Computer data processing method and system based on big data

Abstract

The invention relates to the technical field of data processing, in particular to a computer data processing method and system based on big data. The method comprises the following steps: firstly, collecting production and marketing historical data and market fluctuation data, respectively constructing a first data set and a second data set, and carrying out data processing; secondly, constructing a raw material demand prediction model by using the data in the first data set; and then, constructing a product score prediction model by using the second data set to obtain a product market score. Thirdly, according to the raw material requirement and the first data set, a cost objective function is constructed, and a minimum cost value is obtained; according to the first data set and the product market score, constructing an income objective function, and obtaining a maximum income value; and according to the first data set and the raw material demand, constructing a replenishment objective function, and obtaining a minimum replenishment value. And finally, carrying out optimization calculation on the cost objective function, the income objective function and the replenishment objective function to obtain an optimal solution, and obtaining a supply chain strategy of the minimum cost, the maximum income and the minimum replenishment.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a computer data processing method and system based on big data. Background Art

[0002] In today's e-commerce era, accurate demand forecasting can help companies understand customer purchasing trends and needs, so as to adjust inventory in a timely manner and ensure adequate supply of products. In addition, companies can use effective supply chain management to ensure that the entire process from suppliers to consumers is smooth and efficient. This includes optimizing the coordination of various links in the supply chain, reducing delivery time and transportation costs, and improving response speed and flexibility. These measures not only help companies reduce the risk of business interruption, but also improve their market competitiveness and adaptability.

[0003] In the prior art, traditional forecasting models include moving average method, exponential smoothing method and simple regression analysis model based on historical sales data, which can help enterprises to forecast customer demand to a certain extent. However, these traditional methods usually ignore some key external factors, such as market trends, seasonal changes and other environmental variables, which may make the forecast results inaccurate. In terms of inventory management, the traditional economic order quantity (EOQ) model and the method based on safety stock level are widely used. These methods help enterprises determine the optimal order quantity and inventory level, thereby reducing inventory costs to a certain extent. However, in the face of uncertain market environment, demand fluctuations and emergencies, these traditional methods are often insufficient and cannot fully cope with complex market conditions.

[0004] With the development of big data technology, the limitations of many traditional methods have been significantly improved. For example, advanced machine learning algorithms such as time series forecasting and neural network models can more effectively analyze historical sales data and consumer behavior, thereby providing more accurate future demand forecasts. Modern technology also helps companies make more data-driven decisions by analyzing large amounts of structured and unstructured data to discover potential demand patterns and trends. In addition, companies have begun to use streaming data processing technologies such as Apache Kafka and Apache Flink to monitor inventory levels and supply chain status in real time so that they can quickly adjust strategies to respond to rapidly changing market environments. The application of these technologies has improved the timeliness and accuracy of forecasts and decisions, however, there are still some defects, such as failure to fully take into account market fluctuations, emergencies, and other external uncertainties, which may affect the comprehensiveness and effectiveness of forecasts.

[0005] Therefore, a computer data processing method and system based on big data are proposed. Summary of the invention

[0006] The object of the present invention is to provide a computer data processing method and system based on big data, which is used to improve the efficiency of goods management of enterprises. First, historical production and sales data are collected to construct a first data set; market fluctuation data are collected as a second data set. Then, the first data set and the second data set are processed, and different data are time-aligned. Next, a raw material demand forecasting model is constructed, and the data in the first data set is used to predict the raw material demand N; and the second data set is used to construct a product score forecasting model to obtain the market score P of the product. Then, according to the raw material demand N and the first data set, a cost objective function F1 is constructed to obtain the minimum cost value min(F1); according to the first data set and the market score P of the product, a revenue objective function F2 is constructed to obtain the maximum revenue value max(F2); according to the first data set and the raw material demand N, a replenishment objective function F3 is constructed to obtain the minimum replenishment value min(F3). Finally, the cost objective function, the revenue objective function and the replenishment objective function are optimized to obtain the optimal solution.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A computer data processing method based on big data, comprising:

[0009] Collect historical production and sales data to construct the first data set; collect market fluctuation data as the second data set;

[0010] Performing data processing on the first data set and the second data set, and performing time alignment on different data;

[0011] Constructing a raw material demand prediction model, and using the data in the first data set to predict the raw material demand N;

[0012] Using the second data set, construct a product score prediction model to obtain the market score P of the product;

[0013] According to the raw material demand N and the first data set, a cost objective function F1 is constructed to obtain a minimum cost value min(F1);

[0014] According to the first data set and the market score P of the product, a profit objective function F2 is constructed to obtain a maximum profit value max(F2);

[0015] According to the first data set and the raw material demand N, a replenishment objective function F3 is constructed to obtain a minimum replenishment value min(F3);

[0016] The cost objective function, the revenue objective function and the replenishment objective function are optimized and calculated to obtain an optimal solution.

