Automatic control system and method for pneumatic control valve

Abstract

The invention relates to the technical field of equipment control, in particular to an automatic control system and method for a pneumatic control valve, and the method comprises the steps that a displacement sensing module collects a valve element displacement signal through a magnetostrictive sensor and carries out Kalman filtering processing; the signal conditioning module fuses displacement and pressure data by adopting a fuzzy logic rule; the intelligent conversion module performs state identification by using a support vector machine algorithm; the communication gateway module realizes data protocol encapsulation and transmission; the state diagnosis module performs fault diagnosis based on a hidden Markov model; the digital twin module realizes simulation prediction through multi-domain physical modeling; the optimization decision module adopts a particle swarm optimization algorithm to generate a control instruction; and the closed-loop execution module drives the execution mechanism and feeds back displacement data. The problem that an existing pneumatic valve is insufficient in control precision due to lack of displacement feedback is solved, and real-time monitoring and high-precision closed-loop control over the position of the valve are achieved.

Description

Technical Field

[0001] This invention relates to the field of equipment control technology, and in particular to an automated control system and method for pneumatic control valves. Background Technology

[0002] Pneumatic control valve technology uses compressed air as the power medium. A pneumatic actuator converts control signals into mechanical displacement, driving the valve core to adjust the cross-sectional area of ​​the fluid passage, thereby controlling flow, pressure, or direction. Due to the simple structure, rapid action, and adaptability to harsh environments, pneumatic control valves are commonly used in industrial automation processes where rapid response and reliable operation are required, such as maintaining the stability of process parameters in process control.

[0003] Existing pneumatic control valve technology suffers from the following technical challenges: Current electromagnetically controlled pneumatic valves rely solely on current on / off signals to infer the valve core or push rod's position, failing to directly obtain the actuator's actual displacement feedback. Long-term operation of solenoid valves can lead to mechanical wear, coil aging, or interference from power fluctuations, causing deviations between the actual push rod stroke and the intended command, resulting in incomplete valve closure or incomplete opening. In process industries such as chemical material transport or energy pipeline control, if critical shut-off valves fail to close completely due to internal jamming or wear, and operators cannot remotely verify valve status, there is a risk of media leakage or process interruption. This often necessitates on-site manual verification, increasing maintenance costs and safety uncertainties, thus limiting the reliability and accuracy of the automatic control system. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an automated control system and method for pneumatic control valves, solving the technical problem that the lack of reliable feedback on the switching status of existing solenoid valves prevents the automatic control system from achieving closed-loop monitoring and precise control of valve position.

[0005] To solve the above-mentioned technical problems, the specific contents of the present invention are as follows: In a first aspect, the present invention provides an automated control system for a pneumatic control valve, comprising: The displacement sensing module is configured to acquire valve core displacement signals and process them to obtain digital displacement values. The signal conditioning module connects to the displacement sensing module, receives digital displacement and pressure data, performs fusion processing, and outputs standard industrial communication protocol data frames. The intelligent conversion module, connected to the signal conditioning module, extracts displacement and pressure information from the data frame and performs status recognition to generate a digital signal with status identifier; The communication gateway module connects to the intelligent conversion module, encapsulates digital signals according to protocols, and transmits them to the status diagnosis module and the digital twin module; The status diagnosis module connects to the communication gateway module, receives real-time displacement data and simulation data from the digital twin module, performs fault diagnosis, and outputs the diagnosis results. The digital twin module connects the communication gateway module and the status diagnosis module, constructs a virtual model of the valve and performs simulation, and outputs model correction parameters and performance predictions; The optimization decision-making module connects the status diagnosis module and the digital twin module, and generates control commands based on the diagnosis results and performance predictions; The closed-loop execution module connects the optimization decision module and the displacement sensing module. It drives the actuator according to the control command and feeds back the actual displacement data to the displacement sensing module.

[0006] Furthermore, the automated control system for pneumatic control valves according to the present invention further includes: The displacement sensing module acquires the original voltage signal of the valve core displacement through a magnetostrictive sensor, performs analog-to-digital conversion on the original voltage signal, processes the converted signal using a Kalman filter algorithm, and fuses the filtered output with the predicted value of the pneumatic actuator motion model to output a digital displacement quantity. The signal conditioning module uses fuzzy logic rules to fuse displacement digital quantity and pressure data, inputs the fusion result into a programmable gain amplifier for amplitude conditioning, performs analog-to-digital conversion on the conditioned signal, and encapsulates the converted data into standard industrial communication protocol data frames. The intelligent conversion module uses a support vector machine classification algorithm to extract trajectory features from displacement information and perform pattern recognition to generate a valve status identifier, which is then embedded into the IO-Link digital signal. The communication gateway module encapsulates the IO-Link digital signal embedded with the valve status identifier using the OPC UA protocol, adds a timestamp and a cyclic redundancy check code, and transmits the encapsulated data to the status diagnosis module and the digital twin module. The state diagnosis module constructs a residual sequence analysis queue, applies a hidden Markov model to perform probabilistic modeling of valve behavior sequences, calculates the probability of abnormal states through forward and backward algorithms, and outputs a fault warning index and a health status assessment. The digital twin module uses multi-domain physical modeling technology to construct a virtual image of the valve, receives real-time displacement data to drive computational fluid dynamics simulation, predicts the force distribution and flow characteristics of the valve core, compares the simulation results with the measured data for consistency, and outputs model correction parameters and performance degradation trend prediction. The optimization decision module uses the particle swarm optimization algorithm to search for the optimal combination of control parameters under constraints, and adopts the model predictive control strategy to continuously optimize the valve position setpoint sequence and output control commands. The closed-loop execution module converts control commands into pneumatic control signals, drives the actuator to adjust the valve core opening, and collects actual displacement data to feed back to the displacement sensing module.

[0007] Furthermore, the automated control system for pneumatic control valves according to the present invention further includes: The Kalman filter algorithm of the displacement sensing module establishes a state-space model with valve core displacement and velocity as state variables. Based on the motion model, the prior estimate is calculated and then fused with the current sensor measurement value according to the Kalman gain. The process noise covariance matrix of the state-space model is compensated in real time based on the temperature sensor reading and the historical wear database.

[0008] Furthermore, the automated control system for pneumatic control valves according to the present invention further includes: The fuzzy logic processing of the signal conditioning module defines displacement deviation and pressure change rate as input fuzzy variables, establishes a rule base, which includes rules that increase the weight of the displacement signal if the displacement deviation is large and the pressure is higher than the preset value, performs defuzzification processing on the rule base output, calculates the coordinate value of the center point of the area enclosed by the membership function curve, uses the coordinate value as PID control parameters, and uses the PID control parameters to dynamically adjust the signal conditioning characteristics.

