E-commerce inventory risk assessment method and system based on big data

Abstract

The invention relates to the technical field of risk management, in particular to an e-commerce inventory risk assessment method and system based on big data, and the method comprises the following steps: collecting transaction data, supply chain link data and festival and holiday sales fluctuation data of an e-commerce platform, extracting the sales volume, the inventory volume, the replenishment time interval, the logistics distribution time and the demand prediction error, and calculating the risk of the e-commerce platform; and calculating fluctuation amplitude values in the time sequence item by item to obtain a multi-dimensional inventory scene data set. According to the invention, comprehensive risk analysis is carried out on multi-dimensional nodes such as inventory level, demand fluctuation and logistics distribution duration through a big data technology, risk distribution is accurately identified and quantified, the accuracy and coverage range of risk identification are improved, the upper limit and lower limit of inventory are dynamically adjusted, and the demand fluctuation amplitude is responded in real time. The flexibility and scientificity of inventory regulation and control are guaranteed, the adaptability of inventory management is improved, the volatility and loss risk in inventory management are remarkably reduced, and data-driven support is provided for scientific decision-making of enterprises.

Description

Technical Field

[0001] The present invention relates to the technical field of risk management, and particularly to an e-commerce inventory risk assessment method and system based on big data. Background Art

[0002] The technical field of risk management includes relevant methods and technical systems for identifying, assessing, and controlling various risks. The core content lies in quantifying and analyzing potential risks through scientific methods and technical means, and formulating effective coping strategies through constructing risk models to mitigate or avoid risk losses. In the overall technical field of risk management, the key points include four main links: risk identification, risk analysis, risk assessment, and risk control. Using technical tools such as statistical analysis, model establishment, data mining, and real-time monitoring, risks in different scenarios are systematically processed and controlled, and are widely applied in multiple fields such as financial insurance, logistics supply chain management, and enterprise operation.

[0003] Among them, the big data-based e-commerce inventory risk assessment method refers to using big data technology to analyze and evaluate potential risks in the inventory management process of e-commerce enterprises. This patent theme mainly aims at problems such as overstocking, out-of-stock, and supply chain fluctuations faced by e-commerce enterprises in inventory management, including technical matters such as collection and storage of inventory data, inventory risk identification, inventory risk assessment, and early warning. Specifically, by constructing a big data analysis model to process and mine inventory data, using inventory prediction algorithms combined with multi-dimensional information such as historical sales data, market demand fluctuations, and supply chain delivery capabilities to identify potential inventory risks, and by establishing a risk assessment index system to quantitatively evaluate the inventory status, outputting risk levels or risk early warning signals to assist e-commerce enterprises in realizing scientific management and decision-making of inventory risks.

[0004] In the process of existing technology for inventory risk identification, it relies on historical data for risk assessment, lacking comprehensive consideration of real-time multi-dimensional dynamic factors, resulting in the difficulty of the risk assessment result to quickly reflect actual demand and market fluctuations. Existing methods have deficiencies in path risk analysis and distribution identification, ignoring the relevance of multiple nodes in the supply chain and the complexity of risk propagation, and it is difficult to accurately predict the risk source and influence range. In terms of inventory adjustment, existing technology lacks the flexible response ability to real-time demand fluctuations, and it is easy to have situations of overstocking or understocking, leading to an increase in the risk of inventory overstocking or out-of-stock. For example, during promotional activities or in the case of sudden demand fluctuations, existing technology cannot quickly adjust the inventory structure, resulting in a decline in enterprise operation efficiency and economic losses. The linkage optimization of associated inventory is not fully considered, easily leading to an imbalance in the inventory levels of associated products and increasing inventory risks. These deficiencies make the existing technology lack flexibility and accuracy in the face of complex inventory management scenarios and are difficult to meet the needs of e-commerce enterprises for scientific inventory management. Summary of the Invention

[0005] In order to solve the technical problem that the existing technology lacks comprehensive consideration of real-time multi-dimensional dynamic factors, resulting in the risk assessment results being difficult to quickly reflect the actual demand and market fluctuations. There are deficiencies in the path risk analysis and distribution identification of the existing methods, ignoring the relevance of multiple nodes in the supply chain and the complexity of risk propagation, and it is difficult to accurately predict the risk source and the scope of influence. In terms of inventory adjustment, the existing technology lacks the flexible response ability to real-time demand fluctuations, and it is easy to have the situation of overstocking or understocking, resulting in the aggravation of inventory backlog or out-of-stock risks. For example, during promotional activities or sudden demand fluctuations, the existing technology cannot quickly adjust the inventory structure, resulting in the decline of enterprise operation efficiency and economic losses. The linkage optimization of associated inventory is not fully considered, which easily leads to the imbalance of the inventory levels of associated commodities and increases the inventory risk. These deficiencies make the existing technology lack flexibility and accuracy in the face of complex inventory management scenarios and are difficult to meet the needs of e-commerce enterprises for scientific inventory management. Embodiments of the present invention provide an e-commerce inventory risk assessment method and system based on big data. The technical solutions are as follows:

[0006] On the one hand, an e-commerce inventory risk assessment method based on big data is provided, and the method includes:

[0007] S1: Collect transaction data of the e-commerce platform, data of supply chain links, and holiday sales fluctuation data, extract the sales volume, inventory quantity, replenishment time interval, logistics distribution time, and demand forecasting error, calculate the fluctuation amplitude value in the time series item by item, and obtain a multi-dimensional inventory scenario data set;

[0008] S2: Use the multi-dimensional inventory scenario data set to extract nodes in the e-commerce inventory supply chain link, calculate the weight value of the nodes, and for the node connection relationship in the supply chain path, combine the weight difference value and the goods circulation frequency in the path to evaluate the path connection risk, and obtain the risk propagation path analysis result;

[0009] S3: Use the risk propagation path analysis result to extract e-commerce inventory associated nodes, evaluate the weight distribution between nodes, use big data, perform distribution analysis through the risk weights of the nodes, and combine the risk distribution position and frequency annotation to obtain the risk distribution identification result;

[0010] S4: According to the risk distribution identification result, screen the risk level nodes, screen the risk level nodes, divide the inventory fluctuation cycle into intervals according to the delivery speed, compare the inventory dynamic change with the demand fluctuation amplitude, and adjust the upper and lower limits of the e-commerce inventory level period by period to obtain the inventory dynamic adjustment result;

[0011] S5: Calculate the inventory linkage impact coefficient of associated products based on the inventory dynamic adjustment result, adjust the linkage parameters of e-commerce inventory according to the correlation coefficient, optimize the e-commerce inventory level, and generate an inventory linkage risk assessment result.

[0012] As a further solution of the present invention, the multi-dimensional inventory scenario data set includes sales volume data, inventory volume data, replenishment time interval data, logistics distribution time data, and demand forecast error data. The risk propagation path analysis result includes the weight value of the warehousing node, the weight value of the transportation node, the weight value of the demand-side node, and the corresponding risk propagation path. The risk distribution identification result includes the risk distribution of the inventory level node, the risk distribution of the demand quantity node, and the risk distribution of the logistics distribution duration node. The inventory dynamic adjustment result includes the risk level node screening result, the inventory upper limit adjustment value, and the inventory lower limit adjustment value. The inventory linkage risk assessment result includes product sales correlation data, inventory fluctuation linkage impact coefficient, and linkage parameter optimization adjustment value.

