Material demand analysis and prediction method and device based on big data, and storage medium

Abstract

The invention provides a big data-based material demand analysis and prediction method and device and a storage medium, and the method comprises the steps: obtaining a multi-source historical material data set of a supply chain link where a target enterprise is located, carrying out the cross-link integration processing of the multi-source historical material data set, and obtaining a cross-link integrated data set, performing demand feature extraction processing on the cross-link integrated data set to obtain a material demand dynamic feature and a material trend evolution feature of each material circulation sequence, and based on a preset dynamic analysis model, performing joint trend matching processing on the material demand dynamic feature and the material trend evolution feature to obtain a material demand dynamic feature and a material trend evolution feature; and generating a predicted demand distribution result of the material circulation sequence, generating a material supply optimization strategy according to the predicted demand distribution result, and feeding back the material supply optimization strategy to the supply chain management system to trigger material allocation operation. According to the method, the hysteresis of manual decision making can be avoided, and meanwhile, the accuracy and the anti-interference capability of strategy execution are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to a method, device, and storage medium for analyzing and predicting material requirements based on big data. Background Art

[0002] With the increasing complexity of supply chain management, the technology of material requirement analysis and prediction has become a key link in enterprise resource planning. Existing technologies usually use historical data in the supply chain (such as warehouse inventory or purchase order records) for independent modeling and analysis, and generate local demand prediction results through time series prediction or regression algorithms. However, due to ignoring data conflicts between different links (such as timestamp misalignment and inconsistent measurement units) and the conduction effect of material flow states, the prediction results cannot accurately reflect the real demand changes under the multi-link linkage; at the same time, traditional methods mostly use methods based on periodic mean analysis, lacking the ability to capture the dynamic association between sudden demand fluctuations and the load of supply path nodes, making it difficult for supply strategies to adapt to real-time supply and demand imbalance scenarios; in addition, the output results of existing prediction models mostly stay at the demand value level, and do not form a deep coupling with the multi-node resource allocation mechanism of the supply chain system, resulting in a break in the conversion of prediction results to execution strategies, further amplifying the market response risk brought by strategy lag. Summary of the Invention

[0003] In view of this, the present invention provides a method, device, and storage medium for analyzing and predicting material requirements based on big data.

[0004] The technical solution of the embodiment of the present invention is implemented as follows: On the one hand, the embodiment of the present invention provides a method for analyzing and predicting material requirements based on big data, including the following steps: Obtain a multi-source historical material data set of the supply chain link where the target enterprise is located, where the multi-source historical material data set includes multiple material flow sequences, and each material flow sequence is composed of at least one material demand event and a corresponding material supply event; Perform cross-link integration processing on the multi-source historical material data set to obtain a cross-link integration data set, where the cross-link integration data set is used to reflect the change in the material flow state between different supply chain links; Perform demand feature extraction processing on the cross-link integration data set to obtain the material demand dynamic feature and the material trend evolution feature of each material flow sequence; Based on a preset dynamic analysis model, perform joint trend matching processing on the material demand dynamic feature and the material trend evolution feature to generate a predicted demand distribution result of the material flow sequence; A material supply optimization strategy is generated according to the predicted demand distribution result, and the material supply optimization strategy is fed back to the supply chain management system to trigger a material allocation operation.

[0005] In a second aspect, the present invention provides a material demand analysis and forecasting device, comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and the processor implements the steps in the above-described method when executing the program.

[0006] In a second aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the above-described method when executed by a processor.

[0007] The material demand analysis and prediction method based on big data provided by the present invention obtains the full-link historical material data of multiple links in the supply chain and performs cross-link conflict analysis and data alignment to construct an integrated data set reflecting the real flow status, effectively solving the prediction deviation problem caused by the fragmentation of simple link data in traditional methods; by synchronously extracting the dynamic characteristics of material demand that characterize the real-time fluctuation state and the material trend evolution characteristics of the cross-link transmission effect from the integrated data, and combining them with the dynamic analysis model for joint matching, it realizes multi-dimensional modeling of the coupling effect of demand mutation and supply bottleneck, so that the prediction results can simultaneously capture the combined impact of local supply and demand imbalance and global transmission delay, significantly improving the accuracy of demand distribution prediction in complex supply chain scenarios; further based on the prediction results, an optimization strategy including the dynamic weights of inventory allocation and external procurement is generated and fed back to the supply chain system in real time, forming a closed-loop response mechanism that links demand fluctuation perception with resource allocation, while avoiding the lag of manual decision-making, enhancing the accuracy and anti-interference ability of strategy execution, and ensuring the stable operation of the material supply system in a changing market environment.

[0008] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and are not intended to limit the technical solutions of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A schematic diagram of the implementation flow of a material demand analysis and forecasting method based on big data provided in an embodiment of the present invention.

[0010] Figure 2 A hardware entity diagram of a material demand analysis and forecasting device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0011] In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention are further elaborated in detail below with reference to the accompanying drawings and embodiments. The described embodiments should not be regarded as limiting the present invention. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0012] An embodiment of the present invention provides a material demand analysis and forecasting method based on big data, which can be executed by a processor of a material demand analysis and forecasting device. The material demand analysis and forecasting device can be a device with data processing capabilities, such as a server, laptop, tablet computer, or desktop computer.

[0013] Figure 1 A schematic diagram of the implementation flow of a material demand analysis and forecasting method based on big data provided by an embodiment of the present invention is shown in FIG. Figure 1 As shown, the method includes the following steps: Step S100: Acquire a multi-source historical material data set of the supply chain link where the target enterprise is located. The multi-source historical material data set includes multiple material flow sequences, and each material flow sequence consists of at least one material demand event and a corresponding material supply event.

[0014] The target enterprise's supply chain segment refers to its specific position and functions within the entire supply chain system. The supply chain encompasses a series of processes, from raw material procurement, manufacturing, product distribution, and ultimately, the end customer. The target enterprise may be a raw material supplier, manufacturer, distributor, or retailer. A multi-source historical material data set is a collection of historical material-related data collected from multiple data sources. These data sources can include an enterprise's internal inventory management system, production record system, sales order system, as well as external supplier data and market research data. A material flow sequence is a series of data records describing the flow of materials within the supply chain, capturing the entire process from material demand to supply fulfillment. A material demand event refers to a demand for a material within the supply chain, such as a production plan requiring a certain raw material or a customer ordering a corresponding product. A material supply event is the supply response to a material demand event, such as a supplier shipment or a transfer from an internal warehouse.

[0015] For internal enterprise data sources, relevant data can be extracted from inventory management systems, production record systems, and sales order systems using data interfaces. For example, through the data interface of an enterprise resource planning (ERP) system, inventory change data, production material requisition data, sales order data, etc. can be extracted according to preset time ranges and data field requirements. For external data sources, a data sharing agreement can be established with suppliers to obtain their supply records, inventory data, etc. through a secure data transmission channel. Additionally, web crawler technology can be used to obtain relevant market data from public market research websites, industry report databases, etc. When obtaining data, it is necessary to clean and preprocess the data to remove duplicate data, error data, and incomplete data to ensure data quality. For example, for inventory data recorded in the inventory management system, if there are duplicate records or data missing, deduplication and supplementation processing are required. Finally, the processed data is integrated into a data set to form a multi-source historical material data set, which contains multiple material flow sequences, and each sequence clearly records material demand events and corresponding material supply events.

[0016] Step S200: Perform cross-link integration processing on the multi-source historical material data set to obtain a cross-link integration data set, which is used to reflect the change in the flow state of materials between different supply chain links.

[0017] Cross-link integration processing refers to comprehensively processing data from different supply chain links in the multi-source historical material data set, eliminating differences and conflicts between the data, so that the data can accurately reflect the flow of materials in the entire supply chain. The cross-link integration data set is a data set obtained after integration processing, which contains detailed information on the flow of materials between different supply chain links and can reflect changes in the material flow state, such as the flow direction of materials, flow speed, changes in inventory levels, etc.

[0018] As an implementation method, in step S200, performing cross-link integration processing on the multi-source historical material data set to obtain a cross-link integration data set can specifically include the following steps S210~S240: Step S210: Obtain the original data source identifier of each material flow sequence in the multi-source historical material data set, and determine the type of supply chain link corresponding to the material flow sequence according to the original data source identifier.

[0019] The original data source identifier is information used to identify the initial source of each material flow sequence data. It can be the name, number, address, etc. of the data source. Through this identifier, the source of the data can be traced. The type of supply chain link refers to different functional links in the supply chain, such as the raw material supply link, production and manufacturing link, product distribution link, retail link, etc.

[0020] The acquisition of the original data source identifier can be achieved by adding an identifier field to each data record during the data collection process. For example, when collecting data from different systems, add a "data source identifier" field to each data record to record which system the data comes from. After obtaining the original data source identifier, determine the supply chain link type corresponding to each material flow sequence according to the pre-defined mapping relationship between the data source and the supply chain link type. For example, it is pre-defined that the data of the inventory management system corresponds to the production and manufacturing link, and the data of the sales order system corresponds to the product distribution link, etc. If the original data source identifier of a material flow sequence indicates that it comes from the inventory management system, then it can be determined that the supply chain link type corresponding to this material flow sequence is the production and manufacturing link.

[0021] Step S220: For each supply chain link type, extract the abnormal data segments in the material flow sequence. The abnormal data segments include the conflict data segments where there is a temporal conflict between the material demand event and the material supply event.

[0022] The abnormal data segment refers to the part of the data in the material flow sequence that does not conform to the normal logic or rule. The temporal conflict means that the material demand event and the material supply event appear in an unreasonable time order, such as the material supply event occurring before the material demand event, or the time interval of the material supply event not conforming to the normal production or supply cycle. The conflict data segment is the specific data range containing the temporal conflict.

[0023] The abnormal data segments can be extracted by using the time series analysis method. First, sort the material demand events and material supply events in each material flow sequence in chronological order. Then, check the time relationship between adjacent material demand events and material supply events. For example, if it is found that the time of a certain material supply event is earlier than the time of its corresponding material demand event, or the time interval between material supply events is too short or too long, exceeding the normal fluctuation range, mark this part of the data as an abnormal data segment. When specifically implementing, a threshold for the time interval can be set, and when the time interval between adjacent events is less than or greater than this threshold, it is determined that there is a temporal conflict.

[0024] Step S230: Perform conflict resolution processing on the abnormal data segments, generate a conflict resolution result, and correct the data of the abnormal data segments according to the conflict resolution result to obtain the corrected material flow sequence.

[0025] Conflict resolution processing analyzes and interprets the temporal conflicts existing in abnormal data segments to find out the causes of the conflicts. The conflict resolution result is the information about the conflict causes and solutions obtained after conflict resolution processing. Data correction adjusts the data in the abnormal data segment according to the conflict resolution result to make it conform to normal logic and rules.

