An efficient traceability method for important products

Abstract

An efficient traceability method for important products belongs to the field of product circulation traceability technology. First, the enterprises within the regulatory scope are taken as connected nodes, and an enterprise connectivity network is constructed based on the nodes and the edges between nodes. Then, before the product starts to circulate, the traceability codes of the products of the same batch are continuously associated before and after the starting node. Afterwards, when a batch of products is in circulation, each circulation node samples and scans part of the products of a batch according to the sampling ratio requirements, and calculates the probability distribution of all products of the batch at each node of the connectivity network based on the partial sampling results. Finally, a problem product is found at a certain circulation node, and based on the product probability distribution data of each node of the connectivity network, all possible circulation links of the problem product are calculated, and the maximum probability circulation path is calculated to achieve the purpose of tracing the source and tracing the end. This method can significantly reduce the workload of enterprise traceability information entry. It only needs to sample and scan some products to restore the circulation information of all products.

Description

Technical Field

[0001] The present invention belongs to the technical field of product circulation tracing, and in particular is an efficient product circulation path tracing method. Background Art

[0002] In the field of product traceability, product codes are primarily assigned at the point of origin or manufacturer, with intermediaries scanning the codes and registering purchases, sales, and inventory on the traceability platform. This establishes a product distribution chain and achieves traceability. Each distribution link, such as manufacturers, wholesalers, and retailers, is required to register product codes individually upon entry and exit, using methods known as the "scan every code, verify every code, and verify every code." Current product traceability solutions require every node in the entire product distribution network to register every product's distribution code and supply flow information, resulting in significant operational costs and difficulties in widespread implementation. If data is not shared and inconsistent among participating entities, traditional traceability methods become labor-intensive and inefficient. Furthermore, businesses, the primary actors in the traceability system, are reluctant to upload traceability information or participate in distribution traceability due to privacy concerns and operational costs. Establishing separate traceability systems based on businesses also fails to achieve the goal of information sharing in the supply chain. Therefore, promoting traceability of critical products is imperative to truly achieve full traceability.

[0003] In order to promote the construction of an important product traceability system and ensure public consumption safety, the present invention proposes an efficient traceability method for important products to address the above problems. Summary of the Invention

[0004] The purpose of the present invention is to provide an efficient product traceability method. This method can significantly reduce the workload of enterprises in entering traceability information. There is no need to register the source and destination information of all circulating products, and there is no need to scan and record the traceability code of each product. Only a sampling of some products is required to scan the code to restore the circulation information of all products. This reduces the workload of traditional traceability methods, achieves flexible and controllable sampling intensity under different traceability requirements, and significantly improves the efficiency of traceability work. The present invention adopts the following technical solutions:

[0005] Step 1: The product production, circulation and use enterprises within the regulatory scope are regarded as connected nodes. There is a directed edge between the nodes with product flow relationship. The enterprise connectivity network is constructed based on the nodes and the edges between nodes.

[0006] Step 2: Before the product begins to circulate, a traceability code is assigned to each product in a batch at the starting node. The traceability codes of products in the same batch are continuously linked.

[0007] Step 3: When a batch of products is in circulation, each circulation node samples and scans a portion of the products in the batch according to the sampling ratio requirements. Based on the partial sampling results, the probability distribution of all products in the batch at each node in the connected relationship network is calculated.

[0008] Step 4: When a problem product is found at a certain circulation node, all possible circulation links of the problem product are calculated based on the product probability distribution data of each node in the connected relationship network, and the circulation path with the maximum probability is calculated to achieve the purpose of tracing the source and the end.

[0009] Furthermore, step 1 is as follows:

[0010] Use directed graph structure to abstractly represent enterprise connectivity network G = (V, E), where the node set V = {V0, ..., V t ,...,V m} represents each enterprise, and the edge set E is a directed edge. The direction of the edge represents that the upstream and downstream enterprises have a connected transaction relationship. Definition: Path(V t ) is the node V t Connected path set. Path(V t Each connected path in ) consists of a node V t It is composed of upstream nodes with connectivity.

