Research On Nonlinear Correction Of Fiber Bragg Grating Voltage Sensor

As the core component of the voltage measurement system,the accuracy of the voltage sensor plays a decisive role in the performance of the voltage measurement system.The key to improve the performance of the sensor system is to improve the accuracy of the voltage measurement.During the process of studying high voltage sensors,due to the unique advantages of optical sensors,they have been gradually applied to the power grid and high voltage laboratories.In the research of optical voltage sensor,and analyzing the theoretical knowledge of measuring voltage of optical sensor,it is found that Fiber Bragg Grating(FBG)sensor is sensitive not only to strain but also to temperature.In the experiment,only the strain characteristic of the sensor is used,so the experiment of temperature self-compensation is designed.After the realization of FBG temperature self-compensation,the voltage measurement model of optical fiber optical sensor is established according to electrostatic mechanics and elastic mechanics,also established the sensor measurement system,and the theoretical mathematical model of applied voltage and FBG wavelength drift is obtained.On the basis of the established sensor measurement system,the calibration experiment of the sample data is carried out and a set of data is obtained.Sensors plays an important role in the measurement system,and the reliability of its output signal is very important to the whole measurement system.However,the sensor is always affected by a variety of environmental factors in actual production,resulting in the sensor measurement accuracy and reliability is not satisfactory.This thesis proposes a correction algorithm of ant colony optimization last square support vector regression(ACO-LS-SVR).Because the search speed of traditional ant colony algorithm in signal processing is not unsatisfactory,the accuracy of Least Square Support Regression(LS-SVR)algorithm is also unsatisfactory.So this thesis bases on combines ant colony algorithm to search the best parameters of LS-SVR to stage regression of sample space according to data distribution,in order to reduce the running time of the algorithm.First,the parameters of support vector machine(SVM)are optimized by ant colony algorithm,then the complete characteristic of sensor is obtained by stage regression.The algorithm is simulated in MATLAB software,and the curve fitting accuracy is 98.39%.This algorithm overcomes the problem of the local optimal solution in the traditional SVR algorithm and has a good global convergence effect,Thus,the non-linear correction of optical voltage sensor is realized.