Data analysis system based on digital enterprise management

Abstract

The invention relates to the technical field of digital management, in particular to a data analysis system based on digital enterprise management. The system comprises a data acquisition module, a model establishment module, an inventory adjustment module and a risk assessment module. According to the invention, competitor data, market promotion data and raw material supply chain data are collected through the data acquisition module, a neural network model for predicting enterprise sales is established by using the model establishment module, and the inventory adjustment module calculates the order quantity of the product by using an economic order quantity model according to the predicted enterprise sales. The maximum inventory level of an enterprise is determined through a reorder point algorithm and safe inventory setting, the risk assessment module comprehensively analyzes various indexes by using an analytic hierarchy process to obtain a comprehensive risk assessment result, an inventory optimization strategy is formulated according to the assessment result, and the risk of the enterprise is optimized. The inventory turnover rate is improved, the customer satisfaction is improved, and the market competitiveness is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of digital management, and more specifically, to a data analysis system based on digital enterprise management. Background Art

[0002] Digital enterprise management refers to the management mode in which enterprises use digital technologies such as big data, artificial intelligence, and cloud computing. Big data technology can collect, store, and process massive amounts of data, including structured and unstructured data such as text, images, and audio. Artificial intelligence technology, especially machine learning and deep learning algorithms, can achieve automated prediction, classification, and optimization based on big data. Cloud computing technology provides powerful infrastructure support for the application of big data and artificial intelligence, optimizing, coordinating, and automating internal management processes and business processes to improve enterprise operation efficiency and decision-making quality.

[0003] With the rapid development of the Internet, market data is changing every moment. If an enterprise always maintains the same inventory level, it will lead to a large number of products stacking up in the warehouse when the market demand decreases, resulting in waste of resources, and when the market demand increases, the inventory quantity of products will be insufficient, resulting in the loss of customer sources. In order to be able to predict the sales volume of an enterprise based on the real-time data changes in the market, dynamically adjust the inventory level of the enterprise, and use normative analysis to conduct risk assessment on the inventory, thereby improving inventory turnover, enhancing customer satisfaction, and strengthening market competitiveness. Therefore, we propose a data analysis system based on digital enterprise management. Summary of the Invention

[0004] The purpose of the present invention is to solve the problem that the single inventory management method is difficult to cope with the changing market due to the constantly changing market demand, resulting in waste of enterprise resources and loss of customer sources. In order to be able to predict the sales volume of an enterprise based on the real-time data changes in the market, dynamically adjust the inventory level of the enterprise, and use normative analysis to conduct risk assessment on the inventory, thereby improving inventory turnover, enhancing customer satisfaction, and strengthening market competitiveness.

[0005] To achieve the above object, the present invention provides a data analysis system based on digital enterprise management, including a data collection module, a model establishment module, an inventory adjustment module, and a risk assessment module;

[0006] The data collection module uses a big data platform to collect competitor data, market promotion data, and raw material supply chain data, cleans and standardizes the collected data, and transfers the processed data to the model establishment module;

[0007] The model establishment module uses the principal component analysis method to extract features from the data collected by the data collection module. Taking the promotional activities of competitors, the overall demand trend of the industry, and the raw material price fluctuations as the market dynamic information features, a neural network model for predicting the enterprise sales volume is established. The Adam optimization algorithm is adopted to train this neural network model with the training set data, and the performance of this neural network model is evaluated using the validation set data.

[0008] The inventory adjustment module calculates the order quantity of the product using the economic order quantity model according to the enterprise sales volume predicted by the neural network model, combined with the unit cost of the product and the warehouse storage cost, and determines the maximum inventory level of the enterprise through the reorder point algorithm and safety inventory setting.

[0009] The risk assessment module takes the inventory turnover rate, safety inventory level, and inventory cost as evaluation indicators to conduct a risk assessment on the inventory level determined by the inventory adjustment module. The analytic hierarchy process is used to comprehensively analyze each indicator to obtain a comprehensive risk assessment result, and inventory optimization strategies, supply chain collaboration strategies, and emergency management strategies are formulated according to the assessment results.

[0010] As a further improvement of this technical solution, the model establishment module includes a model establishment unit and a training optimization unit.