[0011] Establish a node transfer probability matrix based on the number of nodes and edge connectivity of the enterprise connectivity network G in, Represents node V t The transfer probability to each node G on the enterprise connectivity network can be calculated by the following formula:

[0012]

[0013] Where, For node V t To node V j The transition probability, V tnext ∈V,V tnext For node V t child node set, V tnext Number of internal child nodes;

[0014] Furthermore, step 2 is as follows:

[0015] Definition: The total number of a batch of products M to be circulated is n

[0016] In the enterprise connectivity network G, each node is a connectivity node, and the nodes participating in the circulation of product M are circulation nodes.

[0017] At the circulation starting node V0, a ​​traceability code {M1, M2, ..., M n}. The coding span is set to c(M i ,M j ), indicating two different traceability codes M i ,Mj The span between 1≤c≤n-1, the maximum code M n The coding span c(M1,M n )=n-1, the coding span of two consecutive codes c(M i ,M i+1 )=1.

[0018] Definition: Minimum product sampling ratio θ, 0<θ≤1.

[0019] Definition: Product M circulation probability matrix Represents the connected nodes {V0,...,V t ,...,V m}, the circulation probability record of the batch of product M. Among them, Indicates that product M is at node V t The circulation probability matrix,

[0020] Furthermore, step 3 is as follows:

[0021] When products are in circulation, each circulation node will sample and scan the codes of part of a batch of products according to the inspection intensity requirements, and calculate the circulation probability of all products in a batch at each node based on the partial inspection results.

[0022] The minimum sampling ratio θ of product M can be adjusted according to the actual traceability requirements. It can be stipulated that 100% sampling inspection should be carried out on key products, then θ=1.

[0023] Definition: Flow to node V t The total number of products M is The sampling ratio is θ ’ , then node V t The number of products that should be sampled is Through node V t Spot check scanning allows circulation supervision departments to be 100% certain The product flows through node V t , which have not been sampled The product is determined by calculation at node V t The circulation probability of is achieved through the following methods:

[0024] Definition: Node V t Each scan product traceability code value is M s , M1≤M s ≤M n

[0025] Node V t The xth random inspection, To update the probability, the specific steps are as follows:

[0026] (1) The product code for the xth sampling inspection is M s ,

[0027] (2) Calculate the probability to update the code segment M D : The xth random inspection, M s The relevant probability update code segment is The x-1th random inspection, M s′ The relevant probability update code segment is Merge M sd and M s′d , and remove the repeated code to get node V t The xth sampling probability update code segment M D .

[0028] (3) Calculate the span matrix C d ,C d Each coding span c ′ Represents the probability update code segment M D The minimum distance between each internal code and the code of the product being sampled;

[0029] (4) According to the span matrix C d Update code segment M for probability D Perform two groupings; the first grouping: group the code segments with continuous spans and a circulation probability not equal to 1; the second grouping: based on the first grouping, if there are repeated values ​​in the coding span c′ within a group, group the code segments again at the maximum span position.

[0030] After 2 groupings, the code span c′ of each group of traceability codes is continuous and non-repeated, and the minimum value of code span c′ is c min ≥1,

[0031] Definition: N g The total number of groups obtained by grouping the node twice during the x-th random inspection.

[0032] (5) Calculate the probability of the products in each group. The specific steps are as follows:

[0033] (5.1) Calculate the minimum coding span c of each set of traceability codes min , maximum coding span c max , the sum of each group of coding spans is c sum ;

[0034] (5.2) The probability of circulation of each product in the group is calculated according to formula (2):

[0035]

[0036] After the probability calculation, the node V t Probability Matrix The sum is equal to the node V t Total number of products in circulation As shown in formula (3):

[0037]

[0038] (5.3) Same span probability average: sum the circulation probabilities of products with the same coding span c′ in different groups and perform probability average;

[0039] Each circulation node samples a product each time, and after calculating and updating the circulation probability matrix through steps (1) to (5) above, performs the reverse update in step (6).

