to the relative proportion of a supplier in the total supply volume, and is used to measure the importance of the supplier in supplying a specific commodity. Its setting requires comprehensive consideration of factors in multiple dimensions, such as the quantity of goods provided by the supplier (directly reflecting the contribution to supply volume), geographical distance (indicating the supplier's supply convenience and risk level), and supply frequency (assessing the stability of the supplier). Supply weight is usually calculated through a weighted formula to integrate these factors in a reasonable manner. For example, the supply weight can be set as the weighted ratio of the supply volume of a single supplier to the supply volume of all suppliers, and the weight can be adjusted based on geographical distance and other key characteristics. This setting method not only reflects the direct contribution of the supplier, but also incorporates additional risk and efficiency considerations, making it more scientific and practical in the analysis of supply concentration; By combining supply volume and supply distance, this step not only reflects the direct contribution of suppliers but also incorporates geographical distribution differences into the weight assessment. If the distance sensitivity factor increases, the supply weight of distant suppliers will decrease significantly, further highlighting the importance of nearby suppliers.

[0026] Based on the supply weight of each supplier, the supply concentration factor is calculated to quantify the dominant position of a single supplier in the total supply. The calculation expression of the supply concentration factor is: in: is the supply concentration factor, is a weight-sensitive factor used to nonlinearly amplify the impact of high-weight suppliers, usually , For suppliers k The supply frequency indicates the proportion of times the supplier supplies goods within the monitoring window; By introducing supply frequency, we can capture the supplier's activeness during the monitoring period. If a supplier has both a high supply weight and high supply frequency, its influence in the supply concentration factor will be significantly amplified, thus more accurately reflecting the risk of high dependence.

[0027] By dynamically updating the supply concentration reference value within the monitoring window, a time series is generated to capture the dynamic change trend of the dependency relationship. The time series expression of the dynamic supply concentration reference value is: in: For the current monitoring window The generated supply concentration reference value, The previous monitoring window The generated supply concentration reference value, For suppliers k In the current monitoring window The supply weight of For suppliers k In the current monitoring window The supply frequency, It is a time smoothing factor used to control the weight distribution of historical supply concentration factors and current supply concentration reference values. The range is ; By dynamically combining historical and current data, the formula can reflect short-term fluctuations and capture long-term trends, providing a comprehensive time dimension perspective for supply chain risk assessment. The introduction of can adjust the sensitivity of the supply concentration reference value to adapt to the needs of different monitoring cycles.

[0028] The Supply Concentration Reference Value (SCV) dynamically quantifies the concentration of a single supplier's supply relative to the total supply of a particular commodity within a monitoring window. A higher SCV indicates a higher reliance on a particular supplier. This is because the SCV quantifies a supplier's dominance within the total supply by comprehensively assessing its supply weight (e.g., supply ratio, frequency, etc.). A high SCV indicates a significant single supplier's weight and a strong supply chain dependency. Conversely, a low SCV indicates a more dispersed supply or less reliance, making the supply chain more resilient to risk. Therefore, the SCV serves as an important indicator for measuring the strength of the dependency relationship between a commodity and its supplier, helping to identify potential supply chain risks.

[0029] When a product's historical delivery success rate is highly dependent on a single supplier, it indicates a high level of reliance on that supplier. This is because the delivery success rate reflects the supplier's stability and reliability in meeting product demand. If a single supplier successfully delivers the majority of orders for a product, and other suppliers have low participation or are unable to meet the same quality and timeliness requirements, the product's supply chain will face significant disruption risks should that supplier encounter supply issues (such as insufficient production capacity, logistics delays, or natural disasters). This high level of reliance not only reduces supply chain resilience but also amplifies the potential impact of a single point of failure on business operations. Therefore, it is a crucial metric for assessing product-supplier dependency.

[0030] The specific steps for extracting the degree to which the historical delivery success rate of a product depends on a single supplier and dynamically quantifying the extracted features according to the monitoring window to generate a fulfillment dependency reference value are as follows: Extract the supplier's fulfillment contribution factor to measure each supplier's relative contribution to the fulfillment of product demand, identify the supplier's performance in meeting product order demand, and provide basic data for subsequent calculations. The calculation expression of the fulfillment contribution factor is: in: For suppliers k In the monitoring window The contribution factor of the internal performance of the product reflects its direct impact on performance. For suppliers k In the monitoring window The number of orders successfully delivered within a certain period indicates its actual fulfillment capability. For products in the monitoring window The total order demand within is used to normalize the supplier's performance. For suppliers k The fulfillment weight factor takes into account the comprehensive score of multi-dimensional fulfillment capabilities such as order delivery time, delivery quality, and ability to meet special needs; This step can quantify the supplier's performance contribution and take into account the differences in their performance capabilities, providing accurate input for subsequent concentration calculations.

