Equipment sampling inspection method and device, electronic equipment and storage medium
By clustering and training the detection model of equipment monitoring data, and determining the sampling targets, the problem of lack of targeted and costly equipment sampling in the existing technology is solved, and more efficient and accurate sampling results are achieved.
Patent Information
- Application Number
- CN202510240887.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-03
- Publication Date
- 2025-06-20
AI Technical Summary
In the prior art, the equipment sampling method is not targeted, resulting in unstable product quality representativeness and high cost.
By clustering the monitoring data of the processed equipment, the data are divided into qualified and unqualified, the monitoring data detection model is trained, and the sampling targets of the to-process equipment are determined.
It improves the direction of equipment sampling, increases the probability of inspection of unqualified products, and reduces the cost of sampling and labor costs.
Smart Images
Figure CN120180124A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data processing, and particularly to a method and device for randomly inspecting equipment, an electronic device, and a storage medium. Background Art
[0002] In industrial production, random inspection is an important link to ensure the product quality level and standardize the behavior of producers. In the prior art, operators usually randomly select a certain number of products from the completed products for inspection at regular intervals. However, this all-round and aimless random inspection method has unstable representativeness of the selected products for the quality of this batch of products, resulting in poor random inspection effects and also requiring labor costs. Summary of the Invention
[0003] The present invention provides a method and device for randomly inspecting equipment, an electronic device, and a storage medium to improve the quality of random inspection and reduce the cost of random inspection.
[0004] In a first aspect, an embodiment of the present invention provides a method for randomly inspecting equipment, the method including:
[0005] Dividing each piece of monitoring data into qualified monitoring data and unqualified monitoring data according to the clustering result of the monitoring data of the processed equipment;
[0006] Training a monitoring data detection model according to the qualified monitoring data and the unqualified monitoring data;
[0007] Inputting the monitoring data of the equipment to be processed into the monitoring data detection model to obtain the monitoring data detection result output by the monitoring data detection model;
[0008] Regarding the equipment to be processed with the monitoring data detection result being unqualified monitoring data as the target randomly inspected equipment, and determining the target randomly inspected equipment among the equipment to be processed with the monitoring data detection result being qualified monitoring data.
[0009] In a second aspect, an embodiment of the present invention further provides a device for randomly inspecting equipment, the device including:
[0010] A monitoring data division module, configured to divide each piece of monitoring data into qualified monitoring data and unqualified monitoring data according to the clustering result of the monitoring data of the processed equipment;
[0011] A monitoring data detection model training module, configured to train a monitoring data detection model according to the qualified monitoring data and the unqualified monitoring data;
[0012] A monitoring data detection result determination module, configured to input the monitoring data of the equipment to be processed into the monitoring data detection model to obtain the monitoring data detection result output by the monitoring data detection model;
[0013] The sampling inspection device determination module is used to take the device to be processed with unqualified monitoring data in the monitoring data detection result as the target sampling inspection device, and to determine the target sampling inspection device among the devices to be processed with qualified monitoring data in the monitoring data detection result.
[0014] In a third aspect, an embodiment of the present invention further provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the device sampling inspection method as described in any one of the embodiments of the present invention.
[0015] In a fourth aspect, an embodiment of the present invention further provides a storage medium storing computer-executable instructions, and the computer-executable instructions are used to execute the device sampling inspection method as described in any one of the embodiments of the present invention when executed by a computer processor.
[0016] The technical solution of the embodiment of the present invention performs clustering processing on the monitoring data of the processed devices, divides each monitoring data into qualified monitoring data and unqualified monitoring data, trains a monitoring data detection model according to the qualified monitoring data and unqualified monitoring data, inputs the monitoring data of the device to be processed into the monitoring data detection model, obtains the monitoring data detection result output by the monitoring data detection model, takes the device to be processed with unqualified monitoring data in the monitoring data detection result as the target sampling inspection device, and determines the target sampling inspection device among the devices to be processed with qualified monitoring data in the monitoring data detection result. The present invention improves the directivity of device sampling inspection, is more likely to detect unqualified products, thereby reversely urging the improvement of product quality, improves the sampling inspection quality, and reduces the sampling inspection cost.
[0017] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present invention, nor is it used to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. Description of the Drawings
[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.
[0019] Figure 1 is a flowchart of a device sampling inspection method provided in Embodiment 1 of the present invention;
[0020] Figure 2 is a schematic diagram of the distribution of different types of monitoring data in different clustering types provided in Embodiment 1 of the present invention;
[0021] Figure 3 It is a flowchart of a device sampling inspection method provided in the second embodiment of the present invention;
[0022] Figure 4 It is a schematic structural diagram of a device sampling inspection device provided in the third embodiment of the present invention;
[0023] Figure 5 It is a schematic structural diagram of an electronic device provided in the fourth embodiment of the present invention. Detailed implementation manners
[0024] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.