[0040] (6) Reverse probability update: take the current node V t As the root node, the connected path set Path(V t ) as child nodes to form a reverse probability update tree; if on the enterprise connectivity network G, node V j For node V i Upstream node, in the reverse probability update tree, node V j That is node V i The specific reverse probability update steps are as follows:

[0041] (6.1) For the parent node V i Circulation probability matrix Solution Large product probability value and its corresponding traceability code segment M A

[0042] (6.2) Solve the flow probability matrix of each child node Large product probability value and its corresponding traceability code segment M B

[0043] (6.3) Solve the circulation probability matrix of each child node The probability value of small products and its corresponding traceability code segment M C

[0044] (6.4) Calculate M A With M B Coding similarity S:

[0045]

[0046] Where, Indicates M A and M B The number of identical traceability codes;

[0047] (6.5) For the child node V with the highest encoding similarity S j M A The product probability within the code segment is updated in reverse, according to formula (5):

[0048]

[0049] Where M a M A A traceability code in the code segment, for The probability value after reverse update, For node V i To node V j The transition probability.

[0050] (6.6) For the child node V with the highest encoding similarity S j M C The product probability within the code segment is updated in reverse order:

[0051]

[0052] Where, Reverse update of the child node V in step (6.5) j The circulation probability value, Δ is the probability increment, M c M C Zhongyi traceability code, for The probability value after reverse update.

[0053] Furthermore, step 4 is as follows:

[0054] Considering the actual traceability situation, a problem product is sampled at a circulation node. According to the product probability distribution data of each node in the connected relationship network G, all possible circulation links of the problem product are calculated, and the maximum probability circulation path is solved to achieve the purpose of tracing the source and the end.

[0055] The specific steps are:

[0056] (1) For node V t The connected path set Path(V t ) Dimensionality reduction and duplication removal are used to obtain the connected node set Node(V t ), Node(V t )∈V;

[0057] (2) According to the circulation probability matrix The circulation probability of the problem-encoded product in each node is related to the circulation node set Node(V t ) to perform probability mapping on each node and remove the connected node set Node(V t) in which the probability of circulation of problem products is 0, and remove the connected path set Path(V t ) The connectivity path where the redundant nodes are located;

[0058] (3) The connected path set Path (V t ) is the set of possible circulation paths for the problem product. The node probabilities of each possible circulation path in the set are accumulated and normalized to obtain the probabilities corresponding to all possible paths. The path with the maximum probability can be obtained by comparison.

[0059] The beneficial effects of the present invention are as follows: enterprises only need to sample some products to cooperate with circulation supervision departments in tracing important products, thereby protecting the privacy of enterprise product circulation data, reducing the workload of enterprises in tracing important products, and increasing the enthusiasm of enterprises in participating in tracing. At the same time, the intensity of sampling can be flexibly controlled under different tracing requirements, ensuring the accuracy of the calculation of the circulation link of problem products, and assisting circulation supervision departments in tracing important products. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 This is a flow chart of an efficient tracing method for an important product of the present invention

[0061] Figure 2 A connectivity network diagram of a specific embodiment of the present invention

[0062] Figure 3 This is a product simulation circulation network diagram for a specific embodiment of the present invention

[0063] Figure 4 The probability reverse update tree diagram of the specific embodiment of the present invention

[0064] Figure 5 This is the distribution diagram of the circulation probability matrix obtained from the random inspection of node V5

[0065] Figure 6 This is the distribution diagram of the circulation probability matrix obtained by sampling node V3

[0066] Figure 7 This is the distribution diagram of the circulation probability matrix obtained from the random inspection of node V4

[0067] Figure 8 The distribution diagram of the circulation probability matrix after node V4 is reversely updated DETAILED DESCRIPTION

[0068] A product circulation traceability method proposed by the present invention is described in detail as follows with reference to the accompanying drawings and examples.

