Intelligent manufacturing enterprise multi-material joint purchase optimization method and system

Abstract

The invention relates to the field of material purchase optimization, and discloses an intelligent manufacturing enterprise multi-material joint purchase optimization method and system, which is used for solving the problem that real-time adjustment cannot be carried out during material joint purchase, and comprises the steps of obtaining historical purchase related data, carrying out prediction by using a time sequence method, obtaining the demand quantity of each material, and carrying out multi-material joint purchase optimization. Screening suppliers according to the material demands, obtaining data of the suppliers, obtaining a priority selection index according to the data of the suppliers through an optimal sequence graph method, selecting an initial supplier according to the priority selection index, obtaining the purchasing cost, the transportation cost and the inventory cost of the initial supplier, and obtaining an optimal purchasing scheme by using a mathematical optimization model. According to the method and the system, combined purchasing is carried out, historical purchasing related data and supplier data are continuously acquired in real time, and real-time purchasing adjustment is carried out, so that the accuracy of material combined purchasing is effectively improved, and the production efficiency of enterprises is improved.

Description

Technical Field

[0001] The present invention relates to the field of material procurement optimization, and more specifically to a multi-material joint procurement optimization method and system for an intelligent manufacturing enterprise. Background Art

[0002] With the continuous development of the procurement model of smart manufacturing enterprises, enterprises have an increasing demand for joint procurement of multiple materials to improve supply chain efficiency, reduce costs and enhance supply chain stability. In traditional procurement methods, enterprises usually adopt a single material independent procurement model, that is, placing orders with suppliers for different materials separately. Existing procurement optimization methods mainly rely on historical procurement data analysis and supplier evaluation models. By analyzing historical procurement orders and supplier delivery status, suppliers are scored and ranked, and top-ranked suppliers are selected for long-term cooperation.

[0003] However, in actual applications, the prediction of procurement demand is mainly based on historical order data, and is not adjusted according to the production plan, resulting in delayed procurement decisions and difficulty in adapting to actual production needs. The existing procurement model lacks the ability to monitor real-time data and adjust procurement optimization strategies, which makes it difficult to effectively control procurement costs and even affects the normal production of the enterprise.

[0004] In view of the above problems, the present invention proposes a solution. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides a method and system for optimizing joint procurement of multiple materials in an intelligent manufacturing enterprise to solve the problems existing in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions: A multi-material joint procurement optimization method for an intelligent manufacturing enterprise comprises the following steps: Step 1: Obtain historical procurement related data, the procurement related data includes material types, warehouse material inventory and each material demand, and use the time series method to predict according to the historical procurement related data to obtain each material demand; Step 2: According to each material demand, select suppliers whose material inventory is greater than the material demand, and obtain the data of each supplier, the supplier data includes the type of supplied materials, the historical material delivery on-time rate, the quality pass rate, the material price and the shipping distance, and obtain the priority selection index through the priority diagram method according to the supplier data; Step 3: Select the initial supplier according to the priority selection index, obtain the procurement cost, transportation cost and inventory cost of the initial supplier, and use the mathematical optimization model according to the procurement cost, transportation cost and inventory cost of the initial supplier to obtain the optimal procurement plan; Step 4: Conduct joint procurement according to the optimal procurement plan and each material demand; Step 5: Continue to collect historical procurement related data and supplier data in real time, and make real-time procurement adjustments.

[0007] Preferably, the step of using the time series method to predict based on historical procurement related data to obtain the demand for each material is: setting a collection time period, obtaining historical procurement related data within the collection time period, the procurement related data including material type, warehouse material inventory and each material demand; performing data preprocessing on the historical procurement related data, the data preprocessing includes data cleaning and time series formatting; selecting an ARIMA time series model, training the ARIMA time series model with historical procurement related data to obtain a final prediction model, and using the final prediction model to predict the demand for each material.

[0008] Preferably, the steps of training the ARIMA time series model with historical procurement-related data to obtain the final prediction model are: using the ADF unit root test to determine whether the data is stable, and if the data is determined to be unsteady, using multi-order difference conversion data until the data is stable to obtain the difference order; using the autocorrelation function graph and the partial autocorrelation function graph to obtain the autoregressive order, and obtaining the moving average order through the autocorrelation function graph; dividing the historical procurement-related data into training set data and test set data, and using the training set data to fit the ARIMA model; selecting the autoregressive order, the difference order and the moving average order, and using the least squares method to obtain the autoregressive coefficient number and the moving average coefficient to obtain the initial prediction model; use the initial prediction model to predict the demand of the training set data to obtain the predicted demand, and calculate the mean square error, root mean square error and mean absolute percentage error according to the predicted demand; if the mean absolute percentage error is less than the preset threshold, the initial prediction model is judged to have high accuracy, and the initial prediction model is used as the final prediction model. If the mean absolute percentage error is greater than or equal to the preset threshold, the autoregressive order, difference order and moving average order of the initial prediction model are adjusted, and the model accuracy judgment is continued until the mean absolute percentage error is less than the preset threshold to obtain the final prediction model.