[0026] For conflict resolution processing of abnormal data segments, a method combining rule reasoning and machine learning can be adopted. For example, define conflict resolution rules. If the material supply event is earlier than the material demand event, it may be caused by data entry errors or advance stockpiling. Then, conduct a preliminary analysis of the abnormal data segment according to these rules. At the same time, use machine learning algorithms to learn from historical data and establish a prediction model for conflict causes. For example, use the decision tree algorithm, taking the features of the abnormal data segment (such as time interval, event type, etc.) as input and the conflict cause as output to train the model. During actual parsing, input the features of the abnormal data segment into the trained model to obtain the prediction result of the conflict cause. Perform data correction according to the conflict resolution result. If it is a data entry error, correct the data by manually checking the original record; if it is caused by advance stockpiling, adjust the time sequence and quantity of the demand event and supply event according to the actual situation. For example, adjust the supply event of advance stockpiling after the demand event and adjust the supply quantity accordingly. After data correction, obtain the corrected material flow sequence.

[0027] Step S240: Align the corrected material flow sequence with other material flow sequences that do not contain abnormal data segments to generate an aligned cross-link data set, and use the aligned cross-link data set as the cross-link integration data set; where data alignment includes associating the material flow sequences of different supply chain links according to the material identifier and unifying the timestamp format and measurement unit in the material flow sequence.

[0028] Data alignment refers to organizing and matching different material flow sequences to make them consistent in some key information for subsequent analysis and processing. The material identifier is the information used to uniquely identify each material, such as material number, name, etc. Through the material identifier, the flow sequences of the same material involved in different supply chain links can be associated. The timestamp format is the format for recording the occurrence time of an event. Different data sources may use different timestamp formats. Unifying the timestamp format can facilitate the analysis of the time sequence of material flow. The measurement unit is the unit for measuring the quantity, weight, volume, etc. of materials. Different links may use different measurement units. Unifying the measurement unit can ensure the accuracy and comparability of data.

[0029] Align the corrected material flow sequence with other material flow sequences that do not contain abnormal data segments. First, group all the material flow sequences according to the material identification, and group the sequences related to the same material into one group. Then, uniformly process the timestamp formats in each group of sequences. Data conversion functions can be used to convert timestamps in different formats into a unified standard format. For example, convert formats such as "YYYY-MM-DD HH:MM:SS" and "DD / MM / YYYY HH:MM" into the "YYYY-MM-DD HH:MM:SS" format. For the unification of measurement units, conversion can be performed according to predefined unit conversion rules. For example, unify kilograms and grams into kilograms, and liters and milliliters into liters, etc. After completing the identification association, timestamp format unification, and measurement unit unification, generate an aligned cross-link data set, which is the cross-link integration data set and can accurately reflect the change in the flow state of materials between different supply chain links.

[0030] Step S300: Perform demand feature extraction processing on the cross-link integration data set to obtain the dynamic material demand characteristics and material trend evolution characteristics of each material flow sequence.

[0031] Demand feature extraction processing is a process of extracting key information from the cross-link integration data set that can reflect the characteristics of material demand. The dynamic material demand characteristics are the characteristics that describe the non-uniform distribution state of material demand in the time dimension, which reflects the change of material demand over time, such as demand fluctuations and mutations. The material trend evolution characteristics are the characteristics that reflect the conduction delay effect and cumulative influence weight of material demand between different supply chain links, which can reveal the propagation law and change trend of material demand in the supply chain.

[0032] As an implementation method, in step S300, perform demand feature extraction processing on the cross-link integration data set to obtain the dynamic material demand characteristics and material trend evolution characteristics of each material flow sequence, which can specifically include the following steps S310 to S340: Step S310: Extract the time-series fluctuation data of each material flow sequence from the cross-link integration data set. The time-series fluctuation data includes the periodic change amplitude of the material demand quantity and the material supply response delay duration.

[0033] The time-series fluctuation data is the data that describes the fluctuation of the material flow sequence in the time dimension. The periodic change amplitude of the material demand quantity refers to the change range of the material demand quantity within a certain period, which reflects the periodic fluctuation of the material demand. The material supply response delay duration refers to the time interval from the occurrence of the material demand event to the start of the corresponding material supply event, which reflects the response speed of the supply chain to the material demand.

[0034] To extract the time-series fluctuation data from the cross-link integration data set, the material demand quantity and the material supply response time in each sequence can be statistically analyzed in the chronological order of the material flow sequence. For the periodic change range of the material demand quantity, first determine an appropriate cycle length, such as monthly, quarterly, or annual cycles. Then, calculate the maximum and minimum values of the material demand quantity within each cycle, and the difference between the two is the change range within that cycle. For the material supply response delay duration, by comparing the timestamps of each material demand event and the corresponding material supply event, calculate the time difference between the two as the supply response delay duration of this demand-supply pair. For example, for a certain material flow sequence, count the material demand quantity for each month, calculate the maximum and minimum values of the demand quantity for each month, and obtain the monthly periodic change range; at the same time, record the times of each demand event and the corresponding supply event, and calculate the supply response delay duration.

[0035] Step S320: Perform pattern recognition processing on the time-series fluctuation data to determine the demand mutation nodes and supply bottleneck nodes in the material flow sequence.

[0036] Pattern recognition processing is a process of analyzing and mining the time-series fluctuation data to find the hidden patterns and rules. The demand mutation node refers to the time point at which the material demand quantity in the material flow sequence suddenly changes significantly, which may be caused by a sudden increase or decrease in market demand, the launch of new products, etc. The supply bottleneck node refers to the node in the supply chain where the material supply is delayed or restricted due to certain factors, which may be caused by insufficient supplier production capacity, transportation problems, etc.

[0037] As an implementation method, in step S320, performing pattern recognition processing on the time-series fluctuation data to determine the demand mutation nodes and supply bottleneck nodes in the material flow sequence may specifically include the following steps S321 to S326: Step S321: Based on the periodic change range of the material demand quantity in the time-series fluctuation data, divide the material flow sequence into multiple fluctuation data segments, and each fluctuation data segment corresponds to a periodic change range interval.

[0038] The fluctuation data segment is a data segment obtained by dividing the material flow sequence according to the periodic change range of the material demand quantity. The periodic change range interval is a predefined change range, and the periodic change range of the material demand quantity within each fluctuation data segment falls within the same interval.

[0039] Cluster analysis can be used to segment material flow sequences into multiple fluctuating data segments based on the cyclical fluctuations in material demand. First, determine the intervals for the cyclical fluctuations, such as 0-10%, 10%-20%, and 20%-30%. Then, statistically analyze the cyclical fluctuations within each material flow sequence and segment the sequence into different fluctuating data segments based on the intervals to which they belong. For example, for a material flow sequence, calculate the cyclical fluctuations for each cycle and assign data with a fluctuation range of 0-10% to one fluctuating data segment, data with a fluctuation range of 10%-20% to another, and so on.

[0040] Step S322: Calculate the difference between adjacent time periods for the material demand in each fluctuation data segment to generate a difference value sequence, and determine the candidate mutation position based on the continuously increasing or decreasing difference direction change in the difference value sequence.

[0041] Adjacent period variance calculation calculates the difference in material demand between two adjacent periods. This calculation can reflect the changing trend of material demand. A variance sequence is a sequence of variances calculated from adjacent period variance calculations. Candidate mutation locations are locations in the variance sequence where the direction of the variance changes from increasing to decreasing, potentially indicating a demand mutation.

[0042] For each fluctuation data segment, the material demand quantity difference between adjacent time periods is calculated, sequentially calculating the difference between the material demand quantities in two adjacent time periods in chronological order. For example, for a fluctuation data segment with a material demand quantity sequence of [100, 120, 130, 110, 90], the difference values between adjacent time periods are calculated to be [20, 10, -20, -20]. Next, the difference value sequence is examined for changes in direction, either increasing or decreasing. If the difference value changes from increasing to decreasing or from decreasing to increasing, the position is marked as a candidate mutation location. In the above example, the position where the difference value changes from 10 to -20 is a candidate mutation location.

[0043] Step S323: extract the material demand change rate within the time series window corresponding to the candidate mutation position. If the change rate exceeds the average change rate threshold of the fluctuation data segment, the candidate mutation position is marked as a demand mutation node.

[0044] The time window is a time range set around a candidate mutation location and is used to analyze changes in material demand near that location. The material demand change rate refers to the percentage change in material demand within the time window, reflecting the rate of change. The average change rate threshold is a critical value set based on the average change in material demand within the fluctuation data segment and is used to determine whether a change in material demand is a mutation.

[0045] To extract the rate of change of material demand within the time series window corresponding to the candidate mutation location, first determine the size of the time series window. For example, centered around the candidate mutation location, select three time periods before and after as the time series window. Then, calculate the initial and final values of the material demand within the time series window. Subtract the initial value from the final value and divide it by the initial value to obtain the rate of change of material demand. For example, within a time series window, if the material demand changes from 100 to 150, the rate of change is 50%. Simultaneously, calculate the average rate of change of material demand within the corresponding fluctuation data segment and use this as the average rate of change threshold. If the rate of change of material demand corresponding to the candidate mutation location exceeds the average rate of change threshold, mark the candidate mutation location as a demand mutation node.

[0046] Step S324: for the material supply response delay duration in the time series fluctuation data, identify the abnormal delay period in which the delay duration exceeds the preset delay benchmark, and extract the supply event identifier associated with the abnormal delay period.

[0047] The preset delay benchmark is a pre-set time threshold used to determine whether a material supply response delay is abnormal. The abnormal delay period is the time period when the material supply response delay exceeds the preset delay benchmark. The supply event identifier uniquely identifies each material supply event. By extracting the supply event identifiers associated with the abnormal delay period, we can further analyze the specific supply event that caused the supply delay.

[0048] Abnormal delay periods are identified based on the length of material supply response delays. Each material supply response delay is compared to a preset delay benchmark. If the delay exceeds the preset delay benchmark, the time period corresponding to the delay is marked as an abnormal delay period. For example, if the preset delay benchmark is 7 days and a material supply response delay is 10 days, then the time period corresponding to this 10-day delay is an abnormal delay period. Supply event identifiers associated with the abnormal delay period are then extracted. These identifiers can be obtained from the cross-link integrated data set and record the specific supply events that occurred during the abnormal delay period.

[0049] Step S325: Obtain the material supply path node information corresponding to the supply event identifier, and based on the supply interruption reason type and path node load status recorded in the node information, filter out the target path nodes whose supply is delayed due to node load exceeding the limit, and mark the target path nodes as supply bottleneck nodes.