[0013] As a further solution of the present invention, collect the transaction data of the e-commerce platform, the data of the supply chain link, and the holiday sales fluctuation data, extract the sales volume, inventory volume, replenishment time interval, logistics distribution time, and demand forecast error, and calculate the fluctuation amplitude value in the time series item by item. The specific steps to obtain the multi-dimensional inventory scenario data set are as follows:

[0014] S101: Collect the transaction data of the e-commerce platform, the data of the supply chain link, and the holiday sales fluctuation data, perform field screening on the sales volume, inventory volume, replenishment time interval, logistics distribution time, and demand forecast error item by item, arrange the data in chronological order by timestamp and perform unified formatting processing, mark and analyze the corresponding time node relationship according to the time interval, and generate a time series marked data set;

[0015] S102: According to the time series marked data set, calculate the difference and change rate of the sales volume, inventory volume, and replenishment time interval at adjacent time nodes item by item, statistically calculate the standard deviation by time node segment, calculate the fluctuation amplitude value, and generate a sales and inventory fluctuation data table;

[0016] S103: Use the sales and inventory fluctuation data table to extract the fluctuation values of the logistics distribution time and demand forecast error and the time node item by item, perform matching on the time axis, analyze the corresponding relationship between the logistics distribution and the demand forecast error, and integrate the fluctuation data of the time node according to the supply chain link to obtain a multi-dimensional inventory scenario data set.

[0017] As a further solution of the present invention, the formula for calculating the fluctuation amplitude value is as follows:

[0018]

[0019] Among them, VA is the amplitude value, x 1 and x 2 represent the sales volumes at two adjacent time nodes, y 1 and y 2 represent the inventory levels at two adjacent time nodes, z 1 and z 2 represent the replenishment time intervals at two adjacent time nodes, w 1 、w 2 、w 3 are weight coefficients, and k is a smoothing adjustment coefficient.

[0020] As a further solution of the present invention, by using the multi-dimensional inventory scenario data set, nodes in the e-commerce inventory supply chain link are extracted, the weight values of the nodes are calculated, and for the node connection relationships in the supply chain path, the path connection risk is evaluated by combining the weight difference value and the goods flow frequency in the path, and the steps for obtaining the risk propagation path analysis result are specifically as follows:

[0021] S201: Based on the multi-dimensional inventory scenario data set, the time series data of the warehousing nodes, transportation nodes, and demand-side nodes are extracted, the sales volumes, inventory levels, and amplitude values of the differentiated nodes in the time series are statistically summarized and standardized, and the logistics distribution time and demand prediction error of the corresponding nodes are summarized in segments according to the time series to generate a node feature data set;

[0022] S202: According to the node feature data set, the proportion of the amplitude value of the warehousing nodes, transportation nodes, and demand-side nodes in the total fluctuation value is calculated item by item, and the weighted calculation of the proportion weight of the logistics distribution time and the normalized value of the demand prediction error is carried out for each node according to the supply chain link to obtain a node weight data table;

[0023] S203: By using the node weight data table, the connection paths between the differentiated nodes in the supply chain are extracted, the cumulative value of the weight values of each node on the path is calculated item by item, the amplitude value of the path is analyzed according to the change amount of the logistics distribution time between the nodes, and the path risk level is classified to obtain the risk propagation path analysis result.

[0024] As a further solution of the present invention, by using the risk propagation path analysis result, the e-commerce inventory associated nodes are extracted, the weight distribution between the nodes is evaluated, and by using big data, the distribution analysis is carried out through the risk weights of the nodes, and the steps for obtaining the risk distribution recognition result by combining the risk distribution position and frequency annotation are specifically as follows:

[0025] S301: Based on the analysis result of the risk propagation path, extract the risk weight values of the inventory level, demand volume, and logistics distribution duration nodes, match the weight value of each node with the time series fluctuation amplitude value, segment and screen the risk propagation path between nodes, and mark the node types to obtain the node risk weight distribution data set;

[0026] S302: According to the node risk weight distribution data set, use big data to perform multi-dimensional cross-analysis on the time series fluctuation amplitude value and the node weight value, classify and summarize the risk values of each type of node according to different time periods, and generate the e-commerce inventory risk classification result;

[0027] S303: Use the e-commerce inventory risk classification result to perform multi-dimensional distribution statistics on the inventory level node, demand volume node, and logistics distribution duration node, calculate the risk distribution value of the node under multi-dimensional risk conditions, and obtain the risk distribution identification result.

[0028] As a further solution of the present invention, the formula for calculating the risk distribution value of the node under multi-dimensional risk conditions is as follows:

[0029]

[0030] Among them, R ij (t) represents the risk distribution value of node i at time t, D ij (t) represents the demand volume data value corresponding to node i at time t, μ ij represents the average demand volume of node i at time t, L ij (t) represents the logistics distribution duration of node i at time t, represents the average logistics distribution duration of node i at time t, T ij (t) represents the inventory level of node i at time t, ∈ is a smoothing factor, ω 1 、ω 2 、ω 3 are weight coefficients.

[0031] As a further solution of the present invention, according to the risk distribution identification result, the steps of screening the risk level nodes, screening the risk level nodes, dividing the inventory fluctuation cycle into intervals according to the delivery speed, comparing the inventory dynamic change with the demand fluctuation amplitude, and adjusting the upper and lower limits of the e-commerce inventory level by time period to obtain the inventory dynamic adjustment result are specifically as follows:

[0032] S401: Adopt the risk distribution identification result, screen the inventory level nodes, demand volume nodes, and logistics distribution speed nodes with the highest risk levels, sort and segment and count the risk values and time series fluctuation amplitude values of the nodes, mark the nodes whose risk values exceed the target threshold, and generate the risk node data set;

[0033] S402: Based on the risk node dataset, extract the dynamic change value of the inventory level node and the fluctuation amplitude value of the demand node period by period, calculate the difference and proportional change between the two, and generate an inventory adjustment interval;

[0034] S403: Through the inventory adjustment interval, dynamically adjust the upper and lower limits of the inventory level period by period. Combining the demand fluctuation amplitude value and the change trend of the distribution speed, screen the key nodes and time periods in the supply chain where the inventory level needs to be optimized to obtain the inventory dynamic adjustment result.

[0035] As a further solution of the present invention, the steps of calculating the inventory linkage impact coefficient of associated products according to the inventory dynamic adjustment result, adjusting the linkage parameters of e-commerce inventory according to the correlation coefficient, optimizing the e-commerce inventory level, and generating the inventory linkage risk assessment result are specifically as follows:

[0036] S501: According to the inventory dynamic adjustment result, extract the product sales correlation data and inventory fluctuation data, perform multi-dimensional matching on the sales volume and inventory fluctuation value of associated products, evaluate the synchronization of the change in product sales volume and the e-commerce inventory level, and generate a product correlation dataset;

[0037] S502: Using the product correlation dataset, calculate the inventory linkage impact coefficient between associated products item by item, segment by time series, analyze the contribution value of inventory fluctuation to the linkage impact coefficient, and quantify the linkage degree to generate an inventory linkage impact coefficient table;

[0038] S503: Using the inventory linkage impact coefficient table, normalize the inventory linkage impact coefficient of associated products, adjust the linkage parameters of e-commerce inventory according to the linkage relationship between product sales and inventory fluctuation, optimize and adjust the inventory level, and optimize the inventory configuration of the supply chain period by period to obtain the inventory linkage risk assessment result.