[0009] Preferably, the step of obtaining the preference index is as follows: obtaining the historical on-time delivery rate of materials within the supplier's inspection time period, and calculating the on-time delivery coefficient based on the historical on-time delivery rate of materials by the singular value decomposition method; obtaining the quality pass rate within the supplier's inspection time period, and calculating the quality pass coefficient based on the quality pass rate by the K-means clustering method; obtaining the supplier's material price and the market average price, and calculating the price competition coefficient by subtracting the supplier's material price from the market average price; constructing a supplier evaluation matrix based on the types of supplied materials, on-time delivery coefficient, quality pass coefficient, price competition coefficient and shipping distance; normalizing the types of supplied materials, on-time delivery coefficient, quality pass coefficient, price competition coefficient and shipping distance to construct a normalized supplier evaluation matrix; calculating the index comparison difference between suppliers based on the normalized supplier evaluation matrix; calculating the preference function based on the index comparison difference between suppliers, and calculating the overall preference value of every two suppliers based on the preference function. The specific acquisition steps are as follows: ; In the formula, Expressed as the overall preference value of supplier i relative to supplier k, It is expressed as the weight of indicator j, m is the number of indicators, is the preference function; the preference index is calculated based on the overall preference value of every two suppliers.

[0010] Preferably, the step of calculating the preference index according to the overall preference value of every two suppliers is: according to the normalized supplier evaluation matrix, calculating the positive flow of the normalized supplier evaluation matrix, and the specific acquisition steps are: ; In the formula, is represented by the positive flow of the evaluation matrix of the i-th supplier, w is represented by the number of suppliers, It is expressed as the overall preference value of supplier i relative to supplier k. According to the normalized supplier evaluation matrix, the negative flow of the normalized supplier evaluation matrix is ​​calculated. The specific acquisition steps are: ; In the formula, Represented as the negative flow of the i-th supplier evaluation matrix, It is expressed as the overall preference value of supplier k relative to supplier i. The net flow of the normalized supplier evaluation matrix is ​​obtained by calculating the difference between the positive flow of the normalized supplier evaluation matrix and the negative flow of the normalized supplier evaluation matrix, and the net flow of the normalized supplier evaluation matrix is ​​used as the preference index.

[0011] Preferably, the steps for obtaining the on-time delivery coefficient are: construct a historical on-time delivery rate vector based on the time series of the historical on-time delivery rate of the material, use the singular value decomposition method, and set the window size; perform singular value decomposition on the nested time window matrix to obtain a singular value matrix; select singular values ​​whose energy accumulation exceeds the set value for retention, and reconstruct the data according to the retained singular values ​​to obtain a denoised on-time delivery rate matrix; perform mean calculation on the data in the denoised on-time delivery rate matrix to obtain the on-time delivery coefficient.

[0012] Preferably, the steps for obtaining the quality qualification coefficient are as follows: Step 2.1: taking the quality qualification rate as a clustering feature, taking all the quality qualification rates within the detection time period as a data set, and each quality qualification rate in the data set as a data point; Step 2.2: using the silhouette coefficient method to determine the number of clusters K of the data set; Step 2.3: randomly selecting K data points in the data set as initial cluster centers, for each data point, calculating its Euclidean distance to each initial cluster center, for each data point, traversing the K initial cluster centers, and assigning it to the initial cluster center closest to it corresponding clusters; Step 2.4: After traversing all data points, the initial clusters are obtained. For each initial cluster, the data points in it are averaged to obtain a new cluster center; Step 2.5: Repeat Step 2.3 and Step 2.4 until the cluster center no longer changes, and the final clusters and final cluster centers are obtained; Step 2.6: Calculate the ratio of the number of data points in each final cluster to the total number of data points to obtain the weight of each final cluster, and perform weighted summation of the weight of each final cluster and the final cluster center to obtain the quality qualification coefficient.

