基于最小二乘支持向量机的传感器非线性动态补偿

提出了一种基于最小二乘支持向量机的非线性传感器动态测量误差的校正方法，使得通过该方法补偿的传感器具有理想的输入输出特性。先将传感器的非线性动态系统分解成线性动态子环节和非线性静态子环节串联；与之对应，非线性动态补偿过程也包含2个阶段：线性动态补偿和非线性静态校正。然后，通过函数展开将补偿器的非线性传递函数转换为等价的类线性形式一中间模型；再通过LS-SVM回归算法求取中间模型参数；最后，推导出中间模型参数与补偿器2个子模型参数之间的关系，并通过该关系实现非线性静态校正和线性动态补偿环节的同时辨识。与常规非线性动态补偿方法比较，该方法优点是明显的：（1）只需进行一次动态标定实验；（2）能给出非线性动态补偿器的数学解析表达式；（3）充分利用LS—SVM的优点，使辨识的补偿器具有更好的抗干扰能力。仿真与实际实验结果均表明该传感器非线性动态补偿方法有效。

Compensation for nonlinear dynamic system of transducer based on least squares support vector machine

A new approach to correct dynamic measurement errors of nonlinear transducer systems base on least squares support vector machine(LS-SVM)was investigated,through which the desired characteristics can be obtained from compensated one.The original nonlinear dynamic system of the transducer is expressed as a linear dynamic subunit followed by a nonlinear static subunit;in accordance with that,the nonlinear dynamic compensation process includes two stages:nonlinear static correction and linear dynamic compensation.Then with function expansion,the nonlinear transfer function of the compensator is converted into a class of linear one-intermediate model.And the parameters of the intermediate model are obtained through LS-SVM regression algorithm.Finally,the relationship between the parameters of the intermediate model and those of the two parts of the transducer is derived,through which the nonlinear static part and linear dynamic part of the compensator are identified simultaneously.Compared with conventional nonlinear dynamic compensation methods,the presented approach has prominent advantages:(1)Only one dynamic calibration experiment is needed.(2)The analytic expressions of nonlinear dynamic compensator can be derived.(3)The compensator is more robust in noise interference rejection capability due to the good features of LS-SVM.Simulations and experimental results show that the compensation method for nonlinear dynamic system of transducer is effective.