[0017] Furthermore, the production and sales historical data include the raw material supply of the production workshop, raw material cost, product production volume, product production process cost, raw material replenishment, warehouse inventory, warehouse storage cost, product sales volume and product sales price; the market fluctuation data include seasonal impact rate, brand impact rate, market average price and technical impact factor.

[0018] Furthermore, the specific process of the time alignment is: first, ensure that the timestamps of all data use the same format and time zone; second, interpolate the data whose timestamps do not completely match; then, set a time window to align the timestamps of different data to the closest time point; finally, process the missing values ​​and blank points of the aligned data.

[0019] Furthermore, the raw material demand prediction model uses a time series model LSTM, and the raw material supply of the first data set is used to train the time series model LSTM, and finally the raw material demand prediction model is obtained to predict the raw material demand N.

[0020] Furthermore, the seasonal impact rate in the second data set is calculated by averaging the sales of each month and averaging the sales of all months; then, dividing the average sales of each month by the overall average sales to obtain a seasonal index, and finally standardizing the seasonal index of each month to obtain the seasonal impact rate of each month.

[0021] The calculation method of the brand influence rate in the second data set is: evaluate the influence of the product brand on consumer purchase decisions through market research and consumer surveys, and divide the number of people who think there is brand influence by the total number of surveys and investigations to obtain the brand influence rate;

[0022] The calculation method of the market average price in the second data set is: refer to the product market average price data in the industry report as the market average price;

[0023] The calculation method of the technical impact factor in the second data set is as follows: first, the frequency of technical indicator upgrades of the product and the consumer rating feedback on different technical indicators are collected, and the ratings are normalized; second, the ratings are weighted averaged according to the importance of different technical indicators and the frequency of technical indicator upgrades, and finally the technical impact factor is obtained;

[0024] Construct the product score prediction model to obtain the product's market score P:

[0025]

[0026] Among them, S is the seasonal impact rate; B is the brand impact rate; M is the market average price; T is the technical impact factor; It includes the interaction effects of season, brand, market average price and technical factors; ζ1 is the parameter controlling the intensity parameter of the interaction effect; α1 is the scaling factor used to adjust the entire formula; μ and κ represent the importance parameters of the market average price and technical influencing factors respectively.

[0027] Furthermore, the minimum cost value min(F1) is calculated using the raw material cost, product production volume, product production process cost, raw material replenishment volume, warehouse inventory and warehouse storage cost and the raw material demand N in the first data set; wherein the formula of the cost objective function F1 is:

[0028]

[0029] Wherein, the cost objective function F1 represents the cost; C m is the unit raw material cost; PN is the product production volume; C P is the unit production process cost; C w is the unit warehouse storage cost; W is the warehouse inventory; R is the raw material replenishment quantity; P is the market score of the product; α2, γ1 and They are scaling factors for adjusting the cost of each part, obtained by testing the costs under standardized production conditions.

[0030] Furthermore, the maximum revenue value max(F2) is calculated using the product sales volume, product sales price and product market score P in the first data set; wherein the formula of the revenue objective function F2 is:

[0031] F2=γ2·P·P r SN+C v SN-F1;

[0032] Wherein, the profit objective function F2 represents the profit; SN is the product sales volume; P r is the product sales price; SN is the product sales volume; C v is the unit sales cost; F1 is the cost; γ2 is the profit influencing parameter, which is obtained by conducting standardized production conditions test on the profit.

[0033] Furthermore, the minimum replenishment value min(F3) is calculated using the warehouse inventory and product sales in the first data set and the raw material demand N; wherein the formula of the replenishment objective function F3 is:

[0034]

[0035] Among them, the replenishment objective function F3 represents the replenishment quantity; Ni is the raw material demand of the ith product; W i is the warehouse inventory of the ith product; SN i is the sales volume of the ith product.

[0036] Furthermore, first, we need to ensure that there is an initial online raw material total amount when no production is carried out. The formula for the online raw material total amount is ON t =ON t-1 +N-f3-W; among them, ON t is the total amount of online raw materials in period t; t-1 is the total amount of online original raw materials in period t-1; f3 is the replenishment value calculated by the replenishment objective function; W is the warehouse inventory;

[0037] Constraints are imposed on the total amount of raw materials online: min(ON t )≤ON t ≤max(ON t ); among which, min(ON t ) is the minimum amount of the total amount of online raw materials; max(ON t ) is the maximum amount of the total amount of raw materials online;

[0038] The warehouse inventory W is constrained as follows: W≤max(W); wherein max(W) is the maximum inventory of the warehouse;

[0039] Constrain the return: F2 ≥ min(F2); where min(F2) is the minimum return;

[0040] The raw material demand N is constrained as follows: min(N)≤N≤max(N); wherein min(N) is the minimum amount of the total amount of online raw materials; and max(N) is the maximum amount of the total amount of online raw materials;

[0041] Through the above constraints, the cost objective function F1, the revenue objective function F2 and the replenishment objective function F3 are optimized and calculated to obtain the optimal solution.