[0039] On the other hand, an electric vehicle status monitoring system is provided. The electric vehicle status monitoring system is used to execute the above-mentioned electric vehicle status monitoring method. The system includes:

[0040] The fluctuation analysis module extracts the sales volume, inventory quantity, replenishment time interval, logistics distribution time, and demand prediction error based on the transaction data of the e-commerce platform, the data of the supply chain link, and the holiday sales fluctuation data to generate a multi-dimensional inventory scenario dataset;

[0041] The association relationship analysis module extracts the warehousing, transportation, and demand-side nodes in the e-commerce inventory supply chain based on the multi-dimensional inventory scenario dataset, calculates the node weight value, analyzes the association relationship between the warehousing and transportation nodes, and generates a risk propagation path analysis result;

[0042] The risk value calculation module uses the risk propagation path analysis results to extract inventory level nodes, demand nodes and logistics delivery time nodes, calculates the risk values ​​of inventory level, demand and logistics delivery time nodes, and generates inventory risk distribution identification results;

[0043] The inventory optimization module uses the inventory risk distribution identification result to calculate the difference between the dynamic change value of the e-commerce inventory and the demand fluctuation amplitude value, adjusts the upper and lower limits of the multi-period inventory level based on the difference result, and generates the inventory dynamic adjustment result;

[0044] The linkage impact assessment module extracts commodity sales correlation data and inventory fluctuation data based on the inventory dynamic adjustment result, calculates the inventory linkage impact coefficient of the related commodities, optimizes the commodity inventory level, and generates the inventory linkage risk assessment result.

[0045] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0046] Through the comprehensive collection and processing of transaction data, supply chain link data, and holiday sales fluctuation data, an inventory scenario data set is constructed from a multi-dimensional perspective to achieve accurate quantification of each key link of e-commerce inventory. Through the weight analysis of warehousing, transportation and demand nodes, the risk propagation path is identified, the potential source and propagation mechanism of inventory risk are deeply explored, and the overall picture of risk distribution is revealed. Using big data technology, a comprehensive risk analysis of multi-dimensional nodes such as inventory level, demand fluctuation and logistics delivery time is carried out to accurately identify and quantify risk distribution, and improve the accuracy and coverage of risk identification. The upper and lower limits of inventory are dynamically adjusted to respond to the fluctuation range of demand in real time, ensure the flexibility and scientificity of inventory regulation, and improve the adaptability of inventory management. The calculation of the correlation of commodity sales and the linkage coefficient of inventory fluctuation enables e-commerce companies to optimize the inventory structure, reduce inventory redundancy and out-of-stock risks, and achieve overall optimization of related inventory. The real-time optimization method based on multi-dimensional data analysis improves the accuracy of risk assessment, dynamic response capability and global optimization effect, significantly reduces volatility and loss risk in inventory management, and provides data-driven support for scientific decision-making of enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a schematic diagram of the workflow of the present invention;

[0048] Figure 2 This is a detailed flow chart of S1 of the present invention;

[0049] Figure 3 This is a detailed flow chart of S2 of the present invention;

[0050] Figure 4 This is a detailed flow chart of S3 of the present invention;

[0051] Figure 5 It is the detailed flowchart of S4 of the present invention;

[0052] Figure 6 It is the detailed flowchart of S5 of the present invention;

[0053] Figure 7 It is the system flowchart of the present invention. Specific embodiments

[0054] The technical solutions in the present invention will be described below with reference to the accompanying drawings.

[0055] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to represent examples, illustrations or explanations. Any embodiment or design solution described as "example" in the present invention should not be construed as being more preferred or more advantageous than other embodiments or design solutions. Rather, the use of the word "example" is intended to present concepts in a specific manner. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one of the two can be selected.

[0056] To make the technical problems, technical solutions and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

[0057] Please refer to Figure 1 , the embodiments of the present invention provide an e-commerce inventory risk assessment method based on big data, and the processing flow of this method can include the following steps:

[0058] S1: Collect transaction data, supply chain link data and holiday sales fluctuation data of the e-commerce platform, extract sales volume, inventory, replenishment time interval, logistics distribution time and demand forecast error, calculate the fluctuation amplitude value in the time series item by item, analyze the correlation degree between parameters, and obtain a multi-dimensional inventory scenario data set;

[0059] S2: Use the multi-dimensional inventory scenario data set, extract the warehousing nodes, transportation nodes and demand-side nodes in the e-commerce inventory supply chain link, calculate the weight value of the nodes through the node relationship information, and analyze the risk degree of the path in combination with the node weight value to obtain the risk propagation path analysis result;

[0060] S3: Use the risk propagation path analysis result, extract the e-commerce inventory associated nodes, evaluate the weight distribution between the nodes, use big data, perform distribution analysis through the risk weights of the nodes, and combine the risk distribution position and frequency annotation to obtain the risk distribution recognition result;

[0061] S4: According to the risk distribution identification result, screen the risk level nodes, and based on the distribution speed, divide the inventory fluctuation cycle into intervals, compare the dynamic change of inventory with the demand fluctuation amplitude, and adjust the upper and lower limits of the e-commerce inventory level period by period to obtain the inventory dynamic adjustment result;

[0062] S5: Through the inventory dynamic adjustment result, calculate the inventory linkage impact coefficient of related products, adjust the linkage parameters of the e-commerce inventory according to the correlation coefficient, optimize the e-commerce inventory level, and generate the inventory linkage risk assessment result.

[0063] The multi-dimensional inventory scenario data set includes sales volume data, inventory quantity data, replenishment time interval data, logistics distribution time data, and demand forecast error data. The risk propagation path analysis result includes the weight value of the warehousing node, the weight value of the transportation node, the weight value of the demand-side node, and the corresponding risk propagation path. The risk distribution identification result includes the risk distribution of the inventory level node, the risk distribution of the demand quantity node, and the risk distribution of the logistics distribution duration node. The inventory dynamic adjustment result includes the screening result of the risk level node, the inventory upper limit adjustment value, and the inventory lower limit adjustment value. The inventory linkage risk assessment result includes the product sales correlation data, the inventory fluctuation linkage impact coefficient, and the linkage parameter optimization adjustment value.

[0064] Please refer to Figure 2 , collect the transaction data, supply chain link data, and holiday sales fluctuation data of the e-commerce platform, extract the sales volume, inventory quantity, replenishment time interval, logistics distribution time, and demand forecast error, and calculate the fluctuation amplitude value in the time series item by item. The specific steps to obtain the multi-dimensional inventory scenario data set are as follows:

[0065] S101: Collect the transaction data, supply chain link data, and holiday sales fluctuation data of the e-commerce platform, screen the fields of sales volume, inventory quantity, replenishment time interval, logistics distribution time, and demand forecast error one by one, arrange the data in chronological order through the time stamp and perform unified formatting processing, mark and analyze the corresponding time node relationship according to the time interval, and the execution process of generating the time series marked data set is as follows;

[0066] For the data in the supply chain link and the data on sales fluctuations during holidays, clean and screen the original data, extract the data content including fields such as sales volume, inventory, replenishment time interval, logistics distribution time, and demand forecasting error. Standardize the fields in different data sources in a unified format, arrange the data in chronological order through timestamps, and complete the time interpolation for the missing timestamp data. Mark the data in time intervals such as daily, weekly, or monthly, divide the data into different time segments. For the transaction data and supply chain data in each time segment, analyze the correlation between sales volume, inventory, and replenishment cycle, specially mark the data on sales fluctuations during holidays, and combine the data trends of transaction peaks and replenishment intervals to analyze the dynamic relationships of time nodes one by one, generating a time series marked data set.

