System operation control method and device, electronic equipment and readable storage medium

Abstract

The invention discloses a system operation control method and device, electronic equipment and a readable storage medium. The method comprises the following steps: acquiring multiple pieces of real-time state data of a system; generating a plurality of first control parameters corresponding to the plurality of operation components according to the plurality of real-time state data and a pre-trained control parameter prediction model; fuzzy processing is carried out on the multiple pieces of real-time state data, and multiple second control parameters corresponding to the multiple operation components are generated; according to the multiple pieces of real-time state data and a pre-trained reinforcement learning model, multiple third control parameters corresponding to the multiple operation components are generated; and according to the multiple first control parameters, the multiple second control parameters, the multiple third control parameters and the target weight proportion, generating a control action of each operation component, and controlling each operation component to work based on the control action. Automatic control over system operation is achieved, errors caused by human factors are reduced, and the stability and safety of system operation are improved.

Description

Technical Field

[0001] The present application relates to the field of automatic control technology, and in particular to a system operation control method, device, electronic device and readable storage medium. Background Art

[0002] During the production of hydrofluoric acid, a hydrofluoric acid production tail gas treatment system is needed to treat the large amount of tail gas generated by the production of hydrofluoric acid. However, the production of hydrofluoric acid involves the synthesis of hydrogen fluoride, which usually requires high temperature and high pressure conditions, which may cause the temperature and pressure of the tail gas to fluctuate violently, affecting the stability of the system. In the related technology, relevant professionals usually record and analyze the operating status of the system regularly to determine whether the control parameters of the operating components need to be adjusted. However, manual monitoring and adjustment are usually slow and cannot respond to the rapidly changing parameters in the system in real time. Manual operation is easily affected by human factors, resulting in inaccurate adjustments, affecting the stability and safety of the system. Summary of the invention

[0003] In view of this, the present application provides a system operation control method, device, electronic device and readable storage medium, the main purpose of which is to solve the technical problems in the prior art that manual monitoring and adjustment are usually slow and cannot respond to rapidly changing parameters in the system in real time, and manual operation is easily affected by human factors, resulting in inaccurate adjustments and affecting the stability and safety of the system.

[0004] According to the first aspect of the present application, a system operation control method is provided, which is applicable to a hydrofluoric acid production tail gas treatment system, the hydrofluoric acid production tail gas treatment system includes multiple operation components, and the method includes:

[0005] Get multiple real-time status data of the system;

[0006] Generate a plurality of first control parameters corresponding to the plurality of operating components according to the plurality of real-time status data and the pre-trained control parameter prediction model;

[0007] Performing fuzzy processing on a plurality of real-time status data to generate a plurality of second control parameters corresponding to a plurality of operating components;

[0008] Generate a plurality of third control parameters corresponding to the plurality of operating components according to the plurality of real-time state data and the pre-trained reinforcement learning model;

[0009] According to a plurality of first control parameters, a plurality of second control parameters, a plurality of third control parameters and a target weight ratio, a control action for each operating component is generated, and each operating component is controlled to operate based on the control action.

[0010] Optionally, the step of generating a plurality of first control parameters corresponding to a plurality of operating components according to a plurality of real-time status data and a pre-trained control parameter prediction model specifically includes:

[0011] Based on a preset feature extraction method of a monitoring item corresponding to each real-time status data and a plurality of real-time status data, feature data corresponding to each monitoring item is generated;

[0012] A plurality of characteristic data are input into a control parameter prediction model to obtain a plurality of first control parameters corresponding to a plurality of operating components.

[0013] Optionally, the step of performing fuzzy processing on a plurality of real-time status data to generate a plurality of second control parameters corresponding to a plurality of operating components specifically includes:

[0014] Based on the fuzzy set mapping table and the multiple real-time status data, determine the fuzzy set corresponding to each monitoring item, wherein the fuzzy set table includes the corresponding relationship between the fuzzy set of each monitoring item and the preset range;

[0015] Based on a fuzzy logic rule base, determining fuzzy variables corresponding to multiple fuzzy sets of multiple monitoring items, wherein the fuzzy logic rule base includes a corresponding relationship between the fuzzy sets of the monitoring items and the fuzzy variables of the operating components;

[0016] The fuzzy variables are defuzzified to generate a plurality of second control parameters corresponding to the plurality of operating components.

[0017] Optionally, the steps of generating a control action for each operating component according to the plurality of first control parameters, the plurality of second control parameters, the plurality of third control parameters and the target weight ratio, and controlling the operation of each operating component based on the control action specifically include:

[0018] Get the real-time running time of the system and the real-time running parameters of each running component;

[0019] Determine the target weight ratio based on the real-time running time;

[0020] Based on the target weight ratio, weighted summing of the plurality of first control parameters, the plurality of second control parameters and the plurality of second control parameters is performed to generate a plurality of target control parameters;

[0021] Generate control actions corresponding to each operating component according to multiple target control parameters and real-time operating parameters;

[0022] Control the work of each running component based on the control action.

