Industrial model automatic deployment and updating method, system and equipment and storage medium

Abstract

The invention relates to an industrial model automatic deployment and updating method, system and device and a storage medium, and the method comprises the steps: carrying out the preprocessing and cleaning of various types of original data in a specified production environment, and obtaining training data; the method comprises the following steps: generating a corresponding industrial model for a specified production environment, training the industrial model to obtain a trained industrial model, automatically deploying the industrial model related to a deployment instruction and a data acquisition tool to the corresponding production environment through a cloud edge cooperation technology according to the deployment instruction, and carrying out real-time monitoring according to real-time data in the production environment. And the data acquisition tool performs real-time acquisition on parameters during operation of the industrial model to obtain operation parameters of the industrial model, evaluates the industrial model according to the operation parameters of the industrial model, and performs online optimization on the industrial model if an evaluation result meets an optimization condition. By adopting the method, the industrial production environment can be effectively detected efficiently and intelligently.

Description

Technical Field

[0001] The present application relates to the technical field of deep learning and industrial data evaluation, and in particular to an industrial model automatic deployment and update method, system, device and storage medium. Background Art

[0002] In recent years, with the rise of the concept of Industry 4.0, intelligent manufacturing has become an important direction for the transformation of the global manufacturing industry. In this context, how to use advanced information technology to improve the efficiency of industrial production, reduce costs, and improve product quality has become the focus of many companies. In particular, deep learning, as a powerful artificial intelligence technology, has demonstrated its great potential in many fields such as image recognition and speech processing. However, in actual industrial applications, traditional model training and deployment methods face many challenges: First, due to the diversity and complexity of industrial data, traditional statistical analysis methods are difficult to effectively extract features, while deep learning can better process high-dimensional, unstructured data and provide more accurate prediction and decision support. However, the training of deep learning models often requires a lot of computing resources and support, which is a problem for traditional industrial enterprises with limited resources. Secondly, the update of industrial models usually relies on manual operation, which is not only time-consuming but also prone to errors. Especially in a dynamically changing industrial environment, the model needs to be quickly adjusted according to new data to maintain its effectiveness. After the trained industrial model is deployed, there is a lack of continuous monitoring and evaluation of the model operation. The current solutions mostly use periodic evaluation methods and log recording and analysis. Both methods have a certain lag and cannot provide real-time feedback on the model operation effect. Summary of the invention

[0003] Based on this, it is necessary to provide an industrial model automatic deployment and update method, system, device and storage medium that can improve the efficiency of industrial production in response to the above technical problems.

[0004] An industrial model automatic deployment and update method, the method comprising: Acquire various types of raw data in a specified production environment, obtain training data after preprocessing and cleaning the raw data, and construct a corresponding training data set; Using online development tools to generate a corresponding industrial model for the specified production environment; Using the training data set to train the industrial model to obtain a trained industrial model; According to the deployment instructions, the industrial models and data collection tools related to the deployment instructions are automatically deployed to the corresponding production environment through cloud-edge collaboration technology, and real-time monitoring is performed based on the real-time data in the production environment; The data acquisition tool acquires the parameters of the industrial model during operation in real time to obtain the operating parameters of the industrial model; The industrial model is evaluated according to the operating parameters of the industrial model, and if the evaluation result meets the optimization condition, the industrial model is optimized online.

[0005] In one embodiment, the multiple types of raw data in the specified generation environment include sensor data, video data, and business data.

[0006] In one embodiment, the online development tool includes canvas modeling and notebook modeling.

[0007] In one embodiment, the industrial model operation parameters are pre-set model performance indicators, including feasibility, failure rate and response time.

[0008] In one embodiment, evaluating the industrial model according to the industrial model operating parameters includes: Based on the industrial model operation parameters and the model operation evaluation grade model, the operation status of the industrial model is evaluated, and the evaluation results include good operation status, average operation status, and poor operation status; If the evaluation result is that the operating state is deficient, the optimization condition is met.

[0009] In one embodiment, when the industrial model is optimized online, the real-time data currently acquired in the production environment is used as training data to perform online training on the industrial model to obtain an optimized industrial model.