[0013] Preferably, a multi-material joint procurement optimization system for an intelligent manufacturing enterprise, the system comprising: a historical data acquisition module, for acquiring historical procurement related data, the procurement related data including material types, warehouse material inventory and each material demand, using the time series method to predict based on the historical procurement related data to obtain each material demand; a priority index acquisition module, for screening out suppliers whose material inventory is greater than the material demand according to each material demand, and acquiring each supplier data, the supplier data including the type of supplied materials, the material historical delivery on-time rate, the quality pass rate, the material price and the shipping distance, and obtaining the priority index through the priority diagram method based on the supplier data; an optimal solution acquisition module, for selecting the initial supplier according to the priority index, obtaining the procurement cost, transportation cost and inventory cost of the initial supplier, and obtaining the optimal procurement solution based on the procurement cost, transportation cost and inventory cost of the initial supplier using a mathematical optimization model; a joint procurement module, for conducting joint procurement according to the optimal procurement solution and each material demand; a real-time adjustment module, for continuing to collect historical procurement related data and supplier data in real time, and performing real-time procurement adjustments.

[0014] Technical effects and advantages of the present invention: Obtain historical procurement-related data, use the time series method to make predictions, obtain the demand for each material, screen suppliers based on the demand for each material, obtain the data of each supplier, evaluate the supplier data through the priority diagram method to obtain the preference index, select the initial supplier based on the preference index, obtain the procurement cost, transportation cost and inventory cost of the initial supplier, use the mathematical optimization model to obtain the optimal procurement plan, conduct joint procurement, continue to collect historical procurement-related data and supplier data in real time, and conduct real-time procurement adjustments, which effectively improves the accuracy of joint material procurement and improves enterprise production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A flow chart of a multi-material joint procurement optimization method for an intelligent manufacturing enterprise provided in an embodiment of the present application.

[0016] Figure 2 A structural diagram of a multi-material joint procurement optimization system for an intelligent manufacturing enterprise provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the present invention. In addition, the forms of the various structures recorded in the following embodiments are only examples. The multi-material joint procurement optimization method and system for an intelligent manufacturing enterprise involved in the present invention are not limited to the various structures recorded in the following embodiments. All other implementations obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.

[0018] The present invention provides a multi-material joint procurement optimization method for an intelligent manufacturing enterprise, such as Figure 1 As shown, the following steps are included: Step 1: Obtain historical procurement-related data, which includes material types, warehouse material inventory, and material demand. Use the time series method to predict the demand for each material based on the historical procurement-related data. Time series method is a statistical method based on the trend of historical data over time. It is mainly used to analyze and predict the future trend of time series data. In multi-material procurement optimization, time series method uses the company's historical procurement data to predict the future demand for each material. Common methods include autoregressive moving average model, exponential smoothing method, and long short-term memory network. Through time series analysis, companies can plan procurement plans more scientifically, avoid inventory backlogs or shortages, reduce procurement costs, and improve supply chain efficiency.

[0019] In this embodiment, it should be specifically explained that the steps of predicting the demand for each material using the time series method based on historical procurement related data are as follows: Set the collection time period to obtain historical procurement-related data within the collection time period. Procurement-related data includes material types, warehouse material inventory, and demand for each material; Preprocess the historical procurement data. Data preprocessing includes data cleaning and time series formatting. Data cleaning includes missing value processing and outlier processing. Time series formatting is to reconstruct the data so that the data is arranged in chronological order to form a time series to ensure the timeliness and continuity of the data. Select the ARIMA time series model, train the ARIMA time series model with historical procurement-related data to obtain the final forecast model, and use the final forecast model to predict the demand for each material.

[0020] The ARIMA time series model is a statistical modeling method widely used for time series forecasting. It is suitable for data with trends, seasonality or non-stationarity. ARIMA combines three parts: autoregression, difference and moving average. The autoregression part uses past observations to predict current values, the difference part is used to stabilize non-stationary data, and the moving average part is used to reduce the cumulative impact of forecast errors. The model can effectively predict future trends and is widely used in economics, market analysis, inventory management and demand forecasting.

[0021] In this embodiment, it should be specifically explained that the steps of training the ARIMA time series model through historical procurement related data to obtain the final prediction model are: The ADF unit root test was used to determine whether the data was stationary. If the data was not stationary, multiple order differences were used to transform the data until the data was stationary and the order of differences was obtained; The autoregressive order is obtained using the autocorrelation function graph and the partial autocorrelation function graph, and the moving average order is obtained using the autocorrelation function graph; The autoregressive order represents the relationship between the current value and the value at a past time point, the difference order is the number of differences used to make the data smooth, and the moving average order represents the relationship between the current value and the prediction error at a past time point.

