Research On Fa Ult Diagnosis And Health Evaluation Algorithm Of Self-validating Pneumatic Actuator

Actuator,as the terminal execution device of industrial automation control system,is closely connected with the production process.Its operating environment is relatively harsh.which is easy to cause the performance degradation of the actuator and increase the probability of failure.Therefore,in order to prevent the occurrence of catastrophic accidents.it is very important to improve the system reliability.The technology of fault diagnosis and health status assessment is an effective way to ensure the stable operation of the system.In order to solve the problem of fault diagnosis and health evaluation about self-validating pneumatic actuators,firstly,a pneumatic actuator fault diagnosis method based on multi-kernel multi-class relevance vector machine was proposed.The simulation model of pneumatic valve was established based on DAMADICS platform and fault data was generated by DABLIB module.Relevance vector machine regression was used to establish the recovery model according to normal sample sequence.When establishing the data recovery model.different kernel functions are selected to map the characteristic parameters of different output types,and the parameters are optimized based on K-fold cross validation algorithm.The residuals were generated by comparing the output of the actual actuator,and the feature extraction was achieved.Taking the feature as the input,the combination of gaussian kernel function and polynomial kernel function was selected.A hybrid algorithm of adaptive particle swarm optimization algorithm and genetic algorithm was used to achieve optimization of multi-objective kernel parameters.A multi-kernel multi-class relevance vector machine was established to diagnose the fault type of pneumatic actuator.Then a data-driven method based on the normal operation of the actuator was proposed.The residual feature was obtained as the event set.The evaluation indexes of health,sub-health,marginal failure and failure were defined as the target countermeasure set.The normal distribution function and the semi-trapezoidal function were selected as the membership function to establish the benchmark models that express the performance degradation degree of the actuator.The weight distribution models of local health degree and comprehensive health degree were established by using analytic hierarchy process,grey relation algorithm and entropy method.Finally,the least squares support vector machine was used to determine the health level.Experimental results show that the method has higher modeling accuracy and better practicability,and realizes multi-fault diagnosis and classification of self-validating pneumatic actuators.It can realize the overall and partial health assessment of pneumatic actuators and reflect the performance status of self-validating pneumatic actuators.