An automobile parts supply chain management method and system

Abstract

This invention discloses a method and system for automotive parts supply chain management, relating to the field of supply chain management technology. The method includes: acquiring operational status data of each node in the supply chain; performing status perception and anomaly identification on the operational status of each node; generating a node status assessment, wherein the operational status data includes actual consumption data; performing risk correction on the actual consumption data based on the node status assessment to generate virtual consumption; performing short-cycle forecasting of parts demand based on the node status assessment to generate forecasted demand data, and outputting parts pull decision instructions in conjunction with the virtual consumption; coordinating and scheduling execution of each node according to the pull decision instructions; providing feedback on the results of the coordinated scheduling execution process, and dynamically correcting the virtual consumption and pull decision instructions based on the feedback results. This application solves the problems of inaccurate supply chain forecasting and scheduling lag by mapping node risk perception with virtual consumption.

Description

Technical Field

[0001] This invention relates to the field of supply chain management technology, and more specifically, to a method and system for managing the automotive parts supply chain. Background Technology

[0002] Automotive parts supply chain management involves multiple stages, including automobile production, parts manufacturing, logistics, and warehousing management, and is a crucial foundation for ensuring the smooth operation of vehicle production. With the rapid development of the automotive industry and the continuous increase in product models, the types and quantities of automotive parts are becoming increasingly diverse and complex, leading to ever-increasing demands for correlation and collaboration among various parts supply nodes. In the automobile manufacturing process, just-in-time parts supply, inventory control, and production planning coordination directly impact overall vehicle production efficiency, cost control, and customer delivery cycles. Therefore, establishing a management method capable of comprehensively sensing the operational status of supply chain nodes, predicting parts demand, and dynamically adjusting replenishment plans has become a key technological direction at this stage.

[0003] For example, the invention patent with publication number CN118052590A discloses an AI-based automotive parts supply chain management system, which includes: a supply chain parameter acquisition module, used to collect the working parameters of all manufacturers in the parts market for the current model of automobile at preset time intervals m within a preset time period L; and a graph network data generation module, used to generate graph network data based on the working parameters of all manufacturers in the supply chain. This invention generates graph network data based on the historical working parameters of all manufacturers in the current automotive parts market, and analyzes the historical change information of each manufacturer recorded in the graph network data to predict the future of the current supply chain and feasible supply chains. It then uses an ant colony algorithm to find the supply chain with the highest predicted future score. Compared to common supply chain construction methods, this invention incorporates the historical working parameters of each manufacturer to predict future trends.

[0004] For example, the invention patent with publication number CN118917628A discloses a supply chain management system and method based on operational data. Through a data acquisition module, it collects multi-dimensional data such as market demand, production efficiency, and logistics status. This data is then used to generate predictive data through a multi-task learning model and quantum Bayesian inference. Based on this predictive data, a distributed collaborative decision-making module utilizes a multi-agent system and a self-organizing criticality algorithm to optimize resource allocation across all nodes of the supply chain and generate a final scheduling scheme. Through resource pooling and virtualization technologies, the system achieves unified allocation of resources across regions and dynamically self-corrects through a feedback mechanism. This improves the scheduling efficiency of the supply chain, reduces the risk of inventory backlog, and enhances the ability to cope with market fluctuations.

[0005] The above-disclosed technical solutions have at least the following technical problems:

[0006] Traditional technologies rely on fixed inventory strategies or pull models based on JIT / Kanban, which cannot quantitatively analyze the real-time operating status of each node in the supply chain, lack mechanisms for identifying anomalies or potential risks, and usually treat component consumption as fixed or historical averages, failing to fully consider the impact of node operating status on actual consumption, resulting in large deviations between inventory plans and actual demand.

[0007] To address the above problems, this invention proposes a solution. Summary of the Invention

[0008] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide an automotive parts supply chain management method and system that solves the problems of inaccurate supply chain forecasting and scheduling lag by using node risk perception and virtual consumption mapping.

[0009] To achieve the above objectives, the present invention provides the following technical solution:

[0010] A method for managing an automotive parts supply chain includes: acquiring operational status data of each node in the supply chain; performing status perception and anomaly identification on the operational status of each node; generating a node status assessment, wherein the operational status data includes actual consumption data; performing risk correction on the actual consumption data based on the node status assessment to generate virtual consumption quantities; performing short-cycle forecasting of parts demand based on the node status assessment to generate forecasted demand data, and outputting parts pull decision instructions in conjunction with the virtual consumption quantities; coordinating and scheduling execution of each node according to the pull decision instructions; providing feedback on the results of the coordinated scheduling and execution process, and dynamically correcting the virtual consumption quantities and pull decision instructions based on the feedback results.

[0011] In a preferred embodiment, the step of acquiring the operational status data of each node in the supply chain, performing status perception and anomaly identification on the operational status of each node, and generating a node status assessment is as follows: acquiring the operational status data of each node in the automotive parts supply chain, and preprocessing the operational status data to form a node operational status dataset; extracting the operational status features of the nodes based on the node operational status dataset, assigning weight coefficients to each feature, and generating a node comprehensive status score through a weighted algorithm; comparing the node comprehensive status score with the corresponding multi-level thresholds to determine the node status, and recording the key feature items and their values ​​that cause the status change; and dynamically correcting the actual consumption based on the node status assessment results and corresponding feature items to generate a virtual consumption.