[0025] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily need to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units does not necessarily need to be limited to those clearly listed steps or units, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices. In the embodiments of the present application, some industry-existing solutions such as certain software, components, models, etc. may be mentioned, and they should be regarded as exemplary. The purpose is only to illustrate the feasibility in the implementation of the technical solution of the present application, but it does not mean that the applicant has already or necessarily used this solution.
[0026] In the technical solution of the present application, the acquisition, transmission, storage, use, processing, etc. of data all comply with the relevant regulations of national laws and regulations.
[0027] Embodiment 1
[0028] Figure 1 This is a flowchart of a device sampling inspection method provided in the first embodiment of the present invention. This embodiment is applicable to the situation of sampling inspection of products. This method can be executed by a device sampling inspection device, which can be implemented in the form of hardware and / or software, and the device sampling inspection device can be configured in an electronic device or a server.
[0029] As shown Figure 1 below, the method includes:
[0030] S110. Divide each piece of monitoring data into qualified monitoring data and unqualified monitoring data according to the clustering result of the monitoring data of the processed devices.
[0031] The processed devices refer to each device corresponding to the batches that have completed sampling inspection, or the devices that have been put into application, etc. That is, there is no need to conduct sampling inspection on the processed devices. The purpose of collecting the monitoring data of the processed devices in this embodiment is to train the monitoring data detection model.
[0032] The monitoring data may include preset parameter data and real-time collected parameter data. Taking the device type in this embodiment as an injection molding device as an example, the monitoring data may include residence time, nozzle temperature, barrel temperature, VP (Velocity / Pressure) pressure, maximum injection time, metering time, injection time, and minimum buffer, etc. The type and quantity of the monitoring data can be flexibly set according to the actual needs of the application scenario.
[0033] The clustering result refers to the result that after clustering the monitoring data of the processed devices, each piece of monitoring data is divided into different clustering types. In this embodiment, the monitoring data of the processed devices is clustered through a clustering algorithm to obtain at least two different clustering types. The clustering algorithm can be a supervised clustering algorithm and / or an unsupervised clustering algorithm. The number of clustering algorithms can be one or more. When there are multiple clustering algorithms, the monitoring data of the processed devices is clustered through different clustering algorithms respectively. Exemplarily, through clustering algorithms A, B, and C, the monitoring data of the processed devices is clustered in an unsupervised manner respectively. Among them, after being processed by clustering algorithm A, the clustering types are two categories, namely A1 and A2; after being processed by clustering algorithm B, the clustering types are three categories, namely B1, B2, and B3; after being processed by clustering algorithm C, the clustering types are two categories, namely C1 and C2.
[0034] Among them, labels can be added to the monitoring data, and the labels include qualified monitoring data and unqualified monitoring data. In this embodiment, it can be considered that the unqualified monitoring data is the monitoring data generated by unqualified devices, or the abnormal monitoring data generated by qualified devices; the qualified monitoring data is the normal monitoring data generated by qualified devices. It should be noted that the qualified monitoring data and unqualified monitoring data in this embodiment are only labels for the detection data, and do not represent the true and objective nature of the monitoring data.
[0035] In this embodiment, according to the clustering result of the monitoring data of the processed devices, each piece of monitoring data is divided into qualified monitoring data and unqualified monitoring data. The purpose is to use the qualified monitoring data and unqualified monitoring data as training data to train a monitoring data detection model, so as to detect the monitoring data of the newly generated devices to be processed through the monitoring data detection model, and determine whether the newly generated devices to be processed are qualified, so as to achieve targeted device sampling inspection.
[0036] In an alternative embodiment, according to the clustering result of the monitoring data of the processed devices, each piece of monitoring data is divided into qualified monitoring data and unqualified monitoring data. It can be to pre-determine the unqualified devices with labels and use their monitoring data as unqualified monitoring data. At the same time, perform supervised clustering processing on the monitoring data of all processed devices to obtain at least two clustering types; if it is determined that the ratio of the monitoring data of unqualified devices in the target clustering type to all the monitoring data corresponding to the target clustering type is greater than or equal to the preset threshold, then the monitoring data corresponding to the target clustering type is used as unqualified monitoring data. The other monitoring data in all the monitoring data except the above unqualified monitoring data is used as qualified monitoring data.
[0037] Furthermore, to ensure the proportional balance between the qualified monitoring data and the unqualified monitoring data, and at the same time, to ensure the training effect of the monitoring data detection model, it is necessary to preset the proportion of unqualified monitoring data, that is, the proportion of unqualified detection data in all the monitoring data. This proportion of unqualified monitoring data can be either a specific proportion value or a proportion range.
[0038] Based on the above embodiment, to ensure that the proportion of unqualified monitoring data meets the preset proportion, the intersection of the target clustering type and other clustering types can be taken, and the monitoring data in the intersection is used as unqualified monitoring data to reduce the number of unqualified monitoring data and ensure that the proportion of unqualified monitoring data meets the preset proportion.