[0031] Calculate the fulfillment concentration factor of a single supplier to quantify the concentration of fulfillment demand on the supplier and identify the potential risks of high dependence. The calculation expression of the fulfillment concentration factor is: in: For suppliers k The performance concentration factor reflects its dominant position in the overall performance demand. For suppliers k The square of the performance contribution factor emphasizes the impact of large performance contribution on concentration. is the sum of the squares of all suppliers' performance contribution factors, indicating the overall performance distribution and used to normalize the concentration. For suppliers k In the monitoring window The cumulative supply volume (ratio) within the period is used to reflect the impact of actual supply capacity on concentration. is the performance distribution bias correction term, which is used to capture the enhanced effect of performance dependence when the supply volume is large, where Adjust the parameter for sensitivity, controlling the rate of change of the exponential function; This step can identify the concentration of single suppliers in fulfillment and consider the combined impact of supply volume and fulfillment performance, highlighting the risks of high dependence.

[0032] Combining the fulfillment concentration factor and the fulfillment contribution factor, we generate a fulfillment dependence reference value that reflects the dynamic degree of dependence of a product on a single supplier, and construct a time series for trend analysis. The time series expression of the fulfillment dependence reference value is: in: For monitoring window The fulfillment dependency reference value generated within the system comprehensively reflects the dynamic dependency relationship between the product and all suppliers. is the performance intensity adjustment item, which amplifies the impact of high-intensity performance suppliers on overall dependence, where is the adjustment coefficient, which controls the adjustment range. For monitoring window The total number of successful deliveries of goods within a period is used to normalize fulfillment contributions and ensure that results across different time windows are comparable; This step can dynamically quantify the overall dependence of goods on suppliers, and construct a time series based on the performance dependence reference values ​​of multiple monitoring windows to capture the trend changes in the dependence relationship between goods and suppliers, providing a basis for subsequent risk prediction and management.

[0033] The fulfillment dependency reference value shows that the degree to which a product's historical delivery success rate depends on a single supplier is dynamically quantified within the monitoring window. A higher performance value for the generated fulfillment dependency reference value indicates a higher level of dependency on a single supplier, while a lower performance value indicates a lower level of dependency. The fulfillment dependency reference value quantifies a product's dependency on a single supplier within the dynamic monitoring window by comprehensively analyzing multiple characteristics, including the product's historical delivery success rate, the supplier's fulfillment contribution factor, and supply concentration. A higher performance value for the fulfillment dependency reference value indicates a single supplier holds a dominant position in the product's supply and fulfillment, with the product highly dependent on the stability and continuity of its supply. Conversely, a lower performance dependency reference value indicates a more dispersed supply chain and less reliance on a single supplier, resulting in greater supply chain resilience and risk mitigation. Therefore, the fulfillment dependency reference value is an important quantitative indicator for measuring the strength of the dependency relationship between a product and its supplier.

[0034] The dynamically quantified features are used as input variables into a trained machine learning model (such as a random forest, support vector machine, or deep learning model). The dependencies between products and suppliers are quantitatively evaluated based on the training results. The supply concentration reference value and fulfillment dependence reference value generated by dynamically quantifying the concentration degree of the proportion of goods supplied by a single supplier in the total supply and the degree to which the historical delivery success rate of goods depends on a single supplier are used as input variables and input into the trained machine learning model to generate a dependency index. The dependency relationship between goods and suppliers is quantitatively evaluated based on the dependency index output by the machine learning model.

[0035] A trained machine learning model is an algorithmic model that, after data input, feature processing, model training, and optimization, is capable of quantitatively assessing the dependencies between products and suppliers. It uses dynamically generated supply concentration reference values ​​and fulfillment dependency reference values ​​as input variables. By learning dependency patterns from historical data, it outputs a dependency index that reflects the degree of dependency between products and suppliers. The model training process involves analyzing extensive historical data, including characteristic data such as supply concentration, fulfillment stability, and supply fluctuations between different products and suppliers, as well as the correlation between these characteristics and actual supply chain performance (such as supply disruptions and fulfillment anomalies). During training, an optimization algorithm adjusts model parameters to effectively capture the nonlinear and complex relationships between input variables and dependencies. Ultimately, the trained model exhibits high predictive and generalization capabilities, enabling it to accurately assess the degree of dependency under new inputs of supply concentration reference values ​​and fulfillment dependency reference values.

