Sequential sampling inspection model construction method based on prior information

Abstract

The invention discloses a prior information-based sequential sampling inspection model construction method. The method comprises the steps of S1, determining an inspection target and prior information of a product; s2, constructing and optimizing a sequential sampling inspection model, establishing an influence relationship between an inspection target and prior information based on the sequential sampling inspection model, and obtaining a prior information variable X; s3, analyzing the influence degree Y of the prior information on the inspection target based on the prior information variable X in the S2, and presetting a qualified critical value H; the invention belongs to the technical field of sequential sampling inspection model construction, and aims at solving the problems that in the prior art, a sampling inspection result has a large deviation with an actual situation, the accuracy is poor, and the inspection efficiency is reduced. The method has the technical effects that on the premise that the preset precision requirement is met, the higher inspection standard is achieved with the smaller sampling amount, and therefore the efficiency and accuracy of sampling inspection are remarkably improved.

Description

Technical Field

[0001] The present application belongs to the technical field of sequential sampling test model construction, and specifically relates to a method for constructing a sequential sampling test model based on prior information. Background Art

[0002] In the fields of industrial production, quality control, etc., sampling inspection is an important means to ensure product quality. Sequential sampling inspection is one of the more common inspection methods among sampling inspection methods. Sequential sampling inspection, also known as successive sampling inspection or continuous sampling inspection, is a statistical sampling method used for sampling inspection without predetermining the total number of samples. The basic idea is that it is a method that does not specify the total number of samples in advance, only draws one or a group of samples each time, and decides whether to continue sampling or stop sampling based on the probability of the results of the previous n tests. This method can dynamically adjust the number of samples according to the actual situation, so as to minimize the number of samples while ensuring the accuracy of the inspection, improve the inspection efficiency, and reduce the number of test samples and the test time.

[0003] However, most of the current sequential sampling test models often ignore the guidance and influence of prior information on the test target, which leads to the blindness of the sequential sampling test model in the sampling process, resulting in a large deviation between the sampling test results and the actual situation, and poor accuracy. At the same time, more samples are required in the sampling process to achieve the expected test accuracy, which increases the time and economic costs of the test and reduces the test efficiency. Therefore, there is an urgent need for a method to construct a sequential sampling test model based on prior information to improve the efficiency and accuracy of sampling tests. Summary of the invention

[0004] The present application provides a method for constructing a sequential sampling inspection model based on prior information, aiming to solve the problems in the prior art that the sampling inspection results have a large deviation from the actual situation, the accuracy is poor and the inspection efficiency is reduced.

[0005] A method for constructing a sequential sampling inspection model based on prior information, the method comprising:

[0006] S1: Determine the inspection objectives and prior information of the product;

[0007] S2: Construct and optimize the sequential sampling test model, establish the influence relationship between the test target and the prior information based on the sequential sampling test model, and obtain the prior information variable X;

[0008] S3: Analyze the influence degree Y of the prior information on the inspection target based on the prior information variable X in S2, and preset the qualified critical value H;

[0009] S4: Incorporate prior information and qualified critical value H into the sequential sampling inspection model to determine the sampling strategy;

[0010] S5: Obtain relevant information of the product to be tested, and perform sampling operations based on the sequential sampling inspection model;

[0011] S6: Determine the termination condition according to the impact degree Y of the product to be tested.

[0012] Optionally, the specific steps of S2 are as follows:

[0013] S2.1: Data processing: pre-processing and integration of the collected prior information data;

[0014] S2.2: Model construction: Select the sequential probability ratio test (SPRT) method and statistical model to construct the sequential sampling test model;

[0015] S2.3: Setting parameters: According to the historical test data in the prior information, set the preliminary sampling quantity, sampling frequency, sampling error range and regression coefficient for the sequential sampling test model, and convert the statistical model into a specific algorithm;

[0016] S2.4: Optimization model: The production parameters and material properties of the product in the prior information are used as simulation data input, and the prior information is processed and analyzed through the sequential sampling inspection model to obtain the inspection target prediction results;

[0017] S2.5: Through data mining technology, a large amount of prior information data is divided into n groups, where (n=1, 2, 3...n), so that the similarity of data points in the same category is high, and the similarity of data points between different groups is low, and n groups of prior information variables X are obtained.

