Intelligent purchasing system of supply chain

Abstract

The invention relates to the technical field of supply chain management, and discloses an intelligent purchase system of a supply chain, which comprises a control center, a data classification storage module, a new purchase project evaluation module, a supplier evaluation recommendation module and a post-evaluation recording module. The data classification storage module is used for storing past historical purchase data and information data of a plurality of suppliers; the new purchase project evaluation module is used for calculating a new purchase project bid evaluation / comparison score value; the supplier evaluation recommendation module is used for screening a plurality of suppliers by adopting an improved k-means clustering analysis algorithm; and the post-evaluation recording module is used for jointly evaluating the product / service score of the bid-winning supplier by the purchase implementation department and the purchase demand department, and the product / service score is used as a post-evaluation score. According to the method, the improved k-means algorithm is adopted, comprehensive evaluation recommendation is carried out on numerous suppliers, meanwhile, post-evaluation scores after purchasing are evaluated and recorded, multi-aspect and full-process evaluation on the suppliers is achieved, and the purchasing efficiency is improved.

Description

Technical Field

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

[0002] With the advancement of economic globalization and the increasingly fierce market competition, the game between enterprises is not only a competition of the company's own strength, but also related to the suppliers who directly cooperate with the company. In the current supply chain procurement industry, the mainstream procurement methods can be divided into: negotiation (single source negotiation, competitive negotiation), bidding (open bidding, invited bidding), comparison (open comparison, invited comparison), bidding, etc. Among them, the most commonly used procurement method is the comparison, including offline comparison and online comparison. For online comparison, due to the rapid development of the Internet, more and more purchasing companies rely on convenient network resources and use some e-commerce platforms for procurement. However, the suppliers of online trading platforms are mixed, which easily leads to asymmetric transaction information. It is difficult to ensure that buyers can choose the right supplier every time, which requires the platform to manage and evaluate the suppliers.

[0003] At present, different evaluation methods are used for supplier evaluation according to different situations, such as the hesitant multi-attribute decision-making model for supplier selection based on the HTFWGBM operator. In order to reduce the evaluation of supplier performance by expert experience, an intelligent supplier evaluation model based on data-driven support vector regression (SVR) is proposed. For the optimal selection of suppliers in a fuzzy environment, a group decision-making method based on intuitive fuzzy set hyperlink induced topic search method and intuitive fuzzy cross entropy and grey correlation degree is proposed for the supplier selection problem with subjective and objective mixed evaluation information. However, these evaluation methods basically focus on the optimization of the methods. However, there are massive and dynamic supplier data in the trading platform. How to quickly and conveniently screen, classify and evaluate is an important problem that needs to be solved in supplier selection.

[0004] In addition, in the current procurement process such as bidding and comparison, the comprehensive score of the supplier's product price, technical strength, and overall corporate strength during the bid evaluation stage is used as an objective evaluation value to determine whether the supplier has won the bid. However, this scoring method depends on different categories of services and products, and the weight ratio of each part such as price, technical strength, and overall corporate strength in the total score is only vaguely determined based on past project experience estimates. This bid evaluation process is generally not related to the actual post-evaluation score obtained after the product / service is delivered. That is, the evaluation of suppliers during the procurement implementation (bid evaluation) process only stops at the release of procurement results, and no evaluation is made on the behavior after the procurement is completed. Summary of the invention

[0005] The present invention provides an intelligent procurement system for a supply chain, which adopts an improved K-means algorithm to comprehensively evaluate and recommend a large number of suppliers, and at the same time evaluates and records the post-evaluation scores after procurement, thereby realizing multi-faceted and full-process evaluation of suppliers and improving procurement efficiency.