[0069] like Figure 1The flowchart of the efficient product traceability method shown in the figure shows that in step 1, the enterprises producing, distributing, and using products within the regulatory scope are taken as connected nodes. There is a directed edge between the nodes with product flow relationships. The enterprise connectivity network is constructed based on the nodes and the edges between the nodes. Specifically:

[0070] Use directed graph structure to abstractly represent enterprise connectivity network G = (V, E), where the node set V = {V0, ..., V t ,...,V m} represents each enterprise, and the edge set E is a directed edge. The direction of the edge represents the connectivity between upstream and downstream enterprises. Definition: Path(V t ) is the node V t Connected path set. Path(V t ) is connected to the node V t It is composed of upstream nodes with connectivity. Figure 2 In the enterprise connectivity network diagram shown in Figure G, V = {V0, V1, V2, V3, V4, V5, V6}, Path(V3) = {V0->V1->V3, V0->V2->V3}.

[0071] Establish a node transfer probability matrix based on the number of nodes and edge connectivity of the enterprise connectivity network G in, Represents node V t The transfer probability to each node in the enterprise connectivity network can be calculated by the following formula:

[0072]

[0073] Where, For node V t To node V j The transition probability, V tnext ∈V,V tnext For node V t child node set, V tnext Number of internal child nodes;

[0074] like Figure 2 As shown,

[0075] Step 2: Before the product begins to circulate, a traceability code is assigned to each product in a batch at the starting node. The traceability codes of products in the same batch are continuously linked. Specifically:

[0076] Definition: The total number of a batch of products M to be circulated is n, and the traceability code {M1, M2, ..., M n}. The coding span is set to c(M i,M j ), represents the distance between two different traceability codes. 1≤c≤n-1, the maximum code M n The coding span c(M1,M n )=n-1, the coding span of two consecutive codes c(M i ,M i+1 )=1. For example, if the total number of product M is 30, then the code {M1,,M2,…,M 30}, maximum code M 30 , the minimum code M1, then the code span c(M1,M 30 )=29, M1 and M2 are consecutively coded, then the coding span c(M1, M2)=1.

[0077] Definition: Minimum product sampling ratio θ, 0<θ≤1.

[0078] Definition: Product M circulation probability matrix Represents each node {V0,...,V t ,...,V m}The circulation probability record of the batch of product M. Among them, Indicates that product M is at node V t The circulation probability matrix, like Figure 2 The enterprise connectivity network diagram shown in the figure, is the circulation probability matrix of node V3,

[0079] Step 3: When the products are in circulation, each circulation node will sample and scan the codes of a portion of the products in a batch according to the sampling inspection intensity requirements. Based on the partial sampling inspection results, the circulation probability of all products in a batch at each node is calculated, specifically:

[0080] The minimum sampling ratio θ of product M can be adjusted according to the actual traceability requirements. It can be stipulated that 100% sampling inspection should be carried out on key products, then θ=1.

[0081] Definition: Product M flows to node V t The total number is The sampling ratio is θ', then the node V t The number of products that should be sampled is Through node V t Spot check scanning makes the circulation supervisor and the enterprise nodes involved in the circulation of product M 100% certain The product flows through node V t , which have not been sampled The product is determined by calculation at node V t The circulation probability of is achieved through the following methods:

[0082] Node V t The xth random inspection To update the probability, the specific steps are as follows:

[0083] (1) The product code for the xth sampling inspection is M s ,

[0084] (2) Calculate the probability to update the code segment M D : The xth random inspection, M s The relevant probability update code segment is The x-1th random inspection, M s′ The relevant probability update code segment is Merge M sd and M s′d , and remove the repeated code to get node V t The xth sampling probability update code segment M D .

[0085] (3) Calculate the span matrix C d ,C d Each coding span c′ represents the probability update code segment M D The minimum distance between each internal code and the code of the product being sampled;

[0086] (4) According to the span matrix C d Update code segment M for probability D Perform two groupings: In the first grouping, code segments with continuous spans and a circulation probability not equal to 1 are grouped together; in the second grouping, based on the first grouping, if there are repeated values ​​in the coding span c′ within a group, the code segments are grouped again at the maximum span position.

