Intelligent purchasing system and method of supply chain

Abstract

The invention belongs to the technical field of supply chain management, and discloses an intelligent purchasing system and method of a supply chain. Comprising the steps of collecting demand prediction data and supplier performance data; carrying out deep mining on the demand prediction data, and predicting future purchase demands; generating a capability portrait based on the performance data of the suppliers, and performing quantitative evaluation on performance scores of different suppliers according to a pre-constructed evaluation system; historical order data is acquired, a supplier relation network topology is constructed in combination with the capability portraits, and a synergistic effect between suppliers is identified; fusing the performance score, the future purchase demand and the synergistic effect, and recommending an optimal supplier combination; acquiring purchase management data, and intelligently formulating a purchase strategy in combination with the supplier performance data of the optimal supplier combination; according to the invention, the purchase efficiency can be improved, and the flexibility and toughness of the supply chain are enhanced, so that the overall response speed, cost control capability and operation quality of the supply chain are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of supply chain management. More specifically, the present invention relates to an intelligent procurement system and method for a supply chain. Background Art

[0002] With the in-depth development of global economic integration and digital transformation, supply chain management has become an important link for enterprises to enhance their core competitiveness; procurement, as a key link in the supply chain, is directly related to cost control, production efficiency, and service level; traditional procurement systems mostly adopt manual management or rule-based static decision-making models, relying on human experience for supplier selection and procurement strategy formulation, with many problems such as lagging response, low efficiency, and information fragmentation, and unable to form a closed-loop management; especially in the face of large-scale, multi-category, and cross-regional procurement scenarios, it is difficult to support intelligent analysis and dynamic optimization of multi-dimensional data, restricting the scientificity and forward-looking nature of procurement decisions; therefore, there is an urgent need for an intelligent procurement system to achieve intelligent management of the entire supply chain process and improve enterprise procurement efficiency and supply chain resilience.

[0003] The patent with the publication number CN118674366A discloses an intelligent procurement system for a supply chain and a supply chain procurement method; including: a supply chain management system, a procurement system, an inventory management system, and a user permission management system; the supply chain management system includes a supplier management module, a supply chain monitoring module, and a supplier evaluation module; the procurement system includes a procurement demand management module, a procurement order management module, a cost control module, a data analysis and reporting module, an artificial intelligence decision support module, and an automated procurement module; the inventory management system includes an inventory management module; the user management system includes a user permission management module. With the use of this intelligent procurement system, the present invention can reduce manual procurement, effectively reduce the procurement error rate, improve procurement efficiency, reduce procurement costs, reduce procurement risks, and enhance the competitiveness of enterprises.

[0004] However, although the above technology realizes intelligent procurement of the supply chain, it mainly focuses on the organizational structure division of system modules, lacking in-depth elaboration on the core technical implementation means, especially the lack of specific technical methods in key links such as supplier evaluation and procurement decision-making; at the same time, the above technology fails to fully consider the collaborative relationship between suppliers, lacking modeling analysis of the supply chain network structure, and it is difficult to identify potential synergistic effects therein, resulting in difficulty in achieving dynamic perception and high-dimensional refined optimization in complex and changing procurement environments.

[0005] In view of this, the present invention proposes an intelligent procurement system and method for a supply chain to solve the above problems. Summary of the Invention

[0006] To overcome the above-mentioned defects of the prior art and achieve the above object, the present invention provides the following technical solutions: An intelligent procurement method for a supply chain, including: Collect supply chain management data, where the supply chain management data includes demand forecast data and supplier performance data; Deeply mine the demand forecast data and use deep learning technology to dynamically predict future procurement demands; Based on the supplier performance data, generate ability portraits of different suppliers, and quantitatively evaluate the performance scores of different suppliers according to a pre-constructed evaluation system; Obtain historical order data, integrate the historical order data with the ability portraits of different suppliers, construct a supplier relationship network topology, and identify the synergy effects among suppliers; Integrate the performance scores, future procurement demands, and synergy effects, and apply an intelligent matching algorithm to recommend the optimal supplier combination; Obtain procurement management data, and in combination with the supplier performance data corresponding to the optimal supplier combination, use multi-objective optimization technology to intelligently formulate personalized procurement strategies.

[0007] Further, the demand forecast data includes historical procurement data, market trend data, and demand procurement categories; the historical procurement data includes historical procurement time, historical procurement categories, and historical procurement quantities; the market trend data includes raw material price fluctuation data and market demand change data, the raw material price fluctuation data includes the market prices of raw materials required for different supply chain products at different time points, and the market demand change data includes the total demand for different supply chain products at different time points; the supplier performance data includes the performance data corresponding to each supplier;

[0008] The method for dynamically predicting future procurement demands includes: Take the historical procurement quantities with the same corresponding historical procurement categories as a set of quantity sets; preset a time interval, and construct a time axis for each set of quantity sets based on the time interval; where, on each time axis, the time difference between every two adjacent time points is equal to the time interval, and each time point corresponds to a historical procurement quantity in the corresponding quantity set; According to the historical procurement quantity corresponding to each time point on each time axis, construct a quantity prediction model for each set of quantity sets, and the quantity prediction model is a recurrent neural network model; preset a time window with a length of which is an integer greater than 1; mark each set of quantity sets corresponding to each demand procurement category as a demand set, and mark the time axis corresponding to the demand set as a demand axis; starting from the last time point on each demand axis, obtain a set of time sets from the demand axis corresponding to each demand set based on the time window, and each set of time sets includes which is an integer greater than 1; Time points; input the historical purchase quantities corresponding to each set of time collections into the corresponding quantity prediction models respectively to predict the corresponding future purchase quantities; According to the market trend data, calculate the first factor and the second factor corresponding to each set of demand collections in sequence; set the weight coefficients of each first factor and each second factor respectively, multiply each first factor and each second factor by the corresponding weight coefficients in sequence to obtain the influence weights, and take the mean of the influence weights with the same corresponding demand collections as the comprehensive influence factor; multiply each future purchase quantity by the corresponding comprehensive influence factor to obtain the predicted purchase quantity of each set of demand collections and use it as the future purchase demand.

[0009] Furthermore, the method for constructing a time axis for the quantity collection includes: Mark the earliest historical purchase time corresponding to the historical purchase quantity in the quantity collection as the earliest time, and use the earliest time as the first time point of the time axis; starting from the earliest time, generate consecutive time points on the time axis according to the time interval; compare the time points on the time axis with the historical purchase times corresponding to each historical purchase quantity in the quantity collection respectively; map the historical purchase quantity corresponding to the historical purchase time that is the same as the time point to the corresponding time point in sequence; mark the time points that are not the same as all historical purchase times as interpolation points, and use polynomial interpolation to calculate the historical purchase quantity corresponding to each interpolation point and map it to the corresponding interpolation point; The method for calculating the first factor and the second factor corresponding to each set of demand collections includes: Map the market trend data to each time point on each demand axis respectively, and each time point corresponds to a set of market prices and a total demand quantity; construct a market impact model for each set of demand collections according to the market price, total demand quantity, and historical purchase quantity corresponding to each time point on each demand axis; the market impact model is a deep neural network model; mark the earliest time point in each set of time collections as the temporary early point, and starting from the time point one before the corresponding temporary early point on each demand axis, obtain a set of pre-time collections on each demand axis based on the time window, and each set of pre-time collections includes Time points; input the market price, total demand quantity, and historical purchase quantity corresponding to each set of time collections into the corresponding market impact models respectively to predict the corresponding first factor; input the market price, total demand quantity, and historical purchase quantity corresponding to each set of pre-time collections into the corresponding market impact models respectively to predict the corresponding second factor.