[0050] The material supply path node information is the detailed information about each node in the material supply process, including the name, location, function, load status, etc. of the node. The supply interruption reason type classifies the reasons for material supply interruption or delay, such as insufficient supplier production capacity, transportation failure, raw material shortage, etc. The path node load status refers to the workload situation of each supply path node, such as the utilization rate of equipment and the workload of personnel. The target path node refers to the node whose supply is delayed due to node load exceeding the limit.

[0051] To obtain the material supply path node information corresponding to the supply event identifier, it can be achieved by establishing an association relationship between the supply event identifier and the material supply path node information in the cross-link integration data set. For example, add an association field to each supply event to record the supply path node information corresponding to the event. Then, based on the supply interruption reason type and path node load status recorded in the node information, filter out the target path nodes whose supply is delayed due to node load exceeding the limit. If the load status of a certain node shows that the utilization rate of its equipment exceeds 90% and the supply interruption reason type is "insufficient production capacity", it can be determined that this node is the target path node whose supply is delayed due to node load exceeding the limit, and mark it as a supply bottleneck node.

[0052] Step S326: Conduct a spatio-temporal correlation analysis on the demand mutation node and the supply bottleneck node to determine the causal correlation strength between the two in the material flow sequence. If the causal correlation strength reaches the preset correlation threshold, configure a supply bottleneck impact identifier for the demand mutation node.

[0053] The spatio-temporal correlation analysis analyzes the correlation relationship between the demand mutation node and the supply bottleneck node in terms of time and space to determine their causal connection. The causal correlation strength is an indicator to measure the tightness of the causal relationship between the demand mutation node and the supply bottleneck node. The preset correlation threshold is a critical value set in advance to judge whether the causal correlation strength is large enough. The supply bottleneck impact identifier is information used to mark whether the demand mutation node is affected by the supply bottleneck.

[0054] To conduct a spatio-temporal correlation analysis of demand mutation nodes and supply bottleneck nodes, an association rule mining method can be adopted. First, match the time information and location information of demand mutation nodes and supply bottleneck nodes to find node pairs that are close in time and space. Then, calculate the causal association strength between each node pair. For example, the method of conditional probability can be used to calculate the probability of a demand mutation node occurring under the condition that a supply bottleneck node occurs, and this probability is used as the causal association strength. If the causal association strength reaches the preset association threshold, a supply bottleneck impact identifier is configured for the demand mutation node.

[0055] Step S330: Generate dynamic material demand characteristics based on demand mutation nodes and supply bottleneck nodes. The dynamic material demand characteristics are used to characterize the non-uniform distribution state of material demand in the time dimension.

[0056] The dynamic material demand characteristics are the characteristics that describe the non-uniform distribution state of material demand in the time dimension. It reflects the changes in material demand over time, such as demand fluctuations and mutations. Generating dynamic material demand characteristics based on demand mutation nodes and supply bottleneck nodes can comprehensively consider the location and amplitude of demand mutation nodes and the influence of supply bottleneck nodes.

[0057] The specific generation process is as follows. First, conduct a quantitative analysis of demand mutation nodes and record information such as the occurrence time of each demand mutation node and the change amplitude of material demand quantity. For example, record that the demand mutation node occurs on the 10th day, the material demand quantity increases from 100 to 200, and the change amplitude is 100%. Then, consider the influence of supply bottleneck nodes on demand mutation nodes. If a supply bottleneck impact identifier is configured for a certain demand mutation node, it means that this demand mutation may be restricted by the supply bottleneck, and this influence needs to be considered when generating dynamic material demand characteristics. The influence degree of the supply bottleneck can be reflected by setting different weights. For example, for a demand mutation node configured with a supply bottleneck impact identifier, multiply its change amplitude by a weight coefficient less than 1 to reflect the inhibitory effect of the supply bottleneck on demand realization. Finally, integrate the information of all demand mutation nodes to form dynamic material demand characteristics, which can comprehensively reflect the non-uniform distribution state of material demand in the time dimension.

[0058] Step S340: Extract the transfer rate difference data between different supply chain links of the material transfer sequence, and perform trend fitting processing on the transfer rate difference data to generate material trend evolution characteristics; the material trend evolution characteristics are used to reflect the conduction delay effect and cumulative influence weight of material demand between different supply chain links.

[0059] The transfer rate difference data is data that describes the difference in the transfer speed of materials between different supply chain links. It reflects how fast or slow materials flow from one link to another in the supply chain. Trend fitting processing is the process of analyzing and modeling the transfer rate difference data to find the changing trends and patterns in the data. The material trend evolution characteristics are characteristics that reflect the conduction delay effect and cumulative influence weight of material demand between different supply chain links. It can reveal the propagation laws and changing trends of material demand in the supply chain.

[0060] As an implementation method, in step S340, extract the transfer rate difference data of the material transfer sequence between different supply chain links, and perform trend fitting processing on the transfer rate difference data to generate material trend evolution characteristics, which can specifically include the following steps S341 - S346: Step S341: Based on the periodic change amplitude of the material demand quantity and the material supply response delay duration in the time - series fluctuation data, calculate the rate difference between the input material rate and the output material rate of each supply chain link to obtain the initial transfer rate of each supply chain link.

[0061] The input material rate refers to the speed at which materials enter a certain supply chain link, usually expressed by the quantity of materials entering the link per unit time. The output material rate refers to the speed at which materials flow out of a certain supply chain link, also expressed by the quantity of materials flowing out of the link per unit time. The initial transfer rate is obtained by calculating the difference between the input material rate and the output material rate, which reflects the net flow speed of materials in this supply chain link.

[0062] Calculate the initial transfer rate of each supply chain link based on the periodic change amplitude of the material demand quantity and the material supply response delay duration. First, determine the change in the material demand quantity within each period according to the periodic change amplitude of the material demand quantity. For example, if the periodic change amplitude of the material demand quantity within a certain period is 20% and the initial demand quantity is 100, then the change in the demand quantity within this period is 20. Then, combined with the material supply response delay duration, determine the time points when materials enter and leave this link. For example, if the material supply response delay duration is 3 days, then materials will enter this link 3 days after the demand event occurs. Based on this information, calculate the input material rate and the output material rate within each period. For example, within a time period with a period of 7 days, the input material quantity is 80, the output material quantity is 60, then the input material rate is 80 / 7, the output material rate is 60 / 7, and the initial transfer rate is (80 / 7)-(60 / 7)=20 / 7.

[0063] Step S342: Compare the initial transfer rates of adjacent supply chain links in terms of rate differences to generate transfer rate difference data between adjacent links, and record the supply chain link connection identifiers corresponding to the transfer rate difference data.

[0064] Adjacent supply chain links refer to two directly adjacent links in the supply chain, such as the raw material supply link and the production and manufacturing link, the production and manufacturing link and the product distribution link, etc. The transfer rate difference data is the difference between the initial transfer rates of adjacent supply chain links, which reflects the change in the transfer speed of materials between adjacent links. The supply chain link connection identifier is information used to uniquely identify the connection relationship between adjacent supply chain links. By recording this identifier, the specific link connection corresponding to the transfer rate difference data can be clarified.

[0065] Compare the initial transfer rates of adjacent supply chain links in terms of rate differences, and calculate the difference in the initial transfer rates of adjacent links. For example, if the initial transfer rate of a certain raw material supply link is 10 and the initial transfer rate of the adjacent production and manufacturing link is 8, then the transfer rate difference data between these two adjacent links is 2. At the same time, record the supply chain link connection identifier corresponding to the transfer rate difference data, such as "raw material supply - production and manufacturing". By performing such comparisons and records for all adjacent supply chain links, a complete set of transfer rate difference data between adjacent links is generated.

[0066] Step S343: Analyze the fluctuation amplitude of the transfer rate difference data in the time dimension, extract the abnormal fluctuation intervals where the fluctuation amplitude exceeds the preset fluctuation benchmark, and locate the corresponding pairs of supply chain links according to the connection identifiers within the abnormal fluctuation intervals.

[0067] The fluctuation amplitude analysis is to analyze the change of the transfer rate difference data in the time dimension and calculate its fluctuation amplitude. The abnormal fluctuation interval is the time period when the fluctuation amplitude of the transfer rate difference data exceeds the preset fluctuation benchmark. The preset fluctuation benchmark is a preset fluctuation amplitude threshold used to determine whether the fluctuation is abnormal. The pair of supply chain links is composed of a pair of adjacent supply chain links corresponding to the connection identifiers within the abnormal fluctuation interval.

[0068] To analyze the fluctuation range of the convective transfer rate difference data in the time dimension, a moving window method can be adopted. First, set the size of a moving window. For example, take 7 days as a window. Then, calculate the maximum and minimum values of the convective transfer rate difference data within each window, and the difference between the two is the fluctuation range within that window. Compare the fluctuation range of each window with a preset fluctuation benchmark. If it exceeds the preset fluctuation benchmark, mark the time period corresponding to that window as an abnormal fluctuation interval. According to the connection identifier within the abnormal fluctuation interval, locate the corresponding supply chain link pair. For example, if the connection identifier within the abnormal fluctuation interval is "production manufacturing - product distribution", then the corresponding supply chain link pair is the production manufacturing link and the product distribution link.

[0069] Step S344: Identify the supply - demand correlation degree between the material supply response delay duration and the periodic change amplitude of the material demand volume in the supply chain link pair. If the supply - demand correlation degree is lower than the preset correlation benchmark, mark the supply chain link pair as a conduction abnormal path.

[0070] The supply - demand correlation degree is an index that measures the degree of association between the material supply response delay duration and the periodic change amplitude of the material demand volume. It reflects the coordination degree between material supply and demand. The preset correlation benchmark is a preset critical value used to judge whether the supply - demand correlation degree is high enough. The conduction abnormal path refers to the supply chain link pair with a supply - demand correlation degree lower than the preset correlation benchmark, indicating that there may be abnormal conduction of material demand in this link pair.

[0071] To identify the supply - demand correlation degree between the material supply response delay duration and the periodic change amplitude of the material demand volume in the supply chain link pair, a correlation analysis method can be used. For example, calculate the Pearson correlation coefficient between the material supply response delay duration and the periodic change amplitude of the material demand volume, and use this coefficient as an index of the supply - demand correlation degree. If this coefficient is lower than the preset correlation benchmark, mark this supply chain link pair as a conduction abnormal path. For example, if the preset correlation benchmark is 0.5 and the Pearson correlation coefficient between the material supply response delay duration and the periodic change amplitude of the material demand volume in a certain supply chain link pair is 0.3, then mark this supply chain link pair as a conduction abnormal path.

[0072] Step S345: Conduct multi - stage state tracking on the convective transfer rate difference data in the conduction abnormal path to obtain the state change trajectory of the difference data within a preset time span.

