Intelligent signal I / O control system and method for industrial automation

Abstract

The invention relates to the technical field of signal control, and discloses an intelligent signal I / O control system and method for industrial automation. The method comprises the following steps: carrying out acquisition and feature analysis on various signals in the industrial automation system to obtain a signal feature model; designing a two-stage I / O control scheme to obtain a first control strategy; constructing a state-trend matrix, and introducing an evaluation function to quantify a control effect to perform dynamic iteration updating to obtain a target control parameter; performing PLC autonomous fault diagnosis to obtain a fault diagnosis result; performing network cooperative control analysis on the plurality of PLCs based on a fault diagnosis result to obtain a hierarchical central coordination mechanism; and establishing a multi-dimensional performance index system of a hierarchical central coordination mechanism, and performing evaluation and adaptive optimization on the first control strategy to obtain a second control strategy, thereby realizing intelligent perception, accurate control and fault diagnosis of a complex industrial environment, thereby improving the performance and stability of the whole automation system.

Description

Technical Field

[0001] The present application relates to the field of signal control technology, and in particular to an intelligent signal I / O control system and method thereof for industrial automation. Background Art

[0002] The complexity and intelligence of industrial automation systems are constantly increasing. Traditional signal I / O control methods often fail to meet the requirements of efficient, accurate, and reliable control in complex and changing industrial environments. Industrial automation systems involve a large number of sensors, actuators, and controllers. These devices generate and process a wide variety of signals and a huge amount of data, which poses a huge challenge to signal acquisition, processing, and control.

[0003] At the same time, various interference factors, equipment failures and system anomalies in the industrial production process also increase the difficulty of control. Traditional control methods usually adopt fixed control strategies, lack the ability to adapt to dynamic changes in the system and predict future states, and are difficult to cope with complex and changing working conditions. In addition, collaborative control and fault diagnosis between multiple programmable logic controllers are also important challenges facing industrial automation systems. Summary of the invention

[0004] The present application provides an intelligent signal I / O control system and method for industrial automation, which are used to achieve intelligent perception, precise control and fault diagnosis of complex industrial environments, thereby improving the performance and stability of the entire automation system.

[0005] In a first aspect, the present application provides an intelligent signal I / O control method for industrial automation, the intelligent signal I / O control method for industrial automation comprising:

[0006] Collect and analyze the characteristics of various signals in industrial automation systems to obtain signal characteristic models;

[0007] Designing a two-level I / O control scheme for the signal characteristic model to obtain a first control strategy including a predictive control layer and a real-time response layer;

[0008] Constructing a state-trend matrix and introducing an evaluation function to quantify the control effect, dynamically iterating and updating the first control strategy to obtain target control parameters;

[0009] Perform PLC autonomous fault diagnosis on the various signals according to the target control parameters to obtain fault diagnosis results;

[0010] Based on the fault diagnosis results, a network collaborative control analysis is performed on multiple PLCs to obtain a hierarchical central coordination mechanism;

[0011] A multi-dimensional performance indicator system for the hierarchical central coordination mechanism is established, and the first control strategy is evaluated and adaptively optimized to obtain a second control strategy.

[0012] In a second aspect, the present application provides an intelligent signal I / O control system for industrial automation, the intelligent signal I / O control system for industrial automation comprising:

[0013] The acquisition module is used to collect and analyze the characteristics of various signals in the industrial automation system to obtain the signal characteristic model;

[0014] A design module, used for designing a two-level I / O control scheme for the signal characteristic model to obtain a first control strategy including a predictive control layer and a real-time response layer;

[0015] An updating module, used for constructing a state-trend matrix and introducing an evaluation function to quantify the control effect, dynamically iterating and updating the first control strategy to obtain target control parameters;

[0016] A diagnosis module, used for performing PLC autonomous fault diagnosis on the various signals according to the target control parameters to obtain a fault diagnosis result;

[0017] An analysis module, used for performing network collaborative control analysis on multiple PLCs based on the fault diagnosis results to obtain a hierarchical central coordination mechanism;

[0018] The optimization module is used to establish a multi-dimensional performance indicator system for the hierarchical central coordination mechanism, and to evaluate and adaptively optimize the first control strategy to obtain a second control strategy.

[0019] A third aspect of the present application provides an intelligent signal I / O control device for industrial automation, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the intelligent signal I / O control device for industrial automation executes the above-mentioned intelligent signal I / O control method for industrial automation.

[0020] A fourth aspect of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned intelligent signal I / O control method for industrial automation.

[0021] In the technical solution provided by the present application, a signal feature model including signal type, change characteristics and interference factors is established by conducting a comprehensive feature analysis of various types of signals in the industrial automation system. A two-level control structure of a predictive control layer and a real-time response layer is adopted, which can not only predict the trend of signal changes, but also respond quickly to emergencies, thereby improving the forward-looking and real-time nature of the control. The introduction of the state-trend matrix and the evaluation function realizes the dynamic iterative update of the control strategy, so that the control parameters can be continuously optimized according to the system state and the control objectives. The PLC autonomous fault diagnosis method based on the particle swarm optimization algorithm improves the accuracy and efficiency of fault detection. Through the hierarchical central coordination mechanism, efficient collaborative control between multiple PLCs is achieved, and the overall performance and reliability of the system are enhanced. A multi-dimensional performance indicator system is established to realize the continuous adaptive optimization of the control strategy, so that the system can continuously adapt to the changing industrial environment. The present application realizes intelligent perception, precise control and fault diagnosis of complex industrial environments, thereby improving the performance and stability of the entire automation system. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0023] Figure 1 A schematic diagram of an embodiment of an intelligent signal I / O control method for industrial automation in an embodiment of the present application;

[0024] Figure 2 This is a schematic diagram of an embodiment of an intelligent signal I / O control system for industrial automation in an embodiment of the present application. DETAILED DESCRIPTION

[0025] The present application embodiment provides a kind of intelligent signal I / O control system and method thereof for industrial automation. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or precedence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described here can be implemented in an order other than the content illustrated or described here. In addition, the term "including" or "having" and any variation thereof are intended to cover non-exclusive inclusions, for example, the process, method, system, product or equipment comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0026] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 , an embodiment of the intelligent signal I / O control method for industrial automation in the embodiment of the present application includes:

[0027] Step S101, collecting and analyzing the characteristics of various signals in the industrial automation system to obtain a signal characteristic model;

[0028] It is understandable that the execution subject of the present application may be an intelligent signal I / O control system for industrial automation, or a terminal or a server, which is not specifically limited here. The present application embodiment is described by taking a server as the execution subject as an example.