[0042] A computer data processing system based on big data, comprising:

[0043] A data collection and processing unit collects production and sales history data to construct a first data set; collects market fluctuation data as a second data set; processes the first data set and the second data set, and performs time alignment on different data;

[0044] A raw material demand prediction model building unit is used to train a time series model LSTM using the data in the first data set to predict the raw material demand N;

[0045] A product score prediction model building unit, using the second data set to build a product score prediction model to obtain a market score P of the product;

[0046] The objective function construction unit constructs a cost objective function F1 according to the raw material demand N and the first data set to obtain a minimum cost value min(F1); constructs a revenue objective function F2 according to the first data set and the market score P of the product to obtain a maximum revenue value max(F2); constructs a replenishment objective function F3 according to the first data set and the raw material demand N to obtain a minimum replenishment value min(F3);

[0047] The optimal solution calculation unit performs optimization calculation on the cost objective function, the revenue objective function and the replenishment objective function to obtain an optimal solution.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] 1. The present invention calculates the market score of products by comprehensively considering seasonal influence rate, brand influence rate, market average price and technical influence factors, thereby significantly optimizing the management of goods. The market scores of these factors are combined to enable enterprises to make more scientific and accurate decisions in the process of goods management, thereby improving overall management efficiency.

[0050] 2. The present invention constructs a comprehensive model including cost objective function, revenue objective function and replenishment objective function, and models the raw material supply, product supply and costs of each link involved in production and sales in detail. Through the comprehensive application of these objective functions, enterprises can improve resource utilization efficiency, reduce waste, ensure efficient operation of production and sales processes, and thus improve overall economic benefits.

[0051] 3. The present invention sets multiple constraints to guide key factors such as the total amount of online raw materials, warehouse inventory and production revenue to improve the efficiency of goods circulation. Constraints help to adjust production and sales strategies in real time, adapt to market changes, ensure that enterprises can maintain efficient operations in a dynamic environment, maximize resource utilization and improve profitability. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A method flow chart of a computer data processing method based on big data of the present invention;

[0053] Figure 2 is a supply chain flow chart of the present invention;

[0054] Figure 3 This is a system structure diagram of a computer data processing system based on big data of the present invention. DETAILED DESCRIPTION

[0055] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0056] In order to improve the efficiency of goods management of enterprises, the present invention provides a computer data processing method and system based on big data. In order to illustrate the function of the present invention, the effectiveness of the present invention will be explained from the following examples.

[0057] Case 1

[0058] In order to improve the production and sales management efficiency of smart home products, a smart home manufacturing company applied a computer data processing method based on big data.

[0059] Reference Figure 1 , which is a method flow chart of a computer data processing method based on big data. Figure 1 In step S01, historical production and sales data are collected to construct a first data set; market fluctuation data are collected as a second data set.

[0060] Among them, the production and sales historical data include the raw material supply of the production workshop, raw material costs, smart home product production volume, product production process costs, raw material replenishment, warehouse inventory, warehouse storage costs, smart home product sales volume and product sales prices; the market fluctuation data include seasonal impact rate, brand influence rate of the smart home manufacturing enterprise, market average price and technology impact factor.

[0061] The first dataset provides detailed information about the production and sales process, which helps optimize inventory management, cost control, and production planning. The second dataset can help companies understand market dynamics and demand trends, thereby adjusting pricing strategies and product positioning. Combining these two datasets, companies can achieve more accurate demand forecasting, inventory optimization, and cost management, improving overall operational efficiency and market competitiveness.

[0062] Reference Figure 1 In step S02, data processing is performed on the first data set and the second data set, and time alignment is performed on different data.

[0063] Specifically, the specific process of the time alignment is:

[0064] First, make sure all data is timestamped in the same format and time zone;

[0065] Secondly, interpolation is performed on data where timestamps do not completely match;

[0066] Then, set a time window to align the timestamps of different data to the closest time point;

[0067] Finally, missing values ​​and blank spots in the aligned data are processed.

[0068] The time alignment process is a prior art and will not be described in detail here.

[0069] Data processing and time alignment of the first and second data sets helps ensure data synchronization and consistency. This step can accurately match the timing of production and sales activities with market fluctuations, making analysis more accurate, thereby improving demand forecasts, optimizing inventory, and adjusting production plans.