[0009] Furthermore, the automated control system for pneumatic control valves according to the present invention further includes: The intelligent conversion module's support vector machine classification extracts three statistical features—mean, variance, and kurtosis—from the displacement trajectory to form a feature vector. This feature vector is then mapped to a high-dimensional feature space using a kernel function. In this high-dimensional feature space, the optimal classification hyperplane is found using the principle of minimizing structural risk. Based on the optimal classification hyperplane, the classification confidence of the valve state is output.

[0010] Furthermore, the automated control system for pneumatic control valves according to the present invention further includes: The hidden Markov model analysis of the condition diagnosis module defines the valve as normal, slightly worn, and severely stuck as implicit states. The displacement residual sequence is used as the observation sequence. The state transition probability matrix is ​​learned based on the observation sequence by the Baum-Welch algorithm. The Viterbi algorithm is used to decode the state sequence using the state transition probability matrix. The forward probability is calculated based on the state sequence to obtain the fault warning index.

[0011] Furthermore, the automated control system for pneumatic control valves according to the present invention further includes: The digital twin module uses the Modelica language to establish a coupled model of mechanical structure, aerodynamics, and fluid characteristics. The coupled model is then input into the DASSL solver for numerical integration. The friction coefficient and flow coefficient in the coupled model are calibrated online using actual displacement data based on the numerical integration results.

[0012] Furthermore, the automated control system for pneumatic control valves according to the present invention further includes: The optimization decision module uses control parameters as particle positions and system performance indicators as fitness functions. Based on the fitness function, it guides particle swarm search through individual extreme values ​​and global extreme values. The particle swarm search results are combined with the rolling optimization mechanism of model predictive control to solve the constrained quadratic programming problem in the prediction time domain.

[0013] Furthermore, the automated control system for pneumatic control valves according to the present invention further includes: The closed-loop execution module uses pulse width modulation technology to convert digital setpoints into pulse signals with adjustable duty cycles. The pulse signals are then converted into proportional pneumatic pressure through an electro-pneumatic converter. The proportional pneumatic pressure is used to drive the actuator to position itself, while monitoring air source pressure fluctuations and providing feedforward compensation for the fluctuations. The model correction parameters output by the digital twin module are fed back to the displacement sensing module to adjust the Kalman filter parameters. The health assessment results output by the state diagnosis module are synchronized to the intelligent conversion module to update the classification model. The predicted trajectory output by the optimization decision module is fed forward to the closed-loop execution module for compensation control.

[0014] Secondly, the present invention provides an automated control method for a pneumatic control valve, applied to an automated control system for the pneumatic control valve as described above, comprising: Step 1: Acquire the original voltage signal of valve core displacement using a magnetostrictive sensor, perform analog-to-digital conversion on the original voltage signal, process the converted signal using a Kalman filter algorithm, fuse the filtered output with the predicted value of the pneumatic actuator motion model, and output the digital displacement value. Step 2: Receive the digital displacement signal and the pressure data collected by the air source pressure sensor. Use fuzzy logic rules to fuse the digital displacement signal and the pressure data. Input the fusion result into a programmable gain amplifier for amplitude conditioning. Perform analog-to-digital conversion on the conditioned signal. Encapsulate the converted data into a standard industrial communication protocol data frame. Step 3: Analyze the displacement and pressure information in the data frame, apply the support vector machine classification algorithm to extract trajectory features from the displacement information and perform pattern recognition to generate valve status identifiers, and embed the valve status identifiers into the IO-Link digital signal; Step 4: Encapsulate the IO-Link digital signal with embedded valve status identifier using the OPC UA protocol, add a timestamp and cyclic redundancy check code, and transmit the encapsulated data to the status diagnosis and digital twin process. Step 5: Receive real-time displacement data and simulation data of the digital twin process, construct a residual sequence analysis queue, apply a hidden Markov model to perform probabilistic modeling of the valve behavior sequence, calculate the probability of abnormal state through forward and backward algorithms, and output the fault warning index and health status assessment. Step 6: Construct a virtual image of the valve using multi-domain physical modeling technology, receive real-time displacement data to drive computational fluid dynamics simulation, predict the force distribution and flow characteristics of the valve core, compare the consistency of the simulation results with the measured data, and output model correction parameters and performance degradation trend prediction. Step 7: Receive fault warning index, health status assessment and performance prediction data, apply particle swarm optimization algorithm to search for the optimal combination of control parameters under constraints, use model predictive control strategy to continuously optimize the valve position setpoint sequence, and output control commands. Step 8: Receive control commands, convert the control commands into pneumatic control signals, drive the actuator to adjust the valve core opening, and collect actual displacement data to feed back to Step 1.

[0015] Beneficial effects of this invention; This invention utilizes a displacement sensing module with a magnetostrictive sensor to directly acquire the raw voltage signal of the valve core displacement. After analog-to-digital conversion and Kalman filtering, the signal is fused with the predicted value from the pneumatic actuator motion model to output a digital displacement value. This solves the problem of existing solenoid valves relying on current on / off signals to infer valve position and lacking direct feedback. A signal conditioning module receives the digital displacement and pressure data, performs fuzzy logic rules for fusion processing, and conditions the signal amplitude using a programmable gain amplifier to output a standard industrial communication protocol data frame, enhancing data adaptability and robustness. An intelligent conversion module parses the displacement and pressure information in the data frame, applies a support vector machine classification algorithm to extract trajectory features and perform pattern recognition, generates a valve status identifier, and embeds it into the IO-Link digital signal to achieve intelligent classification of the valve status. A communication gateway module performs OPC on the digital signal. The UA protocol is encapsulated, with timestamps and cyclic redundancy check codes added, and transmitted to the status diagnosis module and digital twin module to ensure data integrity and synchronization. The status diagnosis module constructs a residual sequence analysis queue, applies a hidden Markov model to probabilistically model the valve behavior sequence, calculates the probability of abnormal states through forward and backward algorithms, and outputs a fault warning index and health status assessment to achieve real-time fault diagnosis. The digital twin module uses multi-domain physical modeling technology to construct a virtual image of the valve, drives computational fluid dynamics simulation, predicts the force distribution and flow characteristics of the valve core, and outputs model correction parameters and performance degradation trend predictions by comparing with actual data, providing a basis for virtual verification and optimization. The optimization decision-making module integrates fault warning index, health status assessment, and performance prediction data, applies particle swarm optimization algorithm to search for optimal control parameters, and uses model predictive control strategy to continuously optimize the valve position setpoint sequence, outputting control commands to achieve intelligent decision-making. The closed-loop execution module converts the control commands into pneumatic control signals, drives the actuator to adjust the valve core opening, and collects actual displacement data to feed back to the displacement sensing module, forming a closed-loop control loop. Through the data flow and feedback mechanism between modules, the system achieves real-time monitoring, precise control, fault diagnosis, and predictive maintenance of valve position, improving system reliability and control accuracy, thereby solving the technical problems of closed-loop monitoring and precise control. Attached Figure Description

[0016] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on the drawings without creative effort.