[0029] Specifically, the temperature, pressure, flow, position, speed, switch quantity and analog quantity in the industrial automation system are collected to obtain the original signal data set. These data originate from various key nodes of the system and reflect different aspects of the system operation. The original signal data set is preprocessed. The preprocessing steps include signal denoising, smoothing and normalization to eliminate noise and interference in the data and obtain more accurate preprocessed signal data. According to the nature of the signal, the preprocessed signal data is classified into discrete signals and continuous signals to obtain the signal type identification. The preprocessed signal data is analyzed in the time domain, and the basic characteristics of the signal are preliminarily understood by calculating the time domain statistical characteristics such as the mean, variance, and peak value of the signal. At the same time, the preprocessed signal data is Fourier transformed to obtain the spectrum data of the signal, which reflects the frequency component and energy distribution of the signal. The signal spectrum data is analyzed to extract the frequency component and energy distribution characteristics of the signal to obtain the frequency domain characteristics. The preprocessed signal data is subjected to wavelet transform to obtain multi-scale wavelet coefficients. Through wavelet transform, the change of the signal can be observed at different scales, and the characteristics of the signal can be captured more carefully. According to the multi-scale wavelet coefficients, the energy proportion and entropy value of each scale are calculated to obtain the time-frequency domain features. These features can describe the energy distribution and complexity of the signal at different time scales. The time domain statistical features, frequency domain features and time-frequency domain features are fused to obtain a comprehensive feature vector. According to the signal type identification and the comprehensive feature vector, a signal feature model including signal type, change characteristics and interference factors is constructed. This model can not only accurately describe the various characteristics of the signal, but also identify the interference factors in the signal.

[0030] Step S102, designing a two-level I / O control scheme for the signal feature model to obtain a first control strategy including a predictive control layer and a real-time response layer;

[0031] Specifically, the signal feature model is subjected to signal time series feature extraction to obtain a signal time series feature set. By analyzing the historical data and change rules of the signal, important features that can reflect the dynamic changes of the signal are extracted. According to the signal time series feature set, a predictive control layer model is constructed using a long short-term memory network. The long short-term memory network captures the long-term dependency and short-term change characteristics of the signal, and effectively predicts the signal change trend. The predictive control layer model is used to predict the signal change trend and obtain a signal prediction sequence. These prediction sequences reveal the possible change trend of the signal at a future moment. Based on the signal prediction sequence, a predictive control layer parameter set is generated. At the same time, the signal feature model is subjected to real-time response feature extraction to obtain an initial real-time response layer model. The model is established based on the current state and instantaneous characteristics of the signal so as to promptly respond to various emergencies that may occur in the system. The initial real-time response layer model is rule defined and optimized to form a regularized real-time response layer model. By setting a series of response rules and optimization strategies, the real-time response layer can respond quickly and accurately. Based on the regularized real-time response layer model, the emergency processing flow of the real-time response layer is analyzed to obtain the real-time response control strategy, and the specific operation steps that the real-time response layer needs to take in different emergency situations are clarified to ensure the stable operation and rapid recovery capability of the system. The predictive control layer parameter set and the real-time response control strategy are integrated and dynamically weighted to form a two-level control coordination mechanism. According to the two-level control coordination mechanism, the outputs of the predictive control layer and the real-time response layer are integrated to obtain the first control strategy.

[0032] Step S103, constructing a state-trend matrix and introducing an evaluation function to quantify the control effect, dynamically iterating and updating the first control strategy to obtain target control parameters;

[0033] Specifically, various signals in the industrial automation system are sampled in real time to obtain instant data of the signals and form a signal state data set. The data set is analyzed to calculate the change rate and acceleration of various signals to obtain a signal trend data set. A state-trend matrix is ​​constructed based on the signal state data set and the signal trend data set. The design of the matrix makes each row represent a signal, and each column contains the state characteristics and trend characteristics of the signal. In order to quantify the control effect, a multi-objective evaluation function is created according to the system control goal of the industrial automation system. The function comprehensively considers multiple control goals, such as system stability, response speed, accuracy, etc., and generates a control effect quantitative model through the comprehensive evaluation of these goals. Based on this quantitative model, the effect of the first control strategy is evaluated and the control effect score is calculated. The scoring system can objectively reflect the performance of the current control strategy in different dimensions. According to the control effect score, it is determined whether the control strategy needs to be updated. If the scoring result shows that the current strategy does not achieve the expected effect, it is necessary to enter the optimization process; if the scoring result is good, the strategy remains unchanged. In the optimization process, the control parameters are iteratively optimized with the goal of minimizing the multi-objective evaluation function. By continuously adjusting the parameter values, we can find the parameter combination that can maximize the control effect and obtain the optimal control parameters. The optimal control parameters are applied to the first control strategy, including updating the parameters of the predictive control layer and the real-time response layer to adapt to new control requirements and environmental changes, and obtain the target control parameters.

[0034] Step S104, performing PLC autonomous fault diagnosis on various signals according to target control parameters to obtain fault diagnosis results;

[0035] Specifically, according to the target control parameters, the features of various signals are extracted to form a signal feature vector set. Based on the signal feature data set, the parameters of the particle swarm optimization algorithm are initialized to generate an initial particle swarm. The generation of the initial particle swarm determines the scope of the search space and the direction of optimization. The fitness of the initial particle swarm is evaluated to calculate the initial fitness value of each particle. The fitness value reflects the quality of the particle solution in the current state. According to the fitness value, the particles in the particle swarm are classified into global optimal solutions, local optimal solutions and random search solutions to form a classified particle swarm. Classification helps different particles adopt different search strategies during the optimization process, thereby improving the optimization efficiency. The classified particle swarm is iteratively optimized to update the position and velocity of the particles and generate a new generation of particle swarms. In this process, an adaptive inertia weight mechanism is introduced to dynamically adjust the inertia weight according to the number of iterations. The adaptive inertia weight mechanism can increase the search range in the early stage of optimization and enhance the accuracy of the solution in the later stage of optimization to obtain an optimized particle swarm. A redundancy evaluation function is designed to evaluate the particles in each iteration, determine the features with redundant information, and screen out the target feature set. The design of the redundancy evaluation function can effectively eliminate redundant features and improve the accuracy and efficiency of fault diagnosis. Based on the obtained target feature set and the optimized particle swarm, fault pattern recognition is performed to generate a list of potential fault points. The probability evaluation of the list of potential fault points is performed, and the confidence of each fault point is calculated to obtain the final fault diagnosis result. The confidence reflects the reliability of each fault point and provides a basis for fault handling and decision-making.