[0010] In one embodiment, multiple industrial models are generated for different production environments, and the specified industrial models are deployed to the corresponding production environments according to the deployment instructions.

[0011] The present application also provides an industrial model automatic deployment and update system, the system comprising: The edge-side data collection and preprocessing module is used to obtain various types of raw data in a specified production environment, preprocess and clean the raw data to obtain training data, and construct a corresponding training data set; A cloud-side model development module is used to provide online development tools to generate corresponding industrial models for the specified production environment; A cloud-side model training and optimization module, used to train the industrial model using the training data set to obtain a trained industrial model; The cloud-side model deployment module is used to deploy the industrial model and data acquisition tool related to the deployment instruction to the corresponding production environment through the cloud-edge collaboration module according to the deployment instruction, perform real-time monitoring according to the real-time data in the production environment, and use the data acquisition tool to collect the parameters of the industrial model during operation in real time to obtain the industrial model operation parameters; The cloud-side model management and operation monitoring module is used to evaluate the industrial model according to the industrial model operation parameters. If the evaluation result meets the optimization conditions, the model training and optimization module will perform online optimization on the industrial model.

[0012] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented: Acquire various types of raw data in a specified production environment, obtain training data after preprocessing and cleaning the raw data, and construct a corresponding training data set; Using online development tools to generate a corresponding industrial model for the specified production environment; Using the training data set to train the industrial model to obtain a trained industrial model; According to the deployment instructions, the industrial models and data collection tools related to the deployment instructions are automatically deployed to the corresponding production environment through cloud-edge collaboration technology, and real-time monitoring is performed based on the real-time data in the production environment; The data acquisition tool acquires the parameters of the industrial model during operation in real time to obtain the operating parameters of the industrial model; The industrial model is evaluated according to the operating parameters of the industrial model, and if the evaluation result meets the optimization condition, the industrial model is optimized online.

[0013] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps: Acquire various types of raw data in a specified production environment, obtain training data after preprocessing and cleaning the raw data, and construct a corresponding training data set; Using online development tools to generate a corresponding industrial model for the specified production environment; Using the training data set to train the industrial model to obtain a trained industrial model; According to the deployment instructions, the industrial models and data collection tools related to the deployment instructions are automatically deployed to the corresponding production environment through cloud-edge collaboration technology, and real-time monitoring is performed based on the real-time data in the production environment; The data acquisition tool acquires the parameters of the industrial model during operation in real time to obtain the operating parameters of the industrial model; The industrial model is evaluated according to the operating parameters of the industrial model, and if the evaluation result meets the optimization condition, the industrial model is optimized online.

[0014] The above-mentioned industrial model automatic deployment and update method, system, device and storage medium obtain training data by preprocessing and cleaning various types of raw data in the specified production environment, and construct a corresponding training data set. The online development tool is used to generate the corresponding industrial model for the specified production environment, and the industrial model is trained with the training data set to obtain a trained industrial model. According to the deployment instruction, the industrial model and the data acquisition tool related to the deployment instruction are automatically deployed to the corresponding production environment through the cloud-edge collaborative technology, and real-time monitoring is performed according to the real-time data in the production environment. The data acquisition tool collects the parameters of the industrial model in real time during operation to obtain the industrial model operation parameters, and the industrial model is evaluated according to the industrial model operation parameters. If the evaluation result meets the optimization conditions, the industrial model is optimized online. This method can effectively detect industrial production scenarios efficiently and intelligently. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic diagram of a process of an industrial model automatic deployment and update method in one embodiment; Figure 2 It is a structural block diagram of an industrial model automatic deployment and update system in one embodiment; Figure 3 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0017] In the existing technology, the deployment of many industrial models still requires manual intervention, including selecting the best model version, configuring parameters, monitoring the running status, etc. This manual operation not only increases the workload, but also easily introduces human errors, reducing the reliability and efficiency of the entire process. In this application, Figure 1 As shown, an industrial model automatic deployment and update method is proposed, which specifically includes the following steps: Step S100, obtaining various types of raw data in a specified production environment, preprocessing and cleaning the raw data to obtain training data, and constructing a corresponding training data set.

