Method, device and medium for visual construction of mechanism model based on industrial data

Abstract

The application discloses a mechanism model visualization construction method and device based on industrial data and a medium, to solve the problem that the existing mechanism model construction needs algorithm engineers to program, and the construction cost is high, the cycle is long, and secondary development is difficult. The method comprises the following steps: connecting a local database with an industrial data set to obtain industrial data; determining the data processing logic of the industrial data according to the data type to determine the arrangement order of a plurality of commonly used algorithm calculation components, dragging and splicing the plurality of commonly used algorithm calculation components in a visual interface, and preprocessing the industrial data; based on an analysis algorithm calculation component, dimensionally transforming the preprocessed industrial data, and based on a feature calculation component, extracting key features of the dimensionally transformed industrial data; based on a learning algorithm calculation component, constructing a mechanism model according to the key features in the dimensionally transformed industrial data, and triggering the debugging of the mechanism model based on the dragging of each configuration information in the mechanism model to complete the visual construction of the mechanism model.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to methods, devices and media for constructing visual mechanisms based on industrial data. Background Technology

[0002] Mechanism models are computational models based on principles of physics and chemistry, leveraging machine learning, engineering knowledge, and expert experience to solve and optimize specific industrial problems. They can be used in industrial scenarios such as production process optimization, trend prediction and analysis, and decision support. Industrial enterprises rely on data flow and application in all aspects, including R&D, supply chain management, manufacturing, quality control, and operations management. Mechanism models are typically used to transform and preserve the value inherent in industrial data, thereby empowering enterprise production and management processes. Most mechanism models are generally encapsulated in toolboxes, suites, and components within mature commercial software, such as the thermodynamic analysis, fluid mechanics analysis, finite element simulation, and circuit design simulation models built into commercial software like CAE, CAD, and EDA. However, due to the procurement costs and closed nature of commercial software, many enterprises find it difficult to access high-quality, customized mechanism model services.

[0003] Currently, building mechanistic models using traditional code programming requires algorithm engineers to not only possess extensive knowledge of equipment mechanisms and technological processes in the industrial field, but also to have technical skills in machine learning and programming. Industrial enterprises generally do not employ such interdisciplinary talent, making it difficult for enterprise users to independently design, develop, and apply mechanistic models according to their actual needs. The mainstream approach to building mechanistic models primarily relies on code programming by algorithm engineers, which generally suffers from high development costs, long development cycles, and difficulties in secondary development. Summary of the Invention

[0004] This application provides a method, device, and medium for constructing a visual mechanism model based on industrial data, which solves the technical problems of existing mechanism model construction methods requiring algorithm engineers to program code, resulting in high construction costs, long cycles, and difficulties in secondary development.

[0005] On the one hand, embodiments of this application provide a method for constructing a visual mechanism model based on industrial data, including:

[0006] Based on JDBC technology and database open interfaces, the local database is connected to the industrial dataset, and industrial data in the industrial dataset is obtained.

[0007] The data types corresponding to the industrial data are determined, and the arrangement order of multiple pre-encapsulated commonly used algorithm calculation components is determined according to the data processing logic corresponding to the data types of the industrial data; the data types of the industrial data include time series data and image data;

[0008] In the visualization interface, multiple commonly used algorithm calculation components corresponding to the industrial data are spliced ​​together in the order of the arrangement by dragging and dropping, and the industrial data is preprocessed sequentially by the spliced ​​multiple commonly used algorithm calculation components.

[0009] Based on pre-packaged analysis algorithm computing components, the pre-processed industrial data is subjected to dimensional transformation, and based on pre-packaged feature computing components, key features are extracted from the dimensionally transformed industrial data.

[0010] Based on the pre-packaged learning algorithm computing components, a corresponding mechanism model is constructed according to the key features in the industrial data after dimensional transformation. The mechanism model is then debugged based on the drag-and-drop triggering of each configuration information in the mechanism model, so as to complete the visualization construction of the mechanism model.