[0036] Step S105: Based on the fault diagnosis result, a network collaborative control analysis is performed on multiple PLCs to obtain a hierarchical central coordination mechanism;

[0037] Specifically, the fault diagnosis results are analyzed to identify the fault status and severity of each PLC, and a PLC fault status matrix is ​​established. Based on this matrix, multiple PLCs are functionally divided, the specific tasks and roles of each PLC are determined, and an initial PLC collaborative control scheme is formed. This scheme aims to optimize resource allocation and ensure that each PLC can perform the most appropriate task according to its status. Based on the initial PLC collaborative control scheme, a PLC network topology diagram is constructed. The construction of the topology diagram takes into account the communication links and data flows between each PLC to form a comprehensive PLC network structure model. This model not only shows the physical connection relationship between each PLC, but also reveals the specific flow path of data flow in the network. The PLC network structure model is analyzed to evaluate the reliability and stability of the network, and the initial hierarchical coordination structure is designed based on the obtained network reliability evaluation report. The structure includes a data exchange layer, a task coordination layer, and a fault tolerance layer. Each layer is responsible for different functions to improve the efficiency and reliability of the overall system. The data exchange layer in the initial hierarchical coordination structure is optimized to establish an efficient data exchange mechanism to ensure that data can be transmitted quickly and accurately between each PLC. At the same time, for the task coordination layer, the task allocation strategy is adjusted in real time according to the load and network conditions of each PLC to form an adaptive task coordination strategy. The flexibility of this strategy enables the system to dynamically adjust the task allocation of each PLC according to the actual situation, thereby improving the operating efficiency of the overall system. For the fault tolerance layer, a distributed decision-making model based on multi-agents is constructed. The core of this model is the distributed fault handling mechanism, which can make decisions and respond quickly when a fault is detected to ensure the continuity and stability of the system. Through the integration of efficient data exchange mechanism, adaptive task coordination strategy and distributed fault handling mechanism, a hierarchical central coordination mechanism is finally formed. This mechanism can not only effectively coordinate the task allocation and fault handling of each PLC, but also improve the overall efficiency and reliability of the system by optimizing data exchange and task coordination.

[0038] Step S106: Establish a multi-dimensional performance indicator system for the hierarchical central coordination mechanism, and evaluate and adaptively optimize the first control strategy to obtain a second control strategy.

[0039] Specifically, the hierarchical central coordination mechanism is decomposed to form an initial performance indicator set. The initial performance indicator set includes various key aspects of the system, including response time, system stability, resource utilization, etc. The hierarchical analysis model is used to assign weights to each performance indicator, determine their relative importance among multiple performance indicators, and form a weighted performance indicator system. The weighted performance indicator system is subjected to multi-dimensional quantitative analysis, and a multi-dimensional performance indicator system is constructed through quantitative analysis of each performance indicator. Based on the multi-dimensional performance indicator system, the strategy performance indicator of the first control strategy is calculated to generate a strategy performance scoring matrix. The matrix covers the performance of the control strategy in different dimensions. According to the strategy performance scoring matrix, the strategy optimization direction of the first control strategy is determined, and the specific areas that need to be improved and the potential optimization paths are pointed out. After the optimization direction is clarified, it is adaptively optimized to form a strategy optimization model. The model proposes specific optimization measures by combining the current state and goal of the system. Based on the strategy optimization model, a Markov decision process is constructed, the state space, action space and reward function are defined, and a reinforcement learning environment is established. In this environment, multiple candidate optimization strategy sets are generated through iterative optimization of the first control strategy. Each candidate strategy represents a possible optimization solution. After optimizing the candidate optimization strategy set, the optimal control strategy is obtained. The optimal control strategy is integrated with the first control strategy, specifically including adjusting the control parameters and decision logic to ensure that the new strategy can be effectively applied to the actual system, and the second control strategy is obtained.

[0040] In the embodiment of the present application, by conducting a comprehensive feature analysis of various types of signals in the industrial automation system, a signal feature model including signal type, change characteristics and interference factors is established. A two-level control structure of a predictive control layer and a real-time response layer is adopted, which can not only predict the trend of signal changes, but also quickly respond to emergencies, thereby improving the forward-looking and real-time nature of the control. The introduction of the state-trend matrix and the evaluation function realizes the dynamic iterative update of the control strategy, so that the control parameters can be continuously optimized according to the system state and the control target. The PLC autonomous fault diagnosis method based on the particle swarm optimization algorithm improves the accuracy and efficiency of fault detection. Through the hierarchical central coordination mechanism, efficient collaborative control between multiple PLCs is achieved, and the overall performance and reliability of the system are enhanced. A multi-dimensional performance indicator system is established to realize the continuous adaptive optimization of the control strategy, so that the system can continuously adapt to the changing industrial environment. The present application realizes intelligent perception, precise control and fault diagnosis of complex industrial environments, thereby improving the performance and stability of the entire automation system.

[0041] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0042] (1) Collect temperature, pressure, flow, position, speed, switch quantity and analog quantity in industrial automation system to obtain original signal data set;

[0043] (2) preprocessing the original signal data set to obtain preprocessed signal data, and classifying the preprocessed signal data into discrete signals and continuous signals to obtain a signal type identifier;

[0044] (3) performing time domain analysis on the preprocessed signal data to obtain time domain statistical features, and performing Fourier transform on the preprocessed signal data to obtain signal spectrum data;

[0045] (4) extracting the frequency components and energy distribution of the signal from the signal spectrum data to obtain frequency domain features, and performing wavelet transform on the preprocessed signal data to obtain multi-scale wavelet coefficients;

[0046] (5) According to the multi-scale wavelet coefficients, the energy proportion and entropy value of each scale are calculated to obtain the time-frequency domain features, and the time-domain statistical features, frequency domain features and time-frequency domain features are fused to obtain a comprehensive feature vector;

[0047] (6) Based on the signal type identification and comprehensive feature vector, a signal feature model including signal type, change characteristics and interference factors is constructed.

[0048] Specifically, various signals such as temperature, pressure, flow, position, speed, switch quantity and analog quantity are collected to obtain the original signal data set. The collection process is usually achieved through sensors installed at different positions. These sensors monitor various physical quantities of the system in real time and generate continuous data streams. The original signal data set contains the original readings of various signals during the operation of the system. These data have noise, fluctuations and other interferences. The original signal data set is preprocessed, including denoising, smoothing and normalization processes to obtain preprocessed signal data. Denoising can eliminate noise introduced by sensor accuracy problems or external interference; smoothing is used to eliminate short-term fluctuations to make the data more stable; normalization unifies data of different units and ranges into a standard range to facilitate subsequent analysis. The preprocessed signal data is classified into discrete signals and continuous signals to obtain signal type identification. Discrete signals usually include switch quantity signals, such as switch status (on / off), etc., while continuous signals include physical quantities such as temperature and pressure, which change continuously on the time axis. The preprocessed signal data is analyzed in the time domain to extract a series of time domain statistical features. These features include mean, variance, skewness, kurtosis, etc., which describe the statistical characteristics of the signal in the time dimension. Perform Fourier transform on the preprocessed signal data to obtain the spectrum data of the signal. Fourier transform converts the time domain signal into the frequency domain signal and reveals the frequency components of the signal. Let X(f) be the Fourier transform result of the signal x(t). Each frequency component f in the spectrum data corresponds to a complex number X(f), whose modulus represents the amplitude of the frequency component and the angle represents the phase. By analyzing the spectrum data, the frequency components and energy distribution of the signal are extracted to form frequency domain features. These features help identify the periodic components and main frequencies in the signal. Perform wavelet transform on the preprocessed signal data to obtain multi-scale wavelet coefficients. Wavelet transform provides a multi-resolution analysis method that can locate the characteristics of the signal in time and frequency at the same time. According to the multi-scale wavelet coefficients, the energy proportion and entropy value of each scale are calculated to obtain the time-frequency domain features. The energy proportion reflects the energy distribution of the signal at different scales, and the entropy value represents the complexity and uncertainty of the signal. The time domain statistical features, frequency domain features and time-frequency domain features are fused to obtain a comprehensive feature vector. According to the signal type identification and the comprehensive feature vector, a signal feature model including signal type, change characteristics and interference factors is constructed. For example, in an industrial control system, the feature model of a temperature signal may include information such as average temperature, temperature fluctuation amplitude, main fluctuation frequency, and interference factors that may affect temperature changes, such as environmental noise and electromagnetic interference. This information is used for further analysis, such as fault prediction and status monitoring, to improve the stability and efficiency of the system.