[0036] Specifically, trained models may include various machine learning methods, such as regression analysis, support vector machines (SVMs), decision trees, random forests, or deep learning models (such as neural networks). Model selection depends on the characteristics and complexity of the data. For example, for time series data, deep learning models based on recurrent neural networks (RNNs) or long short-term memory networks (LSTMs) can be used. These models are capable of handling dynamically changing input features, capturing fluctuations in supply concentration and fulfillment dependency over time, and generating a dependency index as output. During training, the model repeatedly optimizes a loss function (such as mean squared error or cross-entropy loss) to bring the predicted dependency index closer to the actual level of dependency. The trained machine learning model enables a precise quantitative assessment of the dependency relationship between products and suppliers. Ultimately, the model's output not only reflects the current dependency situation but also predicts future trends, helping companies identify potential supply chain risks and develop effective response strategies.

[0037] The machine learning model is not specifically limited here, and can realize the monitoring window Generated supply concentration reference value and monitoring window Generated fulfillment dependency reference value Conduct comprehensive analysis to generate a dependency index In order to realize the technical solution of the present invention, the present invention provides a specific implementation method; the dependency index The resulting calculation formula is: Where, Supply concentration reference values and performance-dependent reference values The preset scaling factor of Both are greater than 0.

[0038] It can be seen from the dependency index that the concentration characteristics of the proportion of the extracted goods supplied by a single supplier in the total supply are dynamically quantified according to the monitoring window. The larger the performance value of the generated supply concentration reference value, the larger the performance value of the generated fulfillment dependence reference value is. In other words, the larger the performance value of the dependency index generated when the dependency relationship between the product and the supplier is quantitatively evaluated by the trained machine learning model, the higher the degree of dependence of the product on the supplier, and vice versa.

[0039] Machine learning models can capture nonlinear relationships between features and combine historical trends to predict future dependency trends. This process can uncover hidden risks in dependency relationships and identify products with potential problems due to high supplier dependence.

[0040] Based on the assessment results and the set dependency risk threshold, it determines whether there is a potential risk of high dependency between the product and its supplier. Through intelligent analysis, it promptly identifies high dependency risks and provides early warning information to buyers so that they can take countermeasures, thereby alleviating the product's high dependency on a single supplier and reducing the risk of supply disruptions. The dependency index generated by the trained machine learning model when quantitatively evaluating the dependency relationship between products and suppliers is compared with the pre-set dependency index reference threshold. Intelligent analysis can promptly identify high dependency risks. The specific steps are as follows: If the dependency index is greater than the dependency index reference threshold, it indicates that the dependency relationship between the product and the supplier is high. The system will generate an early warning message so that the purchaser can take countermeasures to alleviate the product's high dependency on a single supplier and reduce the risk of supply disruption. If the dependency index is less than or equal to the dependency index reference threshold, it indicates that the dependency relationship between the product and the supplier is low, and no early warning information will be generated through the system. At this time, the risk of supply interruption of the product is low.

[0041] If a product's reliance on a single supplier exceeds a threshold, or if dependency trend forecasts indicate a potential increase in future risk, the system will immediately issue an alert. This alert will include a detailed dependency assessment report and targeted response recommendations, such as developing new suppliers, increasing inventory, or adjusting procurement strategies.