[0010] Furthermore, the method for generating the ability portraits of different suppliers includes: Based on the performance data corresponding to different suppliers, evaluate the ability indicators corresponding to each supplier. The ability indicators include delivery ability, quality ability, price ability, and service ability. According to the ability indicators corresponding to each supplier, construct a corresponding bar chart for each supplier, and use the bar chart as the ability portrait of the corresponding supplier.

[0011] Further, the steps for evaluating the delivery ability corresponding to each supplier include: Step S101: Construct multiple fuzzy sets for each data in the delivery performance data respectively; Step S102: Convert the delivery performance data of each supplier into the membership degrees of the corresponding fuzzy sets respectively through fuzzy technology; Step S103: Define fuzzy rules; Step S104: Match each group of fuzzified technical feature data with the fuzzy rules respectively, and perform fuzzy inference using the fuzzy inference method to obtain the fuzzy inference results corresponding to each supplier. The fuzzy inference results are the membership degrees of each delivery ability level; Step S105: Set the ability interval. The range of the ability interval is , is an integer greater than 1; evenly divide the ability interval into four grade intervals, and the grade intervals correspond one by one to the grades in the delivery ability level; add the maximum value and the minimum value of each grade interval and divide by 2 to obtain the interval mean of each grade interval; Step S106: Multiply the membership degree of each delivery ability level corresponding to each supplier by the corresponding interval mean respectively, and then add them up in turn to obtain the total ability value of each supplier; add up the membership degrees of each delivery ability level corresponding to each supplier in turn to obtain the total membership degree; divide the total ability value of each supplier by the corresponding total membership degree as the delivery ability of each supplier; The steps for evaluating the quality ability, price ability, and service ability corresponding to each supplier are the same as the steps for evaluating the delivery ability corresponding to each supplier; The method for quantitatively evaluating the performance scores of different suppliers includes: Preset a set of ratios, and the set of ratios includes the ratio coefficients corresponding to each ability indicator; multiply each ability indicator of each supplier by the corresponding ratio coefficient in the set of ratios respectively, and then add them up in turn to obtain the performance score corresponding to each supplier.

[0012] Further, the historical order data includes the performance data corresponding to each historical cooperation order; The method for constructing the supplier relationship network topology includes: Regarding multiple suppliers corresponding to each historical cooperation order as a set of supplier sets, and regarding historical cooperation orders with the same supplier set as a set of order sets; inputting the performance data corresponding to each set of order sets into a trained performance evaluation model respectively to predict the corresponding performance data, where the performance evaluation model is a deep neural network model; evaluating the ability indicators corresponding to each set of order sets based on the performance data corresponding to different order sets; regarding each supplier as a node, establishing an edge between the nodes corresponding to every two suppliers in each set of supplier sets, and the weight of each edge is the ability indicator corresponding to each order set; constructing a supplier relationship network topology according to the nodes, edges and the weights of the edges, and taking the ability portrait of each supplier as the attribute of the corresponding node. The steps for identifying the synergy effect between suppliers include: Step S201: Mark the ability indicators in the attributes corresponding to each node as the first indicator, and mark the ability indicators corresponding to the weights of the edges as the second indicator. Step S202: Randomly select a set of supplier sets that have not been marked as selected sets and mark it as the current set. Step S203: Compare each first indicator corresponding to each node in the current set with the second indicator corresponding to the corresponding order set in turn; if all first indicators are less than or equal to the corresponding second indicators, then all nodes in the corresponding current set have a positive synergy effect, and mark the corresponding current set as a synergy set; if there is a first indicator greater than the corresponding second indicator, then there are nodes in the corresponding current set with a negative synergy effect, and do not mark the corresponding current set. Step S204: Mark the current set as a selected set. Step S205: Loop through steps S202 to S204 until all supplier sets are marked as selected sets, and when the loop ends, obtain all synergy sets.

[0013] Furthermore, the method for recommending the optimal supplier combination includes: Based on historical cooperation orders, obtain the historical procurement categories corresponding to each collaboration set and mark them as cooperative procurement categories; based on historical procurement data, obtain the historical procurement categories corresponding to each supplier and mark them as independent procurement categories; compare each demand procurement category with each cooperative procurement category; if there is a demand procurement category that is the same as the cooperative procurement category, mark the corresponding collaboration set as a recommended set; if all demand procurement categories are different from the cooperative procurement categories, do not mark the corresponding collaboration set; mark the demand procurement categories that are different from all cooperative procurement categories as remaining procurement categories, and mark the suppliers corresponding to the independent procurement categories that are the same as the remaining procurement categories as candidate suppliers; group the candidate suppliers with the same corresponding independent procurement categories as a group of candidate sets, and the candidate sets correspond one-to-one with the independent procurement categories; According to future procurement requirements, obtain the estimated procurement quantity of each remaining procurement category; according to historical procurement data, obtain the historical procurement quantity corresponding to each candidate supplier, and mark the largest historical procurement quantity corresponding to each candidate supplier as the maximum quantity; compare each maximum quantity with the corresponding estimated procurement quantity, delete the candidate suppliers whose maximum quantity is less than the corresponding estimated procurement quantity from the corresponding candidate set, and the candidate suppliers whose maximum quantity is greater than or equal to the estimated procurement quantity remain in the corresponding candidate set; sort each candidate supplier in each candidate set from largest to smallest according to the corresponding performance score to generate a supplier ranking list; mark the top candidate suppliers in each supplier ranking list as the best suppliers, where is an integer greater than 0; combine each best supplier with each recommended set to obtain the optimal supplier combination.

[0014] Furthermore, the procurement management data includes category inventory, procurement budget, and supply cycle; the category inventory includes the inventory quantity of each demand procurement category; the procurement budget includes the budget amount of each demand procurement category; the supply cycle includes the delivery cycle of each demand procurement category; The method for formulating a personalized procurement strategy includes: Construct groups of different procurement sets, and set sequentially increasing numerical labels for the groups of procurement sets and mark them as procurement labels, and the range of the procurement labels is ; randomly select a procurement label as the initial iteration center; Define the iteration process, and the iteration process is: according to the iteration center, generate candidate solutions within the range of the procurement labels, and calculate the procurement performance evaluation value corresponding to each candidate solution, , the candidate solutions and the procurement labels correspond one by one; mark the candidate solution with the largest procurement performance evaluation value as the local solution, move the iteration center to the local solution, and calculate the index change amount; Execute the iteration process. When each iteration process is completed, compare the index change amount with the preset change threshold; if the index change amount is less than the change threshold, generate a completion instruction; if the index change amount is greater than or equal to the change threshold, do not generate a completion instruction; count the number of generated completion instructions and mark it as the instruction count; when the instruction count is greater than or equal to the preset count threshold, stop executing the iteration process; mark the procurement label corresponding to the iteration center as the best label, and use the procurement set corresponding to the best label as the procurement strategy.