[0049] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0050] (1) Extracting signal time series features from the signal feature model to obtain a signal time series feature set, and constructing a predictive control layer model based on the signal time series feature set and the long short-term memory network;

[0051] (2) predicting the signal change trend through the predictive control layer model to obtain a signal prediction sequence, and generating a predictive control layer parameter set based on the signal prediction sequence;

[0052] (3) extracting real-time response features from the signal feature model to obtain an initial real-time response layer model, and defining and optimizing rules for the initial real-time response layer model to obtain a regularized real-time response layer model;

[0053] (4) Based on the regularized real-time response layer model, the emergency processing flow of the real-time response layer is analyzed to obtain the real-time response control strategy, and the predictive control layer parameter set and the real-time response control strategy are integrated and dynamically weighted to obtain a two-level control coordination mechanism;

[0054] (5) The outputs of the predictive control layer and the real-time response layer are integrated according to the two-level control coordination mechanism to obtain the first control strategy.

[0055] Specifically, the signal time series feature set is obtained by extracting the signal time series feature from the signal feature model. The process of signal time series feature extraction includes the collection and analysis of signal history data. Assume that x(t) represents the signal value at time t, and the time series data consists of continuous signal values ​​{x(t1), x(t2), ..., x(t n )}. After smoothing and denoising, these data are extracted to reflect the characteristics of signal changes, such as trends, periodicity and seasonal components. A predictive control layer model is constructed based on the signal time series feature set and the long short-term memory network (LSTM). The LSTM network is a special recursive neural network that is suitable for processing and predicting time-based sequence data. Its structure can capture long-term dependencies and help accurately predict signal change trends. Assume h t is the hidden state at time t, and the update formula of LSTM is:

[0056] h t =σ(W h ·[h t-1 , x t ]+b h );

[0057] Among them, σ is the activation function, W h is the weight matrix, b h is the bias vector, h t -1 is the hidden state at the previous moment, x tis the input signal value at the current moment. By training the LSTM network, the future trend of the signal is predicted and a signal prediction sequence is generated. These predicted values ​​constitute the predictive control layer parameter set. The signal feature model is subjected to real-time response feature extraction to obtain the initial real-time response layer model. This process involves analyzing and extracting the instantaneous features of the signal, focusing on features such as sudden changes and abnormal fluctuations of the signal. For example, in a temperature control system, the real-time response features may include the instantaneous rate of change of temperature and the detection of sudden rises and falls. Based on these features, the initial real-time response layer model is rule defined and optimized to form a regularized real-time response layer model. The rule definition process includes setting response rules for different scenarios, such as starting a cooling mechanism when the temperature rises above a set threshold. Based on the regularized real-time response layer model, the emergency processing process is analyzed to obtain a real-time response control strategy, which includes specific operating steps in various emergency situations to ensure that the system can respond quickly and take appropriate measures. For example, if the pressure is detected to be too high, the system should immediately reduce the pressure or sound an alarm. The predictive control layer parameter set is integrated with the real-time response control strategy, and dynamic weight allocation is performed to form a two-level control coordination mechanism. The dynamic adjustment of weight allocation can be determined based on the current system state and prediction results to ensure the optimal control effect. The outputs of the predictive control layer and the real-time response layer are integrated according to the two-level control coordination mechanism to form the first control strategy. The strategy combines the foresight of predictive control and the immediacy of real-time response, enabling the system to foresee future changes and make adjustments in advance, as well as respond quickly in emergency situations. Dynamic weight allocation is performed using the following formula:

[0058] u(t)=α·u pred (t)+(1-α)·u resp (t);

[0059] Among them, u(t) is the final control output, u pred (t) is the output of the predictive control layer, u resp (t) is the output of the real-time response layer, and a is the dynamic weight, ranging from 0 to 1. By adjusting α, a balance is achieved between predictive control and real-time response to ensure the optimal control strategy of the system. Through dynamic weight allocation, under normal circumstances, predictive control is mainly relied on, and the weight α is large; in emergencies, the weight of the real-time response layer is increased, and α is reduced to ensure that the system responds quickly.

[0060] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0061] (1) Perform real-time sampling of various signals in the industrial automation system to obtain a signal state data set, and calculate the signal change rate and acceleration of the signal state data set to obtain a signal trend data set;

[0062] (2) constructing a state-trend matrix based on the signal state data set and the signal trend data set, where each row in the state-trend matrix represents a signal and each column contains state and trend features;

[0063] (3) creating a multi-objective evaluation function according to the system control objectives of the industrial automation system, generating a control effect quantification model according to the multi-objective evaluation function, and evaluating the effect of the first control strategy based on the control effect quantification model to obtain a control effect score;

[0064] (4) According to the control effect score, determine whether the control strategy needs to be updated. If it needs to be updated, enter the optimization process. If it does not need to be updated, keep the first control strategy unchanged;

[0065] (5) In the optimization process, the control parameters are iteratively optimized with the goal of minimizing the multi-objective evaluation function to obtain the optimal control parameters, and the optimal control parameters are applied to the first control strategy to update the parameters of the predictive control layer and the real-time response layer to obtain the target control parameters.