[0012] In a preferred embodiment, the step of extracting node operational status features from the node operational status dataset and assigning weight coefficients to each feature, and generating a comprehensive node status score through a weighted algorithm, is as follows: Based on historical operational records, operational status data of each node within a continuous stable operational cycle is obtained, and this operational status data within the specified stable operational cycle is used as the normal operation sample dataset; statistical analysis is performed on the operational status features in the normal operation sample dataset, outputting the corresponding mean range and fluctuation range, which are used as the normal threshold range for each status feature item; abnormal event data of nodes during historical operation is obtained, and node operational status data before and after the occurrence of the abnormal event is extracted to form an abnormal operation sample dataset; statistical analysis is performed on the operational status features in the abnormal operation sample dataset to determine... The value range of status feature items under abnormal operation conditions is divided into warning threshold range and abnormal threshold range, according to the specified value range exceeding the normal threshold range. Based on the normal operation sample dataset and the abnormal operation sample dataset, the value distribution characteristics of each operation status feature under stable operation and abnormal operation conditions are statistically analyzed, and the frequency and magnitude of each operation status feature exceeding the normal threshold range in the abnormal operation sample dataset are output. Based on the normal threshold range, warning threshold range, and abnormal threshold range, corresponding weight coefficients are configured for different operation status feature items according to the influence intensity of each operation status feature on the node operation stability, and a node comprehensive status score is generated by weighted calculation based on the operation status feature value and the weight coefficient. The node comprehensive status score is numerically compared with the preset multi-level status thresholds to extract key feature items.

[0013] In a preferred embodiment, the step of performing risk correction on actual consumption data based on node status assessment to generate virtual consumption is as follows: Based on the node status assessment results, the actual consumption data of the corresponding node in the current statistical period is obtained, and the actual consumption data is used as the baseline consumption input for the node in the current operating state; based on the node status assessment results, the node's comprehensive status score and corresponding key feature items are extracted; based on the positional relationship of the node's comprehensive status score relative to the normal state threshold interval, the warning state threshold interval, and the abnormal state threshold interval, the node's current operating state is mapped to a status risk value; based on the time series processing mechanism, the status risk value of the node in multiple consecutive statistical periods is smoothed; based on the status risk value and its changing trend, a risk correction coefficient is constructed, and the risk correction coefficient is enhanced based on the identified key feature items and feature values ​​to obtain a risk gain correction coefficient; based on the risk gain correction coefficient, the actual consumption data is risk-mapped and corrected to generate virtual consumption.

[0014] In a preferred embodiment, the step of constructing a risk correction coefficient based on the state risk value and its changing trend, and enhancing the risk correction coefficient based on the identified key features and feature values ​​to obtain a risk gain correction coefficient, is as follows: The continuous state risk values ​​of the node in the current and previous n statistical periods are obtained, and the risk values ​​of each period are weighted and averaged according to preset weights to generate a smoothed risk value; the difference between the smoothed risk value of the current statistical period and the smoothed risk value of the previous statistical period is compared to obtain the magnitude of the risk value change; the magnitude of the risk value increase is mapped to the corresponding risk correction coefficient based on the magnitude of the risk value change; the key features identified in the node state assessment and their current feature values ​​are obtained, the deviation of each key feature from its threshold is output, and the deviation of each key feature is superimposed on the risk correction coefficient according to its weight to generate a risk gain correction coefficient; the actual consumption data is corrected for risk mapping based on the risk gain correction coefficient to generate a virtual consumption amount reflecting the node's operational risk.

[0015] In a preferred embodiment, the step of forecasting component demand based on node status assessment, generating forecasted demand data, and outputting component pull decision instructions in conjunction with virtual consumption is as follows: Virtual consumption is used as a risk-corrected consumption benchmark reflecting the risk of node operating status, resulting in a risk-corrected consumption dataset; based on the risk-corrected consumption dataset and combined with the continuous status risk value of the corresponding node, the consumption data of each component is weighted and adjusted to generate a risk-weighted consumption dataset; within a preset short-cycle forecast window, trend analysis is performed on the risk-weighted consumption dataset to extract the direction and magnitude of component consumption changes, generating short-cycle consumption change characteristic data; based on the short-cycle consumption change characteristic data, time-series analysis is performed on component demand in the next short cycle to obtain short-cycle forecasted demand data; the short-cycle forecasted demand data is compared and analyzed with current inventory, in-transit replenishment quantity, and safety stock threshold to determine whether component pull conditions are triggered; when pull conditions are met, the pull replenishment quantity of the corresponding component is determined based on the difference between the short-cycle forecasted demand data and the risk-corrected consumption dataset; and a component pull decision instruction is generated based on the pull replenishment quantity.

[0016] In a preferred embodiment, the step of adjusting the consumption data of each component based on the risk-corrected consumption dataset and the continuous state risk value of the corresponding node to generate a risk-weighted consumption dataset is as follows: After obtaining the risk-corrected consumption dataset, for each type of component in the dataset, the continuous state risk value of its corresponding node is obtained; the corresponding consumption weight adjustment range is determined according to the magnitude of the continuous state risk value; based on the consumption weight adjustment ratio determined for each component, the risk-corrected consumption amount of each type of component in the risk-corrected consumption dataset within the statistical period is used as the basic consumption value; the basic consumption value is processed according to the consumption weight adjustment ratio determined for the continuous state risk value of the corresponding node of each component to generate a weighted consumption result that incorporates the differences in node operating risk; the weighted consumption results corresponding to each component are summarized to form a risk-weighted consumption dataset.

[0017] In a preferred embodiment, the step of coordinating and scheduling execution of each node according to the pull decision instruction specifically includes: parsing the pull decision instruction to form a scheduling task parameter set; based on the scheduling task parameter set, splitting the pull replenishment demand into corresponding supplier execution demand and warehousing execution demand, and generating supply execution instructions corresponding to each supplier node; sending the supply execution instructions to the corresponding supplier nodes, and adjusting the execution priority of the supplier nodes according to the continuous state risk value to obtain the supplier scheduling execution result; generating the corresponding warehousing warehousing scheduling plan according to the warehousing execution demand and the predicted arrival time, and sending the warehousing warehousing scheduling plan to the warehousing node to form the warehousing scheduling execution result; after obtaining the expected delivery information of the supplier nodes, combining the predicted arrival time and the continuous state risk value, generating the corresponding logistics transportation scheduling instruction, and sending the logistics transportation scheduling instruction to the logistics transportation node to form the logistics scheduling execution result; and linking and integrating the supplier scheduling execution result, the warehousing scheduling execution result, and the logistics scheduling execution result to form a collaborative scheduling execution scheme.