[0070] Reference Figure 1 In step S03, a raw material demand prediction model is constructed to predict raw material demand using the data in the first data set.

[0071] Specifically, the raw material demand prediction model uses a time series model LSTM, and uses the raw material supply in the first data set to train the time series model LSTM, and finally obtains the raw material demand prediction model to predict the raw material demand N.

[0072] For the training process of the time series model LSTM, the following parameters are set: the learning rate is set to 0.001, and a lower learning rate helps to stabilize the training process; the optimizer uses the Adam optimizer, which has an adaptive learning rate and momentum correction, and can effectively handle gradient sparsity and imbalance; the number of training rounds is set to 300, and the model is saved at the end of every 50 rounds of training, and the model performance is verified on the test set, and the model with the best performance is selected as; the batch size is set to 64, and a larger batch size can accelerate training; the time series length of the input network can be set to 7, 30, 90, 180, 360, and the sequence length refers to the number of time steps of the input sequence. Here, 7, 30, 90, 180, and 360 represent the number of raw material demands in the past day, month, quarter, half a year, and year, respectively; the hidden layer unit is set to 128; the Dropout rate is set to 0.2 to prevent overfitting; the loss function uses the mean square error loss function. The training process is all existing technology, and no further elaboration is made here.

[0073] The time series model LSTM is good at processing time series data and can identify and remember long-term dependencies, thereby improving the accuracy of demand forecasting.

[0074] Reference Figure 1In step S04, a product score prediction model is constructed using the second data set to obtain a market score for the product.

[0075] The calculation method of the seasonal impact rate in the second data set is as follows: average the sales of each month and average the sales of all months; then, divide the average sales of each month by the overall average sales to obtain the seasonal index; finally, standardize the seasonal index of each month to obtain the seasonal impact rate of each month.

[0076] Furthermore, the calculation method of the brand influence rate in the second data set is: evaluate the influence of the brand of smart home products on customer purchasing decisions through market research and customer surveys, and divide the number of customers who believe that there is brand influence by the total number of customers surveyed and investigated to obtain the brand influence rate.

[0077] Furthermore, the method for calculating the market average price in the second data set is: referring to the market average price data of smart home products in the industry report as the market average price.

[0078] Furthermore, the calculation method of the technical impact factor in the second data set is: first, collect the frequency of technical indicator upgrades of smart home products and customers' rating feedback on different technical indicators, and normalize the scores; second, according to the importance of different technical indicators and using the frequency of technical indicator upgrades, perform weighted averaging on the scores to finally obtain the technical impact factor.

[0079] Based on the calculated seasonal influence rate, brand influence rate, market average price and technical influence factor, a smart home product score prediction model is constructed to obtain the market score P of smart home products:

[0080]

[0081] Among them, S is the seasonal impact rate; B is the brand impact rate; M is the market average price; T is the technical impact factor; It includes the interaction effects of season, brand, market average price and technical factors; ζ1 is the intensity parameter of the parameter controlling the interaction effect; α1 is the scaling factor used to adjust the entire formula, which is set to 0.7 in this embodiment; μ and κ represent the importance parameters of the market average price and technical influencing factors, which are set to 0.5 and 0.9 respectively.

[0082] Building a smart home product score prediction model based on seasonal influence rate, brand influence rate, market average price and technical influence factors can provide accurate market scores. This model comprehensively analyzes multiple market factors to comprehensively evaluate the market performance and potential of products, helping smart home manufacturers adjust production plans, pricing strategies and market promotion to maximize commercial value.

[0083] Reference Figure 1 In step S05, a cost objective function is constructed according to the raw material demand and the first data set to obtain a minimum cost value.

[0084] Specifically, the minimum cost value min(F1) is calculated using the raw material cost, smart home product production volume, product production process cost, raw material replenishment volume, warehouse inventory and warehouse storage cost, and the raw material demand N in the first data set; wherein the formula of the cost objective function F1 is:

[0085]

[0086] Wherein, the cost objective function F1 represents the cost; C m is the unit raw material cost; PN is the product production volume; C P is the unit production process cost; C w is the unit warehouse storage cost; W is the warehouse inventory; R is the raw material replenishment quantity; P is the market score of the product; α2, γ1 and are scaling factors for adjusting the cost of each part, obtained by testing the cost under standardized production conditions. The standardized production conditions include the setting of production equipment, the control of the production environment (such as temperature and humidity), and operating procedures, etc., to ensure that the test is carried out under the same conditions. Verify whether it is effective in actual production. If the verification results do not meet expectations, it is necessary to adjust the test process, reset the standardized conditions, or re-test to ensure the accuracy and applicability of the parameters.

[0087] This method of fitting the cost objective function F1 and calculating the minimum cost value can optimize cost management and reduce unnecessary expenses. The calculation of the minimum cost value helps companies identify cost-saving opportunities, achieve more efficient resource allocation, and effectively control production and inventory costs to improve overall profitability.

