Artificial intelligence-based abnormal production control method and device, and computer equipment

Abstract

Embodiments of the present application disclose an abnormal production control method and device based on artificial intelligence and a computer device. The method comprises: obtaining production-related data; performing production state recognition according to the production-related data to obtain a recognition result; determining whether the recognition result is an abnormal state that cannot be handled by a conventional program; if so, inputting state features, production-related data and a problem object / event into a problem diagnosis model to diagnose a problem cause and obtain a diagnosis result; inputting the diagnosis result, state features and the problem object / event into a control content analysis model to determine a controlled object and a control action; and selecting automatic control or manual control according to the determination result by using an internal control strategy to perform action control on the controlled object. By implementing the method of the embodiments of the present application, an abnormal automatic recognition and repair mechanism can be added in conventional automatic production control, the timeliness of abnormal handling is improved, and production losses caused by abnormalities are reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to a production control method, in particular to an abnormal production control method based on artificial intelligence, device and computer equipment. BACKGROUND

[0002] In the traditional production execution process, two control modes are usually adopted, one is manual control, and the other is programmed automatic control. The first mode relies on manual management of the production process, including normal and abnormal situations, but the efficiency is low, and it is difficult to manage in large-scale production. The second mode is automatically managed through a control program, which can achieve efficient production in normal state and is suitable for large-scale promotion. However, when an exception beyond the program processing range occurs, manual intervention is still required for repair, resulting in a decline in timeliness and affecting overall production efficiency, so this mode is actually a semi-automatic production method.

[0003] Therefore, it is necessary to design a new method to increase the abnormal automatic identification and repair mechanism in the conventional automatic production control, improve the timeliness of abnormal handling, and reduce the production loss caused by abnormality. SUMMARY

[0004] The present application aims to overcome the defects of the prior art and provide an abnormal production control method based on artificial intelligence, device and computer equipment.

[0005] To achieve the above-mentioned purpose, the following technical solutions are adopted: an abnormal production control method based on artificial intelligence, comprising:

[0006] obtaining production-related data;

[0007] performing production state recognition according to the production-related data to obtain a recognition result; wherein the recognition result includes any one of a producible state and an abnormal state that cannot be handled by a conventional program, and when the recognition result is a problem object / event corresponding to the abnormal state that cannot be handled by the conventional program;

[0008] determining whether the recognition result is an abnormal state that cannot be handled by a conventional program;

[0009] if the recognition result is an abnormal state that cannot be handled by a conventional program, inputting state features, production-related data and a problem object / event into a problem diagnosis model to diagnose the cause of the problem, to obtain a diagnosis result;

[0010] inputting the diagnosis result, the state features and the problem object / event into a control content analysis model to determine a controlled object and a control action, to obtain a determination result; the determination result includes a controlled object, a control action and a confidence level;

[0011] Performing simulation according to the determination result to obtain a simulation result;

[0012] Determining whether the simulation results meet the set requirements;

[0013] If the simulation result meets the set requirements, then according to the determination result, the built-in control strategy is used to select automatic control or manual control to control the motion of the controlled object;

[0014] Feeding back the simulation results to the problem diagnosis model so that the problem diagnosis model can be learned and updated;

[0015] If the simulation result does not meet the set requirements, the state characteristics, production-related data and problem objects / events are input into the problem diagnosis model to diagnose the cause of the problem and obtain a diagnosis result.

[0016] A further technical solution is: selecting automatic control or manual control by using a built-in control strategy according to the determination result to control the motion of the controlled object, including:

[0017] When the confidence level is less than a set threshold, the controlled object and the corresponding control action in the determination result are manually adjusted, and the built-in control strategy is used to select automatic control or manual control based on the adjusted controlled object and the corresponding control action to perform action control on the controlled object.

[0018] Its further technical solution is: the problem diagnosis model is obtained by extracting objects and events from production factors, production processes, and product data, establishing associations, and setting normal state features and abnormal state features as well as corresponding problem objects / events as sample sets to train the initial model.

[0019] Its further technical solution is: the initial model includes one of a rule-based model, a statistics-based model, a deep learning-based model and a supervised learning-based model.

[0020] Its further technical solution is: the control content analysis model is obtained by training a deep learning model with several problem objects / events, state characteristics, problem causes and corresponding controlled objects and control actions as sample sets.

[0021] Its further technical solution is: the deep learning model includes a deep reinforcement learning model based on value, strategy, and model.

[0022] A further technical solution is: after determining whether the recognition result is an abnormal state that cannot be handled by a conventional program, the method further includes:

[0023] If the recognition result is not an abnormal state that cannot be handled by conventional procedures, the controlled object is automatically controlled.