[0011] The model establishment unit takes the promotional activities of competitors, the overall demand trend of the industry, and the raw material price fluctuations as the input layer nodes, and takes the prediction of the enterprise sales volume as the output layer node to establish a neural network model for predicting the enterprise sales volume.

[0012] The training optimization unit uses the Adam optimization algorithm to update the weight parameters of the neural network model to minimize the loss function, evaluates the performance of this neural network model using the validation set data, and adjusts the hyperparameters through the loss value and evaluation indicators on the validation set.

[0013] As a further improvement of this technical solution, when the model establishment unit establishes a neural network model for predicting the enterprise sales volume, it selects the ReLU function as the activation function of the hidden layer and uses the Sigmoid function to normalize the output layer.

[0014] As a further improvement of this technical solution, the model establishment unit uses the mean square error as the loss function to measure the difference between the predicted value and the true value, and its formula is:

[0015] ;

[0016] where is the mean square error, is the number of samples, is the true value. is the predicted value.

[0017] As a further improvement of this technical solution, the training optimization unit inputs the training set data into the neural network model, calculates the predicted value through forward propagation, then calculates the loss value according to the loss function, and updates the weight parameters of the model through the backpropagation algorithm until the loss function converges.

[0018] As a further improvement of this technical solution, the inventory adjustment module converts the predicted sales amount into the predicted product demand quantity according to the unit selling price of the product, and calculates the order quantity of this product using the economic order quantity model. The formula is:

[0019] ;

[0020] where is the economic order quantity, is the annual demand quantity, is the cost per order, is the annual storage cost per unit product.

[0021] As a further improvement of this technical solution, the inventory adjustment module determines the safety stock quantity according to the standard deviation of historical demand data and the service level. The formula is:

[0022] ;

[0023] where is the safety stock quantity, is the safety factor, is the daily demand quantity of the standard deviation, is the order lead time.

[0024] As a further improvement of this technical solution, the risk assessment module includes a comprehensive assessment unit and an inventory optimization unit;

[0025] The comprehensive assessment unit uses the inventory turnover rate, safety stock level, and inventory cost as evaluation indicators to conduct a risk assessment on the determined inventory level, and uses the analytic hierarchy process to comprehensively analyze each indicator to obtain a comprehensive risk assessment result;

[0026] The inventory optimization unit formulates inventory optimization strategies, supply chain collaboration strategies, and emergency management strategies according to the evaluation results.

[0027] As a further improvement of this technical solution, when the comprehensive assessment unit comprehensively analyzes each indicator using the analytic hierarchy process, it uses the scale method to compare the elements of different layers and constructs a judgment matrix.

[0028] As a further improvement of the present technical solution, when formulating an optimization strategy based on the evaluation results, the inventory optimization unit divides the evaluation results into low-risk situations, medium-risk situations, and high-risk situations.

[0029] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0030] 1. The data analysis system based on digital enterprise management collects competitor data, market promotion data, and raw material supply chain data through the data collection module, cleans and standardizes the collected data, establishes a neural network model for predicting enterprise sales through the model establishment module, trains the neural network model with the training set data, and evaluates the performance of the neural network model with the validation set data. The inventory adjustment module calculates the order quantity of the product using the economic order quantity model based on the predicted enterprise sales, determines the maximum inventory level of the enterprise through the reorder point algorithm and safety inventory setting, predicts the enterprise sales according to the real-time data changes in the market, and dynamically adjusts the inventory level of the enterprise, improving the inventory turnover rate, enhancing customer satisfaction, and strengthening market competitiveness.

[0031] 2. The risk assessment module comprehensively analyzes various indicators using the analytic hierarchy process to obtain a comprehensive risk assessment result, formulates an inventory optimization strategy based on the evaluation results to optimize the inventory level. At the same time, in different risk situations, corresponding supply chain collaboration strategies and emergency management strategies are selected to optimize the production plan and logistics distribution plan, improving the supply efficiency of the supply chain and reducing costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a schematic diagram of the overall process of the present invention;

[0033] Figure 2 It is a schematic diagram of the overall detailed process of the present invention.