[0088] Reference Figure 1 In step S06, a revenue objective function is constructed according to the first data set and the market score of the smart home product to obtain a maximum revenue value.

[0089] Specifically, the maximum revenue value max(F2) is calculated using the sales volume and product sales price of the smart home products in the first data set and the market score P of the smart home products; wherein the formula of the revenue objective function F2 is:

[0090] F2=γ2·P·P r SN+C v SN-F1;

[0091] Wherein, the profit objective function F2 represents the profit; SN is the product sales volume; P r is the product sales price; SN is the product sales volume; C v is the unit sales cost; F1 is the cost; γ2 is the profit influencing parameter, which is obtained by conducting standardized production conditions test on the profit.

[0092] The standardized production conditions include production environment, operating procedures and raw materials to ensure the consistency of the test process. The parameter γ2 obtained from the test can be verified by subsequent data or actual production results to check the accuracy of the model and verify whether it is effective in actual production. If the verification results do not meet expectations, it is necessary to adjust the test process, reset the standardized conditions or re-test to ensure the accuracy and applicability of the parameters.

[0093] By calculating the maximum revenue value of smart home products, combined with the sales volume, product sales price and market score of smart home products, the profit potential of products can be accurately evaluated. The calculation of the maximum revenue value can help companies develop more effective marketing plans, increase sales revenue and market share, and maximize overall profitability.

[0094] Reference Figure 1 In step S07, a replenishment objective function is constructed according to the first data set and the raw material demand to obtain a minimum replenishment value.

[0095] Specifically, the minimum replenishment value min(F3) is calculated using the warehouse inventory and product sales in the first data set and the raw material demand N; wherein the formula of the replenishment objective function F3 is:

[0096]

[0097] Among them, the replenishment objective function F3 represents the replenishment quantity; N i is the raw material demand of the ith product; W i is the warehouse inventory of the ith product; SN i is the sales volume of the ith product.

[0098] The calculation of the minimum replenishment value can help companies determine the optimal replenishment quantity, avoid excessive inventory, improve inventory turnover and operational efficiency, and achieve more efficient supply chain management.

[0099] Reference Figure 1 In step S08, the cost objective function, the revenue objective function and the replenishment objective function are optimized and calculated to obtain the optimal solution.

[0100] First, ensure that there is an initial online raw material total amount when production is not in progress. The calculation formula for the online raw material total amount is ON t =ON t-1 +N-f3-W; among them, ON t is the total amount of online raw materials in period t; t-1 is the total amount of online original raw materials in period t-1;

[0101] Constraints are imposed on the total amount of raw materials online: min(ON t )≤ON t ≤max(ON t ); among which, min(ON t ) is the minimum amount of the total amount of online raw materials; max(ON t ) is the maximum amount of the total amount of raw materials online;

[0102] The warehouse inventory W is constrained as follows: W≤max(W); wherein max(W) is the maximum inventory of the warehouse;

[0103] Constrain the production revenue: F2≥min(F2); where min(F2) is the minimum production revenue;

[0104] The raw material demand N is constrained as follows: min(N)≤N≤max(N); wherein min(N) is the minimum amount of the total amount of online raw materials; and max(N) is the maximum amount of the total amount of online raw materials.

[0105] Based on the above constraints, the cost objective function F1, the revenue objective function F2 and the replenishment objective function F3 are optimized and calculated to obtain the optimal solution. Then, the optimal solution is checked to see if it meets all the constraints, and a sensitivity analysis is performed to evaluate the stability of the solution. The model is adjusted according to the analysis results, and the solution is repeated when necessary to obtain a better solution. The supply chain process of the smart home manufacturing company is as follows: Figure 2 As shown, this method can adjust the cost of the smart home manufacturing enterprise's supply chain investment and obtain maximum benefits.

[0106] By finding the optimal solution to these objective functions, companies can maximize revenue while reducing costs and ensure the effectiveness of replenishment strategies. This approach helps improve overall operational efficiency, reduce unnecessary expenses, and avoid inventory shortages or excesses.

[0107] In this embodiment, the computer data processing method based on big data used by the smart home manufacturing enterprise can achieve accurate raw material demand forecasting and market score evaluation by integrating production and sales history data and market fluctuation data. In addition, the method can help enterprises effectively reduce operating costs and optimize inventory management by fitting and optimizing cost, revenue and replenishment objective functions.

[0108] Embodiment 2

[0109] A power equipment manufacturing company has adopted a computer data processing system based on big data. The system structure diagram of a computer data processing system based on big data is as follows: Figure 3 As shown, it includes a data collection and processing unit, which collects production and sales history data to construct a first data set; collects market fluctuation data as a second data set; processes the first data set and the second data set, and performs time alignment on different data.

