论文部分内容阅读
提出一种基于最小二乘支持向量机(LS-SVM)构造函数链接型神经网络(FLANN)的方法,并根据正反馈原理将该FLANN应用於热电偶传感器非线性校正.讨论LS-SVM构造FLANN的基本原理和具体算法,给出了非线性补偿器的数学模型.与常规BP迭代算法构造的FLANN比较,该方法构造的FLANN补偿器具有如下优点:①利用LS-SVM将迭代逼近问题转化为直接求解多元线性方程,因此具有更快的速度;②整个训练过程中有且仅有一个全局极值点,确定了所构造FLANN补偿器的唯一性,提高了补偿精度.最后以Pt-Rh30-Pt-Rh6热电偶(B型)为例进行非线性校正实验,结果验证了上述结论.
This paper proposes a method based on Least Squares Support Vector Machine (LS-SVM) constructor linked neural network (FLANN), and applies the FLANN to the nonlinear correction of thermocouple sensors according to the principle of positive feedback. The basic principle and the specific algorithm are given, and the mathematical model of the nonlinear compensator is given. Compared with the FLANN constructed by the conventional BP iterative algorithm, the FLANN compensator constructed by this method has the following advantages: (1) Using LS-SVM, the iterative approximation problem Directly solve the multivariate linear equation, so it has a faster speed. (2) There is only one global extremum point in the whole training process, which confirms the uniqueness of the FLANN compensator and improves the compensation accuracy.Finally, the Pt-Rh30- Pt-Rh6 thermocouple (B type) as an example for non-linear calibration experiments, the results verify the above conclusion.