[0024] The present invention also provides an abnormal production control device based on artificial intelligence, comprising:

[0025] A data acquisition unit, used to acquire production-related data;

[0026] an identification unit, configured to identify the production status based on the production-related data to obtain an identification result; wherein the identification result includes any one of a processable production status and an abnormal status that cannot be processed by a conventional procedure, and, when the identification result is an abnormal status that cannot be processed by a conventional procedure, a problem object / event corresponding to the problem object / event;

[0027] A judgment unit, configured to judge whether the recognition result is an abnormal state that cannot be handled by a conventional program;

[0028] a cause diagnosis unit for inputting the state characteristics, production-related data, and problem objects / events into a problem diagnosis model to diagnose the cause of the problem and obtain a diagnosis result if the identification result indicates an abnormal state that cannot be handled by conventional procedures;

[0029] a control content analysis unit, configured to input the diagnosis result, the state characteristics, and the problem object / event into a control content analysis model to determine a controlled object and a control action, thereby obtaining a determination result; the determination result includes the controlled object, the control action, and a confidence level;

[0030] a simulation unit, configured to perform a simulation based on the determination result to obtain a simulation result;

[0031] a judgment unit, configured to judge whether the simulation result meets the set requirements; if the simulation result does not meet the set requirements, inputting the state characteristics, production-related data, and problem objects / events into the problem diagnosis model to diagnose the cause of the problem and obtain a diagnosis result;

[0032] A control unit, configured to, if the simulation result meets the set requirements, select automatic control or manual control based on the determination result using a built-in control strategy to control the motion of the controlled object;

[0033] The feedback unit is used to feed back the simulation results to the problem diagnosis model so that the problem diagnosis model can be learned and updated.

[0034] The present invention further provides a computer device, comprising a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the above method when executing the computer program.

[0035] The application further provides a storage medium storing a computer program, which is executed by a processor to implement the method.

[0036] Compared with the prior art, the application has the beneficial effects that: the application determines whether it is an abnormal state that cannot be handled by a regular program by acquiring production related data and performing state recognition. The recognition result includes state characteristics and corresponding problem objects / events. If it is an abnormal state that cannot be handled by a regular program, the related data is input into a problem diagnosis model to perform cause analysis, to generate a diagnosis result. Then, the result is transmitted to a control content analysis model to determine a controlled object and a control action, and to generate a determination result. Finally, according to the determination result, the system selects automatic control or manual control; the abnormal automatic recognition and repair mechanism is realized in the regular automatic production control, and the option of manual intervention is retained, so that the timeliness of abnormal processing is improved, the production loss caused by the abnormality is reduced, the characteristics of large-scale and efficient production are maintained, in addition, the simulation link is added, and the result of the control content analysis model and manual auxiliary determination is simulated and confirmed, so that the given answer is as correct as possible, and the control accuracy can be ensured.

[0037] The application will be further described below in combination with the drawings and specific embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0039] Figure 1 An application scenario schematic diagram of the abnormal production control method based on artificial intelligence provided by the embodiments of the application is shown in the figure.

[0040] Figure 2 A flowchart of the abnormal production control method based on artificial intelligence provided by the embodiments of the application is shown in the figure.

[0041] Figure 3 An abstract related object and event schematic diagram provided by the embodiments of the application is shown in the figure.

[0042] Figure 4 A schematic block diagram of the abnormal production control device based on artificial intelligence provided by the embodiments of the application is shown in the figure.

[0043] Figure 5 A schematic block diagram of the computer device provided by the embodiments of the application is shown in the figure. DETAILED DESCRIPTION

[0044] With reference to the accompanying drawings, the technical solutions in the embodiments of the present application will be clearly and completely described in the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0045] It should be understood that when used in the specification and the appended claims, the terms "comprise" and "include" indicate the presence of described features, integers, steps, operations, elements, and / or components, but do not exclude one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0046] It should also be understood that the terms used in the present application specification are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the present application specification and the appended claims, the singular forms "a", "an" and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0047] It should be further understood that the term "and / or" used in the present application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0048] Please refer to Figure 1 and Figure 2 , Figure 1 The application scenario schematic diagram of the abnormal production control method based on artificial intelligence provided by the embodiments of the present application. Figure 2 The schematic flowchart of the abnormal production control method based on artificial intelligence provided by the embodiments of the present application. The abnormal production control method based on artificial intelligence is applied to a server. The server interacts with a terminal to obtain corresponding production related data from the terminal, identifies the production state, and when an abnormality that cannot be handled by a conventional program occurs, diagnoses the problem and reason, and further determines the controlled object and the corresponding control action, and performs manual control or automatic control. At this time, the controlled object and the control action are the controlled object and the control action in the processing scheme of the problem object / event that appears abnormally, that is, the corresponding solution, so as to increase the abnormal automatic identification and repair mechanism in the conventional automatic production control, improve the timeliness of abnormal handling, and reduce the production loss caused by abnormality.