[0011] In one implementation of this application, the step of connecting a local database with an industrial dataset based on JDBC technology and an open database interface, and obtaining industrial data from the industrial dataset, specifically includes:

[0012] Based on preset sensors and production equipment in the industrial scenario, the time-series data in the industrial scenario is collected to the IoT platform through the first preset protocol. Based on JDBC technology and database open interface, the local database is connected to the IoT platform to obtain the time-series data in the IoT platform.

[0013] Furthermore, through a second preset protocol, based on the preset production line quality inspection equipment in the industrial scenario, image data in the industrial scenario is collected to the data platform, and based on JDBC technology and database open interface, the local database is connected to the data platform to obtain image data in the data platform.

[0014] In one implementation of this application, determining the data type corresponding to the industrial data and, based on the data processing logic corresponding to the data type of the industrial data, determining the arrangement order of multiple pre-encapsulated commonly used algorithm calculation components specifically includes:

[0015] Analyze several industrial data points in the industrial dataset, determine the data types corresponding to the several industrial data points, and determine the data processing logic corresponding to the industrial data based on the data types corresponding to the industrial data.

[0016] According to the data processing logic corresponding to the industrial data, among a number of pre-packaged common algorithm calculation components, multiple common algorithm calculation components corresponding to the industrial data are identified, and the arrangement order among the multiple common algorithm calculation components corresponding to the industrial data is determined according to the data processing logic corresponding to the industrial data.

[0017] In one implementation of this application, the preprocessing of the industrial data by sequentially using multiple commonly used algorithm computing components after splicing specifically includes:

[0018] Read the industrial data from the industrial dataset and parse the industrial data in the industrial dataset to obtain the corresponding parsing results;

[0019] Based on the analysis results corresponding to each industrial data, a data quality analysis report corresponding to the industrial data is determined, and several industrial data that meet the quality standards in the data quality analysis report are identified.

[0020] According to a preset splitting ratio and through random sampling, the several industrial data are split into a training dataset and a test dataset with the corresponding splitting ratio.

[0021] In one implementation of this application, the pre-packaged analysis algorithm calculation component performs dimensional transformation on the preprocessed industrial data, specifically including:

[0022] Based on pre-packaged analysis algorithm computing components, data analysis is performed on pre-processed industrial data, and the data dimensions corresponding to each pre-processed industrial data are determined.

[0023] The analysis algorithm calculation component includes at least: a principal component analysis algorithm calculation component, a linear discriminant analysis algorithm calculation component, and a factor analysis algorithm calculation component;

[0024] At least one industrial data point with a dimension greater than a preset dimension threshold is identified, and dimensionality reduction processing is performed on the at least one industrial data point to achieve dimensionality transformation of the industrial data.

[0025] In one implementation of this application, the extraction of key features from dimensionally transformed industrial data based on a pre-packaged feature calculation component specifically includes:

[0026] Based on pre-packaged feature calculation components, the feature attributes corresponding to the industrial data after dimensional transformation are determined, and the degree of influence of each feature attribute in the industrial data on the mechanism model is determined respectively.

[0027] Based on the degree of influence of each feature attribute on the mechanism model, the feature weight of the corresponding feature attribute in the industrial data is determined, and at least one feature attribute with a feature weight greater than a preset weight threshold is identified, so as to use the at least one feature attribute as a key feature.

[0028] In one implementation of this application, after the pre-packaged learning algorithm computing component constructs a corresponding mechanism model based on the key features in the industrial data after dimensional transformation, the method further includes:

[0029] Based on the pre-packaged model verification calculation component, the independent variable industrial data in the pre-processed test dataset is input into the mechanism model, and the corresponding test dependent variable industrial data is obtained.

[0030] The industrial data of the test dependent variable is compared with the industrial data of the preset target dependent variable, and the accuracy of the mechanism model is determined based on the corresponding similarity.