[0067] S102: According to the time series marked data set, calculate the differences and change rates of sales volume, inventory, and replenishment time interval at adjacent time nodes item by item, statistically calculate the standard deviation by time node segments, calculate the volatility amplitude value, and the execution process for generating the sales and inventory volatility data table is as follows;

[0068] The formula for calculating the volatility amplitude value is as follows:

[0069]

[0070] Among them, VA is the volatility amplitude value, x 1 and x 2 represent the sales volumes at two adjacent time nodes, y 1 and y 2 represent the inventories at two adjacent time nodes, z 1 and z 2 represent the replenishment time intervals at two adjacent time nodes, w 1 、w 2 、w 3 are weight coefficients, and k is a smoothing adjustment coefficient;

[0071] Parameter meanings and setting values

[0072] x 1 is the sales volume at the current node, which is monitored through the sales records in the time series and is set to 1500 pieces;

[0073] x 2 is the sales volume at the previous node, which is monitored through the sales records in the time series and is set to 1350 pieces;

[0074] y 1 is the inventory at the current node, which is obtained through the inventory records and is set to 5000 pieces;

[0075] y 2is the inventory of the previous node, and the data is obtained from the inventory record, set to 5200 pieces;

[0076] z 1 is the replenishment time interval of the current node, and the data is obtained from the replenishment plan record, set to 5 days;

[0077] z 2 is the replenishment time interval of the previous node, and the data is obtained from the replenishment plan record, set to 3 days;

[0078] w 1 is the weight coefficient of the sales volume change, with a set value of 0.5, set according to the importance of the fluctuation range, and the weight value fluctuates with the importance ratio of the sales volume in the overall fluctuation range;

[0079] w 2 is the weight coefficient of the inventory change, with a set value of 0.3, set according to the importance of the fluctuation range, and the weight value fluctuates with the importance ratio of the inventory in the overall fluctuation range;

[0080] w 3 is the weight coefficient of the replenishment time interval change, with a set value of 0.2, set according to the importance of the fluctuation range, and the weight value fluctuates with the importance ratio of the replenishment interval in the overall fluctuation range;

[0081] k is the smoothing adjustment coefficient, with a set value of 0.1, used to control the sensitivity of the overall fluctuation, and this value is set through the mean and variance analysis of historical data;

[0082] Substitute the parameters into the formula for calculation

[0083] Calculate the squared term of the sales volume change:

[0084] w 1 |x 1 -x 2 | 2 = 0.5·|1500 - 1350| 2 = 0.5·150 2 = 0.5·22500 = 11250;

[0085] Calculate the squared term of the inventory change:

[0086] w 2 |y 1 -y 2 | 2 = 0.3·|5000 - 5200| 2 = 0.3·200 2 = 0.3·40000 = 12000;

[0087] Calculate the squared term of the change in replenishment time interval:

[0088] w 3 |z 1 -z 2 | 2 = 0.2 · |5 - 3| 2 = 0.2 · 2 2 = 0.2 · 4 = 0.8;

[0089] Substitute the above results into the numerator of the formula for calculation:

[0090] 11250 + 12000 + 0.8 = 23250.8;

[0091] Calculate the denominator:

[0092] 1 + k = 1 + 0.1 = 1.1;

[0093] Calculate the fluctuation amplitude value:

[0094]

[0095] The results show that the comprehensive fluctuation amplitude of the current time node in terms of sales volume, inventory level, and replenishment time interval is 145.42. This result is used to quantify the fluctuation amplitude and eliminate abnormal data that exceeds the set target range, providing a more stable data basis for the analysis of time series.

[0096] S103: Using the sales and inventory fluctuation data table, extract the fluctuation values of logistics delivery time and demand forecasting error with respect to the time node item by item, perform matching on the time axis, analyze the corresponding relationship between logistics delivery and demand forecasting error, and integrate the fluctuation data of the time node according to the supply chain link. The execution process of obtaining the multi-dimensional inventory scenario dataset is as follows;

[0097] Extract the time node data of logistics delivery time and demand forecasting error. For the data fluctuation characteristics in the time series, perform field extraction and data integration on the fluctuation amplitudes of logistics delivery time and demand forecasting error, establish a matching relationship of fluctuation values on the time axis, align the time data of logistics delivery and demand forecasting error data according to the time series, analyze the fluctuation trend on the time axis, and quantify the corresponding relationship between the two through statistical analysis methods. Divide the node data according to the supply chain link, perform hierarchical processing and integration on the fluctuation data of each node, and conduct time series segmentation statistics and clustering analysis on the delivery time fluctuation and demand forecasting error in different supply chain links to provide data support for the dynamic adjustment between supply chain nodes and obtain the multi-dimensional inventory scenario dataset.

[0098] Please refer to Figure 3, using a multi-dimensional inventory scenario dataset, extract the nodes in the e-commerce inventory supply chain link, calculate the weight values of the nodes, and for the node connection relationships in the supply chain path, evaluate the path connection risk by combining the weight difference value and the goods flow frequency in the path, and the steps to obtain the risk propagation path analysis result are specifically as follows:

[0099] S201: Based on the multi-dimensional inventory scenario dataset, extract the time series data of the warehousing nodes, transportation nodes, and demand-side nodes, statistically summarize and standardize the sales volume, inventory volume, and fluctuation amplitude values of the differential nodes in the time series, and segment and summarize the logistics distribution time and demand prediction error of the corresponding nodes according to the time series, and the execution process of generating the node feature dataset is as follows;

[0100] To extract the time series data of the warehousing nodes, transportation nodes, and demand-side nodes, it is necessary to classify and identify various nodes in the inventory scenario, including the inventory storage data of the warehousing nodes, the logistics status data of the transportation nodes, and the sales data of the demand-side nodes. For the data characteristics of different nodes, extract the sales volume, inventory volume, and fluctuation amplitude values in the time series respectively, uniformly organize the time series data of the nodes into a structured data table, perform fluctuation statistics on the inventory data of the warehousing nodes and transportation nodes in the time series, calculate the change amplitude of the inventory volume of the nodes in each time period, use the sliding window method to obtain the standard deviation of the sales volume and inventory volume in different time periods, and perform standardization processing according to the difference value between time periods, and summarize the logistics distribution time and demand prediction error data of the warehousing nodes, transportation nodes, and demand-side nodes in each time period to generate the node feature dataset.

[0101] S202: According to the node feature dataset, calculate the proportion of the fluctuation amplitude value of the warehousing nodes, transportation nodes, and demand-side nodes in the total fluctuation value item by item, and perform weighted calculation on the proportion weight of the logistics distribution time and the normalized value of the demand prediction error for each node according to the supply chain link to obtain the execution process of the node weight data table as follows;

[0102] According to the node feature dataset, calculate the proportion of the fluctuation amplitude value of the warehousing nodes, transportation nodes, and demand-side nodes in the total fluctuation value item by item, according to the formula:

[0103]

[0104] Calculate the proportion of the fluctuation amplitude value of node p in the total fluctuation value. In the formula, R p represents the proportion of the fluctuation amplitude of node p, V p is the fluctuation amplitude value of node p, N is the total number of nodes in the supply chain network, is the sum of the fluctuation amplitude values of all nodes;

[0105] Formula detailed explanation and formula calculation derivation process:

[0106] Extract data samples of three nodes from the node feature dataset, and set the fluctuation amplitude value of warehouse node A to V A = 120, the fluctuation amplitude value of transportation node B to V B = 150, and the fluctuation amplitude value of demand-side node C to V C = 230;

[0107] Calculate the total sum of the fluctuation amplitude values of the total number of nodes:

[0108]

[0109] Calculate the proportion of the fluctuation amplitude of each node respectively:

[0110] Proportion of the fluctuation amplitude of warehouse node A:

[0111]

[0112] Proportion of the fluctuation amplitude of transportation node B:

[0113]

[0114] Proportion of the fluctuation amplitude of demand-side node C:

[0115]

[0116] The results show that the proportions of the fluctuation amplitudes of warehouse node A, transportation node B, and demand-side node C are 0.24, 0.30, and 0.46 respectively, indicating that demand-side node C contributes the most to the overall fluctuation amplitude. In subsequent analyses, the supply chain optimization strategy can be adjusted according to this proportion for priority setting.