[0049] Figure 2 The flowchart of the abnormal production control method based on artificial intelligence provided by the embodiments of the present application. As shown in Figure 2 , the method comprises the following steps S110 to S170.

[0050] S110, acquire production-related data.

[0051] In this embodiment, the production-related data includes production factors, production processes, products, and data, and also includes state features of objects / events, such as rotation speed, etc. The production factors refer to resources such as raw materials, equipment, personnel, and energy. The production processes refer to process flow, operation steps, time nodes, and related parameters. The products refer to product specifications, quality standards, and performance indicators. The data refers to real-time data collected in production, historical records, and quality test results, etc.

[0052] S120, identify the production state according to the production-related data to obtain an identification result; wherein the identification result includes any one of a processable production state and an abnormal state that cannot be processed by a conventional program, and when the identification result is an abnormal state that cannot be processed by a conventional program, a problem object / event corresponding to the abnormal state.

[0053] In this embodiment, real-time monitoring and analysis are performed on the production-related data (such as production factors, processes, and product data) to identify the current production state. The identification result includes any one of a processable production state and an abnormal state that cannot be processed by a conventional program, and when the identification result is an abnormal state that cannot be processed by a conventional program, a problem object / event corresponding to the abnormal state.

[0054] In the process of identification, the normal and abnormal states of each device and the objects and associated events output by the device can be identified. Real-time monitoring of device data ensures that potential faults can be found in a timely manner.

[0055] Moreover, the identified states can be classified based on an embedded rule set to identify common abnormal patterns. When a new abnormal pattern or state is found, the rule set can be automatically expanded in combination with historical data to adapt to changing production environments.

[0056] Through multi-threading technology, the control program and the exception handling program are called in real time and concurrently to ensure efficient processing of various production situations. When an exception occurs, the system can quickly respond and take appropriate measures to reduce production downtime.

[0057] According to the incremental update part of the problem diagnosis model and the control content analysis model, and historical control data, synchronous learning is performed to identify new states. Through continuous updating, it is ensured that the identification process can adapt to new production conditions and technical changes.

[0058] Real-time identification of production status and anomalies enables rapid response and action to reduce downtime and improve overall production efficiency. Through data-driven identification and classification, management can make more accurate decisions and optimize production processes and resource allocation. Early identification of potential equipment failures or production anomalies can effectively reduce production risks and reduce maintenance and downtime costs. Through automatic expansion of rule sets and synchronous learning, the system can continuously adapt to new production environments and maintain efficient and accurate status identification capabilities. By promptly detecting and handling anomalies, it ensures that products meet quality standards, thereby improving customer satisfaction and market competitiveness.

[0059] This systematic production status identification and management method helps companies maintain flexibility and efficiency in complex production environments.

[0060] In this embodiment, normal state refers to the normal operating parameters (such as temperature, pressure, speed, etc.) of the equipment or production process, determined based on historical data and industry standards. Abnormal state refers to the deviation from the normal standard, which is usually defined by setting a threshold (such as an upper and lower limit).

[0061] S130, determining whether the recognition result is an abnormal state that cannot be handled by a conventional program;

[0062] S140. If the identification result is an abnormal state that cannot be handled by conventional procedures, the state characteristics, production-related data, and problem objects / events are input into a problem diagnosis model to diagnose the cause of the problem and obtain a diagnosis result.

[0063] In this embodiment, the diagnosis result refers to the cause of the abnormal situation.

[0064] Specifically, the problem diagnosis model is developed by extracting objects and events from production factors, production processes, and product data, establishing associations, and using normal state characteristics, abnormal state characteristics, and the corresponding problem objects / events as sample sets to train an initial model. Normal state characteristics refer to processes that can be handled by conventional programs, while abnormal state characteristics refer to processes that cannot be handled by conventional programs. The initial model can be a rule-based model, a statistics-based model, a deep learning-based model, or a supervised learning-based model.

[0065] In this embodiment, if Figure 3As shown, various production factors, processes, and products are modeled, abstracting related objects and events to simplify complex production processes and facilitate subsequent analysis. Each object and event is assigned a unique identifier and interconnected relationships are established to effectively manage the production process. Based on the company's quality standards and historical data, normal and abnormal state characteristics are defined for each type of object and event to help identify anomalies in production. Algorithmic modeling is performed on this basic data, and the model is trained to identify and predict states, improving its effectiveness and reliability. This model can be used for real-time monitoring and anomaly detection. Data is collected during actual production processes to initially train the problem diagnosis model. Dynamic incremental training is then performed through online learning to ensure the model continuously adapts to real-world conditions. After training, the model is fed with relevant object, event, abnormal state, and characteristic data, and it outputs potential cause of the problem and its confidence level.

[0066] Specifically, when detecting anomalies and diagnosing problems in the production process, the following models and algorithms can be selected:

[0067] Rule-based model: Utilizes expert systems or rule engines to perform anomaly detection based on preset rules.