[0087] After 2 groupings, the code span c′ of each group of traceability codes is continuous and non-repeated, and the minimum value of code span c′ is c min ≥1,

[0088] Definition: N g The total number of groups obtained by performing two groupings for the x-th random inspection of the node;

[0089] (5) Calculate the probability of the products in each group. The specific steps are as follows:

[0090] (5.1) Calculate the minimum coding span c of each set of traceability codes min , maximum coding span c max , the sum of each group of coding spans is c sum ;

[0091] (5.2) The probability of circulation of each product in the group is calculated according to formula (2):

[0092]

[0093] After the probability calculation, the node V t Probability Matrix The sum is equal to the node V t Total number of products in circulation As shown in formula (3):

[0094]

[0095] (5.3) Same span probability average: Accumulate the circulation probability of products with the same coding span c′ in different groups and perform probability average;

[0096] Each time a product is sampled at each node, after the circulation probability matrix is ​​calculated and updated by the above steps (1) to (5), step (6) is performed to reversely update each child node on the reverse update tree.

[0097] (6) Perform reverse probability update: take the current node V t As the root node, the connected path set Path(V t ) as child nodes to form a reverse probability update tree; if on the enterprise connectivity network G, node V j For node V i Upstream node, in the reverse probability update tree, node V j That is node V i The specific reverse probability update steps are as follows:

[0098] (6.1) For the parent node V i Circulation probability matrix Solution Large product probability value and its corresponding traceability code segment M A

[0099] (6.2) Solve the flow probability matrix of each child node Large product probability value and its corresponding traceability code segment M B

[0100] (6.3) Solve the circulation probability matrix of each child node The probability value of small products and its corresponding traceability code segment M C

[0101] (6.4) Calculate M A With M B Coding similarity S:

[0102]

[0103] Where, Indicates M A and M B The number of identical traceability codes;

[0104] (6.5) For the child node V with the highest encoding similarity S j M A The product probability within the code segment is updated in reverse according to the following formula:

[0105]

[0106] Where M a M A A traceability code in the code segment, for The probability value after reverse update, For node V i To node V j The transition probability.

[0107] (6.6) For the child node V with the highest encoding similarity S j M C The product probability within the code segment is updated in reverse order:

[0108]

[0109] Where, Reverse update of the child node V in step (6.5) j The circulation probability value, Δ is the probability increment, M c M C Zhongyi traceability code, for The probability value after reverse update.

[0110] Each sampling operation of each circulation node goes through the above 6 steps of probability calculation, such as Figure 3 The circulation simulation result of product M is shown, and the traceability code is {M 16 ~M 25} products circulate to node V5, i.e. Product M sampling ratio θ′=0.2, then Node V5 randomly checks at least two products. The first time node V5 scans the code, the product code is M. 20 , scan the code for the second time and select M 24 , the product code that was not scanned is {M 16 ~M 19},{M 21 ~M 23},{M 25}, the circulation probability at node V5 is determined by calculation. The specific calculation example is as follows:

[0111] (1) Node V5 first random inspection, x = 1, the traceability code of the inspected item is M 20 , M s =M 20 ,

[0112]

[0113] (2) Calculate the probability to update the code segment M D : Extended code segment M D ={M 11 ~M 29};

[0114] (3) Calculate the span matrix C d : Node V5 first random inspection, M 20 M 11 The maximum span of the traceability code, M 11 With M 20 The span is 9, M 24 With M 20 The span is 4, C d =[9,8,7,6,5,4,3,2,1,0,1,2,3,4,5,6,7,8,9] T , C d The span value in is the extended code segment M D Each traceability code is separated from the traceability code M 20 span;

[0115] (4) According to the span matrix C d Update code segment M for probability D Make 2 groups: M D =[M 11 ~M 29 ], according to the span matrix C d Grouping, {M 11 ~M 19} has a span value of [9~1], {M 21 ~M 29} has a span value of [1 to 9], and the grouping result is {M 11 ~M 19},{M 21 ~M 29}, N g =2.