[0117] S203: Use the node weight data table to extract the connection paths between different nodes in the supply chain, calculate the cumulative value of the weight values of each node on the path item by item, analyze the fluctuation amplitude value of the path according to the change amount of the logistics distribution time between nodes, and classify the risk level of the path. The execution process for obtaining the risk propagation path analysis result is as follows;

[0118] To extract the connection paths between differentiated nodes in the supply chain, it is necessary to label the connection relationships of each node in the supply chain network, clarify the path associations among warehousing nodes, transportation nodes, and demand-side nodes, gradually extract the node weight values involved in each path, accumulate the node weight values, calculate the importance of the path using the cumulative weight value of the path, and at the same time, combine the change amount of the logistics distribution time, analyze the fluctuation amplitude value of the path in different time periods through the time difference method, use the fluctuation amplitude value as the main evaluation index of the path risk, grade and mark the parts with larger fluctuation amplitude values in the path, classify the risks as high-risk, medium-risk, and low-risk, analyze the main node weights and fluctuation values of the high-risk paths, calculate the influence contribution rate of the fluctuation of each node on the overall risk of the path, and use it as a reference for the decision-making optimization of supply chain management to obtain the analysis results of the risk propagation path.

[0119] Please refer to Figure 4 , and the steps to extract e-commerce inventory association nodes, evaluate the weight distribution among nodes, use big data to conduct distribution analysis through the risk weights of nodes, and combine the risk distribution position and frequency annotation to obtain the risk distribution recognition result by using the analysis result of the risk propagation path are as follows:

[0120] S301: Based on the analysis result of the risk propagation path, extract the risk weight values of the inventory level, demand quantity, and logistics distribution duration nodes, match the weight value of each node with the time series fluctuation amplitude value, segment and screen the risk propagation paths among nodes and mark the node types. The execution process of obtaining the node risk weight distribution data set is as follows;

[0121] Extract the risk weight values of the inventory level, demand quantity, and logistics distribution duration nodes, classify and label the nodes in the risk propagation path, divide the nodes into inventory nodes, demand nodes, and logistics nodes, extract the risk weight value of each node, for the time series fluctuation amplitude value of the node, use the time axis to match the risk weight value and the fluctuation amplitude segment by segment, according to the dynamic change trend of the fluctuation amplitude, segment and screen the risk propagation path of the node, label and classify the key features of the high-fluctuation nodes, extract the features of the risk propagation paths of different node types, divide the risk propagation paths among nodes in the form of multiple time series, and mark the type information and dynamic change characteristics of each node one by one to provide data support for supply chain optimization and generate the node risk weight distribution data set.

[0122] S302: According to the node risk weight distribution data set, use big data to conduct multi-dimensional cross-analysis of the time series fluctuation amplitude value and the node weight value, classify and summarize the risk values of each type of node according to different time periods, and the execution process of generating the e-commerce inventory risk classification result is as follows;

[0123] Using big data analysis technology, perform multi-dimensional cross-analysis on the time series fluctuation amplitude value and the node weight value, construct a cross-data table for the risk weight value and the fluctuation amplitude value of the node, classify and summarize the risk values of the nodes according to the fluctuation characteristics in different time periods, through the time segmentation statistical method, conduct differential analysis and comparison on the risk values of each type of node, establish corresponding correlation analysis models for the risk values of the inventory nodes and the risk values of the demand nodes and logistics nodes respectively, cluster the risk characteristics of the nodes in the differential time periods, and form a classification result table according to the node type and time period, which is used to further improve the risk control strategies of different nodes in supply chain management and generate the e-commerce inventory risk classification result.

[0124] S303: Using the e-commerce inventory risk classification result, perform multi-dimensional distribution statistics on the inventory level node, the demand quantity node, and the logistics distribution duration node, calculate the risk distribution value of the node under multi-dimensional risk conditions, and the execution process for obtaining the risk distribution identification result is as follows;

[0125] The formula for calculating the risk distribution value of the node under multi-dimensional risk conditions is as follows:

[0126]

[0127] Among them, R ij (t) represents the risk distribution value of node i at time t, D ij (t) represents the demand quantity data value corresponding to node i at time t, μ ij represents the average demand quantity of node i at time t, L ij (t) represents the logistics distribution duration of node i at time t, represents the average logistics distribution duration of node i at time t, T ij (t) represents the inventory level of node i at time t, ∈ is the smoothing factor, ω 1 、ω 2 、ω 3 are the weight coefficients;

[0128] Parameter meaning and setting value:

[0129] D ij (t) represents the observed value of the demand quantity of the node at a certain time, and the monitoring range is calculated based on the historical demand data in the past 12 months. The observed value of the demand quantity in a specific time period is 100 units;

[0130] μ ij represents the average value of the demand quantity of the node in the past 12 months, which is calculated by statistically analyzing historical data. The average value is 85 units, reflecting the central tendency of historical demand;

[0131] L ij(t) represents the observed value of the logistics distribution duration of a node in a specific time period. The monitoring range is obtained based on the actual data of the logistics distribution duration, and the distribution duration is 6 hours;

[0132] represents the average value of the logistics distribution duration of the node in the past 12 months. The average value is obtained through statistical analysis of historical data and is 8 hours, reflecting the average stable level of the distribution duration;

[0133] T ij (t) represents the inventory level of a node in a specific time period. The monitoring range is output based on the real-time data of the inventory management system, and the inventory level is 1500 units;

[0134] ∈ represents the smoothing factor, and the set value is 0.1, which is used to avoid abnormal denominator calculation when the inventory level is zero;

[0135] ω 1 represents the weight parameter of the demand quantity in the formula. It is set according to the sensitivity of historical demand fluctuations, and the weight setting value is 0.5, which fluctuates with the dispersion degree of historical demand data;

[0136] ω 2 represents the weight parameter of the logistics distribution duration in the formula. It is set according to the influence degree of the distribution duration fluctuation, and the weight setting value is 0.3, which fluctuates with the standard deviation of the distribution duration;

[0137] ω 3 represents the weight parameter of the inventory level in the formula. It is set according to the importance of inventory management, and the weight setting value is 0.2, which fluctuates with the safety critical value of the inventory level;

[0138] Substitute the parameters into the formula for calculation:

[0139] Calculate the squared deviation term of the demand quantity:

[0140] |D ij (t)-μ ij | 2 =|100 - 85| 2 =15 2 =225;

[0141] Calculate the squared deviation term of the logistics distribution duration:

[0142]

[0143] Substitute the above squared deviation terms into the numerator part of the formula:

[0144]

[0145] Calculate the denominator part and complete the formula calculation:

[0146]

[0147] The results show that under the multi-dimensional fluctuations of demand volume, logistics distribution duration, and inventory level, the risk distribution value of this node is relatively high, mainly affected by the relatively large deviation value of demand volume. Combining with the setting of weight coefficients, the accuracy of risk assessment can be further optimized or the node priority can be adjusted.