[0066] Specifically, various signals in the industrial automation system are sampled in real time to obtain a signal status data set. These signals include temperature, pressure, flow, position, speed, etc. The signal status data set collected in real time through the sensor network reflects the current working state of the system. The signal change rate and acceleration are calculated for the signal status data set. The change rate is usually used to measure the growth or decrease speed of the signal, while the acceleration indicates the change in the speed of signal change, that is, the acceleration of the change. The calculation formulas for the change rate and acceleration are:

[0067]

[0068]

[0069] Among them, ΔS represents the change of the signal within the time interval Δt. These calculations provide trend data for each signal to form a signal trend data set. Based on the signal state data set and the signal trend data set, a state-trend matrix is ​​constructed. Each row in the matrix represents an independent signal, and each column contains the state characteristics (such as current value, rate of change) and trend characteristics (such as acceleration) of the signal. For example, in a factory cooling system, temperature sensor data can show the current temperature state, while pressure sensor data provides the change trend of system pressure. In order to quantitatively evaluate the control strategy, a multi-objective evaluation function is created based on the signal characteristics and the control objectives of the industrial automation system. These objectives may include system stability, response time, energy efficiency, etc. The multi-objective evaluation function quantifies multiple objectives into a comprehensive evaluation index, usually in the form of a weighted sum:

[0070]

[0071] Among them, F is the comprehensive evaluation value, f ( (x) represents the i-th objective function, w ( is the corresponding weight coefficient, which indicates the importance of the target. Based on the multi-objective evaluation function, a control effect quantification model is generated, which is used to evaluate the effect of the current control strategy, that is, the first control strategy. The first control strategy is evaluated through the control effect quantification model to obtain a control effect score. The score reflects the performance of the current strategy in achieving various goals. According to the control effect score, it is determined whether the control strategy needs to be updated. If the score indicates that the current strategy fails to effectively achieve the system goal, it is necessary to enter the optimization process; if the score is high, the current strategy remains unchanged. In the optimization process, the control parameters are iteratively optimized with the goal of minimizing the multi-objective evaluation function. This process involves adjusting each control parameter to find the optimal combination that can improve the comprehensive evaluation value. The optimization algorithm may use methods such as genetic algorithms and particle swarm optimization to obtain the optimal control parameters through continuous trial and iteration. The optimal control parameters are applied to the first control strategy, and the parameters of the predictive control layer and the real-time response layer are updated to obtain new target control parameters. For example, in the above cooling system, if the temperature control is unstable, the optimization process may adjust the control parameters of the cooling water flow to improve the system response speed and temperature control accuracy. Through iterative optimization, the system can continuously improve its control strategy to adapt to different operating conditions and target requirements.

[0072] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0073] (1) According to the target control parameters, the features of various signals are extracted to obtain the signal feature vector set, and based on the signal feature data set, the parameters of the particle swarm optimization algorithm are initialized to generate the initial particle swarm;

[0074] (2) Perform fitness evaluation on the initial particle swarm to obtain the initial fitness value of each particle, and classify the particles into global optimal solution, local optimal solution and random search solution according to the initial fitness value to obtain the classified particle swarm;

[0075] (3) Iteratively optimize the classified particle swarm, update the particle position and velocity, obtain a new generation of particle swarm, and introduce an adaptive inertia weight mechanism to dynamically adjust the inertia weight according to the number of iterations to obtain the optimized particle swarm;

[0076] (4) Design a redundancy evaluation function to evaluate the particles in each iteration to obtain the target feature set, and perform fault mode recognition based on the target feature set and the optimized particle swarm to obtain a list of potential fault points;

[0077] (5) Perform probability evaluation on the list of potential fault points and calculate the confidence level of each fault point to obtain the fault diagnosis result.

[0078] Specifically, by real-time acquisition of temperature, pressure, flow, position, speed and other signals in the system, the characteristics of these signals such as mean, variance, frequency component, etc. are extracted to form a signal feature vector set. Each feature vector contains a set of eigenvalues ​​that reflect the state and dynamic behavior of the signal. Let x i represents the eigenvector of the i-th signal, x i ={f i1 , f i2 , ..., f in}, where f ij Represents the jth eigenvalue. Based on the signal feature data set, the parameters of the particle swarm optimization (PSO) algorithm are initialized, including the number of particles, the maximum number of iterations, the initial speed and position, etc. The particle swarm optimization algorithm is an optimization algorithm based on swarm intelligence, which simulates the foraging behavior of bird flocks in nature. Each particle represents a possible solution, its position represents the parameter value of the current solution, and its speed represents the changing trend of the solution. The generation of the initial particle swarm is completed by randomly distributing the position and speed of the particles in the solution space. The formula is expressed as:

[0079]

[0080]

[0081] in, and are the initial velocity and position of the ith particle in the jth dimension, v min , v max are the minimum and maximum values ​​of the velocity, x min , x max is the range of positions, and rand() is a random number between 0 and 1. The initial particle swarm is evaluated for fitness to measure the quality of the solution represented by each particle. The fitness function is selected based on specific optimization goals, such as minimizing energy consumption, maximizing production efficiency, etc. The higher the fitness value, the better the solution. According to the initial fitness value, the particles are classified into global optimal solutions, local optimal solutions, and random search solutions. The global optimal solution is the particle with the highest fitness value in the current particle swarm, the local optimal solution is the particle with the highest fitness value in a subgroup, and the random search solution is used to explore new solution spaces. Based on the classified particle swarm, iterative optimization is performed. By updating the position and velocity of the particles, the particle swarm continuously approaches the optimal solution. The formula for updating the particle velocity is:

[0082]

[0083]

[0084] Among them, ω is the inertia weight, c1, c2 are learning factors, r1, r2 are random numbers, p ij is the historical optimal position of the particle, g j is the global optimal position. In order to improve the performance of the algorithm, an adaptive inertia weight mechanism is introduced. The inertia weight ω changes with the number of iterations, usually decreasing linearly from a larger value to a smaller value, encouraging the algorithm to search extensively in the early stage and finely in the later stage. In each iteration, the particles are evaluated using the designed redundancy evaluation function to obtain the target feature set. The redundancy evaluation function is used to identify and eliminate redundant information in signal features and retain the most valuable features for fault detection. Based on the optimized particle swarm and target feature set, fault pattern identification is performed to form a list of potential fault points. This process includes clustering analysis of signal features, classification model training, etc. The list of potential fault points is probabilistically evaluated, the confidence of each fault point is calculated, and the fault diagnosis result is obtained. The confidence calculation takes into account factors such as historical data of the fault mode and feature consistency. The formula is as follows:

[0085]

[0086] Among them, P(fault|feature ( ) is the probability of failure under a specific feature, and n is the number of features.

[0087] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0088] (1) Analyze the fault diagnosis results, identify the fault status and severity of each PLC, obtain the PLC fault status matrix, and divide the functions and assign tasks to multiple PLCs based on the PLC fault status matrix to obtain the initial PLC collaborative control solution;

[0089] (2) Based on the initial PLC collaborative control scheme, a PLC network topology diagram is constructed, and a PLC network structure model is generated according to the communication links and data flows in the PLC network topology diagram;

[0090] (3) Analyze the PLC network structure model to obtain a network reliability evaluation report, and design an initial hierarchical coordination structure based on the network reliability evaluation report. The initial hierarchical coordination structure includes a data exchange layer, a task coordination layer, and a fault tolerance layer;

[0091] (4) The data exchange layer in the initial hierarchical coordination structure is optimized to obtain an efficient data exchange mechanism. For the task coordination layer, task allocation is adjusted in real time according to PLC load and network conditions to obtain an adaptive task coordination strategy.