[0017] Figure 1 This is a flowchart of the automated control method for a pneumatic control valve according to the present invention. Detailed Implementation

[0018] To make the technical solution of the present invention clearer, the present invention will be clearly and completely described below with reference to specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. The present invention provided by various embodiments will be described in detail below with reference to the accompanying drawings. To better understand the purpose of the present invention, the present invention will be described in further detail below.

[0019] In a first aspect, the present invention provides an automated control system for a pneumatic control valve, comprising: The displacement sensing module is configured to acquire valve core displacement signals and process them to obtain digital displacement values. The signal conditioning module connects to the displacement sensing module, receives digital displacement and pressure data, performs fusion processing, and outputs standard industrial communication protocol data frames. The intelligent conversion module, connected to the signal conditioning module, extracts displacement and pressure information from the data frame and performs status recognition to generate a digital signal with status identifier; The communication gateway module connects to the intelligent conversion module, encapsulates digital signals according to protocols, and transmits them to the status diagnosis module and the digital twin module; The status diagnosis module connects to the communication gateway module, receives real-time displacement data and simulation data from the digital twin module, performs fault diagnosis, and outputs the diagnosis results. The digital twin module connects the communication gateway module and the status diagnosis module, constructs a virtual model of the valve and performs simulation, and outputs model correction parameters and performance predictions; The optimization decision-making module connects the status diagnosis module and the digital twin module, and generates control commands based on the diagnosis results and performance predictions; The closed-loop execution module connects the optimization decision module and the displacement sensing module. It drives the actuator according to the control command and feeds back the actual displacement data to the displacement sensing module.

[0020] The displacement sensing module acquires the original voltage signal of the valve core displacement through a magnetostrictive sensor, performs analog-to-digital conversion on the original voltage signal, processes the converted signal using a Kalman filter algorithm, and fuses the filtered output with the predicted value of the pneumatic actuator motion model to output a digital displacement quantity. The signal conditioning module uses fuzzy logic rules to fuse displacement digital quantity and pressure data, inputs the fusion result into a programmable gain amplifier for amplitude conditioning, performs analog-to-digital conversion on the conditioned signal, and encapsulates the converted data into standard industrial communication protocol data frames. The intelligent conversion module uses a support vector machine classification algorithm to extract trajectory features from displacement information and perform pattern recognition to generate a valve status identifier, which is then embedded into the IO-Link digital signal. The communication gateway module encapsulates the IO-Link digital signal embedded with the valve status identifier using the OPC UA protocol, adds a timestamp and a cyclic redundancy check code, and transmits the encapsulated data to the status diagnosis module and the digital twin module. The state diagnosis module constructs a residual sequence analysis queue, applies a hidden Markov model to perform probabilistic modeling of valve behavior sequences, calculates the probability of abnormal states through forward and backward algorithms, and outputs a fault warning index and a health status assessment. The digital twin module uses multi-domain physical modeling technology to construct a virtual image of the valve, receives real-time displacement data to drive computational fluid dynamics simulation, predicts the force distribution and flow characteristics of the valve core, compares the simulation results with the measured data for consistency, and outputs model correction parameters and performance degradation trend prediction. The optimization decision module uses the particle swarm optimization algorithm to search for the optimal combination of control parameters under constraints, and adopts the model predictive control strategy to continuously optimize the valve position setpoint sequence and output control commands. The closed-loop execution module converts control commands into pneumatic control signals, drives the actuator to adjust the valve core opening, and collects actual displacement data to feed back to the displacement sensing module.

[0021] The displacement sensing module acquires the raw voltage signal of the valve spool displacement through a magnetostrictive sensor, performs analog-to-digital conversion on the raw voltage signal, and then applies a Kalman filter algorithm to process the converted signal. The Kalman filter algorithm fuses the filtered output with the predicted value of the pneumatic actuator motion model to output a high-precision digital displacement value. This process eliminates sensor noise and interference, providing reliable valve spool position feedback.

[0022] The signal conditioning module connects to the displacement sensing module, receiving digital displacement data and integrating pressure data collected by a gas source pressure sensor. The signal conditioning module uses fuzzy logic rules to fuse the digital displacement and pressure data, and inputs the fusion result into a programmable gain amplifier for amplitude conditioning. The conditioned signal undergoes analog-to-digital conversion, and the converted data is encapsulated into standard industrial communication protocol data frames. This fusion processing enhances the robustness of the data and adapts to changes in different operating conditions.

[0023] The intelligent conversion module connects to the signal conditioning module and parses the displacement and pressure information in the data frame. The intelligent conversion module applies a support vector machine classification algorithm to extract trajectory features from the displacement information and perform pattern recognition to generate a valve status identifier. This valve status identifier is embedded in the IO-Link digital signal for easy subsequent transmission and processing. This step achieves intelligent classification and identification of the valve status.

[0024] The communication gateway module connects to the intelligent conversion module and encapsulates the IO-Link digital signal embedded with valve status identifiers using the OPC UA protocol. The communication gateway module adds timestamps and cyclic redundancy check codes to ensure data transmission integrity and timing accuracy. The encapsulated data is then transmitted to the status diagnosis module and the digital twin module, achieving efficient data distribution.

[0025] The condition diagnosis module connects to the communication gateway module, receiving real-time displacement data and simulation data output from the digital twin module. The condition diagnosis module constructs a residual sequence analysis queue and applies a hidden Markov model to probabilistically model the valve behavior sequence. It calculates the probability of abnormal states using forward and backward algorithms, outputting a fault warning index and a health status assessment. This process enables real-time fault diagnosis and health monitoring of the valve.

[0026] The digital twin module connects the communication gateway module and the condition diagnosis module, employing multi-domain physical modeling technology to construct a virtual image of the valve. The digital twin module receives real-time displacement data to drive computational fluid dynamics simulation, predicting the force distribution and flow characteristics of the valve core. The simulation results are compared with measured data for consistency, outputting model correction parameters and performance degradation trend predictions. The digital twin module provides a virtual representation of the valve, supporting performance prediction and model optimization.

[0027] The optimization decision-making module connects the status diagnosis module and the digital twin module, receiving fault warning index, health status assessment, and performance prediction data. It applies particle swarm optimization to search for the optimal combination of control parameters under constraints and uses model predictive control to continuously optimize the valve position setpoint sequence. Finally, the optimization decision-making module outputs control commands to achieve intelligent valve control.

[0028] The closed-loop execution module connects the optimization decision module and the displacement sensing module, receiving control commands. The closed-loop execution module converts these commands into pneumatic control signals, driving the actuator to adjust the valve core opening. Simultaneously, it collects actual displacement data and feeds it back to the displacement sensing module, forming a closed-loop control circuit. This process enables precise control and real-time adjustment of the valve position.