[0148] Please refer to Figure 5 , according to the risk distribution identification results, filter the nodes with risk levels, filter the nodes with risk levels, divide the inventory fluctuation cycle into intervals according to the distribution speed, compare the dynamic changes of inventory with the amplitude of demand fluctuations, and adjust the upper and lower limits of the e-commerce inventory level period by period. The specific steps to obtain the dynamic inventory adjustment results are as follows:

[0149] S401: Using the risk distribution identification results, filter the inventory level nodes, demand volume nodes, and logistics distribution speed nodes with priority risk levels, sort and segment the statistics of the risk values and the amplitude values of time series fluctuations of the nodes, and mark the nodes whose risk values exceed the target threshold. The execution process of generating the risk node dataset is as follows;

[0150] Filter the inventory level nodes, demand volume nodes, and logistics distribution speed nodes with priority risk levels, classify and filter the nodes in the risk distribution identification results, sort the nodes with priority risk levels in descending order of risk values, and at the same time extract the amplitude values of the time series of the nodes. Through the segmented statistical method, the fluctuation amplitude and risk value of each node are processed in time segments. Mark the nodes with large fluctuation amplitude and risk values exceeding the target threshold, gradually form a list of high-risk nodes, and at the same time classify and organize the risk sources of high-risk nodes, record the risk characteristics of inventory level nodes, demand volume nodes, and logistics distribution speed nodes respectively, and generate a risk node dataset.

[0151] S402: Based on the risk node dataset, extract the dynamic change value of the inventory level node and the amplitude value of the demand volume node period by period, calculate the difference and proportional change between the two, and the execution process of generating the inventory adjustment interval is as follows;

[0152] Extract the dynamic change values of the inventory level nodes and the fluctuation amplitude values of the demand nodes for each time period. For the data of the inventory level nodes, extract the dynamic change values of the inventory within each time period, calculate the time series trend of the inventory change values, and at the same time extract the fluctuation amplitude values of the demand nodes within the same time period. Align and match the data of both along the time axis. Calculate the difference between the inventory change value and the demand fluctuation value through the difference method, and at the same time conduct a quantitative analysis of the relationship between the two according to the proportional change calculation formula. Conduct a segmented statistics on the differences and proportional changes within each time period to form an inventory adjustment indicator within the time interval, and classify it according to the difference range and the interval of the proportional change to generate an inventory adjustment interval.

[0153] S403: Through the inventory adjustment interval, dynamically adjust the upper and lower limits of the inventory level for each time period. Combining the demand fluctuation amplitude value and the change trend of the distribution speed, screen the key nodes and time periods in the supply chain where the inventory level needs to be optimized. The execution process of obtaining the inventory dynamic adjustment result is as follows;

[0154] Dynamically adjust the upper and lower limits of the inventory level for each time period. Use the data in the inventory adjustment interval to set the upper and lower limit values of the inventory level. Combining the demand fluctuation amplitude value and the change trend of the logistics distribution speed, screen the key nodes in the supply chain where the inventory level needs to be optimized, and conduct an analysis of the dynamic range of the inventory level for each key node for each time period. At the same time, combining the dynamic change trend of the distribution speed, identify the time periods with low distribution speed and large demand fluctuations, and further adjust the upper and lower limit ranges of the inventory to dynamic values. By integrating and optimizing the inventory adjustment interval data for each time period, screen out the inventory nodes and time periods in the supply chain that need to be optimized key points, provide specific adjustment strategies for the optimization of supply chain management, and obtain the inventory dynamic adjustment result.

[0155] Please refer to Figure 6 , through the inventory dynamic adjustment result, calculate the inventory linkage impact coefficient of the associated products, adjust the linkage parameters of the e-commerce inventory according to the correlation coefficient, optimize the e-commerce inventory level, and the steps to generate the inventory linkage risk assessment result are specifically as follows:

[0156] S501: According to the inventory dynamic adjustment result, extract the product sales correlation data and inventory fluctuation data, conduct a multi-dimensional matching of the sales volume and inventory fluctuation value of the associated products, evaluate the synchronization of the change in the sales volume of the products and the e-commerce inventory level, and the execution process of generating the product correlation data set is as follows;

[0157] To extract the relevant data on commodity sales and inventory fluctuations, it is necessary to collect the commodity sales data and the dynamic changes in inventory within the e-commerce platform, construct a time series dataset in a time-segmented manner, extract the time series records of the daily average sales volume and inventory fluctuations by comparing the change trends of the sales data and inventory fluctuation data of each commodity. For the extracted time series record data, calculate the growth rate or change rate for the change value of sales volume and the change value of inventory respectively within each time period. Apply time series analysis methods to conduct a preliminary screening of the correlation between the change in commodity sales volume and inventory fluctuations, and then verify the accuracy by calculating the correlation coefficient between the two (such as the Pearson correlation coefficient), and gradually screen out the commodity pairs with significant correlations, eliminate the commodity pairs that do not meet the set correlation threshold, conduct multiple verifications on the screened commodity pairs, and further construct a set of highly correlated commodity pairs based on the verification results to ensure the effectiveness and rationality of the data results, and generate a commodity correlation dataset.

[0158] S502: Using the commodity correlation dataset, calculate the inventory linkage impact coefficient between associated commodities item by item, segment by time series, analyze the contribution value of inventory fluctuations to the linkage impact coefficient, and quantify the degree of linkage. The execution process for generating the inventory linkage impact coefficient table is as follows;

[0159] Calculate the inventory linkage impact coefficient between associated commodities item by item using the commodity correlation dataset, according to the formula:

[0160]

[0161] In the formula, C po represents the inventory linkage impact coefficient between commodity p and commodity o, S p,t and S o,t respectively represent the inventory fluctuation values of commodity p and commodity o in the time period t, and respectively represent the average inventory fluctuation values of commodity p and commodity o, and T is the total number of time periods for analysis;

[0162] Detailed explanation of the formula and the derivation process of formula calculation:

[0163] Select two commodities from the e-commerce platform, collect the daily inventory fluctuation data within one month. Let the inventory fluctuation value of commodity X be:

[0164] S p,t ={12, 18, 15, 10, 16, 14, 13, 17, 20, 18,

[0165] 12, 14, 16, 19, 13, 11, 15, 18, 19, 14, 16, 15, 13, 14, 12, 17, 19, 15, 16, 14};

[0166] The inventory fluctuation value of Product Y is:

[0167] S o,t ={10, 15, 12, 8, 14, 13, 11, 16, 18, 17, 11, 13, 15, 18, 12, 10, 14, 17, 18, 12, 15, 14, 12,

[0169] 13, 11, 16, 18, 14, 15, 13};

[0170] Calculate the average inventory fluctuations of Product X and Product Y respectively:

[0171] Product X:

[0172] Product Y:

[0173] Calculate the deviation value for each time period:

[0174] Deviation value of Product X: That is, {-2.5, 3.5, 0.5, -4.5, 1.5, -0.5, -1.5, 2.5, 5.5, 3.5, -2.5, -0.5, 1.5, 4.5, -1.5, -3.5, 0.5, 3.5, 4.5, -0.5, 1.5, 0.5, -1.5, -0.5, -2.5, 2.5, 4.5, 0.5, 1.5, -0.5};

[0175] Deviation value of Product Y: That is, {-3.6, 1.4, -1.6, -5.6, 0.4, -0.6, -2.6, 2.4, 4.4, 3.4, -2.6, -0.6, 1.4, 4.4, -1.6, -3.6, 0.4, 3.4, 4.4, -1.6, 1.4, 0.4, -1.6, -0.6, -2.6, 2.4, 4.4, 0.4, 1.4, -0.6};

[0176] Calculate the variance and covariance of Product X and Product Y respectively:

[0177] Variance of Product X:

[0178] Variance of Product Y:

[0179] Covariance:

[0180] Calculate the inventory linkage impact coefficient:

[0181]

[0182] The results show that the inventory linkage impact coefficient between product X and product Y is 0.918, indicating a significant positive correlation between the inventory fluctuations of the two products. The inventory adjustment strategy can be further optimized.