[0023] Optionally, the step of generating a control action corresponding to each operating component according to the multiple target control parameters and the real-time operating parameters specifically includes:

[0024] Compare the target control parameter and the real-time operating parameter of each operating component to determine whether the target control parameter and the real-time operating parameter are the same;

[0025] For any operating component, if the target control parameter and the real-time operating parameter are the same, the control action of the operating component is determined to remain unchanged;

[0026] If the target control parameter and the real-time operation parameter are different, a control action of the operation component is generated according to the target control parameter and the real-time operation parameter.

[0027] Optionally, the method further comprises:

[0028] Acquire multiple historical status data and multiple historical control parameters of the system;

[0029] Build a neural network model;

[0030] Based on multiple historical state data and multiple historical control parameters, the neural network model is trained, and the trained neural network model is used as a control parameter prediction model.

[0031] Optionally, the method further comprises:

[0032] Set multiple fuzzy sets for each monitoring item;

[0033] Determine a preset range corresponding to each fuzzy set according to multiple historical status data of the system;

[0034] According to the corresponding relationship between the fuzzy set and the preset range, a fuzzy set table is established;

[0035] Based on multiple historical state data and multiple historical control parameters of the system, a corresponding relationship between fuzzy sets of multiple monitoring items and fuzzy variables of multiple operating components is generated;

[0036] Based on the corresponding relationship between fuzzy sets and fuzzy variables, a fuzzy logic rule base is established.

[0037] According to the second aspect of the present application, a system operation control device is provided, which is applicable to a hydrofluoric acid production tail gas treatment system. The hydrofluoric acid production tail gas treatment system includes multiple operation components, and the device includes:

[0038] An acquisition module is used to acquire multiple real-time status data of the system;

[0039] A first generating module, used to generate a plurality of first control parameters corresponding to a plurality of operating components according to a plurality of real-time status data and a pre-trained control parameter prediction model;

[0040] A second generating module is used to perform fuzzy processing on the multiple real-time status data to generate multiple second control parameters corresponding to the multiple operating components;

[0041] A third generating module, used to generate a plurality of third control parameters corresponding to the plurality of operating components according to the plurality of real-time state data and the pre-trained reinforcement learning model;

[0042] The fourth generation module is used to generate a control action for each operating component according to multiple first control parameters, multiple second control parameters, multiple third control parameters and a target weight ratio; the control module is used to control the operation of each operating component based on the control action.

[0043] According to a third aspect of the present application, an electronic device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the methods described in the first aspect when executing the computer program.

[0044] According to a fourth aspect of the present application, a readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the methods in the first aspect are implemented.

[0045] By means of the above technical scheme, the present application provides a system operation control method, device, electronic device and readable storage medium. Specifically, during the operation of the hydrofluoric acid production tail gas treatment system, real-time status data in the system is collected, and the control parameters that meet the actual working conditions of the system are calculated by using the neural network method, fuzzy logic control method and reinforcement learning strategy respectively, and then the control parameters output by fuzzy logic control, neural network and reinforcement learning are comprehensively processed to obtain the optimal control action to control the system operation. Compared with the control method of manual monitoring and regulation in the relevant technology, the present application can dynamically and continuously monitor and optimize according to real-time data combined with environmental conditions, maintain the long-term stability of the system, reduce errors caused by human factors, improve the stability and safety of the system operation, and reduce dependence on manual operation, saving human resources.

[0046] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present application. Also, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0048] Figure 1 A schematic diagram of a system operation control method provided in an embodiment of the present application is shown;

[0049] Figure 2 A schematic diagram of another system operation control method provided in an embodiment of the present application is shown;

[0050] Figure 3 A schematic diagram of the structure of a system operation control device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0051] The exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art.

[0052] The present application embodiment provides a system operation control method, such as Figure 1 As shown, the method includes:

[0053] S101. Acquire multiple real-time status data of the system.

[0054] The system operation control method provided in the embodiment of the present application is mainly used in the control scenario of multiple operating components in the hydrofluoric acid production exhaust gas treatment system. The execution subject of the embodiment of the present application is a device or equipment that can automatically control the system operation, which can be specifically set on the server side. The server judges the real-time load situation based on the real-time status data of the hydrofluoric acid production exhaust gas treatment system during operation, and calculates the adjustment amount of multiple operating components to ensure the stable operation of the system.

[0055] The system operation control method provided in the present application is applicable to a hydrofluoric acid production tail gas treatment system, which is used to recycle and treat tail gas generated during the production of hydrofluoric acid. Specifically, during the operation of the system, real-time status data fed back by multiple sensors (such as pressure sensors, temperature sensors, and flow sensors) arranged at key positions in the tail gas pipeline in the receiving system are used to monitor changes in the system operation status and understand the system load situation.

[0056] Optionally, the real-time status data includes at least one of the following: a real-time pressure value, a real-time temperature value, and a real-time flow value.