[0042] After implementing countermeasures, continue to obtain supply chain data on products and suppliers, and analyze changes in real-time dependencies during multiple subsequent monitoring windows to verify the effectiveness of countermeasures. Continuously monitor the real-time dependencies between products and suppliers to ensure that dependency risks continue to decrease. After implementing countermeasures, continue to obtain supply chain data for products and suppliers, and analyze changes in real-time dependencies over multiple subsequent monitoring windows. The specific steps are as follows: After the response measures are implemented, continue to obtain supply chain data on commodities and suppliers to analyze changes in dependencies over multiple subsequent monitoring windows; After obtaining the dependency index for multiple monitoring windows, a comprehensive analysis of the changing trends of the obtained dependency index is conducted to preliminarily assess whether the response measures have effectively alleviated the product's dependence on a single supplier. By calculating the dependency index change rate for each supplier, the degree of improvement or deterioration in the degree of dependence is quantified. The calculation expression for the dependency index change rate is: in: Indicates supplier k The rate of change of dependency index during multiple monitoring windows, T is the total number of monitoring windows obtained, indicating the time span of the evaluation, is a supplier k In the window The dependency index within It is the reference threshold of the dependency index, which is used to assess the supplier dependency risk. is the time weight factor, which is used to give higher weight to the data in the recent time window, reflecting the impact of the current data on the evaluation; By calculating the rate of change of the dependence index , we can quantitatively evaluate the degree of improvement in the dependency relationship between products and suppliers. , indicating that the dependency has been alleviated; if , it means that the dependency has not been effectively alleviated and more countermeasures may be needed. This process helps companies identify whether they have successfully reduced their dependence on a single supplier for their products.

[0043] After completing the dependency index change analysis for a single supplier, we further evaluate the dependency changes for all suppliers to measure the effectiveness of the overall response measures. By combining the importance weights of each supplier, we calculate the overall response effectiveness evaluation value, thereby providing guidance for the optimization strategy. The calculation expression for the effectiveness evaluation value is: in, It is the effectiveness evaluation value of the overall response measures, reflecting the comprehensive effect of changes in all supplier dependencies. is a supplier k Importance weights are usually set based on the supplier's strategic position or criticality to the supply of goods; Calculate the changes in all supplier dependencies by weighted average to obtain an evaluation of the overall response effect .if , indicating that the dependency has been effectively alleviated and the response measures are effective; if If it is positive, it means that the response measures are not effective and the supply chain strategy may need to be further optimized.

[0044] The goal of this phase is to ensure that the response measures not only effectively reduce dependency risks but also provide a basis for decision-making in future supply chain management.

[0045] Summarize and optimize the effectiveness of response measures and establish long-term mechanisms to prevent the recurrence of high-dependency risks; After analyzing and evaluating the effectiveness of countermeasures in improving product-supplier dependencies, successful experiences and shortcomings will be summarized, current strategies will be optimized and adjusted, and a systematic and sustainable management mechanism will be developed to ensure that the risks of high dependency can be effectively avoided in future supply chain management. Specifically, this includes timely identifying potential dependency issues through data monitoring and dynamic analysis, establishing alternative supplier networks, adjusting procurement ratios, optimizing contract terms, and other means to reduce excessive reliance on a single supplier at the source. At the same time, through standardized risk management processes and automated early warning systems, supply chain dependencies will be continuously monitored, forming a long-term management mechanism that can be dynamically adjusted and optimized, thereby enhancing the resilience and risk resistance of the supply chain and preventing the recurrence of similar risk issues.

[0046] The aforementioned AI-powered big data-based supply chain risk analysis method enables dynamic monitoring and intelligent assessment of the dependencies between products and suppliers, effectively identifying and mitigating high-dependency risks within the supply chain. By integrating internal enterprise systems with external data sources, this method ensures comprehensive data support, providing a reliable foundation for accurately assessing the dependencies between products and suppliers. Through the application of feature engineering and machine learning models, it is possible to intelligently quantify the degree of dependency of products on suppliers and monitor changes in these dependencies in real time. This allows buyers to take effective countermeasures before potential risks arise, reducing the risk of supply disruptions and ensuring production line continuity. Over the long term, this method not only optimizes the effectiveness of countermeasures but also establishes a continuous risk management mechanism to prevent the recurrence of similar risks and enhance the resilience and stability of the supply chain.