[0183] S503: The following is the execution process of using the inventory linkage impact coefficient table to normalize the inventory linkage impact coefficients of associated products, adjusting the linkage parameters of e-commerce inventory according to the linkage relationship between product sales and inventory fluctuations, optimizing and adjusting the inventory level, optimizing the inventory configuration of the supply chain by time period, and obtaining the inventory linkage risk assessment results;

[0184] Normalize the inventory linkage impact coefficients of associated products. According to the data in the inventory linkage impact coefficient table, select the inventory linkage impact coefficients within a certain time period, calculate the maximum and minimum values, and map the inventory linkage impact coefficients to the range of 0 to 1 according to the normalization formula. The normalization formula is

[0185]

[0186] where N po is the normalized value of the inventory linkage impact coefficient, min(C) is the minimum value of the linkage impact coefficient, and max(C) is the maximum value of the linkage impact coefficient;

[0187] After normalization, use the weighted average algorithm to integrate the linkage coefficients of multiple product pairs, generate an optimized inventory configuration table by dynamically adjusting the linkage parameters of e-commerce inventory, and apply the inventory configuration table to the supply chain by time period to optimize the inventory allocation strategy in supply chain management, achieve more efficient inventory optimization, and obtain the inventory linkage risk assessment results.

[0188] Please refer to Figure 7 , on the other hand, an electric vehicle status monitoring system is provided. The electric vehicle status monitoring system is used to execute the above-mentioned electric vehicle status monitoring method. The system includes:

[0189] The fluctuation analysis module extracts sales volume, inventory, replenishment time interval, logistics distribution time, and demand forecasting error based on the transaction data of the e-commerce platform, supply chain link data, and holiday sales fluctuation data, and generates a multi-dimensional inventory scenario dataset;

[0190] The association relationship analysis module extracts the warehousing, transportation, and demand-side nodes in the e-commerce inventory supply chain based on the multi-dimensional inventory scenario dataset, calculates the node weight values, analyzes the association relationship between the warehousing and transportation nodes, and generates a risk propagation path analysis result;

[0191] The risk value calculation module uses the results of risk propagation path analysis, extracts the inventory level node, demand volume node, and logistics distribution duration node, calculates the risk values of the inventory level, demand volume, and logistics distribution duration nodes, and generates the inventory risk distribution identification result;

[0192] The inventory optimization module uses the inventory risk distribution identification result, calculates the difference between the dynamic change value of e-commerce inventory and the demand fluctuation amplitude value, and adjusts the upper and lower limits of the multi-period inventory level based on the difference result to generate the inventory dynamic adjustment result;

[0193] The linkage impact assessment module extracts the commodity sales correlation data and inventory fluctuation data based on the inventory dynamic adjustment result, calculates the inventory linkage impact coefficient of related commodities, optimizes the commodity inventory level, and generates the inventory linkage risk assessment result.

[0194] As mentioned above, it is only the specific implementation manner 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 all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. An e-commerce inventory risk assessment method based on big data, characterized in that: The following steps are involved: S1: Collect transaction data from e-commerce platforms, supply chain data, and holiday sales fluctuation data, extract sales volume, inventory volume, replenishment time interval, logistics delivery time, and demand forecast error, calculate the fluctuation amplitude value in the time series item by item, and obtain a multi-dimensional inventory scenario data set; S2: Using the multidimensional inventory scenario data set, extracting nodes in the e-commerce inventory supply chain, calculating the weight values ​​of the nodes, and evaluating the path connection risk based on the node connection relationship in the supply chain path, combining the weight difference value and the frequency of goods circulation in the path, to obtain the risk propagation path analysis results; S3: Using the risk propagation path analysis results, extract e-commerce inventory related nodes, evaluate the weight distribution between nodes, use big data to perform distribution analysis through the risk weights of nodes, and combine the risk distribution location and frequency annotation to obtain the risk distribution identification results; S4: According to the risk distribution identification result, the risk level nodes are screened, the inventory fluctuation cycle is divided into intervals according to the delivery speed, the dynamic change of inventory and the fluctuation range of demand are compared, and the upper and lower limits of the e-commerce inventory level are adjusted period by period to obtain the inventory dynamic adjustment result; S5: Calculate the inventory linkage impact coefficient of the associated products through the inventory dynamic adjustment result, adjust the linkage parameters of the e-commerce inventory according to the correlation coefficient, optimize the e-commerce inventory level, and generate the inventory linkage risk assessment result.

2. The e-commerce inventory risk assessment method based on big data according to claim 1 is characterized in that: The multidimensional inventory scenario data set includes sales volume data, inventory volume data, replenishment time interval data, logistics delivery time data, and demand forecast error data. The risk propagation path analysis result includes storage node weight value, transportation node weight value, demand-side node weight value, and corresponding risk propagation path. The risk distribution identification result includes inventory level node risk distribution, demand volume node risk distribution, and logistics delivery time node risk distribution. The inventory dynamic adjustment result includes risk level node screening result, inventory upper limit adjustment value, and inventory lower limit adjustment value. The inventory linkage risk assessment result includes commodity sales correlation data, inventory fluctuation linkage impact coefficient, and linkage parameter optimization adjustment value.

3. The e-commerce inventory risk assessment method based on big data according to claim 1 is characterized in that: Collect transaction data from e-commerce platforms, supply chain data, and holiday sales fluctuation data, extract sales volume, inventory volume, replenishment time interval, logistics delivery time, and demand forecast error, calculate the fluctuation amplitude value in the time series item by item, and obtain the multidimensional inventory scenario data set in the following steps: S101: Collect transaction data, supply chain data, and holiday sales fluctuation data from e-commerce platforms, screen sales volume, inventory, replenishment time interval, logistics delivery time, and demand forecast error one by one, arrange the data in chronological order by timestamp and format them uniformly, mark them by time interval and analyze the corresponding time node relationship, and generate a time series labeled data set; S102: Calculate the difference and change rate of sales volume, inventory volume and replenishment time interval at adjacent time nodes item by item according to the time series labeled data set, calculate the standard deviation by time node segment, calculate the fluctuation amplitude value, and generate a sales and inventory fluctuation data table; S103: Utilize the sales and inventory fluctuation data table to extract the fluctuation values ​​of logistics delivery time and demand forecast error and time nodes one by one, match them on the time axis, analyze the corresponding relationship between logistics delivery and demand forecast error, integrate the fluctuation data of time nodes according to the supply chain links, and obtain a multi-dimensional inventory scenario data set.