[0018] In a preferred embodiment, the step of providing feedback on the results of the collaborative scheduling execution process and dynamically correcting the virtual consumption and pull decision instructions based on the feedback results is as follows: After the collaborative scheduling execution plan completes its execution task, execution status information is collected in real time to form a collaborative scheduling execution feedback dataset; the collaborative scheduling execution feedback dataset is compared with the expected supply, warehousing, and logistics execution data in the original pull decision instructions to identify the deviation between the actual execution results and the expected plan, forming execution deviation information; based on the execution deviation information, the virtual consumption of the corresponding nodes is dynamically corrected to obtain an updated virtual consumption dataset; based on the execution deviation information and the updated virtual consumption dataset, the next round of component pull decision instructions is dynamically adjusted to generate corrected pull decision instructions.

[0019] A system for automotive parts supply chain management includes a status assessment module, a risk correction module, an instruction module, a scheduling module, and a feedback module, with interconnections between the modules. The status assessment module acquires operational status data of each node in the supply chain, performs status perception and anomaly identification on the operational status of each node, and generates a node status assessment. The operational status data includes actual consumption data. The risk correction module corrects the actual consumption data based on the node status assessment, generating a virtual consumption quantity. The instruction module performs short-cycle forecasting of parts demand based on the node status assessment, generates forecasted demand data, and outputs parts pull decision instructions in conjunction with the virtual consumption quantity. The scheduling module coordinates and executes the execution of each node according to the pull decision instructions. The feedback module provides feedback on the results of the coordinated scheduling execution process and dynamically corrects the virtual consumption quantity and pull decision instructions based on the feedback results.

[0020] The technical effects and advantages of the automotive parts supply chain management method and system of this invention are as follows:

[0021] 1. This invention achieves dynamic correlation and control between supply chain operational risks and component demand by comprehensively perceiving and assessing the operational status of each node in the supply chain, thereby significantly improving the agility, reliability, and responsiveness of the supply chain. Specifically, this invention first acquires operational status data of each node in the supply chain, including the actual consumption of each node, and generates node status assessments through status perception and anomaly identification methods, realizing quantitative analysis of node operational stability and potential anomalies. Based on the node status assessment, this invention performs risk correction on the actual consumption data. By smoothing the continuous cycle status risk values ​​of nodes and combining the deviation of key characteristic items to generate a risk gain correction coefficient, a virtual consumption quantity is obtained. This allows component consumption to reflect potential node risks in advance, reducing the risk of inventory gaps and production interruptions caused by node anomalies or supply instability.

[0022] 2. This invention establishes a result feedback and dynamic correction mechanism. During collaborative scheduling, execution status information is collected in real time, generating execution deviation information. Based on this information, virtual consumption and the next round of pull decision instructions are dynamically corrected, forming an adaptive adjustment closed loop. This enables the supply chain to respond in real time to changes in node operation and actual execution deviations, improving short-cycle scheduling accuracy. This method and system organically combine node risk quantification, virtual consumption correction, short-cycle demand forecasting, pull decision-making, collaborative scheduling, and dynamic adjustment. This not only reduces inventory holdings and shortens response cycles but also improves the supply chain's resilience to sudden events and abnormal nodes. Attached Figure Description

[0023] Figure 1This is a flowchart illustrating a method for managing the automotive parts supply chain according to the present invention.

[0024] Figure 2 This is a schematic diagram of the system structure of an automotive parts supply chain management method according to the present invention. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] Example 1, Figure 1 This invention provides a method for managing the automotive parts supply chain, comprising:

[0027] S1, acquire the operating status data of each node in the supply chain, perform status perception and anomaly identification on the operating status of each node, and generate a node status assessment. The operating status data includes actual consumption data.

[0028] In this embodiment, the operational status data of each node in the supply chain is acquired, and the operational status of each node is assessed for status awareness and anomaly identification to generate a node status evaluation, as detailed below:

[0029] The system acquires operational status data for each node in the automotive parts supply chain. These nodes include at least OEM production nodes, parts supplier nodes, warehousing nodes, and logistics and transportation nodes. The system also acquires operational status data for each node, including but not limited to production consumption, production cycle time, capacity utilization rate, inventory level, inventory turnover rate, logistics and transportation timeliness, and transportation delays.

[0030] The acquired operational status data is preprocessed, including removing abnormal samples, filling in missing data, and standardizing data with different units and dimensions. At the same time, all node data are aligned according to a unified time granularity to form a unified node operational status dataset.

[0031] Based on the node operation status dataset, the operation status features of the nodes are extracted and weighted coefficients are assigned to each feature. A weighted algorithm is used to generate a comprehensive node status score. The operation status features include actual consumption change rate, inventory consumption ratio, logistics timeliness deviation rate and capacity load deviation.

[0032] The node's overall status score is compared with the corresponding multi-level thresholds to perform status perception and anomaly identification, and the node is determined to be in a normal state, a warning state, or an abnormal state. During the determination process, the key feature items that cause the status change and their values ​​are recorded. The multi-level status thresholds include normal threshold range, warning threshold range, and abnormal threshold range, and each threshold range is defined in the form of a numerical range.

[0033] Based on the node status assessment results and corresponding feature items, the actual consumption is dynamically corrected to generate virtual consumption.

[0034] In this embodiment, the running status features of nodes are extracted based on the node running status dataset, and weight coefficients are assigned to each feature. A weighted algorithm is then used to generate a comprehensive node status score, as follows:

[0035] Based on the historical operation records, obtain the operation status data of each node within a continuous and stable operation cycle. Use the operation status data within the specified stable operation cycle as a normal operation sample dataset to characterize the operation characteristics of each node under no obvious abnormalities.

[0036] Statistical analysis is performed on the operating status characteristics in the normal operating sample dataset to calculate the corresponding mean interval and fluctuation interval, and the mean interval and fluctuation interval are used as the normal threshold interval for each status characteristic item.

[0037] Further, abnormal event data of nodes during historical operation are obtained, and node operation status data before and after the occurrence of abnormal events are extracted to form an abnormal operation sample dataset, which is used to represent the operation characteristics of nodes under risk or abnormal conditions.

[0038] Statistical analysis is performed on the operational status characteristics in the abnormal operation sample dataset to determine the value range of the status characteristic items under abnormal operation conditions. The specified value ranges exceeding the normal threshold range are divided into warning threshold ranges and abnormal threshold ranges.