[0073] Multi-stage state tracking is a process of continuously monitoring and recording the states of the flow rate difference data in the conduction abnormal path at different stages. The preset time span is a pre-set time range used to determine the observation time period of the state change trajectory. The state change trajectory is a record of the changes of the flow rate difference data within the preset time span, which can reflect the change trends and rules of the data.

[0074] For multi-stage state tracking of the flow rate difference data in the conduction abnormal path, the method of time series analysis can be adopted. First, divide the preset time span into multiple stages in chronological order, for example, divide it by month. Then, record the states of the flow rate difference data within each stage, such as the average value, maximum value, minimum value, etc. By continuously tracking and recording these state data, the state change trajectory of the flow rate difference data within the preset time span is obtained. For example, within a preset time span of 12 months, record the average value of the flow rate difference data in the conduction abnormal path every month to form a state change trajectory containing 12 data points.

[0075] Step S346: Perform piecewise slope fitting based on the state change trajectory to generate a trend conduction curve corresponding to the conduction abnormal path, and smooth the slope mutation points in the trend conduction curve to generate the material trend evolution characteristics; among them, the material trend evolution characteristics include a conduction path stability index and a trend diffusion direction vector. The conduction path stability index is used to quantify the anti-interference ability of the supply and demand fluctuations between supply chain links, and the trend diffusion direction vector is used to indicate the conduction priority order of the material demand changes in the supply chain.

[0076] Piecewise slope fitting is a process of analyzing and modeling the state change trajectory, dividing it into multiple line segments, and calculating the slope of each line segment. The trend conduction curve is a curve generated by piecewise slope fitting, which reflects the change trend of the flow rate difference data in the conduction abnormal path. The slope mutation point is a point where the slope of the trend conduction curve suddenly changes greatly. Smoothing processing is a process of adjusting these mutation points to make the curve smoother and more continuous. The conduction path stability index is an index used to quantify the anti-interference ability of the supply and demand fluctuations between supply chain links, which can be measured by the stability of the trend conduction curve. The trend diffusion direction vector is a vector used to indicate the conduction priority order of the material demand changes in the supply chain, which can be determined by the slope direction and magnitude of the trend conduction curve.

[0077] Based on the piecewise slope fitting of the state change trajectory, first divide the state change trajectory into multiple line segments. The division points of the line segments can be determined by using clustering analysis or breakpoint detection methods. Then, calculate the slope of each line segment and draw a trend conduction curve according to the slope. Smooth the slope mutation points in the trend conduction curve. The moving average method or spline interpolation method can be used. For example, using the moving average method, average several data points around the slope mutation point and replace the value of the mutation point with the average value to make the curve smoother. The generated material trend evolution characteristics include the conduction path stability index and the trend diffusion direction vector. The conduction path stability index can be obtained by calculating the standard deviation or variance of the trend conduction curve. The smaller the standard deviation or variance, the more stable the conduction path and the stronger the anti-interference ability of the supply and demand fluctuations between supply chain links. The trend diffusion direction vector can be determined according to the slope direction and magnitude of the trend conduction curve. A positive slope indicates an upward trend in the material demand change, and the larger the slope, the higher the conduction priority; a negative slope indicates a downward trend in the material demand change, and the larger the absolute value of the slope, the lower the conduction priority.

[0078] Step S400: Based on a preset dynamic analysis model, perform joint trend matching processing on the dynamic characteristics of material demand and the evolution characteristics of material trends to generate a predicted demand distribution result for the material flow sequence.

[0079] The preset dynamic analysis model is a model established in advance for analyzing the trend of material demand changes. It can be a model constructed based on machine learning algorithms, such as a neural network model, a support vector machine model, etc. The joint trend matching processing is a process of matching and analyzing the dynamic characteristics of material demand and the evolution characteristics of material trends with the preset dynamic analysis model to find the associations and rules between them. The predicted demand distribution result is the prediction result of the future demand distribution of the material flow sequence obtained after the joint trend matching processing, which includes information such as the material demand priority sequence of each supply chain link within a specified future period.

[0080] As an implementation method, in step S400, based on a preset dynamic analysis model, perform joint trend matching processing on the dynamic characteristics of material demand and the evolution characteristics of material trends to generate a predicted demand distribution result for the material flow sequence, which can specifically include the following steps S410 to S450: Step S410: Input the dynamic characteristics of material demand into the demand quantification module of the dynamic analysis model to generate a set of demand quantification indicators, which includes the predicted value of the demand peak and the demand trough fluctuation coefficient.

[0081] The demand quantification module is a module in the dynamic analysis model used to quantitatively analyze material requirements. It can convert the dynamic characteristics of material requirements into specific quantification indicators. The set of demand quantification indicators is a set of a series of quantification indicators generated by the demand quantification module. Among them, the predicted demand peak value is the prediction of the maximum value that the material requirements may reach in the future, and the demand trough fluctuation coefficient is an indicator to measure the degree of fluctuation of the material demand trough value.

[0082] Input the dynamic characteristics of material requirements into the demand quantification module of the dynamic analysis model. The demand quantification module can adopt the method of regression analysis for quantitative analysis. For example, using a linear regression model, take the amplitude, time, etc. of the demand mutation nodes in the dynamic characteristics of material requirements as input variables, and take the historical demand peak and trough data as output variables to train the model. When inputting new dynamic characteristics of material requirements, the model makes predictions according to the parameters obtained from training, and generates the predicted demand peak value and the demand trough fluctuation coefficient. Assume that the trained linear regression model is y = a * x1 + b * x2 + c, where x1 and x2 are the amplitude and time of the demand mutation nodes in the dynamic characteristics of material requirements, and a, b, and c are the parameters of the model. After inputting the new dynamic characteristics of material requirements, the calculated y value is the predicted demand peak value, and at the same time, the demand trough fluctuation coefficient is calculated according to the error analysis of the model.

[0083] Step S420: Input the material trend evolution characteristics into the trend conduction module of the dynamic analysis model to generate a set of trend conduction indicators, which includes the trend diffusion rate and the trend decay threshold.

[0084] The trend conduction module is a module in the dynamic analysis model used to analyze the trend conduction of material requirements. It can convert the material trend evolution characteristics into specific trend conduction indicators. The set of trend conduction indicators is a set of a series of indicators generated by the trend conduction module. Among them, the trend diffusion rate refers to the speed at which the change in material requirements spreads in the supply chain, and the trend decay threshold refers to the critical value at which the change in material requirements begins to decay during the propagation process.

[0085] Input the material trend evolution characteristics into the trend conduction module of the dynamic analysis model. The trend conduction module can adopt the method of time series analysis for analysis. For example, using the autoregressive integrated moving average model (ARIMA), take the state change trajectory data in the material trend evolution characteristics as input to predict the trend of the change in material requirements. Calculate the trend diffusion rate and the trend decay threshold according to the prediction results. Assume that the prediction result of the ARIMA model for the change in material requirements is a time series. The trend diffusion rate is obtained by calculating the change rate between adjacent time points, and at the same time, the trend decay threshold is determined according to the change trend of the series.

[0086] Step S430: Calculate the demand transfer matching degree of the material flow sequence between different supply chain links according to the demand quantification index set and the trend conduction index set.

[0087] The demand transfer matching degree is an index to measure the coordination of material demand transfer between different supply chain links, which reflects the matching degree of the demand quantification index and the trend conduction index between different links.

[0088] To calculate the demand transfer matching degree according to the demand quantification index set and the trend conduction index set, the method of weighted summation can be adopted. First, determine the weights of the demand quantification index and the trend conduction index. For example, set the weight of the demand peak prediction value to 0.6, the weight of the demand trough fluctuation coefficient to 0.2, the weight of the trend diffusion rate to 0.1, and the weight of the trend decay threshold to 0.1. Then, for each supply chain link, perform weighted summation on the demand quantification index and the trend conduction index of this link according to the weights to obtain the demand transfer score of this link. Finally, calculate the difference between the demand transfer scores of adjacent supply chain links, and take the absolute value of the difference as the demand transfer matching degree. For example, the demand transfer score of a certain supply chain link is 80, and the demand transfer score of the adjacent link is 70, then the demand transfer matching degree between these two links is |80 - 70| = 10.

[0089] Step S440: If the demand transfer matching degree is lower than the preset matching threshold, adjust the weights of the dynamic characteristics of material demand and the evolution characteristics of material trends to generate an adjusted combined feature set.

[0090] The preset matching threshold is a critical value set in advance to judge whether the demand transfer matching degree is high enough. Weight adjustment is a process of reallocating the weights of the dynamic characteristics of material demand and the evolution characteristics of material trends in the joint analysis. The adjusted combined feature set is a combination of the dynamic characteristics of material demand and the evolution characteristics of material trends obtained after weight adjustment.

[0091] If the demand transfer matching degree is lower than the preset matching threshold, it indicates that the coordination of material demand transfer between different supply chain links is poor, and it is necessary to adjust the weights of the dynamic characteristics of material demand and the evolution characteristics of material trends. An optimization algorithm, such as a genetic algorithm, can be used to determine the optimal weight allocation scheme. The genetic algorithm simulates the biological evolution process and continuously iterates to find the optimal solution. Taking the demand transfer matching degree as the objective function and the weights of the dynamic characteristics of material demand and the evolution characteristics of material trends as variables, through the selection, crossover, and mutation operations of the genetic algorithm, the weights are continuously adjusted until the demand transfer matching degree reaches the preset matching threshold or is close to the optimal value.

[0092] Step S450: Input the adjusted combined feature set into the distribution prediction module of the dynamic analysis model to generate a predicted demand distribution result, where the predicted demand distribution result includes the material demand priority sequence for each supply chain link within a specified future period.

[0093] The distribution prediction module is a module in the dynamic analysis model used to predict the material demand distribution based on the input feature set. The predicted demand distribution result is the prediction of the distribution of material demand for each supply chain link within a specified future period after being processed by the distribution prediction module. The material demand priority sequence is a sequence obtained by ranking each supply chain link according to the importance or urgency of the material demand.

[0094] Input the adjusted combined feature set into the distribution prediction module of the dynamic analysis model. The distribution prediction module can use classification algorithms in machine learning, such as the decision tree algorithm, for prediction. The decision tree algorithm constructs a decision tree model and classifies and predicts the material demand based on the input feature set. Taking the adjusted combined feature set as the input of the decision tree model, the model predicts the material demand for each supply chain link within a specified future period according to the rules obtained from training, and generates a predicted demand distribution result. For example, the decision tree model determines that the material demand for a certain supply chain link is relatively high within the next month based on the input features and ranks it at the forefront of the material demand priority sequence.

[0095] Step S500: Generate a material supply optimization strategy based on the predicted demand distribution result and feedback the material supply optimization strategy to the supply chain management system to trigger material allocation operations.

