电场传感器性能改善算法研究

Research on performance improvement algorithm of electric field sensor

针对目前国内用于检测电场的传感器产品很少,且在实际应用中难以满足检测精度,抗干扰能力较差等问题,该文采用森馥SEM-600低频电磁辐射测量仪,对一台标准电场发生器产生的电场进行三维分量检测,并在此基础上开展电场传感器性能改善研究。研究采用最小二乘法对该电场传感器的原始检测数据进行分析,计算结果作为传感器的基本性能技术指标。同时研究采用BP神经网络、支持向量机、RBF和遗传算法等结构映射方法,建立电场强度的读数模型,对电场传感器的性能进行优化,以提高传感器的检测准确度。通过对这5种算法读数结果的对比,最终确定采用误差最小的支持向量机算法,建立电场传感器的准确读数模型。优化后,在0~20 kV/m的电场范围内,线性度由原始检测数据的4.33%提升至0.01%,误差由13.6%降低至1.9%,使传感器的性能得到显著提升。

At present,there are less domestic sensor products used to detect electric fields,and in the meantime,the products are also limited by the detection accuracy in application and poor anti-interference ability.In the article,a Senfu SEM-600 of low-frequency electromagnetic radiation was used to detect the threedimensional components of electric field which were generated by a standard electric field generator,and then the research was conducted on the performance improvement of the electric field sensor of Senfu SEM-600.The least-square method was used to analyze the original detection data of the electric field sensor,and the calculation results were used for the basic performance indices of the sensor.In the meantime,the structure mapping methods such as BP neural network,support vector machine,RBF and genetic algorithm were studied to establish the reading model of the electric field intensity,optimize the performance of the electric field sensor,and further improve the accuracy of the sensor.By comparing the reading results of the five algorithms,the SVM algorithm with the minimum reading error was determined to establish the accurate reading model of the electric field sensor.In the electric field range of 0-20 kV/m,the linearity of the original test data increased from 4.33% to 0.01%, and the error decreased from 13.6% to 1.9%. It shows that the performance of the SenfuSEM-600 has been significantly improved after the above optimization.