[0034] The meanings of the various reference numerals in the figure are as follows:

[0035] 100, data collection module; 200, model establishment module; 210, model establishment unit; 220, training and optimization unit; 300, inventory adjustment module; 400, risk assessment module; 410, comprehensive evaluation unit; 420, inventory optimization unit. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the protection scope of the present invention.

[0037] At present, the market demand is constantly changing. A single inventory management method is difficult to cope with market changes, resulting in waste of enterprise resources and loss of customer sources. In order to predict the sales volume of an enterprise based on real-time data changes in the market, dynamically adjust the inventory level of the enterprise, and use normative analysis to evaluate the inventory risk, thereby improving the inventory turnover rate, enhancing customer satisfaction, and strengthening market competitiveness.

[0038] Therefore, the present invention proposes to collect competitor data, market promotion data, and raw material supply chain data through a data collection module, establish a neural network model for predicting the sales volume of an enterprise using a model establishment module, calculate the order quantity of the product using the economic order quantity model according to the predicted sales volume of the enterprise by the inventory adjustment module, determine the maximum inventory level of the enterprise through the reorder point algorithm and safety inventory setting, and use the analytic hierarchy process for comprehensive analysis of various indicators by the risk assessment module to obtain a comprehensive risk assessment result, and formulate an inventory optimization strategy according to the assessment result.

[0039] Specifically as follows:

[0040] Please refer to Figure 1 As shown, the present invention provides a data analysis system based on digital enterprise management, including a data collection module 100, a model establishment module 200, an inventory adjustment module 300, and a risk assessment module 400;

[0041] The data collection module 100 uses a big data platform to collect competitor data, market promotion data, and raw material supply chain data, cleans and standardizes the collected data, and transfers the processed data to the model establishment module 200;

[0042] The model establishment module 200 uses the principal component analysis method to extract features from the data collected by the data collection module 100, takes the promotional activities of competitors, the overall demand trend of the industry, and the raw material price fluctuations as the market dynamic information features, establishes a neural network model for predicting the sales volume of an enterprise, adopts the Adam optimization algorithm, trains the neural network model with the training set data, and evaluates the performance of the neural network model with the validation set data;

[0043] The inventory adjustment module 300 calculates the order quantity of the product using the economic order quantity model according to the predicted sales volume of the enterprise by the neural network model, combines the unit cost of the product and the warehouse storage cost, and determines the maximum inventory level of the enterprise through the reorder point algorithm and safety inventory setting;

[0044] The risk assessment module 400 takes the inventory turnover rate, safety inventory level, and inventory cost as evaluation indicators, conducts a risk assessment on the inventory level determined by the inventory adjustment module 300, comprehensively analyzes each indicator using the analytic hierarchy process, obtains a comprehensive risk assessment result, and formulates inventory optimization strategies, supply chain collaboration strategies, and emergency management strategies based on the assessment result;

[0045] The data collection module 100 uses a big data platform to collect competitor data, market promotion data, and raw material supply chain data, checks for missing values and outliers in the data. For missing values, methods such as mean filling, median filling, or prediction filling based on other features can be used. For outliers, they can be corrected or deleted according to the actual situation, and the data is standardized so that the mean of each feature is 0 and the standard deviation is 1.

[0046] As Figure 2 shown, among them, the model establishment module 200 includes a model establishment unit 210 and a training and optimization unit 220;

[0047] The model establishment unit 210 takes the promotional activities of competitors, the overall demand trend in the industry, and raw material price fluctuations as input layer nodes, and takes predicting the enterprise sales amount as the output layer node, and establishes a neural network model for predicting the enterprise sales amount;

[0048] The training and optimization unit 220 uses the Adam optimization algorithm to update the weight parameters of the neural network model, which is used to minimize the loss function, evaluates the performance of the neural network model using the validation set data, and adjusts the hyperparameters through the loss value and evaluation indicators on the validation set;

[0049] The preprocessed data is divided into a training set, a validation set, and a test set, and is divided according to a ratio of 7:2:1. The training set is used to train the model, the validation set is used to adjust the model hyperparameters and prevent overfitting, and the test set is used to evaluate the final performance of the model;

[0050] Set hyperparameters such as the learning rate, the number of iterations, and the batch size. The learning rate is generally between 0.001 - 0.1 and can be adjusted according to the training situation. The number of iterations is determined according to the size of the dataset and the complexity of the model, usually between several hundred and several thousand times. The batch size is generally selected as a power of 2, such as 32, 64, 128, etc.