[0057] In actual application scenarios, before the system is started, the initial control parameters of each operating component are set according to the design parameters and predetermined safety standards. When the hydrofluoric acid production tail gas treatment system is started, each operating component is controlled according to the initial control parameters. During the operation of the system, the status data fed back by each sensor is collected in real time to perform trend analysis and judge the system performance.

[0058] S102: Generate a plurality of first control parameters corresponding to a plurality of operating components according to a plurality of real-time status data and a pre-trained control parameter prediction model.

[0059] In this step, the production of hydrofluoric acid involves the synthesis of hydrogen fluoride, which usually requires high temperature and high pressure conditions, which will cause the temperature and pressure of the tail gas to fluctuate violently. In addition, the quality of the raw materials, the fluctuation of the flow rate, and the changes in operating conditions (such as feed rate and reaction temperature) will affect the temperature and pressure of the tail gas during the production process. In order to ensure the stable operation of the system, the control parameters of the operating components (such as fans, valves, etc.) need to be accurately adjusted to cope with the changes in the state of temperature, pressure and flow.

[0060] In order to achieve precise control of the system, this application proposes a pre-trained control parameter prediction model, which is used to predict the optimal control parameters of each operating component in the real-time operating state based on the system operating state. During the operation of the system, multiple real-time state data monitored are input into the pre-trained control parameter prediction model, and the model uses these input data for calculation to give an optimal first control parameter to meet the stable operation requirements of the system.

[0061] Optionally, the hydrofluoric acid production tail gas treatment system is provided with multiple operating components, including fans, valves, etc.

[0062] S103: Perform fuzzy processing on a plurality of real-time status data to generate a plurality of second control parameters corresponding to a plurality of operating components.

[0063] In this step, during the operation of the hydrofluoric acid production tail gas treatment system, due to real-time environmental changes and some uncertainties in the system, the control parameters output by a single neural network may not be fully applicable. In order to ensure the reliability of the control effect, this application proposes to use a fuzzy logic control method based on a neural network model to fuzzify the real-time state data to convert it into fuzzy variables corresponding to multiple operating components, and finally defuzzify it to obtain the actual control parameters (i.e., the second control parameters), so that the final control action can better cope with complex and dynamic operating environments.

[0064] S104: Generate multiple third control parameters corresponding to multiple operating components according to multiple real-time operating data and a pre-trained reinforcement learning model.

[0065] In this step, the neural network provides a prediction of the control parameters based on the real-time state data, while the fuzzy logic control adjusts the control parameters according to empirical rules. However, these adjustments may not guarantee long-term optimization effects. Therefore, the present application also proposes to continuously adjust the control parameters through a reinforcement learning strategy, and then continuously learn the most effective control strategy over a period of time, and then judge the real-time optimal control parameters (i.e., the third control parameters) of each operating component based on the real-time operating state, so as to maximize the processing effect and reduce energy consumption in the long term.

[0066] S105. Generate a control action for each operating component according to a plurality of first control parameters, a plurality of second control parameters, a plurality of third control parameters and a preset weight ratio, and control each operating component to operate based on the control action.

[0067] In this step, in order to ensure that the adjustment amount of each operating component of the system is more in line with the current operating conditions of the system, the weight ratio of the control parameters calculated by each method is pre-set, and then after each control strategy is used to calculate the optimal control parameters in turn, multiple control parameters are comprehensively calculated based on the preset weight ratio, combined with the real-time operating parameters of each operating component, and the final control action is generated and applied to each operating component. While giving full play to the respective advantages of these methods, while making up for the shortcomings of a single method, it can dynamically optimize system performance in the long term, adapt to complex and changing environments, optimize system control performance, and greatly improve the stability of the operation of the hydrofluoric acid production tail gas treatment system.

[0068] The system operation control method provided in the embodiment of the present application, during the operation of the hydrofluoric acid production tail gas treatment system, collects real-time status data in the system, and respectively uses a neural network method, a fuzzy logic control method, and a reinforcement learning strategy to calculate control parameters that meet the actual working conditions of the system, and then comprehensively processes the control parameters output by the fuzzy logic control, neural network, and reinforcement learning to obtain the optimal control action to control the system operation. Compared with the control method of manual monitoring and regulation in the relevant technology, the present application can dynamically and continuously monitor and optimize according to real-time data combined with environmental conditions, maintain the long-term stability of the system, reduce errors caused by human factors, improve the stability and safety of the system operation, and reduce dependence on manual operation, saving human resources.

[0069] Further, as a refinement and expansion of the specific implementation of the above embodiment, in order to fully illustrate the specific implementation process of this embodiment, the embodiment of the present application provides another system operation control method, such as Figure 2 As shown, the method includes:

[0070] S201. Acquire multiple real-time status data of the system.

[0071] This step and Figure 1 The method of step S101 shown is the same and will not be repeated here.