[0110] The specific process of time alignment is: first, ensure that the timestamps of all data use the same format and time zone; second, interpolate the data whose timestamps do not completely match; then, set a time window to align the timestamps of different data to the closest time point; finally, handle missing values ​​and blank points of the aligned data. The process of time alignment is an existing technology and will not be described in detail here.

[0111] A computer data processing system based on big data also includes a raw material demand prediction model construction unit, which uses the data in the first data set to train the time series model LSTM to predict the raw material demand N.

[0112] The raw material demand prediction model uses the time series model LSTM, and uses the raw material supply in the first data set to train the time series model LSTM, and finally obtains the raw material demand prediction model to predict the raw material demand N.

[0113] For the training process of the time series model LSTM, the following parameters are set: the learning rate is set to 0.001, and a lower learning rate helps to stabilize the training process; the optimizer uses the Adam optimizer, which has an adaptive learning rate and momentum correction, and can effectively handle gradient sparsity and imbalance; the number of training rounds is set to 300, and the model is saved at the end of every 50 rounds of training, and the model performance is verified on the test set, and the model with the best performance is selected as; the batch size is set to 64, and a larger batch size can accelerate training; the time series length of the input network can be set to 7, 30, 90, 180, 360, and the sequence length refers to the number of time steps of the input sequence. Here, 7, 30, 90, 180, and 360 represent the number of raw material demands in the past day, month, quarter, half a year, and year, respectively; the hidden layer unit is set to 128; the Dropout rate is set to 0.2 to prevent overfitting; the loss function uses the mean square error loss function. The training process is all existing technology, and no further elaboration is made here.

[0114] A computer data processing system based on big data also includes a product score prediction model construction unit, which uses the second data set to construct a product score prediction model to obtain the market score P of the product.

[0115] The calculation method of the seasonal impact rate in the second data set is as follows: average the sales of each month and average the sales of all months; then, divide the average sales of each month by the overall average sales to obtain the seasonal index; finally, standardize the seasonal index of each month to obtain the seasonal impact rate of each month.

[0116] Furthermore, the calculation method of the brand influence rate in the second data set is: evaluate the influence of the brand of smart home products on customer purchasing decisions through market research and customer surveys, and divide the number of customers who believe that there is brand influence by the total number of customers surveyed and investigated to obtain the brand influence rate.

[0117] Furthermore, the method for calculating the market average price in the second data set is: referring to the market average price data of smart home products in the industry report as the market average price.

[0118] Furthermore, the calculation method of the technical impact factor in the second data set is: first, collect the frequency of technical indicator upgrades of smart home products and customers' rating feedback on different technical indicators, and normalize the scores; second, according to the importance of different technical indicators and using the frequency of technical indicator upgrades, perform weighted averaging on the scores to finally obtain the technical impact factor.

[0119] Based on the calculated seasonal influence rate, brand influence rate, market average price and technical influence factor, a smart home product score prediction model is constructed to obtain the market score P of smart home products:

[0120]

[0121] Among them, S is the seasonal impact rate; B is the brand impact rate; M is the market average price; T is the technical impact factor; It includes the interaction effects of season, brand, market average price and technical factors; ζ1 is the intensity parameter of the parameter controlling the interaction effect; α1 is the scaling factor used to adjust the entire formula, which is set to 0.8 in this embodiment; μ and κ represent the importance parameters of the market average price and technical influencing factors, which are set to 0.7 and 0.8 respectively.

[0122] A computer data processing system based on big data also includes an objective function construction unit, which constructs a cost objective function F1 according to the raw material demand N and the first data set to obtain a minimum cost value min(F1); constructs a profit objective function F2 according to the first data set and the market score P of the product to obtain a maximum profit value max(F2); and constructs a replenishment objective function F3 according to the first data set and the raw material demand N to obtain a minimum replenishment value min(F3).

[0123] Among them, the formula of cost objective function F1 is:

[0124]

[0125] Among them, C m is the unit raw material cost; PN is the product production volume; C P is the unit production process cost; C w is the unit warehouse storage cost; W is the warehouse inventory; R is the raw material replenishment quantity; P is the market score of the product; α2, γ1 and They are scaling factors for adjusting the cost of each part, obtained by testing the costs under standardized production conditions.

[0126] The formula of the profit objective function F2 is:

[0127] F2=γ2·P·P r SN+C v SN-F1;

[0128] Wherein, the profit objective function F2 represents the profit; SN is the product sales volume; P r is the product sales price; SN is the product sales volume; C v is the unit sales cost; F1 is the cost; γ2 is the profit influencing parameter, which is obtained by conducting standardized production conditions test on the profit.