[0029] The system achieves automated control of pneumatic control valves through the collaborative work of its modules. The displacement sensing module provides initial displacement data, which, after signal conditioning and intelligent conversion, is distributed via a communication gateway. The status diagnosis module and digital twin module process the data in parallel, providing diagnostic and predictive information. The optimization decision-making module generates control commands based on this information, and the closed-loop execution module executes the commands and feeds back the data. This architecture solves the technical challenges of closed-loop monitoring and precise control of valve position.

[0030] The Kalman filtering algorithm of the displacement sensing module of this invention establishes a state-space model with valve core displacement and velocity as state variables. Based on the motion model, it calculates prior estimates and then fuses these prior estimates with the current sensor measurements using Kalman gain weighting. The process noise covariance matrix of the state-space model is compensated in real-time based on temperature sensor readings and a historical wear database. The state-space model describes the valve's dynamic behavior by defining displacement and velocity variables, the motion model predicts the valve core's trajectory based on physical laws, the prior estimates provide predicted values, and the Kalman gain weighted fusion combines the predicted values ​​with sensor measurements to reduce errors. The process noise covariance matrix is ​​dynamically adjusted based on temperature changes and wear data to improve filtering accuracy and adaptability.

[0031] The fuzzy logic processing of the signal conditioning module in this invention defines displacement deviation and pressure change rate as input fuzzy variables, establishes a rule base, and includes rules that increase the weight of the displacement signal if the displacement deviation is large and the pressure is higher than a preset value. The output of the rule base is defuzzified, and the coordinates of the center point of the area enclosed by the membership function curve are calculated. These coordinates are used as PID control parameters to dynamically adjust the signal conditioning characteristics. Fuzzy logic processing quantifies input uncertainty through fuzzy variables, the rule base defines control strategies based on expert experience, defuzzification converts the fuzzy output into precise values, the coordinates of the area center point represent the optimal control point, and the PID control parameters are used to adjust the gain and response characteristics of the signal conditioning module to achieve adaptive signal optimization.

[0032] The intelligent conversion module of this invention uses support vector machine (SVM) classification to extract three statistical features—mean, variance, and kurtosis—from the displacement trajectory to form feature vectors. These feature vectors are then mapped to a high-dimensional feature space using a kernel function. In this high-dimensional feature space, the optimal classification hyperplane is found using the principle of structural risk minimization. Based on this optimal classification hyperplane, the classification confidence score of the valve state is output. The feature extraction process captures key patterns from the displacement data; the kernel function mapping solves the linear inseparability problem; structural risk minimization balances model complexity and generalization ability; the optimal classification hyperplane provides the decision boundary; and the classification confidence score reflects the reliability of state identification, thus improving the accuracy of valve state monitoring.

[0033] The Hidden Markov Model (HMM) analysis of the condition diagnosis module in this invention defines the valve's normal, slightly worn, and severely jammed states as hidden states. The displacement residual sequence is used as the observation sequence. The Baum-Welch algorithm learns the state transition probability matrix based on the observation sequence, and the Viterbi algorithm decodes the optimal state sequence using the state transition probability matrix. A fault warning index is obtained by calculating the forward probability based on the state sequence. Hidden states represent the invisible internal states of the valve, while the observation sequence provides visible data. The Baum-Welch algorithm estimates model parameters through iterative optimization, the Viterbi algorithm finds the optimal state path, the forward probability quantifies the state characteristics, and the fault warning index enables early fault detection and risk assessment.

[0034] This invention's digital twin module uses the Modelica language to establish a coupled model of mechanical structure, aerodynamics, and fluid characteristics. The coupled model is input into the DASSL solver for numerical integration. The numerical integration results are then used to calibrate the friction and flow coefficients in the coupled model online using actual displacement data. The Modelica language supports multi-domain physical modeling, the coupled model integrates mechanical and fluid behaviors, the DASSL solver handles differential-algebraic equations, numerical integration provides simulation results, and online calibration adjusts model parameters using actual data, maintaining consistency between the digital twin model and the physical valve and enhancing prediction accuracy.

[0035] This invention's optimization decision module uses control parameters as particle positions and system performance indicators as fitness functions. Based on the fitness function, it guides particle swarm optimization (PSO) through individual and global extrema. The PSO search results are then combined with the rolling optimization mechanism of model predictive control (MMC) to solve a constrained quadratic programming problem in the prediction time domain. Particle positions represent control variables, the fitness function evaluates control effectiveness, PSO simulates swarm intelligence, MMC uses rolling optimization to handle dynamic constraints, and the quadratic programming problem solves for the optimal control sequence, achieving precise and robust valve position control.

[0036] This invention's closed-loop execution module employs pulse width modulation (PWM) technology to convert digital setpoints into pulse signals with adjustable duty cycles. These pulse signals are then used by an electro-pneumatic converter to generate proportional pneumatic pressure, which drives the actuator for positioning. Simultaneously, it monitors pneumatic pressure fluctuations and performs feedforward compensation. Model correction parameters output from the digital twin module are fed back to the displacement sensing module to adjust the Kalman filter parameters. Health assessment results from the state diagnosis module are synchronized to the intelligent conversion module to update the classification model. The predicted trajectory output from the optimization decision module is fed forward to the closed-loop execution module for compensatory control. PWM enables digital-to-analog conversion, the electro-pneumatic converter generates pneumatic signals, feedforward compensation counteracts pressure interference, and the feedback mechanism achieves system self-adaptation through parameter adjustment and model updates. The predicted trajectory feedforward improves control response speed, forming a multi-level closed-loop optimization.

[0037] Secondly, please refer to Figure 1 This invention provides an automated control method for a pneumatic control valve, applied to an automated control system for the pneumatic control valve as described above, comprising: Step 1: Acquire the original voltage signal of valve core displacement using a magnetostrictive sensor, perform analog-to-digital conversion on the original voltage signal, process the converted signal using a Kalman filter algorithm, fuse the filtered output with the predicted value of the pneumatic actuator motion model, and output the digital displacement value. Step 2: Receive the digital displacement signal and the pressure data collected by the air source pressure sensor. Use fuzzy logic rules to fuse the digital displacement signal and the pressure data. Input the fusion result into a programmable gain amplifier for amplitude conditioning. Perform analog-to-digital conversion on the conditioned signal. Encapsulate the converted data into a standard industrial communication protocol data frame. Step 3: Analyze the displacement and pressure information in the data frame, apply the support vector machine classification algorithm to extract trajectory features from the displacement information and perform pattern recognition to generate valve status identifiers, and embed the valve status identifiers into the IO-Link digital signal; Step 4: Encapsulate the IO-Link digital signal with embedded valve status identifier using the OPC UA protocol, add a timestamp and cyclic redundancy check code, and transmit the encapsulated data to the status diagnosis and digital twin process. Step 5: Receive real-time displacement data and simulation data of the digital twin process, construct a residual sequence analysis queue, apply a hidden Markov model to perform probabilistic modeling of the valve behavior sequence, calculate the probability of abnormal state through forward and backward algorithms, and output the fault warning index and health status assessment. Step 6: Construct a virtual image of the valve using multi-domain physical modeling technology, receive real-time displacement data to drive computational fluid dynamics simulation, predict the force distribution and flow characteristics of the valve core, compare the consistency of the simulation results with the measured data, and output model correction parameters and performance degradation trend prediction. Step 7: Receive fault warning index, health status assessment and performance prediction data, apply particle swarm optimization algorithm to search for the optimal combination of control parameters under constraints, use model predictive control strategy to continuously optimize the valve position setpoint sequence, and output control commands. Step 8: Receive control commands, convert the control commands into pneumatic control signals, drive the actuator to adjust the valve core opening, and collect actual displacement data to feed back to Step 1.