[0006] The present invention provides an intelligent procurement system for a supply chain, comprising a control center, a data classification storage module, a new procurement project evaluation module, a supplier evaluation recommendation module, and a post-evaluation recording module, wherein the control center is connected to the data classification storage module, the new procurement project evaluation module, the supplier evaluation recommendation module, and the post-evaluation recording module respectively;

[0007] The data classification storage module is used to store a large amount of past historical purchase data according to different categories of products, as well as store information data of multiple suppliers and historical price data of suppliers;

[0008] The new procurement project evaluation module is used to extract the characteristic parameters of the training old procurement project supplier samples, and then calculate the reasonable weighted values ​​of each parameter to accurately and effectively evaluate the subsequent new procurement project bid evaluation / comparison score;

[0009] The supplier evaluation and recommendation module is used to screen multiple suppliers using an improved k-means clustering analysis algorithm to select the best supplier from the multiple suppliers for recommendation;

[0010] The post-evaluation record module is used to deliver the purchased products / services to the user department after the procurement implementation stage is over. After the procurement result shares are allocated and the products / services are provided, the procurement implementation department and the procurement demand department jointly evaluate the product / service scores of the winning suppliers and use them as post-evaluation scores.

[0011] Furthermore, the new procurement project evaluation module includes an evaluation sample unit, a parameter extraction unit and a pre-training unit. The evaluation sample unit is used to obtain and store the post-evaluation score as the expected output value of the pre-training unit; the parameter extraction unit is used to extract the characteristic parameters used for bid evaluation / comparison in the new / old procurement projects as the input node value of the pre-training unit; the pre-training unit is used to use the characteristic parameters to evaluate the new procurement projects and calculate the scores.

[0012] Furthermore, the supplier evaluation and recommendation module includes a data acquisition unit, a data preprocessing unit and a supplier recommendation unit. The data acquisition unit is used to collect the information data of the supplier, the data processing unit is used to process the information data of the supplier to eliminate data differentiation, and the supplier recommendation unit is used to classify the quotations of suppliers using an improved k-means clustering analysis algorithm, divide the suppliers into several different categories according to evaluation indicators, calculate the weight value of the indicator through the hierarchical analysis method, and find the optimal supplier according to the weighted score of the indicator.

[0013] Furthermore, the data preprocessing unit processes the supplier's information data using mean-standard deviation standardization, and the calculation formula is:

[0014] x * =(x-μ) / σ

[0015] Among them, ∈ is the mean and σ is the standard deviation.

[0016] Further, the supplier recommendation unit includes a cluster analysis subunit and a hierarchical analysis evaluation subunit;

[0017] The cluster analysis subunit uses a cluster analysis algorithm to analyze suppliers and divide the suppliers into different clusters, so that the similarity of the indicator data of suppliers in the same cluster is high, and the difference of the indicator data of suppliers between clusters is high. The indicator is derived from the established supplier evaluation system. Several attributes are selected from the evaluation system for cluster analysis, and suppliers are divided into different categories according to the attributes;

[0018] After cluster analysis of the selected suppliers, the hierarchical analysis evaluation subunit calculates the weight of the first-level index relative to the target layer through the hierarchical analysis method, then calculates the relative weight of the second-level index relative to the first-level index, and then calculates the absolute weight of the second-level index, and calculates the horizontal index; calculates the score of each category according to the weight and index value, and selects the best supplier group. Finally, the score of each supplier in the best supplier group is calculated one by one, and the best supplier is selected for recommendation.

[0019] Furthermore, in the improved k-means clustering analysis algorithm,

[0020] The k-means clustering analysis algorithm finds k cluster centers for a data set containing n objects, and divides the data in the data set into k clusters according to the proximity principle, so that the k clusters meet two requirements: the number of data in each cluster cannot be less than 1, and each data object belongs to and only belongs to one cluster, and then adjusts the data attributes according to the decision function;

[0021] The algorithm first randomly selects k objects from n data objects to be classified. The k objects represent the average value in the cluster during the first classification. The Euclidean distance between the remaining nk objects and the k objects is calculated. The data object with the smallest distance and the average object are divided into the same cluster. In this way, n data are divided into k different clusters. Then, the average value in each cluster is recalculated, and the above classification steps are iterated until the result of the last classification is the same as the result of the second to last classification. The algorithm is the convergence of the criterion function, and the convergence function is:

[0022]

[0023] Where E represents the sum of square differences of all data objects, x represents the data object, and x i Represents cluster C i The smaller the convergence function is, the more similar the data in the cluster is, the higher the independence of the clusters is, and the lower the similarity correlation between clusters is;

[0024] Improvements include:

[0025] Create a two-dimensional array sum[k][m] to store the data objects in each cluster and a one-dimensional array count[k] to store the number of data objects in each cluster. In this way, the new average value of each cluster can be calculated directly through the formula, 0<i≤k without traversing the database;

[0026] Modify the condition at the end of the k-means algorithm to indicate that the mean of each cluster no longer changes or changes significantly, so that the mean of the cluster can be obtained, and use a two-dimensional array to calculate sum[k][m] and a one-dimensional array count[k] instead of traversing the database to calculate the criterion function.