[0015] Furthermore, the method for calculating the procurement performance evaluation value corresponding to the candidate solution includes: Obtain the procurement set corresponding to the procurement label of the candidate solution and mark it as the calculation set; obtain the supplier performance data corresponding to each supplier in the calculation set and mark it as the calculation data; use the calculation data, the calculation set, and the procurement management data as the performance prediction data; input the performance prediction data into the trained performance prediction model respectively to predict the corresponding procurement performance data; among them, the performance prediction model includes a cost prediction model, a supply prediction model, and a quality prediction model, and the performance prediction models are all deep neural network models; the procurement performance data includes procurement cost, supply cycle, and product quality; Subtract the procurement cost from the procurement budget to obtain the budget balance; subtract the supply cycle from the delivery cycle to obtain the delivery deviation; preset an index weight set, and the index weight set includes the index weights corresponding to the budget balance, the delivery deviation, and the product quality; multiply the budget balance, the delivery deviation, and the product quality by the corresponding index weights in the index weight set respectively to obtain the budget coefficient, the delivery coefficient, and the quality coefficient; add the budget coefficient, the quality coefficient, and the delivery coefficient to obtain the procurement performance evaluation value; The method for calculating the index change amount is: subtract the procurement performance evaluation value corresponding to the iteration center in the current iteration process from the procurement performance evaluation value corresponding to the iteration center in the previous iteration process, and take the absolute value to obtain the index change amount.

[0016] An intelligent procurement system for a supply chain, implementing the described intelligent procurement method for a supply chain, includes: A data collection module for collecting supply chain management data, and the supply chain management data includes demand forecast data and supplier performance data; A demand forecasting module for deeply mining the demand forecast data and dynamically forecasting future procurement demands using deep learning technology; A supplier evaluation module, which is used to generate ability portraits of different suppliers based on supplier performance data, and quantitatively evaluate the performance scores of different suppliers according to a pre-constructed evaluation system; A supplier recommendation module, which is used to obtain historical order data, integrate the historical order data with the ability portraits of different suppliers, construct a supplier relationship network topology, and identify the synergy effects among suppliers; fuse the performance scores, future procurement requirements and synergy effects, and apply an intelligent matching algorithm to recommend the optimal supplier combination; A strategy formulation module, which is used to obtain procurement management data, and combine the supplier performance data corresponding to the optimal supplier combination, and use multi-objective optimization technology to intelligently formulate personalized procurement strategies.

[0017] The technical effects and advantages of an intelligent procurement system and method for a supply chain according to the present invention: Through deep learning technology, integrating multi-dimensional factors such as historical procurement data and market trend data, realizing accurate perception and prediction of future procurement requirements; constructing ability portraits based on supplier performance data, and comprehensively evaluating suppliers by combining fuzzy evaluation and quantitative analysis methods, which can more objectively quantify the performance of suppliers in key capabilities such as delivery, quality, and price; by analyzing the supplier relationship network, identifying the synergy effects among suppliers, providing an important basis for recommending the optimal supplier combination, being beneficial to giving play to the synergy advantages among suppliers, and improving the overall efficiency of the supply chain; in the procurement strategy formulation stage, integrating multi-dimensional constraint conditions such as supplier performance, procurement requirements, and inventory, and using an iterative optimization method to dynamically generate personalized procurement plans, achieving refined balance and dynamic adjustment among multiple objectives such as cost, delivery, and quality, and maximizing the scientificity and effectiveness of procurement decisions; not only improving procurement efficiency, reducing costs, but also enhancing the flexibility and resilience of the supply chain, thereby effectively improving the overall response speed, cost control ability and operation quality of the supply chain, and enhancing the market competitiveness of enterprises. Description of the Drawings

[0018] Figure 1 It is a schematic diagram of an intelligent procurement system for a supply chain according to Embodiment 1 of the present invention; Figure 2 It is a flowchart of an intelligent procurement method for a supply chain according to Embodiment 2 of the present invention. Detailed Embodiments

[0019] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0020] Example 1

[0021] Please refer to Figure 1 As shown, an intelligent procurement system for a supply chain in this embodiment includes a data collection module, a demand forecasting module, a supplier evaluation module, a supplier recommendation module, and a strategy formulation module; each module is connected by wired and / or wireless means to achieve data transmission between modules.

[0022] The data collection module is used to collect supply chain management data, and the supply chain management data includes demand forecasting data and supplier performance data.

[0023] The demand forecasting data refers to the relevant data used to predict the changing trend of future procurement demands, which helps enterprises formulate reasonable procurement plans in advance and achieve the optimal allocation of supply chain resources; the demand forecasting data includes historical procurement data, market trend data, and demand procurement categories; the historical procurement data refers to the historical procurement records of enterprises, which are used to reflect past procurement behavior patterns and demand laws; the historical procurement data includes historical procurement time, historical procurement categories, and historical procurement quantities; the historical procurement categories refer to various supply chain products involved in the historical procurement activities of enterprises; the market trend data refers to the external macro factors related to enterprise procurement, which are used to evaluate the external changes affecting procurement demands in the future; the market trend data includes raw material price fluctuation data and market demand change data. The raw material price fluctuation data is the market price change of the raw materials required for supply chain products, including the market prices of the raw materials required for different supply chain products at different time points; the market demand change data is the fluctuation of the total market demand for supply chain products, including the total demand for different supply chain products at different time points; the demand procurement categories refer to various supply chain products involved in the current procurement process of enterprises based on actual business needs; The historical procurement data is obtained from the enterprise resource planning system (such as the ERP system) or the supply chain management system (such as the SCM system) within the enterprise; the market trend data is obtained from the market reports regularly released by industry analysis companies (such as LeadLeo Research Institute, iResearch, etc.) or the public data of futures exchanges (such as Shanghai Futures Exchange, Zhengzhou Commodity Exchange, etc.).

[0024] Supplier performance data refers to the relevant data that measures the performance of suppliers in historical cooperation, which helps to quantitatively evaluate and intelligently screen suppliers, and further provides data support for subsequent supplier recommendation and procurement strategy formulation; supplier performance data includes the performance data corresponding to each supplier, and the performance data includes delivery performance data, quality performance data, price performance data, and service performance data; delivery performance data reflects the delivery ability and time control level of suppliers during the order execution process, including on-time delivery rate, average delivery cycle, order fulfillment rate, etc.; quality performance data is used to measure the quality stability and qualification level of products provided by suppliers, including product qualification rate, quality defect rate, repair rate, etc.; price performance data reflects the performance of suppliers in cost control and price stability, including average quotation, quotation floating situation, etc.; service performance data is used to evaluate the soft capabilities of suppliers such as communication efficiency and after-sales service during the cooperation process, including after-sales service satisfaction, problem handling efficiency, etc.; Delivery performance data is obtained through the enterprise resource planning system (i.e., ERP system) or the supply chain management system (i.e., SCM system) within the enterprise; quality performance data is obtained through the quality management system (i.e., QMS system) within the enterprise; price performance data is obtained through historical procurement contracts; service performance data is obtained through the supplier relationship management system (i.e., SRM system) within the enterprise.

[0025] The demand forecasting module is used to deeply mine demand forecasting data and dynamically predict future procurement demands using deep learning technology.