[0018] Step S110: using an online development tool to generate a corresponding industrial model for a specified production environment.

[0019] Step S120: train the industrial model using the training data set to obtain a trained industrial model.

[0020] Step S130, according to the deployment instructions, the industrial models and data collection tools related to the deployment instructions are automatically deployed to the corresponding production environment through cloud-edge collaboration technology, and real-time monitoring is performed based on the real-time data in the production environment.

[0021] Step S140: The data collection tool collects the parameters of the industrial model during operation in real time to obtain the industrial model operation parameters.

[0022] Step S150, evaluating the industrial model according to the industrial model operation parameters, and if the evaluation result meets the optimization condition, performing online optimization on the industrial model.

[0023] In this embodiment, a method for industrial model training and automatic deployment based on deep learning technology is proposed. This method makes full use of the powerful computing power of cloud computing to accelerate model training, and realizes the automatic distribution and deployment of models and model operation data collection tools through cloud-edge collaboration technology, thus providing a new solution for industrial intelligence.

[0024] In step S100, multiple types of data from a specified production environment can be collected by edge devices set in the environment, and the production environment can be monitored based on the multiple types of data. Therefore, in this method, the data used to train the industrial model is also multiple types of historical data from a production environment, or real-time data. The multiple types of raw data in the specified production environment include sensor data, video data, and business.

[0025] In this embodiment, the production environment refers to the environment in which the industrial model actually runs. In the following example, the track detection model actually runs in the track detection system to generate the corresponding industrial model.

[0026] In this embodiment, these different types of data need to be preprocessed and cleaned to ensure the quality and accuracy of the data, and then a training data set is constructed based on the processed data.

[0027] Specifically, when preprocessing and cleaning different types of data, for structured data, rules such as null value detection and illegal data removal are used to clean the data. For unstructured data, such as text data and video data, machine learning is used for processing, including data cleaning, data conversion, data integration and data mining to ensure the quality and accuracy of the data, and then a training data set is constructed based on the processed data. Industrial model training data generally includes sensor data, production process data, environmental data, and maintenance and fault record data. These data come from various physical sensors, industrial control systems, and business systems in the production environment.

[0028] While the training data set is constructed in step S100, an online development tool can be used to construct a corresponding industrial model for a specified production environment in step S110. In this embodiment, the online development tool includes canvas modeling and notebook modeling, that is, the industrial model can be constructed by dragging and dropping or writing code.

[0029] In step S120, after constructing the training data set and the industrial model, the powerful computing power of cloud computing is used to train the industrial model through the training data set, and the learnable parameters in the working model are continuously adjusted until convergence, so as to obtain an industrial model capable of evaluating the status of the production environment according to the input data. Specifically, according to the above three steps, by utilizing relevant data of different production environments, multiple industrial models corresponding to different production environments can be trained. In the subsequent industrial model deployment process, one of the industrial models can be deployed to the corresponding production environment according to specific deployment instructions.

[0030] Furthermore, the production environments are all in the field of industrial production, including but not limited to industrial manufacturing scenarios, chemical industry, general scenarios, etc. Among them, the industrial manufacturing scenario refers to the use of statistical process control (SPC) models to monitor key indicators in the production process, promptly identify quality problems and take corrective measures. The chemical industry uses chemical kinetic models to simulate the chemical reaction process in the reactor and optimize the reaction conditions to improve the yield. The general scenario refers to the use of condition monitoring data and machine learning models to predict equipment failures and take maintenance measures in advance.

[0031] In step S130, after receiving the deployment instruction, the industrial model specified in the deployment instruction is deployed to the corresponding production environment.

[0032] Specifically, for a trained model, after defining the attributes or parameters of the model at runtime, such as input and output parameters. Then, through the cloud-edge collaboration method, it is sent to the edge device with one click in the platform. After sending, the model service will be automatically started according to the running script. Different from the traditional method: connecting to the device through various software, deploying the model to the device, and debugging after deployment. In this automated deployment method, the underlying differences between various types of devices are shielded. As long as the device can be connected to the platform, it can run the industrial model as an edge environment, reducing a lot of manual operations.

