Product parameter statistical analysis method and system

Abstract

The invention provides a product parameter statistical analysis method and system. The method comprises the following steps: collecting measured values of product parameters of a tested product under each production process; the measured value of the product parameter comprises a measured value of an output signal of the tested product and a measured value of a device parameter of each customized component in the tested product; performing classified statistics on the measured values of the product parameters to obtain a device sub-parameter data set of the customized component and an output sub-parameter data set of the tested product; performing influence coefficient analysis on the device sub-parameter data set of the customized component and the output sub-parameter data set of the tested product to obtain an influence coefficient of the device sub-parameter of each customized component on the output sub-parameter of the tested product; and according to each influence coefficient, an adjustment scheme of the tested product is determined, and the device sub-parameters of the customized components are rapidly and accurately adjusted, so that the replacement frequency in the production process is reduced, the production efficiency is improved, and the production cost is saved.

Description

Technical Field

[0001] The present invention belongs to the field of data technology, and more specifically, relates to a product parameter statistical analysis method and system. Background Art

[0002] Due to functional requirements, several customized components are added to the LEU circuit design. The customized components include transformers and current transformers. The transformer has a decisive influence on the key parameters of the product output signal. The key parameters of the transformer include inductance, leakage inductance and turns ratio. The product output signal includes C1 signal and C6 signal. The key parameters of the product output signal include the amplitude of the C1 signal, the eye diagram parameters of the C1 signal, the amplitude of the C6 signal, the frequency of the C6 signal, the harmonic amount of the C6 signal and the output short-circuit critical value. The C1 signal is a square wave signal and the C6 signal is a sine wave signal. Due to the large difference in the same parameter between customized components, the consistency of the key parameters of the product output signal is unstable, and even does not meet the performance requirements. It is necessary to replace customized components to ensure product performance. The existing technology for replacing customized components is to manually analyze each parameter to replace customized components. This method has high labor costs, is limited by manual experience, and cannot quickly and accurately determine the adjustment plan. Summary of the invention

[0003] In view of this, the object of the present invention is to provide a product parameter statistical analysis method and system for quickly and accurately adjusting the device sub-parameters of customized components to reduce the number of replacements in the production process, improve production efficiency, and save production costs.

[0004] The first aspect of the present application discloses a product parameter statistical analysis method, comprising:

[0005] Collecting the measured values ​​of product parameters of the tested product in each production process; the measured values ​​of the product parameters include the measured values ​​of the output signal of the tested product and the measured values ​​of the device parameters of each customized component in the tested product; the output signal includes at least one output sub-parameter, and the device parameters of each customized component include at least one device sub-parameter;

[0006] Classifying and counting the measured values ​​of the product parameters to obtain a device sub-parameter data set of the customized components and an output sub-parameter data set of the tested product;

[0007] Performing influence coefficient analysis on the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product to obtain the influence coefficient of the device sub-parameter of each customized component on the output sub-parameter of the tested product;

[0008] An adjustment plan for the tested product is determined based on each of the influence coefficients.

[0009] Optionally, the classified statistics of the measured values ​​of the product parameters to obtain the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product include:

[0010] Taking each of the device sub-parameters and each of the output sub-parameters as statistical targets, classifying and counting the measured values ​​of each of the device parameters and each of the output sub-parameters;

[0011] taking a set of each measured value of the device sub-parameter as a data set of the corresponding device sub-parameter;

[0012] The set of each measured value of the output sub-parameter is used as the data set of the corresponding output sub-parameter.

[0013] Optionally, the performing of influence coefficient analysis on the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product to obtain the influence coefficient of the device sub-parameter of each customized component on the output sub-parameter of the tested product includes:

[0014] For any of the device sub-parameter data sets and any of the output sub-parameter data sets: determine the dissimilarity between the corresponding device sub-parameter data set and the corresponding output sub-parameter data set using a K-means algorithm, and use the dissimilarity as an influence coefficient between the corresponding device sub-parameter and the corresponding output sub-parameter.

[0015] Optionally, determining the dissimilarity between the corresponding device sub-parameter data set and the corresponding output sub-parameter data set by using a K-means algorithm includes:

[0016] Determine the difference between the measured value of the device sub-parameter and the measured value of the output sub-parameter collected in the same measurement batch, and perform square algorithm processing on the difference to obtain a first value of the corresponding measurement batch;

[0017] The first values ​​of different measurement batches are summed and square rooted to obtain the dissimilarity between the device sub-parameter data set and the output sub-parameter data set.

[0018] Optionally, determining an adjustment plan for the tested product according to each of the influence coefficients includes:

[0019] Determining, based on the influence coefficient and the mapping relationship between the influence coefficient and the adjustment scheme, an adjustment scheme for corresponding device sub-parameters of the customized component corresponding to the influence coefficient;

[0020] The adjustment schemes of the device sub-parameters of the customized components are summarized as the adjustment scheme of the product under test.

[0021] Optionally, after classifying and counting the measured values ​​of the product parameters to obtain the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product, the method further includes:

[0022] Linear regression processing is performed on each of the device sub-parameter data sets and each of the output sub-parameter data sets to obtain the change trends between each of the device sub-parameters and the output sub-parameters.

[0023] Optionally, after performing influence coefficient analysis on the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product to obtain the influence coefficient of the device sub-parameter of each customized component on the output sub-parameter of the tested product, the method further includes:

[0024] The influence coefficient of the device sub-parameter of each of the customized components on the output sub-parameter of the tested product is displayed to the system user or production management personnel.

[0025] Optionally, before classifying and counting the measured values ​​of the product parameters to obtain the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product, the method further includes:

[0026] The measured values ​​of the product parameters are preprocessed.

[0027] The second aspect of the present application discloses a product parameter statistical analysis system, comprising:

[0028] The terminal module is used to collect the measured values ​​of the product parameters of the tested product in each production process; the measured values ​​of the product parameters include the measured values ​​of the output signal of the tested product and the measured values ​​of the device parameters of each customized component in the tested product; the output signal includes at least one output sub-parameter, and the device parameter of each customized component includes at least one device sub-parameter;

[0029] An integration module, used for classifying and counting the measured values ​​of the product parameters to obtain a device sub-parameter data set of the customized components and an output sub-parameter data set of the tested product;

[0030] An analysis module, used to perform influence coefficient analysis on the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product, to obtain the influence coefficient of the device sub-parameter of each customized component on the output sub-parameter of the tested product;

[0031] The countermeasure module is used to determine the adjustment plan of the tested product according to each of the influence coefficients.

