基于QPSO稀疏化LSSVM的压力传感器温度补偿研究

针对硅压阻式压力传感器的温度补偿问题,提出一种量子粒子群(Quantum Particle Swarm Optimization,QPSO)稀疏化最小二乘支持向量机(Least Squares Support Vector Machine,LSSVM)策略,目的在于能够保证温度补偿性能的同时获得较为精简的补偿模型。结合10 MPa绝压压力传感器标定实验数据进行仿真试验,研究结果表明,该方法的补偿效果优于QPSO优化的LSSVM、经典稀疏化LSSVM和QPSO优化的稀疏化LSSVM,补偿后测试样本集的最大相对误差,平均误差和误差方差分别为1.104×10^-3、4.819×10^-4和1.197×10^-7。在满足高精度测试要求的前提下,达到提升补偿效率的目的。

Research on Temperature Compensation for Pressure Sensor Based on QPSO sparse LSSVM

To address the temperature compensation issue of piezo-resistive pressure sensor,a strategy of quantum particle swarm optimization(QPSO)sparse least square support vector machine(LSSVM)was proposed,the aim of which is to obtain a relative compact compensation model without any loss of the performance of temperature compensation. A calibration experiment of 10MPa absolute pressure sensor was conducted,the simulation results of which demonstrate the presented compensation method can obtain a more satisfactory compensation precision compared with the results come from QPSO improved LSSVM,typical sparse LSSVM and QPSO improved sparse LSSVM. The maximum relative error,mean error and error variance of the testing set after compensation is 1.104×10^-3,4.819×10^-4 and 1.197×10^-7 respectively. The compensation results also show that the presented temperature compensation method is able to reach a balance between compensation efficiency and compensation effectiveness.