[0068] Statistical-based models, such as principal component analysis (PCA), PCA combined with K-nearest neighbor (KNN), and isolation forest, are used to identify outliers.

[0069] Deep learning-based models: Use reinforcement learning models such as Deep Q-Network (DQN) and A2C to learn complex control strategies, or use sequence models such as RNN and LSTM to process time series data.

[0070] Supervised learning-based models: support vector machines (SVM), random forests, and XGBoost are used for classification.

[0071] The training process of the problem diagnosis model is as follows:

[0072] Offline training (based on static data):

[0073] Collect historical production data under normal and abnormal conditions and perform preprocessing (such as missing value filling and normalization); extract key features (such as temperature, pressure, and equipment status) and construct feature vectors; select a deep learning framework (such as TensorFlow or PyTorch), build a DQN or A2C model, and define the state space, action space, and reward function; use historical data for offline training, optimize model parameters, and evaluate performance (such as accuracy and recall).

[0074] Online Learning:

[0075] Apply the offline-trained model to the production environment; continuously acquire new data in the production environment; use online learning with new data to update model parameters regularly (such as daily or weekly); regularly evaluate model performance and adjust parameters or training strategies according to the results, and retrain or adjust the model architecture if necessary.

[0076] S150, input the diagnostic results, state features, and problem objects / events into the control content analysis model to determine the controlled objects and control actions, and obtain a determination result; the determination result includes controlled objects, control actions, and confidence levels.

[0077] In this embodiment, the control content analysis model is obtained by training a deep learning model using a sample set of problem objects / events, state features, problem causes, and corresponding controlled objects and control actions. The deep learning model includes value-based, policy-based, and model-based deep reinforcement learning models.

[0078] Controlled objects refer to objects that need to be controlled, and control actions refer to control methods to solve abnormal situations.

[0079] Specifically, the input of the control content analysis model includes problem objects / events, state features, and possible causes. These information help the model understand the current system state and potential problems.

[0080] Output: involves controlled objects and control actions, generates specific operation instructions, and facilitates the model to directly participate in the control of production processes. Such design enables the model to propose feasible solutions for specific production abnormalities, and ensures that the output results are easy to understand and execute.

[0081] Two-stage design of training of control content analysis model:

[0082] First stage: training based on static data, in the initial stage, use historical data to train the model, so that it has the basic ability to handle problems. Optimize the model through a large amount of historical data to ensure that it can effectively generalize to unseen data.

[0083] Second stage: online learning, after the model is deployed, continue to update it using newly generated data. This method enables the model to continuously adapt to environmental changes, thereby improving long-term performance. Combining the advantages of offline training and online learning, it can quickly deploy pre-trained models and continuously improve their performance after going online.

[0084] In addition, it also allows manual adjustment of model output and feedback of these adjustments to the model for further training. The benefits include:

[0085] Improve flexibility: when the model decision is not ideal, manual intervention can be used to correct it.

[0086] Enhanced learning: Collecting human-adjusted data helps the model make more accurate decisions in the future.

[0087] Promote human-machine collaboration: Make full use of human experience and machine learning capabilities to establish an effective collaboration model.

[0088] S160: Perform simulation according to the determination result to obtain a simulation result.

[0089] In this embodiment, the simulation result refers to a result obtained by performing an environmental simulation on the determination result.

[0090] Specifically, based on specific trigger conditions (such as the timeliness of exception handling and the severity of the problem), the system can choose to start or shut down the simulation environment to better meet actual control needs.

[0091] Specifically, simulation is to build a virtual environment that is consistent with the real environment and perform control in the virtual environment based on the determination results.

[0092] S170: Determine whether the simulation result meets the set requirements.

[0093] In this embodiment, the simulation results include two types: the abnormal situation is resolved, or the abnormal situation still exists. When the abnormal situation is resolved, it indicates that the simulation result meets the set requirements; otherwise, it indicates that the simulation result does not meet the set requirements.

[0094] S180: If the simulation result meets the set requirements, select automatic control or manual control using the built-in control strategy according to the determination result to control the motion of the controlled object.

[0095] Specifically, when the confidence level is less than a set threshold, the controlled object and the corresponding control action in the determination result are manually adjusted, and the built-in control strategy is used to select automatic control or manual control based on the adjusted controlled object and the corresponding control action to perform action control on the controlled object.

[0096] Automated control is the preferred approach in stable and predictable environments. This strategy, based on pre-set rules and algorithms, automatically adjusts and optimizes the operation of the controlled object, ensuring efficiency and consistency. The model analyzes input data, determines the optimal control strategy, and automatically implements it. For example, in a production line, automated control can monitor equipment status in real time and make adjustments, reducing human intervention.