[0031] In one implementation of this application, after the pre-packaged learning algorithm computing component constructs a corresponding mechanism model based on the key features in the industrial data after dimensional transformation, the method further includes:

[0032] Based on cross-validation, the preprocessed test dataset is split into a preset number of mutually exclusive subsets, and a first number of mutually exclusive subsets are used as the training subset, and a second number of mutually exclusive subsets are used as the test subset; wherein, the preset number is the sum of the first number and the second number;

[0033] The test results corresponding to the first number of training subsets and the second number of test subsets are determined, and the generalization error value corresponding to the mechanism model is determined based on the average value of the specified number of test results.

[0034] On the other hand, embodiments of this application also provide a device for constructing a mechanism model visualization based on industrial data, the device comprising:

[0035] At least one processor;

[0036] And, a memory communicatively connected to the at least one processor;

[0037] The memory stores instructions that can be executed by the at least one processor, which are then executed by the at least one processor to enable the at least one processor to perform the above-described method for constructing a mechanism model based on industrial data visualization.

[0038] On the other hand, embodiments of this application also provide a non-volatile computer storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured as follows:

[0039] The above describes a method for constructing a visualization of mechanistic models based on industrial data.

[0040] This application provides a method, device, and medium for constructing a visual mechanism model based on industrial data, which has at least the following beneficial effects:

[0041] Based on pre-packaged computing components, the system can integrate materialized calculation formulas, industrial mechanism knowledge, and expert experience to meet the needs of algorithm engineers. By connecting the local database with the industrial dataset, it can receive industrial data from the dataset and determine the corresponding data processing logic based on the data type. The computing components for processing industrial data are then sorted according to the determined data processing logic. Based on the determined sorting, the computing components can be dragged and stitched together in the visualization interface to preprocess the industrial data sequentially. The preprocessed industrial data undergoes dimensionality transformation, and key features of the transformed industrial data are extracted. Based on these key features and the pre-packaged computing components, a mechanism model can be constructed. The mechanism model can then be debugged by dragging and dropping in the visualization interface, achieving visualized construction of the mechanism model. Attached Figure Description

[0042] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0043] Figure 1 A flowchart illustrating the method for constructing a mechanism model visualization based on industrial data, as provided in this application embodiment;

[0044] Figure 2 A schematic diagram of the internal structure of a device for constructing a mechanism model based on industrial data, provided in an embodiment of this application. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0046] This application provides a method, device, and medium for visually constructing mechanistic models based on industrial data. Based on pre-packaged computing components, it can embed materialized calculation formulas, industrial mechanism knowledge, and expert experience into the system, meeting the needs of algorithm engineers. By connecting a local database with an industrial dataset, it can receive industrial data from the dataset. Then, based on the data type of the industrial data, it determines the corresponding data processing logic. The computing components for processing the industrial data are sorted according to the determined data processing logic. Based on this sorting, the computing components can be dragged and assembled in the visualization interface to preprocess the industrial data sequentially. The preprocessed industrial data undergoes dimensional transformation, and key features are extracted. Based on these key features and the pre-packaged computing components, a mechanistic model can be constructed. The mechanistic model can then be debugged by dragging and dropping in the visualization interface, achieving visual construction of the mechanistic model. This solves the technical problems of existing mechanistic model construction methods that require algorithm engineers to program code, resulting in high construction costs, long cycles, and difficulties in secondary development.

[0047] The technical solutions provided by the various embodiments of this application are described in detail below with reference to the accompanying drawings.

[0048] Figure 1 This is a flowchart illustrating the method for constructing a visual mechanism model based on industrial data, as provided in an embodiment of this application. Figure 1 As shown in the embodiments of this application, the method for constructing a mechanism model visualization based on industrial data mainly includes the following steps:

[0049] 101. Based on JDBC technology and database open interfaces, the local database is connected to the industrial dataset, and industrial data in the industrial dataset is obtained.