[0096] The material supply optimization strategy is a strategy formulated based on the predicted demand distribution result for optimizing material supply, which includes decisions on aspects such as inventory transfer ratio and external procurement weight. The supply chain management system is an information system used by enterprises to manage supply chain operations. It can receive the material supply optimization strategy and trigger corresponding material allocation operations, such as inventory transfer and procurement.

[0097] As an implementation, in step S500, generating a material supply optimization strategy based on the predicted demand distribution result and feedbacking the material supply optimization strategy to the supply chain management system to trigger material allocation operations can specifically include the following steps S510 - S550: Step S510: Extract the priority sequence identifier from the predicted demand distribution result and determine the urgency level classification of material supply according to the priority sequence identifier.

[0098] The priority sequence identifier is information used to identify the priority of material requirements in each supply chain link in the predicted demand distribution result. It can be a numerical number, an alphabetic code, etc. The grading of the urgency of material supply is the level divided according to the priority sequence identifier of the material supply. For example, it is divided into three levels: high, medium, and low.

[0099] The priority sequence identifier in the predicted demand distribution result can be directly obtained from the data structure of the predicted demand distribution result. For example, the predicted demand distribution result is stored in tabular form, and one of the columns records the priority sequence identifiers of each supply chain link. To determine the grading of the urgency of material supply according to the priority sequence identifier, the corresponding relationship between the priority sequence identifier and the urgency grading can be preset. For example, the supply chain links with priority sequence identifiers from 1 to 3 correspond to a high urgency grading, those from 4 to 6 correspond to a medium grading, and those from 7 to 10 correspond to a low grading. If the priority sequence identifier of a certain supply chain link is 2, then the urgency grading of the material supply for this link is high.

[0100] Step S520: Based on the urgency grading, configure the corresponding supply resource quota for each supply chain link. The supply resource quota includes the inventory transfer ratio and the external procurement weight.

[0101] The supply resource quota is the quantity and proportion of resources allocated to each supply chain link to meet material requirements. Among them, the inventory transfer ratio refers to the ratio of transferring materials from the enterprise's internal inventory, and the external procurement weight refers to the ratio of purchasing materials from external suppliers.

[0102] Based on the urgency grading, the corresponding supply resource quota can be configured for each supply chain link by using a preset allocation rule. For example, for supply chain links with a high urgency grading, the inventory transfer ratio is set to 70% and the external procurement weight is set to 30%; for supply chain links with a medium urgency grading, the inventory transfer ratio is set to 50% and the external procurement weight is set to 50%; for supply chain links with a low urgency grading, the inventory transfer ratio is set to 30% and the external procurement weight is set to 70%. According to the urgency grading of each supply chain link, configure its supply resource quota according to the above rules.

[0103] Step S530: Generate an initial supply strategy according to the inventory transfer ratio and the external procurement weight, and perform resource conflict detection on the initial supply strategy.

[0104] The initial supply strategy is a preliminary material supply plan formulated according to the inventory transfer ratio and the external procurement weight. It includes information such as the quantity of inventory transfer and the quantity of external procurement. Resource conflict detection is to check the initial supply strategy to determine whether there are situations of resource shortage or conflict, such as insufficient inventory to meet the transfer demand, and external suppliers being unable to provide sufficient materials, etc.

[0105] Generate an initial supply strategy based on the inventory transfer ratio and the external procurement weight. First, calculate the total material requirements for each supply chain link. For example, the total material requirements for a certain supply chain link are 100 pieces. Then, calculate the inventory transfer quantity and the external procurement quantity according to the inventory transfer ratio and the external procurement weight. If the inventory transfer ratio is 60% and the external procurement weight is 40%, then the inventory transfer quantity is 100 * 60% = 60 pieces, and the external procurement quantity is 100 * 40% = 40 pieces. After generating the initial supply strategy, perform a resource conflict detection on it.

[0106] As an implementation manner, in step S530, performing a resource conflict detection on the initial supply strategy may specifically include the following steps S531 to S537: Step S531: Obtain the real-time inventory status data of the current supply chain link. The real-time inventory status data includes the available inventory quantity and the in-transit inventory quantity.

[0107] The real-time inventory status data is data reflecting the actual inventory situation of the current supply chain link. The available inventory quantity refers to the inventory quantity that can be immediately used for transfer currently, and the in-transit inventory quantity refers to the inventory quantity that has been ordered but not yet arrived.

[0108] Obtaining the real-time inventory status data of the current supply chain link can be achieved by interacting with the enterprise's inventory management system. The inventory management system records the inventory situation of each supply chain link in real time. By calling the data interface of the inventory management system, the corresponding real-time inventory status data is obtained according to the identifier of the supply chain link. For example, for a certain production and manufacturing link, the available inventory quantity obtained through the data interface is 80 pieces, and the in-transit inventory quantity is 20 pieces.

[0109] Step S532: Calculate the target transfer quantity according to the inventory transfer ratio, and determine whether the target transfer quantity exceeds the sum of the available inventory quantity and the in-transit inventory quantity.

[0110] The target transfer quantity is the quantity of materials that need to be transferred from the inventory calculated according to the inventory transfer ratio. Determining whether the target transfer quantity exceeds the sum of the available inventory quantity and the in-transit inventory quantity can determine whether there is a situation of insufficient inventory.

[0111] Calculate the target transfer quantity according to the inventory transfer ratio, which can be obtained by multiplying the total material requirement by the inventory transfer ratio. For example, if the total material requirement is 100 pieces and the inventory transfer ratio is 60%, then the target transfer quantity is 100 * 60% = 60 pieces. Compare the target transfer quantity with the sum of the available inventory and the in-transit inventory. If the available inventory is 80 pieces and the in-transit inventory is 20 pieces, and the sum is 100 pieces, and the target transfer quantity of 60 pieces is less than the sum, it means the inventory can meet the transfer requirement; if the target transfer quantity is 120 pieces, which is greater than the sum, it means there is a situation of insufficient inventory.

[0112] Step S533: If it exceeds, determine that there is a first type of resource conflict and generate a first conflict identifier.

[0113] The first type of resource conflict refers to the conflict situation where the inventory transfer quantity exceeds the sum of the available inventory and the in-transit inventory. The first conflict identifier is information used to mark the existence of the first type of resource conflict, which can be a code or a mark.

[0114] If the target transfer quantity exceeds the sum of the available inventory and the in-transit inventory, determine that there is a first type of resource conflict. For example, if the target transfer quantity is 120 pieces and the sum of the available inventory and the in-transit inventory is 100 pieces, there is a first type of resource conflict at this time. Generate a first conflict identifier, such as "C1", indicating the existence of the first type of resource conflict.

[0115] Step S534: Obtain the real-time supply capacity data of external suppliers. The real-time supply capacity data includes the maximum committed quantity and the minimum supply cycle.

[0116] The real-time supply capacity data is data reflecting the current actual supply capacity of external suppliers. Among them, the maximum committed quantity refers to the maximum quantity of materials that the supplier can provide within a certain period of time, and the minimum supply cycle refers to the shortest time required for the supplier to provide materials from receiving the order.

[0117] To obtain the real-time supply capacity data of external suppliers, it can be achieved by docking data with the supplier's information system. The supplier's information system records its supply capacity situation in real time, and obtains data such as the maximum committed quantity and the minimum supply cycle of the supplier through a secure data transmission channel. For example, after docking with a certain supplier, the maximum committed quantity obtained is 50 pieces and the minimum supply cycle is 5 days.

[0118] Step S535: Calculate the target purchase quantity according to the external procurement weight, and judge whether the target purchase quantity exceeds the maximum committed quantity or whether the minimum supply cycle is later than the demand time window.

[0119] The target purchase quantity is the quantity of materials that need to be purchased from external suppliers calculated based on the external purchase weight. The demand time window refers to the time range within which the material demand must be met. By determining whether the target purchase quantity exceeds the maximum committed quantity or whether the minimum supply cycle is later than the demand time window, it can be determined whether there is a shortage of external procurement resources or a supply delay.

[0120] The target purchase quantity is calculated based on the external purchase weight, which can be obtained by multiplying the total material demand by the external purchase weight. For example, if the total material demand is 100 pieces and the external purchase weight is 40%, then the target purchase quantity is 100 * 40% = 40 pieces. Compare the target purchase quantity with the maximum committed quantity, and at the same time compare the minimum supply cycle with the demand time window. If the maximum committed quantity is 50 pieces, the target purchase quantity of 40 pieces is less than the maximum committed quantity, and the minimum supply cycle is 5 days and the demand time window is 7 days, it means that external procurement can meet the demand; if the target purchase quantity is 60 pieces, which is greater than the maximum committed quantity, or the minimum supply cycle is 10 days, which is later than the demand time window of 7 days, it means that there is a shortage of external procurement resources or a supply delay.

[0121] Step S536: If it exceeds or is later than, determine that there is a type-II resource conflict and generate a second conflict identifier.

[0122] The type-II resource conflict refers to the conflict situation where the target purchase quantity exceeds the maximum committed quantity or the minimum supply cycle is later than the demand time window. The second conflict identifier is information used to mark the existence of a type-II resource conflict, which can be a code or a mark.

[0123] If the target purchase quantity exceeds the maximum committed quantity or the minimum supply cycle is later than the demand time window, determine that there is a type-II resource conflict. For example, if the target purchase quantity is 60 pieces and the maximum committed quantity is 50 pieces, there is a type-II resource conflict at this time; or if the minimum supply cycle is 10 days and the demand time window is 7 days, there is also a type-II resource conflict. Generate a second conflict identifier, such as "C2", indicating the existence of a type-II resource conflict.

[0124] Step S537: Determine the priority order of conflict resolution according to the type combination of the first conflict identifier and the second conflict identifier.

[0125] The priority order of conflict resolution is the order of handling conflicts determined according to the type combination of the first conflict identifier and the second conflict identifier, which can help enterprises reasonably arrange resources and resolve important conflicts first.

[0126] Based on the type combination of the first conflict identifier and the second conflict identifier, determine the priority order of conflict resolution, and the priority rules corresponding to different type combinations can be preset in advance. For example, if there is only the first type of resource conflict (only the first conflict identifier "C1"), then prioritize resolving the problem of insufficient inventory; if there is only the second type of resource conflict (only the second conflict identifier "C2"), then prioritize resolving the problem of insufficient external procurement resources or supply delay; if there are both the first type and the second type of resource conflicts (both "C1" and "C2"), then first try to resolve the problem of insufficient inventory, and then resolve the problem of external procurement. According to the specific type combination of conflict identifiers, determine the priority order of conflict resolution according to the above rules.

