LS-SVM在单通道颜色测温中的应用

针对复杂工业环境中高温难以直接测量的问题，提出一种基于最小二乘支持向量机（LS—SVM）的单通道颜色测温方法。利用高温物体的颜色与温度之间复杂的非线性映射关系，采用绿色单通道图像样本特征值建立LS—SVM回归模型实现颜色测温，并与BP神经网络模型进行比较。仿真结果表明，LS—SVM测温模型具有良好的非线性建模和泛化能力。该方法能够间接得到温度测量值，并且具有较高的预测精度，为高温测量提供一个新的有效手段。

Application of LS-SVM in Temperature Measurement Using Single Colour

A method based on least square- support vector machine （LS- SVM） in temperature measurement using sing colour is proposed to overcome the difficulty of the high temperature measurement which is hard to measure directly in complex industrial environment. The method uses the image feature of single colour of green, in the light of the complex nonlinear relationship between the colour of the high- temperature object and the temperature, to build the regression model. And it is compared with the BP neural network at last. The simulation result shows that, the LS- SVM model has good ability of modeling nonlinear process and good generalization. The proposed method can predict the temperature value indirectly and the predictive accuracy is satisfied in application. And a novel approach is provided for the high - temperature measurement field.