[0039] In another alternative embodiment, according to the clustering result of the monitoring data of the processed devices, each piece of monitoring data is divided into qualified monitoring data and unqualified monitoring data. It can be to perform unsupervised clustering processing on the monitoring data of the processed devices to obtain at least two clustering types; analyze the distribution status of different types of monitoring data for each clustering type; take the clustering type with abnormal distribution status of at least one type of monitoring data as the target clustering type; and use the monitoring data corresponding to the target clustering type as unqualified monitoring data. The other monitoring data in all the monitoring data except the above unqualified monitoring data is used as qualified monitoring data.
[0040] Exemplarily, Figure 2A schematic diagram of the distribution of different types of monitoring data in different clustering types is provided. For example, Figure 2 as shown, blue represents clustering type 0, red represents clustering type 1, green represents clustering type 2, and the types of monitoring data include minimum buffer, metering time, VP pressure, and residence time. According to Figure 2 the distribution of different monitoring data in the three clustering types in
[0041] it can be seen that clustering type 2 (green) is significantly an outlier data. Therefore, clustering type 2 is used as the target clustering type, and then the unqualified monitoring data is determined.
[0042] Similarly, to ensure that the proportion of unqualified monitoring data meets the pre-set proportion, the intersection of the target clustering type and other clustering types can be taken, and the monitoring data in the intersection is used as the unqualified monitoring data. This embodiment will not repeat the above process.
[0043] In this embodiment, through the clustering process of the monitoring data, the division of qualified and unqualified monitoring data in the monitoring data is realized, so that the divided monitoring data can be used as the data basis for model training.
[0044] S120. According to the qualified monitoring data and the unqualified monitoring data, a monitoring data detection model is trained.
[0045] In this embodiment, the qualified monitoring data and the unqualified monitoring data can be used as sample data to optimize and train a pre-set neural network model. When the model converges, a trained monitoring data detection model is obtained.
[0046] Furthermore, S120 can further include: training a multi-layer perceptron neural network model according to the qualified monitoring data and the unqualified monitoring data to obtain a monitoring data detection model; wherein, the monitoring data detection model runs on an inference server.
[0047] Among them, the multi-layer perceptron neural network model refers to the MLP (Multilayer Perceptron), which is a feedforward artificial neural network model composed of multiple neuron layers and can model complex non-linear relationships.
[0048] In this embodiment, training based on the multi-layer perceptron neural network model has good applicability for the binary classification (qualified monitoring data or unqualified monitoring data) mode in this embodiment.
[0049] The inference server refers to a server based on the browser / server (Browser / Server) architecture.
[0050] In this embodiment, running the monitoring data detection model on the inference server, the logical processing, data storage, etc. of the monitoring data detection model are all completed on the server side, which has good maintainability and compatibility.
[0051] S130: Input the monitoring data of the device to be processed into the monitoring data detection model to obtain the monitoring data detection result output by the monitoring data detection model.
[0052] Among them, the device to be processed refers to each device in the batch that needs to be sampled. The monitoring data detection results output by the monitoring data detection model include two types: qualified monitoring data and unqualified monitoring data. In this embodiment, the monitoring data detection model processes the input monitoring data of the device to be processed, infers the qualified or unqualified labels of the monitoring data, and outputs the label results.
[0053] In this embodiment, before sampling each device in this batch, first input the monitoring data of each device to be processed into the monitoring data detection model, and determine the target sampling devices to be finally sampled among each device to be processed according to its monitoring data detection results. The advantage of this setting is that it makes the sampling more targeted, and the suspected unqualified devices are more likely to be sampled, thereby promoting the improvement of device quality in reverse.
[0054] S140: Use the device to be processed with the monitoring data detection result of unqualified monitoring data as the target sampling device, and determine the target sampling device among the devices to be processed with the monitoring data detection result of qualified monitoring data.
[0055] In this embodiment, the device to be processed with the monitoring data detection result of unqualified monitoring data is a suspected unqualified device, and it is measured to determine whether it is a truly objectively unqualified device. For the devices to be processed with the monitoring data detection result of qualified monitoring data, the conventional sampling method is still used for sampling to obtain the target sampling device.
[0056] Further, based on the sampling inspection ratio and the number of devices to be processed with unqualified monitoring data in the monitoring data inspection results, the target sampling inspection devices can be determined from the devices to be processed with qualified monitoring data in the monitoring data inspection results. Specifically, based on the total number of all devices to be processed and the sampling inspection ratio, the total number of target sampling inspection devices is determined; the difference between the total number of target sampling inspection devices and the first number of devices to be processed with unqualified monitoring data in the monitoring data inspection results is used as the second number; among the devices to be processed with qualified monitoring data in the monitoring data inspection results, the second number of devices to be processed is selected as the target sampling inspection devices. Among them, when selecting the second number of devices to be processed, conventional sampling inspection methods such as random selection can be adopted.
