Research On Health Assessment Of Control Valve Based On Hybrid Model And Feature Transfer

As one of the essential terminal actuators in the automatic control of the industrial production process,the pressure,flow rate,temperature,and liquid level can automatically be controlled by the pneumatic control valve.The control valve’s smooth operation can even affect the factory’s optimization control,production quality,and process safety.Therefore,the pneumatic control valve’s health management is enormously significant to the factory’s efficient operation.In this paper,a health assessment system of the control valve is designed based on the national standard of state monitoring and health management.The specific research contents are as follows:Firstly,the pneumatic control valve’s theoretical process model is established based on the series mixed model.The pneumatic control valve mechanism establishes its flow model,unbalanced force model,and actuator model.According to the least square support vector machine,the predictive regulator model is established,and the penalty coefficient and kernel function parameters are optimized by particle swarm optimization.The data model is applied to identify the simplified model’s internal parameters,such as flow characteristics,pressure recovery coefficient,etc.The DAMADICS simulation reference platform was used for simulation research,and the mechanism model,data model,and precision of the mixed model of the pneumatic control valve were analyzed.Secondly,to study and analyze the pneumatic control valve’s health condition,the experimental bench and practical mathematical model of a pneumatic control valve are established.The pneumatic control valve’s performance parameters are analyzed through the characteristic valve test,dynamic load performance test,and characteristic flow test,such as spring stiffness,spring preload,sliding friction coefficient,Coulomb friction force,flow characteristic,and so on.Then the actual mixing model of the pneumatic control valve is established according to the system identification data.According to the study of a valve fault,the simulation scheme of valve fault is designed from two aspects of software simulation and hardware simulation.Experiments verify the model’s accuracy,and the influence of the mechanism model and the mixed model on the accuracy of the steady-state and transition state is analyzed.Thirdly,based on principal component analysis and feature migration,the control valve’s intelligent decision-making framework is established,including anomaly detection and fault diagnosis.The threshold was established using the mixed index of square prediction error(SPE)and Hotelling’s T~2 statistics to detect anomalies.The process condition of anomalies and the change of sample distance were also taken into account.Based on the joint probability distribution and the structural risk minimization principle,the flow pattern regularization theory and the weight of condition distribution and edge distribution are adjusted by the balance factor to optimize the fault diagnosis.Through experimental analysis,abnormal detection algorithms’detection ability in regular operation,different fault degrees,and minor faults are verified.Verify the fault diagnosis algorithm’s detection capability under the same data distribution and different data distribution under different working conditions and fault degrees.Based on time series analysis and status evaluation,the pneumatic control valve status prediction is realized by analyzing and calculating the control valve’s reliability index.Then the performance evaluation index of the regulating valve is calculated by the standardization and entropy method.Time series prediction is realized utilizing the autoregressive Integrated moving average model(ARIMA),and the fault degree is analyzed through a support vector machine.Fault data are generated through DAMADICS simulation to verify the accuracy of the state prediction model.Finally,the pneumatic control valve’s health evaluation experiment was carried out to analyze the control valve’s overall health evaluation process and verify the health evaluation system’s performance.