[0018] Optionally, the prior information includes historical inspection data, production parameters and material properties of the product.

[0019] Optionally, the preprocessing requires cleaning and testing the data, identifying and processing missing values ​​and duplicate values ​​in the prior information data, and processing the prior information data.

[0020] Optionally, S3 specifically includes the following steps:

[0021] S3.1: Based on statistical analysis techniques, the influence of the prior information after grouping on the test target is quantified to determine the relationship between the prior information and the test target, and to provide a basis for the subsequent critical value;

[0022] The calculation formula of the impact degree Y is:

[0023] Y=E0+E1X1+E2X2+……+E n X n

[0024] Among them, E0, E1, E2, ..., E n are regression coefficients, X1, X2, ..., X n is n sets of prior information;

[0025] S3.2: Based on the prior information of the product, set an a priori qualified critical value H and input it into the sequential sampling inspection model.

[0026] Optionally, the sampling strategy includes setting of sampling quantity, sampling interval and updating frequency parameters of prior information.

[0027] Optionally, the termination condition is that the impact degree Y≥qualified critical value H.

[0028] Optionally, if the impact degree Y≥qualified critical value H, it means that the product meets the quality standard, that is, the product is qualified; if the impact degree Y<qualified critical value H, it means that the product does not meet the quality standard, that is, the product is unqualified. Compared with the prior art, the present application has at least the following beneficial effects:

[0029] This application is based on further analysis and research of existing technical problems. By incorporating prior information into the sequential sampling test model, the calculation accuracy of the sequential sampling test model is improved. Prior information effectively utilizes previously accumulated data and experience to lay a solid foundation for the subsequent sampling process; combined with the sequential sampling method, the sampling strategy is adjusted to achieve higher inspection standards with less sampling while meeting the predetermined accuracy requirements, thereby significantly improving the efficiency and accuracy of sampling inspection. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 A flowchart of a method for constructing a sequential sampling inspection model based on prior information provided in one embodiment of the present application; DETAILED DESCRIPTION

[0031] In order to make the objectives, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0032] like Figure 1 As shown, the present application provides a method for constructing a sequential sampling inspection model based on prior information, comprising the following steps:

[0033] S1: Determine the inspection objectives and prior information of the product; the inspection objectives include the product's non-conforming rate and quality characteristics; the prior information includes the product's historical inspection data, production parameters, and material characteristics, which helps to understand the distribution and influencing factors of product quality;

[0034] S2: Construct and optimize the sequential sampling test model, establish the influence relationship between the test target and the prior information based on the sequential sampling test model, and obtain the prior information variable X;

[0035] The specific steps of S2 are as follows:

[0036] S2.1: Data processing: pre-processing and integration of the collected prior information data;

[0037] During preprocessing, the data needs to be cleaned and tested, missing values ​​and duplicate values ​​in the prior information data need to be identified and processed, and the prior information data needs to be processed. For example, the fillna() function of the Pandas library can be used to fill missing values, and the drop_duplicates() function of the Pandas library can be used to delete duplicate values.

[0038] The preprocessed prior information data is normalized using standardized tools, and the data from different data sources are integrated to form a unified and complete data set.

[0039] S2.2: Model construction: Select the sequential probability ratio test (SPRT) method and statistical model to construct a sequential sampling inspection model; so that the model can calculate and describe the product's non-conforming rate and quality characteristics.

[0040] S2.3: Setting parameters: Based on the historical test data in the prior information, set the preliminary sampling quantity, sampling frequency, sampling error range and regression coefficient for the sequential sampling inspection model, and convert the statistical model into a specific algorithm to facilitate calculation and judgment in the actual sampling inspection process.

[0041] S2.4: Optimization model: The production parameters and material properties of the products in the prior information are used as simulation data input, and the prior information is processed and analyzed through the sequential sampling inspection model to obtain the inspection target prediction results; by comparing and analyzing the relationship between the historical inspection data of the products in the prior information and the inspection target prediction results, the accuracy, sensitivity and specificity indicators of the model are evaluated to avoid deviations or deficiencies in the sequential sampling inspection model and complete the optimization process.