[0127] Step S540: If a resource conflict is detected in the initial supply strategy, then dynamically adjust the inventory transfer ratio and the external procurement weight based on the conflict type, and generate an optimized material supply strategy after conflict resolution.

[0128] Dynamic adjustment refers to the process of real-time adjustment of the inventory transfer ratio and the external procurement weight according to the type and degree of resource conflict. The optimized material supply strategy after conflict resolution is a material supply strategy that can solve the resource conflict obtained after dynamic adjustment.

[0129] As an implementation method, in step S540, dynamically adjust the inventory transfer ratio and the external procurement weight based on the conflict type, and generate an optimized material supply strategy after conflict resolution, which can specifically include the following steps S541~S545: Step S541: If there is the first type of resource conflict, then reduce the inventory transfer ratio, and recalculate the target transfer volume according to the reduced inventory transfer ratio until the target transfer volume does not exceed the sum of the available inventory and the in-transit inventory.

[0130] If it is detected that there is the first type of resource conflict, it means that the inventory transfer volume exceeds the sum of the available inventory and the in-transit inventory, and the inventory transfer ratio needs to be reduced. The inventory transfer ratio can be reduced gradually, for example, by 10% each time. After reducing the inventory transfer ratio, recalculate the target transfer volume according to the new ratio. For example, the original inventory transfer ratio is 60%, the target transfer volume is 120 pieces, and the sum of the available inventory and the in-transit inventory is 100 pieces, there is the first type of resource conflict. Reduce the inventory transfer ratio to 50%, then the new target transfer volume is 100*50% = 50 pieces, which is less than the sum and meets the requirements. Keep repeating this process until the target transfer volume does not exceed the sum of the available inventory and the in-transit inventory.

[0131] Step S542: If there is a second - type resource conflict, reduce the external procurement weight, and recalculate the target procurement quantity based on the reduced external procurement weight until the target procurement quantity does not exceed the maximum committed quantity and the minimum supply cycle is earlier than the demand time window.

[0132] If it is detected that there is a second - type resource conflict, it means that the target procurement quantity exceeds the maximum committed quantity or the minimum supply cycle is later than the demand time window, and the external procurement weight needs to be reduced. The external procurement weight can be reduced gradually, for example, by 10% each time. After reducing the external procurement weight, recalculate the target procurement quantity according to the new weight. For example, originally the external procurement weight was 40% and the target procurement quantity was 60 pieces, and the maximum committed quantity was 50 pieces, so there was a second - type resource conflict. Reduce the external procurement weight to 30%, then the new target procurement quantity is 100 * 30% = 30 pieces, which is less than the maximum committed quantity and meets the requirements. At the same time, check whether the minimum supply cycle is earlier than the demand time window. If not, continue to reduce the external procurement weight until the target procurement quantity does not exceed the maximum committed quantity and the minimum supply cycle is earlier than the demand time window.

[0133] Step S543: If there are both first - type resource conflicts and second - type resource conflicts, first reduce the inventory transfer ratio, and after the target transfer quantity corresponding to the reduced inventory transfer ratio stabilizes, then reduce the external procurement weight.

[0134] If there are both first - type resource conflicts and second - type resource conflicts, according to the priority order of conflict resolution, first reduce the inventory transfer ratio. Reduce the inventory transfer ratio according to the method in Step S541 until the target transfer quantity does not exceed the sum of the available inventory and the in - transit inventory. After the target transfer quantity stabilizes, then reduce the external procurement weight according to the method in Step S542 until the target procurement quantity does not exceed the maximum committed quantity and the minimum supply cycle is earlier than the demand time window. For example, when there are both first - type and second - type resource conflicts, first reduce the inventory transfer ratio from 60% to 50% to make the target transfer quantity meet the requirements. Then, reduce the external procurement weight from 40% to 30% to make the target procurement quantity and the supply cycle meet the requirements.

[0135] Step S544: Regenerate the supply resource quota according to the adjusted inventory transfer ratio and external procurement weight, and update the initial supply strategy based on the regenerated supply resource quota.

[0136] Recalculate the inventory transfer quantity and external purchase quantity according to the adjusted inventory transfer ratio and external purchase weight to obtain the regenerated supply resource quota. For example, if the adjusted inventory transfer ratio is 50% and the external purchase weight is 30%, and the total material demand is 100 pieces, then the inventory transfer quantity is 100 * 50% = 50 pieces, and the external purchase quantity is 100 * 30% = 30 pieces. Update the initial supply strategy based on the regenerated supply resource quota, and update information such as the new inventory transfer quantity and external purchase quantity into the initial supply strategy.

[0137] Step S545: Use the updated initial supply strategy as the optimized material supply strategy after conflict resolution.

[0138] Use the updated initial supply strategy as the optimized material supply strategy after conflict resolution. This strategy can solve the resource conflict problem in the initial supply strategy and achieve the optimization of material supply.

[0139] Step S550: Encode the inventory transfer instructions and purchase instructions in the optimized material supply strategy into executable operation codes, and transmit the executable operation codes to the strategy execution interface of the supply chain management system.

[0140] The inventory transfer instruction is a specific instruction regarding inventory transfer in the optimized material supply strategy, including information such as the type, quantity, source, and destination of the transferred materials. The purchase instruction is a specific instruction regarding external purchase, including information such as the type, quantity, and supplier of the purchased materials. The executable operation code is the code that converts the inventory transfer instruction and purchase instruction into a code that can be recognized and executed by a computer, and it can realize the automatic control of material allocation operations. The strategy execution interface is the interface in the supply chain management system for receiving and executing the optimized material supply strategy.

[0141] As an implementation manner, the method provided by the embodiment of the present invention may further include the following steps S600 - S900: Step S600: Real-time monitor the strategy execution status of the supply chain management system and obtain the actual execution effect data of the material allocation operation.

[0142] Real-time monitoring is a process of continuously tracking and monitoring the execution of the optimized material supply strategy in the supply chain management system. The strategy execution status refers to the execution progress and results of the optimized material supply strategy in the supply chain management system, such as whether the inventory transfer is completed, whether the purchase has arrived, etc. The actual execution effect data is the data reflecting the actual effect generated by the material allocation operation, including the actual demand satisfaction rate, the actual supply delay duration, etc.

[0143] The strategy execution status of the real-time monitoring supply chain management system can be achieved by interacting with the log system of the supply chain management system. The log system of the supply chain management system records the execution process and results of each material allocation operation. By regularly reading the log data, the strategy execution status information can be obtained. For example, read the log data every 1 hour to check whether the inventory transfer operation is completed and whether the purchase order has been shipped. To obtain the actual execution effect data of the material allocation operation, relevant information can be extracted from the business data of the supply chain management system. For example, obtain the actual sales quantity from the sales order system and calculate the actual demand satisfaction rate; obtain the material arrival time from the logistics management system and calculate the actual supply delay duration.

[0144] Step S700: Conduct a deviation analysis on the actual execution effect data and the predicted demand distribution result to generate a deviation analysis report.

[0145] Deviation analysis is a process of comparing and analyzing the actual execution effect data and the predicted demand distribution result to find the differences between the two. The deviation analysis report is a document that summarizes and records the deviation analysis results, which includes information such as deviation types and deviation impact levels.

[0146] As an implementation method, in step S700, conducting a deviation analysis on the actual execution effect data and the predicted demand distribution result to generate a deviation analysis report can specifically include the following steps S710 to S750: Step S710: Extract the actual demand satisfaction rate and the actual supply delay duration from the actual execution effect data.

[0147] The actual demand satisfaction rate refers to the ratio of the actual quantity of materials provided to the actual demand quantity, which reflects the degree of satisfaction of material supply for demand. The actual supply delay duration refers to the time interval from the occurrence of the material demand event to the completion of the actual supply event, which reflects the timeliness of material supply.

[0148] To extract the actual demand satisfaction rate and the actual supply delay duration from the actual execution effect data, statistical calculations can be performed from relevant business data. For example, obtain the actual sales quantity and the order demand quantity from the sales order system and calculate the actual demand satisfaction rate; obtain the material demand time and the arrival time from the logistics management system and calculate the actual supply delay duration.

[0149] Step S720: Extract the predicted demand satisfaction rate range and the predicted supply delay threshold from the predicted demand distribution result.

[0150] The predicted demand satisfaction rate range is the proportion interval in which the estimated material demand in the predicted demand distribution result can be satisfied, which reflects an expected range of future demand satisfaction. The predicted supply delay threshold is a pre-set time limit used to measure the maximum allowable duration of the predicted material supply delay.

[0151] The predicted demand satisfaction rate range and the predicted supply delay threshold in the predicted demand distribution result can be directly obtained from the data structure of the predicted demand distribution result. The predicted demand distribution result is usually stored in a structured data form, such as a database table or a data file. For example, in a database table storing the predicted demand distribution result, there are dedicated fields to record the predicted demand satisfaction rate range (such as 85%-95%) and the predicted supply delay threshold (such as 5 days). By querying this database table, the corresponding data can be accurately extracted.

[0152] Step S730: Calculate the first deviation amount between the actual demand satisfaction rate and the median of the predicted demand satisfaction rate range, and the second deviation amount between the actual supply delay duration and the predicted supply delay threshold.

[0153] The first deviation amount is the difference between the actual demand satisfaction rate and the median of the predicted demand satisfaction rate range, which reflects the degree of difference in proportion between the actual demand satisfaction situation and the predicted situation. The second deviation amount is the difference between the actual supply delay duration and the predicted supply delay threshold, which reflects the degree of difference in time between the actual supply delay situation and the predicted situation.

[0154] To calculate the first deviation amount, first calculate the median of the predicted demand satisfaction rate range. Assume the predicted demand satisfaction rate range is a%-b%, then the median is (a + b) / 2 %. Then subtract this median from the actual demand satisfaction rate to get the first deviation amount. For example, if the predicted demand satisfaction rate range is 85%-95%, the median is (85 + 95) / 2 = 90%, and the actual demand satisfaction rate is 80%, then the first deviation amount is 80% - 90% = -10%. To calculate the second deviation amount, directly subtract the predicted supply delay threshold from the actual supply delay duration.

[0155] Step S740: Determine the deviation coefficient according to the weighted sum of the first deviation amount and the second deviation amount.

[0156] The deviation coefficient is an indicator that comprehensively considers the influence degree of the first deviation amount and the second deviation amount on the overall deviation. To accurately measure the overall deviation, different weights need to be assigned to the first deviation amount and the second deviation amount, and the weights can be determined according to the actual business situation and the importance of each indicator.