[0097] When the environment is complex or unusual circumstances arise, manual control becomes necessary. Through manual intervention, operators can adjust the controlled object based on their experience and intuition, resolving issues that automated control struggles with. In these situations, the model identifies situations requiring manual intervention and provides recommendations to help operators make decisions.

[0098] Ultimately, the choice of control strategy depends on the system's real-time state, environmental dynamics, and task complexity. Built-in control strategies dynamically assess these factors and select the most appropriate control approach to ensure stable and flexible system operation. This allows the model to leverage the advantages of automation while also mobilizing human resources when necessary to achieve optimal results.

[0099] In dynamic data training, data from manual decision-making is used as incremental data for the control content analysis model. After obtaining the final controlled object and control action data, the exception handler selects automated or manual control based on the built-in control strategy. If automated control is selected, the control program is called to execute the action control. This control program has the following technical features:

[0100] It is capable of controlling the actions of controlled objects, which are derived from production factors, and the control actions are based on built-in rule sets. It has the ability to receive feedback, can perform remedial processing when the action is executed abnormally, and combine AI capabilities to enhance the processing mechanism. It supports multi-dimensional action control, including time, space, force and speed, etc. It has multi-threaded concurrent control capabilities, can control production factors across network segments, and supports interface authentication and encrypted transmission.

[0101] S190 , feeding back the simulation results to the problem diagnosis model so that the problem diagnosis model can be learned and updated.

[0102] In this embodiment, the simulation environment is a simulation system that is highly consistent with the real environment. This environment includes various objects, events, and states to ensure that it is as similar as possible to the actual operation scenario, such as Figure 1 This simulation environment is used to evaluate the response of the exception handler.

[0103] In this process, the exception handler assesses the situation based on feedback from the simulation environment. If the feedback is normal, the exception handler continues to execute subsequent processing steps according to the established logic. Furthermore, to enhance the system's intelligence, the exception handler transmits the simulation environment's feedback to the problem diagnosis model, enabling it to learn from these results and optimize future decision-making and processing capabilities.

[0104] However, if the feedback results are abnormal, the situation changes. In this case, the exception handler forwards the feedback results returned from the simulation environment to the problem diagnosis model. The problem diagnosis model combines the previously output problem causes with the simulation results to reanalyze the problem. During this process, the problem diagnosis model outputs new problem causes, allowing for more accurate identification and handling of the abnormal situation.

[0105] Through this feedback mechanism, the system can continuously learn and adjust, improve its response to abnormal situations, and ensure effective processing and decision-making in complex and changing environments.

[0106] If the simulation result does not meet the set requirements, step S140 is executed.

[0107] S200: If the recognition result is not an abnormal state that cannot be handled by a conventional program, automatically control the controlled object.

[0108] In addition, the method of this embodiment also monitors the timeliness of the entire exception handling process, including the problem cause analysis of the problem diagnosis model, the processing solution determination and simulation process of the control content analysis model, to ensure that alarms are issued in time when processing each link and the priority of problem handling is optimized.

[0109] The problem diagnosis model is allowed to output multiple sets of problem causes so that the control content analysis model can process these inputs in parallel. The control content analysis model will provide multiple sets of results based on multiple sets of inputs, facilitating multiple sets of tests in a simulation environment.

[0110] The aforementioned AI-based abnormal production control method acquires production-related data and performs state identification to determine whether the state is abnormal and cannot be handled by conventional procedures. The identification results include state characteristics and the corresponding problem object / event. If the abnormal state cannot be handled by conventional procedures, the relevant data is input into the problem diagnosis model for cause analysis, generating diagnostic results. These results are then passed to the control content analysis model to determine the controlled object and control action, and generate a determination result. Finally, based on the determination result, the system selects automated or manual control. This adds an automatic abnormality identification and repair mechanism to conventional automated production control, while retaining the option of manual intervention. This improves the timeliness of abnormality handling, reduces production losses caused by abnormalities, and maintains the characteristics of large-scale and efficient production. Furthermore, a simulation step is added to simulate and confirm the results of the control content analysis model and the manual assistance judgment, minimizing the possibility of incorrect answers and ensuring control accuracy.

[0111] Figure 4 FIG is a schematic block diagram of an abnormal production control device 300 based on artificial intelligence provided by an embodiment of the present invention. Figure 4As shown, corresponding to the above abnormal production control method based on artificial intelligence, the present invention also provides an abnormal production control device 300 based on artificial intelligence. The abnormal production control device 300 based on artificial intelligence includes a unit for executing the above abnormal production control method based on artificial intelligence, and the device can be configured in a server. Specifically, please refer to Figure 4 The abnormal production control device 300 based on artificial intelligence includes a data acquisition unit 301, an identification unit 302, a judgment unit 303, a cause diagnosis unit 304, a control content analysis unit 305, a simulation unit 306, a judgment unit 307, a control unit 308, a feedback unit 309 and a normal processing unit 310.