[0026] The methods for dynamically predicting future procurement demands include: Taking the historical procurement quantities with the same corresponding historical procurement categories as a set of quantity collections, and the quantity collections correspond one-to-one with the historical procurement categories; presetting a time interval, and constructing a time axis for each set of quantity collections based on the time interval, and the time interval is preset by those skilled in the art according to the actual situation; wherein, on each time axis, the time difference between every two adjacent time points is equal to the time interval, and each time point corresponds to a historical procurement quantity in the corresponding quantity collection; it should be noted that the time difference between the last time point on each time axis and the corresponding latest time is less than the time interval, and the latest time is the latest historical procurement time corresponding to the historical procurement quantity in the quantity collection; Based on the historical procurement quantities corresponding to each time point on each time axis, constructing a quantity forecasting model for each set of quantity collections, and the quantity forecasting models correspond one-to-one with the quantity collections; wherein, the quantity forecasting model is a recurrent neural network model, and the recurrent neural network model is prior art, and the specific construction process will not be elaborated here. The recurrent neural network model can process time series data and capture the changing trend of historical procurement quantities, so as to predict the procurement quantities at future time points; presetting the length as time window, where \(n\) is an integer greater than 1, and the time window is preset by those skilled in the art according to the actual situation; mark the quantity set corresponding to each demand purchase category as a demand set, and mark the time axis corresponding to the demand set as a demand axis; starting from the last time point on each demand axis, obtain a set of time sets from each demand axis based on the time window, and each set of time sets includes time points; input the historical purchase quantities corresponding to each set of time sets into the corresponding quantity prediction model respectively to predict the corresponding future purchase quantities, and the future purchase quantities correspond one-to-one with the demand purchase categories; According to the market trend data, calculate the first factor and the second factor corresponding to each set of demand sets in turn; set the weight coefficients of each first factor and each second factor respectively, multiply each first factor and each second factor by the corresponding weight coefficients in turn to obtain the influence weights, and take the average value of the influence weights with the same corresponding demand set as the comprehensive influence factor; multiply each future purchase quantity by the corresponding comprehensive influence factor to obtain the predicted purchase quantity of each set of demand sets and use it as the future purchase demand.

[0027] The method for constructing a time axis for the quantity set includes: Mark the earliest historical purchase time corresponding to the historical purchase quantity in the quantity set as the earliest time, and use the earliest time as the first time point of the time axis; starting from the earliest time, generate continuous time points on the time axis according to the time interval; compare the time points on the time axis with the historical purchase time corresponding to each historical purchase quantity in the quantity set respectively; map the historical purchase quantity corresponding to the historical purchase time that is the same as the time point to the corresponding time point in turn; mark the time points that are not the same as all historical purchase times as interpolation points, and use polynomial interpolation methods (such as Lagrange interpolation method, Newton interpolation method, etc.) to calculate the historical purchase quantity corresponding to each interpolation point and map it to the corresponding interpolation point.

[0028] The method for calculating the first factor and the second factor corresponding to each set of demand sets includes: Map the market trend data to each time point on each demand axis respectively, and each time point corresponds to a set of market prices and a total demand; construct a market impact model for each set of demand sets according to the market price, total demand and historical purchase quantity corresponding to each time point on each demand axis, and the market impact models correspond one-to-one with the demand sets; the market impact model is a deep neural network model, and the deep neural network model is prior art, and the specific construction process will not be elaborated here too much. The deep neural network model can effectively capture the influence of non-linear factors such as market price fluctuations and changes in total demand on the purchase quantity; Mark the earliest time point in each set of time collections as the tentative early point. Starting from the time point one before the corresponding tentative early point on each demand axis, obtain a set of pre-time collections from each demand axis based on the time window. Each set of pre-time collections includes time points; input the market price, total demand, and historical purchase quantity corresponding to each set of time collections into the corresponding market impact model respectively to predict the corresponding first factor; input the market price, total demand, and historical purchase quantity corresponding to each set of pre-time collections into the corresponding market impact model respectively to predict the corresponding second factor.

[0029] Exemplarily, there are 7 time points on the demand axis, the length of the time window is 3, where time point 1 is the first time point and time point 7 is the last time point; thus, the time collection includes time point 7, time point 6, and time point 5; since the earliest time point in the time collection is time point 5, time point 5 is the tentative early point, and the pre-time collection includes time point 4, time point 3, and time point 2.

[0030] The method for separately setting the weight coefficients of each first factor and each second factor includes: Take the market price and total demand corresponding to the time collection of each set of demand collections as a set of volatility analysis data. The volatility analysis data corresponds one-to-one with the demand collections; input each set of volatility analysis data into the trained market analysis model respectively to predict the corresponding market volatility coefficient; among them, the market analysis model is a deep neural network model, and the market volatility coefficient is a comprehensive index used to quantitatively describe the volatility degree of the market price and market demand within the time collection, and is used to reflect the stability or turmoil degree of the market; preset a coefficient threshold and a weight set, and both the coefficient threshold and the weight set are pre-set by those skilled in the art according to the actual situation; among them, the weight set includes weight coefficient and weight coefficient , , and ; compare the market volatility coefficient with the coefficient threshold; if the market volatility coefficient is greater than the coefficient threshold, then take the weight coefficient as the weight coefficient of the first factor, and take the weight coefficient as the weight coefficient of the second factor; if the market volatility coefficient is less than or equal to the coefficient threshold, then take the weight coefficient as the weight coefficient of the first factor, and take the weight coefficient as the weight coefficient of the second factor.

[0031] The supplier evaluation module is used to generate the ability portraits of different suppliers based on the supplier performance data, and quantitatively evaluate the performance scores of different suppliers according to the pre-constructed evaluation system.

[0032] The method for generating the capability portraits of different suppliers includes: Based on the performance data corresponding to different suppliers, evaluate the capability indicators corresponding to each supplier. The capability indicators include delivery capability, quality capability, price capability, and service capability; according to the capability indicators corresponding to each supplier, construct a corresponding bar chart for each supplier, and use the bar chart as the capability portrait of the corresponding supplier; among them, each bar in the bar chart corresponds to a capability indicator, which is used to intuitively display the performance of each supplier in different capability dimensions, and show the comprehensive capability characteristics, advantages and disadvantages of the supplier.

[0033] The steps for evaluating the delivery capability corresponding to each supplier include: Step S101: Construct multiple fuzzy sets for each data in the delivery performance data respectively; for example, the fuzzy sets corresponding to the on-time delivery rate are high on-time delivery rate, medium on-time delivery rate, low on-time delivery rate, etc., and the fuzzy sets corresponding to the average delivery cycle are long average delivery cycle, medium average delivery cycle, short average delivery cycle, etc.; Step S102: Convert the delivery performance data of each supplier into the membership degrees of the corresponding fuzzy sets respectively through the fuzzyfication technology; the fuzzyfication technology is the process of converting accurate numerical values into the membership degrees corresponding to fuzzy sets. The fuzzyfication technology includes, for example, triangular membership functions, trapezoidal membership functions, etc.; for example, if the numerical value of the on-time delivery rate is relatively high, it is inferred that the membership degree of high on-time delivery rate is 0.9, the membership degree of medium on-time delivery rate is 0.3, and the membership degree of low on-time delivery rate is 0; Step S103: Define fuzzy rules, and the fuzzy rules are defined according to expert knowledge or relevant literature; for example, if the on-time delivery rate is high, the average delivery cycle is short, and the order fulfillment rate is high, it is inferred that the membership degree of the delivery capability level of the corresponding supplier belonging to excellent is high; if the on-time delivery rate is low, the average delivery cycle is long, and the order fulfillment rate is low, it is inferred that the membership degree of the delivery capability of the corresponding supplier belonging to failing is high; Step S104: Match each group of fuzzified technical feature data with the fuzzy rules respectively, and use fuzzy inference methods (such as Mamdani fuzzy inference model, Sugeno fuzzy inference model, etc.) for fuzzy inference to obtain the fuzzy inference results corresponding to each supplier. The fuzzy inference results are the membership degrees of each delivery capability level, and the delivery capability levels include excellent, good, passing, and failing; the fuzzy inference results are, for example, the membership degree of excellent is 0.6, the membership degree of good is 0.8, the membership degree of passing is 0.1, and the membership degree of failing is 0; Step S105: Set the capability interval, and the range of the capability interval is , is an integer greater than 1. In this embodiment, it is preferably is 100; evenly divide the ability interval into four grade intervals, which correspond one by one to the grades in the delivery ability level; add the maximum value of each grade interval to the corresponding minimum value and then divide by 2 to obtain the interval mean of each grade interval; Step S106: Multiply the membership degree of each delivery ability level corresponding to each supplier by the corresponding interval mean, and then add them up in turn to obtain the total ability value of each supplier; add up the membership degrees of each delivery ability level corresponding to each supplier in turn to obtain the total membership degree; divide the total ability value of each supplier by the corresponding total membership degree as the delivery ability of each supplier.