[0057] The technical solution of this embodiment conducts sampling inspection on the devices to be processed based on the monitoring data inspection results, realizes targeted sampling inspection, is more likely to detect unqualified devices, improves the directivity of extracting suspected unqualified devices, thereby promoting the improvement of device quality in reverse and improving the sampling inspection effect.
[0058] Further, after S140, it further includes:
[0059] S1. Determine the sampling inspection results of each target sampling inspection device;
[0060] S2. Use the monitoring data of the target sampling inspection devices with unqualified sampling inspection results as unqualified monitoring data to retrain the monitoring data inspection model.
[0061] Among them, the sampling inspection results of the target sampling inspection devices are real and objective results, which can include qualified devices and unqualified devices. Specifically, the sampling inspection results can be obtained through ORT testing (Ongoing Reliability Test, product reliability test), and the specific items of ORT testing in this embodiment are not limited.
[0062] In this embodiment, for the target sampling inspection devices with unqualified sampling inspection results, their detection data is used as unqualified detection data and incorporated into the unqualified monitoring data in S120 to form a new training data set, and the monitoring data inspection model is retrained.
[0063] It can be understood that since the unqualified monitoring data and qualified monitoring data in S120 are obtained by reasoning based on the clustering processing results of the monitoring data of the processed devices, and the monitoring data of the target sampling inspection devices with unqualified sampling inspection results are real and objective unqualified monitoring data, therefore, in this embodiment, the monitoring data of the target sampling inspection devices with unqualified sampling inspection results is incorporated into the unqualified monitoring data and the monitoring data inspection model is retrained, which can improve the accuracy of the monitoring data inspection model.
[0064] In this embodiment, when a non-conforming device is randomly inspected, the monitoring data of the non-conforming device can be incorporated into the non-conforming monitoring data, and the monitoring data detection model can be retrained; alternatively, at preset time intervals, the monitoring data of the non-conforming devices detected within this time interval can be incorporated into the non-conforming monitoring data, and the monitoring data detection model can be retrained. Or, every time a preset number of non-conforming devices are detected, their monitoring data can be incorporated into the non-conforming monitoring data, and the monitoring data detection model can be retrained; alternatively, in response to a retraining instruction for the monitoring data detection model on the user interface, the monitoring data of the non-conforming devices that have not been processed before receiving the retraining instruction for the monitoring data detection model can be incorporated into the non-conforming monitoring data, and the monitoring data detection model can be retrained.
[0065] Further, S2 can further include:
[0066] S20. Determine the first quantity of the devices to be processed whose monitoring data detection results output by the monitoring data detection model are non-conforming monitoring data, and, after using the devices to be processed whose monitoring data detection results are non-conforming monitoring data as the target randomly inspected devices, determine the second quantity of the target randomly inspected devices whose random inspection results are non-conforming devices;
[0067] S21. Use the ratio of the second quantity to the first quantity as the accuracy rate of the monitoring data detection model;
[0068] S22. Use the monitoring data of the target randomly inspected devices whose random inspection results are non-conforming devices as non-conforming monitoring data, and retrain the monitoring data detection model until the accuracy rate of the monitoring data detection model is greater than or equal to a preset accuracy rate threshold.
[0069] Based on the above retraining of the monitoring data detection model, this embodiment also provides a judgment condition for stopping the retraining of the monitoring data detection model.
[0070] It can be understood that as the monitoring data of non-conforming devices is continuously incorporated into the non-conforming monitoring data and the monitoring data detection model is retrained, the accuracy rate of the monitoring data detection model will increase. However, to avoid overtraining of the monitoring data detection model and save computing resources and costs, the retraining of the monitoring data detection model can be stopped when the accuracy rate of the monitoring data detection model reaches a certain threshold.
[0071] Specifically, the first quantity of the devices to be processed with unqualified monitoring data detection results is the number of devices inferred as unqualified by the monitored data detection model. Among the devices inferred as unqualified by the monitored data detection model, the second quantity of the target sampled devices with unqualified sampling results is the number of devices that are truly objectively unqualified. Therefore, the value obtained by dividing the second quantity by the first quantity is the accuracy rate of the monitored data detection model.
[0072] In this embodiment, the monitored data detection model is continuously retrained with the monitoring data of devices that are truly objectively unqualified to improve the accuracy rate of the monitored data detection model, thereby improving the sampling efficiency. When the accuracy rate of the monitored data detection model reaches a certain threshold, the retraining of the monitored data detection model is stopped to avoid overtraining of the monitored data detection model and save computing resources and costs.