[0051] In order to better select the activation functions of the hidden layer and the output layer, among them, when the model establishment unit 210 establishes a neural network model for predicting the enterprise sales amount, it selects the ReLU function as the activation function of the hidden layer and uses the Sigmoid function to normalize the output layer;

[0052] The ReLU function has advantages such as simple calculation and fast convergence speed. It can effectively solve the problem of gradient disappearance, making the neural network easier to train. According to the complexity of the problem and the characteristics of the data, determine the number of hidden layers of the neural network and the number of neurons in each layer. Generally speaking, for the enterprise sales prediction problem, you can first try using 1-3 hidden layers, and the number of neurons in each layer is between dozens and hundreds, such as the common 64, 128, 256, etc. Adjust through experiments to find the optimal structure;

[0053] During the forward propagation process, each neuron in the hidden layer receives the output of the previous layer as input. After linear transformation (such as matrix multiplication and addition), it is then activated through the ReLU function;

[0054] The Sigmoid function can map any real number to a relatively small interval to achieve a normalization effect. It is often used to convert the output of the neural network into a probability or proportion form. According to the specific requirements of sales prediction, determine the number of neurons in the output layer. If it is to predict a single sales value, usually there is only one neuron in the output layer. If you want to predict multiple related sales indicators, such as the sales of different products or the sales in different regions, etc., then the number of neurons in the output layer is equal to the number of indicators to be predicted;

[0055] During the forward propagation process, the neurons in the output layer receive the output of the hidden layer as input. After linear transformation, they are then activated through the Sigmoid function to obtain the normalized output result.

[0056] In order to better measure the difference between the predicted value and the true value, among them, the model unit 210 uses the mean square error as the loss function to measure the difference between the predicted value and the true value. Its formula is:

[0057] ;

[0058] Among them, is the mean square error, is the number of samples, is the true value, is the predicted value.

[0059] The mean square error (MSE) refers to the average of the squares of the differences between the predicted value and the true value. During the training process of models such as neural networks, first through forward propagation, the input data passes through the calculations of each hidden layer and the action of the activation function, and finally the predicted value of the output layer is obtained . For each sample, calculate the corresponding predicted value according to the structure and parameters of the model;

[0060] According to the above formula, for each sample, the true value and the predicted value Substitute and calculate the sum of squared errors for all samples, and then divide by the number of samples , to obtain the mean squared error . This value reflects the overall fitting degree of the model to the training data under the current parameters. The smaller the value, the closer the predicted value is to the true value, and the better the performance of the model.

[0061] In order to better approximate the predicted result to the true value, among them, the training optimization unit 220 inputs the training set data into the neural network model, calculates the predicted value through forward propagation, then calculates the loss value according to the loss function, and then updates the weight parameters of the model through the backpropagation algorithm until the loss function converges;

[0062] Taking the mean squared error as the objective function, calculate the gradients of the mean squared error with respect to each parameter (such as weights and biases) in the model through the backpropagation algorithm. According to the direction and magnitude of the gradients, use the Adam optimization algorithm to update the parameters of the model, so that the mean squared error gradually decreases during the iteration process, thereby making the predicted result of the model continuously approach the true value;

[0063] Repeat the above process of forward propagation, calculating the loss value, and backpropagation to update the weight parameters. Each iteration will make the loss function value change in the decreasing direction. By continuously iterating, observe the change of the loss function value. When the loss function value no longer decreases significantly or reaches conditions such as the preset number of iterations and accuracy requirements, it is considered that the model converges and the training process ends.

[0064] In order to better calculate the order quantity of the product, among them, the inventory adjustment module 300 converts the predicted sales amount into the predicted product demand according to the sales unit price of the product, and uses the economic order quantity model to calculate the order quantity of the product. The formula is:

[0065] ;

[0066] Among them, is the economic order quantity, is the annual demand, is the cost per order, is the annual storage cost per unit product.