[0027] Furthermore, it also includes a supplier quotation management module, which is connected to the control center and is used to store quotations of all existing suppliers for various products, including historical quotations and latest quotations.

[0028] Furthermore, it also includes a user interaction interface, which is connected to the control center and is used to display the recommended best supplier, quotations from all suppliers, and perform human-computer interaction.

[0029] The beneficial effects of the present invention are:

[0030] 1. The present invention adopts an improved k-means algorithm to improve the deficiency of frequent traversal of the database in the traditional algorithm. By establishing an index system, the cluster center is easy to set, is not affected by noise data, has practical significance, has a faster calculation speed and good clustering effect, and also reduces the time complexity of the algorithm. The actual data of the network is collected quantitatively to improve the solution efficiency.

[0031] 2. Use the original post-evaluation data analysis of procurement projects to correct the existing bidding / comparison project review method. The system effectively corrects the problem of disconnection between the procurement process and the product use process, so that the optimized procurement weight scheme can better fit the evaluation and perception of suppliers by the current network demand units. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a schematic diagram of the structure of the intelligent procurement system of the supply chain of the present invention.

[0033] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0034] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0035] like Figure 1 As shown, the present invention provides an intelligent procurement system for a supply chain, including a control center, a data classification storage module, a new procurement project evaluation module, a supplier evaluation recommendation module, and a post-evaluation recording module, wherein the control center is respectively connected to the data classification storage module, the new procurement project evaluation module, the supplier evaluation recommendation module, and the post-evaluation recording module;

[0036] The data classification storage module is used to store a large amount of past historical procurement data according to different categories of products, as well as information data of multiple suppliers and historical price data of suppliers; at the same time, it stores basic information of suppliers and related information of supplier transactions.

[0037] The new procurement project evaluation module is used to extract the characteristic parameters of the training old procurement project supplier samples, and then calculate the reasonable weighted values ​​of each parameter to accurately and effectively evaluate the subsequent new procurement project bid evaluation / comparison score;

[0038] The supplier evaluation and recommendation module is used to screen multiple suppliers using an improved k-means clustering analysis algorithm to select the best supplier from the multiple suppliers for recommendation;

[0039] The post-evaluation record module is used to deliver the purchased products / services to the user department after the procurement implementation stage is over. After the procurement result shares are allocated and the products / services are provided, the procurement implementation department and the procurement demand department jointly evaluate the product / service scores of the winning suppliers and use them as post-evaluation scores.

[0040] In one embodiment, the new procurement project evaluation module includes an evaluation sample unit, a parameter extraction unit and a pre-training unit. The evaluation sample unit is used to obtain and store a post-evaluation score as an expected output value of the pre-training unit; the parameter extraction unit is used to extract characteristic parameters used for bid evaluation / comparison in new / old procurement projects as input node values ​​of the pre-training unit; and the pre-training unit is used to use the characteristic parameters to evaluate and calculate scores for new procurement projects.

[0041] The workflow of the new procurement project evaluation module is as follows:

[0042] 1) Extract characteristic parameters used for bid evaluation / comparison in old procurement projects (including but not limited to: price, technical indicators, comprehensive strength of enterprises, etc.) from the data storage module and store them for backup.

[0043] 2) The extracted objective feature parameters are input into the pre-training unit, and the evaluation sample unit selects project materials of a specific category from the data storage module and extracts the post-purchase evaluation result scores for future use.

[0044] 3) The evaluation sample unit transmits the extracted post-procurement evaluation result score to the pre-training unit for standby. The pre-training unit calculates the relationship between the post-procurement evaluation result and its corresponding evaluation / comparison feature parameters based on the old procurement project evaluation / comparison feature parameters obtained from step 2) and the post-procurement evaluation result score of the corresponding project through the pre-training unit (actually not limited to this method).