[0032] Optionally, applied to the cyber-physical system (CPS) architecture; where:

[0033] The terminal module is an actuator in the CPS architecture;

[0034] The integration module is a manufacturing execution EMS system in the CPS architecture;

[0035] The analysis module is a full life cycle management PLM system in the CPS architecture;

[0036] The countermeasure module is an integrated platform in the CPS architecture.

[0037] It can be seen from the above technical scheme that a product parameter statistical analysis method provided by the present invention includes: collecting the measured values ​​of the product parameters of the tested product under each production process; the measured values ​​of the product parameters include the measured values ​​of the output signal of the tested product, and the measured values ​​of the device parameters of each customized component in the tested product; the output signal includes at least one output sub-parameter, and the device parameter of each customized component includes at least one device sub-parameter; classifying and counting the measured values ​​of the product parameters to obtain the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product; classifying and counting the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product The output sub-parameter data set of the product is analyzed for influence coefficients to obtain the influence coefficient of the device sub-parameters of each customized component on the output sub-parameters of the product under test; the adjustment plan for the product under test is determined based on each influence coefficient; the influence relationship between the device sub-parameters of the customized components and the output sub-parameters of the product under test is obtained through data statistics and analysis, and the adjustment plan for the product under test is further determined, which is convenient for guiding the adjustment of the device sub-parameters of the customized components. No manual analysis is required, and unified analysis and adjustment standards are provided to achieve fast and accurate adjustment of the device sub-parameters of the customized components, so as to reduce the number of replacements in the production process, improve production efficiency, and save production costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0039] Figure 1 is a flow chart of a product parameter statistical analysis method provided by an embodiment of the present invention;

[0040] Figure 2 It is a schematic diagram of physical data storage involved in a product parameter statistical analysis method provided by an embodiment of the present invention;

[0041] Figure 3 It is a flowchart of multi-dimensional viewing involved in a product parameter statistical analysis method provided by an embodiment of the present invention;

[0042] Figure 4 and Figure 5 is a schematic diagram of a star model involved in a product parameter statistical analysis method provided in an embodiment of the present invention;

[0043] Figure 6 is a schematic diagram of a regression line involved in a product parameter statistical analysis method provided by an embodiment of the present invention;

[0044] Figure 7 is a schematic diagram of a product parameter statistical analysis system provided by an embodiment of the present invention;

[0045] Figure 8 is a schematic diagram of a CPS architecture provided by an embodiment of the present invention;

[0046] Fig. 9 is a schematic diagram of a product parameter statistical analysis system provided by an embodiment of the present invention applied to a CPS architecture;

[0047] Fig.10 It is a query schematic diagram involved in a product parameter statistical analysis system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] In this application, the terms "comprises", "comprising" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element.

[0050] Explanation of relevant terms:

[0051] Line-side Electric Unit (LEU) is a part of the train control system in the railway communication system. It is used to receive train control messages and transmit them to the transponder inside the rails.

[0052] CPS: Cyber ​​physical system, or information-physical system, is the product of the integration of traditional automation control systems and new information technology. It is the necessary foundation for realizing intelligent manufacturing in the industrial field. Intelligent manufacturing is realized through the deep integration of information technology and traditional manufacturing. The core of the integration is CPS, which can connect the virtual digital world and the displayed physical world. It is the key technology of intelligent manufacturing. The key technologies for realizing CPS include industrial manufacturing intelligent technology, industrial software intelligent technology, data integration and analysis technology, high-performance network communication technology and security protection technology. This application uses CPS's high-performance network communication technology to realize product performance monitoring.

[0053] FCS: Centralized control system.

[0054] SCADA: Data monitoring system and acquisition system.

[0055] DCS: Distributed Control System.

[0056] MES: Manufacturing Execution System.

[0057] SCM: Enterprise supply chain management.

[0058] ERP: Enterprise Resource Planning.

[0059] PLM: Full life cycle management.

[0060] Data warehouse: Data warehouse is a strategic collection of all types of data support for decision-making processes at all levels of the enterprise. It is a single data store created for analytical reporting and decision support purposes. It provides guidance for business process improvement, monitoring time, cost, quality and control for enterprises that need business intelligence.

[0061] ROLAP technology: Online Analytical Processing, online analytical processing technology, is one of the main analytical tools of the data warehouse system. Its core concept is multidimensional model, focusing on data analysis. Its main use is to allow users to obtain useful data from various angles, whether it is slicing, dicing or drilling multidimensional data. There are two types of OLAP classification: one is that the technology has a multidimensional framework, which can temporarily store data in it, which is equivalent to storing it in the client-server database; the other is that the data is extracted from the relational database and the format is converted at the same time, that is, relational OLAP (usually referred to as ROLAP technology).

[0062] Data mining: Data mining is an important intelligent tool to support decision making and has developed rapidly in the field of data information mining. Data mining is an in-depth study of data information, aiming to discover unknown additional information from the data. From a scientific research perspective, this is a relatively new discipline based on disciplines such as computer science and statistics.

[0063] The embodiment of the present application provides a product parameter statistical analysis method for solving the problems of the prior art, such as high labor cost, limitation of manual experience, and inability to quickly and accurately determine adjustment plans.

[0064] See also Figure 1 , the product parameter statistical analysis method includes:

[0065] S101. Collect measured values ​​of product parameters of the tested product in each production process.

[0066] The measured values ​​of product parameters include: the measured values ​​of the output signal of the product under test, and the measured values ​​of the device parameters of each customized component in the product under test. The output signal includes at least one output sub-parameter, and the device parameters of each customized component include at least one device sub-parameter. For example, the device sub-parameters of a customized component transformer include inductance, leakage inductance and turns ratio, and the output sub-parameters include C1 signal amplitude, C6 signal amplitude, C6 signal frequency, C6 signal harmonics and output short-circuit critical value. Among them, the C1 signal is a square wave signal, which is used to carry the output message information of the LEU product, and the C6 signal is a sine wave signal, which is an energy signal and is mainly used for long-distance transmission of C1 signal-carried information.

[0067] Customized components can be transformers and current transformers, etc. Different customized components have different device parameters. For example, taking a transformer as an example, device parameters may include temperature, leakage inductance, etc. The temperature measurement values ​​of the transformer may be different in different processes, and the temperature measurement values ​​may also be different in different measurement batches. The various parameters of the transformer are collectively used as the device parameters of the transformer; the various parameters of the transformer are respectively used as device sub-parameters, such as the temperature of the transformer as a device sub-parameter of the transformer, and the leakage inductance of the transformer as another device sub-parameter of the transformer.