[0039] Based on the normal operation sample dataset and the abnormal operation sample dataset, the value distribution characteristics of each operation state feature in the stable operation state and the abnormal operation state are statistically analyzed. The frequency and magnitude of each operation state feature exceeding the normal threshold range in the abnormal operation sample dataset are output to characterize the influence intensity of each operation state feature on the abnormal operation of the node.

[0040] Based on the normal threshold range, the early warning threshold range, and the abnormal threshold range, according to the influence intensity of each operating state characteristic on the node's operating stability, corresponding weight coefficients are configured for different operating state characteristic items, and a comprehensive node status score is generated by weighted calculation based on the operating state characteristic value and the weight coefficient.

[0041] The node's overall status score is compared numerically with a preset multi-level status threshold, wherein the multi-level status threshold includes a normal threshold range, a warning threshold range, and an abnormal threshold range, and each threshold range is defined in the form of a numerical range.

[0042] When the node's overall status score falls within the normal threshold range, the corresponding node's operating status is determined to be normal, and the node's overall status score and the current values ​​of each status feature item are recorded as the node status evaluation result.

[0043] When the overall status score of a node falls into the warning threshold range, the operating status of the corresponding node is determined to be a warning state. While recording the node status evaluation results, the feature values ​​of each status feature item are further compared with their corresponding warning thresholds to identify status feature items that exceed or are close to the warning threshold. The identified status feature items and their corresponding values ​​are recorded as key feature items that trigger status changes.

[0044] When the overall status score of a node falls into the abnormal threshold range, the operating status of the corresponding node is determined to be abnormal. The feature value of each status feature item is compared with its corresponding abnormal threshold item by item to identify the status feature items that exceed the abnormal threshold. The status feature items and their corresponding values ​​are recorded as the key feature items that cause the node to enter the abnormal state.

[0045] S2, based on the node status assessment, performs risk correction on the actual consumption data and generates virtual consumption;

[0046] In this embodiment, the actual consumption data is risk-corrected based on the node status assessment to generate a virtual consumption amount, as detailed below:

[0047] Based on the node status assessment results, obtain the actual consumption data of the corresponding node in the current statistical period, and use the actual consumption data as the baseline consumption input of the node in the current operating state;

[0048] Based on the node status assessment results, the node's comprehensive status score and corresponding key feature items are extracted to characterize the degree of deviation of the node's current operating status from the normal operating baseline.

[0049] Based on the positional relationship between the node's comprehensive status score and the normal status threshold interval, the early warning status threshold interval, and the abnormal status threshold interval, the node's current operating status is mapped to a continuous status risk value, so that the more the node's comprehensive status score deviates from the normal status interval, the greater the corresponding status risk value.

[0050] Based on the time series processing mechanism, the state risk value of a node is smoothed over multiple consecutive statistical periods to ensure that the node state risk value reflects the changing trend of the operating state, thereby reducing the impact of single-period fluctuations on the consumption correction results. Specifically, in each statistical period, the state risk values ​​of the current period and several previous periods are taken as window data. The smoothed risk value of the current statistical period is generated by calculating the average or weighted average of the risk values ​​within the window. In the weighted averaging process, the weight is set according to the time distance between the risk value of each period and the current period, so that the risk values ​​of recent periods have a greater impact on the smoothing result, while the risk values ​​of more distant periods have a smaller impact. By applying the above sliding window smoothing process to the entire time series period by period, the smoothed risk value sequence of the node in consecutive statistical periods can be obtained.

[0051] Based on the state risk value and its changing trend, a risk correction coefficient is constructed, and the risk correction coefficient is enhanced according to the identified key feature items and feature values ​​to obtain a risk gain correction coefficient, so that the key features that have a high impact on the stability of node operation have a higher weight in the risk correction coefficient.

[0052] Based on the risk gain correction coefficient, the actual consumption data is corrected by risk mapping, so that the risk of node operation status is reflected in the consumption in a continuous and controllable manner, thereby generating virtual consumption.

[0053] In this embodiment, based on the positional relationship of the node's comprehensive status score relative to the normal status threshold interval, the early warning status threshold interval, and the abnormal status threshold interval, the node's current operating status is mapped to a continuous status risk value, as follows:

[0054] The node comprehensive status score is set with corresponding normal state threshold range, early warning state threshold range and abnormal state threshold range, and each threshold range is arranged sequentially in numerical order to characterize the different stages of the node's operating status evolution from stable to abnormal.

[0055] When the node's overall status score falls into the normal state threshold range, the position of the node's overall status score within the normal state threshold range is mapped to the corresponding risk value range through linear mapping, forming the corresponding basic risk value. The closer the node's overall status score is to the upper limit of the normal state threshold range, the greater the corresponding basic risk value.

[0056] When the node's overall status score falls into the warning state threshold range, the position of the node's overall status score within the warning state threshold range is mapped to the corresponding risk value range through a linear mapping of the deviation magnitude, forming a corresponding enhanced risk value. This causes the risk value to continuously increase as the node's overall status score changes from the lower limit to the upper limit of the warning state threshold range.

[0057] When the node's overall status score falls into the abnormal status threshold range, the deviation of the node's overall status score from the upper limit of the warning status threshold range is mapped to the corresponding risk value range, forming a corresponding high-risk value. The further the node's overall status score deviates from the lower limit of the abnormal status threshold range, the greater the corresponding risk value increase.

[0058] Among them, the basic risk value, enhanced risk value and high risk value are numerically continuous and uniformly normalized to a preset risk value range, thus forming a continuous state risk value.

[0059] In this embodiment, a risk correction coefficient is constructed based on the state risk value and its changing trend. This risk correction coefficient is then enhanced based on the identified key features and their values ​​to obtain a risk gain correction coefficient, as detailed below:

[0060] Obtain the continuous state risk value of the node in the current and previous n statistical periods, and perform a weighted average of the risk values ​​of each period according to the preset weight to generate a smooth risk value, so as to reduce the impact of single-period fluctuations on risk correction.