[0033] In this embodiment, multiple industrial models are generated for different production environments, and the specified industrial models are deployed to the corresponding production environments according to the deployment instructions.

[0034] In this embodiment, when the industrial model is deployed, a suitable data collection tool is also selected for synchronous deployment.

[0035] Specifically, select a collection tool based on the data you need to collect. In this example, you need to collect the CPU usage and memory usage of the model running environment, so you choose the system indicator collection tool. At the same time, you also need to collect indicators such as the model response time and error rate, so you choose the application indicator collection tool accordingly. In addition, there are also collection tools such as middleware data collectors and database indicator collectors.

[0036] Furthermore, while the industrial model is monitoring the production environment in real time, in step S140, a synchronously deployed data collection tool also collects the operating parameters of the industrial model in real time. The collected operating parameters are pre-set model performance indicators, including feasibility, failure rate and response time. At the same time, the data collection tool also collects model basic information management, model version, model operation status and update strategy.

[0037] In this embodiment, the basic information includes the model name, model domain, model version, model description, model performance indicators, and model update rules. The model operation status is evaluated based on the collected operation data using the operation status evaluation model. The update strategy defines the rules for automatic model updates, such as triggering the model update process when the operation status is general. This process is to collect the latest model operation environment data, form a data set, retrain the model, and redeploy the trained and updated model to the edge device.

[0038] Considering that after the industrial model is deployed to the corresponding production environment, the production environment will change over time. After a period of time, the industrial model that was suitable for the previous production environment may no longer be suitable for the current situation. This situation is likely to cause the current industrial model to no longer be able to perform effective detection and cannot give accurate prediction results for the production environment status.

[0039] In this embodiment, in view of the above situation, in step S150, an evaluation is performed based on the industrial model operating parameters collected in real time by the data collection tool. If the evaluation result shows that optimization is required, the industrial model is optimized in real time.

[0040] Specifically, when evaluating the industrial model operating parameters collected in real time by the data collection tool: the operating status of the industrial model is evaluated based on the industrial model operating parameters and the model operation evaluation level model, and the evaluation results include good operating status, general operating status, and poor operating status. If the evaluation result is that the operating status is poor, the optimization conditions are met.

[0041] Specifically, the operation status of the model can be evaluated by using two evaluation methods: mean and benchmark comparison. The average value over a period of time is calculated to measure the general level of the model output, and the real-time data is compared with historical data or data under ideal conditions to evaluate performance changes.

[0042] Furthermore, when the industrial model is optimized online, the real-time data currently acquired in the production environment is used as training data to perform online training on the industrial model to obtain an optimized industrial model.

[0043] Specifically, when the model evaluation result is insufficient, the model update operation will be triggered. The latest rail inspection vehicle data is obtained through the data acquisition and preprocessing module, and then the data is preprocessed and cleaned, and then the model is retrained, and the optimized model is pushed to the model management model module. If automatic update is selected in the model update rule, the optimized model will be automatically updated to the terminal side through the cloud-edge collaboration module.

[0044] In this embodiment, after the optimized industrial model is obtained, its version is marked and stored. Then, through the optimization instruction, the optimized version replaces the old version, and the parameters of multiple versions of the industrial model are retained to facilitate traceability during later maintenance.

[0045] In the above-mentioned method for automatic deployment and update of industrial models, training data is obtained by preprocessing and cleaning various types of raw data in a specified production environment, and a corresponding training data set is constructed. An online development tool is used to generate a corresponding industrial model for a specified production environment, and the industrial model is trained using the training data set to obtain a trained industrial model. According to the deployment instruction, the industrial model and the data acquisition tool related to the deployment instruction are automatically deployed to the corresponding production environment through cloud-edge collaboration technology, and real-time monitoring is performed according to the real-time data in the production environment. The data acquisition tool collects the parameters of the industrial model in real time during operation to obtain the industrial model operation parameters, and the industrial model is evaluated according to the industrial model operation parameters. If the evaluation result meets the optimization condition, the industrial model is optimized online. This method is based on deep learning technology to carry out industrial model training, improve the speed and efficiency of model training, and thus achieve rapid iteration and update. At the same time, the automatic deployment of the model is realized through automated tools and processes, which reduces the need for manual operation and improves the flexibility and response speed of deployment. The model is also updated by monitoring and evaluating the deployed industrial model to adapt to the new environment or optimize performance.