[0068] Taking a transformer as an example, the output sub-parameters of the transformer may include the parameters shown in Table 1.

[0069] Table 1: Output sub-parameters of transformer

[0070]

[0071]

[0072] The device sub-parameters may be key parameters of the corresponding customized components, i.e., parameters that will affect the output signal of the product under test. The output sub-parameters of the product under test may be key parameters of the product under test, i.e., parameters that will affect the operation of the product under test.

[0073] It should be noted that when testing the product under test, multiple measurements need to be performed, and each measurement is regarded as a measurement batch. The number of measurements under different production processes can be the same or different, and is not specifically limited here.

[0074] The measured values ​​of product parameters in the functional test process, high temperature aging test process, and room temperature copying process of the product production can be obtained. Of course, the measured values ​​of product parameters in other processes can also be obtained, and the production process is not specifically limited here. The measured values ​​of product parameters under different production processes are collected to make the measured values ​​more comprehensive and the influence coefficients more accurately determined.

[0075] The product under test may be a LEU product, or other products, which is not specifically limited here.

[0076] S102 , classify and count the measured values ​​of the product parameters to obtain a device sub-parameter data set of customized components and an output sub-parameter data set of the measured product.

[0077] Specifically, the measured values ​​of the product parameters can be first classified and counted according to each customized component to obtain the output signal data set of the tested product and the device parameter data set of each customized component; then, each data set of each customized component can be classified and counted according to the device sub-parameters of the customized component to obtain the device sub-parameter data set of each customized component.

[0078] The device sub-parameter data set includes: a plurality of measured values ​​of the device sub-parameters; and the output sub-parameter data set includes: a plurality of measured values ​​of the output sub-parameters.

[0079] For example, the temperature of the transformer is a device sub-parameter of the transformer, and the temperature data set of the transformer includes multiple temperature measurement values. The leakage inductance value of the transformer is another device sub-parameter of the transformer, and the leakage inductance value data set of the transformer includes multiple leakage inductance measurement values.

[0080] Each measurement value may carry a corresponding device identifier and sub-parameter identifier, and then classification statistics may be performed based on the identifier. Of course, other methods may also be used for classification statistics, which will not be elaborated here one by one. It depends on the actual situation and is within the scope of protection of this application.

[0081] S103 , performing influence coefficient analysis on the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product to obtain the influence coefficient of the device sub-parameter of each customized component on the output sub-parameter of the tested product.

[0082] That is to say, any device sub-parameter is taken as an influencing factor and the corresponding output sub-parameter is taken as an output result, and the influence coefficient of the influencing factor on the output result is calculated according to the specific value of the influencing factor and the specific value of the output result.

[0083] Specifically, any device sub-parameter and any output sub-parameter can be used as a set of data to determine the influence coefficient of the device sub-parameter on the output sub-parameter, that is, the influence coefficients of all combinations of all device sub-parameters and all output sub-parameters need to be calculated.

[0084] It is also possible to set the device sub-parameters and the corresponding output sub-parameters to form a set of data, that is, only the influence coefficient of the set combination is calculated.

[0085] The influence coefficient can be understood as the influence of adjusting the specific value of the device sub-parameter on the corresponding output sub-parameter. For example, when the temperature of the transformer is adjusted from 30 degrees to 40 degrees, what will be the change of the corresponding output current, such as from 10A to 10.5A, etc. This is just an example for explanation, and the specific situation will not be repeated here. Of course, other explanations can also be given, which will not be repeated here.

[0086] S104. Determine an adjustment plan for the tested product based on each influencing coefficient.

[0087] Specifically, a mapping relationship between the influence coefficient and the adjustment scheme can be set, and then the adjustment scheme can be determined by the influence coefficient. Of course, other methods can also be used, which are not specifically limited here. The mapping relationship between each device sub-parameter and the adjustment scheme of different customized components can be different, and the corresponding mapping relationship can be set respectively.

[0088] The adjustment scheme can adjust the corresponding device sub-parameters of each customized component, so that the output sub-parameters of the tested product meet the corresponding requirements, such as adjusting the turns ratio of the transformer, such as the turns ratio is 5:1, so that the output voltage of the tested product meets the requirements, such as adjusting the output voltage to 12V, etc. In other words, the device sub-parameters of customized components are adjusted according to the purpose.

[0089] It should be noted that the adjustment scheme may be an adjustment scheme for device sub-parameters of multiple customized components. Of course, when the influence coefficient is within the corresponding range, the device sub-parameters of the customized components corresponding to the influence coefficient do not need to be adjusted.

[0090] In addition, the present application can be applied to the CPS architecture, and then by leveraging the advantages of CPS technology, it can realize the collection of parameters of customized components in the tested products, the collection of key performance parameters of products, and obtain the influence relationship between parameters through data statistics and analysis.

[0091] Taking LEU products as an example, a CPS system for LEU products is built, that is, the device sub-parameters of the customized components of the products and the output sub-parameters of the LEU products themselves are collected in real time during the three processes of normal temperature testing, normal temperature copying and high temperature testing in the production process of LEU products. The data are transmitted to the MES system through the communication terminal of the workshop MES system for data aggregation, and then the MES system uploads the data set to the industrial data integration platform for data statistics and analysis. The relationship between parameters is obtained through the algorithm, which is convenient for guiding the component parameter adjustment plan.

[0092] In this embodiment, the measured values ​​of product parameters of the tested product under each production process are collected; the measured values ​​of the product parameters include the measured values ​​of the output signal of the tested product and the measured values ​​of the device parameters of each customized component in the tested product; the output signal includes at least one output sub-parameter, and the device parameter of each customized component includes at least one device sub-parameter; the measured values ​​of the product parameters are classified and counted to obtain a device sub-parameter data set of the customized components and an output sub-parameter data set of the tested product; the influence coefficient analysis is performed on the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product to obtain the influence coefficient of the device sub-parameter of each customized component on the output sub-parameter of the tested product; based on each influence coefficient, an adjustment plan for the tested product is determined; the influence relationship between the device sub-parameters of the customized components and the output sub-parameters of the tested product is obtained through data statistics and analysis, and the adjustment plan for the tested product is further determined, which is convenient for guiding the adjustment of the device sub-parameters of the customized components, without manual analysis, and unified analysis and adjustment standards, so as to achieve fast and accurate adjustment of the device sub-parameters of the customized components, so as to reduce the number of replacements in the production process, improve production efficiency, and save production costs.