[0047] The present invention provides Figure 2 The supply chain risk analysis system shown here, based on artificial intelligence and big data, includes a data collection and integration module, a feature engineering and quantification module, a machine learning assessment module, a risk assessment and early warning module, a response measure effectiveness verification module, and a long-term mechanism establishment and optimization module. Data collection and integration module, which obtains all data related to products and suppliers from internal enterprise systems and external sources, providing comprehensive data support for subsequent analysis and evaluation; The feature engineering and quantification module performs feature engineering based on comprehensive data to extract key features that reflect the dependency between products and suppliers. This module extracts core features that reflect the dependency between products and suppliers from big data, providing data dimension support for quantitative analysis. The extracted features are dynamically quantified according to the monitoring window to construct a time series of dependency relationships. The machine learning evaluation module uses the dynamically quantified features as input variables into the trained machine learning model and quantitatively evaluates the dependency between products and suppliers based on the training results. The risk assessment and early warning module determines whether there is a potential risk of high dependence between products and suppliers based on the assessment results and the set dependency risk threshold. It promptly identifies high dependency risks through intelligent analysis and provides early warning information to buyers so that they can take countermeasures and alleviate the high dependence of products on a single supplier. The countermeasure effectiveness verification module continues to obtain supply chain data of goods and suppliers after the countermeasures are implemented, and analyzes changes in real-time dependencies during multiple subsequent monitoring windows to verify the effectiveness of the countermeasures. It continuously monitors the real-time dependencies between goods and suppliers to ensure that dependency risks continue to decrease. The long-term mechanism establishment and optimization module summarizes and optimizes the effectiveness of response measures and establishes long-term mechanisms to prevent the recurrence of high-dependency risks; The supply chain risk analysis method based on artificial intelligence big data provided in an embodiment of the present invention is implemented through the above-mentioned supply chain risk analysis system based on artificial intelligence big data. The specific methods and processes of the supply chain risk analysis system based on artificial intelligence big data are detailed in the embodiment of the above-mentioned supply chain risk analysis method based on artificial intelligence big data, and will not be repeated here.

[0048] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0049] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present application should be included within the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

[0050] The above description is merely illustrative of certain exemplary embodiments of the present invention. It goes without saying that those skilled in the art will be able to modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims.

Claims

1. A supply chain risk analysis method based on artificial intelligence big data, characterized in that: The following steps are involved: Acquire all data related to products and suppliers from internal enterprise systems and external sources to provide comprehensive data support for subsequent analysis and evaluation; Based on comprehensive data, feature engineering is performed to extract key features that can reflect the dependency relationship between products and suppliers. The core features that can reflect the dependency relationship between products and suppliers are extracted from big data to provide data dimension support for quantitative analysis. The extracted features are dynamically quantified according to the monitoring window to construct a time series of dependency relationships. The dynamically quantified features are used as input variables and input into the trained machine learning model. The dependency relationship between products and suppliers is quantitatively evaluated based on the training results. Based on the assessment results and the set dependency risk threshold, determine whether there is a potential risk of high dependency between the product and the supplier. Through intelligent analysis, timely identify the high dependency risk and provide early warning information to the purchaser so that he can take countermeasures to alleviate the high dependency of the product on a single supplier. After the countermeasures are implemented, continue to obtain supply chain data of goods and suppliers, and analyze the changes in real-time dependencies during multiple subsequent monitoring windows to verify the effectiveness of the countermeasures, and continuously monitor the real-time dependencies between goods and suppliers to ensure that dependency risks continue to decrease; Summarize and optimize the effectiveness of response measures and establish long-term mechanisms to prevent the recurrence of high-dependency risks.

2. The supply chain risk analysis method based on artificial intelligence big data according to claim 1 is characterized in that: Obtain all data related to products and suppliers from internal enterprise systems and external sources. The specific steps are as follows: First, integrate internal enterprise data, including extracting key information from ERP, supplier management systems, and inventory and logistics management systems; Next, collect external data sources, including industry reports, market monitoring platforms, logistics tracking services, and publicly available supplier credit assessment and financial health data; Finally, through data cleaning and standardization, the integrity and consistency of internal enterprise data and external data sources are ensured, laying the foundation for subsequent feature extraction and analysis.

3. The supply chain risk analysis method based on artificial intelligence big data according to claim 1 is characterized in that: The key features that can reflect the dependency relationship between goods and suppliers are extracted, among which the extracted features include the concentration degree of the proportion of goods supplied by a single supplier in the total supply and the degree to which the historical delivery success rate of goods depends on a single supplier. The extracted features are dynamically quantified according to the monitoring window to generate supply concentration reference values ​​and fulfillment dependence reference values ​​respectively. The time series of the dependency relationship is constructed through the supply concentration reference value and the fulfillment dependence reference value, and the dependency relationship between goods and suppliers is dynamically quantified.

4. The supply chain risk analysis method based on artificial intelligence big data according to claim 3 is characterized in that: The supply concentration reference value and fulfillment dependence reference value generated by dynamically quantifying the proportion of goods supplied by a single supplier in the total supply and the degree to which the historical delivery success rate of goods depends on a single supplier are used as input variables and input into the trained machine learning model to generate a dependency index. The dependency relationship between goods and suppliers is quantitatively evaluated based on the dependency index output by the machine learning model.