[0073] The technical solution of the embodiment of the present invention performs clustering processing on the monitoring data of the processed devices, divides each monitoring data into qualified monitoring data and unqualified monitoring data, trains a monitored data detection model based on the qualified monitoring data and unqualified monitoring data, inputs the monitoring data of the devices to be processed into the monitored data detection model, obtains the monitoring data detection result output by the monitored data detection model, uses the devices to be processed with unqualified monitoring data detection results as the target sampled devices, and determines the target sampled devices among the devices to be processed with qualified monitoring data detection results. The present invention improves the directivity of device sampling, is more likely to detect unqualified products, thereby reversely urging the improvement of product quality, improves the sampling quality, and reduces the sampling cost.
[0074] Embodiment 2
[0075] Figure 3 FIG. is a flowchart of a device sampling method provided by the second embodiment of the present invention. On the basis of the above embodiment, the present embodiment further specifies the process of dividing the monitoring data and the process of training the monitored data detection model.
[0076] As Figure 3 shown, the method includes:
[0077] S210. Perform clustering processing on the monitoring data of the processed devices according to at least two clustering algorithms to obtain at least two types of monitoring data.
[0078] This embodiment describes the process of dividing each monitoring data into qualified monitoring data and unqualified monitoring data based on the clustering processing results of multiple clustering algorithms.
[0079] Specifically, the clustering algorithm referred to in this embodiment is an unsupervised clustering algorithm. In the case where the monitored data of the processed device has no labels and prior knowledge, the monitored data is divided into different clusters according to the characteristics and similarities of the monitored data itself, that is, different types of monitored data.
[0080] The types and quantities of the monitored data types obtained by different clustering algorithms may be different. Exemplarily, after being processed by clustering algorithm A, the clustering types are two types, namely A1 and A2; after being processed by clustering algorithm B, the clustering types are three types, namely B1, B2, and B3; after being processed by clustering algorithm C, the clustering types are two types, namely C1 and C2.
[0081] In this embodiment, by using multiple unsupervised clustering algorithms to process the monitored data of the processed device and combining the subsequent analysis of the monitored data distribution of the monitored data types, the abnormal monitored data types can be effectively and accurately located, thereby realizing the accurate division of qualified monitored data and unqualified monitored data.
[0082] S220. Determine the abnormal monitored data types among the various monitored data types.
[0083] In this embodiment, the abnormal monitored data types can be determined by performing multi-dimensional analysis on the monitored data distribution of the various monitored data types. Taking Figure 2 as an example, clustering type 2 (green) is obviously an outlier data, so clustering type 2 is the abnormal monitored data type.
[0084] S230. Divide the various monitored data into qualified monitored data and unqualified monitored data according to the abnormal monitored data types.
[0085] In an optional embodiment, the monitored data corresponding to the abnormal monitored data types can be directly used as unqualified monitored data, and the other monitored data except the unqualified monitored data can be used as qualified monitored data.
[0086] Further, S230 can further include: dividing the various monitored data into qualified monitored data and unqualified monitored data according to the monitored data corresponding to at least one abnormal monitored data type and the preset unqualified monitored data ratio.
[0087] In this embodiment, the unqualified monitored data ratio is preset to avoid the influence of excessive unqualified monitored data on the training of the monitored data detection model, thereby avoiding the misdetection of unqualified monitored data in the future.
[0088] In an alternative embodiment, if there is one type of abnormal monitoring data, and the proportion of each piece of monitoring data corresponding to this type of abnormal monitoring data in all the monitoring data is less than or equal to the proportion of unqualified monitoring data, then each piece of monitoring data corresponding to this type of abnormal monitoring data can be directly used as unqualified monitoring data.
[0089] If there is one type of abnormal monitoring data, and the proportion of each piece of monitoring data corresponding to this type of abnormal monitoring data in all the monitoring data is greater than the proportion of unqualified monitoring data, then the processed devices corresponding to this type of abnormal monitoring data can be intersected with the processed devices corresponding to other types of monitoring data. Among the processed devices corresponding to this type of abnormal monitoring data, the processed devices other than those in the intersection are used as abnormal processed devices. Each piece of monitoring data corresponding to the abnormal processed devices is used as unqualified monitoring data.
[0090] In another alternative embodiment, if there are at least two types of abnormal monitoring data, and the sum of the monitoring data corresponding to these types of abnormal monitoring data accounts for a proportion of all the monitoring data that is less than or equal to the proportion of unqualified monitoring data, then each piece of monitoring data corresponding to these types of abnormal monitoring data can be directly used as unqualified monitoring data.
[0091] Furthermore, the monitoring data corresponding to at least two types of abnormal monitoring data are intersected and / or unioned to obtain unqualified monitoring data; among them, the proportion of the unqualified monitoring data in all the monitoring data matches the pre-set proportion of unqualified monitoring data; the other monitoring data except the unqualified monitoring data are used as qualified monitoring data.
[0092] In this embodiment, if the number of pieces of monitoring data corresponding to each type of abnormal monitoring data is large, then at least one intersection and / or union operation can be performed on the monitoring data corresponding to different types of abnormal monitoring data until the proportion of the monitoring data obtained after processing in all the monitoring data matches the pre-set proportion of unqualified monitoring data.