[0072] In actual application scenarios, high-precision pressure sensors, flow sensors, and temperature sensors are arranged at key locations of the tail gas pipeline of the hydrofluoric acid production tail gas treatment system. The pressure sensor monitors the pressure value in the pipeline in real time; the flow sensor monitors the tail gas flow value in real time; and the temperature sensor monitors the tail gas temperature value in real time. During the operation of the system, various sensors are used to collect real-time status data. The collected real-time data is pre-processed (such as removing noise, outliers, etc.) using filtering algorithms, data smoothing, and other methods.

[0073] S202: Generate feature data corresponding to each monitoring item based on a preset feature extraction method of the monitoring item corresponding to each real-time status data and a plurality of real-time status data.

[0074] In this step, each state data corresponds to a monitoring item, and a preset feature extraction method corresponding to each monitoring item is obtained. According to each prediction feature extraction method, key features are extracted from the real-time state data of each monitoring item, and the feature data corresponding to each monitoring item is calculated.

[0075] In actual application scenarios, the key feature of the pressure monitoring item is the pressure change trend. It is necessary to calculate the pressure change rate within a period of time based on the collected pressure value. Specifically, a time window (such as 10 minutes) is set, and the change rate of the pressure value in the window is calculated based on the real-time pressure value. The key feature of the temperature monitoring item is the temperature change trend. It is necessary to calculate the temperature change rate within a period of time based on the collected temperature value. Specifically, a time window is set, and the change rate of the temperature value in the window is calculated based on the real-time temperature value. The key feature of the flow monitoring item is the flow fluctuation amplitude. It is necessary to calculate the flow standard deviation within a period of time based on the real-time flow value. Specifically, a time window is set, and the standard deviation of the flow value in the window is calculated based on the real-time flow value.

[0076] Through the above methods, key features are extracted from real-time status data, unnecessary information is removed, and the data structure is simplified, so as to better reflect the dynamic behavior and trend of the system.

[0077] S203: Input a plurality of characteristic data into a control parameter prediction model to obtain a plurality of first control parameters corresponding to a plurality of operating components.

[0078] In this step, the characteristic data of each monitoring item is input into the trained control parameter prediction model. The model predicts and outputs the optimal control parameters of each operating component based on the real-time operating status.

[0079] In an embodiment of the present application, optionally, in order to accurately generate the first control parameter of each control item, it is necessary to pre-train a control parameter prediction model, based on a preset feature extraction method and multiple real-time status data corresponding to the monitoring item of each real-time status data, before generating the feature data corresponding to each monitoring item, it also includes: obtaining multiple historical status data and multiple historical control parameters of the system; constructing a neural network model; based on multiple historical status data and multiple historical control parameters, training the neural network model, and using the trained neural network model as a control parameter prediction model.

[0080] In this embodiment, multiple historical state data and corresponding historical control parameters of the hydrofluoric acid production tail gas treatment system under different production conditions are collected to ensure the diversity and representativeness of the data. Thereafter, a large amount of historical state data and historical control parameters collected are pre-sorted by using methods such as mean replacement method and linear interpolation method to remove obvious outliers. For example, if the difference between a data point and the adjacent data points before and after exceeds a certain range, it is marked as an outlier and processed. The processed historical state data is used as input data, and the historical control parameters are used as output data, and are divided into training set and test set according to a certain ratio.

[0081] Further, select a suitable neural network type, such as a multilayer perceptron (MLP) or a recurrent neural network (RNN), and determine the parameters such as the number of layers and the number of neurons of the network. For example, an MLP with a three-layer structure can be selected, including an input layer, a hidden layer, and an output layer, and the number of neurons in each layer can be determined according to the complexity of the system. Use the training set to train the neural network, and adjust the network weights and biases so that the network output is as close to the actual control target as possible. Specifically, a large amount of characteristic data of historical temperature, historical flow, and historical pressure is used as input data, and a large amount of control parameters of historical valve opening and historical fan speed are used as output data to train the neural network. Use the test set to verify the trained neural network and evaluate its performance. If the performance does not meet the requirements, adjust the network structure and parameters and retrain; if the performance meets the requirements. Specifically, calculate the prediction error of the neural network on the test set, such as the mean square error, etc. If the error is large, adjust the number of layers, the number of neurons, the learning rate and other parameters of the network, retrain, and finally output the trained neural network model as the control parameter prediction model.

[0082] S204: Determine the fuzzy set corresponding to each monitoring item based on the fuzzy set mapping table and multiple real-time status data.

[0083] The fuzzy set table includes the corresponding relationship between the fuzzy set of each monitoring item and the preset range.

[0084] S205. Determine fuzzy variables corresponding to multiple fuzzy sets of multiple monitoring items based on a fuzzy logic rule base.

[0085] The fuzzy logic rule base includes the corresponding relationship between the fuzzy sets of monitoring items and the fuzzy variables of the operating components.

[0086] S206: Defuzzify the fuzzy variables to generate multiple second control parameters corresponding to the multiple operating components.