[0112] The data acquisition unit 301 is used to acquire production-related data; the identification unit 302 is used to identify the production status according to the production-related data to obtain an identification result; wherein the identification result includes any one of the production status that can be processed and the abnormal status that cannot be processed by the conventional program, and the problem object / event corresponding to the abnormal status that cannot be processed by the conventional program when the identification result is; the judgment unit 303 is used to judge whether the identification result is an abnormal status that cannot be processed by the conventional program; the cause diagnosis unit 304 is used to input the status characteristics, production-related data and problem object / event into the problem diagnosis model to diagnose the cause of the problem if the identification result is an abnormal status that cannot be processed by the conventional program to obtain a diagnosis result; the control content analysis unit 305 is used to input the diagnosis result, the status characteristics and problem object / event into the control content analysis model to determine the controlled object and the control action to obtain a determination result. result; the determination result includes the controlled object, control action and confidence level; a simulation unit 306 is used to perform simulation according to the determination result to obtain a simulation result; a judgment unit 307 is used to judge whether the simulation result meets the set requirements; if the simulation result does not meet the set requirements, the state characteristics, production-related data and problem objects / events are input into the problem diagnosis model to diagnose the cause of the problem to obtain a diagnosis result; a control unit 308 is used to select automatic control or manual control based on the built-in control strategy to control the action of the controlled object if the simulation result meets the set requirements; a feedback unit 309 is used to feed back the simulation result to the problem diagnosis model so that the problem diagnosis model can be learned and updated; a normal processing unit 310 is used to automatically control the controlled object if the recognition result is not an abnormal state that cannot be handled by the conventional program.

[0113] In one embodiment, the control unit 308 is used to manually adjust the controlled object and the corresponding control action in the determination result when the confidence level is less than a set threshold, and adopt a built-in control strategy to select automatic control or manual control according to the adjusted controlled object and the corresponding control action to perform action control of the controlled object.

[0114] It should be noted that technical personnel in the relevant field can clearly understand that the specific implementation process of the above-mentioned artificial intelligence-based abnormal production control device 300 and each unit can refer to the corresponding description in the aforementioned method embodiment. For the convenience and conciseness of the description, it will not be repeated here.

[0115] The abnormal production control device 300 based on artificial intelligence can be implemented in the form of a computer program. The computer program can be used in Figure 5 Runs on the computer device shown.

[0116] See also Figure 5 , Figure 5 1 is a schematic block diagram of a computer device provided in an embodiment of the present application. The computer device 500 may be a server, wherein the server may be an independent server or a server cluster composed of multiple servers.

[0117] See Figure 5 The computer device 500 includes a processor 502 , a memory, and a network interface 505 connected via a system bus 501 , wherein the memory may include a non-volatile storage medium 503 and an internal memory 504 .

[0118] The non-volatile storage medium 503 can store an operating system 5031 and a computer program 5032. The computer program 5032 includes program instructions, which, when executed, can enable the processor 502 to execute an abnormal production control method based on artificial intelligence.

[0119] The processor 502 is used to provide computing and control capabilities to support the operation of the entire computer device 500.

[0120] The internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute an abnormal production control method based on artificial intelligence.

[0121] The network interface 505 is used to communicate with other devices through the network. Figure 5The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device 500 to which the scheme of the present application is applied. Specifically, the computer device 500 can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0122] The processor 502 is configured to run the computer program 5032 stored in the memory to implement the following steps:

[0123] obtaining production-related data; identifying a production state according to the production-related data to obtain an identification result; wherein the identification result includes any one of a processable production state and an abnormal state that cannot be processed by a conventional program, and when the identification result is an abnormal state that cannot be processed by a conventional program, a problem object / event corresponding to the abnormal state; determining whether the identification result is an abnormal state that cannot be processed by a conventional program; if the identification result is an abnormal state that cannot be processed by a conventional program, inputting state characteristics, production-related data, and a problem object / event into a problem diagnosis model to diagnose a problem cause to obtain a diagnosis result; inputting the diagnosis result, the state characteristics, and the problem object / event into a control content analysis model to determine a controlled object and a control action to obtain a determination result; simulating according to the determination result to obtain a simulation result; determining whether the simulation result meets a set requirement; if the simulation result meets the set requirement, selecting automatic control or manual control according to the determination result using an embedded control strategy to control the action of the controlled object; feeding back the simulation result to the problem diagnosis model to enable the problem diagnosis model to learn and update; and if the simulation result does not meet the set requirement, performing the inputting of the state characteristics, the production-related data, and the problem object / event into the problem diagnosis model to diagnose the problem cause to obtain the diagnosis result.

[0124] The problem diagnosis model is obtained by establishing a correlation relationship after extracting objects and events from production factors, production processes, and product data, and setting normal state characteristics and abnormal state characteristics and corresponding problem objects / events as a sample set to train an initial model.