[0061] The magnitude of change in risk value is obtained by comparing the difference between the smoothed risk value of the current statistical period and the smoothed risk value of the previous statistical period.

[0062] Based on the magnitude of the change in risk value, the increase in risk value is mapped to the corresponding risk correction coefficient, so that the greater the change in risk value, the higher the risk correction coefficient, while when the risk value remains stable or decreases, the basic risk correction coefficient remains at its original value or is appropriately reduced.

[0063] Obtain the key feature items and their current feature values ​​identified in the node status assessment, output the deviation of each key feature item from its threshold, and add the deviation of each key feature item to the risk correction coefficient according to its weight to generate the risk gain correction coefficient.

[0064] The actual consumption data is corrected by risk mapping based on the risk gain correction coefficient, generating a virtual consumption amount that reflects the operational risk of the node.

[0065] In this embodiment, the actual consumption data is corrected by risk mapping based on the risk gain correction coefficient, so that the node operating status risk is reflected in the consumption in a continuous and controllable manner, thereby generating virtual consumption, as follows:

[0066] The direction and magnitude of consumption correction are determined based on the magnitude of the risk gain correction coefficient. When the risk gain correction coefficient is in the low-risk range, the actual consumption remains unchanged or is slightly adjusted to reflect the normal impact of stable node operation on consumption.

[0067] When the risk gain correction coefficient increases as the node operation risk increases, the actual consumption is amplified and corrected step by step according to the state interval corresponding to the risk gain correction coefficient, so that the potential risks such as unstable supply, logistics delay or production fluctuation of the node are reflected in the consumption in advance.

[0068] By using the aforementioned continuous and hierarchical correction method, the actual consumption data is smoothly adjusted under different risk levels, thereby generating a virtual consumption amount that can characterize the risk of node operation status.

[0069] S3, based on node status assessment, performs short-cycle forecasting of component demand, generates forecast demand data, and outputs component pull decision instructions in conjunction with virtual consumption.

[0070] In this embodiment, based on node status assessment, short-cycle forecasts of component demand are made, generating forecast demand data. Combined with virtual consumption data, component pull decision instructions are output, as follows:

[0071] After completing the node status assessment and generating the virtual consumption, the virtual consumption is used as the risk correction consumption benchmark to reflect the risk of node operation status, and a risk correction consumption dataset for short-cycle analysis is obtained.

[0072] Based on the risk-corrected consumption dataset, and combined with the continuous state risk values ​​of the corresponding nodes, the consumption data of each component is weighted and adjusted to generate a risk-weighted consumption dataset, which is used to characterize the risk-sensitive consumption level of components under different node states.

[0073] Within a preset short-cycle prediction window, trend analysis is performed on the risk-weighted consumption dataset to extract the direction and magnitude of changes in component consumption and generate short-cycle consumption change characteristic data.

[0074] Based on short-cycle consumption change characteristic data, time series analysis is performed on the component demand in the next short cycle, and extrapolation forecasts are made for the component demand in the next short cycle to obtain short-cycle forecast demand data that can reflect the node operation risk in advance. Specifically, the short-cycle consumption change characteristic data is used as an input parameter to reflect the consumption trend. When the consumption is on an upward trend, the forecast demand in the next short cycle is adjusted upward accordingly. When the consumption remains stable or downward, the forecast demand is maintained or moderately adjusted downward. At the same time, the continuous state risk value of the corresponding node is introduced into the forecast process to correct the risk of extrapolation results. When the node operation risk is high, the forecast demand is further increased based on the trend forecast result to cover the impact of potential supply instability or delivery delay in advance, thereby generating short-cycle forecast demand data that can reflect the node operation risk in advance.

[0075] Compare and analyze short-cycle forecast demand data with current inventory levels, in-transit replenishment levels, and safety stock thresholds to determine whether the conditions for triggering component demand are met.

[0076] When the pull conditions are met, the pull replenishment quantity of the corresponding parts is determined based on the difference between the short-cycle forecast demand data and the risk correction consumption data set, so that the pull quantity can both cover the short-cycle forecast demand and provide redundant compensation for the node operation risk.

[0077] Based on the quantity of replenishment required, a component pull decision instruction is generated, and the component identifier, node identifier, prediction cycle, and continuous status risk value are associated in the pull decision instruction for subsequent collaborative scheduling and execution by suppliers, warehouses, and logistics nodes.

[0078] In this embodiment, based on the risk-corrected consumption dataset and combined with the continuous state risk values ​​of the corresponding nodes, the consumption data of each component is weighted and adjusted to generate a risk-weighted consumption dataset, as detailed below:

[0079] After obtaining the risk correction consumption dataset, for each type of component in the dataset, the continuous state risk value of its corresponding node is obtained, and the continuous state risk value is used as the basis for risk adjustment reflecting the uncertainty of node operation.

[0080] The corresponding consumption weight adjustment range is determined based on the magnitude of the continuous state risk value, so that the component consumption data corresponding to the node with a higher continuous state risk value has a higher weight in the calculation, while the component consumption data corresponding to the node with a lower continuous state risk value maintains its original weight or is slightly adjusted.

[0081] The consumption weight adjustment ratio is determined for each component, and the risk correction consumption of each type of component in the risk correction consumption data within the statistical period is used as the basic consumption value.

[0082] Based on the consumption weight adjustment ratio determined by the continuous state risk value of each component's corresponding node, the basic consumption value is amplified or maintained to generate a weighted consumption result that incorporates the differences in node operation risk.

[0083] The weighted consumption results for each component are summarized to form a risk-weighted consumption dataset.

[0084] In this embodiment, short-cycle forecast demand data is compared and analyzed with current inventory levels, in-transit replenishment levels, and safety stock thresholds to determine whether the conditions for triggering component demand are met, as detailed below:

[0085] After obtaining the short-cycle forecast demand data for each component, we first obtain the available inventory of the corresponding component in the current statistical period, and then obtain the in-transit replenishment quantity of the component that has been ordered but not yet put into storage.