[0087] In steps S204 to S206, fuzzy sets are used to describe parameters that are fuzzy or uncertain in actual situations. Taking temperature as an example, fuzzy sets can be used to represent "slightly higher" or "very low" temperatures rather than an exact temperature value. Based on the preset range in the fuzzy set table, the real-time status data is mapped to the fuzzy set. For example, the real-time temperature of the system is 35 degrees Celsius. According to the boundary value (i.e., the preset range) of each fuzzy set, it is determined that the real-time temperature can be mapped to the "high temperature" set.

[0088] Furthermore, the fuzzy logic rule base is used to describe the rules between different fuzzy sets and fuzzy variables, for example, if "the temperature is high and the pressure is moderate", then "increase the cooling device power"; if "the flow rate is low and the temperature is low", then "reduce the fan speed".

[0089] Finally, the fuzzy variables of the operating components are defuzzified, and the fuzzy decisions obtained by fuzzy reasoning are converted into specific and clear control parameters.

[0090] In actual application scenarios, in order to obtain clear control parameters, after the fuzzy decision is obtained by fuzzy reasoning, the fuzzy decision can be sent to the terminal of the relevant professional, and the second control parameter corresponding to the fuzzy variable based on the real-time running time feedback from the relevant professional can be accepted.

[0091] In an embodiment of the present application, optionally, in order to accurately generate the second control parameter of each control item, before determining the fuzzy set corresponding to each monitoring item based on the fuzzy set table, the fuzzy set mapping table and the real-time status data, it also includes: setting multiple fuzzy sets for each monitoring item; determining the preset range corresponding to each fuzzy set according to multiple historical status data of the system; establishing a fuzzy set table according to the correspondence between the fuzzy set and the preset range; generating the correspondence between the fuzzy sets of multiple monitoring items and the fuzzy variables of multiple operating components based on the multiple historical status data and multiple historical control parameters of the system; and establishing a fuzzy logic rule base based on the correspondence between the fuzzy sets and the fuzzy variables.

[0092] In this embodiment, first, according to the actual range of each monitoring item, a fuzzy set corresponding to each monitoring item is defined, such as "low temperature", "moderate temperature", and "high temperature". Secondly, relevant experts set the preset range corresponding to each fuzzy set of each monitoring item based on their experience, professional knowledge and the historical operation of the system. Taking pressure as an example, the fuzzy set corresponding to pressure in the range of [0, P1] is "low pressure", the fuzzy set corresponding to pressure in the range of [P1, P2] is "moderate pressure", and the fuzzy set corresponding to pressure in the range of [P2, ∞] is "high pressure". Finally, the fuzzy sets of all monitoring items in the system and their corresponding preset ranges are summarized to form a fuzzy set table.

[0093] Furthermore, the relevant experts, based on the correspondence between the historical status data of the system's monitoring items and the historical control parameters of the control items, combined with their experience and expertise, formulate the correspondence between the fuzzy sets of the monitoring items and the fuzzy variables of the system's operating components, such as "if the pressure is high and the flow is large, reduce the valve opening", "if the temperature is high and the pressure is medium, increase the cooling device power", "if the flow is low and the temperature is low, reduce the fan speed", etc. Finally, the fuzzy logic rule base is summarized and generated.

[0094] S207: Generate multiple third control parameters corresponding to multiple operating components according to multiple real-time status data and a pre-trained reinforcement learning model.

[0095] In this step, the collected multiple real-time status data are used as input of the reinforcement learning model, and the model predicts and outputs the optimal action (ie, the third control parameter).

[0096] In actual application scenarios, the reinforcement learning model is pre-trained. First, the state space is defined. Specifically, multiple real-time state data of the system and the operating data of multiple operating components are represented as a vector, including real-time temperature value, real-time flow value, real-time pressure value, real-time valve opening, and real-time fan speed. Secondly, the action space is defined, and the control actions that can be taken are represented as the changes in parameters such as fan speed and valve opening. For example, the control action can be represented as a vector, including the increase in fan speed, the adjustment of valve opening, etc. Thirdly, the reward function is defined. Specifically, according to the control target of the system, the reward function is defined to evaluate the effect of a certain control action taken by the system in a certain state. For example, if the control target of the system is to keep the negative pressure of the system stable, the reward function can be defined as the degree of proximity between the negative pressure and the set value, as well as the stability of parameters such as temperature, flow, and pressure. Further, reinforcement learning training, specifically, select a suitable reinforcement learning algorithm (such as Q-learning, deep Q network (DQN), etc.) for training, and define how the state is transferred to the next state according to the real-time state and the selected action, and the reward calculated according to the results of state transfer and action execution. Afterwards, the strategy of the reinforcement learning model is randomly initialized, and the strategy is updated according to the reward and state transfer information obtained from each interaction. The updated rules defined in the algorithm are used to obtain the trained reinforcement learning model. The changes in the system state after the action is implemented are collected, and these data will serve as the basis for the next prediction of the model. Continue to collect data and implement actions, and fine-tune parameters according to actual results, so as to continuously improve the performance of the model.