[0050] Mechanistic knowledge mainly refers to the fundamental principles and mechanisms of natural sciences such as mechanics, physics, chemistry, and biology, as well as the operational principles and mechanisms of various technologies and processes, such as the law of conservation of momentum, the laws of thermodynamics, fluid mechanics equations, and chemical reaction equations. It can be acquired by consulting relevant books and asking experts in related technical fields.

[0051] To achieve low-cost and efficient development, management, and application of mechanistic models, this application discloses a method for visually constructing mechanistic models based on industrial data. It provides a complete library of computational components, including modules for data access, data preprocessing, feature extraction, algorithm configuration, model building, model validation, and model application. Simultaneously, it offers a zero-code, visual model orchestration and development tool that encapsulates complex mechanistic knowledge, expert experience, and machine learning algorithms into computational components, enabling the reuse and accumulation of knowledge, experience, and algorithms. Users can independently drag, connect, configure, and debug various computational components according to their business needs, thus achieving visual development of mechanistic models. After model validation, it supports saving and publishing as a standard RESTful API interface, supporting calls from third-party business systems.

[0052] Expert experience mainly refers to the experiential knowledge possessed by professional and skilled personnel with professional knowledge, skills and rich practical experience in the production and management process of industrial enterprises. Generally, it is necessary to communicate and record with experienced workers or domain experts in the factory or workshop to obtain expert experience in relevant fields or production processes. This expert experience is then encapsulated into industry-specific function algorithms in the form of programming code and used in the model building step to realize the platform-based accumulation and modeling reuse of expert experience.

[0053] Specifically, in one embodiment of this application, the server collects time-series data from the industrial scenario to an IoT platform based on preset sensors and production equipment in the industrial scenario using a first preset protocol. It then connects its local database to the IoT platform using JDBC technology and an open database interface to obtain the time-series data from the IoT platform. Simultaneously, the server collects image data from the industrial scenario to a data platform based on preset production line quality inspection equipment using a second preset protocol. It then connects its local database to the data platform using JDBC technology and an open database interface to obtain the image data from the data platform.

[0054] It should be noted that the first preset protocol in this application embodiment is, for example, OPC, Mudbus, MQTT, etc., and the second preset protocol is, for example, HTTP and HTTPS. Industrial datasets include, for example, MySQL, Oracle, DB2, PostgreSQL, MongoDB, PI, TDengine, etc. Time-series data can be used for training and evaluating mechanistic models for equipment fault diagnosis and remaining life prediction, while image data can be used for testing and evaluating image models for image classification and object detection.

[0055] 102. Determine the data type corresponding to the industrial data, and based on the data processing logic corresponding to the data type of the industrial data, determine the arrangement order of multiple pre-packaged commonly used algorithm calculation components.

[0056] Specifically, in one embodiment of this application, the server analyzes several industrial data in the industrial dataset, determines the data types corresponding to the several industrial data, and determines the data processing logic corresponding to the industrial data based on the data types corresponding to the industrial data. Then, according to the data processing logic corresponding to the industrial data, the server determines multiple commonly used algorithm calculation components corresponding to the industrial data from several pre-encapsulated commonly used algorithm calculation components, and determines the arrangement order among the multiple commonly used algorithm calculation components corresponding to the industrial data based on the data processing logic corresponding to the industrial data.

[0057] 103. In the visualization interface, multiple commonly used algorithm calculation components corresponding to industrial data are spliced ​​together in the order of arrangement by dragging and dropping, and the industrial data is preprocessed sequentially by the spliced ​​multiple commonly used algorithm calculation components.

[0058] Specifically, in one embodiment of this application, the server reads industrial data from an industrial dataset, parses the industrial data in the industrial dataset to obtain corresponding parsing results, and then determines a data quality analysis report corresponding to each industrial data based on the parsing results, and identifies several industrial data that meet the quality standards in the data quality analysis report. It should be noted that the data quality analysis report in this embodiment includes analysis results on data accuracy, uniqueness, and completeness.