[0125] The initial model includes one of a rule-based model, a statistical-based model, a deep learning-based model, and a supervised learning-based model.

[0126] The control content analysis model is obtained by training a deep learning model using a plurality of problem objects / events, state characteristics, problem causes, and corresponding controlled objects and control actions as a sample set.

[0127] The deep learning model includes a deep reinforcement learning model based on value, policy, and model.

[0128] In one embodiment, when implementing the step of selecting automatic control or manual control based on the determination result using the built-in control strategy to control the motion of the controlled object, the processor 502 specifically implements the following steps:

[0129] When the confidence level is less than a set threshold, the controlled object and the corresponding control action in the determination result are manually adjusted, and the built-in control strategy is used to select automatic control or manual control based on the adjusted controlled object and the corresponding control action to perform action control on the controlled object.

[0130] In one embodiment, after implementing the step of determining whether the recognition result is an abnormal state that cannot be processed by a conventional program, the processor 502 further implements the following steps:

[0131] If the recognition result is not an abnormal state that cannot be handled by conventional procedures, the controlled object is automatically controlled.

[0132] It should be understood that in the embodiment of the present application, the processor 502 may be a central processing unit (CPU), and the processor 502 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0133] Those skilled in the art will appreciate that all or part of the steps in the method of the above-described embodiment can be implemented by instructing the relevant hardware through a computer program. The computer program includes program instructions, which can be stored in a storage medium that is computer-readable. The program instructions are executed by at least one processor in the computer system to implement the steps in the method of the above-described embodiment.

[0134] Therefore, the present invention also provides a storage medium. The storage medium may be a computer-readable storage medium. The storage medium stores a computer program, wherein when the computer program is executed by a processor, the processor performs the following steps:

[0135] Acquire production-related data; identify the production status according to the production-related data to obtain an identification result; wherein the identification result includes any one of a processable production status and an abnormal status that cannot be processed by a conventional procedure, and a problem object / event corresponding to the abnormal status that cannot be processed by a conventional procedure when the identification result is; determine whether the identification result is an abnormal status that cannot be processed by a conventional procedure; if the identification result is an abnormal status that cannot be processed by a conventional procedure, input the status characteristics, production-related data and problem object / event into a problem diagnosis model to diagnose the cause of the problem to obtain a diagnosis result; input the diagnosis result, the status characteristics and problem object / event into a control content analysis model to determine the controlled object and control action to obtain a determination result; the determination result includes a controlled object, a control action and a confidence level; a simulation is performed according to the determination result to obtain a simulation result; it is judged whether the simulation result meets the set requirements; if the simulation result meets the set requirements, automatic control or manual control is selected by using the built-in control strategy according to the determination result to control the action of the controlled object; the simulation result is fed back to the problem diagnosis model so that the problem diagnosis model is learned and updated; if the simulation result does not meet the set requirements, the state characteristics, production-related data and problem objects / events are input into the problem diagnosis model to diagnose the cause of the problem to obtain a diagnosis result.

[0136] The problem diagnosis model is obtained by extracting objects and events from production factors, production processes, and product data, establishing associations, and setting normal state features and abnormal state features as well as corresponding problem objects / events as sample sets to train the initial model.

[0137] The initial model includes one of a rule-based model, a statistics-based model, a deep learning-based model, and a supervised learning-based model.

[0138] The control content analysis model is obtained by training a deep learning model using a number of problem objects / events, state characteristics, problem causes, and corresponding controlled objects and control actions as sample sets.

[0139] The deep learning model includes a deep reinforcement learning model based on value, strategy, and model.

[0140] In one embodiment, when the processor executes the computer program to implement the step of selecting automatic control or manual control based on the determination result using the built-in control strategy to control the motion of the controlled object, the processor specifically implements the following steps:

[0141] When the confidence level is less than a set threshold, the controlled object and the corresponding control action in the determination result are manually adjusted, and the built-in control strategy is used to select automatic control or manual control based on the adjusted controlled object and the corresponding control action to perform action control on the controlled object.

[0142] In one embodiment, after executing the computer program to implement the step of determining whether the recognition result is an abnormal state that cannot be processed by a conventional program, the processor further implements the following steps:

[0143] If the recognition result is not an abnormal state that cannot be handled by conventional procedures, the controlled object is automatically controlled.

[0144] The storage medium may be any computer-readable storage medium that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a magnetic disk, or an optical disk.

[0145] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the composition and steps of each example according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0146] In the several embodiments provided herein, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the various units is merely a logical functional division, and actual implementation may employ other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be omitted or not implemented.

[0147] The steps in the methods of the embodiments of the present invention may be adjusted in order, combined, or deleted as needed. The units in the devices of the embodiments of the present invention may be combined, divided, or deleted as needed. Furthermore, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit.