[0093] Optionally, the specific process of the above step S102, classifying and counting the measured values ​​of the product parameters to obtain the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product, may be:

[0094] Each device sub-parameter and each output sub-parameter is used as a statistical target, and the measured values ​​of each device parameter and each output sub-parameter are counted. The set of each measured value of the device sub-parameter is used as a data set of the corresponding device sub-parameter. The set of each measured value of the output sub-parameter is used as a data set of the corresponding output sub-parameter.

[0095] That is to say, when the measured values ​​of product parameters are collected, each measured value is not distinguished and collected uniformly. However, the impact of device sub-parameters on output sub-parameters needs to be analyzed later, so each measured value needs to be classified and counted and divided into corresponding data sets.

[0096] Specifically, a device sub-parameter is used as a statistical target, and in combination with corresponding software tools, the measurement value corresponding to the device sub-parameter is found in each measurement value. The specific search method can be that the identifier corresponding to each measurement value is the identifier of the device sub-parameter, and of course other methods can also be used. All measurement values ​​corresponding to the device sub-parameter are placed in the device sub-parameter data set. The same is true for other device sub-parameters and output sub-parameters, which will not be described here one by one.

[0097] Taking customized components as an example, their device sub-parameters include temperature, humidity, leakage inductance, turns ratio and other parameters. If the user needs to observe the data set of the above parameters, a list of measured values ​​of each parameter on the same test date can be established, as shown in Table 2. A and B represent the result sets obtained after scanning by column, which is a data clipping / integration method that can be determined by the user, such as slicing, dicing, drilling, rotating, etc.

[0098] Table 2: List of measurement values

[0099]

[0100] Table 2 lists the distribution of the parameters required by the user, which completes the preliminary observation of the relationship between the key parameters of customized components and products. The influence relationship between the sub-parameters of customized components and the output sub-parameters of the tested product needs to be further explored through algorithms, which will not be described in detail here.

[0101] According to your own needs, you can perform single-layer or multi-layer queries in the multidimensional database, and obtain the Result framework, such as the structure in Table 3 to display the results.

[0102] Table 3: Multidimensional structure diagram result table

[0103]

[0104] As shown in Table 3, it is a multidimensional result set framework. The dimension members on the horizontal and vertical axes intersect to form a cell. Logically, the cells do not belong to the same hierarchical relationship. The cells obtained by intersecting these specific hierarchical members can be obtained in SQL queries. These cell sets are referred to as segments below. The so-called "segment" is a set of cells with the same measurement, dimension and value. Table 4 is a result framework table after the cell segmentation in the similar results shown in Table 3.

[0105] Table 4: Split cell frame result table

[0106]

[0107] The data mining algorithm involved in the next step will be calculated based on the fragments. Since some data does not appear in the cache area, it is necessary to process the uncached data into fragments and then classify them using the algorithm. Non-fragmented data sets are called "stripes". The call stack in Mondrian can decompose "stripes" into "fragments" and splice "fragments" into "stripes" during the return process. Among them, Mondrian is an open source server developed in Java.

[0108] The following is part of the Mondrian call stack program code:

[0109] load()-mondrian.rolap.agg.Segment;

[0110] load()-mondrian.rolap.agg.Aggregation;

[0111] loadAggregation()-mondrian.rolap.agg.AggregationManager;

[0112] loadAggregation()-mondrian.rolap.FastBatchingCellReader.Batch;

[0113] loadAggregation()-mondrian.rolap.FastBatchingCellreader;

[0114] executeBody(Query)-mondrian.rolap.Result;

[0115] The present application can use ROLAP technology to realize data statistics. The ROLAP technology breaks the traditional data statistics mode, adopts a multi-dimensional data statistics mode, increases the correlation exploration between parameters, and facilitates in-depth understanding of the relationship between parameters.

[0116] The received data can be used as metadata of the ROLAP technology. The data is divided into various device sub-parameters of customized components (each parameter includes multiple sets of values, obtained by testing multiple products on the production line) and various output sub-parameters of the tested product (each parameter includes multiple sets of values, obtained by testing multiple products on the production line). The data sets of various device sub-parameters of customized components and the data of various output sub-parameters of the tested product are subsets of the metadata. The data received here can be the measured values ​​of product parameters or the pre-processed measured values ​​of product parameters.

[0117] See also Figure 2 ,The physical data storage is realized through a relational database server. ,The source data includes the sub-parameter data sets of each device of ,customized components and the output sub-parameter data sets of the ,tested products. Further calculations can be performed in the “relational ,database server” part, such as the difference of the same key parameter during ,different periods, etc. It can be set according to user ,requirements. In this solution, no further calculation is required and ,can be used directly as metadata to enable further calculation ,supported by the system.

[0118] Figure 2 The "source data" here refers to the input data of the ROLAP system, which can be the measured value of the product parameter or the data after preprocessing the measured value of the product parameter; "metadata" and "index" are the functions of the "relational database server" part, and "metadata" is the processed data set of "source data". In this solution, "source data" can be directly used as "metadata"; "index" is a data query function, that is, the server can realize the query of "source data" and "metadata", which is not involved in this solution.

[0119] In addition, after completing the physical data storage, multi-dimensional viewing function can also be provided.

[0120] Specifically, Figure 3 As shown, first, use the software tool to establish a connection with the SQL database. Secondly, create a ROLAP system, enter the device sub-parameters of the customized components into the SQL database as metadata using the output sub-parameters of the tested product, and form a multi-parameter fusion data set under the classification of customized components and tested products, which can be viewed on the database interface. "MDX query to generate Query objects" and "establish Result framework" are technical preparations for the next step. Then, the software obtains the parameters required by the user in the database through MDX instructions. Finally, the final result is formed into a framework and displayed to the user.

[0121] The ROLAP data model is based on a relational database. It is a flat structure that uses a star model to represent multidimensional data using a relational database (sql server, mysql). It uses complex SQL to obtain data from a data warehouse and then uses a relational database to display multidimensional data sets.

[0122] The star model uses two types of tables. One is the fact table. The fact table can be a measure or a measure group, which stores the values ​​of the variables we are looking for, such as Figure 4 The InternetSales measure group shown is the fact table. The second type of table is the dimension table, such as Figure 1 The measure group of Customer, Date, and Product shown in the figure is the dimension table. The fact table is in the middle, and multiple dimension tables radiate from it, so it is called a star model.