[0046] This method utilizes deep learning and edge-cloud collaborative technology to increase the speed of model training, reduce the complexity of model deployment, and shorten the model deployment time, thereby improving the efficiency of industrial model research and development and accelerating the input of models into the production environment.

[0047] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0048] In this application, if Figure 2As shown, an industrial model automatic deployment and update system is also provided, which includes: an edge-side data acquisition and preprocessing module, a cloud-side model development module, a cloud-side model training and optimization module, a cloud-side model deployment module, and a cloud-side model management and operation monitoring module. Among them, the edge-side data acquisition and preprocessing module is used to obtain various types of raw data in a specified production environment, obtain training data after preprocessing and cleaning the raw data, and construct a corresponding training data set. The cloud-side model development module is used to use online development tools to generate corresponding industrial models for a specified production environment. The cloud-side model training and optimization module is used to train the industrial model using the training data set to obtain a trained industrial model. The cloud-side model deployment module is used to deploy the industrial model and data acquisition tool related to the deployment instruction to the corresponding production environment through the cloud-edge collaborative model according to the deployment instruction, perform real-time monitoring according to the real-time data in the production environment, and the data acquisition tool collects the parameters of the industrial model in real time during operation to obtain the industrial model operation parameters. The cloud-side model management and operation monitoring module is used to evaluate the industrial model based on the industrial model operation parameters. If the evaluation result meets the optimization conditions, the cloud-side model training and optimization module will perform online optimization on the industrial model.

[0049] Furthermore, when there are multiple industrial models, these models are placed in the cloud-side model management and operation monitoring module. The cloud-side model deployment module extracts the corresponding industrial model from the model management and operation monitoring module for deployment according to the deployment instructions.

[0050] Specifically, the edge-side data collection and preprocessing module includes an edge-side data collection unit, an edge-side data preprocessing unit, and an edge-side data management unit. The edge-side data collection unit is used to collect the operating data generated by the model in the production environment, and the data preprocessing unit receives the operating data, converts and cleans the data to form a training data set. The edge-side data management unit is used to manage the data set and store the data set in the database of the cloud server.

[0051] Specifically, the cloud-side model development module provides online development tools to create industrial models using data-driven modeling technology based on the data collected from the specified production environment.

[0052] Specifically, the cloud-side model training and optimization module includes a data asset module, a task management module, and a model training module. The data asset module manages various metadata and data sets, and the task management module includes model training task management and scheduled task management. The model training module includes a model training unit and a model monitoring unit. The model training unit carries out model training tasks, and the model detection unit monitors the model training process.

[0053] Furthermore, the cloud-edge collaboration module automatically deploys the corresponding industrial model and data collection tools to the production environment. Among them, the cloud-edge collaboration module includes an edge control module, an end-side deployment module, and a cloud-edge communication module. The edge control module is used to receive deployment instructions and send them to the end-side deployment module through the cloud-edge communication module. The cloud-edge communication module is used for message communication between the edge control module and the end-side deployment module. It includes the sending of industrial models and the transmission of industrial model operation data. The model end-side deployment module executes the deployment instructions and deploys the received industrial models and data collection tools.

[0054] In one of the embodiments, taking the construction of an industrial model for track defect tracking, namely a track defect tracking model, to achieve tracking and prediction of track health as an example, each module in the above-mentioned industrial model automatic deployment and update system is further explained.