[0086] The available inventory and in-transit replenishment are combined to calculate the available supply of parts in the next short cycle;

[0087] By comparing and analyzing short-cycle forecasted demand data with available supply, when the short-cycle forecasted demand data is greater than the sum of available supply and the corresponding safety stock threshold, it is determined that there is a risk of insufficient supply for the component in the next short cycle, thereby triggering the component pull condition.

[0088] When the short-cycle forecast demand data is less than or equal to the sum of the available supply and the safety stock threshold, it is determined that the component does not need to be replenished in the next short cycle, and the existing inventory and supply status is maintained.

[0089] S4, based on the pull decision command, coordinates and executes the corresponding supplier nodes, warehousing nodes and logistics transportation nodes;

[0090] In this embodiment, based on the pull decision command, the corresponding supplier nodes, warehousing nodes, and logistics transportation nodes are coordinated and scheduled for execution, as follows:

[0091] After generating the component pull decision instruction, the component identifier, pull replenishment demand, target node identifier, prediction cycle, and continuous state risk value carried in the pull decision instruction are parsed to form a set of scheduling task parameters for scheduling execution.

[0092] Based on the scheduling task parameter set, the pull replenishment demand is broken down into corresponding supplier execution demand and warehousing execution demand, and a supply execution instruction corresponding to each supplier node is generated. The supply execution instruction includes at least the supply quantity, delivery time window and risk identification information.

[0093] The supply execution instruction is sent to the corresponding supplier node, and the execution priority of the supplier node is adjusted according to the continuous state risk value to obtain the supplier scheduling execution result;

[0094] Based on the required volume of inbound operations and the predicted delivery time, a corresponding warehousing inbound scheduling plan is generated and sent to the warehousing nodes to reserve storage space resources and arrange inbound operation capacity, thus forming the warehousing scheduling execution result;

[0095] After obtaining the expected delivery information from the supplier nodes, the corresponding logistics transportation scheduling instructions are generated by combining the predicted delivery time and continuous status risk value. The logistics transportation scheduling instructions are then sent to the logistics transportation nodes to arrange the transportation mode, transportation route and transportation priority, thus forming the logistics scheduling execution result.

[0096] The results of supplier scheduling, warehouse scheduling, and logistics scheduling are linked and integrated to form a collaborative scheduling execution plan for the same pull decision instruction. The collaborative scheduling execution plan is then uniformly issued and executed.

[0097] In this embodiment, the supplier scheduling execution results, warehouse scheduling execution results, and logistics scheduling execution results are linked and integrated to form a collaborative scheduling execution scheme for the same pull decision command, as detailed below:

[0098] After obtaining the supplier scheduling execution results, warehouse scheduling execution results, and logistics scheduling execution results, the three types of execution results are grouped according to the component identification, pull replenishment demand, target node identification, and execution time window contained in each result. The supply, warehousing, and logistics execution tasks of the same component are then linked.

[0099] In each group of related tasks, the execution results are aligned according to the demand for replenishment, ensuring that the expected shipment quantity of the supply node is consistent with the warehousing capacity of the storage node and the transportation capacity of the logistics node, thereby forming a collaborative scheduling execution unit that can cover the entire supply chain.

[0100] Prioritize and optimize the timing of each collaborative scheduling execution unit. Based on the continuous state risk value of the nodes, prioritize the scheduling of high-risk component tasks. Specifically, for each collaborative scheduling execution unit, compare the continuous state risk value of the nodes corresponding to the components it contains with a pre-set risk threshold. When the risk value is greater than the risk threshold, the component and its corresponding execution task are identified as high-risk tasks. For tasks with risk values ​​lower than the risk threshold, they are identified as normal priority tasks. On this basis, all collaborative scheduling execution units are sorted according to task priority, and the component tasks identified as high-risk are placed in the priority execution sequence. At the same time, the timing of high-risk tasks is optimized by combining the execution time window of each task to ensure that high-risk components can be scheduled first in all aspects of supply, warehousing and logistics, thereby reducing the impact of potential supply interruptions or delays.

[0101] All collaborative scheduling execution units are integrated and summarized to generate a complete collaborative scheduling execution plan for the same pull decision command. The collaborative scheduling execution plan includes the specific execution order, quantity, time window and risk priority information of each component in the supply, warehousing and logistics links, and is distributed to the corresponding nodes for unified execution and monitoring to ensure that the pull replenishment task is completed in a coordinated manner throughout the entire supply chain.

[0102] S5 provides feedback on the results of the collaborative scheduling execution process and dynamically corrects the virtual consumption and pull decision instructions based on the feedback results.

[0103] In this embodiment, the results of the collaborative scheduling execution process are fed back, and the virtual consumption and pull decision instructions are dynamically corrected based on the feedback results, as follows:

[0104] After each supplier node, warehousing node, and logistics transportation node completes the task execution according to the collaborative scheduling execution plan, the execution status information is collected in real time, including the actual quantity shipped, arrival time, inbound quantity, transportation delay, and possible abnormal events, forming a collaborative scheduling execution feedback dataset.

[0105] The collaborative scheduling execution feedback dataset is compared with the expected supply, warehousing and logistics execution data in the original pull decision instruction to identify the deviation between the actual execution results and the expected plan, including insufficient parts, delivery delays or advances, and the occurrence of abnormal events, and the deviation information is recorded as execution deviation information.

[0106] Based on the execution deviation information, the virtual consumption of the corresponding node is dynamically corrected to obtain an updated virtual consumption dataset.

[0107] If the actual consumption is lower than expected, the virtual consumption is reduced to reflect the high efficiency of downstream execution in the supply chain.

[0108] If actual consumption is higher than expected or there are abnormal delivery delays, the virtual consumption will be increased to reflect potential supply or shortage risks.

[0109] Based on the execution deviation information and the updated virtual consumption dataset, the next round of component pull decision instructions are dynamically adjusted, including modifying the replenishment quantity, adjusting the replenishment time window, and reallocating priorities, and a revised pull decision instruction is generated.

[0110] The revised pull decision instructions are reissued to the corresponding suppliers, warehouses and logistics nodes, and feedback information is continuously collected during the execution process to form a closed-loop dynamic correction mechanism.