[0034] The steps for evaluating the quality ability, price ability, and service ability corresponding to each supplier are the same as the steps for evaluating the delivery ability corresponding to each supplier.

[0035] The method for quantitatively evaluating the performance scores of different suppliers includes: Preset a set of ratios, where the set of ratios includes the ratio coefficients corresponding to each ability index, and the set of ratios is preset by those skilled in the art according to the actual importance of each ability index; multiply each ability index of each supplier by the corresponding ratio coefficient in the set of ratios, and then add them up in turn to obtain the performance score corresponding to each supplier.

[0036] The supplier recommendation module is used to obtain historical order data, integrate the historical order data with the ability portraits of different suppliers, construct a supplier relationship network topology, and identify the synergy effects among suppliers; fuse the performance scores, future procurement requirements, and synergy effects, and apply an intelligent matching algorithm to recommend the optimal supplier combination.

[0037] The historical order data includes the performance data corresponding to each historical cooperation order. The historical cooperation order is an order in which multiple suppliers cooperate together in the enterprise's historical procurement activities; the performance data refers to the relevant data for measuring the performance of multiple suppliers during collaborative cooperation in the same cooperation order, and is used for subsequent calculation of the performance data of multiple suppliers during collaborative cooperation; the performance data includes the scheduled delivery time, delivery time, delivery cycle, number of qualified products, number of defective products, quantity of product purchases, product quotation, problem handling duration, etc.; the historical order data is obtained through the enterprise's internal enterprise resource planning system, quality management system, historical procurement contracts, and supplier management system.

[0038] The method for constructing a supplier relationship network topology includes: Take the multiple suppliers corresponding to each historical cooperation order as a set of supplier sets, and the supplier sets correspond one-to-one with the historical cooperation orders; take the historical cooperation orders with the same supplier set as a set of order sets, and the order sets correspond one-to-one with the supplier sets; input the performance data corresponding to each set of order sets into the trained performance evaluation model respectively to predict the corresponding performance data, and the performance evaluation model is a deep neural network model; evaluate the ability indicators corresponding to each set of order sets based on the performance data corresponding to different order sets; take each supplier as a node, and establish an edge between the nodes corresponding to every two suppliers in each set of supplier sets, and the weight of each edge is the ability indicator corresponding to each order set; construct a supplier relationship network topology according to the nodes, edges and the weights of the edges, and use the ability portrait of each supplier as the attribute of the corresponding node.

[0039] Exemplarily, supplier A and supplier B are supplier set 1, and supplier A, supplier B and supplier C are supplier set 2; therefore, in the supplier relationship network topology, supplier A, supplier B and supplier C are all one node; since both supplier set 1 and supplier set 2 contain supplier A and supplier B, there are two edges between supplier A and supplier B, one edge between supplier B and supplier C, and one edge between supplier A and supplier C.

[0040] The steps for identifying the synergy effect between suppliers include: Step S201: Mark the ability indicators in the attributes corresponding to each node as the first indicator, and mark the ability indicators corresponding to the weights of the edges as the second indicator; Step S202: Randomly select a set of supplier sets that have not been marked as the selected set and mark it as the current set; Step S203: Compare each first indicator corresponding to each node in the current set with the second indicator corresponding to the corresponding order set in turn; if all the first indicators are less than or equal to the corresponding second indicators, then all the nodes in the corresponding current set have a positive synergy effect, and mark the corresponding current set as the synergy set; if there is a first indicator greater than the corresponding second indicator, then there are nodes in the corresponding current set that have a negative synergy effect, and do not mark the corresponding current set; Step S204: Mark the current set as the selected set; Step S205: Loop through steps S202 to S204 until all supplier sets are marked as the selected set, the loop ends, and all synergy sets are obtained.

[0041] The method for recommending the optimal supplier combination includes: According to historical cooperation orders, obtain the historical procurement categories corresponding to each collaborative set and mark them as cooperative procurement categories; according to historical procurement data, obtain the historical procurement categories corresponding to each supplier and mark them as independent procurement categories; compare each demand procurement category with each cooperative procurement category respectively; if there is a demand procurement category that is the same as a cooperative procurement category, mark the corresponding collaborative set as a recommended set; if all demand procurement categories are different from cooperative procurement categories, do not mark the corresponding collaborative set; mark the demand procurement categories that are different from all cooperative procurement categories as remaining procurement categories, and mark the suppliers corresponding to the independent procurement categories that are the same as the remaining procurement categories as candidate suppliers; group the candidate suppliers with the same corresponding independent procurement categories as a group of candidate sets, and the candidate sets correspond one-to-one with the independent procurement categories; According to future procurement requirements, obtain the estimated procurement quantity of each remaining procurement category; according to historical procurement data, obtain the historical procurement quantity corresponding to each candidate supplier, and mark the largest historical procurement quantity corresponding to each candidate supplier as the maximum quantity; compare each maximum quantity with the corresponding estimated procurement quantity respectively, delete the candidate suppliers whose maximum quantity is less than the corresponding estimated procurement quantity from the corresponding candidate set, and the candidate suppliers whose maximum quantity is greater than or equal to the estimated procurement quantity remain in the corresponding candidate set; sort each candidate supplier in each candidate set from largest to smallest according to the corresponding performance score to generate a supplier ranking list; mark the top candidate suppliers in each supplier ranking list as the best suppliers, where is an integer greater than 0; combine each best supplier with each recommended set to obtain the optimal supplier combination.

[0042] The strategy formulation module is used to obtain procurement management data, and in combination with the supplier performance data corresponding to the optimal supplier combination, use multi-objective optimization technology to intelligently formulate personalized procurement strategies.

[0043] The procurement management data includes category inventory, procurement budget, and supply cycle; the category inventory includes the inventory quantity of each demand procurement category; the procurement budget includes the budget amount of each demand procurement category; the supply cycle includes the delivery cycle of each demand procurement category; the procurement management data is obtained through the enterprise resource planning system within the enterprise.

