基于支持向量机的传感器非线性动态补偿方法

Nonlinear Dynamic Compensation of Sensors Based on Least Squares Support Vector Machines

提出了应用支持向量机(LS-SVM)实现传感器非线性动态补偿方法.LS-SVM的训练过程遵循的是结构风险最小化原则,而不是通常神经网络的经验误差最小化,可获得更好的泛化性能,不易发生局部最优及过拟合现象,因此可弥补应用人工神经网络进行传感器非线性动态补偿的缺陷.通过实例验证了该方法的可行性,结果表明,即使当传感器动态模型存在严重非线性,且有测量噪声存在,该方法也仍然有效.

The least squares support vector machine （LS--SVM） is proposed for nonlinear dynamic compensation of sensors based on the structural risk minimization principle rather than the empirical error minimization principle commonly implemented in the neural networks, the LS-SVM can achieve higher generalization performance, the local minima and over fitting are unlikely to occur. Therefore, the LS-SVM can overcome the shortcomings of neural networks in nonlinear dynamic compensation of sensors. The feasibility of the method is demonstrated by applying it to a practical example. The experimental results show that the method is still effective even if the sensor＇s dynamic model is of high nonlinearity and there exists additive measuring noise.