[0116] (5) Calculate the probability of each group. The specific calculation example is as follows:

[0117] (5.1) Calculate the minimum coding span c for each group min , maximum coding span c max , the total coding span of each group csum :

[0118] Group {M 11 ~M 19}, the span value is [9~1], c min =1,c max =9,c sum =45,

[0119] Group {M 21 ~M 29}, the span value is [1~9], c min =1,c max =9,c sum =45;

[0120] (5.2) The probability of circulation of each product in the group is calculated based on formula (2). For example:

[0121]

[0122] After probability accumulation calculation, the following node probability matrix can be obtained:

[0123]

[0124] (5.3) Average probability of the same span: N g =2,M 16 、M 24 Belong to different groups, and M 20 The span of is 4, then Calculate the average:

[0125]

[0126] The first random inspection of node V5 is calculated through steps (1) to (5). like Figure 5 shown.

[0127] (6) Perform reverse probability update, such as Figure 4 The probability of node V5 is reversely updated in the tree diagram shown. The child nodes of node V5 are V3 and V4. The specific calculation example is as follows:

[0128] (6.1) Parent node V5 circulation probability matrix The cumulative probability value of the top 10 products corresponds to the code M A ={M 16 ~M 25}

[0129] (6.2) Figure 6 The subnode V3 shown is obtained by sampling and calculating the value in steps (1) to (5). MB ={M4~M8,M 13 ~M 17},like Figure 7 The subnode V4 shown is obtained by sampling and calculating the value in steps (1) to (5). M B ={M1~M4,M 15 ~M 20}

[0130] (6.3) Figure 6 As shown, the M of child node V3 C ={M 21 ~M 30},like Figure 7 As shown, the M of child node V4 C ={M7~M 11 ,M 26 ~M 30}

[0131] (6.4) The encoding similarity between node V5 and child node V3 is 0.2. Similarly, the encoding similarity of child node V4 is 0.5;

[0132] (6.5) For the child node V4 with the highest encoding similarity S, A The corresponding product probability is updated in reverse:

[0133] From formula (1),

[0134] If the reverse update like Figure 7 As shown, Through steps (1) to (5), we can get

[0135] From formula (5), we can get: but

[0136] (6.6) For the child node V4 with the highest encoding similarity S, C The product probability within the code segment is updated in reverse order:

[0137] From formula (6), we can get:

[0138] If the reverse update like Figure 7 As shown, Through steps (1) to (5), we can get

[0139] From formula (7), we can get:

[0140] After reverse update The calculation results are as follows Figure 8 shown.

[0141] Step 4: When a problem product is sampled at a distribution node, all possible distribution links of the problem product are calculated based on the product probability distribution data of each node in the connected relationship network, and the maximum probability distribution path is calculated. The specific steps are as follows:

[0142] The specific steps are:

[0143] (1) For node V t The connected path set Path(V t ) Dimensionality reduction and duplication removal are used to obtain the connected node set Node(V t ), Node(V t )∈V;

[0144] (2) According to the circulation probability matrix The cumulative probability of circulation of the problem-coded products in each node is related to the circulation node set Node(V t ) each node to do probability mapping, remove the connected node set Node(V t ) in which the probability of circulation of problem products is 0, and remove the connected path set Path(V t ) The connectivity path where the redundant nodes are located;

[0145] (3) The connected path set Path (V t ) is the set of possible circulation paths for the problem product. The node probabilities of each possible circulation path in the set are accumulated and normalized to obtain the probabilities corresponding to all possible paths. The path with the maximum probability can be obtained by comparison.

[0146] If the traceability code M is sampled at the circulation node V5 20 For the problem product, calculate M 20 The maximum probability flow path in the enterprise connectivity network G is:

[0147] (1) Figure 2 In the enterprise connectivity network diagram shown, the connectivity path set Path(V5) of node V5 is reduced in dimension and deduplicated to obtain the connectivity node set Node(V5);

[0148] Path(V5)={V0->V1->V3->V5,V0->V1->V4->V5,V0->V2->V3->V5,V0->V2->V4->V5}

[0149] Node(V5)={V0,V1,V2,V3,V4,V5}

[0150] (2) According to the circulation probability matrix PM The problem coding product M of each node 20 The circulation probability of the product is mapped with the probability of each node in the connected node set Node(V5), and the redundant nodes with the circulation probability of the problem product of 0 in the connected node set Node(V5) are removed. Then node V3 is a redundant node, and the connected paths where the redundant nodes are located in the connected path set Path(V5) are removed, that is, V0->V1->V3->V5 and V0->V2->V3->V5 are removed;

[0151] (3) The connected path set Path (V5) after redundancy removal is the possible circulation path set of the problem product. The node probabilities of each possible circulation path in the possible circulation path set are accumulated and normalized to obtain the probabilities corresponding to all possible paths. By comparison, the path with the maximum probability can be obtained.