4. The e-commerce inventory risk assessment method based on big data according to claim 3 is characterized in that: The formula for calculating the fluctuation amplitude value is as follows: Among them, VA is the fluctuation amplitude value, x1 and x2 represent the sales volume at two adjacent time nodes, y1 and y2 represent the inventory volume at two adjacent time nodes, z1 and z2 represent the replenishment time interval between two adjacent time nodes, w1, w2, w3 are weight coefficients, and k is the smoothing adjustment coefficient.

5. The e-commerce inventory risk assessment method based on big data according to claim 1 is characterized in that: The multidimensional inventory scenario dataset is used to extract nodes in the e-commerce inventory supply chain link, calculate the weight value of the node, and evaluate the path connection risk based on the node connection relationship in the supply chain path, combined with the weight difference value and the frequency of goods circulation in the path. The specific steps for obtaining the risk propagation path analysis result are as follows: S201: Based on the multidimensional inventory scenario data set, extract the time series data of the storage node, the transportation node and the demand-side node, statistically summarize and standardize the sales volume, inventory volume and fluctuation amplitude values ​​of the differentiated nodes in the time series, summarize the logistics delivery time and demand forecast error of the corresponding nodes in segments according to the time series, and generate a node feature data set; S202: According to the node feature data set, the proportion of the fluctuation amplitude value of the storage node, the transportation node and the demand-side node to the total fluctuation value is calculated item by item, and the nodes are divided into supply chain links, and the weight of the proportion of logistics distribution time and the normalized value of the demand forecast error are weighted and calculated to obtain a node weight data table; S203: Using the node weight data table, extract the connection paths between differentiated nodes in the supply chain, calculate the cumulative weight value of each node on the path item by item, analyze the fluctuation amplitude value of the path according to the change in logistics delivery time between nodes, grade the path risk degree, and obtain the risk propagation path analysis result.

6. The e-commerce inventory risk assessment method based on big data according to claim 1 is characterized in that: Using the risk propagation path analysis results, extracting e-commerce inventory related nodes, evaluating the weight distribution between nodes, using big data, performing distribution analysis through the risk weights of nodes, and combining the risk distribution location and frequency annotations to obtain the risk distribution identification results are as follows: S301: Based on the risk propagation path analysis results, extract the risk weight values ​​of the inventory level, demand and logistics delivery time nodes, match the weight value of each node with the time series fluctuation amplitude value, segment the risk propagation path between nodes and mark the node type to obtain the node risk weight distribution data set; S302: Based on the node risk weight distribution data set, using big data, a multi-dimensional cross-analysis is performed on the time series fluctuation amplitude value and the node weight value, and the risk value of each type of node is classified and summarized according to the differentiated time period to generate an e-commerce inventory risk classification result; S303: Using the e-commerce inventory risk classification result, perform multidimensional distribution statistics on inventory level nodes, demand nodes and logistics delivery time nodes, calculate the risk distribution value of the node under multidimensional risk conditions, and obtain a risk distribution identification result.

7. The method for e-commerce inventory risk assessment based on big data according to claim 6 is characterized in that: The formula for calculating the risk distribution value of the node under multi-dimensional risk conditions is as follows: Among them, R ij (t) represents the risk distribution value of node i at time t, D ij (t) represents the demand data value corresponding to node i at time t, μ ij represents the average demand of node i at time t, L ij (t) represents the logistics delivery time of node i at time t, represents the average logistics delivery time of node i at time t, T ij (t) represents the inventory level of node i at time t, ∈ is the smoothing factor, and ω1, ω2, ω3 are weight coefficients.

8. The e-commerce inventory risk assessment method based on big data according to claim 1 is characterized in that: According to the risk distribution identification result, the risk level nodes are screened, the inventory fluctuation cycle is divided into intervals according to the delivery speed, the dynamic changes of inventory and the fluctuation range of demand are compared, and the upper and lower limits of the e-commerce inventory level are adjusted period by period. The steps of obtaining the inventory dynamic adjustment result are as follows: S401: Using the risk distribution identification result, screening the inventory level nodes, demand nodes and logistics delivery speed nodes with priority risk levels, sorting and segmenting the risk values ​​and time series fluctuation amplitude values ​​of the nodes, marking the nodes whose risk values ​​exceed the target threshold, and generating a risk node data set; S402: Based on the risk node data set, extract the dynamic change value of the inventory level node and the fluctuation amplitude value of the demand node in each period, calculate the difference and proportional change between the two, and generate the inventory adjustment interval; S403: Through the inventory adjustment interval, the upper and lower limits of the inventory level are dynamically adjusted time period by time period, and the key nodes and time periods where the inventory level needs to be optimized in the supply chain are screened in combination with the demand fluctuation amplitude value and the distribution speed change trend, to obtain the inventory dynamic adjustment result.

9. The e-commerce inventory risk assessment method based on big data according to claim 1 is characterized in that: The steps of calculating the inventory linkage impact coefficient of the associated products through the inventory dynamic adjustment result, adjusting the linkage parameters of the e-commerce inventory according to the correlation coefficient, optimizing the e-commerce inventory level, and generating the inventory linkage risk assessment result are as follows: S501: extracting commodity sales correlation data and inventory fluctuation data according to the inventory dynamic adjustment result, performing multi-dimensional matching on the sales volume and inventory fluctuation value of the related commodities, evaluating the synchronization between the sales volume change of commodities and the inventory level of e-commerce, and generating a commodity correlation data set; S502: using the commodity correlation data set, calculating the inventory linkage influence coefficients between the related commodities one by one, segmenting by time series, analyzing the contribution value of inventory fluctuation to the linkage influence coefficient, and quantifying the linkage degree, and generating an inventory linkage influence coefficient table; S503: Using the inventory linkage impact coefficient table, the inventory linkage impact coefficients of related commodities are normalized, the linkage parameters of the e-commerce inventory are adjusted according to the linkage relationship between commodity sales and inventory fluctuations, the inventory level is optimized, the supply chain inventory configuration is optimized time period by time, and the inventory linkage risk assessment result is obtained.

10. The e-commerce inventory risk assessment system based on big data is characterized by: According to the big data-based e-commerce inventory risk assessment method according to any one of claims 1 to 9, the system comprises: The fluctuation analysis module extracts sales volume, inventory volume, replenishment time interval, logistics delivery time and demand forecast error based on the transaction data of the e-commerce platform, supply chain link data and holiday sales fluctuation data, and generates a multi-dimensional inventory scenario data set; The association analysis module extracts the warehousing, transportation and demand-side nodes in the e-commerce inventory supply chain based on the multi-dimensional inventory scenario data set, calculates the node weight value, analyzes the association relationship between the warehousing and transportation nodes, and generates risk propagation path analysis results; The risk value calculation module uses the risk propagation path analysis results to extract inventory level nodes, demand nodes and logistics delivery time nodes, calculates the risk values ​​of inventory level, demand and logistics delivery time nodes, and generates inventory risk distribution identification results; The inventory optimization module uses the inventory risk distribution identification result to calculate the difference between the dynamic change value of the e-commerce inventory and the demand fluctuation amplitude value, adjusts the upper and lower limits of the multi-period inventory level based on the difference result, and generates the inventory dynamic adjustment result; The linkage impact assessment module extracts commodity sales correlation data and inventory fluctuation data based on the inventory dynamic adjustment result, calculates the inventory linkage impact coefficient of the related commodities, optimizes the commodity inventory level, and generates the inventory linkage risk assessment result.

Images