[0055] First, the raw data of a rail inspection vehicle in the subway line is collected through the edge-side data collection and preprocessing module. The data content includes rail parameter measurement values, defect records, TQI, etc. These data are preprocessed and cleaned to ensure the quality and accuracy of the data, and the corresponding training data set is constructed.

[0056] At the same time, in the cloud-side model development module, modeling is performed based on various indicators that affect the health of the track. In this implementation, the drag-and-drop method can be used to create and develop the model to obtain the corresponding industrial model.

[0057] Next, the processed training data set in the cloud-side data acquisition and preprocessing module and the industrial model built in the model development module are transferred to the cloud-side model training and optimization module. For the industrial model created for the first time, only model training is required. After the training is completed, the available model is output to the cloud-side model management module to train the industrial model using the training data set and manage the trained track defect tracking model.

[0058] The cloud-side model management and operation monitoring module also manages the version, parameters, and hyperparameters of the rail defect tracking model, and monitors the operating status of the deployed industrial model. At the same time, in this module, a variety of data collection tools are built-in to collect structured and unstructured data in the production environment.

[0059] In this implementation, the version of the track defect tracking model is the initialization version, and its corresponding parameters are all from the model training module. In order to evaluate the operating efficiency of the model, it is necessary to collect data such as the delay, throughput, and proportion of track defects recognized by the model. Model data collection tools and network collection tools can be used.

[0060] Furthermore, the cloud-side model deployment and reasoning module performs automated deployment of the track defect tracking model. The track defect tracking model is obtained from the cloud-side model management module according to the deployment instructions. If there are multiple versions, you can select the version that needs to be applied. Then select the model data acquisition tool and the network acquisition tool, and deploy the model and the data acquisition tool together to the production environment through the cloud-edge collaboration module. In the production environment, the equipment health index is calculated based on the current key parameters of the equipment, so as to realize the real-time display of the track health of the line, that is, the health status.

[0061] At the same time, through the data collection tool, the predefined data in the tool is collected and transmitted back to the cloud-side model management and operation monitoring module.

[0062] Next, in the cloud-side model management and operation monitoring module, after the industrial model related to track defect tracking is deployed to the production environment, the current health status of the track can be obtained based on real-time data. Generally speaking, the situation given by the industrial model is consistent with the actual situation. However, due to factors such as the increase in the age of the track and changes in the operating environment, the results of the industrial model deviate from the actual situation. If this situation causes the accuracy of the model to drop to the set threshold, an industrial model operation warning will be triggered. When a warning appears, the latest data can be used to optimize and retrain the model so that it is higher than the set accuracy under the new conditions.

[0063] At this time, the training data currently collected and processed in the edge-side data collection and preprocessing module is pushed to the cloud-side model training and optimization module, and the model to be optimized is selected from the cloud-side model management and operation monitoring module and passed to the cloud-side model training and optimization module. The industrial model is trained again using the new training data, that is, the data collected in real time during the current period, to obtain a new version of the industrial model after iteration. The new version of the model is pushed to the cloud-side model management module. Its version number, such as the iterative optimization version, is recorded, and its corresponding parameters are stored in this module.

[0064] Finally, the track defect tracking model is obtained from the cloud-side model management and operation monitoring module, and the iterative optimized version is selected to update the model to the production environment through the cloud-edge collaboration module to realize the replacement of the old and new models.

[0065] For the specific limitations of the industrial model automatic deployment and update device, please refer to the limitations of the industrial model automatic deployment and update method above, which will not be repeated here. Each module in the above-mentioned industrial model automatic deployment and update device can be implemented in whole or in part through software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0066] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, an industrial model automatic deployment and update method is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0067] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0068] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented: Acquire various types of raw data in a specified production environment, obtain training data after preprocessing and cleaning the raw data, and construct a corresponding training data set; Using online development tools to generate a corresponding industrial model for the specified production environment; Using the training data set to train the industrial model to obtain a trained industrial model; According to the deployment instructions, the industrial models and data collection tools related to the deployment instructions are automatically deployed to the corresponding production environment through cloud-edge collaboration technology, and real-time monitoring is performed based on the real-time data in the production environment; The data acquisition tool acquires the parameters of the industrial model during operation in real time to obtain the operating parameters of the industrial model; The industrial model is evaluated according to the operating parameters of the industrial model, and if the evaluation result meets the optimization condition, the industrial model is optimized online.