[0093] Specifically, whether to perform an intersection operation or a union operation on the monitoring data corresponding to every two types of abnormal monitoring data, and how many intersection and / or union operations to perform, can be flexibly set according to the amount of data of the monitoring data corresponding to each type of abnormal monitoring data and the requirement of the proportion of unqualified monitoring data.
[0094] Exemplarily, taking the above clustering process using clustering algorithms A, B, and C as an example, if the types of abnormal monitoring data are A1, B2, and C2, and there is a relatively large amount of monitoring data corresponding to types A1 and B2, while a relatively small amount of monitoring data corresponding to type C2, then the intersection processing can be first performed on the monitoring data corresponding to types A1 and B2, and then the union processing is performed on the obtained intersection and the monitoring data corresponding to type C2 to obtain the final unqualified monitoring data.
[0095] S240. Train a multi-layer perceptron neural network model based on the qualified monitoring data and the unqualified monitoring data to obtain a monitoring data detection model.
[0096] Among them, the monitoring data detection model runs on an inference server.
[0097] S250. Input the monitoring data of the device to be processed into the monitoring data detection model to obtain the monitoring data detection result output by the monitoring data detection model.
[0098] S260. Take the device to be processed with the monitoring data detection result being unqualified monitoring data as the target sampling inspection device, and determine the target sampling inspection device among the devices to be processed with the monitoring data detection result being qualified monitoring data.
[0099] The specific process of extracting the target sampling inspection device according to the monitoring data detection result of the monitoring data detection model has been described in the above embodiments, and will not be repeated here in this embodiment.
[0100] S270. Determine the sampling inspection results of each target sampling inspection device.
[0101] S280. Take the monitoring data of the target sampling inspection device with the sampling inspection result being an unqualified device as the unqualified monitoring data, and retrain the monitoring data detection model.
[0102] The specific process of retraining the monitoring data detection model has been described in the above embodiments, and will not be repeated here in this embodiment.
[0103] The technical solution of this embodiment clusters the monitoring data of the processed devices through different clustering algorithms to obtain various types of monitoring data clusters. By performing data analysis on the distribution of the monitoring data within the same type of monitoring data, abnormal monitoring data types are identified. Through the intersection and / or union processing of multiple abnormal monitoring data types, the final unqualified monitoring data is obtained. This realizes the effective and accurate division of monitoring data, thereby providing a data basis for the training of the monitoring data detection model. When sampling inspection is required, the monitoring data of the device to be processed is input into the monitoring data detection model. Based on the detection result of the monitoring data detection model, the device to be processed with the monitoring data detection result of unqualified monitoring data is used as the target sampling inspection device, and, the target sampling inspection device is determined among the devices to be processed with the monitoring data detection result of qualified monitoring data. This realizes the targeted sampling inspection of devices, making it easier to detect suspected unqualified devices, thereby improving the sampling inspection effect and promoting the improvement of device quality in reverse. By incorporating the monitoring data of the devices with unqualified sampling inspection results into the unqualified monitoring data, the monitoring data detection model is retrained to improve the accuracy of the monitoring data detection model, thereby improving the sampling inspection efficiency. When the accuracy of the monitoring data detection model reaches a certain threshold, the retraining of the monitoring data detection model is stopped to avoid overtraining of the monitoring data detection model and save computing resources and costs.
[0104] Embodiment III
[0105] Figure 4 It is a schematic structural diagram of a device sampling inspection device provided in Embodiment III of the present invention. As Figure 4 shown, the device includes:
[0106] A monitoring data division module 310, configured to divide each monitoring data into qualified monitoring data and unqualified monitoring data according to the clustering processing result of the monitoring data of the processed device;
[0107] A monitoring data detection model training module 320, configured to train a monitoring data detection model according to the qualified monitoring data and the unqualified monitoring data;
[0108] A monitoring data detection result determination module 330, configured to input the monitoring data of the device to be processed into the monitoring data detection model to obtain the monitoring data detection result output by the monitoring data detection model;
[0109] A sampling inspection device determination module 340, configured to use the device to be processed with the monitoring data detection result of unqualified monitoring data as the target sampling inspection device, and, determine the target sampling inspection device among the devices to be processed with the monitoring data detection result of qualified monitoring data.
[0110] In the technical solution of the embodiment of the present invention, by clustering the monitoring data of the processed device, each piece of monitoring data is divided into qualified monitoring data and unqualified monitoring data. According to the qualified monitoring data and unqualified monitoring data, a monitoring data detection model is trained. The monitoring data of the device to be processed is input into the monitoring data detection model, and the monitoring data detection result output by the monitoring data detection model is obtained. The device to be processed with the monitoring data detection result being unqualified monitoring data is used as the target sampling inspection device, and, a target sampling inspection device is determined among the devices to be processed with the monitoring data detection result being qualified monitoring data. The present invention improves the directivity of device sampling inspection, is more likely to detect unqualified products, thereby reversely urging the improvement of product quality, improves the sampling inspection quality, and reduces the sampling inspection cost.