[0042] S2.5: Through data mining technology, a large amount of prior information data is divided into n groups, where (n = 1, 2, 3...n), so that the similarity of data points in the same category is high, and the similarity of data points between different groups is low, and n groups of prior information variables X are obtained. It is easy to identify data clusters related to the test target, so as to discover potential patterns or trends.

[0043] S3: Based on the prior information variable X in S2, the influence degree Y of the prior information on the inspection target is analyzed, and the qualified critical value H is preset; the influence degree Y is used to represent a certain quality characteristic of the product.

[0044] The impact degree Y is used to describe the probability of defective products in product quality. As the decision-making basis of the sequential sampling inspection model, the impact degree Y can minimize the inspection cost while ensuring the inspection accuracy.

[0045] S3 specifically includes the following steps:

[0046] S3.1: Based on statistical analysis techniques, the influence of the prior information after grouping on the test target is quantified to determine the relationship between the prior information and the test target, and to provide a basis for the subsequent critical value;

[0047] The calculation formula of the impact degree Y is:

[0048] Y=E0+E1X1+E2X2+……+E n X n

[0049] Among them, E0, E1, E2, ..., E n are regression coefficients, X1, X2, ..., X n is n sets of prior information.

[0050] S3.2: According to the prior information of the product, set a priori qualified critical value H and input it into the sequential sampling inspection model. The qualified critical value H is used as a benchmark for judging whether the product is qualified. Before setting it, it is necessary to comprehensively consider multiple factors such as product quality volatility, production costs, and customer demand to ensure that it is neither too loose to cause inferior products to enter the market, nor too strict to cause unnecessary waste.

[0051] S4: Incorporate prior information and qualified critical value H into the sequential sampling inspection model to determine the sampling strategy.

[0052] The sampling strategy includes the setting of parameters such as the number of samples, sampling interval, and the frequency of updating the prior information. During the sampling process, the prior information is updated in real time, and the sampling strategy is dynamically adjusted according to the updated prior information to ensure the representativeness and effectiveness of the sampling results.

[0053] The sample size is one of the core parameters in the sampling strategy, which directly affects the statistical power. A sufficient sample size can improve the statistical power and reduce the risk of failing to detect the actual effect; at the same time, the sample size is also closely related to the sampling error. An appropriate sample size can control the sampling error within an acceptable range.

[0054] Reasonable sampling intervals can avoid data redundancy and duplication and improve data utilization efficiency. Too short sampling intervals may lead to high correlation between data and increase the complexity of analysis; while too long sampling intervals may miss important information.

[0055] The setting of sampling interval needs to consider the speed of data change. For fast-changing data, a shorter sampling interval is required to capture the trend of data changes; for slowly changing data, the sampling interval can be appropriately extended.

[0056] The update frequency of prior information should reflect the timeliness of the data. In a rapidly changing environment, prior information needs to be updated frequently to maintain its accuracy and effectiveness; in a stable environment, the update frequency can be appropriately reduced.

[0057] S5: Obtain the associated information of the products to be tested, and perform sampling operations based on the sequential sampling inspection model, wherein the associated information includes the volume, production parameters, and material properties of each product to be tested.

[0058] During the sampling operation, a certain number of samples are randomly selected from the products to be tested for testing. The testing process should be strictly carried out in accordance with the methods and standards of the sequential sampling inspection model to ensure the accuracy of the test results.

[0059] S6: Determine the termination condition according to the impact degree Y of the product to be tested. The termination condition is that the impact degree Y ≥ the qualified critical value H.

[0060] If the impact degree Y ≥ the qualified critical value H, it means that the product meets the quality standards, that is, the product is qualified, and sampling is terminated to minimize the number of samplings while ensuring accuracy; if the impact degree Y < the qualified critical value H, it means that the product does not meet the quality standards, that is, the product is unqualified, and subsequent sampling is continued, and sampling is stopped when the requirements are met.