[0152] Calculated by step 3 of the present invention,

[0153] Then the maximum probability path is V0->V1->V4->V5, which is the same as Figure 3 The simulated flow paths of the products shown are identical.

Claims

1. An efficient traceability method for important products, characterized by: The details are as follows: Step 1: Enterprises involved in the production, distribution, and use of products within the regulatory scope are considered as connected nodes. There is a directed edge between nodes with product flow relationships. An enterprise connectivity network is constructed based on the nodes and the edges between them. Step 2: Before the product begins to circulate, a traceability code is assigned to each product in a batch at the starting node. The traceability codes of products in the same batch are continuously linked. Step 3: When a batch of products is in circulation, each circulation node samples and scans the codes of a portion of the products in the batch according to the sampling ratio requirements. Based on the partial sampling results, the probability distribution of all products in the batch at each node in the connected relationship network is calculated; Step 4: When a problem product is found at a certain circulation node, all possible circulation links of the problem product are calculated based on the product probability distribution data of each node in the connected relationship network, and the circulation path with the maximum probability is calculated to achieve the purpose of tracing the source and the end.

2. The method for efficient tracing of important products according to claim 1, characterized in that: Step 1 is as follows: Use directed graph structure to abstractly represent enterprise connectivity network G = (V, E), where the node set V = {V0, ..., V t ,...,V m } represents each enterprise, and the edge set E is a directed edge. The direction of the edge represents that the upstream and downstream enterprises have a connected transaction relationship. Definition: Path(V t ) is the node V t Connected path set; Path(V t Each connected path in ) consists of a node V t It is composed of upstream nodes with connectivity; Establish a node transfer probability matrix based on the number of nodes and edge connectivity of the enterprise connectivity network G in, Represents node V t The transfer probability to each node G on the enterprise connectivity network can be calculated by the following formula: Where, For node V t To node V j The transition probability, V tnext ∈V,V tnext For node V t child node set, V tnext The number of internal child nodes.