[0092] (5) For the fault-tolerant layer, a distributed decision-making model based on multi-agents is constructed to generate a distributed fault handling mechanism. The efficient data exchange mechanism, adaptive task coordination strategy and distributed fault handling mechanism are integrated to obtain a hierarchical central coordination mechanism.

[0093] Specifically, the fault diagnosis system collects the diagnostic data of each PLC (programmable logic controller) to form a fault diagnosis result set. The result set includes the status information of each PLC, such as online status, fault type, fault frequency and severity. Using this information, a PLC fault status matrix is ​​constructed. Each row in the matrix represents a PLC, and each column represents a fault state or attribute. For example, the matrix element M (+Indicates the severity of the i-th PLC under the j-th fault state. According to the PLC fault state matrix, multiple PLCs are functionally divided and tasked. Functional division takes into account the health status and capabilities of each PLC, assigns tasks to the most suitable PLC to ensure the overall stability and efficiency of the system, and forms an initial PLC collaborative control scheme, which defines the responsibilities and operating scope of each PLC, such as which PLCs are responsible for the main control tasks and which PLCs are used as backup or support systems. Based on the initial PLC collaborative control scheme, a PLC network topology diagram is constructed. The topology diagram shows the connection relationship and communication links between each PLC, reflecting the physical and logical structure of the entire network. Based on the topology diagram, a PLC network structure model is generated. The model includes information about all PLC nodes and their connections, especially data flow and communication paths. Analysis of data flow helps to understand how information propagates in the network, thereby optimizing network performance. The PLC network structure model is analyzed to generate a network reliability assessment report. The report evaluates key indicators such as network stability, communication efficiency, and fault tolerance. Through the assessment, potential bottlenecks and vulnerabilities in the network are identified. Based on the network reliability assessment report, the initial hierarchical coordination structure is designed. The initial hierarchical coordination structure usually includes a data exchange layer, a task coordination layer, and a fault tolerance layer. The data exchange layer is responsible for efficient data transmission and processing, the task coordination layer manages PLC task allocation and coordination, and the fault tolerance layer provides fault detection and response mechanisms. The data exchange layer in the initial hierarchical coordination structure is optimized. The optimization goal is to establish an efficient data exchange mechanism, which includes optimizing communication protocols, data compression technology, and redundant path planning. Through these optimization measures, data transmission delays and packet loss rates can be significantly reduced, and the real-time performance of the system can be enhanced. In the task coordination layer, the task allocation strategy is adjusted in real time according to the load and network status of the PLC to form an adaptive task coordination strategy. The adaptive strategy allows the system to dynamically adjust the tasks of the PLC according to real-time data to avoid overloading or idleness of some PLCs. For the fault tolerance layer, a distributed decision-making model based on multi-agents is constructed. The multi-agent system simulates the collaboration and coordination capabilities of multiple autonomous agents (PLCs), allowing the system to automatically adjust when a fault is detected. For example, when a PLC fails, other PLCs can automatically take over its tasks to ensure the continuity of the system. The generated distributed fault handling mechanism includes the steps of fault detection, fault isolation and fault recovery. The effectiveness of these mechanisms depends on the collaboration and information sharing among multiple agents. The hierarchical central coordination mechanism is formed by integrating the efficient data exchange mechanism, the adaptive task coordination strategy and the distributed fault handling mechanism. This system ensures that the system can work efficiently and stably under various operating conditions. For example, in a complex manufacturing system, if a PLC cannot communicate due to network problems, the hierarchical central coordination mechanism will reallocate tasks and use other PLCs to ensure the continuous operation of the production line.In this way, the system not only improves fault recovery capabilities, but also enhances overall response speed and efficiency.

[0094] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0095] (1) Decompose the hierarchical central coordination mechanism to obtain an initial performance indicator set, and assign weights to each performance indicator based on the initial performance indicator set and the hierarchical analysis model to obtain a weighted performance indicator system;

[0096] (2) performing a multi-dimensional quantitative analysis on the weighted performance indicator system to obtain a multi-dimensional performance indicator system, and calculating a strategy performance indicator for the first control strategy based on the multi-dimensional performance indicator system to obtain a strategy performance scoring matrix;

[0097] (3) generating a strategy optimization direction of the first control strategy according to the strategy performance scoring matrix, and adaptively optimizing the strategy optimization direction to obtain a strategy optimization model;

[0098] (4) Based on the policy optimization model, a Markov decision process is constructed and the state space, action space and reward function are defined to obtain a reinforcement learning environment;

[0099] (5) In a reinforcement learning environment, the first control strategy is iteratively optimized to obtain a candidate optimization strategy set, and the candidate optimization strategy set is optimized to obtain an optimal control strategy;

[0100] (6) Integrate the optimal control strategy with the first control strategy, adjust the control parameters and decision logic, and obtain the second control strategy.

[0101] Specifically, by analyzing the hierarchical central coordination mechanism, an initial set of performance indicators is extracted. These indicators may include system response time, resource utilization, fault recovery speed, data transmission delay, etc. Each performance indicator reflects the performance of a certain aspect of the system. The hierarchical analysis model is used to assign weights to each performance indicator. The hierarchical analysis model quantifies qualitative judgments into quantitative analysis by constructing a judgment matrix to determine the relative importance of each indicator. The formula is as follows:

[0102]

[0103] Among them, W ( is the weight of the i-th performance indicator, a (j is an element in the judgment matrix, which indicates the importance of the i-th indicator relative to the j-th indicator, and n is the total number of indicators. In this way, a weighted performance indicator system is obtained. The weighted performance indicator system is subjected to multi-dimensional quantitative analysis, including standardization of the specific values ​​of each indicator to facilitate comparison and integration between different indicators. The result of the multi-dimensional quantitative analysis is a multi-dimensional performance indicator system. Based on the multi-dimensional performance indicator system, the strategy performance indicator of the first control strategy is calculated to obtain a strategy performance scoring matrix. Each element of the scoring matrix represents the performance score of the current control strategy under a specific dimension. According to the strategy performance scoring matrix, the shortcomings of the current control strategy are identified and the strategy optimization direction is generated. The strategy optimization direction points out the specific areas that need to be improved, such as a faster adjustment mechanism may be required for system response speed, and a better scheduling algorithm may be required for resource utilization. The strategy optimization direction is adaptively optimized to develop a strategy optimization model. Adaptive optimization means that the system can dynamically adjust its parameters to adapt to environmental changes and optimization goals, which usually involves complex algorithm design and parameter tuning. According to the strategy optimization model, a Markov decision process is constructed, and the state space, action space and reward function are defined to form a reinforcement learning environment. In this environment, the state space contains all possible states of the system, such as the current load situation, network status, etc.; the action space defines all possible actions that the system can take, such as adjusting load distribution, modifying network parameters, etc.; the reward function gives rewards or penalties based on the effect of each action to guide the optimization of the strategy. The formula is expressed as:

[0104] R(s,a)=E[r t61 ∣s t =s,a t =a];

[0105] Among them, R(s,a) is the expected reward obtained by performing action a in state s, r t +1 is the actual reward at time t+1. In a reinforcement learning environment, multiple candidate optimization strategy sets are generated by iteratively optimizing the first control strategy. Each candidate strategy achieves a specific optimization goal through different combinations and parameter settings. The candidate optimization strategy set is optimized and the optimal control strategy is found using reinforcement learning algorithms such as Q-learning or deep Q network (DQN). The optimal control strategy can not only maximize the system performance indicators, but also adapt to various environmental changes and improve the robustness of the system. The optimal control strategy is integrated with the first control strategy, and the control parameters and decision logic are adjusted to form a second control strategy. For example, in an intelligent manufacturing system, if the first control strategy focuses mainly on production efficiency, then through the above optimization, the second control strategy may simultaneously enhance the health monitoring capabilities and energy efficiency management of the equipment, thereby comprehensively improving the overall performance and reliability of the system.

[0106] The above describes the intelligent signal I / O control method for industrial automation in the embodiment of the present application. The following describes the intelligent signal I / O control system for industrial automation in the embodiment of the present application. Figure 2 In the embodiment of the present application, an embodiment of an intelligent signal I / O control system for industrial automation includes:

[0107] The acquisition module 201 is used to collect and analyze the characteristics of various signals in the industrial automation system to obtain a signal characteristic model;

[0108] A design module 202 is used to design a two-level I / O control scheme for the signal characteristic model to obtain a first control strategy including a predictive control layer and a real-time response layer;

[0109] An updating module 203 is used to construct a state-trend matrix and introduce an evaluation function to quantify the control effect, dynamically iterate and update the first control strategy, and obtain target control parameters;

[0110] The diagnosis module 204 is used to perform PLC autonomous fault diagnosis on various signals according to target control parameters to obtain fault diagnosis results;

[0111] An analysis module 205 is used to perform network collaborative control analysis on multiple PLCs based on the fault diagnosis results to obtain a hierarchical central coordination mechanism;

[0112] The optimization module 206 is used to establish a multi-dimensional performance indicator system of the hierarchical central coordination mechanism, and to evaluate and adaptively optimize the first control strategy to obtain the second control strategy.

[0113] Through the collaboration of the above components, a signal feature model including signal type, change characteristics and interference factors is established by conducting a comprehensive feature analysis of various signals in the industrial automation system. The two-level control structure of the predictive control layer and the real-time response layer is adopted, which can not only predict the trend of signal changes, but also respond quickly to emergencies, thereby improving the foresight and real-time performance of the control. The introduction of the state-trend matrix and the evaluation function realizes the dynamic iterative update of the control strategy, so that the control parameters can be continuously optimized according to the system state and control objectives. The PLC autonomous fault diagnosis method based on the particle swarm optimization algorithm improves the accuracy and efficiency of fault detection. Through the hierarchical central coordination mechanism, efficient collaborative control between multiple PLCs is achieved, and the overall performance and reliability of the system are enhanced. A multi-dimensional performance indicator system is established to realize the continuous adaptive optimization of the control strategy, so that the system can continuously adapt to the changing industrial environment. This application realizes intelligent perception, precise control and fault diagnosis of complex industrial environments, thereby improving the performance and stability of the entire automation system.

[0114] The present application also provides an intelligent signal I / O control device for industrial automation, the intelligent signal I / O control device for industrial automation comprising a memory and a processor, the memory storing computer-readable instructions, and when the computer-readable instructions are executed by the processor, the processor executes the steps of the intelligent signal I / O control method for industrial automation in the above-mentioned embodiments.

[0115] The present application also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the intelligent signal I / O control method for industrial automation.

[0116] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0117] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0118] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An intelligent signal I / O control method for industrial automation, characterized in that: The method comprises: Collect and analyze the characteristics of various signals in industrial automation systems to obtain signal characteristic models; Designing a two-level I / O control scheme for the signal characteristic model to obtain a first control strategy including a predictive control layer and a real-time response layer; Constructing a state-trend matrix and introducing an evaluation function to quantify the control effect, dynamically iterating and updating the first control strategy to obtain target control parameters; Perform PLC autonomous fault diagnosis on the various signals according to the target control parameters to obtain fault diagnosis results; Based on the fault diagnosis results, a network collaborative control analysis is performed on multiple PLCs to obtain a hierarchical central coordination mechanism; A multi-dimensional performance indicator system for the hierarchical central coordination mechanism is established, and the first control strategy is evaluated and adaptively optimized to obtain a second control strategy.

2. The intelligent signal I / O control method for industrial automation according to claim 1, characterized in that: The signal characteristic model is obtained by collecting and analyzing the characteristics of various signals in the industrial automation system, including: Collect temperature, pressure, flow, position, speed, switch quantity and analog quantity in industrial automation system to obtain original signal data set; Preprocessing the original signal data set to obtain preprocessed signal data, and classifying the preprocessed signal data into discrete signals and continuous signals to obtain a signal type identifier; Performing time domain analysis on the preprocessed signal data to obtain time domain statistical features, and performing Fourier transform on the preprocessed signal data to obtain signal spectrum data; Extracting the frequency components and energy distribution of the signal from the signal spectrum data to obtain frequency domain features, and performing wavelet transform on the pre-processed signal data to obtain multi-scale wavelet coefficients; According to the multi-scale wavelet coefficients, the energy proportion and entropy value of each scale are calculated to obtain time-frequency domain features, and the time-domain statistical features, the frequency domain features and the time-frequency domain features are subjected to feature fusion to obtain a comprehensive feature vector; A signal feature model including signal type, change feature and interference factor is constructed according to the signal type identifier and the comprehensive feature vector.

3. The intelligent signal I / O control method for industrial automation according to claim 1, characterized in that: The two-level I / O control scheme design is performed on the signal characteristic model to obtain a first control strategy including a predictive control layer and a real-time response layer, including: Extracting signal time series features from the signal feature model to obtain a signal time series feature set, and constructing a predictive control layer model based on the signal time series feature set and a long short-term memory network; Predicting the signal change trend through the predictive control layer model to obtain a signal prediction sequence, and generating a predictive control layer parameter set based on the signal prediction sequence; Extracting real-time response features from the signal feature model to obtain an initial real-time response layer model, and defining and optimizing rules for the initial real-time response layer model to obtain a regularized real-time response layer model; Based on the regularized real-time response layer model, an emergency processing flow analysis of the real-time response layer is performed to obtain a real-time response control strategy, and the predictive control layer parameter set and the real-time response control strategy are integrated and dynamically weighted to obtain a two-level control coordination mechanism; The outputs of the predictive control layer and the real-time response layer are integrated according to the two-level control coordination mechanism to obtain a first control strategy.

