基于遗传算法的RBF神经网络在热敏电阻温度传感器非线性补偿中的应用

A New Approach to Non-Linearity Compensation of Thermistor Temperature Transducer Based on Genetic Algorithm and RBF Neural Network

针对热敏电阻温度传感器应用中存在的非线性问题,提出了应用遗传算法训练径向基函数(RBF)神经网络实现其非线性补偿的遗传神经网络方法,介绍了非线性补偿的原理和网络训练方法。该方法能同时优化网络结构和参数,具有全局寻优能力,补偿精度高。实验结果表明,所提出的热敏电阻温度传感器非线性补偿方法是实用的和可行的。

A method is presented to compensate non-linearity of thermistor temperature transducer using non-linearity compensation model founded by radial basis funcation （RBF） neural network that is trained by genetic algorithms to settle its non-linear problem. The principle and training method of neural network are introduced. In this method, the configuration and parameters of non-linearity compensation model are optimized by genetic algorithm. The results show that the proposed non-linearity compensation method has the advantages of fast training process, good global searching ability, high precision and strong robustness. It makes convenient for thermistor temperature transducer to be applied in the measurement.