[0067] The economic order quantity model is a method for determining the optimal order quantity of an enterprise, aiming to minimize the inventory cost.

[0068] First, according to the sales unit price of the product, convert the predicted sales amount into the predicted product demand. The calculation formula is: Predicted demand = Predicted sales amount ÷ Product sales unit price. For example, if the predicted sales amount of a certain product is 1 million yuan and the product sales unit price is 100 yuan, then the predicted demand is 10,000 pieces;

[0069] Unit cost refers to the procurement cost or production cost per unit of product, including costs directly related to product production or procurement such as raw materials, labor, and manufacturing expenses. This value is determined through the enterprise's financial records, procurement contracts, etc.;

[0070] Storage costs include warehousing expenses, capital occupancy costs, insurance premiums, goods losses, etc. Storage costs are usually expressed as the storage cost per unit of product per year and can be obtained by calculating the sum of various storage costs and allocating them to each unit of product. For example, if the total storage costs such as warehousing expenses and capital occupancy costs of an enterprise in a year are 100,000 yuan and the average annual inventory quantity is 10,000 units, then the annual storage cost per unit of product is 10 yuan;

[0071] Substituting the above data into the economic order quantity formula gives: ; Rounding to 316 units.

[0072] In order to better determine the safety stock quantity, among them, the inventory adjustment module 300 determines the safety stock quantity according to the standard deviation of historical demand data and the service level, and its formula is:

[0073] ;

[0074] Among them, is the safety stock quantity, is the safety factor, is the daily demand is the standard deviation of is the lead time.

[0075] To cope with demand fluctuations and supply uncertainties, enterprises usually set safety stocks. There are various calculation methods for safety stocks. A commonly used method is to determine based on the standard deviation of historical demand data and the service level. For example, according to historical data, the standard deviation of demand is calculated to be 10 units, the enterprise hopes to achieve a service level of 95%, its safety factor is 1.645, and the lead time is 5 days;

[0076] Substituting the above data into the safety stock formula gives: ; Rounding to 37 units.

[0077] The maximum inventory level refers to the upper limit of inventory, and the calculation formula is: ; In the above example, when the EOQ is taken as 316 units and the safety stock is 37 units, the maximum inventory level is: ;

[0078] Among them, the risk assessment module 400 includes a comprehensive assessment unit 410 and an inventory optimization unit 420;

[0079] The comprehensive evaluation unit 410 takes inventory turnover rate, safety inventory level, and inventory cost as evaluation indicators, conducts risk assessment on the determined inventory level, comprehensively analyzes each indicator using the analytic hierarchy process, and obtains a comprehensive risk assessment result;

[0080] The inventory optimization unit 420 formulates inventory optimization strategies, supply chain collaboration strategies, and emergency management strategies according to the evaluation results;

[0081] Using the analytic hierarchy process to comprehensively analyze each indicator and obtain the comprehensive risk assessment result mainly includes steps such as establishing a hierarchical structure model, constructing a judgment matrix, calculating the weight vector, consistency test, and calculating the comprehensive risk assessment result;

[0082] Goal layer: Clearly define the overall goal to be evaluated, that is, the comprehensive risk assessment result;

[0083] Criterion layer: Determine the main factors or criteria affecting risk assessment, such as market risk, technology risk, management risk, etc.;

[0084] Index layer: Further refine each criterion into specific quantifiable or qualitative indicators. Taking market risk as an example, it may include indicators such as the number of competitors, market demand fluctuations, product price fluctuations, etc.;

[0085] In order to better construct the judgment matrix, among them, when the comprehensive evaluation unit 410 comprehensively analyzes each indicator using the analytic hierarchy process, the scale method is used to compare the elements of different layers and construct the judgment matrix;

[0086] For elements at the same level, by comparing them pairwise for their importance to an element in the upper level, a judgment matrix is constructed. Usually, the 1-9 scale method is used to measure the relative importance degree, and the larger the number, the more important the former is than the latter. For example, when comparing the importance of the number of competitors and market demand fluctuations under market risk to market risk, if it is considered that the number of competitors is slightly more important than market demand fluctuations, a value of 3 can be assigned to the corresponding position in the judgment matrix, and vice versa, a value of 1 / 3 is assigned. If the two are equally important, a value of 1 is assigned;