[0123] There are actually many ways to display multidimensional data sets. They can be imported into Excel tables or displayed in the form of a matrix using tools such as powerBI.

[0124] like Figure 5 As shown, it is another form of the star model, where InternetSales is the fact table, and DueDate, Customer, OrderDate, Product, and Ship Date are all dimension tables.

[0125] Optionally, the above step S103, performing influence coefficient analysis on the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product to obtain the influence coefficient of the device sub-parameter of each customized component on the output sub-parameter of the tested product, includes:

[0126] For any device sub-parameter data set and any output sub-parameter data set: determine the dissimilarity between the corresponding device sub-parameter data set and the corresponding output sub-parameter data set through the K-means algorithm, and use the dissimilarity as the influence coefficient between the corresponding device sub-parameter and the corresponding output sub-parameter.

[0127] That is to say, any device sub-parameter and any output sub-parameter are taken as a combination to calculate the influence coefficient. The K-average method can be used to calculate the dissimilarity between the device sub-parameter data set and the output sub-parameter data set, and then the dissimilarity is used as the influence coefficient of the combination.

[0128] For example, for the a1 device sub-parameter and the b1 output sub-parameter: the dissimilarity between the device sub-parameter dataset A1 and the output sub-parameter dataset B1 is determined by the K-average algorithm, and the dissimilarity is used as the influence coefficient between the a1 device sub-parameter and the b1 output sub-parameter. The device sub-parameter dataset A1 includes multiple measured values ​​of the a1 device sub-parameter; the output sub-parameter dataset B1 includes multiple measured values ​​of the b1 output sub-parameter.

[0129] K-means algorithm is a simple and effective statistical clustering technique. The algorithm can classify two sets of data by the dissimilarity between them, or it can be used to determine the dissimilarity.

[0130] This application analyzes the influence relationship between the device sub-parameters of customized components and the output sub-parameters of the tested product, and finally obtains the data of the influence coefficient between each parameter. By quantifying the theoretical relationship between the key parameters of the two, the purpose of optimizing the output sub-parameters by adjusting the device sub-parameters is finally achieved.

[0131] Taking the K-means algorithm as an example, the algorithm includes two calculation quantities, L_id and Short_id. L_id is the leakage inductance value, a device sub-parameter of the customized component transformer, and Short_id is the short-circuit load value, an output sub-parameter of the product under test. The correlation between the two is calculated through the K-means algorithm. The same is true for other key parameters. By comparing the correlations, we can know which quantities among the key parameters of customized components have a greater impact on the output signal. By adjusting these quantities, we can optimize the key parameters of the product.

[0132] Optionally, determining the dissimilarity between the corresponding device sub-parameter data set and the corresponding output sub-parameter data set by a K-means algorithm includes:

[0133] Determine the difference between the measured value of the device sub-parameter and the measured value of the output sub-parameter collected in the same measurement batch, and process the difference by a square algorithm to obtain the first value of the corresponding measurement batch; sum the first values ​​of different measurement batches and perform square root processing to obtain the dissimilarity between the device sub-parameter data set and the output sub-parameter data set.

[0134] It should be noted that during the testing process of the product under test, it is necessary to perform multiple measurements on the product under test, and each measurement is regarded as the same measurement batch.

[0135] Suppose we have two sets of elements: X = {x1, x2...xn}, Y = {y1, y2...yn}, which is the calculation of the dissimilarity between scalars. They each have n characteristic attributes, then the dissimilarity between X and Y is defined as:

[0136]

[0137] X is a device sub-parameter data set, x1, x2...xn are measurement values ​​of different measurement batches of corresponding device sub-parameters, Y is an output sub-parameter data set, y1, y2...yn are measurement values ​​of different measurement batches of corresponding output sub-parameters.

[0138] A data mining algorithm, namely the K-means algorithm, is used to deeply calculate the potential relationship between parameters, namely the influence coefficient between device sub-parameters and output sub-parameters. Each device sub-parameter and output sub-parameter can be reflected through a coordinate mapping diagram.

[0139] Optionally, the above step S104, determining an adjustment plan for the tested product according to each influence coefficient, includes:

[0140] First, according to the influence coefficient and the mapping relationship between the influence coefficient and the adjustment scheme, the adjustment scheme of the corresponding device sub-parameter of the customized component corresponding to the influence coefficient is determined. The adjustment schemes of each device sub-parameter of each customized component are summarized as the adjustment scheme of the tested product.

[0141] That is to say, based on the influence coefficient between each device sub-parameter and the output sub-parameter, suggestions for adjusting the customized components of the product circuit are put forward to reduce the loss during this period of functional testing.

[0142] The adjustment plan can also be presented to the user in an interactive manner, such as through a web page mode, or in other ways, which are not specifically limited here. Using web page development, web page applications have no regional restrictions, are highly promotable, have an interactive interface, and are more convenient to deploy and apply than other software.

[0143] In the present application, since the parameter consistency of the customized components in the tested product is poor and has a significant impact on the output signal of the tested product, the present application proposes real-time monitoring of the device sub-parameters of the customized components and the output sub-parameters of the tested product. By obtaining the potential impact between the parameters, the adjustment direction of the parameters of the customized components is obtained to reduce production costs and improve production efficiency.

[0144] Optionally, after the above step S102, in which the measured values ​​of the product parameters are classified and counted to obtain the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product, the method further includes:

[0145] Linear regression processing is performed on each device sub-parameter data set and each output sub-parameter data set to obtain the change trends between each device sub-parameter and output sub-parameter.

[0146] Specifically, it is possible to determine whether there is an abnormality based on the change trend. For example, when the coordinate point of a device sub-parameter and the corresponding output sub-parameter does not match the change trend formed by the two parameters, it can be said that there is an abnormality in the current device sub-parameter, and then the corresponding custom component is located. When the slope of the linear regression remains unchanged, it means that the consistency of this batch of custom components is good. When the values ​​collected from a production batch of custom components are less consistent, the slope of its linear regression will be abnormal.

[0147] The linear regression algorithm is used to realize the predictive properties of data. Linear regression is a predictive data mining method that derives a prediction model based on one or more response variables of one or more variables. The linear regression algorithm model in this application only considers the case of two variables (binary), that is, the function of the monitoring object parameters and time. Temperature is also used as the research dimension.