4. The intelligent signal I / O control method for industrial automation according to claim 3, characterized in that: The state-trend matrix is ​​constructed and an evaluation function is introduced to quantify the control effect, and the first control strategy is dynamically iterated and updated to obtain the target control parameters, including: Performing real-time sampling on various signals in the industrial automation system to obtain a signal state data set, and performing signal change rate and acceleration calculation on the signal state data set to obtain a signal trend data set; Constructing a state-trend matrix according to the signal state data set and the signal trend data set, wherein each row in the state-trend matrix represents a signal and each column contains state and trend features; Creating a multi-objective evaluation function according to the system control objectives of the industrial automation system, generating a control effect quantification model according to the multi-objective evaluation function, and evaluating the effect of the first control strategy based on the control effect quantification model to obtain a control effect score; According to the control effect score, determine whether the control strategy needs to be updated, if it needs to be updated, enter the optimization process, if it does not need to be updated, keep the first control strategy unchanged; In the optimization process, the control parameters are iteratively optimized with the goal of minimizing the multi-objective evaluation function to obtain the optimal control parameters, and the optimal control parameters are applied to the first control strategy, the parameters of the predictive control layer and the real-time response layer are updated to obtain the target control parameters.

5. The intelligent signal I / O control method for industrial automation according to claim 1, characterized in that: The performing PLC autonomous fault diagnosis on the various types of signals according to the target control parameters to obtain fault diagnosis results includes: According to the target control parameters, feature extraction is performed on the various types of signals to obtain a signal feature vector set, and based on the signal feature data set, parameters of a particle swarm optimization algorithm are initialized to generate an initial particle swarm; Performing fitness evaluation on the initial particle swarm to obtain an initial fitness value of each particle, and classifying the particles into global optimal solutions, local optimal solutions, and random search solutions according to the initial fitness values ​​to obtain a classified particle swarm; Iteratively optimizing the classified particle swarm, updating the particle position and velocity to obtain a new generation of particle swarm, and introducing an adaptive inertia weight mechanism to dynamically adjust the inertia weight according to the number of iterations to obtain an optimized particle swarm; Designing a redundancy evaluation function, evaluating the particles in each iteration to obtain a target feature set, and performing fault mode identification based on the target feature set and the optimized particle swarm to obtain a list of potential fault points; A probability evaluation is performed on the potential fault point list, and the confidence level of each fault point is calculated to obtain a fault diagnosis result.

6. The intelligent signal I / O control method for industrial automation according to claim 5, characterized in that: Based on the fault diagnosis result, a network collaborative control analysis is performed on multiple PLCs to obtain a hierarchical central coordination mechanism, including: Analyze the fault diagnosis results, identify the fault status and severity of each PLC, obtain a PLC fault status matrix, and perform functional division and task allocation on multiple PLCs according to the PLC fault status matrix to obtain an initial PLC collaborative control solution; Based on the initial PLC collaborative control scheme, a PLC network topology diagram is constructed, and according to the communication links and data flows in the PLC network topology diagram, a PLC network structure model is generated; Analyze the PLC network structure model to obtain a network reliability evaluation report, and design an initial hierarchical coordination structure according to the network reliability evaluation report, wherein the initial hierarchical coordination structure includes a data exchange layer, a task coordination layer, and a fault tolerance layer; The data exchange layer in the initial hierarchical coordination structure is optimized to obtain an efficient data exchange mechanism, and for the task coordination layer, task allocation is adjusted in real time according to PLC load and network conditions to obtain an adaptive task coordination strategy; For the fault-tolerant layer, a distributed decision-making model based on multi-agents is constructed to generate a distributed fault handling mechanism, and the efficient data exchange mechanism, the adaptive task coordination strategy and the distributed fault handling mechanism are integrated to obtain a hierarchical central coordination mechanism.

7. The intelligent signal I / O control method for industrial automation according to claim 1, characterized in that: The multi-dimensional performance indicator system of the hierarchical central coordination mechanism is established, and the first control strategy is evaluated and adaptively optimized to obtain the second control strategy, including: Decomposing the hierarchical central coordination mechanism to obtain an initial performance indicator set, and assigning weights to each performance indicator according to the initial performance indicator set and a hierarchical analysis model to obtain a weighted performance indicator system; Performing a multi-dimensional quantitative analysis on the weighted performance indicator system to obtain a multi-dimensional performance indicator system, and based on the multi-dimensional performance indicator system, performing a strategy performance indicator calculation on the first control strategy to obtain a strategy performance scoring matrix; Generating a strategy optimization direction of the first control strategy according to the strategy performance scoring matrix, and adaptively optimizing the strategy optimization direction to obtain a strategy optimization model; According to the strategy optimization model, a Markov decision process is constructed and a state space, an action space and a reward function are defined to obtain a reinforcement learning environment; In the reinforcement learning environment, iteratively optimizing the first control strategy to obtain a candidate optimization strategy set, and optimizing the candidate optimization strategy set to obtain an optimal control strategy; The optimal control strategy is integrated with the first control strategy, and control parameters and decision logic are adjusted to obtain a second control strategy.

8. An intelligent signal I / O control system for industrial automation, characterized in that: For executing the intelligent signal I / O control method for industrial automation according to any one of claims 1 to 7, the system comprises: The acquisition module is used to collect and analyze the characteristics of various signals in the industrial automation system to obtain the signal characteristic model; A design module, used for designing a two-level I / O control scheme for the signal characteristic model to obtain a first control strategy including a predictive control layer and a real-time response layer; An updating module, used for constructing a state-trend matrix and introducing an evaluation function to quantify the control effect, dynamically iterating and updating the first control strategy to obtain target control parameters; A diagnosis module, used for performing PLC autonomous fault diagnosis on the various signals according to the target control parameters to obtain a fault diagnosis result; An analysis module, used for performing network collaborative control analysis on multiple PLCs based on the fault diagnosis results to obtain a hierarchical central coordination mechanism; The optimization module is used to establish a multi-dimensional performance indicator system for the hierarchical central coordination mechanism, and to evaluate and adaptively optimize the first control strategy to obtain a second control strategy.

9. An intelligent signal I / O control device for industrial automation, characterized in that: The intelligent signal I / O control device for industrial automation comprises: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory to enable the intelligent signal I / O control device for industrial automation to execute the intelligent signal I / O control method for industrial automation as described in any one of claims 1-7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the intelligent signal I / O control method for industrial automation as described in any one of claims 1 to 7 is implemented.

Images