[0129] The formula of replenishment objective function F3 is:

[0130]

[0131] Among them, N i is the raw material demand of the ith product; W i is the warehouse inventory of the ith product; SN i is the sales volume of the ith product.

[0132] A computer data processing system based on big data also includes an optimal solution calculation unit, which performs optimal calculation on the cost objective function, the revenue objective function and the replenishment objective function to obtain the optimal solution.

[0133] First, set up the different constraints:

[0134] The calculation formula for the total amount of raw materials online is ON t =ON t-1 +N-f3-W; among them, ON t is the total amount of online raw materials in period t; t-1 is the total amount of online original raw materials in period t-1;

[0135] Constraints are imposed on the total amount of raw materials online: min(ON t )≤ON t ≤max(ON t ); among which, min(ON t ) is the minimum amount of the total amount of online raw materials; max(ON t ) is the maximum amount of the total amount of raw materials online;

[0136] The warehouse inventory W is constrained as follows: W≤max(W); wherein max(W) is the maximum inventory of the warehouse;

[0137] Constrain the production revenue: F2≥min(F2); where min(F2) is the minimum production revenue;

[0138] The raw material demand N is constrained as follows: min(N)≤N≤max(N); wherein min(N) is the minimum amount of the total amount of online raw materials; and max(N) is the maximum amount of the total amount of online raw materials.

[0139] Through the above constraints, the cost objective function F1, the revenue objective function F2 and the replenishment objective function F3 are optimized and calculated to obtain the optimal solution.

[0140] The computer data processing system based on big data adopted in this embodiment includes all the features of any of the aforementioned embodiments.

[0141] This embodiment also includes a data visualization module and a real-time monitoring and alarm module.

[0142] In the data visualization module, real-time charts are used to display the equipment operation status, production speed and output of each production line, helping management analyze production efficiency. Then, pie charts and bar charts are used to display the quality defect rate of different equipment to identify the equipment or production stage that needs the most improvement. Next, energy consumption data is displayed in the form of a line graph to analyze the energy consumption trend of different production links and help optimize energy use.

[0143] Furthermore, in the real-time monitoring and alarm module, the operating parameters of the equipment, including temperature, pressure, etc., are monitored in real time. When the parameters exceed the normal range, the alarm is automatically triggered and the maintenance team is notified in time for processing. Then, the thresholds of key indicators of the production line, such as production speed and output, are set. Once abnormal fluctuations occur, the alarm is immediately issued to reduce the risk of production interruption. Finally, according to the operating time and status of the production equipment, maintenance alarms are automatically generated to remind relevant personnel to perform regular maintenance to prevent equipment failure.

[0144] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A computer data processing method based on big data, characterized in that: include: Collect historical production and sales data to build the first data set; Collect market volatility data as the second data set; performing data processing on the first data set and the second data set; Using the data in the first data set, a raw material demand forecasting model is constructed to forecast the raw material demand N; Using the second data set, construct a product score prediction model to obtain the market score P of the product; According to the raw material demand N and the first data set, a cost objective function F1 is constructed to obtain a minimum cost value min(F1); According to the first data set and the market score P of the product, a profit objective function F2 is constructed to obtain a maximum profit value max(F2); According to the first data set and the raw material demand N, a replenishment objective function F3 is constructed to obtain a minimum replenishment value min(F3); The cost objective function, the revenue objective function and the replenishment objective function are optimized and calculated to obtain an optimal solution.

2. A computer data processing method based on big data according to claim 1, characterized in that: The production and sales historical data include the raw material supply of the production workshop, raw material cost, product production volume, product production process cost, raw material replenishment, warehouse inventory, warehouse storage cost, product sales volume and product sales price; the market fluctuation data include seasonal impact rate, brand impact rate, market average price and technical impact factor.

3. A computer data processing method based on big data according to claim 1, characterized in that: The specific process of time alignment is: first, ensure that the timestamps of all data use the same format and time zone; second, interpolate the data whose timestamps do not completely match; then, set a time window to align the timestamps of different data to the closest time point; finally, process the missing values ​​and blank points of the aligned data.

4. A computer data processing method based on big data according to claim 1, characterized in that: The raw material demand prediction model uses a time series model LSTM, and the raw material supply of the first data set is used to train the time series model LSTM, and finally the raw material demand prediction model is obtained to predict the raw material demand N.