[0148] If this integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product, stored in a storage medium, includes instructions for enabling a computer device (such as a personal computer, terminal, or network device) to execute all or part of the steps of the method described in various embodiments of the present invention.

[0149] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and such modifications or substitutions are intended to be within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. The abnormal production control method based on artificial intelligence is characterized by: include: Obtain production-related data; Performing production status identification based on the production-related data to obtain an identification result; wherein the identification result includes any one of a processable production status and an abnormal status that cannot be processed by a conventional procedure, and when the identification result is a problem object / event corresponding to the abnormal status that cannot be processed by a conventional procedure; Determining whether the recognition result is an abnormal state that cannot be handled by conventional procedures; If the identification result is an abnormal state that cannot be handled by conventional procedures, the state characteristics, production-related data, and problem objects / events are input into the problem diagnosis model to diagnose the cause of the problem and obtain a diagnosis result; Inputting the diagnosis result, the state characteristics, and the problem object / event into a control content analysis model to determine the controlled object and the control action, thereby obtaining a determination result; the determination result includes the controlled object, the control action, and the confidence level; Performing simulation according to the determination result to obtain a simulation result; Determining whether the simulation results meet the set requirements; If the simulation result meets the set requirements, then according to the determination result, the built-in control strategy is used to select automatic control or manual control to control the motion of the controlled object; Feeding back the simulation results to the problem diagnosis model so that the problem diagnosis model can be learned and updated; If the simulation result does not meet the set requirements, the state characteristics, production-related data and problem objects / events are input into the problem diagnosis model to diagnose the cause of the problem to obtain a diagnosis result.

2. The abnormal production control method based on artificial intelligence according to claim 1 is characterized in that: The step of selecting automatic control or manual control based on the determination result using a built-in control strategy to control the motion of the controlled object includes: When the confidence level is less than a set threshold, the controlled object and the corresponding control action in the determination result are manually adjusted, and the built-in control strategy is used to select automatic control or manual control based on the adjusted controlled object and the corresponding control action to perform action control on the controlled object.

3. The abnormal production control method based on artificial intelligence according to claim 1 is characterized in that: The problem diagnosis model is obtained by extracting objects and events from production factors, production processes, and product data, establishing associations, and setting normal state features and abnormal state features as well as corresponding problem objects / events as sample sets to train the initial model.

4. The abnormal production control method based on artificial intelligence according to claim 3 is characterized in that: The initial model includes one of a rule-based model, a statistics-based model, a deep learning-based model, and a supervised learning-based model.

5. The abnormal production control method based on artificial intelligence according to claim 1 is characterized in that: The control content analysis model is obtained by training a deep learning model using a number of problem objects / events, state characteristics, problem causes, and corresponding controlled objects and control actions as sample sets.

6. The abnormal production control method based on artificial intelligence according to claim 5 is characterized in that: The deep learning model includes a deep reinforcement learning model based on value, strategy, and model.

7. The abnormal production control method based on artificial intelligence according to claim 1 is characterized in that: After determining whether the recognition result is an abnormal state that cannot be handled by a conventional program, the method further includes: If the recognition result is not an abnormal state that cannot be handled by conventional procedures, the controlled object is automatically controlled.

8. Abnormal production control device based on artificial intelligence, characterized in that: include: A data acquisition unit, used to acquire production-related data; an identification unit, configured to identify the production status based on the production-related data to obtain an identification result; wherein the identification result includes any one of a processable production status and an abnormal status that cannot be processed by a conventional procedure, and, when the identification result is an abnormal status that cannot be processed by a conventional procedure, a problem object / event corresponding to the problem object / event; A judgment unit, configured to judge whether the recognition result is an abnormal state that cannot be handled by a conventional program; a cause diagnosis unit for inputting the state characteristics, production-related data, and problem objects / events into a problem diagnosis model to diagnose the cause of the problem and obtain a diagnosis result if the identification result indicates an abnormal state that cannot be handled by conventional procedures; a control content analysis unit, configured to input the diagnosis result, the state characteristics, and the problem object / event into a control content analysis model to determine a controlled object and a control action, thereby obtaining a determination result; the determination result includes the controlled object, the control action, and a confidence level; a simulation unit, configured to perform a simulation based on the determination result to obtain a simulation result; a judgment unit, configured to judge whether the simulation result meets the set requirements; if the simulation result does not meet the set requirements, inputting the state characteristics, production-related data, and problem objects / events into the problem diagnosis model to diagnose the cause of the problem and obtain a diagnosis result; A control unit, configured to, if the simulation result meets the set requirements, select automatic control or manual control based on the determination result using a built-in control strategy to control the motion of the controlled object; A feedback unit is used to feed back simulation results to the problem diagnosis model so that the problem diagnosis model can be learned and updated.

9. A computer device, characterized in that: The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the method according to any one of claims 1 to 7 when executing the computer program.

10. A storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images