[0022] The historical procurement data is divided into training set data and test set data, and the training set data is used to fit the ARIMA model. The specific steps are as follows: ; In the formula, Expressed as the demand forecast value at time point t, Expressed as a constant term, Expressed as the autoregressive coefficient, Expressed as the moving average coefficient, is represented as the error term, Expressed as the autoregressive order, Expressed as the moving average order; Select the best autoregressive order, difference order and moving average order, and use the least squares method to obtain the autoregressive coefficient and moving average coefficient to obtain the initial prediction model. The least squares method is a mathematical method for data fitting and parameter estimation, which aims to find a set of optimal parameters to minimize the sum of squared errors between the model's predicted values ​​and actual observed values. Use the initial prediction model to predict the demand of the training set data to obtain the predicted demand. The mean square error is calculated based on the predicted demand. The specific acquisition steps are as follows: ; In the formula, It is expressed as mean square error, n is the number of test samples, It is expressed as the actual demand, that is, the real material demand at time point t. It is expressed as the predicted demand, that is, the material demand at time point t predicted by the initial prediction model. The smaller the mean square error, the lower the prediction error of the model. The root mean square error is calculated based on the predicted demand. The specific acquisition steps are as follows: ; In the formula, It is expressed as root mean square error. The root mean square error solves the problem of mismatch of mean square error units, making the error measurement consistent with the original data. The smaller the root mean square error, the better the prediction effect. The mean absolute percentage error is calculated based on the predicted demand. The specific acquisition steps are as follows: ; In the formula, It is expressed as mean absolute percentage error. The mean absolute percentage error represents the percentage of the prediction error relative to the true value. Therefore, it has dimensionless characteristics and is suitable for comparison of different data scales. The smaller the mean absolute percentage error, the more accurate the prediction model. If the mean absolute percentage error is less than 20%, the initial prediction model is judged to be highly accurate and is used as the final prediction model. If the mean absolute percentage error is greater than or equal to 20%, the autoregressive order, difference order, and moving average order of the initial prediction model are adjusted, and the model accuracy is continued until the mean absolute percentage error is less than 20% to obtain the final prediction model.

[0023] Step 2: Filter out suppliers whose material inventory is greater than the material demand based on the material demand, and obtain the data of each supplier. The supplier data includes the type of supplied materials, the historical on-time delivery rate of materials, the quality pass rate, the material price, and the delivery distance. The priority index is obtained based on the supplier data through the priority diagram method; The precedence diagram method is a multi-criteria decision analysis method used to rank and prioritize multiple candidate solutions. Based on different evaluation indicators, this method calculates the preference between candidate solutions and constructs a precedence diagram to determine the best choice. The core idea is to compare the performance of each candidate solution on different criteria, calculate the preference index, and thus obtain the priority ranking of suppliers. The precedence diagram method is more intuitive than the traditional weighted scoring method, and can provide a more scientific decision-making basis when facing multiple suppliers and complex weight relationships. It is widely used in supply chain optimization, procurement decision-making, investment evaluation and other fields.

[0024] The benefit of evaluating the supplier preference index through the priority diagram method is that it can comprehensively consider multiple key factors (such as the type of supplied materials, on-time delivery rate, quality qualification rate, price competitiveness, shipping distance, etc.), and scientifically quantify the ranking of suppliers to avoid the deviation caused by traditional subjective scoring or single weight scoring. The priority diagram method can compare suppliers in pairs and calculate their relative advantages in different indicators, making the decision more objective and fair. In this way, enterprises can choose suppliers with the best comprehensive capabilities, thereby reducing procurement costs, improving supply chain stability, optimizing logistics distribution, reducing delivery risks, and ultimately improving overall procurement efficiency and operational benefits.

[0025] In this embodiment, it should be specifically explained that the step of obtaining the priority index is: Obtain the historical on-time delivery rate of materials during the supplier's inspection period, and calculate the on-time delivery coefficient based on the historical on-time delivery rate of materials using the singular value decomposition method; Singular value decomposition is a mathematical method used for data dimension reduction, pattern extraction and denoising. It can decompose complex data into different characteristic patterns, extract the core information, and remove random fluctuations and noise. In the analysis of supplier on-time delivery rate, the singular value decomposition method analyzes historical delivery data, identifies long-term stable delivery trends, eliminates abnormal fluctuations, and calculates the on-time delivery coefficient to reflect the overall delivery stability of the supplier. This method can help companies more accurately evaluate the reliability of suppliers, optimize procurement decisions, and reduce supply chain risks.

[0026] The benefit of calculating the on-time delivery coefficient by singular value decomposition is that it can extract key patterns from the historical on-time delivery rate data of materials, remove accidental fluctuations and outliers, and improve the accuracy of supplier delivery capacity assessment. The singular value decomposition method can identify long-term stable delivery trends, making the calculated on-time delivery coefficient more reliable and avoiding misjudgments caused by individual abnormal deliveries or short-term fluctuations.