[0097] S208: Acquire the real-time running time of the system and the real-time running parameters of each running component.

[0098] S209: Determine the target weight ratio according to the real-time running time.

[0099] In steps S208 and S209, after using the neural network, fuzzy logic control method and reinforcement learning strategy to output their respective optimal control parameters in turn, the three control parameters need to be comprehensively sorted out. Paying attention to the fuzzy logic control method in the early stage can help the system run quickly and stably, and deal with the uncertainty and complexity in the early stage. As the system runs and data accumulates, the weights are gradually transferred to the neural network and reinforcement learning algorithms, and the advantages of these methods can be used to further improve the system performance and control accuracy. Therefore, the method of determining the weight ratio in stages can effectively utilize the advantages of different control methods and optimize the overall performance of the system. Specifically, the real-time running time of the system is obtained, and the weight ratio of the three algorithms is determined based on the real-time running time and the preset weight ratio mapping table. The preset weight ratio mapping table contains the corresponding relationship between different running time periods and the weight ratios of the three algorithms. For example, the system running time is 15 days, and in the preset weight ratio mapping table, the target weight ratio of the fuzzy logic control method, neural network and reinforcement learning strategy is called as 4:3:3.

[0100] S210. Based on the target weight ratio, perform weighted summation on the plurality of first control parameters, the plurality of second control parameters, and the plurality of second control parameters to generate a plurality of target control parameters.

[0101] S211. Generate a control action corresponding to each operating component according to multiple target control parameters and real-time operating parameters.

[0102] S212: Control each running component to work based on the control action.

[0103] In steps S210 to S212, the first control parameter, the second control parameter and the third control parameter of each operating component are obtained, and the first control parameter, the second control parameter and the third control parameter are weighted and summed according to the target weight ratio to obtain the target control parameter of the operating component. Thereafter, for any operating component, the target control parameter is compared with the real-time operating parameter of the operating component to determine whether it needs to be adjusted and the adjustment amount of the operating component (such as reducing the speed of the fan 30). Finally, the control action of the operating component is generated according to the adjustment parameter, and the operation of the operating component is controlled according to the control action.

[0104] In an embodiment of the present application, optionally, in order to accurately generate control actions for each operating component, step S211 generates a control action corresponding to each operating component based on multiple target control parameters and real-time operating parameters, specifically including: comparing the target control parameters and the real-time operating parameters of each operating component to determine whether the target control parameters and the real-time operating parameters are the same; for any operating component, if the target control parameters and the real-time operating parameters are the same, determining that the control action of the operating component remains unchanged; if the target control parameters and the real-time operating parameters are different, generating a control action for the operating component based on the target control parameters and the real-time operating parameters.

[0105] In this embodiment, for any operating component, its target control parameter is compared with its real-time operating parameter to determine whether the target control parameter and the real-time operating parameter are the same. If they are the same, it means that the current load condition of the system is stable, and the current operating parameter of the operating component has met the optimal control conditions. There is no need to adjust the operating component, so that it can continue to operate according to the current operating parameters. Therefore, the control action of the operating component is determined to remain unchanged; otherwise, it means that the current operating condition of the system is unstable and the operating component needs to be adjusted. At this time, according to the target control parameter and the real-time operating parameter, the adjustment action and adjustment amount of the operating component are determined, and then the control action of the operating component is determined to adjust the operating component.

[0106] Further, as Figure 1 In the specific implementation of the method, the embodiment of the present application provides a system operation control device 300, which is applicable to a hydrofluoric acid production tail gas treatment system. The hydrofluoric acid production tail gas treatment system includes multiple operation components, such as Figure 3 As shown, the device includes: an acquisition module 301 , a first generation module 302 , a second generation module 303 , a third generation module 304 , a fourth generation module 305 and a control module 306 .

[0107] The acquisition module 301 is used to acquire multiple real-time status data of the system;

[0108] A first generating module 302, configured to generate a plurality of first control parameters corresponding to a plurality of operating components according to a plurality of real-time status data and a pre-trained control parameter prediction model;

[0109] The second generating module 303 is used to perform fuzzy processing on the multiple real-time status data to generate multiple second control parameters corresponding to the multiple operating components;

[0110] A third generating module 304, configured to generate a plurality of third control parameters corresponding to a plurality of operating components according to a plurality of real-time state data and a pre-trained reinforcement learning model;

[0111] The fourth generation module 305 is used to generate the control action of each operating component according to multiple first control parameters, multiple second control parameters, multiple third control parameters and the target weight ratio; the control module 306 is used to control the operation of each operating component based on the control action.

[0112] In a specific application scenario, the first generating module 302 is specifically used to:

[0113] Based on a preset feature extraction method of a monitoring item corresponding to each real-time status data and a plurality of real-time status data, feature data corresponding to each monitoring item is generated;

[0114] A plurality of characteristic data are input into a control parameter prediction model to obtain a plurality of first control parameters corresponding to a plurality of operating components.