[0157] Determine the deviation coefficient based on the weighted sum of the first deviation amount and the second deviation amount. First, set the weight of the first deviation amount as w1 and the weight of the second deviation amount as w2, and w1 + w2 = 1. Then the formula for calculating the deviation coefficient is: Deviation coefficient = First deviation amount × w1 + Second deviation amount × w2. For example, set w1 = 0.6, w2 = 0.4, the first deviation amount is -10% (converted to a decimal -0.1), and the second deviation amount is 2 days. Assume that the number of days is also normalized to a decimal (e.g., based on 10 days, 2 days is normalized to 0.2), then the deviation coefficient = -0.1 × 0.6 + 0.2 × 0.4 = -0.06 + 0.08 = 0.02.

[0158] Step S750: If the deviation coefficient exceeds the deviation threshold, generate a deviation analysis report including the deviation type and the deviation impact level; among them, the deviation type includes demand forecasting deviation and supply response deviation, and the deviation impact level classifies the material shortage risk of the supply chain link according to the weighted sum.

[0159] The deviation threshold is a preset critical value used to determine whether the deviation coefficient exceeds the acceptable range. When the deviation coefficient exceeds the deviation threshold, it indicates that there is a large difference between the actual execution situation and the predicted situation, and in-depth analysis is required and a deviation analysis report is generated. The deviation type is divided into demand forecasting deviation and supply response deviation. The demand forecasting deviation mainly refers to the difference between the actual demand satisfaction situation and the predicted demand satisfaction situation, while the supply response deviation focuses on the difference between the actual supply delay situation and the predicted supply delay situation. The deviation impact level is a classification of the material shortage risk of the supply chain link according to the weighted sum, such as high, medium, and low levels, to evaluate the impact degree of the deviation on the supply chain.

[0160] If the deviation coefficient exceeds the deviation threshold, determine the deviation type according to the specific situations of the first deviation amount and the second deviation amount. If the first deviation amount is large, it indicates that there is mainly a demand forecasting deviation; if the second deviation amount is large, then it is mainly a supply response deviation. For the deviation impact level, it can be divided according to the size of the deviation coefficient. For example, when the deviation coefficient is between 0 - 0.05, the deviation impact level is low; between 0.05 - 0.1, it is medium; greater than 0.1, it is high. Assume that the deviation coefficient is 0.15, exceeding the deviation threshold of 0.1, and the first deviation amount is relatively small while the second deviation amount is large, then the deviation type is supply response deviation and the deviation impact level is high. Generate a deviation analysis report based on this information. The report content details the deviation type, the deviation impact level, and the relevant data analysis results, providing a basis for subsequent adjustments.

[0161] Step S800: If the deviation coefficient in the deviation analysis report exceeds the preset deviation threshold, trigger the parameter correction operation for the dynamic analysis model; wherein, the parameter correction operation includes: adjusting the weight allocation ratio of the demand quantification module and the trend conduction module in the dynamic analysis model according to the deviation type in the actual execution effect data.

[0162] The preset deviation threshold is a pre-set standard value used to determine whether the deviation has reached the level where the dynamic analysis model needs to be adjusted. When the deviation coefficient exceeds this threshold, it indicates that there is a large deviation between the prediction result of the dynamic analysis model and the actual situation, and the model needs to be corrected. The parameter correction operation is a process of adjusting the parameters in the dynamic analysis model. Specifically, according to the deviation type in the actual execution effect data, the weight allocation ratio of the demand quantification module and the trend conduction module is adjusted to improve the prediction accuracy of the model.

[0163] If the deviation coefficient in the deviation analysis report exceeds the preset deviation threshold, make corresponding weight adjustments according to the deviation type. If the deviation type is demand prediction deviation, it means that the prediction effect of the demand quantification module is not good. The weight of the demand quantification module can be appropriately increased, and the weight of the trend conduction module can be reduced. For example, the original weight of the demand quantification module is 0.6, and the weight of the trend conduction module is 0.4. Now, the weight of the demand quantification module is increased to 0.7, and the weight of the trend conduction module is reduced to 0.3. If the deviation type is supply response deviation, it means that the prediction ability of the trend conduction module is insufficient. The weight of the trend conduction module can be increased, and the weight of the demand quantification module can be reduced. For example, the weight of the trend conduction module is increased from 0.4 to 0.5, and the weight of the demand quantification module is reduced from 0.6 to 0.5. Through such weight adjustments, the dynamic analysis model can better adapt to the actual situation and improve the prediction accuracy.

[0164] Step S900: After the corrected dynamic analysis model regenerates the updated predicted demand distribution result, iteratively optimize the material supply optimization strategy based on the updated predicted demand distribution result.

[0165] The corrected dynamic analysis model is the model obtained after the parameter correction operation, which can more accurately reflect the changing trend of material demand. The updated predicted demand distribution result is the prediction of the future distribution of material demand regenerated by the corrected dynamic analysis model. Iteratively optimizing the material supply optimization strategy refers to the process of continuously adjusting and improving the existing material supply optimization strategy according to the updated predicted demand distribution result to ensure that the material supply can better meet the actual demand.

[0166] After the updated predicted demand distribution result is regenerated by the corrected dynamic analysis model, again according to the method of step S500 and its sub-steps, a new material supply optimization strategy is generated based on the updated predicted demand distribution result. During the generation process, the urgency level classification of material supply, supply resource quota, etc. are re-determined. For example, if the updated predicted demand distribution result shows an increase in the material demand of a certain supply chain link, then in the new material supply optimization strategy, the supply resource quota of this link is increased, and the inventory transfer ratio or the weight of external procurement is increased. Through continuous iterative optimization, the material supply optimization strategy can adapt to the changes in material demand in a timely manner, improving the operation efficiency and stability of the supply chain.

[0167] As an implementation manner, the method provided by the embodiment of the present invention may further include the following steps S1000 to S1300: Step S1000: Configure a data life cycle identifier for each material flow sequence, and the data life cycle identifier is used to indicate the data valid period of the material flow sequence.

[0168] The data life cycle identifier is an identification information assigned to each material flow sequence, which clarifies the valid range of the data of this material flow sequence in terms of time. The data valid period refers to the time period during which the data of the material flow sequence has reference value and availability. Beyond this time period, the data may lose its accuracy and effectiveness due to reasons such as market changes and business adjustments.

[0169] Configuring a data life cycle identifier for each material flow sequence can determine the data valid period according to the characteristics of the material flow sequence and business requirements. For some material flow sequences with high stability and small market changes, the data valid period can be set longer, such as one year or two years; for some material flow sequences that are greatly affected by market fluctuations, the data valid period can be set shorter, such as one quarter or half a year. For example, for the material flow sequence of a certain type of commonly used standard parts, its market demand is relatively stable, and the data valid period can be set to one year; while for the material flow sequence of a certain type of fashionable electronic products, due to the rapid market replacement, the data valid period is set to one quarter. Add a "data valid period" field to each material flow sequence to record its start time and end time as the data life cycle identifier.

[0170] Step S1100: When it is detected that the current time exceeds the data valid period, mark the corresponding material flow sequence as expired data, and migrate the expired data to the archived storage space.

[0171] The current time refers to the current timestamp of the system, which is used to compare with the valid time period of the data in the data life cycle identifier. Expired data refers to the material flow sequence data whose current time exceeds its valid time period. Although these data may still have certain historical reference value, they are no longer applicable in the current analysis and prediction. The archival storage space is a storage area specifically used to store expired data. It can be storage devices such as disk arrays and tape libraries within the enterprise, or storage services provided by cloud storage service providers.

[0172] When it is detected that the current time exceeds the valid time period of the data, the system will automatically identify the corresponding material flow sequence as expired data. A scheduled task program can be written to regularly (such as daily) check the relationship between the valid time period of all material flow sequence data and the current time. When it is found that the end time of a certain material flow sequence is earlier than the current time, mark this sequence as expired data. For example, add an "expired flag" field to the record of this sequence in the database table and set its value to "yes". Then, migrate the expired data from the original business database to the archival storage space. A data migration tool can be used to copy the expired data to the archival storage space according to the preset rules and delete these data in the original database to save storage space.

[0173] Step S1200: When generating the predicted demand distribution result, exclude the expired data in the archival storage space and perform feature extraction and trend matching processing only based on the unexpired material flow sequences.

[0174] In the process of generating the predicted demand distribution result, in order to ensure the accuracy and effectiveness of the prediction, it is necessary to exclude the expired data in the archival storage space. Because expired data may no longer be able to reflect the current market situation and business requirements, using these data for analysis and prediction will introduce errors. Performing feature extraction and trend matching processing only based on the unexpired material flow sequences can make the prediction result more in line with the actual situation.

[0175] When generating the predicted demand distribution result, the system will automatically filter out the unexpired material flow sequences. In the data query stage, filter out the material flow sequence data with an "expired flag" through the query conditions. Then, perform feature extraction and trend matching processing on the filtered unexpired material flow sequences. According to the method of step S300 and its sub-steps, extract the dynamic characteristics of material demand and the evolution characteristics of material trends, and perform joint trend matching based on these characteristics to generate the predicted demand distribution result. For example, in the feature extraction process, only analyze the time series fluctuation data of the unexpired material flow sequences to determine the demand mutation nodes and supply bottleneck nodes, so as to obtain accurate dynamic characteristics of material demand.

[0176] Step S1300: Regularly perform statistical analysis on the expired data in the archival storage space, generate a historical trend summary report, and input the historical trend summary report into a dynamic analysis model for long-term trend learning.

[0177] Regularly performing statistical analysis on the expired data in the archival storage space can uncover the hidden historical trends and patterns in the data. The historical trend summary report is a summary and refinement of the results of the statistical analysis of the expired data, which contains information such as the change trend and periodic patterns of material requirements over a certain period in the past. Inputting the historical trend summary report into a dynamic analysis model for long-term trend learning enables the model to absorb the experience and patterns in the historical data and improve its prediction ability for future material requirements.

[0178] Regularly (such as annually) performing statistical analysis on the expired data in the archival storage space can use data analysis tools and algorithms. For example, using time series analysis algorithms to analyze the material requirements in the expired data to find out its periodic change patterns and long-term trends. Generate a historical trend summary report based on the analysis results, and the report content includes the growth trend of material requirements, seasonal fluctuations, etc. Input the historical trend summary report into the dynamic analysis model in a suitable data format (such as CSV file, JSON file, etc.). After receiving the historical trend summary report, the dynamic analysis model combines the information in it with the current training data to adjust and optimize the parameters of the model, thereby achieving long-term trend learning. For example, based on the demand growth pattern of a certain material in a specific season every year found in the historical trend summary report, the model adjusts its prediction parameters for the material demand in that season in the future to improve the prediction accuracy.