5. A computer data processing method based on big data according to claim 1, characterized in that: The calculation method of the seasonal impact rate in the second data set is: average the sales of each month and average the sales of all months; then, divide the average sales of each month by the overall average sales to obtain the seasonal index, and finally standardize the seasonal index of each month to obtain the seasonal impact rate of each month; The calculation method of the brand influence rate in the second data set is: evaluate the influence of the product brand on consumer purchase decisions through market research and consumer surveys, and divide the number of people who think there is brand influence by the total number of surveys and investigations to obtain the brand influence rate; The calculation method of the market average price in the second data set is: refer to the product market average price data in the industry report as the market average price; The calculation method of the technical impact factor in the second data set is as follows: first, the frequency of technical indicator upgrades of the product and the consumer rating feedback on different technical indicators are collected, and the ratings are normalized; second, the ratings are weighted averaged according to the importance of different technical indicators and the frequency of technical indicator upgrades, and finally the technical impact factor is obtained; Construct the product score prediction model to obtain the product's market score P: Among them, S is the seasonal impact rate; B is the brand impact rate; M is the market average price; T is the technical impact factor; It includes the interaction effects of season, brand, market average price and technical factors; ζ1 is the intensity parameter for controlling the interaction effect; α1 is the scaling factor used to adjust the entire formula; μ and κ represent the importance parameters of the market average price and technical influencing factors respectively.

6. A computer data processing method based on big data according to claim 1, characterized in that: The minimum cost value min(F1) is calculated using the raw material cost, product production volume, product production process cost, raw material replenishment volume, warehouse inventory and warehouse storage cost in the first data set and the raw material demand N; wherein the formula of the cost objective function F1 is: Wherein, the cost objective function F1 represents the cost; C m is the unit raw material cost; PN is the product production volume; C P is the unit production process cost; C w is the unit warehouse storage cost; W is the warehouse inventory; R is the raw material replenishment quantity; P is the market score of the product; α2, γ1 and They are scaling factors for adjusting the cost of each part, obtained by testing the costs under standardized production conditions.

7. A computer data processing method based on big data according to claim 1, characterized in that: The maximum revenue value max(F2) is calculated using the product sales volume, product sales price and product market score P in the first data set; wherein the formula of the revenue objective function F2 is: F2=γ2·P·P r ·SN+C v ·SN-F1; Wherein, the profit objective function F2 represents the profit; SN is the product sales volume; P r is the product sales price; SN is the product sales volume; C v is the unit sales cost; F1 is the cost; γ2 is the profit influencing parameter, which is obtained by conducting standardized production conditions test on the profit.

8. A computer data processing method based on big data according to claim 1, characterized in that: The minimum replenishment value min(F3) is calculated using the warehouse inventory and product sales in the first data set and the raw material demand N; wherein the formula of the replenishment objective function F3 is: Among them, the replenishment objective function F3 represents the replenishment quantity; N i is the raw material demand of the ith product; W i is the warehouse inventory of the ith product; SN i is the sales volume of the ith product.

9. A computer data processing method based on big data according to claim 1, characterized in that: First, ensure that there is an initial amount of raw materials online when production is not in progress. The formula is ON t =ON t-1 +N-f3-W; among them, ON t is the total amount of online raw materials in period t; t-1 is the total amount of online original raw materials in period t-1; f3 is the replenishment value calculated by the replenishment objective function; W is the warehouse inventory; Constraints are imposed on the total amount of raw materials online: min(ON t )≤ON t ≤max(ON t ); among which, min(ON t ) is the minimum amount of the total amount of online raw materials; max(ON t ) is the maximum amount of the total amount of raw materials online; The warehouse inventory W is constrained as follows: W≤max(W); wherein max(W) is the maximum inventory of the warehouse; Constrain the production revenue: F2≥min(F2); where min(F2) is the minimum production revenue; The raw material demand N is constrained as follows: min(N)≤N≤max(N); wherein min(N) is the minimum amount of the total amount of online raw materials; and max(N) is the maximum amount of the total amount of online raw materials; Through the above constraints, the cost objective function F1, the revenue objective function F2 and the replenishment objective function F3 are optimized and calculated to obtain the optimal solution.

10. A computer data processing system based on big data, characterized in that: include: A data collection and processing unit, collecting production and sales historical data and constructing a first data set; Collect market volatility data as the second data set; Performing data processing on the first data set and the second data set, and performing time alignment on different data; A raw material demand prediction model building unit is used to train a time series model LSTM using the data in the first data set to predict the raw material demand N; A product score prediction model building unit, using the second data set to build a product score prediction model to obtain a market score P of the product; The objective function construction unit constructs a cost objective function F1 according to the raw material demand N and the first data set to obtain a minimum cost value min(F1); constructs a revenue objective function F2 according to the first data set and the market score P of the product to obtain a maximum revenue value max(F2); constructs a replenishment objective function F3 according to the first data set and the raw material demand N to obtain a minimum replenishment value min(F3); The optimal solution calculation unit performs optimization calculation on the cost objective function, the revenue objective function and the replenishment objective function to obtain an optimal solution.

Images