[0038] This invention directly acquires the raw voltage signal of valve core displacement through a displacement sensing module, solving the problem of existing solenoid valves relying on current on / off signals to infer valve position and lacking direct feedback. The displacement sensing module uses a magnetostrictive sensor to obtain the actual position of the valve core, processes it through analog-to-digital conversion and Kalman filtering algorithms, and fuses the predicted values ​​from the pneumatic actuator motion model to output a high-precision digital displacement quantity, providing a reliable displacement feedback basis for the system. The signal conditioning module receives the digital displacement quantity and pressure data, performs fuzzy logic rule fusion processing, conditions the signal amplitude through a programmable gain amplifier, and outputs standard industrial communication protocol data frames to enhance data robustness. The intelligent conversion module parses the displacement and pressure information in the data frames, applies a support vector machine classification algorithm to extract trajectory features and perform pattern recognition, generates a valve status identifier, and embeds it into the IO-Link digital signal to achieve intelligent classification of valve status. The communication gateway module encapsulates the digital signal using the OPC UA protocol, adds a timestamp and checksum, and transmits it to the status diagnosis module and digital twin module to achieve data integrity and synchronization. The status diagnosis module constructs a residual sequence analysis queue, applies a hidden Markov model to probabilistically model the valve behavior sequence, calculates the probability of abnormal states through forward and backward algorithms, and outputs a fault warning index and health status assessment to achieve real-time fault diagnosis. The digital twin module uses multi-domain physical modeling technology to construct a virtual image of the valve, drives computational fluid dynamics simulation, predicts the valve core force and flow characteristics, and outputs model correction parameters and performance predictions by comparing with actual data, providing a basis for virtual verification and optimization. The optimization decision module integrates fault warning index, health status assessment, and performance prediction data, applies a particle swarm optimization algorithm to search for optimal control parameters, uses a model predictive control strategy to continuously optimize the valve position setpoint sequence, and outputs control commands to achieve intelligent decision-making. The closed-loop execution module converts control commands into pneumatic control signals, drives the actuator to adjust the valve core opening, and collects actual displacement data to feed back to the displacement sensing module, forming a closed-loop control loop. Through the data flow and feedback mechanism between modules, the system achieves real-time monitoring of valve position, fault diagnosis, predictive maintenance, and precise control, thus solving the technical problems of closed-loop monitoring and precise control.

[0039] In industrial automation scenarios, pneumatic control valves are commonly used for chemical material conveying or energy pipeline control. However, traditional solenoid valves lack direct displacement feedback, leading to inaccurate valve core position estimation and causing media leakage or process interruption. This invention achieves real-time monitoring and high-precision closed-loop control of valve position through an automated control system. During system implementation, the displacement sensing module uses a magnetostrictive sensor to directly acquire the raw voltage signal of the valve core displacement. After analog-to-digital conversion, the raw voltage signal is processed using a Kalman filter algorithm. The filtered output is fused with the predicted value of the pneumatic actuator motion model to output a digital displacement value. This process uses a state-space model with valve core displacement and velocity as state variables, calculates prior estimates based on the motion model, and weights and fuses them with sensor measurements. The process noise covariance matrix is ​​compensated in real time based on temperature sensor readings and a historical wear database to cope with environmental fluctuations and equipment aging.

[0040] The signal conditioning module receives digital displacement data and pressure data collected by the air source pressure sensor, and performs fusion processing using fuzzy logic rules. Fuzzy logic processing defines displacement deviation and pressure change rate as input fuzzy variables to establish a rule base. For example, if the displacement deviation is large and the pressure is higher than a preset value, the weight of the displacement signal is increased. The rule base output is defuzzified, and the coordinates of the center point of the area enclosed by the membership function curve are calculated. These center point coordinates are used as PID control parameters to dynamically adjust the signal conditioning characteristics. The fusion result is input to a programmable gain amplifier for amplitude conditioning, and then after analog-to-digital conversion, it is encapsulated into a standard industrial communication protocol data frame to enhance the robustness of the data under varying operating conditions.

[0041] The intelligent conversion module parses the displacement and pressure information in the data frame and uses the support vector machine classification algorithm to extract three statistical features—mean, variance, and kurtosis—from the displacement trajectory to form a feature vector. The feature vector is mapped to a high-dimensional feature space through a kernel function. The optimal classification hyperplane is found by using the principle of minimizing structural risk. Based on the hyperplane, the classification confidence of the valve status is output, a valve status identifier is generated and embedded into the IO-Link digital signal to realize intelligent identification of the valve status, such as normal, worn, or stuck.

[0042] The communication gateway module encapsulates the IO-Link digital signal with embedded status identifiers using the OPC UA protocol, adding timestamps and cyclic redundancy check codes to ensure data transmission integrity and synchronization. The encapsulated data is then distributed to the status diagnosis module and the digital twin module. The status diagnosis module constructs a residual sequence analysis queue, applying a hidden Markov model to define valve status as implicit states such as normal, slight wear, or severe jamming. Using the displacement residual sequence as the observation sequence, the state transition probability matrix is ​​learned through the Baum-Welch algorithm, and the Viterbi algorithm is used to decode the final state sequence. Based on the forward probability, a fault warning index and health status assessment are calculated to achieve early fault detection.

[0043] The digital twin module uses Modelica to build a coupled model of mechanical structure, aerodynamics, and fluid characteristics. The model is input into the DASSL solver for numerical integration. Real-time displacement data is used to drive computational fluid dynamics simulation to predict the force distribution and flow characteristics of the valve core. After comparing the simulation results with the measured data, the model correction parameters and performance degradation trend prediction are output. During online calibration, the friction coefficient and flow coefficient are adjusted according to the actual displacement data to maintain the consistency between the virtual model and the physical valve.

[0044] The optimization decision module receives fault warning index, health status assessment and performance prediction data, and applies particle swarm optimization algorithm with control parameters as particle positions and system performance index as fitness function. It searches for the optimal combination of control parameters by guiding individual extreme values ​​and global extreme values, and combines model predictive control strategy to continuously optimize the valve position setpoint sequence. It solves a constrained quadratic programming problem in the prediction time domain and outputs control commands.