[0111] In this embodiment, the next round of component pull decision instructions is dynamically adjusted based on execution deviation information and the updated virtual consumption dataset, as follows:

[0112] Obtain the execution deviation information of the previous round of coordinated scheduling and the updated virtual consumption dataset of the corresponding nodes;

[0113] For each component, the original planned replenishment quantity, replenishment time window, and allocated nodes in the pull decision instruction are checked item by item. The impact of the actual execution deviation on the next round of pull demand is calculated. For example, if the actual supply is lower than the original plan or there is a delay, the corresponding quantity is increased in the next round of pull replenishment to make up for the potential shortage. If the actual supply is higher than expected or completed ahead of schedule, the quantity of the next round of replenishment is appropriately reduced to avoid inventory backlog.

[0114] By combining the updated virtual consumption dataset, risk corrections are made to the replenishment quantity for each node. For example, for nodes with high continuous risk values, the replenishment quantity is further increased or the replenishment time window is brought forward to reduce the risk of supply interruption. For low-risk nodes, the original plan can be maintained or slightly adjusted.

[0115] After the replenishment quantity and time window are adjusted, resources are coordinated for multiple nodes involved in the pull decision instruction to ensure that the adjusted pull decision can be executed at each node, and the revised pull decision instruction is generated.

[0116] Example 2, Figure 2 The present invention provides a system for automotive parts supply chain management, comprising a status assessment module, a risk correction module, an instruction module, a scheduling module, and a feedback module, with connections between the modules;

[0117] The status assessment module is used to acquire the operational status data of each node in the supply chain, perform status perception and anomaly identification on the operational status of each node, and generate node status assessments. The operational status data includes actual consumption data.

[0118] The risk correction module is used to correct the actual consumption data based on the node status assessment and generate virtual consumption.

[0119] The instruction module is used to make short-cycle predictions of component demand based on node status assessment, generate predicted demand data, and output component pull decision instructions in combination with virtual consumption.

[0120] The scheduling module is used to coordinate and execute the execution of each node according to the pull decision command;

[0121] The feedback module is used to provide feedback on the results of the collaborative scheduling execution process and to dynamically adjust the virtual consumption and pull decision instructions based on the feedback results.

[0122] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0123] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0124] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0125] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0126] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0127] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for managing the automotive parts supply chain, characterized in that, include: The system acquires operational status data of each node in the supply chain, performs status perception and anomaly identification on the operational status of each node, and generates node status assessments. The operational status data includes actual consumption data. Based on the node status assessment, the actual consumption data is risk-corrected to generate virtual consumption. Based on node status assessment, short-cycle forecasts of component demand are made, forecast demand data is generated, and component pull decision instructions are output in combination with virtual consumption. Based on the pull decision command, each node is coordinated and scheduled for execution; The system provides feedback on the results of the coordinated scheduling execution process and dynamically adjusts the virtual consumption and pull decision instructions based on the feedback results.

2. The automotive parts supply chain management method according to claim 1, characterized in that, The process of acquiring operational status data of each node in the supply chain, performing status perception and anomaly identification on the operational status of each node, and generating node status assessments is as follows: Acquire operational status data of each node in the automotive parts supply chain, and preprocess the operational status data to form a node operational status dataset; Based on the node running status dataset, the running status features of the nodes are extracted, and each feature is assigned a weight coefficient. A weighted algorithm is then used to generate a comprehensive node status score. The node's overall status score is compared with the corresponding multi-level threshold to determine the node's status, and the key feature items that cause the status change and their values ​​are recorded. Based on the node status assessment results and corresponding feature items, the actual consumption is dynamically corrected to generate virtual consumption.

3. The automotive parts supply chain management method according to claim 2, characterized in that, The node operation status features are extracted from the node operation status dataset, and weight coefficients are assigned to each feature. A weighted algorithm is then used to generate a comprehensive node status score, as detailed below: Based on the historical operation records, obtain the operation status data of each node within a continuous and stable operation cycle, and use the operation status data within the specified stable operation cycle as the normal operation sample dataset. Perform statistical analysis on the operating status characteristics in the normal operating sample dataset, output the corresponding mean interval and fluctuation interval, and use the mean interval and fluctuation interval as the normal threshold interval for each status characteristic item; Obtain abnormal event data of nodes during their historical operation, and extract node operation status data before and after the occurrence of abnormal events to form an abnormal operation sample dataset; Statistical analysis is performed on the operational status characteristics in the abnormal operation sample dataset to determine the value range of the status characteristic items under abnormal operation conditions. The specified value ranges exceeding the normal threshold range are divided into warning threshold ranges and abnormal threshold ranges. Based on the normal operation sample dataset and the abnormal operation sample dataset, the value distribution characteristics of each operation state feature in the stable operation state and the abnormal operation state are statistically analyzed, and the frequency and magnitude of each operation state feature exceeding the normal threshold range in the abnormal operation sample dataset are output. Based on the normal threshold range, the early warning threshold range, and the abnormal threshold range, according to the influence intensity of each operating state characteristic on the node's operating stability, corresponding weight coefficients are configured for different operating state characteristic items, and a comprehensive node status score is generated by weighted calculation based on the operating state characteristic value and the weight coefficient. The node's overall status score is compared numerically with a preset multi-level status threshold to extract key feature items.

4. The automotive parts supply chain management method according to claim 1, characterized in that, The process of risk correction based on node status assessment to generate virtual consumption data is as follows: Based on the node status assessment results, obtain the actual consumption data of the corresponding node in the current statistical period, and use the actual consumption data as the baseline consumption input of the node in the current operating state; Based on the node status assessment results, the node's comprehensive status score and corresponding key feature items are extracted; Based on the positional relationship of the node's comprehensive status score relative to the normal status threshold interval, the early warning status threshold interval, and the abnormal status threshold interval, the node's current operating status is mapped to a status risk value. Based on the time series processing mechanism, the state risk value of a node is smoothly calculated over multiple consecutive statistical periods. Based on the state risk value and its changing trend, a risk correction coefficient is constructed, and the risk correction coefficient is enhanced based on the identified key feature items and feature values ​​to obtain the risk gain correction coefficient. Based on the risk gain correction coefficient, the actual consumption data is corrected by risk mapping to generate virtual consumption.