[0087] Score each indicator in the index layer: According to actual data or expert judgment, conduct a risk score for each indicator. The 0-10 scoring system can be used, and the higher the score, the greater the risk. Suppose the score of the number of competitors indicator is 7 points, the score of the market demand fluctuation indicator is 6 points, and the score of the product price fluctuation indicator is 5 points;

[0088] Calculate the risk values at the criterion layer: Based on the weights and scores of each indicator in the indicator layer, calculate the risk values of each criterion in the criterion layer. Taking market risk as an example, market risk value = weight of the number of competitors × score of the number of competitors + weight of market demand fluctuation × score of market demand fluctuation + weight of product price fluctuation × score of product price fluctuation;

[0089] Calculate the comprehensive risk assessment result: According to the same method, based on the weights of each criterion in the criterion layer and the calculated risk values, calculate the comprehensive risk assessment result of the target layer.

[0090] In order to better formulate optimization strategies, among them, when the inventory optimization unit 420 formulates optimization strategies according to the evaluation results, the evaluation results are divided into low-risk situations, medium-risk situations, and high-risk situations;

[0091] Low-risk situation: If the comprehensive risk score is relatively low, it indicates that the overall operation status is relatively stable. At this time, classic methods such as the Economic Order Quantity (EOQ) model can be used to determine the order quantity, maintain a relatively low safety inventory level to reduce inventory holding costs. At the same time, the inventory turnover rate can be appropriately increased to speed up the capital turnover;

[0092] Medium-risk situation: When the risk is at a medium level, it is necessary to seek a balance between cost and risk. A strategy combining periodic inventory checks and quantitative replenishment can be adopted, and the safety inventory level can be adjusted according to risk factors. For materials with relatively high risks, the safety inventory is appropriately increased, and for materials with relatively low risks, the normal inventory level is maintained;

[0093] High-risk situation: If the evaluation results show a relatively high risk, such as facing unstable raw material supply, large fluctuations in market demand, etc., the safety inventory level should be increased to ensure the continuity of the supply of key materials. At the same time, shorten the order cycle, increase the order frequency, and adopt flexible ordering methods, such as signing emergency replenishment agreements with suppliers, etc.

[0094] In summary, the working principle of this solution is as follows:

[0095] The data analysis system based on digital enterprise management collects competitor data, market promotion data, and raw material supply chain data through the data collection module 100, cleans and standardizes the collected data, uses the model building module 200 to establish a neural network model for predicting enterprise sales, trains the neural network model with the training set data, and evaluates the performance of the neural network model with the validation set data. The inventory adjustment module 300 calculates the order quantity of the product using the economic order quantity model based on the predicted enterprise sales, determines the maximum inventory level of the enterprise through the reorder point algorithm and safety stock setting, predicts the enterprise sales according to the real-time data changes in the market, and dynamically adjusts the inventory level of the enterprise to improve the inventory turnover rate, enhance customer satisfaction, and strengthen market competitiveness. The risk assessment module 400 uses the analytic hierarchy process to comprehensively analyze various indicators, obtains a comprehensive risk assessment result, formulates an inventory optimization strategy based on the assessment result, optimizes the inventory level. At the same time, in different risk situations, corresponding supply chain collaboration strategies and emergency management strategies are selected to optimize the production plan and logistics distribution plan, improve the supply efficiency of the supply chain, and reduce costs.