[0069] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented: Acquire various types of raw data in a specified production environment, obtain training data after preprocessing and cleaning the raw data, and construct a corresponding training data set; Using online development tools to generate a corresponding industrial model for the specified production environment; Using the training data set to train the industrial model to obtain a trained industrial model; According to the deployment instructions, the industrial models and data collection tools related to the deployment instructions are automatically deployed to the corresponding production environment through cloud-edge collaboration technology, and real-time monitoring is performed based on the real-time data in the production environment; The data acquisition tool acquires the parameters of the industrial model during operation in real time to obtain the operating parameters of the industrial model; The industrial model is evaluated according to the operating parameters of the industrial model, and if the evaluation result meets the optimization condition, the industrial model is optimized online.

[0070] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0071] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0072] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. An industrial model automatic deployment and update method, characterized in that: The method comprises: Acquire various types of raw data in a specified production environment, obtain training data after preprocessing and cleaning the raw data, and construct a corresponding training data set; Using online development tools to generate a corresponding industrial model for the specified production environment; Using the training data set to train the industrial model to obtain a trained industrial model; According to the deployment instructions, the industrial models and data collection tools related to the deployment instructions are automatically deployed to the corresponding production environment through cloud-edge collaboration technology, and real-time monitoring is performed based on the real-time data in the production environment; The data acquisition tool acquires the parameters of the industrial model during operation in real time to obtain the operating parameters of the industrial model; The industrial model is evaluated according to the operating parameters of the industrial model, and if the evaluation result meets the optimization condition, the industrial model is optimized online.

2. The method for automatic deployment and updating of industrial models according to claim 1, characterized in that: The multiple types of raw data in the specified generation environment include sensor data, video data, and business data.

3. The method for automatic deployment and updating of industrial models according to claim 2, characterized in that: The online development tools include canvas modeling and notebook modeling.

4. The method for automatic deployment and updating of industrial models according to claim 3, characterized in that: The industrial model operation parameters are pre-set model performance indicators, including feasibility, failure rate and response time.

5. The method for automatic deployment and updating of industrial models according to claim 4, characterized in that: Evaluating the industrial model according to the industrial model operation parameters includes: Based on the industrial model operation parameters and the model operation evaluation grade model, the operation status of the industrial model is evaluated, and the evaluation results include good operation status, average operation status, and poor operation status; If the evaluation result is that the operating state is deficient, the optimization condition is met.

6. The method for automatic deployment and updating of industrial models according to claim 5, characterized in that: When the industrial model is optimized online, the real-time data currently acquired in the production environment is used as training data to perform online training on the industrial model to obtain an optimized industrial model.

7. The method for automatic deployment and updating of industrial models according to any one of claims 1 to 6, characterized in that: Multiple industrial models are generated for different production environments, and the specified industrial models are deployed to the corresponding production environments according to the deployment instructions.

8. The industrial model automatic deployment and update system is characterized by: The system comprises: The edge-side data collection and preprocessing module is used to obtain various types of raw data in a specified production environment, preprocess and clean the raw data to obtain training data, and construct a corresponding training data set; A cloud-side model development module is used to provide online development tools to generate corresponding industrial models for the specified production environment; A cloud-side model training and optimization module, used to train the industrial model using the training data set to obtain a trained industrial model; The cloud-side model deployment module is used to deploy the industrial model and data acquisition tool related to the deployment instruction to the corresponding production environment through the cloud-edge collaboration module according to the deployment instruction, perform real-time monitoring according to the real-time data in the production environment, and use the data acquisition tool to collect the parameters of the industrial model during operation in real time to obtain the industrial model operation parameters; The cloud-side model management and operation monitoring module is used to evaluate the industrial model according to the industrial model operation parameters. If the evaluation result meets the optimization conditions, the model training and optimization module will perform online optimization on the industrial model.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images