[0027] Obtain the quality pass rate of the supplier during the inspection period, and calculate the quality pass coefficient based on the quality pass rate through the K-means clustering method; K-means clustering is a commonly used unsupervised learning algorithm that is used to divide a data set into K different categories, so that the similarity of data points within the same category is maximized and the similarity between different categories is minimized. Its core idea is to iteratively optimize the distance from the data point to the cluster center, continuously adjust the category assignment, and finally obtain a stable clustering result. In the quality pass rate analysis, K-means clustering can divide suppliers into different quality grade coefficients according to their quality pass rates at different time points, and calculate the quality pass, which is used to measure the long-term quality stability of suppliers.

[0028] Obtain the supplier's material price and the average market price, and calculate the price competition coefficient by subtracting the supplier's material price from the average market price; Construct a supplier evaluation matrix based on the types of supplied materials, on-time delivery coefficient, quality qualification coefficient, price competitiveness coefficient and shipping distance. Take the data shown in Table 1 as an example: Table 1 Supplier evaluation matrix supplier Supply material type Delivery on-time factor Quality qualification factor Price competitiveness coefficient Shipping distance A 15 0.95 0.98 0.90 300 B 10 0.85 0.95 0.88 500 C 20 0.92 0.96 0.87 250 The types of supplied materials, on-time delivery coefficient, quality qualification coefficient, price competitiveness coefficient and delivery distance are normalized to construct a normalized supplier evaluation matrix, as shown in Table 2: Table 2 Normalized supplier evaluation matrix supplier Supply material type Delivery on-time factor Quality qualification factor Price competitiveness coefficient Shipping distance A 0.33 1.00 1.00 1.00 0.50 B 0.00 0.00 0.00 0.50 0.00 C 1.00 0.70 0.33 0.00 1.00 The indicator comparison difference between suppliers is calculated based on the normalized supplier evaluation matrix. The specific acquisition steps are as follows: ; In the formula, It is expressed as the advantage of supplier i over supplier k in indicator j, It is represented as the normalized value of supplier i on indicator j, It is represented as the normalized value of supplier k on indicator j; Calculate the preference function based on the difference in indicators between suppliers ,when hour, ,when hour, ; The overall preference value of every two suppliers is calculated according to the preference function. The specific steps are as follows: ; In the formula, Expressed as the overall preference value of supplier i relative to supplier k, It is represented as the weight of indicator j, and the sum of all indicator weights is 1, m is the number of indicators, is the preference function; The preference index is calculated based on the overall preference value of every two suppliers.

[0029] In this embodiment, it should be specifically explained that the steps of calculating the preference index according to the overall preference values ​​of every two suppliers are as follows: According to the normalized supplier evaluation matrix, the positive flow of the normalized supplier evaluation matrix is ​​calculated. The positive flow is a measure of the supplier's advantage over other suppliers, that is, the degree to which the supplier performs better than other suppliers in multiple evaluation indicators. It is determined by calculating the total preference value of a supplier over other suppliers in all pairwise comparisons. The higher the positive flow value, the stronger the competitiveness of the supplier in the comprehensive evaluation and the more suitable for priority selection. It indicates how much advantage a supplier has in all comparisons, that is, how much better it is than other suppliers. The specific steps to obtain it are: ; In the formula, is represented by the positive flow of the evaluation matrix of the i-th supplier, w is represented by the number of suppliers, It is expressed as the overall preference value of supplier i relative to supplier k; According to the normalized supplier evaluation matrix, the negative flow of the normalized supplier evaluation matrix is ​​calculated. The negative flow is a measure of the supplier's disadvantage relative to other suppliers, that is, the degree to which the supplier is surpassed by other suppliers in multiple evaluation indicators. It calculates the total preference value of the supplier that is inferior to other suppliers in all pairwise comparisons. The higher the negative flow value, the worse the overall performance of the supplier, and the more cautious the selection should be in purchasing decisions. It indicates how much disadvantage a supplier has in all comparisons, that is, how much worse it is than other suppliers. The specific steps to obtain it are: ; In the formula, Represented as the negative flow of the i-th supplier evaluation matrix, It is expressed as the overall preference value of supplier k relative to supplier i; The net flow of the normalized supplier evaluation matrix is ​​obtained by calculating the difference between the positive flow of the normalized supplier evaluation matrix and the negative flow of the normalized supplier evaluation matrix. The net flow is the difference between the positive flow and the negative flow, which is used to measure the comprehensive priority of suppliers. It reflects the net competitiveness of a supplier in all comparisons, that is, the degree to which the overall advantages of the supplier outweigh the disadvantages. The higher the net flow value, the better the performance of the supplier in all indicators, and should be the priority choice for procurement; if the net flow value is low or even negative, it means that the supplier is relatively weak in competitiveness and is not suitable for priority procurement, and the net flow of the normalized supplier evaluation matrix is ​​used as the priority index.