[0059] Then, the server splits several industrial datasets according to a preset splitting ratio and through random sampling, creating training datasets and test datasets with corresponding splitting ratios. It's important to note that random sampling refers to randomly selecting samples from the collected raw industrial datasets to form the training dataset. In random sampling, each data sample has a chance to be selected, and the data samples are independent and unrelated to each other. This ensures the representativeness of the training data and improves the training effect in the subsequent model building process.

[0060] In one embodiment of this application, the server user triggers a drag-and-drop operation in the visual interface, assembling multiple commonly used algorithm calculation components required for the data processing logic of the industrial data according to the determined arrangement order. Then, the parameters of the calculation components are configured. After the configuration is completed, the user clicks "Run," and the system automatically allocates computing resources and executes them sequentially according to the order in which the calculation components are assembled, ultimately completing the preprocessing process of the industrial data.

[0061] 104. Based on pre-packaged analysis algorithm calculation components, perform dimensional transformation on pre-processed industrial data, and extract key features from the dimensionally transformed industrial data based on pre-packaged feature calculation components.

[0062] Specifically, in one embodiment of this application, the server performs data analysis on the pre-processed industrial data based on a pre-packaged analysis algorithm calculation component, and determines the data dimension corresponding to each pre-processed industrial data. It should be noted that the analysis algorithm calculation component in this embodiment includes at least: a principal component analysis algorithm calculation component, a linear discriminant analysis algorithm calculation component, and a factor analysis algorithm calculation component.

[0063] Then, the server identifies at least one industrial data point with a dimensionality greater than a preset threshold and performs dimensionality reduction processing on each of these data points to achieve dimensionality transformation. It's important to note that dimensionality transformation of industrial data typically refers to converting high-dimensional data to low-dimensional data, or complex data to simple data. Industrial production processes are highly complex, and the collected datasets often contain a large number of multidimensional features. Too many dimensional features can hinder the model from finding patterns, and these features may be highly correlated. Multicollinearity can lead to weak generalization ability of the model. To solve these problems, dimensionality reduction processing is necessary. Dimensionality reduction reduces the number of feature attributes in the dataset, eliminates irrelevant or redundant features, reduces the number of features, improves model accuracy, and reduces model training and inference time. The dimensionality-reduced dataset is then used in the subsequent model building process.

[0064] Based on a pre-packaged feature calculation component, the server determines the feature attributes corresponding to the industrial data after dimensional transformation, and determines the degree of influence of each feature attribute in the industrial data on the mechanism model. Then, based on the degree of influence of each feature attribute on the mechanism model, it can determine the feature weight of the corresponding feature attribute in the industrial data, and determine at least one feature attribute whose feature weight is greater than a preset weight threshold, so as to use at least one feature attribute as a key feature.

[0065] It should be noted that the basis for determining key features is that the feature attribute has a significant impact on the dependent variable y, and the key feature has a high correlation coefficient with the input result. The key feature is to extract the most useful feature attributes for machine learning tasks from the original industrial dataset. The extracted key features play a role in reducing the dimensionality of the data, reducing redundancy and noise, thereby improving the generalization ability and predictive performance of the model.

[0066] 105. Based on the pre-packaged learning algorithm computing component, construct the corresponding mechanism model according to the key features in the industrial data after dimensional transformation, and debug the mechanism model based on the drag-and-drop trigger of each configuration information in the mechanism model to complete the visualization construction of the mechanism model.