[0111] Based on the above embodiment, optionally, the monitoring data division module 310 includes:
[0112] A monitoring data clustering unit, configured to cluster the monitoring data of the processed device according to at least two clustering algorithms to obtain at least two types of monitoring data;
[0113] An abnormal monitoring data type determination unit, configured to determine an abnormal monitoring data type among each type of monitoring data;
[0114] A monitoring data division unit, configured to divide each piece of monitoring data into qualified monitoring data and unqualified monitoring data according to the abnormal monitoring data type.
[0115] Based on the above embodiment, optionally, the monitoring data division unit is specifically configured to:
[0116] Divide each piece of monitoring data into qualified monitoring data and unqualified monitoring data according to each piece of monitoring data corresponding to at least one abnormal monitoring data type and the pre-set unqualified monitoring data ratio.
[0117] Based on the above embodiment, optionally, the monitoring data division unit is specifically configured to:
[0118] Perform an intersection operation and / or a union operation on the monitoring data corresponding to at least two abnormal monitoring data types to obtain unqualified monitoring data;
[0119] wherein, the ratio of the unqualified monitoring data to all the monitoring data matches the pre-set unqualified monitoring data ratio;
[0120] Use the other monitoring data except the unqualified monitoring data as the qualified monitoring data.
[0121] Based on the above embodiment, optionally, the monitoring data detection model training module 320 includes:
[0122] A monitoring data detection model training unit, which is used to train a multi-layer perceptron neural network model based on qualified monitoring data and unqualified monitoring data to obtain a monitoring data detection model;
[0123] Among them, the monitoring data detection model runs on an inference server.
[0124] Based on the above embodiments, optionally, the device further includes:
[0125] A sampling inspection result determination module, which is used to determine the sampling inspection results of each target sampling inspection device according to the monitoring data of each target sampling inspection device;
[0126] A monitoring data detection model retraining module, which is used to use the monitoring data of the target sampling inspection device with an unqualified sampling inspection result as unqualified monitoring data to retrain the monitoring data detection model.
[0127] Based on the above embodiments, optionally, the monitoring data detection model retraining module includes:
[0128] A quantity determination unit, which is used to determine the first quantity of the devices to be processed with unqualified monitoring data detection results output by the monitoring data detection model, and the second quantity of the target sampling inspection devices with unqualified sampling inspection results after using the devices to be processed with unqualified monitoring data detection results as target sampling inspection devices;
[0129] A monitoring data detection model accuracy determination unit, which is used to use the ratio of the second quantity to the first quantity as the accuracy of the monitoring data detection model;
[0130] A monitoring data detection model retraining unit, which is used to use the monitoring data of the target sampling inspection device with an unqualified sampling inspection result as unqualified monitoring data to retrain the monitoring data detection model until the accuracy of the monitoring data detection model is greater than or equal to a preset accuracy threshold.
[0131] The device sampling inspection device provided by the embodiments of the present invention can execute the device sampling inspection method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the method.
[0132] Embodiment 4
[0133] Figure 5The structural schematic diagram of an electronic device 10 that can be used to implement the embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.
[0134] As Figure 5 shown, the electronic device 10 includes at least one processor 11, and a memory communicatively connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc. The memory stores a computer program executable by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. The input / output (I / O) interface 15 is also connected to the bus 14.
[0135] Multiple components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disc, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.
[0136] The processor 11 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 11 include but are not limited to a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the device sampling inspection method.
[0137] In some embodiments, the device spot-checking method may be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed onto the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the device spot-checking method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to execute the device spot-checking method by any other suitable means (e.g., by means of firmware).
[0138] The various embodiments of the systems and techniques described above in this document can be implemented in digital electronic circuitry, integrated circuit systems, field-programmable gate arrays (FPGA), application-specific integrated circuits (ASIC), application-specific standard products (ASSP), systems-on-chip (SOC), complex programmable logic devices (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special-purpose or general-purpose programmable processor that receives data and instructions from a storage system, at least one input device, and at least one output device, and transmits the data and instructions to the storage system, the at least one input device, and the at least one output device.
[0139] The computer program for implementing the method of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to the processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowchart and / or block diagram are implemented. The computer program can be executed entirely on the machine, partially on the machine, as a stand-alone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.
[0140] In the context of the present invention, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium can be a machine-readable signal medium. More specific examples of the machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
[0141] To provide for interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can also be used to provide for interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic, speech, or tactile input).
[0142] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.
[0143] A computing system may include a client and a server. The client and the server are generally far from each other and usually interact via a communication network. The client-server relationship is created by computer programs running on respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system, solving the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS services.