5. The supply chain risk analysis method based on artificial intelligence big data according to claim 3 is characterized in that: The specific steps for extracting the concentration of the proportion of goods supplied by a single supplier in the total supply and dynamically quantifying the extracted features according to the monitoring window to generate a reference value for supply concentration are as follows: By calculating the supply weight of a single supplier in the total supply, basic data is provided for the supply concentration reference value, and the contribution of the supplier to the supply of goods is quantified. The calculation expression of the supply weight is: in: For Suppliers k The supply weight, For Suppliers k The number of items offered, For Suppliers k The supply distance is used to reflect the potential impact of the supplier's geographical distribution on supply. is the distance sensitivity factor, which is used to adjust the impact of distance on the supply weight. v represents the index of the supplier, n represents the total number of suppliers; Based on the supply weight of each supplier, the supply concentration factor is calculated to quantify the dominant position of a single supplier in the total supply. The calculation expression of the supply concentration factor is: in: is the supply concentration factor, is a weight-sensitive factor used to nonlinearly amplify the impact of high-weight suppliers. For Suppliers k The supply frequency indicates the proportion of the supplier's supply times within the monitoring window; By dynamically updating the supply concentration reference value in the monitoring window, a time series is generated to capture the dynamic change trend of the dependency relationship. The time series expression of the dynamic supply concentration reference value is: in: For the current monitoring window The generated supply concentration reference value, The previous monitoring window The generated supply concentration reference value, For Suppliers k In the current monitoring window The supply weight, For Suppliers k In the current monitoring window The supply frequency, is the time smoothing factor, which is used to control the weight distribution of the historical supply concentration factor and the current supply concentration reference value. The range is .

6. The supply chain risk analysis method based on artificial intelligence big data according to claim 5 is characterized in that: The specific steps for extracting the degree to which the historical delivery success rate of goods depends on a single supplier and dynamically quantifying the extracted features according to the monitoring window to generate the fulfillment dependency reference value are as follows: The supplier's fulfillment contribution factor is extracted to measure the relative contribution of each supplier in the fulfillment of commodity demand and to identify the supplier's performance in meeting the demand for commodity orders. The calculation expression of the fulfillment contribution factor is: in: For Suppliers k In the monitoring window The contribution factor of the internal commodity to the fulfillment of the contract reflects its direct impact on the fulfillment of the contract. For Suppliers k In the monitoring window The number of orders successfully delivered within a period of time indicates its actual fulfillment capability. For products in the monitoring window The total order demand within is used to normalize the supplier's performance. For Suppliers k The performance weight factor of Calculate the fulfillment concentration factor of a single supplier to quantify the concentration of the fulfillment demand of the product on the supplier and clarify the potential risks of high dependence. The calculation expression of the fulfillment concentration factor is: in: For Suppliers k The performance concentration factor reflects its dominant position in the overall performance demand. For Suppliers k The square of the performance contribution factor emphasizes the impact of large performance contributions on concentration. is the sum of the squares of all suppliers' fulfillment contribution factors, indicating the overall fulfillment distribution and used to normalize the concentration. For Suppliers k In the monitoring window The cumulative supply volume within the period is used to reflect the impact of actual supply capacity on concentration. is the performance distribution bias correction term, where Adjust the parameter for sensitivity, controlling the rate of change of the exponential function; Combining the fulfillment concentration factor and the fulfillment contribution factor, a fulfillment dependence reference value that reflects the dynamic dependence of the commodity on a single supplier is generated, and a time series is constructed for trend analysis. The time series expression of the fulfillment dependence reference value is: in: For monitoring window The fulfillment dependency reference value generated within the system comprehensively reflects the dynamic dependency relationship between the product and all suppliers. is the performance intensity adjustment item, which amplifies the impact of high-intensity performance suppliers on overall dependence, where is the adjustment coefficient, controlling the adjustment range. For monitoring window The total number of successful deliveries of the products in the contract, used to normalize the fulfillment contribution.