[0044] The method for formulating personalized procurement strategies includes: Subtract the corresponding inventory quantity in the category inventory from the estimated procurement quantity of each demand procurement category to obtain the actual procurement quantity of each demand procurement category, and set the quantity range of each demand procurement category , is the actual purchase quantity; randomly select values from the quantity range corresponding to each cooperative procurement category, and construct a set of cooperation sets, with a total of sets of cooperation sets constructed; randomly select values from the quantity range corresponding to each independent procurement category, and construct a set of independent sets, with a total of sets of independent sets constructed; among them, is the number of suppliers in the collaborative set, and the sum of the values in each set of cooperation sets is , and are both integers greater than 1; randomly select a set of cooperation sets and a set of independent sets to construct a set of procurement sets, with a total of sets of different procurement sets, ; for sets of procurement sets, set sequentially increasing digital labels and mark them as procurement labels, and the range of procurement labels is ; randomly select a procurement label as the initial iteration center; Define the iteration process. The iteration process is as follows: According to the iteration center, generate candidate solutions within the range of procurement labels, and calculate the procurement performance evaluation value corresponding to each candidate solution, , and the candidate solutions correspond one-to-one with the procurement labels; mark the candidate solution with the largest procurement performance evaluation value as the local solution, move the iteration center to the local solution, and calculate the index change amount; Execute the iteration process. When each iteration process is completed, compare the index change amount with the preset change threshold; if the index change amount is less than the change threshold, generate a completion instruction; if the index change amount is greater than or equal to the change threshold, do not generate a completion instruction; count the number of times the completion instruction is generated and mark it as the instruction count; when the instruction count is greater than or equal to the preset number threshold, stop executing the iteration process; mark the procurement label corresponding to the iteration center as the best label, and use the procurement set corresponding to the best label as the procurement strategy; both the change threshold and the number threshold are preset by those skilled in the art according to the actual situation.

[0045] The method for calculating the procurement performance evaluation value corresponding to the candidate solution includes: Obtain the procurement set corresponding to the procurement label of the candidate solution and mark it as the calculation set; obtain the supplier performance data corresponding to each supplier in the calculation set and mark it as the calculation data; use the calculation data, the calculation set, and the procurement management data as the performance prediction data; input the performance prediction data into the trained performance prediction model respectively to predict the corresponding procurement performance data; among them, the performance prediction model includes a cost prediction model, a delivery prediction model, and a quality prediction model, and the performance prediction models are all deep neural network models; the procurement performance data includes procurement cost, delivery cycle, and product quality; the cost prediction model is used to predict the procurement cost, the delivery prediction model is used to predict the delivery cycle, and the quality prediction model is used to predict the product quality; Subtract the procurement cost from the procurement budget to obtain the budget surplus; subtract the delivery cycle from the delivery period to obtain the delivery deviation; preset an index weight set, and the index weight set includes the index weights corresponding to the budget surplus, the delivery deviation, and the product quality, and the index weight set is pre-set by those skilled in the art according to the actual situation; multiply the budget surplus, the delivery deviation, and the product quality by the corresponding index weights in the index weight set respectively to obtain the budget coefficient, the delivery coefficient, and the quality coefficient; add the budget coefficient, the quality coefficient, and the delivery coefficient to obtain the procurement performance evaluation value; It should be understood that the delivery period is the time when the enterprise expects the supplier to complete the delivery, and the delivery cycle is the time when the supplier expects to complete the delivery.

[0046] The method for calculating the index change amount is: subtract the procurement performance evaluation value corresponding to the iteration center in the current iteration process from the procurement performance evaluation value corresponding to the iteration center in the previous iteration process, and take the absolute value to obtain the index change amount.

[0047] It should be noted that the purpose of formulating a personalized procurement strategy is to make flexible adjustments according to the actual situation such as the enterprise's budget, inventory, and delivery cycle, so as to optimize procurement decisions, reduce procurement costs, improve delivery efficiency, and ensure product quality; at the same time, through dynamic iterative optimization, improve procurement efficiency, supply chain stability and flexibility, and ultimately achieve the goals of cost control, resource optimization, and supply chain risk management.

[0048] In this embodiment, through deep learning technology, multi-dimensional factors such as historical procurement data and market trend data are integrated to achieve accurate perception and prediction of future procurement requirements; an ability profile is constructed based on supplier performance data, and a comprehensive evaluation of suppliers is carried out by combining fuzzy evaluation and quantitative analysis methods, which can more objectively quantify the performance of suppliers in key capabilities such as delivery, quality, and price; by analyzing the supplier relationship network, the synergy among suppliers is identified, providing an important basis for recommending the optimal supplier combination, which is conducive to giving play to the synergy advantages among suppliers and improving the overall efficiency of the supply chain; in the procurement strategy formulation stage, multi-dimensional constraints such as supplier performance, procurement requirements, and inventory are integrated, and an iterative optimization method is used to dynamically generate personalized procurement plans, achieving refined balance and dynamic adjustment among multiple objectives such as cost, delivery, and quality, and maximizing the scientificity and effectiveness of procurement decisions; it not only improves procurement efficiency, reduces costs, but also enhances the flexibility and resilience of the supply chain, thereby effectively improving the overall response speed, cost control ability, and operation quality of the supply chain, and enhancing the market competitiveness of the enterprise.

[0049] Embodiment 2

[0050] Please refer to Figure 2 as shown. For the parts not described in detail in this embodiment, refer to the description content of Embodiment 1. A smart procurement method for a supply chain is provided, and the method includes: Collect supply chain management data, where the supply chain management data includes demand forecast data and supplier performance data; Deeply mine the demand forecast data, and use deep learning technology to dynamically predict future procurement requirements; Based on the supplier performance data, generate ability profiles of different suppliers, and quantitatively evaluate the performance scores of different suppliers according to the pre-constructed evaluation system; Obtain historical order data, integrate the historical order data with the ability profiles of different suppliers, construct a supplier relationship network topology, and identify the synergy among suppliers; integrate the performance scores, future procurement requirements, and synergy, and apply an intelligent matching algorithm to recommend the optimal supplier combination; Obtain procurement management data, and combine the supplier performance data corresponding to the optimal supplier combination, and use multi-objective optimization technology to intelligently formulate personalized procurement strategies.

[0051] Embodiment 3

[0052] This application also provides an electronic device. The electronic device may include one or more processors and one or more memories. Among them, computer-readable code is stored in the memory, and when the computer-readable code is run by one or more processors, it can execute a smart procurement method for a supply chain as described above.

[0053] The method or system according to an embodiment of the present application can also be implemented by means of the architecture of the electronic device shown in the present application. The electronic device may include a bus, one or more CPUs, a ROM, a RAM, a communication port connected to a network, an input / output, a hard disk, etc. The storage device in the electronic device, such as the ROM or the hard disk, can store an intelligent procurement method for a supply chain provided by the present application. Further, the electronic device may further include a user interface. Of course, the architecture shown in the present application is only exemplary. When implementing different devices, one or more components shown in the electronic device of the present application can be omitted according to actual needs.

[0054] Embodiment 4

[0055] An embodiment of the present application discloses a computer-readable storage medium. Computer-readable instructions are stored on the computer-readable storage medium. When the computer-readable instructions are run by a processor, an intelligent procurement method for a supply chain according to an embodiment of the present application as described with reference to the above drawings can be executed. The storage medium includes, but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory, etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0056] In addition, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the present application provides a non-transitory machine-readable storage medium storing machine-readable instructions that can be run by a processor to execute instructions corresponding to the method steps provided by the present application, such as: an intelligent procurement method for a supply chain. When the computer program is executed by a central processing unit (CPU), the above functions defined in the method of the present application are executed.

[0057] The foregoing is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacement of some of the technical features. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

[0058] It should be noted that in this text, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, such that a process, method, article or device comprising a series of elements not only includes those elements but also other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising one..." does not exclude the presence of additional identical elements in the process, method, article or device comprising said element.