[0061] The data mining technology in step S2.4 generally refers to a clustering algorithm, and is not limited to a K-means algorithm, and may be, for example, a hierarchical clustering algorithm;

[0062] In the above-mentioned method for constructing a sequential sampling test model based on prior information, the calculation accuracy of the sequential sampling test model is improved by incorporating prior information into the sequential sampling test model. Prior information effectively utilizes previously accumulated data and experience to lay a solid foundation for the subsequent sampling process, and then combines the sequential sampling method to adjust the sampling strategy to achieve a higher inspection standard with a smaller sampling amount while meeting the predetermined accuracy requirements, thereby significantly improving the efficiency and accuracy of sampling inspection.

[0063] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A method for constructing a sequential sampling inspection model based on prior information, characterized in that: The method comprises: S1: Determine the inspection objectives and prior information of the product; S2: Construct and optimize the sequential sampling test model, establish the influence relationship between the test target and the prior information based on the sequential sampling test model, and obtain the prior information variable X; S3: Analyze the influence degree Y of the prior information on the inspection target based on the prior information variable X in S2, and preset the qualified critical value H; S4: Incorporate prior information and qualified critical value H into the sequential sampling inspection model to determine the sampling strategy; S5: Obtain relevant information of the product to be tested, and perform sampling operations based on the sequential sampling inspection model; S6: Determine the termination condition according to the impact degree Y of the product to be tested.

2. The method for constructing a sequential sampling inspection model based on prior information according to claim 1, characterized in that: The specific steps of S2 are as follows: S2.1: Data processing: pre-processing and integration of the collected prior information data; S2.2: Model construction: Select the sequential probability ratio test (SPRT) method and statistical model to construct the sequential sampling test model; S2.3: Setting parameters: According to the historical test data in the prior information, set the preliminary sampling quantity, sampling frequency, sampling error range and regression coefficient for the sequential sampling test model, and convert the statistical model into a specific algorithm; S2.4: Optimization model: The production parameters and material properties of the product in the prior information are used as simulation data input, and the prior information is processed and analyzed through the sequential sampling inspection model to obtain the inspection target prediction results; S2.5: Through data mining technology, a large amount of prior information data is divided into n groups, where (n=1, 2, 3...n), so that the similarity of data points in the same category is high, and the similarity of data points between different groups is low, and n groups of prior information variables X are obtained.

3. The method for constructing a sequential sampling inspection model based on prior information according to claim 1, characterized in that: The prior information includes historical test data, production parameters and material properties of the product.

4. The method for constructing a sequential sampling inspection model based on prior information according to claim 2, characterized in that: The preprocessing requires cleaning and testing of the data, identifying and processing missing values ​​and duplicate values ​​in the prior information data, and processing the prior information data.

5. The method for constructing a sequential sampling inspection model based on prior information according to claim 1, characterized in that: The S3 specifically includes the following steps: S3.1: Based on statistical analysis techniques, the influence of the prior information after grouping on the test target is quantified to determine the relationship between the prior information and the test target, and to provide a basis for the subsequent critical value; The calculation formula of the impact degree Y is: Y=E0+E1X1+E2X2+……+E n X n Among them, E0, E1, E2, ..., E n are regression coefficients, X1, X2, ..., X n is n sets of prior information; S3.2: Based on the prior information of the product, set an a priori qualified critical value H and input it into the sequential sampling inspection model.

6. The method for constructing a sequential sampling inspection model based on prior information according to claim 1, characterized in that: The sampling strategy includes the setting of sampling quantity, sampling interval and updating frequency parameters of prior information.

7. The method for constructing a sequential sampling inspection model based on prior information according to claim 1, characterized in that: The termination condition is that the impact degree Y≥the qualified critical value H.

8. The method for constructing a sequential sampling inspection model based on prior information according to claim 7, characterized in that: If the impact degree Y≥qualified critical value H, it means that the product meets the quality standards, that is, the product is qualified; if the impact degree Y<qualified critical value H, it means that the product does not meet the quality standards, that is, the product is unqualified.

Images