7. The supply chain risk analysis method based on artificial intelligence big data according to claim 4 is characterized in that: The dependency index generated by the trained machine learning model when quantitatively evaluating the dependency relationship between products and suppliers is compared with the pre-set dependency index reference threshold, and intelligent analysis is performed to timely identify high dependency risks. The specific steps are as follows: If the dependency index is greater than the dependency index reference threshold, it indicates that the dependency relationship between the product and the supplier is high, and the system generates early warning information so that the purchaser can take countermeasures to alleviate the high dependency of the product on a single supplier and reduce the risk of supply interruption; If the dependency index is less than or equal to the dependency index reference threshold, it will indicate that the dependency relationship between the product and the supplier is low, and no warning information will be generated through the system.

8. The supply chain risk analysis method based on artificial intelligence big data according to claim 7 is characterized in that: After the countermeasures are implemented, continue to obtain supply chain data of goods and suppliers, and analyze the changes in real-time dependencies during multiple subsequent monitoring windows. The specific steps are as follows: After the countermeasures are implemented, continue to obtain supply chain data for commodities and suppliers to analyze changes in dependencies over multiple subsequent monitoring windows; After obtaining the dependency index of multiple monitoring windows, a comprehensive analysis of the change trend of the obtained dependency index is conducted to preliminarily evaluate whether the response measures have effectively alleviated the dependence of the product on a single supplier. By calculating the dependency index change rate of each supplier, the improvement or deterioration of the degree of dependence is quantified. The calculation expression of the dependency index change rate is: in: Indicates supplier k The rate of change of the dependency index during multiple monitoring windows, T is the total number of monitoring windows obtained, indicating the time span of the evaluation, Is a supplier k In the window The dependency index within It is the reference threshold of the dependency index, which is used as a benchmark for assessing supplier dependency risk. is the time weight factor, which is used to give higher weight to the data in the recent time window, reflecting the impact of the current data on the evaluation; if , indicating that the dependency has been alleviated; if , then the dependency has not been alleviated.

9. The supply chain risk analysis method based on artificial intelligence big data according to claim 8 is characterized in that: After completing the dependency index change analysis of a single supplier, further evaluate the dependency changes of all suppliers to measure the effectiveness of the overall response measures. By combining the importance weights of each supplier, calculate the overall response effect evaluation value to provide guidance for the optimization strategy. The calculation expression of the effect evaluation value is: in, It is the effectiveness evaluation value of the overall response measures, reflecting the comprehensive effect of changes in all supplier dependencies. Is a supplier k The importance weight of Calculate the changes in all supplier dependencies by weighted average to obtain an evaluation of the overall response effect ; if , indicating that the dependency has been alleviated and the response measures are effective; if If it is positive, it means that the effect of the countermeasures is not obvious and the supply chain strategy needs to be further optimized.

10. A supply chain risk analysis system based on artificial intelligence big data, used to implement the above-mentioned supply chain risk analysis method based on artificial intelligence big data, characterized in that: It includes data collection and integration module, feature engineering and quantification module, machine learning assessment module, risk assessment and early warning module, response measures effectiveness verification module and long-term mechanism establishment and optimization module; Data collection and integration module, which obtains all data related to goods and suppliers from internal enterprise systems and external sources, providing comprehensive data support for subsequent analysis and evaluation; The feature engineering and quantification module performs feature engineering based on comprehensive data to extract key features that can reflect the dependency relationship between products and suppliers. It extracts core features that can reflect the dependency relationship between products and suppliers from big data, provides data dimension support for quantitative analysis, dynamically quantifies the extracted features according to the monitoring window, and constructs a time series of dependency relationships. The machine learning evaluation module uses the dynamically quantified features as input variables and inputs them into the trained machine learning model, and performs quantitative evaluation of the dependency relationship between products and suppliers based on the training results; The risk assessment and early warning module determines whether there is a potential risk of high dependence between the product and the supplier based on the assessment results and the set dependency risk threshold. It promptly identifies the high dependence risk through intelligent analysis and provides early warning information to the purchaser so that he can take countermeasures to alleviate the high dependence of the product on a single supplier. The countermeasures effectiveness verification module continues to obtain supply chain data of goods and suppliers after the countermeasures are implemented, and analyzes the changes in real-time dependencies during multiple subsequent monitoring windows to verify the effectiveness of the countermeasures, continuously monitor the real-time dependencies between goods and suppliers, and ensure that the dependency risks continue to decrease; The long-term mechanism establishment and optimization module summarizes and optimizes the effects of response measures and establishes long-term mechanisms to prevent the recurrence of high-dependency risks.

Images