5. The automotive parts supply chain management method according to claim 4, characterized in that, The risk correction coefficient is constructed based on the state risk value and its changing trend. Then, the risk correction coefficient is enhanced based on the identified key features and eigenvalues ​​to obtain the risk gain correction coefficient, as detailed below: Obtain the continuous state risk value of the node in the current and previous n statistical periods, and generate a smoothed risk value by weighting the risk value of each period according to the preset weight; By comparing the difference between the smoothed risk value of the current statistical period and the smoothed risk value of the previous statistical period, the magnitude of change in the risk value can be obtained. The magnitude of the increase in risk value is mapped to the corresponding risk correction coefficient based on the magnitude of the change in risk value; Obtain the key feature items and their current feature values ​​identified in the node status assessment, output the deviation of each key feature item from its threshold, and add the deviation of each key feature item to the risk correction coefficient according to its weight to generate the risk gain correction coefficient. The actual consumption data is corrected by risk mapping based on the risk gain correction coefficient, generating a virtual consumption amount that reflects the operational risk of the node.

6. The automotive parts supply chain management method according to claim 1, characterized in that, The process involves assessing node status, making short-term forecasts of component demand, generating forecast demand data, and combining this with virtual consumption data to output component pull decision instructions, as detailed below: Using virtual consumption as a risk correction consumption benchmark to reflect the risk of node operating status, a risk correction consumption dataset is obtained. Based on the risk-corrected consumption dataset, and combined with the continuous state risk value of the corresponding node, the consumption data of each component is weighted and adjusted to generate a risk-weighted consumption dataset. Within a preset short-cycle prediction window, trend analysis is performed on the risk-weighted consumption dataset to extract the direction and magnitude of changes in component consumption and generate short-cycle consumption change characteristic data. Based on the short-cycle consumption change characteristics data, time series analysis is performed on the component demand in the next short cycle to obtain short-cycle forecast demand data. Compare and analyze short-cycle forecast demand data with current inventory levels, in-transit replenishment levels, and safety stock thresholds to determine whether the conditions for triggering component demand are met. When the pull conditions are met, the pull replenishment quantity of the corresponding parts is determined based on the difference between the short-cycle forecast demand data and the risk-corrected consumption data set. Component pull decision instructions are generated based on the quantity of replenishment required.

7. The automotive parts supply chain management method according to claim 6, characterized in that, The risk-corrected consumption dataset, combined with the continuous state risk values ​​of the corresponding nodes, is used to adjust the weights of the consumption data of each component, generating a risk-weighted consumption dataset, as detailed below: After obtaining the risk correction consumption dataset, for each type of component in the dataset, obtain the continuous state risk value of its corresponding node. The corresponding consumption weight adjustment range is determined based on the magnitude of the continuous state risk value. The consumption weight adjustment ratio is determined for each component, and the risk correction consumption of each type of component in the risk correction consumption data within the statistical period is used as the basic consumption value. Based on the consumption weight adjustment ratio determined by the continuous state risk value of each component's corresponding node, the basic consumption value is processed to generate a weighted consumption result that incorporates the differences in node operation risk. The weighted consumption results for each component are summarized to form a risk-weighted consumption dataset.

8. The automotive parts supply chain management method according to claim 1, characterized in that, The coordinated scheduling and execution of each node based on the pull decision command is as follows: Analyze the pull decision instructions to form a set of scheduling task parameters; Based on the scheduling task parameter set, the pull replenishment demand is broken down into the corresponding supplier execution demand and warehousing execution demand, and supply execution instructions corresponding to each supplier node are generated. The supply execution instruction is sent to the corresponding supplier node, and the execution priority of the supplier node is adjusted according to the continuous state risk value to obtain the supplier scheduling execution result; Based on the required amount of goods to be received and the predicted delivery time, a corresponding warehouse receiving schedule is generated and sent to the warehouse nodes to form the warehouse scheduling execution result. After obtaining the expected delivery information from the supplier node, the corresponding logistics transportation scheduling instructions are generated by combining the predicted delivery time and continuous status risk value, and then the logistics transportation scheduling instructions are sent to the logistics transportation node to form the logistics scheduling execution result. The results of supplier scheduling, warehouse scheduling, and logistics scheduling are linked and integrated to form a collaborative scheduling and execution plan.

9. The automotive parts supply chain management method according to claim 1, characterized in that, The process of providing feedback on the results of the collaborative scheduling execution, and dynamically correcting the virtual consumption and pull decision instructions based on the feedback results, is detailed as follows: After the collaborative scheduling execution scheme completes the execution task, it collects execution status information in real time and forms a collaborative scheduling execution feedback dataset. The collaborative scheduling execution feedback dataset is compared with the expected supply, warehousing and logistics execution data in the original pull decision instruction to identify the deviation between the actual execution results and the expected plan, thus forming execution deviation information; Based on the execution deviation information, the virtual consumption of the corresponding node is dynamically corrected to obtain an updated virtual consumption dataset. Based on the execution deviation information and the updated virtual consumption dataset, the next round of component pull decision instructions are dynamically adjusted to generate revised pull decision instructions.

10. A system using the automotive parts supply chain management method as described in any one of claims 1-9, characterized in that, It includes a status assessment module, a risk correction module, an instruction module, a scheduling module, and a feedback module, and the modules are interconnected. The status assessment module is used to acquire the operational status data of each node in the supply chain, perform status perception and anomaly identification on the operational status of each node, and generate node status assessments. The operational status data includes actual consumption data. The risk correction module is used to correct the actual consumption data based on the node status assessment and generate virtual consumption. The instruction module is used to make short-cycle predictions of component demand based on node status assessment, generate predicted demand data, and output component pull decision instructions in combination with virtual consumption. The scheduling module is used to coordinate and execute the execution of each node according to the pull decision command; The feedback module is used to provide feedback on the results of the collaborative scheduling execution process and to dynamically adjust the virtual consumption and pull decision instructions based on the feedback results.

Images