[0045] In the pre-training unit, the evaluation / comparison feature parameters of these old projects will be used as the input of the neural network. The post-evaluation scores of the corresponding procurement projects will be used as the corresponding expected outputs. Then the learning algorithm will be used for training.

[0046] 4) Extract the bid evaluation / comparison feature parameters from the data storage module and store them for future use. Input the extracted bid evaluation / comparison feature parameters; after the pre-training unit is trained, perform the same feature extraction process on the new procurement project to obtain new bid evaluation / comparison feature parameters, and then input these feature parameters.

[0047] 5) According to the characteristic parameter weight calculation formula obtained by the pre-training unit, the evaluation / comparison score of the new procurement project is calculated.

[0048] In one embodiment, the supplier evaluation and recommendation module includes a data collection unit, a data preprocessing unit and a supplier recommendation unit. The data collection unit is used to collect information data of the supplier. The data processing unit is used to process the information data of the supplier to eliminate data differentiation. The supplier recommendation unit is used to classify quotations of suppliers using an improved k-means clustering analysis algorithm, divide suppliers into several different categories according to evaluation indicators, calculate the weight value of the indicator through the hierarchical analysis method, and find the optimal supplier according to the weighted score of the indicator.

[0049] In one embodiment, the data preprocessing unit processes the supplier's information data using mean-standard deviation standardization, and the calculation formula is:

[0050] x * =(x-μ) / σ

[0051] Among them, μ is the mean and σ is the standard deviation.

[0052] In one embodiment, the supplier recommendation unit includes a cluster analysis subunit and a hierarchical analysis evaluation subunit;

[0053] The cluster analysis subunit uses a cluster analysis algorithm to analyze suppliers and divide the suppliers into different clusters, so that the similarity of the indicator data of suppliers in the same cluster is high, and the difference of the indicator data of suppliers between clusters is high. The indicator is derived from the established supplier evaluation system. Several attributes are selected from the evaluation system for cluster analysis, and suppliers are divided into different categories according to the attributes;

[0054] After cluster analysis of the selected suppliers, the hierarchical analysis evaluation subunit calculates the weight of the first-level index relative to the target layer through the hierarchical analysis method, then calculates the relative weight of the second-level index relative to the first-level index, and then calculates the absolute weight of the second-level index, and calculates the horizontal index; calculates the score of each category according to the weight and index value, and selects the best supplier group. Finally, the score of each supplier in the best supplier group is calculated one by one, and the best supplier is selected for recommendation.

[0055] In the improved k-means clustering analysis algorithm:

[0056] The k-means clustering analysis algorithm finds k cluster centers for a data set containing n objects, and divides the data in the data set into k clusters according to the proximity principle, so that the k clusters meet two requirements: the number of data in each cluster cannot be less than 1, and each data object belongs to and only belongs to one cluster, and then adjusts the data attributes according to the decision function;

[0057] The algorithm first randomly selects k objects from n data objects to be classified. The k objects represent the average value in the cluster during the first classification. The Euclidean distance between the remaining nk objects and the k objects is calculated. The data object with the smallest distance and the average object are divided into the same cluster. In this way, n data are divided into k different clusters. Then, the average value in each cluster is recalculated, and the above classification steps are iterated until the result of the last classification is the same as the result of the second to last classification. The algorithm is the convergence of the criterion function, and the convergence function is:

[0058]

[0059] Where E represents the sum of square differences of all data objects, x represents the data object, and x i Represents cluster C i The smaller the convergence function is, the more similar the data in the cluster is, the higher the independence of the clusters is, and the lower the similarity correlation between clusters is; specifically:

[0060] Input: k sets S, containing target categories of k data objects

[0061] Output: k clusters

[0062] Step 1: Randomly select k data objects from the set S as the initial k clusters C1,...,C k The average value P1,...,P k .

[0063] Step 2: Iterate the following steps:

[0064] For each data object O in the data set S:

[0065] Calculate the mean P of the data object to each cluster i distance, find the P closest to the data object O i .

[0066] ||OP i ||=min{||OP j ||,1≤j≤k}

[0067] Assign data object O to P i Belongs to cluster C i .

[0068] For each cluster C j (1≤j≤k):

[0069] Calculate the mean value of each cluster object.

[0070] Calculate the clustering criterion function E,

[0071] Until the k clusters no longer change, the criterion function E converges.