[0148] In binary variables, Y represents the dependent variable or response variable, that is, the corresponding output sub-parameter, and X represents the independent variable or explanatory variable, that is, the corresponding device sub-parameter; the linear regression algorithm shows that there is a linear noise relationship between X and Y, so for each set of observation data (xi, yi), the regression function is defined as follows:

[0149] yi=a+bxi+ei, (i=1, 2,...,n);

[0150] Where a is the intercept of the regression function, b is the slope of the regression function, also called the regression coefficient, and ei is the random error of the regression function for the i-th set of observations.

[0151] A regression function has two main parts: the regression line and the error term. The regression line can be constructed empirically; the error term describes how well the regression line approximates the observed response variable. From an exploratory analysis perspective, determining the regression line can be described as finding a fitted line that makes the scattered points as close to it as possible. The regression line is a linear function:

[0152]

[0153] in, It means that the fitted value of the ith dependent variable is calculated based on the value of the ith explanatory variable xi. For each observation yi, the error term ei (residual) is the sum of the observed response value yi and the fitted value. The difference:

[0154]

[0155] Each error term can be interpreted as the part that cannot be explained by the linear relationship between the corresponding explanatory variables. The least squares method is usually used to solve this problem. The specific steps of the algorithm are:

[0156] First, the key to the linear regression algorithm is to find the y that minimizes the error; then, to solve the sum of squared errors of the fitting values, the least squares method or gradient descent method can be used; finally, the fitting curve is calculated and simulated.

[0157] The present application uses the least square method to calculate so as to minimize the sum of squared errors (SSE).

[0158] First, the SSE method is defined as:

[0159] In order to obtain the minimum value of SSE, we need to find the deviation of the SSE function about a and b and set it to 0. Since the sum of squared errors is a quadratic function, if there is an extreme point, then that point corresponds to the minimum value. Therefore, the parameters of the regression line can be obtained by solving the following standard system of equations:

[0160]

[0161] We get: where μy and μx are the mean values: a = μy - bμx, and substitute it into the following formula:

[0162]

[0163] Where σx and σy are the standard deviations of variables X and Y respectively: r(X, Y) is the correlation coefficient between variables X and Y, and d is the sign for partial derivative.

[0164] In addition, the present application uses a linear regression algorithm to calculate the responsiveness of the system warning. Therefore, in the selection of X and Y objects, X can be taken as the leakage inductance parameter and Y as the change value of the short-circuit impedance critical value parameter. For example, if the leakage inductance value is kept within a constant range, the relationship between the change value of the short-circuit impedance critical value parameter and time, that is, the change rate of the short-circuit impedance critical value parameter is constant, which is reflected on the regression line as the slope value of the straight line is stable and unchanged. When the change value in a certain period of time suddenly changes, the regression line reflected by the data will definitely not be a fitting state. In this way, when the change rate is abnormal, the problem can be discovered and the abnormal point can be found, and it can be restored in time.

[0165] Back to the algorithm operation, since the slope of the regression line is obtained through empirical values, a set of parameter change values ​​under normal conditions is taken and their average value is taken, and their slope b is calculated as the slope of the regression line. Therefore, the regression line can be described as the following equation:

[0166] REND = b WORLD;

[0167] Among them, REND is the response variable and WORLD is the explanatory variable. The regression line obtained by this set of parameter changes can be shown as Figure 6 As shown:

[0168] After analyzing the data, the distribution relationship between the parameters and the prediction and judgment of the expected distribution trajectory of the parameters are obtained. Not only the parameter distribution law is obtained, but also the product performance development can be predicted based on the prediction of the parameter distribution trend, and the flow of products to the next process can be interrupted in time, reducing production costs and improving production efficiency.

[0169] The K-means algorithm and the linear regression algorithm are both types of data mining algorithms, which perform statistical calculations on non-directly related parameters to further explore non-intuitive data relationships.

[0170] In this application, since the distribution of device sub-parameters of customized components of the tested product and the genus sub-parameters of the tested product are relatively scattered, a data mining algorithm, namely a linear regression algorithm, is used to predict the development of product performance according to the data distribution trend.

[0171] Optionally, after performing influence coefficient analysis on the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product in the above step S103 to obtain the influence coefficient of the device sub-parameter of each customized component on the output sub-parameter of the tested product, the following further includes:

[0172] The influence coefficient of the device sub-parameters of each customized component on the output sub-parameters of the tested product is displayed to the system user or production manager.

[0173] Specifically, the corresponding parameters can be displayed through a web page or a client, etc. The specific display method will not be described here one by one, and it will depend on the actual situation, all of which are within the protection scope of this application.

[0174] Optionally, before classifying and counting the measured values ​​of the product parameters to obtain the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product, the following is further included:

[0175] Preprocess the measured values ​​of product parameters.

[0176] The preprocessing can be to further calculate and organize the product parameters. For example, the turns ratio of the transformer of the tested product needs to be calculated based on the product parameters. The eye diagram parameters of the C1 signal of the output signal, the harmonic calculation of the C6 signal and other parameters also need to be calculated based on the product parameters.

[0177] At the same time, statistics can also be performed on the sub-parameters of each device of customized components from the same production batch. The purpose of statistics is to be able to identify the production batch of customized components from the same production batch when displaying the analysis results, or to filter and display the analysis results of customized components from the same production batch.

[0178] Another embodiment of the present application provides a product parameter statistical analysis system.

[0179] See also Figure 7 , the product parameter statistical analysis system includes:

[0180] The terminal module is used to collect the measured values ​​of product parameters of the tested product in each production process; the measured values ​​of the product parameters include the measured values ​​of the output signal of the tested product and the measured values ​​of the device parameters of each customized component in the tested product; the output signal includes at least one output sub-parameter, and the device parameters of each customized component include at least one device sub-parameter.

[0181] The integration module is used to classify and count the measured values ​​of product parameters to obtain the device sub-parameter data set of customized components and the output sub-parameter data set of the tested product.

[0182] The analysis module is used to perform influence coefficient analysis on the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product to obtain the influence coefficient of the device sub-parameter of each customized component on the output sub-parameter of the tested product.

[0183] The countermeasure module is used to determine the adjustment plan of the tested product based on various influencing coefficients.