[0067] It should be noted that the learning algorithm computation components mainly include two categories of algorithms: supervised learning and unsupervised learning. Supervised learning algorithms include Naive Bayes, decision trees, gradient boosting algorithms, random forests, linear regression, logistic regression, relevance vector machines, support vector machines, and backpropagation neural networks. Unsupervised learning algorithms include K-Means clustering and association rule algorithms. These algorithms form the core computational logic of the entire model construction process and are used in the model building phase. Mechanistic knowledge can generally be abstracted into mathematical computation components, while expert experience can be manually encoded into decision tree business recommendation components. Mechanistic models face complex business scenarios, and the entire model computation process requires the integration of the aforementioned mathematical computation components, decision tree business recommendation components, and machine learning algorithm components.

[0068] In one embodiment of this application, after the server constructs a corresponding mechanism model based on the key features in the industrial data after dimensional transformation using a pre-packaged learning algorithm computing component, it inputs the independent variable industrial data in the pre-processed test dataset into the mechanism model based on a pre-packaged model verification computing component, and obtains the corresponding test dependent variable industrial data. Then, it compares the similarity between the test dependent variable industrial data and the preset target dependent variable industrial data, and determines the accuracy of the mechanism model based on the corresponding similarity.

[0069] In one embodiment of this application, after the server constructs a corresponding mechanism model based on the key features in the dimensionality-transformed industrial data using a pre-packaged learning algorithm computing component, it can split the preprocessed test dataset into a preset number of mutually exclusive subsets based on cross-validation. The first number of mutually exclusive subsets are used as the training subset, and the second number of mutually exclusive subsets are used as the test subset. It should be noted that the preset number in this embodiment is the sum of the first and second numbers. In this application, the second number is 1, and the first number is the preset number minus 1. The preset number can be selected as an appropriate value based on the actual situation.

[0070] Then, the server determines the test results corresponding to a first number of training subsets and a second number of test subsets, and determines the generalization error value corresponding to the mechanistic model based on the average of a specified number of test results. It should be noted that the specified number selected in this embodiment is 10; those skilled in the art can determine the specific value of the specified number based on actual circumstances, and this application does not impose a specific limitation on it.

[0071] In one embodiment of this application, the server deploys the mechanism model as an API interface, which includes the following operation process: Flask is a lightweight web application framework written in Python. With the help of the platform's built-in Flask framework, the mechanism model file obtained from training and verification can be published as an API interface. Specifically, a model service application script is created based on the model code file after training and evaluation and imported into the Flask framework. The script specifies the access path and request method of the model API interface, writes the execution action when calling the interface, binds the access IP address and port of the model interface service, and uses the Postman interface testing tool to test the service call.

[0072] The above are embodiments of the method proposed in this application. Based on the same inventive concept, embodiments of this application also provide a device for constructing a mechanism model visualization based on industrial data, the structure of which is as follows: Figure 2 As shown.

[0073] Figure 2 This application provides a schematic diagram of the internal structure of a device constructed using a mechanism model based on industrial data, as shown in the embodiments of this application. Figure 2 As shown, the device includes:

[0074] At least one processor;

[0075] And, a memory that is communicatively connected to at least one processor;

[0076] The memory stores instructions that can be executed by at least one processor, and the instructions, when executed by at least one processor, enable at least one processor to:

[0077] Based on JDBC technology and database open interfaces, the local database is connected to the industrial dataset, and industrial data in the industrial dataset is obtained.

[0078] The data types corresponding to the industrial data are determined, and based on the data processing logic corresponding to the data types of the industrial data, the arrangement order of several pre-encapsulated commonly used algorithm calculation components is determined; the data types of industrial data include time series data and image data;

[0079] In the visualization interface, multiple commonly used algorithm calculation components corresponding to industrial data are spliced ​​together in the order of arrangement by dragging and dropping, and the industrial data is preprocessed in sequence by the spliced ​​multiple commonly used algorithm calculation components.

[0080] Based on pre-packaged analysis algorithm computing components, the pre-processed industrial data is subjected to dimensional transformation, and based on pre-packaged feature computing components, key features are extracted from the dimensionally transformed industrial data.