[0045] The closed-loop execution module converts control commands into pulse signals with adjustable duty cycles using pulse width modulation (PWM) technology. These pulse signals are then converted by an electrical converter to generate proportional pneumatic pressure, which drives the actuator to adjust the valve opening. Simultaneously, it monitors air source pressure fluctuations and performs feedforward compensation. Actual displacement data is fed back to the displacement sensing module, forming a closed-loop control loop. Model correction parameters output from the digital twin module are fed back to the displacement sensing module to adjust the Kalman filter parameters. The health assessment results from the state diagnosis module are synchronized to the intelligent conversion module to update the classification model. The predicted trajectory from the optimization decision module is fed forward to the closed-loop execution module for compensatory control, achieving system adaptive optimization.

[0046] In the implementation of this invention, the system achieves valve position control and predictive maintenance in industrial sites such as chemical processes through data flow and feedback between modules. However, the control accuracy is affected by the accuracy of the sensors or the degree of model matching, and periodic calibration is required to maintain performance.

[0047] Embodiment 1 of the present invention: In chemical production processes, pneumatic control valves are commonly used to regulate the feed flow rate of reactors. In this system, the displacement sensing module directly acquires the raw voltage signal of the valve core displacement via a magnetostrictive sensor. This signal is converted into a digital quantity by a 24-bit high-precision analog-to-digital converter, and then fused with the predicted value from the pneumatic actuator motion model using a Kalman filter algorithm. For example, when the valve experiences slight wear due to long-term use, the process noise covariance matrix is ​​dynamically adjusted based on data from the built-in temperature sensor and a historical wear database to compensate for model errors and output a more stable digital displacement signal. The signal conditioning module receives the displacement data and a 0.7MPa pressure signal acquired by the air source pressure sensor, and fuses them using fuzzy logic rules: if the detected displacement deviation exceeds 5% of the full stroke and the pressure fluctuation is greater than 0.05MPa, the PID parameters calculated using the area center point of the membership function automatically increase the weight of the displacement signal. This signal is then conditioned by a programmable gain amplifier and encapsulated into a Profinet data frame. The intelligent conversion module extracts the mean, variance, and kurtosis features of the displacement trajectory from the data frame, maps them to a high-dimensional space for classification using a support vector machine kernel function, identifies the "valve core stuck" state, and embeds IO-Link signals. The communication gateway module adds millisecond-level timestamps and CRC checksums, and synchronizes the data to the status diagnosis module and digital twin module via the OPC UA protocol. The status diagnosis module uses a hidden Markov model to probabilistically model the residual sequence, decodes the "slight wear" state sequence using the Viterbi algorithm, and outputs a warning when the health index drops to 85%. The digital twin module uses a multi-domain model built with Modelica to simulate and display the uneven force on the valve core, and outputs friction coefficient calibration parameters after comparing it with the actual displacement data. The optimization decision module aims to minimize the flow control error, uses a particle swarm optimization algorithm to search for the optimal PID parameter combination, and combines it with model predictive control to generate valve opening commands. The closed-loop execution module outputs a pulse signal with a duty cycle of 45% through pulse width modulation, which drives the electrical converter to generate proportional pressure. At the same time, it performs feedforward compensation based on the air source pressure fluctuation, and finally controls the valve core positioning accuracy within ±0.1mm.

[0048] Embodiment 2 of the present invention: In natural gas pipeline scenarios, this system is used for leak prevention control of critical shut-off valves. The magnetostrictive sensor of the displacement sensing module is installed on the high-pressure ball valve actuator. The Kalman-filtered state-space model adjusts process noise parameters in real time based on pipeline pressure fluctuations (1.2MPa±0.1MPa) to eliminate vibration interference. The signal conditioning module processes displacement and pressure data using a fuzzy rule base: when a sudden pressure drop and displacement response delay are detected, the displacement signal gain is automatically enhanced, and a standard Modbus TCP data frame is output. The intelligent conversion module extracts the statistical features of the displacement trajectory, classifies and identifies the "valve not in position" state using a support vector machine, and embeds the state identifier into the IO-Link signal. When the communication gateway transmits data using the OPC UA protocol, it simultaneously attaches the valve number and safety level label. The state diagnosis module analyzes the residual sequence using a hidden Markov model, combines it with fluid dynamics simulation data (such as valve core force distribution) provided by the digital twin module, and uses the Baum-Welch algorithm to update the state transition probability matrix, providing a 48-hour advance warning of "sealing surface wear" risk. The digital twin module performs numerical integration on the coupled model using the DASSL solver, calibrates the flow coefficient online based on actual displacement data, and predicts valve performance degradation trends. The optimization decision module uses minimizing sealing leakage as a constraint, employs a particle swarm optimization algorithm to optimize control parameters, and uses model predictive control to continuously adjust the valve action sequence. The closed-loop execution module converts control commands into 4-20mA pneumatic signals to drive the cylinder to position the valve core, while simultaneously using feedforward compensation to offset pipeline pressure fluctuations. The feedback of actual displacement data is used to update the Kalman filter parameters of the displacement sensing module. Even under conditions of sudden pressure changes, the system maintains valve position repeatability error within 0.5%. However, if there is mechanical clearance at the sensor installation location, periodic zero-point calibration is required to maintain accuracy.

[0049] The Hidden Markov Model (HMM) is constructed in the state diagnosis module by defining three implicit states of the valve: normal, slight wear, and severe jamming. The input data is the displacement residual sequence provided by the displacement sensing module and the digital twin module, which serves as the observation sequence for probabilistic modeling. The model uses the Baum-Welch algorithm to learn the state transition probability matrix based on the observation sequence, and decodes the most probable state sequence using the Viterbi algorithm. Then, it combines forward and backward algorithms to calculate the probability of abnormal states. Finally, it outputs a fault warning index and health status assessment data, providing fault risk indicators for optimization decisions.

[0050] The state-space model is constructed in the Kalman filtering algorithm of the displacement sensing module. The state-space equation is established with valve core displacement and velocity as state variables. The input data includes the measured value of the original voltage signal collected by the magnetostrictive sensor after analog-to-digital conversion and the predicted value of the pneumatic actuator motion model. The model calculates the prior estimate based on the motion model and uses Kalman gain to weight and fuse the prior estimate with the current sensor measurement value. The process noise covariance matrix is ​​compensated in real time based on the temperature sensor reading and the historical wear database. The output is a high-precision digital displacement quantity, which is used for subsequent signal conditioning and state recognition.

[0051] The PID control model is indirectly constructed in the signal conditioning module through fuzzy logic processing. The input data are fuzzy variables such as displacement deviation and pressure change rate. After processing by the fuzzy rule base and defuzzification calculation, the coordinates of the center point of the area of ​​the membership function curve are obtained. These coordinates are converted into PID control parameters, which are used to dynamically adjust the signal conditioning characteristics of the programmable gain amplifier. The model output is the optimized PID parameters, which adjust the signal gain and response in real time and enhance the robustness of data fusion.