[0072] Improvements include:

[0073] Create a two-dimensional array sum[k][m] to store the data objects in each cluster and a one-dimensional array count[k] to store the number of data objects in each cluster. In this way, the new average value of each cluster can be calculated directly through the formula, 0<i≤k without traversing the database;

[0074] Modify the condition at the end of the k-means algorithm to indicate that the mean value of each cluster no longer changes or changes significantly, so that the mean value of the cluster can be obtained, and use a two-dimensional array to calculate sum[k][m] and a one-dimensional array count[k] instead of traversing the database to calculate the criterion function. Specifically:

[0075] Input: A set S with n data objects and target classification number k

[0076] Output: k clusters, the mean value of each cluster no longer changes

[0077] step:

[0078] (1) Randomly select k data objects from the set S as the initial k clusters C1,...,C k The average value P1,...,P k .

[0079] (2) Initialize sum[i][j] = 0, count[i] = 0, where sum[i][j] is used to save the sum of the j attribute values ​​of the data objects in the i-th cluster, and count[i] is used to record the number of data objects in the i-th cluster.

[0080] (3) Iterate the following steps:

[0081] The new mean overwrites the old mean

[0082] for t=1 to n

[0083] do{

[0084] Save class label int j

[0085] for i=1 to k

[0086] {Calculate the distance between each data object and the mean of each cluster, find the minimum distance and record the number of clusters p}

[0087] Assign the object to C p

[0088] sum[m][i]=sum[m][i]+O[i][i]

[0089] The total number of objects of this class is count[m] = count[m] + 1

[0090] } / / End of loop

[0091] Get the new mean of each cluster: sum[m][i] / count[m]

[0092] Compare the old and the new;

[0093] (4) Know that the old and new means no longer change.

[0094] In one embodiment, it also includes a supplier quotation management module, which is connected to the control center and is used to store quotations of all existing suppliers for various products, including historical quotations and latest quotations.

[0095] In one embodiment, it also includes a user interaction interface, which is connected to the control center and is used to display the recommended best supplier, quotations from all suppliers, and perform human-computer interaction, that is, record and search supplier transactions.

[0096] The present invention adopts an improved k-means algorithm to improve the deficiency of frequent traversal of the database in the traditional algorithm. By establishing an indicator system, the cluster center is easy to set, is not affected by noise data, has practical significance, has a faster calculation speed and good clustering effect, and also reduces the time complexity of the algorithm, and improves the solution efficiency by collecting actual data of the network in quantity. At the same time, the present invention adopts the original post-evaluation data analysis of the procurement project to correct the existing bidding / comparison project review method. The system effectively corrects the problem of disconnection between the procurement process and the product use process, so that the optimized procurement weight scheme can better fit the evaluation and perception of the supplier by the current network demand unit.

[0097] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0098] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. An intelligent procurement system for a supply chain, characterized in that: It includes a control center, a data classification storage module, a new purchase project evaluation module, a supplier evaluation recommendation module, and a post-evaluation record module, wherein the control center is connected to the data classification storage module, the new purchase project evaluation module, the supplier evaluation recommendation module, and the post-evaluation record module respectively; The data classification storage module is used to store a large amount of past historical purchase data according to different categories of products, as well as store information data of multiple suppliers and historical price data of suppliers; The new procurement project evaluation module is used to extract the characteristic parameters of the training old procurement project supplier samples, and then calculate the reasonable weighted values ​​of each parameter to accurately and effectively evaluate the subsequent new procurement project bid evaluation / comparison score; The supplier evaluation and recommendation module is used to screen multiple suppliers using an improved k-means clustering analysis algorithm to select the best supplier from the multiple suppliers for recommendation; The post-evaluation record module is used to deliver the purchased products / services to the user department after the procurement implementation stage is over. After the procurement result shares are allocated and the products / services are provided, the procurement implementation department and the procurement demand department jointly evaluate the product / service scores of the winning suppliers and use them as post-evaluation scores.