[0081] Based on pre-packaged learning algorithm computing components, a corresponding mechanism model is constructed according to the key features in the industrial data after dimensional transformation. The mechanism model is then debugged based on drag-and-drop triggers of various configuration information in the mechanism model, so as to complete the visualization construction of the mechanism model.

[0082] This application embodiment also provides a non-volatile computer storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured as follows:

[0083] Based on JDBC technology and database open interfaces, the local database is connected to the industrial dataset, and industrial data in the industrial dataset is obtained.

[0084] The data types corresponding to the industrial data are determined, and based on the data processing logic corresponding to the data types of the industrial data, the arrangement order of several pre-encapsulated commonly used algorithm calculation components is determined; the data types of industrial data include time series data and image data;

[0085] In the visualization interface, multiple commonly used algorithm calculation components corresponding to industrial data are spliced ​​together in the order of arrangement by dragging and dropping, and the industrial data is preprocessed in sequence by the spliced ​​multiple commonly used algorithm calculation components.

[0086] Based on pre-packaged analysis algorithm computing components, the pre-processed industrial data is subjected to dimensional transformation, and based on pre-packaged feature computing components, key features are extracted from the dimensionally transformed industrial data.

[0087] Based on pre-packaged learning algorithm computing components, a corresponding mechanism model is constructed according to the key features in the industrial data after dimensional transformation. The mechanism model is then debugged based on drag-and-drop triggers of various configuration information in the mechanism model, so as to complete the visualization construction of the mechanism model.

[0088] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device and medium embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the description of the method embodiments.

[0089] The foregoing has described specific embodiments of this application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0090] The devices and media provided in this application are one-to-one with the methods. Therefore, the devices and media also have similar beneficial technical effects as their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0091] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0092] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0093] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0094] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0095] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0096] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0097] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0098] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0099] The above description is merely an embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of this application should be included within the scope of the claims of this application.

Claims

1. A method for constructing a visual model of a mechanism based on industrial data, characterized in that, The method includes: Based on JDBC technology and database open interfaces, the local database is connected to the industrial dataset, and industrial data in the industrial dataset is obtained. The data types corresponding to the industrial data are determined, and the arrangement order of multiple pre-encapsulated commonly used algorithm calculation components is determined according to the data processing logic corresponding to the data types of the industrial data; the data types of the industrial data include time series data and image data; In the visualization interface, multiple commonly used algorithm calculation components corresponding to the industrial data are spliced ​​together in the order of the arrangement by dragging and dropping, and the industrial data is preprocessed sequentially by the spliced ​​multiple commonly used algorithm calculation components. Based on pre-packaged analysis algorithm computing components, the pre-processed industrial data is subjected to dimensional transformation, and based on pre-packaged feature computing components, key features are extracted from the dimensionally transformed industrial data. Based on the pre-packaged learning algorithm computing components, a corresponding mechanism model is constructed according to the key features in the industrial data after dimensional transformation. The mechanism model is then debugged based on the drag-and-drop trigger of each configuration information in the mechanism model, so as to complete the visualization construction of the mechanism model. The method, based on JDBC technology and open database interfaces, connects the local database with the industrial dataset and obtains industrial data from the industrial dataset, specifically including: Based on preset sensors and production equipment in the industrial scenario, the time-series data in the industrial scenario is collected to the IoT platform through the first preset protocol. Based on JDBC technology and database open interface, the local database is connected to the IoT platform to obtain the time-series data in the IoT platform. And through a second preset protocol, based on the preset production line quality inspection equipment in the industrial scenario, image data in the industrial scenario is collected to the data platform, and based on JDBC technology and database open interface, the local database is connected to the data platform to obtain image data in the data platform; The process involves determining the data type corresponding to the industrial data and, based on the data processing logic corresponding to the data type, determining the arrangement order of multiple pre-packaged commonly used algorithm calculation components. Specifically, this includes: Analyze several industrial data points in the industrial dataset, determine the data types corresponding to the several industrial data points, and determine the data processing logic corresponding to the industrial data based on the data types corresponding to the industrial data. According to the data processing logic corresponding to the industrial data, among a number of pre-packaged common algorithm calculation components, multiple common algorithm calculation components corresponding to the industrial data are identified, and the arrangement order among the multiple common algorithm calculation components corresponding to the industrial data is determined according to the data processing logic corresponding to the industrial data. The pre-packaged analysis algorithm calculation component performs dimensional transformation on the preprocessed industrial data, specifically including: Based on pre-packaged analysis algorithm computing components, data analysis is performed on pre-processed industrial data, and the data dimensions corresponding to each pre-processed industrial data are determined. The analysis algorithm calculation component includes at least: a principal component analysis algorithm calculation component, a linear discriminant analysis algorithm calculation component, and a factor analysis algorithm calculation component; Identify at least one industrial data point whose dimension is greater than a preset dimension threshold, and perform dimensionality reduction processing on the at least one industrial data point to achieve dimensionality transformation of the industrial data. The pre-packaged feature calculation component extracts key features from the dimensionally transformed industrial data, specifically including: Based on a pre-packaged feature calculation component, the feature attributes corresponding to the industrial data after dimensional transformation are determined, and the degree of influence of each feature attribute in the industrial data on the mechanism model is determined respectively. Based on the degree of influence of each feature attribute on the mechanism model, the feature weight of the corresponding feature attribute in the industrial data is determined, and at least one feature attribute with a feature weight greater than a preset weight threshold is identified, so as to use the at least one feature attribute as a key feature.