[0030] In this embodiment, it should be specifically explained that the steps for obtaining the delivery on-time coefficient are: The historical on-time delivery rate of materials is used to construct a historical on-time delivery rate vector according to the time series. The singular value decomposition method is used to set the window size, for example, the window size is 3, and a nested time window matrix is ​​constructed. For example, the matrix is: ; Perform singular value decomposition on the nested time window matrix to obtain a singular value matrix, for example, the matrix is: ; Select singular values ​​with energy accumulation exceeding 90% for retention, and reconstruct the data by dimensionality reduction based on the retained singular values ​​to obtain the denoised delivery on-time rate matrix; The data in the denoised delivery on-time rate matrix is ​​averaged to obtain the delivery on-time coefficient.

[0031] In this embodiment, it should be specifically explained that the steps for obtaining the quality qualification coefficient are: Step 2.1: Take the quality pass rate as the clustering feature, take all the quality pass rates within the detection time period as the data set, and each quality pass rate in the data set as a data point; Step 2.2: Use the silhouette coefficient method to determine the number of clusters K of the data set; The silhouette coefficient method is an unsupervised learning evaluation method used to measure the quality of data clustering to determine the number of clusters K. It calculates the silhouette coefficient of each data point to evaluate its compactness in the current cluster and its degree of separation from the nearest neighbor cluster. In general, the K that maximizes the average silhouette coefficient is selected as the optimal number of clusters to ensure that the data points are tightly clustered in each cluster and far away from other clusters, thereby improving the accuracy and stability of clustering.

[0032] Step 2.3: Randomly select K data points in the data set as the initial cluster centers. For each data point, calculate its Euclidean distance to each initial cluster center. For each data point, traverse the K initial cluster centers and assign it to the cluster corresponding to the nearest initial cluster center. Step 2.4: After traversing all data points, the initial clusters are obtained. For each initial cluster, the mean of the data points in it is calculated to obtain a new cluster center. Step 2.5: Repeat steps 2.3 and 2.4 until the cluster center no longer changes, and obtain the final cluster and the final cluster center; Step 2.6: Calculate the ratio of the number of data points in each final cluster to the total number of data points to obtain the weight of each final cluster, and perform a weighted summation of the weight of each final cluster and the final cluster center to obtain the quality qualification coefficient.

[0033] Step 3: Select an initial supplier according to the preference index, obtain the purchase cost, transportation cost and inventory cost of the initial supplier, and use a mathematical optimization model to obtain an optimal purchase plan based on the purchase cost, transportation cost and inventory cost of the initial supplier. It should be specifically noted that using a mathematical optimization model to obtain an optimal purchase plan based on the purchase cost, transportation cost and inventory cost of the initial supplier is a prior art, and this embodiment does not describe its specific steps in detail; Mathematical optimization model is a mathematical method used to solve the optimal decision-making plan. It calculates the optimal solution that can meet the demand by establishing objective functions (such as minimizing procurement costs) and constraints (such as supply capacity, inventory restrictions, transportation conditions, etc.). In supplier procurement optimization, mathematical optimization models can integrate procurement costs, transportation costs and inventory costs, and use linear programming, integer programming, mixed integer programming and other methods to calculate the optimal procurement plan to ensure cost minimization and maximize supply stability.

[0034] In this embodiment, it should be specifically explained that the steps of selecting the initial supplier according to the preference index are: Set the initial number of suppliers, sort the preference index of each supplier from high to low, and select the suppliers corresponding to the initial number of suppliers as the initial suppliers.

[0035] Step 4: Conduct joint procurement based on the optimal procurement plan and the demand for each material; Step 5: Continue to collect historical procurement-related data and supplier data in real time, and make real-time procurement adjustments.