[0184] In this embodiment, the terminal module is used to collect the measured values ​​of the product parameters of the tested product in each production process; the measured values ​​of the product parameters include the measured values ​​of the output signal of the tested product and the measured values ​​of the device parameters of each customized component in the tested product; the output signal includes at least one output sub-parameter, and the device parameter of each customized component includes at least one device sub-parameter; the integration module is used to classify and count the measured values ​​of the product parameters to obtain the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product; the analysis module is used to classify and count the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product The influence coefficient analysis is performed on the sub-parameter data set to obtain the influence coefficient of the device sub-parameters of each customized component on the output sub-parameters of the product under test; the countermeasure module is used to determine the adjustment plan of the product under test based on each influence coefficient; the influence relationship between the device sub-parameters of the customized components and the output sub-parameters of the product under test is obtained through data statistics and analysis, and the adjustment plan of the product under test is further determined, which is convenient for guiding the adjustment of the device sub-parameters of the customized components. There is no need for manual analysis, and the analysis and adjustment standards are unified to achieve fast and accurate adjustment of the device sub-parameters of the customized components, so as to reduce the number of replacements in the production process, improve production efficiency, and save production costs.

[0185] Optionally, applied to the cyber-physical system (CPS) architecture; where:

[0186] The terminal module is the actuator in the CPS architecture;

[0187] The integration module is the EMS system in the CPS architecture;

[0188] The analysis module is the PLM system in the CPS architecture;

[0189] The countermeasure module is an integrated platform in the CPS architecture.

[0190] The product parameter statistical analysis system may also include a collection module, which is mainly used to pre-process the parameters collected by the terminal module.

[0191] CPS architecture system Figure 8 As shown in the figure, the CPS architecture system includes external data services, enterprise management layer, operation management layer, production site control layer and physical entities. The external data service is equipped with cloud computing, big data analysis, industrial application programs (APP) and other services, as well as industrial data integration platforms; the enterprise management layer is equipped with SCM, ERP, PLM and other systems. The operation management layer is equipped with product R&D virtual simulation, production process virtual simulation, and MES scheduling, management, materials, production, quality and equipment. The production site control layer is equipped with SCADA, DCS / FCS and other field control systems. The physical entity is equipped with controllers, sensors, actuators, traditional production equipment, instruments and other equipment.

[0192] The CPS architecture interacts with the outside world, such as securely interacting with information and devices through an interconnected network.

[0193] The specific overall structure of the CPS architecture applied in this application is as follows Fig. 9 shown.

[0194] The system structure mainly consists of: countermeasure module, analysis module, integration module, acquisition module, and terminal module; the terminal module uses the actuator in the CPS architecture to realize signal acquisition; the acquisition module uses the SCADA system of the CPS architecture to realize parameter calculation and integration; the integration module uses the MES system in the CPS architecture to implement ROLAP technology to realize multi-data display; the analysis module uses the PLM system in the CPS architecture to realize data mining and data fitting algorithms; the integration module uses an integrated platform to realize the terminal mode, that is, display data.

[0195] Specifically, each module is described below:

[0196] (1) The terminal module is the actuator in the CPS architecture.

[0197] The terminal module includes sensors and collectors, which are mainly used to obtain key parameters of product circuit customized components (transformers, current transformers, etc.) and key parameters of product outputs and transmit them to the collection module via Ethernet communication.

[0198] For example, the temperature can be obtained through a temperature sensor, and the leakage inductance value can be obtained through instruments such as an impedance analyzer.

[0199] In addition to the CPS architecture, other IoT terminal collection methods can be used to collect and upload data.

[0200] This application uses Ethernet communication as an example for explanation, other wireless communication modes are also feasible and will not be described in detail here.

[0201] (2) The acquisition module is the ACADA system in the CPS architecture.

[0202] The acquisition module mainly performs preliminary processing and statistical methods on the parameters uploaded by the terminal module, obtains the final parameter values ​​to be analyzed, and transmits the sorted data to the integration module via Ethernet communication.

[0203] In this solution, it mainly refers to the calculation and statistics of the device sub-parameters of customized components obtained by the terminal module and the output signals of the tested product. Taking customized components as an example, the different device sub-parameters of customized components of the same production batch will be counted in the acquisition module; taking the tested product as an example, the main parameters obtained are C1 signal and C6 signal, and the eye diagram parameters of C1 signal and the harmonic component parameters of C6 signal will be obtained by algorithm in the acquisition module. Then, the processed data is transmitted to the integration module through Ethernet communication.

[0204] (3) The integration module is the EMS system in the CPS architecture.

[0205] The integration module combines ROLAP technology to classify and count the final parameter values ​​of the acquisition module.

[0206] Specifically, the integration module uses the ROLAP technology, an analysis tool for the final parameter value, to process the data at the database level. Therefore, in the integration module, a multi-dimensional database system is established for the device sub-parameters of the customized components and the output sub-parameters of the tested product, and their respective data sets are obtained, which facilitates the observation of the potential connections between the parameters from multiple angles.

[0207] This application can use the Mondrian tool to implement ROLAP technology. Mondrian is an open source project built in Java language, including the OLAP engine also written in Java language. The solution itself can use MDX (database query) language to query multidimensional databases, and then display the results in a multidimensional way through Java API (platform external application interface).

[0208] ROLAP technology can realize physical output storage. Specifically, the physical layer is the lowest structure of the database, and the lowest function includes storing the physical input data and the calculated results. Therefore, the physical data storage process occurs in the physical layer of the database. The physical layer is not open to the outside world. This is to ensure the security of the database and to solve the problem of multi-user access. Therefore, access to the physical layer can only be restricted through database management tools.

[0209] ROLAP technology can realize multi-dimensional viewing of objects. Multi-dimensional viewing mainly refers to terminal display, which receives data from the intermediate OLAP layer and then reflects it to the upper-level module. It is also important to establish an application interface based on data characteristics. This requires that in addition to the tools provided by the OLAP system itself, the server can be compatible with other software products, that is, it requires a good OLAP server API interface feature. This part is achieved through the software Mondrian. The advantage of its API interface is mentioned in the introduction to Mondrian.

[0210] ROLAP technology can be used to find clustering calculation targets, and combined with software tools to find data targets, obtain device sub-parameter data sets of customized components and output sub-parameter data sets of the tested products.

[0211] The integration module finally establishes a multi-dimensional model of each parameter and transmits the data fragments to the analysis module via Ethernet communication.

[0212] (4) The analysis module is the PLM system in the CPS architecture.

[0213] The data mining tool used in this application can be the Weka system. Weka is an open source data mining software based on the Java environment. The functions provided by the software include data processing, feature selection, classification, regression, clustering and association rules.