3. The method for efficient tracing of important products according to claim 1, characterized in that: Step 2 is as follows: Definition: The total number of a batch of products M to be circulated is n In the enterprise connectivity network G, each node is a connectivity node, and the nodes involved in the circulation of product M are circulation nodes; At the circulation starting node V0, a ​​traceability code {M1, M2, ..., M n }; The coding span is set to c(M i ,M j ), indicating two different traceability codes M i ,M j The span between; 1≤c≤n-1, the maximum code M n The coding span c(M1,M n )=n-1, the coding span of two consecutive codes c(M i ,M i+1 )=1; Definition: Minimum product sampling ratio θ, 0<θ≤1; Definition: Product M circulation probability matrix Represents the connected nodes {V0,...,V t ,...,V m }, the circulation probability record of the batch of product M; where, Indicates that product M is at node V t The circulation probability matrix, 4. The method for efficient tracing of important products according to claim 1, characterized in that: Step 3 is as follows: When products are in circulation, each circulation node will randomly scan the codes of a batch of products according to the inspection intensity requirements. Based on the partial inspection results, the circulation probability of all products in the batch at each node is estimated; The minimum sampling ratio θ of product M can be adjusted according to the actual traceability requirements. It can be stipulated that 100% sampling inspection should be carried out on key products, then θ = 1; Definition: Flow to node V t The total number of products M is The sampling ratio is θ ’ , then node V t The number of products that should be sampled is Through node V t Spot check scanning allows circulation supervision departments to be 100% certain The product flows through node V t , which have not been sampled The product is determined by calculation at node V t The circulation probability of is achieved through the following methods: Definition: Node V t Each scan product traceability code value is M s , M1≤M s ≤M n Node V t The xth random inspection, the circulation probability matrix of product M To update the probability, the specific steps are as follows: (1) The product code for the xth sampling inspection is M s , (2) Calculate the probability to update the code segment M D : The xth random inspection, M s The relevant probability update code segment is The x-1th random inspection, M s ′ is detected, and the relevant probability update code segment is Merge M sd and M s′d , and remove the repeated code to get node V t The xth sampling probability update code segment M D ; (3) Calculate the span matrix C d ,C d Each coding span c′ represents the probability update code segment M D The minimum distance between each internal code and the code of the product being sampled; (4) According to the span matrix C d Update code segment M for probability D Perform two groupings; the first grouping: group code segments with continuous spans and a non-1 probability of circulation into one group; the second grouping: based on the first grouping, if there are repeated values ​​in the code span c′ within a group, group the code segments again at the position of the maximum span; After 2 groupings, the code span c′ of each group of traceability codes is continuous and non-repeated, and the minimum value of code span c′ is c min ≥1, Definition: N g The total number of groups obtained by performing two groupings for the x-th random inspection of the node; (5) Calculate the probability of the products in each group. The specific steps are as follows: (5.1) Calculate the minimum coding span c of each set of traceability codes min , maximum coding span c max , the sum of each group of coding spans is c sum ; (5.2) The probability of circulation of each product in the group is calculated according to formula (2): After the probability calculation, the node V t Probability Matrix The sum is equal to the node V t Total number of products in circulation As shown in formula (3): (5.3) Same span probability average: sum the circulation probabilities of products with the same coding span c′ in different groups and perform probability average; Each circulation node samples a product each time, and after calculating and updating the circulation probability matrix according to steps (1) to (5) above, performs the reverse update in step (6); (6) Reverse probability update: take the current node V t As the root node, the connected path set Path(V t ) as child nodes to form a reverse probability update tree; if on the enterprise connectivity network G, node V j For node V i Upstream node, in the reverse probability update tree, node V j That is node V i The specific reverse probability update steps are as follows: (6.1) For the parent node V i Circulation probability matrix Solution Large product probability value and its corresponding traceability code segment M A (6.2) Solve the flow probability matrix of each child node Large product probability value and its corresponding traceability code segment M B (6.3) Solve the circulation probability matrix of each child node The probability value of small products and its corresponding traceability code segment M C (6.4) Calculate M A With M B Coding similarity S: Where, Indicates M A and M B The number of identical traceability codes; (6.5) For the child node V with the highest encoding similarity S j M A The product probability within the code segment is updated in reverse, according to formula (5): Where M a M A A traceability code in the code segment, for The probability value after reverse update, For node V i To node V j The transition probability of (6.6) For the child node V with the highest encoding similarity S j M C The product probability within the code segment is updated in reverse order: Where, Reverse update of the child node V in step (6.5) j The probability value, Δ is the probability increment, M c M C Zhongyi traceability code, for The probability value after reverse update.

5. The method for efficient tracing of important products according to claim 1, characterized in that: Step 4 is as follows: Considering the actual traceability situation, a problem product is sampled at a circulation node. Based on the product probability distribution data of each node in the connected relationship network G, all possible circulation links of the problem product are calculated to solve the circulation path with the maximum probability, thus achieving the purpose of tracing back to the source and the end; The specific steps are: (1) For node V t The connected path set Path(V t ) Dimensionality reduction and duplication removal are used to obtain the connected node set Node(V t ), Node(V t )∈V; (2) According to the circulation probability matrix The circulation probability of the problem-encoded product in each node is related to the circulation node set Node(V t ) to perform probability mapping on each node and remove the connected node set Node(V t ) in which the probability of circulation of problem products is 0, and remove the connected path set Path(V t ) The connectivity path where the redundant nodes are located; (3) The connected path set Path (V t ) is the set of possible circulation paths for the problem product. The node probabilities of each possible circulation path in the set are accumulated and normalized to obtain the probabilities corresponding to all possible paths. The maximum probability path can be obtained by comparison.

Images