[0036] In this embodiment, it should be specifically explained that Figure 2 As shown, a multi-material joint procurement optimization system for an intelligent manufacturing enterprise, the system includes: The historical data acquisition module is used to obtain historical procurement-related data, including material types, warehouse material inventory, and material demand. The demand for each material is predicted using the time series method based on the historical procurement-related data; The priority index acquisition module is used to screen out suppliers whose material inventory is greater than the material demand according to the material demand, and obtain the data of each supplier. The supplier data includes the type of supplied materials, the historical on-time delivery rate of materials, the quality qualification rate, the material price and the delivery distance. The priority index is obtained by evaluating the supplier data through the priority diagram method; The optimal solution acquisition module is used to select the initial supplier according to the preference index, obtain the purchase cost, transportation cost and inventory cost of the initial supplier, and use the mathematical optimization model to obtain the optimal purchase solution based on the purchase cost, transportation cost and inventory cost of the initial supplier; Joint purchasing module, used for joint purchasing based on the optimal purchasing plan and the demand for each material; The real-time adjustment module is used to continue to collect historical procurement-related data and supplier data in real time, and to make real-time procurement adjustments.

[0037] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

[0038] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A multi-material joint procurement optimization method for an intelligent manufacturing enterprise, characterized in that: The following steps are involved: Step 1: Obtain historical procurement-related data, which includes material types, warehouse material inventory, and material demand. Use the time series method to predict the demand for each material based on the historical procurement-related data. Step 2: Filter out suppliers whose material inventory is greater than the material demand based on the material demand, and obtain the data of each supplier. The supplier data includes the type of supplied materials, the historical on-time delivery rate of materials, the quality pass rate, the material price, and the delivery distance. The priority index is obtained based on the supplier data through the priority diagram method; Step 3: Select the initial supplier according to the preference index, obtain the purchase cost, transportation cost and inventory cost of the initial supplier, and use the mathematical optimization model to obtain the optimal purchase plan based on the purchase cost, transportation cost and inventory cost of the initial supplier; Step 4: Conduct joint procurement based on the optimal procurement plan and the demand for each material; Step 5: Continue to collect historical procurement-related data and supplier data in real time, and make real-time procurement adjustments.

2. According to claim 1, a multi-material joint procurement optimization method for an intelligent manufacturing enterprise is characterized by: The steps of using the time series method to predict the demand for each material based on historical procurement related data are as follows: Set the collection time period to obtain historical procurement-related data within the collection time period. Procurement-related data includes material types, warehouse material inventory, and demand for each material; Preprocess historical procurement-related data, including data cleaning and time series formatting; Select the ARIMA time series model, train the ARIMA time series model with historical procurement-related data to obtain the final forecast model, and use the final forecast model to predict the demand for each material.

3. The method for optimizing multi-material joint procurement of intelligent manufacturing enterprises according to claim 2 is characterized by: The steps of training the ARIMA time series model through historical procurement related data to obtain the final prediction model are as follows: The ADF unit root test was used to determine whether the data was stationary. If the data was not stationary, multiple order differences were used to transform the data until the data was stationary and the order of differences was obtained; The autoregressive order is obtained using the autocorrelation function graph and the partial autocorrelation function graph, and the moving average order is obtained using the autocorrelation function graph; Divide the historical purchase-related data into training set data and test set data, and use the training set data to fit the ARIMA model; Select the autoregressive order, difference order and moving average order, and use the least squares method to obtain the autoregressive coefficient and moving average coefficient to obtain the initial prediction model; Use the initial forecasting model to forecast the demand of the training set data to obtain the forecast demand, and calculate the mean square error, root mean square error, and mean absolute percentage error based on the forecast demand; If the mean absolute percentage error is less than the preset threshold, the initial prediction model is judged to have high accuracy and the initial prediction model is used as the final prediction model. If the mean absolute percentage error is greater than or equal to the preset threshold, the autoregressive order, difference order and moving average order of the initial prediction model are adjusted, and the model accuracy judgment is continued until the mean absolute percentage error is less than the preset threshold to obtain the final prediction model.

4. The method for optimizing multi-material joint procurement of intelligent manufacturing enterprises according to claim 1 is characterized by: The step of obtaining the preference index is as follows: Obtain the historical on-time delivery rate of materials during the supplier's inspection period, and calculate the on-time delivery coefficient based on the historical on-time delivery rate of materials using the singular value decomposition method; Obtain the quality pass rate of the supplier during the inspection period, and calculate the quality pass coefficient based on the quality pass rate through the K-means clustering method; Obtain the supplier's material price and the average market price, and calculate the price competition coefficient by subtracting the supplier's material price from the average market price; Construct a supplier evaluation matrix based on the types of supplied materials, on-time delivery coefficient, quality compliance coefficient, price competitiveness coefficient, and shipping distance; Normalize the types of supplied materials, on-time delivery coefficient, quality qualification coefficient, price competitiveness coefficient and shipping distance to construct a normalized supplier evaluation matrix; Calculate the comparison difference of indicators between suppliers based on the normalized supplier evaluation matrix; The preference function is calculated based on the difference in the indicators between suppliers, and the overall preference value of every two suppliers is calculated based on the preference function. The specific acquisition steps are as follows: ; In the formula, It is expressed as the overall preference value of supplier i relative to supplier k, It is expressed as the weight of indicator j, m is the number of indicators, is the preference function; The preference index is calculated based on the overall preference value of every two suppliers.