[0179] In addition, it can be understood that in the above introductions of each embodiment of the present invention, various algorithms involved can be obtained from relevant contents in the prior art. For the sake of saving space, they are not elaborated in detail in the embodiments of the present invention. In addition, those skilled in the art can make detailed supplements according to the common general knowledge in the art when implementing the solution of the present invention. For example, according to the general knowledge in the art, normalization can be used to eliminate the dimension conflict before feature fusion, interpolation can be used to eliminate the dimension difference, historical data, experience or business scenario requirements can be combined to reasonably set the threshold, and the model can be trained based on the general model training method, etc. The present invention will no longer give redundant introductions to the overly detailed implementation processes.

[0180] An embodiment of the present invention provides a material demand analysis and prediction device, including a memory and a processor. The memory stores a computer program that can run on the processor, and when the processor executes the program, it implements some or all of the steps in the above method.

[0181] An embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, some or all of the steps in the above method are implemented. The computer-readable storage medium may be transient or non-transient.

[0182] Figure 2 It is a schematic diagram of the hardware entity of the material requirement analysis and prediction device provided by the embodiment of the present invention. As Figure 2 shown, the hardware entity of the material requirement analysis and prediction device 1000 includes: a processor 1001 and a memory 1002. Among them, the memory 1002 stores a computer program that can run on the processor 1001, and when the processor 1001 executes the program, the steps in the method of any of the above embodiments are implemented.

[0183] The memory 1002 stores a computer program that can run on the processor. The memory 1002 is configured to store instructions and applications executable by the processor 1001, and can also cache data to be processed or already processed by the processor 1001 and each module in the material requirement analysis and prediction device 1000 (for example, image data, audio data, voice communication data, and video communication data), and can be implemented by flash memory (FLASH) or random access memory (Random Access Memory, RAM).

[0184] When the processor 1001 executes the program, the steps of the material requirement analysis and prediction method based on big data in any of the above are implemented. The processor 1001 generally controls the overall operation of the material requirement analysis and prediction device 1000.

[0185] An embodiment of the present invention provides a computer storage medium. The computer storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of the material requirement analysis and prediction method based on big data in any of the above embodiments.

[0186] As described above, it is only the embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should be covered by the protection scope of the present invention.

Claims

1. A method for analyzing and predicting material requirements based on big data, characterized in that, Including the following steps: Obtain a multi-source historical material data set of the supply chain link where the target enterprise is located. The multi-source historical material data set includes multiple material transfer sequences, and each material transfer sequence is composed of at least one material demand event and a corresponding material supply event; Perform cross-link integration processing on the multi-source historical material data set to obtain a cross-link integration data set, which is used to reflect the change in the transfer state of materials between different supply chain links; Perform demand feature extraction processing on the cross-link integration data set to obtain the material demand dynamic feature and the material trend evolution feature of each material transfer sequence; Based on a preset dynamic analysis model, perform joint trend matching processing on the material demand dynamic feature and the material trend evolution feature to generate a predicted demand distribution result for the material transfer sequence; Generate a material supply optimization strategy according to the predicted demand distribution result, and feedback the material supply optimization strategy to the supply chain management system to trigger a material allocation operation.

2. The method according to claim 1, wherein The performing cross-link integration processing on the multi-source historical material data set to obtain a cross-link integration data set includes: Obtain the original data source identifier of each material transfer sequence in the multi-source historical material data set, and determine the supply chain link type corresponding to the material transfer sequence according to the original data source identifier; For each supply chain link type, extract the abnormal data segment in the material transfer sequence. The abnormal data segment includes a conflict data segment with a time sequence conflict between the material demand event and the material supply event; Perform conflict resolution processing on the abnormal data segment to generate a conflict resolution result, and perform data correction on the abnormal data segment according to the conflict resolution result to obtain a corrected material transfer sequence; Align the corrected material transfer sequence with other material transfer sequences that do not contain abnormal data segments to generate an aligned cross-link data set, and use the aligned cross-link data set as the cross-link integration data set; wherein, the data alignment includes associating the material transfer sequences of different supply chain links according to the material identifier, and unifying the time stamp format and measurement unit in the material transfer sequence.

3. The method according to claim 2, wherein The performing demand feature extraction processing on the cross-link integration data set to obtain the material demand dynamic feature and the material trend evolution feature of each material transfer sequence includes: Extract the time series fluctuation data of each material transfer sequence from the cross-link integration data set. The time series fluctuation data includes the periodic change amplitude of the material demand quantity and the material supply response delay duration; Perform pattern recognition processing on the time series fluctuation data to determine the demand mutation node and the supply bottleneck node in the material transfer sequence; Based on the demand mutation node and the supply bottleneck node, generate the material demand dynamic feature, which is used to characterize the non-uniform distribution state of the material demand in the time dimension; Extract the transfer rate difference data of the material transfer sequence between different supply chain links, and perform trend fitting on the transfer rate difference data to generate the material trend evolution characteristics; the material trend evolution characteristics are used to reflect the conduction delay effect and cumulative influence weight of material demand between different supply chain links.

4. The method according to claim 3, characterized in that, Based on the preset dynamic analysis model, perform joint trend matching on the material demand dynamic characteristics and the material trend evolution characteristics to generate the predicted demand distribution result of the material transfer sequence, including: Input the material demand dynamic characteristics into the demand quantification module of the dynamic analysis model to generate a set of demand quantification indicators, and the set of demand quantification indicators includes the predicted value of the demand peak and the demand valley fluctuation coefficient; Input the material trend evolution characteristics into the trend conduction module of the dynamic analysis model to generate a set of trend conduction indicators, and the set of trend conduction indicators includes the trend diffusion rate and the trend decay threshold; According to the set of demand quantification indicators and the set of trend conduction indicators, calculate the demand conduction matching degree of the material transfer sequence between different supply chain links; If the demand conduction matching degree is lower than the preset matching threshold, adjust the weights of the material demand dynamic characteristics and the material trend evolution characteristics to generate an adjusted set of joint characteristics; Input the adjusted set of joint characteristics into the distribution prediction module of the dynamic analysis model to generate the predicted demand distribution result, and the predicted demand distribution result includes the material demand priority sequence of each supply chain link within a specified future period.

5. The method according to claim 4, wherein Generate a material supply optimization strategy according to the predicted demand distribution result, and feedback the material supply optimization strategy to the supply chain management system to trigger material allocation operations, including: Extract the priority sequence identifier in the predicted demand distribution result, and determine the urgency level classification of material supply according to the priority sequence identifier; Based on the urgency level classification, allocate corresponding supply resource quotas for each supply chain link, and the supply resource quotas include inventory transfer ratio and external procurement weight; According to the inventory transfer ratio and the external procurement weight, generate an initial supply strategy and perform resource conflict detection on the initial supply strategy; If resource conflicts are detected in the initial supply strategy, dynamically adjust the inventory transfer ratio and the external procurement weight based on the conflict type to generate a material supply optimization strategy after conflict resolution; Encode the inventory transfer instructions and procurement instructions in the material supply optimization strategy into executable operation codes, and transmit the executable operation codes to the strategy execution interface of the supply chain management system.

6. The method according to claim 5, wherein The resource conflict detection of the initial supply strategy includes: Obtain the real-time inventory status data of the current supply chain link, and the real-time inventory status data includes the available inventory quantity and the in-transit inventory quantity; Calculate the target transfer quantity according to the inventory transfer ratio, and determine whether the target transfer quantity exceeds the sum of the available inventory quantity and the in-transit inventory quantity; If it exceeds, it is determined that there is a first - type resource conflict, and a first conflict identifier is generated; Obtain the real - time supply capacity data of external suppliers, where the real - time supply capacity data includes the maximum committed quantity and the minimum supply cycle; Calculate the target purchase quantity according to the external purchase weight, and determine whether the target purchase quantity exceeds the maximum committed quantity or whether the minimum supply cycle is later than the demand time window; If it exceeds or is later than, it is determined that there is a second - type resource conflict, and a second conflict identifier is generated; Determine the priority order of conflict resolution according to the type combination of the first conflict identifier and the second conflict identifier.

7. The method according to claim 6, wherein The dynamic adjustment of the inventory transfer ratio and the external purchase weight based on the conflict type to generate an optimized material supply strategy after conflict resolution includes: If there is a first - type resource conflict, reduce the inventory transfer ratio, and recalculate the target transfer quantity according to the reduced inventory transfer ratio until the target transfer quantity does not exceed the sum of the available inventory and the in - transit inventory; If there is a second - type resource conflict, reduce the external purchase weight, and recalculate the target purchase quantity according to the reduced external purchase weight until the target purchase quantity does not exceed the maximum committed quantity and the minimum supply cycle is earlier than the demand time window; If there are both the first - type resource conflict and the second - type resource conflict, first reduce the inventory transfer ratio, and after the target transfer quantity corresponding to the reduced inventory transfer ratio is stable, then reduce the external purchase weight; Regenerate the supply resource quota according to the adjusted inventory transfer ratio and external purchase weight, and update the initial supply strategy based on the regenerated supply resource quota; Take the updated initial supply strategy as the optimized material supply strategy after conflict resolution.

8. The method according to claim 1, characterized in that, The method further includes: Real - time monitor the policy execution status of the supply chain management system, and obtain the actual execution effect data of the material allocation operation; Conduct a deviation analysis on the actual execution effect data and the predicted demand distribution result to generate a deviation analysis report; If the deviation coefficient in the deviation analysis report exceeds the preset deviation threshold, trigger a parameter correction operation for the dynamic analysis model; The parameter correction operation includes: adjusting the weight distribution ratio of the demand quantification module and the trend conduction module in the dynamic analysis model according to the deviation type in the actual execution effect data; After the corrected dynamic analysis model regenerates an updated predicted demand distribution result, iteratively optimize the material supply optimization strategy based on the updated predicted demand distribution result.

9. The method according to claim 8, wherein The conducting a deviation analysis on the actual execution effect data and the predicted demand distribution result to generate a deviation analysis report includes: Extract the actual demand satisfaction rate and the actual supply delay duration in the actual execution effect data; Extract the predicted demand satisfaction rate range and the predicted supply delay threshold in the predicted demand distribution result; Calculate a first deviation amount between the actual demand satisfaction rate and the median of the predicted demand satisfaction rate range, and a second deviation amount between the actual supply delay duration and the predicted supply delay threshold; Determine the deviation coefficient according to the weighted sum of the first deviation amount and the second deviation amount; If the deviation coefficient exceeds the deviation threshold, generate the deviation analysis report including the deviation type and the deviation impact level; wherein, the deviation type includes demand prediction deviation and supply response deviation, and the deviation impact level classifies the material shortage risk of the supply chain link according to the weighted sum.

10. A material requirements analysis and prediction device, comprising a memory and a processor, wherein the memory stores a computer program that can run on the processor, characterized in that, When the processor executes the program, the steps in the method according to any one of claims 1 to 9 are implemented.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, the steps in the method according to any one of claims 1 to 9 are implemented.

Images