[0052] The coupled model is built in the digital twin module using the Modelica language, integrating the mechanical structure, aerodynamics, and fluid characteristics of the valve, among other physical relationships. The input data is the actual displacement data collected by real-time displacement sensors. The model is numerically integrated using the DASSL solver to drive computational fluid dynamics simulation to predict the force distribution and flow characteristics of the valve core, and the simulation results are compared with the measured data for consistency. The output includes model correction parameters such as friction coefficient and flow coefficient, as well as performance degradation trend predictions, which are used for virtual verification and system optimization.

[0053] The particle swarm optimization model is constructed in the optimization decision module. It uses control parameters as particle positions and system performance indicators as fitness functions. The input data includes the fault warning index output by the state diagnosis module, the health status assessment, and the performance prediction data provided by the digital twin module. The model guides the particle swarm to search for the optimal combination of control parameters under constraints through individual extreme values ​​and global extreme values, and performs rolling optimization by combining the model predictive control strategy. The output is the optimal control parameter sequence, which is used to generate valve position setpoint commands.

[0054] The model predictive control model is constructed in the optimization decision module and adopts a rolling optimization mechanism. The input data are the optimal control parameters obtained by the particle swarm optimization model and the current state of the system. The model solves a constrained quadratic programming problem in the prediction time domain and dynamically adjusts the valve position setpoint sequence to cope with system fluctuations. The output is a real-time control command to ensure that the valve position accurately tracks the target value and improves the closed-loop control performance.

[0055] Kalman filtering is an optimization algorithm used to estimate system state. It reduces noise interference by combining predicted values ​​from the system model and measured values ​​from sensors, and dynamically adjusting weights using Kalman gain. In this invention, the algorithm is applied to the displacement sensing module to process the raw voltage signal of valve core displacement acquired by the magnetostrictive sensor. The algorithm first performs motion prediction using a state-space model with valve core displacement and velocity as state variables. Then, it fuses the predicted values ​​with the sensor data after analog-to-digital conversion. Simultaneously, it adjusts the process noise covariance matrix in real time based on temperature sensor readings and a historical wear database, ultimately outputting a high-precision digital displacement value to provide reliable valve core position feedback for the system.

[0056] Fuzzy logic is an intelligent control method for handling uncertainties and nonlinear problems. It transforms empirical knowledge into executable logical judgments by defining fuzzy variables, constructing a rule base, and performing defuzzification calculations. In this invention, the algorithm is implemented by a signal conditioning module, which fuses digital displacement data with data collected from pressure sensors. The algorithm uses displacement deviation and pressure change rate as input fuzzy variables, and infers using a preset rule base (e.g., increasing displacement weight if displacement deviation is large and pressure is high). It then calculates the area center point of the output membership function, converting the result into signal conditioning characteristics for dynamically adjusting the programmable gain amplifier as PID control parameters. This enhances the adaptability and robustness of the data under fluctuating operating conditions.

[0057] Support Vector Machines (SVMs) are classification algorithms based on the principle of minimizing structural risk. They map data to a high-dimensional space using kernel functions and find the optimal classification hyperplane to achieve accurate pattern recognition. In this invention, the algorithm is executed by an intelligent conversion module for valve status identification. The algorithm extracts three statistical features—mean, variance, and kurtosis—from the displacement trajectory to form a feature vector. After mapping this vector to a high-dimensional feature space using a kernel function, the SVM finds the optimal classification hyperplane, thereby classifying the valve's normal, worn, or stuck states and outputting the classification confidence score. Finally, the status identifier is embedded into the IO-Link digital signal to achieve intelligent judgment and identification of the valve's operating status.

[0058] Particle swarm optimization (PSO) is an optimization algorithm that simulates the behavior of swarm intelligence. It guides a swarm of particles to search for optimal solutions in the solution space through individual and global extrema, making it suitable for optimization problems under multi-parameter constraints. In this invention, the algorithm is implemented by an optimization decision module used to generate control commands. The algorithm uses valve control parameters as particle positions and system performance indicators as fitness functions. It finds the optimal parameter combination through iterative search and combines a rolling optimization mechanism of model predictive control to solve a constrained quadratic programming problem in the prediction time domain. Finally, it outputs a sequence of control commands, dynamically optimizing the valve position setpoint to improve control accuracy and response speed.

[0059] Model predictive control (MMC) is an advanced control strategy based on rolling optimization and feedback correction. It dynamically adjusts control commands to address system constraints and disturbances by solving an optimization problem online within a finite time domain. In this invention, the algorithm works in conjunction with particle swarm optimization (PSO) in the optimization decision module. Based on the optimal parameter combination provided by PSO and the real-time system state, the algorithm continuously solves for the valve position setpoint sequence within the prediction time domain. It then uses quadratic programming to handle constraints such as actuator displacement and air source pressure fluctuations, ultimately outputting smooth control commands that conform to actual operating conditions, ensuring that the valve positioning process is both fast and stable.

Claims

1. An automated control system for a pneumatic control valve, characterized in that, include: The displacement sensing module is configured to acquire valve core displacement signals and process them to obtain digital displacement values. The signal conditioning module connects to the displacement sensing module, receives digital displacement and pressure data, performs fusion processing, and outputs standard industrial communication protocol data frames. The intelligent conversion module, connected to the signal conditioning module, extracts displacement and pressure information from the data frame and performs status recognition to generate a digital signal with status identifier; The communication gateway module connects to the intelligent conversion module, encapsulates digital signals according to protocols, and transmits them to the status diagnosis module and the digital twin module; The status diagnosis module connects to the communication gateway module, receives real-time displacement data and simulation data from the digital twin module, performs fault diagnosis, and outputs the diagnosis results. The digital twin module connects the communication gateway module and the status diagnosis module, constructs a virtual model of the valve and performs simulation, and outputs model correction parameters and performance predictions; The optimization decision-making module connects the status diagnosis module and the digital twin module, and generates control commands based on the diagnosis results and performance predictions; The closed-loop execution module connects the optimization decision module and the displacement sensing module. It drives the actuator according to the control command and feeds back the actual displacement data to the displacement sensing module.