2. The intelligent procurement system for supply chain according to claim 1, characterized in that: The new procurement project evaluation module includes an evaluation sample unit, a parameter extraction unit and a pre-training unit. The evaluation sample unit is used to obtain and store the post-evaluation score as the expected output value of the pre-training unit; the parameter extraction unit is used to extract the characteristic parameters used for bid evaluation / comparison in the new / old procurement project as the input node value of the pre-training unit; the pre-training unit is used to use the characteristic parameters to evaluate the new procurement project and calculate the score.

3. The intelligent procurement system for supply chain according to claim 1, characterized in that: The supplier evaluation and recommendation module includes a data collection unit, a data preprocessing unit and a supplier recommendation unit. The data collection unit is used to collect the information data of the supplier. The data processing unit is used to process the information data of the supplier to eliminate data differentiation. The supplier recommendation unit is used to classify the quotations of suppliers using an improved k-means clustering analysis algorithm, divide the suppliers into several different categories according to the evaluation indicators, calculate the weight values ​​of the indicators through the hierarchical analysis method, and find the optimal supplier according to the weighted scores of the indicators.

4. The intelligent procurement system for supply chain according to claim 3, characterized in that: The data preprocessing unit processes the supplier's information data using mean-standard deviation standardization, and the calculation formula is: x * =(x-μ) / σ Among them, μ is the mean and σ is the standard deviation.

5. The intelligent procurement system for supply chain according to claim 3, characterized in that: The supplier recommendation unit includes a cluster analysis subunit and a hierarchical analysis evaluation subunit; The cluster analysis subunit uses a cluster analysis algorithm to analyze suppliers and divide the suppliers into different clusters, so that the similarity of the indicator data of suppliers in the same cluster is high, and the difference of the indicator data of suppliers between clusters is high. The indicator is derived from the established supplier evaluation system. Several attributes are selected from the evaluation system for cluster analysis, and suppliers are divided into different categories according to the attributes; After cluster analysis of the selected suppliers, the hierarchical analysis evaluation subunit calculates the weight of the first-level index relative to the target layer through the hierarchical analysis method, then calculates the relative weight of the second-level index relative to the first-level index, and then calculates the absolute weight of the second-level index, and calculates the horizontal index; calculates the score of each category according to the weight and index value, and selects the best supplier group. Finally, the score of each supplier in the best supplier group is calculated one by one, and the best supplier is selected for recommendation.

6. The intelligent procurement system for supply chain according to claim 5, characterized in that: In the improved k-means clustering analysis algorithm, The k-means clustering analysis algorithm finds k cluster centers for a data set containing n objects, and divides the data in the data set into k clusters according to the proximity principle, so that the k clusters meet two requirements: the number of data in each cluster cannot be less than 1, and each data object belongs to and only belongs to one cluster, and then adjusts the data attributes according to the decision function; The algorithm first randomly selects k objects from n data objects to be classified. The k objects represent the average value in the cluster during the first classification. The Euclidean distance between the remaining nk objects and the k objects is calculated. The data object with the smallest distance and the average object are divided into the same cluster. In this way, n data are divided into k different clusters. Then, the average value in each cluster is recalculated, and the above classification steps are iterated until the result of the last classification is the same as the result of the second to last classification. The algorithm is the convergence of the criterion function, and the convergence function is: Where E represents the sum of square differences of all data objects, x represents the data object, and x i Represents cluster C i The smaller the convergence function is, the more similar the data in the cluster is, the higher the independence of the clusters is, and the lower the similarity correlation between clusters is; Improvements include: Create a two-dimensional array sum[k][m] to store the data objects in each cluster and a one-dimensional array count[k] to store the number of data objects in each cluster. In this way, the new average value of each cluster can be calculated directly through the formula, 0<i≤k without traversing the database; Modify the condition at the end of the k-means algorithm to indicate that the mean of each cluster no longer changes or changes significantly, so that the mean of the cluster can be obtained, and use a two-dimensional array to calculate sum[k][m] and a one-dimensional array count[k] instead of traversing the database to calculate the criterion function.

7. The intelligent procurement system for supply chain according to claim 1, characterized in that: It also includes a supplier quotation management module, which is connected to the control center and is used to store quotations of all existing suppliers for various products, including historical quotations and latest quotations.

8. The intelligent procurement system for supply chain according to claim 1, characterized in that: It also includes a user interaction interface, which is connected to the control center and is used to display the recommended best supplier, quotations from all suppliers, and perform human-computer interaction.

Images