Research On Modeling Method Of Sensor Dynamic Characteristics And Model Uncertainty

With the development of economy,society,science and technology,more and more dynamic measurements are required in the fields of scientific research,aerospace,industrial control and so on,quantitative and in-depth acquisition of various parameters and physical quantity changes in transient processes are required.As the front end of the dynamic test system,the sensor's dynamic characteristics determine the dynamic performance of the entire test system.This paper mainly studies the dynamic characteristic modeling method and model uncertainty of sensors,the contents are as follows:(1)For the linear sensor model,the QR decomposition method was introduced to determine the order of the sensor dynamic model.The principle of the improved least squares algorithm,recursive maximum likelihood method,particle swarm optimization algorithm was analyzed,and the sensor linear system simulation was carried out.Under different noise models and signal-to-noise ratios,when using the recursive instrumental variable method for modeling,the fitting error between the model output and the real output of the system is less than 3%,which is better than other linear modeling methods.Hammerstein nonlinear system identification algorithm and BP neural network algorithm were introduced,and nonlinear system simulation was carried out.(2)Carry on the linear system simulation,when there was no noise interference,use recursive least square method for parameter identification,and obtain the true value of the system model parameters.The Bootstrap sampling method was used to evaluate the uncertainty of the linear model of the sensor.Through the simulation of the uncertainty of the sensor linear model,the Bessel method,Monte Carlo method and Bootstrap sampling method were compared in the evaluation of the uncertainty of the linear system model.(3)Different linear modeling methods were used to dynamically model a domestic piezoelectric pressure sensor,and the recursive instrumental variable method was used to dynamically compensate the domestic piezoelectric pressure sensor and Endevoc 8530 C piezoresistive pressure sensor.The uncertainty of the pressure sensor dynamic characteristic model was discussed.A K-type thermocouple with a diameter of 1mm was identified by Hammerstein model and BP neural network model,and the BP neural network algorithm was used to dynamically compensate the thermocouple.Through the research on the modeling method of the dynamic characteristics of the sensor and the uncertainty of the model,it can be concluded that the recursive instrumental variable method has strong adaptability and high modeling accuracy in the linear modeling of the sensor.The BP neural network algorithm has a good effect in the nonlinear modeling and dynamic error correction of the sensor.The Bootstrap sampling method is effective for the evaluation of the uncertainty of the linear model of the sensor.