[0214] The analysis module further calculates and analyzes the parameters between the databases. The main idea of ​​the solution is to obtain the connection between the device sub-parameters of customized components and the output sub-parameters of the tested product, and to improve the output sub-parameters of the tested product by adjusting the device sub-parameters of customized components. Therefore, the relevant connection, that is, the influence coefficient, is obtained through the analysis method of the analysis module. The analysis method includes data mining algorithm, data fitting algorithm, etc. This module calculates and analyzes the database data integrated by the integration module to obtain the influence coefficient between the device sub-parameters of customized components and the output sub-parameters of the tested product.

[0215] (5) The countermeasure module is an integrated platform in the CPS architecture.

[0216] The countermeasure module mainly displays the calculation results of the analysis module in a user interactive manner, such as in web page mode, and draws relevant adjustment suggestions for customized components.

[0217] Web page implementation architecture Fig.10 As shown in the figure, the final integrated module proposes adjustment suggestions for the customized components of the product circuit based on the data relationship to reduce the loss during the functional test.

[0218] The countermeasure module may also be implemented by using other technologies such as the client.

[0219] The integration module can be implemented using other technologies such as the client.

[0220] In this embodiment, the performance of the product under test is monitored and adjusted through the CPS architecture. A communication network is established between modules through Ethernet. Ethernet technology is easy to implement and has low requirements for the application environment. It can effectively reduce the cost of building the CPS architecture. The countermeasure module is implemented in web page mode. The distribution and relationship distribution of various parameters can be intuitively viewed through the development of a web terminal. The web terminal also includes an interactive interface, which is also convenient for deploying other software.

[0221] The features recorded in the various embodiments in this specification can be replaced or combined with each other, and the same and similar parts between the various embodiments can refer to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. The system and system embodiments described above are only schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative work.

[0222] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in this application can be implemented by electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0223] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined in this application may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown in this application, but will conform to the widest scope consistent with the principles and novel features disclosed in this application.

Claims

1. A method for statistical analysis of product parameters, characterized in that: include: Collecting the measured values ​​of product parameters of the tested product in each production process; the measured values ​​of the product parameters include the measured values ​​of the output signal of the tested product and the measured values ​​of the device parameters of each customized component in the tested product; the output signal includes at least one output sub-parameter, and the device parameters of each customized component include at least one device sub-parameter; Classifying and counting the measured values ​​of the product parameters to obtain a device sub-parameter data set of the customized components and an output sub-parameter data set of the tested product; Performing influence coefficient analysis on the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product to obtain the influence coefficient of the device sub-parameter of each customized component on the output sub-parameter of the tested product; An adjustment plan for the tested product is determined based on each of the influence coefficients.

2. The product parameter statistical analysis method according to claim 1, characterized in that: The classified statistics of the measured values ​​of the product parameters to obtain the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product include: Taking each of the device sub-parameters and each of the output sub-parameters as statistical targets, classifying and counting the measured values ​​of each of the device parameters and each of the output sub-parameters; taking a set of each measured value of the device sub-parameter as a data set of the corresponding device sub-parameter; The set of each measured value of the output sub-parameter is used as the data set of the corresponding output sub-parameter.

3. The product parameter statistical analysis method according to claim 1, characterized in that: The performing influence coefficient analysis on the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product to obtain the influence coefficient of the device sub-parameter of each customized component on the output sub-parameter of the tested product includes: For any of the device sub-parameter data sets and any of the output sub-parameter data sets: determine the dissimilarity between the corresponding device sub-parameter data set and the corresponding output sub-parameter data set using a K-means algorithm, and use the dissimilarity as an influence coefficient between the corresponding device sub-parameter and the corresponding output sub-parameter.

4. The product parameter statistical analysis method according to claim 3, characterized in that: The determining the dissimilarity between the corresponding device sub-parameter data set and the corresponding output sub-parameter data set by using a K-average algorithm includes: Determine the difference between the measured value of the device sub-parameter and the measured value of the output sub-parameter collected in the same measurement batch, and perform square algorithm processing on the difference to obtain a first value of the corresponding measurement batch; The first values ​​of different measurement batches are summed and square rooted to obtain the dissimilarity between the device sub-parameter data set and the output sub-parameter data set.

5. The product parameter statistical analysis method according to claim 1, characterized in that: Determining the adjustment plan of the tested product according to each of the influence coefficients includes: Determining, based on the influence coefficient and the mapping relationship between the influence coefficient and the adjustment scheme, an adjustment scheme for corresponding device sub-parameters of the customized component corresponding to the influence coefficient; The adjustment schemes of the device sub-parameters of the customized components are summarized as the adjustment scheme of the product under test.

6. The product parameter statistical analysis method according to claim 1, characterized in that: After classifying and counting the measured values ​​of the product parameters to obtain the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product, the method further includes: Linear regression processing is performed on each of the device sub-parameter data sets and each of the output sub-parameter data sets to obtain the change trends between each of the device sub-parameters and the output sub-parameters.

7. The product parameter statistical analysis method according to claim 1, characterized in that: After performing influence coefficient analysis on the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product to obtain the influence coefficient of each device sub-parameter of the customized components on the output sub-parameter of the tested product, the method further includes: The influence coefficient of the device sub-parameter of each of the customized components on the output sub-parameter of the tested product is displayed to the system user or production management personnel.

8. The product parameter statistical analysis method according to claim 1, characterized in that: Before classifying and counting the measured values ​​of the product parameters to obtain the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product, the method further includes: The measured values ​​of the product parameters are preprocessed.

9. A product parameter statistical analysis system, characterized in that: include: The terminal module is used to collect the measured values ​​of the product parameters of the tested product in each production process; the measured values ​​of the product parameters include the measured values ​​of the output signal of the tested product and the measured values ​​of the device parameters of each customized component in the tested product; the output signal includes at least one output sub-parameter, and the device parameter of each customized component includes at least one device sub-parameter; An integration module, used for classifying and counting the measured values ​​of the product parameters to obtain a device sub-parameter data set of the customized components and an output sub-parameter data set of the tested product; An analysis module, used to perform influence coefficient analysis on the device sub-parameter data set of the customized components and the output sub-parameter data set of the tested product, to obtain the influence coefficient of the device sub-parameter of each customized component on the output sub-parameter of the tested product; The countermeasure module is used to determine the adjustment plan of the tested product according to each of the influence coefficients.

10. The product parameter statistical analysis system according to claim 9, characterized in that: Applied to the cyber-physical system (CPS) architecture; among which: The terminal module is an actuator in the CPS architecture; The integration module is a manufacturing execution EMS system in the CPS architecture; The analysis module is a full life cycle management PLM system in the CPS architecture; The countermeasure module is an integrated platform in the CPS architecture.

Images