[0115] In a specific application scenario, the second generation module 303 is specifically used to:

[0116] Based on the fuzzy set mapping table and the multiple real-time status data, determine the fuzzy set corresponding to each monitoring item, wherein the fuzzy set table includes the corresponding relationship between the fuzzy set of each monitoring item and the preset range;

[0117] Based on a fuzzy logic rule base, determining fuzzy variables corresponding to multiple fuzzy sets of multiple monitoring items, wherein the fuzzy logic rule base includes a corresponding relationship between the fuzzy sets of the monitoring items and the fuzzy variables of the operating components;

[0118] The fuzzy variables are defuzzified to generate a plurality of second control parameters corresponding to the plurality of operating components.

[0119] In a specific application scenario, the fourth generating module 305 is specifically used to:

[0120] Get the real-time running time of the system and the real-time running parameters of each running component;

[0121] Determine the target weight ratio based on the real-time running time;

[0122] Based on the target weight ratio, weighted summing of the plurality of first control parameters, the plurality of second control parameters and the plurality of second control parameters is performed to generate a plurality of target control parameters;

[0123] According to multiple target control parameters and real-time operation parameters, a control action corresponding to each operation component is generated.

[0124] In a specific application scenario, the control module 306 is specifically used to:

[0125] Control the work of each running component based on the control action.

[0126] In a specific application scenario, the fourth generating module 305 is further used to:

[0127] Compare the target control parameter and the real-time operating parameter of each operating component to determine whether the target control parameter and the real-time operating parameter are the same;

[0128] For any operating component, if the target control parameter and the real-time operating parameter are the same, the control action of the operating component is determined to remain unchanged;

[0129] If the target control parameter and the real-time operation parameter are different, a control action of the operation component is generated according to the target control parameter and the real-time operation parameter.

[0130] In a specific application scenario, the acquisition module 301 is also used to acquire multiple historical status data and multiple historical control parameters of the system.

[0131] In specific application scenarios, such as Figure 3 As shown, the device also includes:

[0132] Establishing module 307 for constructing a neural network model;

[0133] The training module 308 is used to train the neural network model based on multiple historical state data and multiple historical control parameters, and use the trained neural network model as a control parameter prediction model.

[0134] In specific application scenarios, such as Figure 3 As shown, the device also includes:

[0135] A setting module 309, used to set multiple fuzzy sets for each monitoring item;

[0136] A determination module 310, for determining a preset range corresponding to each fuzzy set according to a plurality of historical state data of the system;

[0137] The establishing module 307 is also used to establish a fuzzy set table according to the corresponding relationship between the fuzzy set and the preset range;

[0138] A fifth generating module 311, for generating a correspondence between a plurality of fuzzy sets of monitoring items and a plurality of fuzzy variables of operating components based on a plurality of historical state data and a plurality of historical control parameters of the system;

[0139] The establishing module 307 is further used to establish a fuzzy logic rule base based on the corresponding relationship between the fuzzy sets and the fuzzy variables.

[0140] The system operation control device 300 provided in the embodiment of the present application, during the operation of the hydrofluoric acid production tail gas treatment system, collects real-time status data in the system, and respectively uses a neural network method, a fuzzy logic control method, and a reinforcement learning strategy to calculate control parameters that meet the actual working conditions of the system, and then comprehensively processes the control parameters output by the fuzzy logic control, the neural network, and the reinforcement learning to obtain the optimal control action to control the system operation. Compared with the control method of manual monitoring and regulation in the relevant technology, the present application can dynamically and continuously monitor and optimize according to real-time data combined with environmental conditions, maintain the long-term stability of the system, reduce errors caused by human factors, improve the stability and safety of the system operation, and reduce dependence on manual operation, saving human resources.

[0141] In an exemplary embodiment, the present application further provides an electronic device, including a memory and a processor. The memory stores a computer program, and the processor is used to execute the program stored in the memory and execute the system operation control method in the above embodiment.

[0142] In an exemplary embodiment, the present application further provides a readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the system operation control method are implemented.

[0143] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented by hardware, or by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile readable storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including a number of instructions for enabling an electronic device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each implementation scenario of the present application.

[0144] Those skilled in the art will appreciate that the accompanying drawings are merely schematic diagrams of a preferred implementation scenario, and the modules or processes in the accompanying drawings are not necessarily required for implementing the present application.

[0145] Those skilled in the art will appreciate that the modules in the devices in the implementation scenario can be distributed in the devices in the implementation scenario according to the implementation scenario description, or can be changed accordingly and located in one or more devices different from the implementation scenario. The modules in the above implementation scenario can be combined into one module, or can be further split into multiple units.

[0146] The above application serial numbers are for description only and do not represent the advantages or disadvantages of the implementation scenarios.

[0147] The above disclosure only discloses several specific implementation scenarios of the present application. However, the present application is not limited thereto, and any changes that can be conceived by technicians in this field should fall within the scope of protection of the present application.