5. The method for optimizing multi-material joint procurement of intelligent manufacturing enterprises according to claim 4 is characterized by: The steps of calculating the preference index according to the overall preference value of every two suppliers are as follows: According to the normalized supplier evaluation matrix, the positive flow of the normalized supplier evaluation matrix is ​​calculated. The specific acquisition steps are: ； In the formula, is represented by the positive flow of the evaluation matrix of the i-th supplier, w is represented by the number of suppliers, Expressed as the overall preference value of supplier i relative to supplier k According to the normalized supplier evaluation matrix, the negative flow of the normalized supplier evaluation matrix is ​​calculated. The specific acquisition steps are: ； In the formula, Represented as the negative flow of the i-th supplier evaluation matrix, It is expressed as the overall preference value of supplier k relative to supplier i; The net flow of the normalized supplier evaluation matrix is ​​obtained by performing a difference calculation between the positive flow of the normalized supplier evaluation matrix and the negative flow of the normalized supplier evaluation matrix, and the net flow of the normalized supplier evaluation matrix is ​​used as the priority selection index.

6. The method for optimizing multi-material joint procurement of intelligent manufacturing enterprises according to claim 4 is characterized by: The steps for obtaining the delivery on-time coefficient are: The historical on-time delivery rate of materials is used to construct a historical on-time delivery rate vector based on the time series, and the singular value decomposition method is used to set the window size; Perform singular value decomposition on the nested time window matrix to obtain a singular value matrix; Select singular values ​​whose energy accumulation exceeds the set value and keep them. Reconstruct the data by dimension reduction based on the retained singular values ​​to obtain the denoised delivery on-time rate matrix. The data in the denoised delivery on-time rate matrix is ​​averaged to obtain the delivery on-time coefficient.

7. The method for optimizing multi-material joint procurement of intelligent manufacturing enterprises according to claim 4 is characterized by: The steps for obtaining the quality qualification coefficient are as follows: Step 2.1: Take the quality pass rate as the clustering feature, take all the quality pass rates within the detection time period as the data set, and each quality pass rate in the data set as a data point; Step 2.2: Use the silhouette coefficient method to determine the number of clusters K of the data set; Step 2.3: Randomly select K data points in the data set as the initial cluster centers. For each data point, calculate its Euclidean distance to each initial cluster center. For each data point, traverse the K initial cluster centers and assign it to the cluster corresponding to the nearest initial cluster center. Step 2.4: After traversing all data points, the initial clusters are obtained. For each initial cluster, the mean of the data points in it is calculated to obtain a new cluster center. Step 2.5: Repeat steps 2.3 and 2.4 until the cluster center no longer changes, and obtain the final cluster and the final cluster center; Step 2.6: Calculate the ratio of the number of data points in each final cluster to the total number of data points to obtain the weight of each final cluster, and perform a weighted summation of the weight of each final cluster and the final cluster center to obtain the quality qualification coefficient.

8. A multi-material joint procurement optimization system for an intelligent manufacturing enterprise, used to implement a multi-material joint procurement optimization method for an intelligent manufacturing enterprise as described in any one of claims 1 to 7, characterized in that: The system comprises: The historical data acquisition module is used to obtain historical procurement-related data, including material types, warehouse material inventory, and material demand. The demand for each material is predicted using the time series method based on the historical procurement-related data; The priority index acquisition module is used to screen out suppliers whose material inventory is greater than the material demand according to the material demand, and obtain the data of each supplier. The supplier data includes the type of supplied materials, the historical on-time delivery rate of materials, the quality qualification rate, the material price and the delivery distance. The priority index is obtained by evaluating the supplier data through the priority diagram method; The optimal solution acquisition module is used to select the initial supplier according to the preference index, obtain the purchase cost, transportation cost and inventory cost of the initial supplier, and use the mathematical optimization model to obtain the optimal purchase solution based on the purchase cost, transportation cost and inventory cost of the initial supplier; Joint purchasing module, used for joint purchasing based on the optimal purchasing plan and the demand for each material; The real-time adjustment module is used to continue to collect historical procurement-related data and supplier data in real time, and to make real-time procurement adjustments.

Images