[0096] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art of this industry should understand that the present invention is not limited by the above embodiments. The above embodiments and the descriptions in the specification are only preferred examples of the present invention and are not used to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. Data analysis system based on digital enterprise management, characterized by: It includes a data collection module (100), a model building module (200), an inventory adjustment module (300) and a risk assessment module (400); The data collection module (100) uses the big data platform to collect competitor data, marketing data and raw material supply chain data, cleans and standardizes the collected data, and transmits the processed data to the model building module (200); The model building module (200) uses the principal component analysis method to extract features from the data collected by the data collection module (100), uses the promotional activities of competitors, the overall demand trend of the industry and the fluctuation of raw material prices as market dynamic information features, establishes a neural network model for predicting corporate sales, uses the Adam optimization algorithm to train the neural network model using training set data, and uses the validation set data to evaluate the performance of the neural network model; The inventory adjustment module (300) calculates the order quantity of the product using the economic order quantity model based on the enterprise sales predicted by the neural network model, combined with the unit cost of the product and the warehouse storage cost, and determines the maximum inventory level of the enterprise through the reorder point algorithm and safety stock setting; The risk assessment module (400) uses inventory turnover rate, safety inventory level and inventory cost as assessment indicators to conduct risk assessment on the inventory level determined by the inventory adjustment module (300), and uses the hierarchical analysis method to conduct a comprehensive analysis of various indicators to obtain a comprehensive risk assessment result, and formulates an inventory optimization strategy, a supply chain collaboration strategy and an emergency management strategy based on the assessment result.

2. The data analysis system based on digital enterprise management according to claim 1, characterized in that: The model building module (200) includes a model building unit (210) and a training optimization unit (220); The model building unit (210) uses competitor promotion activities, industry overall demand trends and raw material price fluctuations as input layer nodes and uses predicted enterprise sales as output layer nodes to build a neural network model for predicting enterprise sales; The training optimization unit (220) uses the Adam optimization algorithm to update the weight parameters of the neural network model to minimize the loss function, uses the validation set data to evaluate the performance of the neural network model, and adjusts the hyperparameters based on the loss value and evaluation index on the validation set.

3. The data analysis system based on digital enterprise management according to claim 2 is characterized in that: When establishing a neural network model for predicting corporate sales, the model building unit (210) selects a ReLU function as an activation function of a hidden layer and uses a Sigmoid function to normalize an output layer.

4. The data analysis system based on digital enterprise management according to claim 2, characterized in that: The model building unit (210) uses mean square error as a loss function to measure the difference between the predicted value and the true value, and its formula is: ； in, is the mean square error, is the sample size, is the true value, is the predicted value.

5. The data analysis system based on digital enterprise management according to claim 2, characterized in that: The training optimization unit (220) inputs the training set data into the neural network model, calculates the prediction value through forward propagation, then calculates the loss value according to the loss function, and then updates the weight parameters of the model through the back propagation algorithm until the loss function converges.

6. The data analysis system based on digital enterprise management according to claim 1, characterized in that: The inventory adjustment module (300) converts the predicted sales volume into the predicted product demand volume according to the sales unit price of the product, and calculates the order quantity of the product using the economic order quantity model, the formula of which is: ； in, is the economic order quantity, The annual demand is is the cost per order, is the annual storage cost per unit product.

7. The data analysis system based on digital enterprise management according to claim 1 is characterized by: The inventory adjustment module (300) determines the safety inventory number according to the standard deviation of historical demand data and the service level, and its formula is: ； in, is the safety stock number, is the safety factor, Daily requirement The standard deviation of For order lead time.

8. The data analysis system based on digital enterprise management according to claim 1, characterized in that: The risk assessment module (400) includes a comprehensive assessment unit (410) and an inventory optimization unit (420); The comprehensive evaluation unit (410) uses inventory turnover rate, safety inventory level and inventory cost as evaluation indicators to conduct risk evaluation on the determined inventory level, and uses the hierarchical analysis method to conduct a comprehensive analysis on various indicators to obtain a comprehensive risk evaluation result; The inventory optimization unit (420) formulates an inventory optimization strategy, a supply chain collaboration strategy and an emergency management strategy according to the evaluation results.

9. The data analysis system based on digital enterprise management according to claim 8, characterized in that: When the comprehensive evaluation unit (410) uses the hierarchical analysis method to conduct a comprehensive analysis on various indicators, a scaling method is used to compare elements at different levels to construct a judgment matrix.

10. The data analysis system based on digital enterprise management according to claim 8, characterized in that: When the inventory optimization unit (420) formulates an optimization strategy based on the evaluation results, the evaluation results are divided into low risk situations, medium risk situations and high risk situations.

Images