[0059] In the description of the present invention, it should be understood that the terms "first", "second", etc. are only used for descriptive distinction and cannot be construed as indicating or implying relative importance.

[0060] In the description of the present invention, unless otherwise specified, the meaning of "a plurality" is two or more.

[0061] In the description of the present invention, the meaning of "several" is one or more, and the meaning of "a large number" is two or more.

[0062] In the description of this specification, the descriptions referring to terms such as "one embodiment", "some embodiments", "examples", "specific examples", or "some examples", etc. mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0063] For the formulas in this specification, the dimensional quantities are removed and only the numerical values are calculated. The formulas are obtained by collecting a large amount of data and performing software simulation to get a formula that is closest to the actual situation. The preset parameters and threshold values in the formulas are set by those skilled in the art according to the actual situation.

[0064] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the claims and their equivalents.

Claims

1. An intelligent procurement method for a supply chain, characterized in that, Including: Collecting supply chain management data, which includes demand forecasting data and supplier performance data; Deeply mining the demand forecasting data and dynamically predicting future procurement demands by using deep learning techniques; Generating ability portraits of different suppliers based on the supplier performance data, and quantitatively evaluating the performance scores of different suppliers according to a pre-constructed evaluation system; Obtaining historical order data, integrating the historical order data with the ability portraits of different suppliers, constructing a supplier relationship network topology, and identifying the synergy effects among suppliers; fusing the performance scores, future procurement demands and synergy effects, and applying an intelligent matching algorithm to recommend an optimal supplier combination; Obtaining procurement management data, and intelligently formulating personalized procurement strategies by using multi-objective optimization techniques in combination with the supplier performance data corresponding to the optimal supplier combination.

2. The intelligent procurement method of the supply chain according to claim 1, wherein The demand forecasting data includes historical procurement data, market trend data and demand procurement categories; the historical procurement data includes historical procurement time, historical procurement categories and historical procurement quantities; the market trend data includes raw material price fluctuation data and market demand change data, the raw material price fluctuation data includes the market prices of raw materials required for different supply chain products at different time points, and the market demand change data includes the total demand quantities of different supply chain products at different time points; the supplier performance data includes the performance data corresponding to each supplier; The method for dynamically predicting future procurement demands includes: Taking the historical procurement quantities with the same corresponding historical procurement categories as a set of quantity sets; presetting a time interval, and constructing a time axis for each set of quantity sets based on the time interval; wherein, on each time axis, the time difference between every two adjacent time points is equal to the time interval, and each time point corresponds to a historical procurement quantity in the corresponding quantity set; Constructing a quantity prediction model for each set of quantity sets according to the historical procurement quantity corresponding to each time point on each time axis, the quantity prediction model being a recurrent neural network model; presetting the length as The time window is an integer greater than 1; Mark the quantity set corresponding to each demand procurement category as a demand set, and mark the time axis corresponding to the demand set as a demand axis; Starting from the last time point on each demand axis, obtain a set of time sets from the demand axis corresponding to each group of demand sets based on the time window, and each set of time sets includes time points; Input the historical procurement quantities corresponding to each set of time sets into the corresponding quantity prediction model to predict the corresponding future procurement quantities; Calculating a first factor and a second factor corresponding to each set of demand sets in sequence according to the market trend data; respectively setting the weight coefficients of each first factor and each second factor, multiplying each first factor and each second factor by the corresponding weight coefficients in sequence to obtain influence weights, taking the mean value of the influence weights with the same corresponding demand sets as the comprehensive influence factor; multiplying each future procurement quantity by the corresponding comprehensive influence factor to obtain the predicted procurement quantity of each set of demand sets, and taking it as the future procurement demand.

3. The intelligent procurement method of the supply chain according to claim 2, wherein, The method for constructing a time axis for a quantity set includes: Mark the earliest historical purchase time corresponding to the historical purchase quantity in the quantity set as the earliest time, and use the earliest time as the first time point on the time axis; starting from the earliest time, generate consecutive time points on the time axis according to the time interval; compare the time points on the time axis with the historical purchase time corresponding to each historical purchase quantity in the quantity set respectively; map the historical purchase quantity corresponding to the historical purchase time that is the same as the time point to the corresponding time point in turn; mark the time points that are not the same as all historical purchase times as interpolation points, and use the polynomial interpolation method to calculate the historical purchase quantity corresponding to each interpolation point and map it to the corresponding interpolation point. The method for calculating the first factor and the second factor corresponding to each group of demand sets includes: Map the market trend data to each time point on each demand axis respectively, and each time point corresponds to a set of market prices and a total demand quantity; construct a market impact model for each set of demand collections according to the market price, total demand quantity and historical purchase quantity corresponding to each time point on each demand axis; the market impact model is a deep neural network model; mark the earliest time point in each set of time collections as a tentative early point, starting from the time point before the corresponding tentative early point on each demand axis, and obtain a set of pre-time collections from each demand axis based on the time window, and each set of pre-time collections includes time points; input the market price, total demand quantity and historical purchase quantity corresponding to each set of time collections into the corresponding market impact model respectively to predict the corresponding first factor; input the market price, total demand quantity and historical purchase quantity corresponding to each set of pre-time collections into the corresponding market impact model respectively to predict the corresponding second factor.

4. The intelligent procurement method of the supply chain according to claim 3, characterized in that The method for generating the ability portraits of different suppliers includes: Based on the performance data corresponding to different suppliers, evaluate the ability indicators corresponding to each supplier. The ability indicators include delivery ability, quality ability, price ability, and service ability; according to the ability indicators corresponding to each supplier, construct a corresponding bar chart for each supplier, and use the bar chart as the ability portrait of the corresponding supplier.

5. The intelligent procurement method of the supply chain according to claim 4, characterized in that, The steps for evaluating the delivery ability corresponding to each supplier include: Step S101: Construct multiple fuzzy sets for each data in the delivery performance data respectively; Step S102: Convert the delivery performance data of each supplier into the membership degree of the corresponding each fuzzy set respectively through the fuzzy technology; Step S103: Define fuzzy rules; Step S104: Match each group of fuzzified technical feature data with the fuzzy rules respectively, and use the fuzzy inference method to perform fuzzy inference to obtain the fuzzy inference result corresponding to each supplier. The fuzzy inference result is the membership degree of each delivery ability level; Step S105: Set the ability range, where the range of the ability range is , is an integer greater than 1; evenly divide the ability range into four grade ranges, which correspond one by one to the grades in the delivery ability level; add the maximum value of each grade range to the corresponding minimum value and then divide by 2 to obtain the interval mean of each grade range; Step S106: Multiply the membership degree of each delivery ability level corresponding to each supplier by the corresponding interval mean value respectively, and then add them up in turn to obtain the total ability value of each supplier; add up the membership degrees of each delivery ability level corresponding to each supplier in turn to obtain the total membership degree; divide the total ability value of each supplier by the corresponding total membership degree as the delivery ability of each supplier. The steps for evaluating the quality ability, price ability, and service ability corresponding to each supplier are the same as the steps for evaluating the delivery ability corresponding to each supplier; The method for quantitatively evaluating the performance scores of different suppliers includes: Preset a ratio set, and the ratio set includes the ratio coefficients corresponding to each ability indicator; multiply each ability indicator of each supplier by the corresponding ratio coefficient in the ratio set respectively, and then add them up in turn to obtain the performance score corresponding to each supplier.