2. The method for constructing a visual mechanism model based on industrial data according to claim 1, characterized in that, The process of preprocessing the industrial data sequentially using multiple commonly used algorithm computing components after splicing specifically includes: Read the industrial data from the industrial dataset and parse the industrial data in the industrial dataset to obtain the corresponding parsing results; Based on the analysis results corresponding to each industrial data, a data quality analysis report corresponding to the industrial data is determined, and several industrial data that meet the quality standards in the data quality analysis report are identified. According to a preset splitting ratio and through random sampling, the several industrial data are split into a training dataset and a test dataset with the corresponding splitting ratio.

3. The method for constructing a visual mechanism model based on industrial data according to claim 1, characterized in that, After constructing a corresponding mechanism model based on the key features in the industrial data after dimensional transformation, the method, based on the pre-packaged learning algorithm computing component, further includes: Based on the pre-packaged model verification calculation component, the independent variable industrial data in the pre-processed test dataset is input into the mechanism model, and the corresponding test dependent variable industrial data is obtained. The industrial data of the test dependent variable is compared with the industrial data of the preset target dependent variable, and the accuracy of the mechanism model is determined based on the corresponding similarity.

4. The method for constructing a visual mechanism model based on industrial data according to claim 1, characterized in that, After constructing a corresponding mechanism model based on the key features in the industrial data after dimensional transformation, the method, based on the pre-packaged learning algorithm computing component, further includes: Based on cross-validation, the preprocessed test dataset is split into a preset number of mutually exclusive subsets, and a first number of mutually exclusive subsets are used as the training subset, and a second number of mutually exclusive subsets are used as the test subset; wherein, the preset number is the sum of the first number and the second number; The test results corresponding to the first number of training subsets and the second number of test subsets are determined, and the generalization error value corresponding to the mechanism model is determined based on the average value of the specified number of test results.

5. A device for constructing a mechanism model visualization based on industrial data, characterized in that, The device includes: At least one processor; And, a memory communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform the mechanism model visualization construction method based on industrial data as described in any one of claims 1-4.

6. A non-volatile computer storage medium storing computer-executable instructions, characterized in that, The computer-executable instructions are set as follows: The method for constructing a mechanism model visualization based on industrial data as described in any one of claims 1-4.

Images