Claims

1. A system operation control method, characterized in that: Applicable to a hydrofluoric acid production tail gas treatment system, the hydrofluoric acid production tail gas treatment system includes multiple operating components, and the method includes: Get multiple real-time status data of the system; Generate a plurality of first control parameters corresponding to a plurality of operating components according to the plurality of real-time status data and a pre-trained control parameter prediction model; Performing fuzzy processing on the multiple real-time status data to generate multiple second control parameters corresponding to the multiple operating components; Generate a plurality of third control parameters corresponding to the plurality of operating components according to the plurality of real-time status data and the pre-trained reinforcement learning model; A control action for each operating component is generated according to the plurality of first control parameters, the plurality of second control parameters, the plurality of third control parameters and the target weight ratio, and each operating component is controlled to operate based on the control action.

2. The method according to claim 1, characterized in that The step of generating a plurality of first control parameters corresponding to a plurality of operating components according to the plurality of real-time status data and the pre-trained control parameter prediction model specifically includes: Based on a preset feature extraction method of the monitoring item corresponding to each real-time status data and the plurality of real-time status data, generating feature data corresponding to each monitoring item; A plurality of characteristic data are input into the control parameter prediction model to obtain a plurality of first control parameters corresponding to a plurality of operating components.

3. The method according to claim 1, characterized in that The step of performing fuzzy processing on the multiple real-time status data to generate multiple second control parameters corresponding to the multiple operating components specifically includes: Determine the fuzzy set corresponding to each monitoring item based on the fuzzy set mapping table and the multiple real-time status data, wherein the fuzzy set table includes the corresponding relationship between the fuzzy set of each monitoring item and the preset range; Determine fuzzy variables corresponding to multiple fuzzy sets of multiple monitoring items based on a fuzzy logic rule base, wherein the fuzzy logic rule base includes a correspondence between the fuzzy sets of the monitoring items and the fuzzy variables of the operating components; The fuzzy variables are defuzzified to generate a plurality of second control parameters corresponding to the plurality of operating components.

4. The method according to claim 1, characterized in that The step of generating a control action for each operating component according to the plurality of first control parameters, the plurality of second control parameters, the plurality of third control parameters and the target weight ratio, and controlling the operation of each operating component based on the control action specifically includes: Get the real-time running time of the system and the real-time running parameters of each running component; Determining a target weight ratio according to the real-time running time; Based on the target weight ratio, weighted summing the plurality of first control parameters, the plurality of second control parameters and the plurality of second control parameters is performed to generate a plurality of target control parameters; Generating the control action corresponding to each operating component according to the multiple target control parameters and the real-time operating parameters; Each operating component is controlled to operate based on the control action.

5. The method according to claim 4, characterized in that The step of generating the control action corresponding to each operating component according to the multiple target control parameters and the real-time operating parameters specifically includes: Comparing the target control parameter of each operating component with the real-time operating parameter to determine whether the target control parameter is the same as the real-time operating parameter; For any operating component, if the target control parameter is the same as the real-time operating parameter, determining that the control action of the operating component remains unchanged; If the target control parameter and the real-time operation parameter are different, a control action of the operation component is generated according to the target control parameter and the real-time operation parameter.

6. The method according to any one of claims 1 to 5, characterized in that Also includes: Acquire multiple historical status data and multiple historical control parameters of the system; Build a neural network model; Based on the multiple historical state data and the multiple historical control parameters, the neural network model is trained, and the trained neural network model is used as a control parameter prediction model.

7. The method according to any one of claims 1 to 5, characterized in that Also includes: Set multiple fuzzy sets for each monitoring item; Determine a preset range corresponding to each fuzzy set according to multiple historical status data of the system; Establishing a fuzzy set table according to the corresponding relationship between the fuzzy set and the preset range; Based on the plurality of historical state data and the plurality of historical control parameters of the system, generating a correspondence between a plurality of fuzzy sets of monitoring items and a plurality of fuzzy variables of operating components; Based on the corresponding relationship between fuzzy sets and fuzzy variables, a fuzzy logic rule base is established.

8. A system operation control device, characterized in that: Applicable to a hydrofluoric acid production tail gas treatment system, the hydrofluoric acid production tail gas treatment system comprises a plurality of operating components, the device comprises: An acquisition module is used to acquire multiple real-time status data of the system; A first generating module, configured to generate a plurality of first control parameters corresponding to a plurality of operating components according to the plurality of real-time status data and a pre-trained control parameter prediction model; A second generating module, configured to perform fuzzy processing on the plurality of real-time status data to generate a plurality of second control parameters corresponding to the plurality of operating components; A third generating module, configured to generate a plurality of third control parameters corresponding to the plurality of operating components according to the plurality of real-time status data and the pre-trained reinforcement learning model; The fourth generation module is used to generate a control action for each operating component according to the multiple first control parameters, the multiple second control parameters, the multiple third control parameters and the target weight ratio; the control module is used to control the operation of each operating component based on the control action.

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

10. A 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