2. The automated control system for a pneumatic control valve according to claim 1, characterized in that, Also includes: The displacement sensing module acquires the original voltage signal of the valve core displacement through a magnetostrictive sensor, performs analog-to-digital conversion on the original voltage signal, processes the converted signal using a Kalman filter algorithm, and fuses the filtered output with the predicted value of the pneumatic actuator motion model to output a digital displacement quantity. The signal conditioning module uses fuzzy logic rules to fuse displacement digital quantity and pressure data, inputs the fusion result into a programmable gain amplifier for amplitude conditioning, performs analog-to-digital conversion on the conditioned signal, and encapsulates the converted data into standard industrial communication protocol data frames. The intelligent conversion module uses a support vector machine classification algorithm to extract trajectory features from displacement information and perform pattern recognition to generate a valve status identifier, which is then embedded into the IO-Link digital signal. The communication gateway module encapsulates the IO-Link digital signal embedded with the valve status identifier using the OPC UA protocol, adds a timestamp and a cyclic redundancy check code, and transmits the encapsulated data to the status diagnosis module and the digital twin module. The state diagnosis module constructs a residual sequence analysis queue, applies a hidden Markov model to perform probabilistic modeling of valve behavior sequences, calculates the probability of abnormal states through forward and backward algorithms, and outputs a fault warning index and a health status assessment. The digital twin module uses multi-domain physical modeling technology to construct a virtual image of the valve, receives real-time displacement data to drive computational fluid dynamics simulation, predicts the force distribution and flow characteristics of the valve core, compares the simulation results with the measured data for consistency, and outputs model correction parameters and performance degradation trend prediction. The optimization decision module uses the particle swarm optimization algorithm to search for the optimal combination of control parameters under constraints, and adopts the model predictive control strategy to continuously optimize the valve position setpoint sequence and output control commands. The closed-loop execution module converts control commands into pneumatic control signals, drives the actuator to adjust the valve core opening, and collects actual displacement data to feed back to the displacement sensing module.

3. The automated control system for a pneumatic control valve according to claim 2, characterized in that, Also includes: The Kalman filter algorithm of the displacement sensing module establishes a state-space model with valve core displacement and velocity as state variables. Based on the motion model, the prior estimate is calculated and then fused with the current sensor measurement value according to the Kalman gain. The process noise covariance matrix of the state-space model is compensated in real time based on the temperature sensor reading and the historical wear database.

4. The automated control system for a pneumatic control valve according to claim 3, characterized in that, Also includes: The fuzzy logic processing of the signal conditioning module defines displacement deviation and pressure change rate as input fuzzy variables, establishes a rule base, which includes rules that increase the weight of the displacement signal if the displacement deviation is large and the pressure is higher than the preset value, performs defuzzification processing on the rule base output, calculates the coordinate value of the center point of the area enclosed by the membership function curve, uses the coordinate value as PID control parameters, and uses the PID control parameters to dynamically adjust the signal conditioning characteristics.

5. The automated control system for a pneumatic control valve according to claim 4, characterized in that, Also includes: The intelligent conversion module's support vector machine classification extracts three statistical features—mean, variance, and kurtosis—from the displacement trajectory to form a feature vector. This feature vector is then mapped to a high-dimensional feature space using a kernel function. In this high-dimensional feature space, the optimal classification hyperplane is found using the principle of minimizing structural risk. Based on the optimal classification hyperplane, the classification confidence of the valve state is output.

6. The automated control system for a pneumatic control valve according to claim 5, characterized in that, Also includes: The hidden Markov model analysis of the condition diagnosis module defines the valve as normal, slightly worn, and severely stuck as implicit states. The displacement residual sequence is used as the observation sequence. The state transition probability matrix is ​​learned based on the observation sequence by the Baum-Welch algorithm. The Viterbi algorithm is used to decode the state sequence using the state transition probability matrix. The forward probability is calculated based on the state sequence to obtain the fault warning index.

7. The automated control system for a pneumatic control valve according to claim 6, characterized in that, Also includes: The digital twin module uses the Modelica language to establish a coupled model of mechanical structure, aerodynamics, and fluid characteristics. The coupled model is then input into the DASSL solver for numerical integration. The friction coefficient and flow coefficient in the coupled model are calibrated online using actual displacement data based on the numerical integration results.

8. The automated control system for a pneumatic control valve according to claim 7, characterized in that, Also includes: The optimization decision module uses control parameters as particle positions and system performance indicators as fitness functions. Based on the fitness function, it guides particle swarm search through individual extreme values ​​and global extreme values. The particle swarm search results are combined with the rolling optimization mechanism of model predictive control to solve the constrained quadratic programming problem in the prediction time domain.

9. The automated control system for a pneumatic control valve according to claim 8, characterized in that, Also includes: The closed-loop execution module uses pulse width modulation technology to convert digital setpoints into pulse signals with adjustable duty cycles. The pulse signals are then converted into proportional pneumatic pressure through an electro-pneumatic converter. The proportional pneumatic pressure is used to drive the actuator to position itself, while monitoring air source pressure fluctuations and providing feedforward compensation for the fluctuations. The model correction parameters output by the digital twin module are fed back to the displacement sensing module to adjust the Kalman filter parameters. The health assessment results output by the state diagnosis module are synchronized to the intelligent conversion module to update the classification model. The predicted trajectory output by the optimization decision module is fed forward to the closed-loop execution module for compensation control.

10. An automated control method for a pneumatic control valve, applied to an automated control system for the pneumatic control valve as described in any one of claims 1 to 9, characterized in that, include: Step 1: Acquire the original voltage signal of valve core displacement using a magnetostrictive sensor, perform analog-to-digital conversion on the original voltage signal, process the converted signal using a Kalman filter algorithm, fuse the filtered output with the predicted value of the pneumatic actuator motion model, and output the digital displacement value. Step 2: Receive the digital displacement signal and the pressure data collected by the air source pressure sensor. Use fuzzy logic rules to fuse the digital displacement signal and the pressure data. Input the fusion result into a programmable gain amplifier for amplitude conditioning. Perform analog-to-digital conversion on the conditioned signal. Encapsulate the converted data into a standard industrial communication protocol data frame. Step 3: Analyze the displacement and pressure information in the data frame, apply the support vector machine classification algorithm to extract trajectory features from the displacement information and perform pattern recognition to generate valve status identifiers, and embed the valve status identifiers into the IO-Link digital signal; Step 4: Encapsulate the IO-Link digital signal with embedded valve status identifier using the OPC UA protocol, add a timestamp and cyclic redundancy check code, and transmit the encapsulated data to the status diagnosis and digital twin process. Step 5: Receive real-time displacement data and simulation data of the digital twin process, construct a residual sequence analysis queue, apply a hidden Markov model to perform probabilistic modeling of the valve behavior sequence, calculate the probability of abnormal state through forward and backward algorithms, and output the fault warning index and health status assessment. Step 6: Construct a virtual image of the valve using multi-domain physical modeling technology, receive real-time displacement data to drive computational fluid dynamics simulation, predict the force distribution and flow characteristics of the valve core, compare the consistency of the simulation results with the measured data, and output model correction parameters and performance degradation trend prediction. Step 7: Receive fault warning index, health status assessment and performance prediction data, apply particle swarm optimization algorithm to search for the optimal combination of control parameters under constraints, use model predictive control strategy to continuously optimize the valve position setpoint sequence, and output control commands. Step 8: Receive control commands, convert the control commands into pneumatic control signals, drive the actuator to adjust the valve core opening, and collect actual displacement data to feed back to Step 1.

Images