A Sensor’s Dynamic Compensation Method Based On ⅡR Digital Filter

whether the signal can be accurately measured is a very important issue in the transientmeasurement field. The signal changes very fast and it contains a wealth of high-frequencycomponents. The dynamic characteristics of the sensor must be first considered for thedetection of this signal. Only in the case of that the sensor’s bandwidth is wider than500Hz,the signal whose spectrum is below500Hz can be measured accurately. In many cases, thedynamic performance of the sensor can not meet the requirements of the measurement. Themanufacturing process of the sensor can be improved or better sensors can be used to solvethis problem. But the costs are higher. If the signal processing method is used to real-timeprocess the output signal of the sensor in the measurement system which comprised amicroprocessor, it is a better choice.Fuzzy neural network is used to improve the dynamic characteristics of a second-ordersystem and the simulation is done. To further enhance the dynamic characters of the sensorthe improved RLS algorithm and genetic algorithm is used to identify the parameters of theself-regression model to establish the inverse model of the sensor. The gradient descentmethod is used to generate the initial parameters of the self-regression model and RLS、genetic algorithm are used to optimize and correct the parameter values.Their results are better than fuzzy neural network after comparison. Two methods have avery similar effect of the parameter identification.The improved genetic algorithm is used to identify the parameters of the ⅡR filteroff-line. The parameters are stored into the FPGA to correct the dynamic characteristics of the sensor. In fact, the realization is possible. The charge amplifier, instrumentation amplifier, alow pass filter, A/D converter and D/A converter are designed. The detection memorymodule which has a negative delay function and a adaptive sampling frequency function isdesigned. The core of the module is FPGA. The Verilog HDL language is used in real-timeimplementation of the algorithm. Experimental results show that the method can extend thebandwidth of the sensor whose damping ratio is0.1and the natural frequency is6283.18rad/sto1/2of the natural frequency.