6. The intelligent procurement method of the supply chain according to claim 5, characterized in that The historical order data includes the performance data corresponding to each historical cooperation order; The method for constructing the supplier relationship network topology includes: Regarding multiple suppliers corresponding to each historical cooperation order as a set of supplier collections, and regarding historical cooperation orders with the same supplier collection as a set of order collections; inputting the performance data corresponding to each set of order collections into the trained performance evaluation model respectively to predict the corresponding performance data, where the performance evaluation model is a deep neural network model; evaluating the ability indicators corresponding to each set of order collections based on the performance data corresponding to different order collections; taking each supplier as a node, establishing an edge between the nodes corresponding to every two suppliers in each set of supplier collections, and the weight of each edge is the ability indicator corresponding to each order collection; constructing a supplier relationship network topology according to the nodes, edges, and the weights of the edges, and taking the ability portrait of each supplier as the attribute of the corresponding node. The steps for identifying the synergy effect among suppliers include: Step S201: Mark the ability indicators in the attributes corresponding to each node as the first indicator, and mark the ability indicators corresponding to the weights of the edges as the second indicator. Step S202: Randomly select a set of supplier collections that have not been marked as the selected set, and mark it as the current set. Step S203: Compare each first indicator corresponding to each node in the current set with the second indicator corresponding to the corresponding order collection in turn; if all first indicators are less than or equal to the corresponding second indicator, then all nodes in the corresponding current set have a positive synergy effect, and mark the corresponding current set as the synergy set; if there is a first indicator greater than the corresponding second indicator, then there are nodes in the corresponding current set with a negative synergy effect, and do not mark the corresponding current set. Step S204: Mark the current set as the selected set. Step S205: Loop through Step S202 to Step S204 until all supplier collections are marked as the selected set. When the loop ends, obtain all synergy sets.

7. The intelligent procurement method of the supply chain according to claim 6, characterized in that, The method for recommending the optimal supplier combination includes: According to the historical cooperation orders, obtain the historical procurement categories corresponding to each set of synergy sets, and mark them as the cooperative procurement categories; according to the historical procurement data, obtain the historical procurement categories corresponding to each supplier, and mark them as the independent procurement categories; compare each demand procurement category with each cooperative procurement category; if there is a demand procurement category that is the same as the cooperative procurement category, then mark the corresponding synergy set as the recommended set; if all demand procurement categories are different from the cooperative procurement categories, then do not mark the corresponding synergy set; mark the demand procurement categories that are different from all cooperative procurement categories as the remaining procurement categories, and mark the suppliers corresponding to the independent procurement categories that are the same as the remaining procurement categories as the candidate suppliers; take the candidate suppliers corresponding to the same independent procurement category as a set of candidate collections, and the candidate collections correspond one-to-one with the independent procurement categories. Obtain the estimated procurement quantity for each remaining procurement category according to future procurement requirements; obtain the historical procurement quantity corresponding to each candidate supplier according to historical procurement data, and mark the largest historical procurement quantity corresponding to each candidate supplier as the maximum quantity; compare each maximum quantity with the corresponding estimated procurement quantity respectively, and delete the candidate suppliers whose maximum quantity is less than the corresponding estimated procurement quantity from the corresponding candidate set, while the candidate suppliers whose maximum quantity is greater than or equal to the estimated procurement quantity remain in the corresponding candidate set; sort each candidate supplier in each candidate set in descending order according to the corresponding performance score to generate a supplier ranking list; for each supplier ranking list, the top All candidate suppliers are marked as the best suppliers, where is an integer greater than 0; Combine each best supplier with each set of recommended sets to obtain the optimal supplier combination.

8. The intelligent procurement method of the supply chain according to claim 7, wherein The procurement management data includes category inventory, procurement budget, and supply cycle; the category inventory includes the inventory quantity of each required procurement category; the procurement budget includes the budget amount of each required procurement category; the supply cycle includes the delivery cycle of each required procurement category; The method for formulating a personalized procurement strategy includes: Build a set of different procurement sets, for the set of procurement sets, set sequentially increasing numerical labels and mark them as procurement labels, and the range of the procurement labels is ; randomly select a procurement label as the initial iteration center; Define the iterative process, which is as follows: According to the iterative center, generate candidate solutions within the range of the procurement label, and calculate the procurement performance evaluation value corresponding to each candidate solution. , where the candidate solutions correspond one-to-one with the procurement label; Mark the candidate solution with the largest procurement performance evaluation value as the local solution, move the iterative center to the local solution, and calculate the index change amount. Execute an iterative process. When each iterative process is completed, compare the index change amount with a preset change threshold; if the index change amount is less than the change threshold, generate a completion instruction; if the index change amount is greater than or equal to the change threshold, do not generate a completion instruction; count the number of generated completion instructions and mark it as the instruction count; when the instruction count is greater than or equal to a preset count threshold, stop executing the iterative process; mark the procurement label corresponding to the iteration center as the best label, and use the procurement set corresponding to the best label as the procurement strategy.

9. The intelligent procurement method of the supply chain according to claim 8, wherein, The method for calculating the procurement performance evaluation value corresponding to a candidate solution includes: Obtain the procurement set corresponding to the procurement label of the candidate solution and mark it as the calculation set; obtain the supplier performance data corresponding to each supplier in the calculation set and mark it as the calculation data; use the calculation data, the calculation set, and the procurement management data as performance prediction data; input the performance prediction data into the trained performance prediction model respectively to predict the corresponding procurement performance data; among them, the performance prediction model includes a cost prediction model, a supply prediction model, and a quality prediction model, and the performance prediction models are all deep neural network models; the procurement performance data includes procurement cost, supply cycle, and product quality; Subtract the procurement cost from the procurement budget to obtain the budget surplus; subtract the supply cycle from the delivery cycle to obtain the delivery deviation; preset an index weight set, and the index weight set includes the index weights corresponding to the budget surplus, the delivery deviation, and the product quality; multiply the budget surplus, the delivery deviation, and the product quality by the corresponding index weights in the index weight set respectively to obtain the budget coefficient, the delivery coefficient, and the quality coefficient; add the budget coefficient, the quality coefficient, and the delivery coefficient to obtain the procurement performance evaluation value; The method for calculating the index change amount is: subtract the procurement performance evaluation value corresponding to the iteration center in the previous iterative process from the procurement performance evaluation value corresponding to the iteration center in the current iterative process, and take the absolute value to obtain the index change amount.

10. An intelligent procurement system for a supply chain, which implements the intelligent procurement method for the supply chain according to any one of claims 1-9, characterized in that including: A data collection module for collecting supply chain management data, where the supply chain management data includes demand forecasting data and supplier performance data; A demand forecasting module for deeply mining the demand forecasting data and dynamically predicting future procurement demands using deep learning techniques; A supplier evaluation module for generating ability portraits of different suppliers based on the supplier performance data and quantitatively evaluating the performance scores of different suppliers according to a pre-built evaluation system; A supplier recommendation module for obtaining historical order data, integrating the historical order data with the ability portraits of different suppliers, constructing a supplier relationship network topology, and identifying the synergy effects among suppliers; fusing the performance scores, future procurement demands, and synergy effects, and applying an intelligent matching algorithm to recommend the optimal supplier combination; A strategy formulation module for obtaining procurement management data and intelligently formulating personalized procurement strategies using multi-objective optimization techniques in combination with the supplier performance data corresponding to the optimal supplier combination.

Images