[0144] It should be understood that various forms of processes shown above can be used, steps can be reordered, added or deleted. For example, the steps recited in the present invention can be executed in parallel, sequentially or in a different order, as long as the desired results of the technical solution of the present invention can be achieved, and no limitation is made herein.
[0145] The above specific embodiments do not constitute a limitation on the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.
Claims
1. A device sampling inspection method, characterized in that: include: According to the cluster processing result of the monitoring data of the processed equipment, each monitoring data is divided into qualified monitoring data and unqualified monitoring data; According to qualified monitoring data and unqualified monitoring data, a monitoring data detection model is trained; Input the monitoring data of the equipment to be processed into the monitoring data detection model to obtain the monitoring data detection results output by the monitoring data detection model; The equipment to be processed whose monitoring data detection result is unqualified monitoring data is taken as the target sampling equipment, and the target sampling equipment is determined from the equipment to be processed whose monitoring data detection result is qualified monitoring data.
2. The method according to claim 1, characterized in that According to the clustering processing results of the monitoring data of the processed equipment, each monitoring data is divided into qualified monitoring data and unqualified monitoring data, including: Performing clustering processing on the monitoring data of the processed equipment according to at least two clustering algorithms to obtain at least two types of monitoring data; Determine abnormal monitoring data types among various monitoring data types; According to the type of abnormal monitoring data, each monitoring data is divided into qualified monitoring data and unqualified monitoring data.
3. The method according to claim 2, characterized in that According to the type of abnormal monitoring data, each monitoring data is divided into qualified monitoring data and unqualified monitoring data, including: According to each monitoring data corresponding to at least one abnormal monitoring data type and a preset unqualified monitoring data ratio, each monitoring data is divided into qualified monitoring data and unqualified monitoring data.
4. The method according to claim 3, characterized in that According to each monitoring data corresponding to at least one abnormal monitoring data type and a preset unqualified monitoring data ratio, each monitoring data is divided into qualified monitoring data and unqualified monitoring data, including: Performing intersection processing and / or union processing on monitoring data corresponding to at least two abnormal monitoring data types to obtain unqualified monitoring data; Among them, the proportion of unqualified monitoring data to all monitoring data matches the pre-set proportion of unqualified monitoring data; All monitoring data except unqualified monitoring data shall be regarded as qualified monitoring data.
5. The method according to claim 1, characterized in that Based on qualified monitoring data and unqualified monitoring data, a monitoring data detection model is trained, including: According to qualified monitoring data and unqualified monitoring data, the multi-layer perceptron neural network model is trained to obtain a monitoring data detection model; Wherein, the monitoring data detection model runs on the inference server.
6. The method according to claim 1, characterized in that After the equipment to be processed whose monitoring data detection result is unqualified monitoring data is used as the target sampling equipment, and the target sampling equipment is determined from the equipment to be processed whose monitoring data detection result is qualified monitoring data, the method further includes: Determine the sampling results of each target sampling equipment; The monitoring data of the target inspection equipment whose inspection results are unqualified equipment is used as unqualified monitoring data, and the monitoring data detection model is re-trained.
7. The method according to claim 6, characterized in that The monitoring data of the target sampling equipment with the sampling results as unqualified equipment is used as unqualified monitoring data, and the monitoring data detection model is retrained, including: Determine a first number of devices to be processed whose monitoring data detection results output by the monitoring data detection model are unqualified monitoring data, and a second number of target sampling devices whose sampling results are unqualified devices after taking the devices to be processed whose monitoring data detection results are unqualified monitoring data as target sampling devices; The ratio of the second quantity to the first quantity is used as the accuracy of the monitoring data detection model; The monitoring data of the target sampling equipment whose sampling results are unqualified equipment is used as unqualified monitoring data, and the monitoring data detection model is retrained until the accuracy of the monitoring data detection model is greater than or equal to the preset accuracy threshold.
8. A device for sampling equipment, characterized in that: include: A monitoring data division module, used to divide each monitoring data into qualified monitoring data and unqualified monitoring data according to the cluster processing result of the monitoring data of the processed equipment; A monitoring data detection model training module is used to train a monitoring data detection model based on qualified monitoring data and unqualified monitoring data; A monitoring data detection result determination module is used to input the monitoring data of the device to be processed into the monitoring data detection model to obtain the monitoring data detection result output by the monitoring data detection model; The sampling device determination module is used to take the devices to be processed whose monitoring data detection results are unqualified monitoring data as target sampling devices, and to determine the target sampling devices among the devices to be processed whose monitoring data detection results are qualified monitoring data.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the device sampling inspection method as described in any one of claims 1-7 is implemented.
10. A storage medium storing computer executable instructions, characterized in that: The computer executable instructions are used to execute the device sampling inspection method as described in any one of claims 1-7 when executed by a computer processor.