论文部分内容阅读
使用微波谐振腔对物料含水率测量过程中,减少谐振参量与含水率多元非线性回归过程的误差是影响测量精度的主要因素。针对这一问题,建立了一种基于支持向量机多元非线性回归模型,并确定了其中谐振频率、品质因数和环境温度的特征值、贡献率。应用SVM-KM对该模型进行实验研究,利用50组数据对模型进行训练并验证其学习性能,利用另外15组数据验证其泛化能力。实验表明,该方法能够实现微波谐振腔物料含水率的软测量,且小样本条件下比神经元网络具有优势。对SVM多元非线性回归泛化性能进行测试,其均方根相对误差为1.06%,平均绝对相对误差为0.96%,最大绝对相对误差为1.16%。
The use of microwave cavity in the material moisture content measurement process to reduce the resonance parameters and moisture content of the multivariate nonlinear regression error is the main factor affecting the measurement accuracy. Aiming at this problem, a multivariate nonlinear regression model based on support vector machine is established and the eigenvalues and contribution rates of resonance frequency, quality factor and ambient temperature are determined. SVM-KM was used to study the model. Fifty groups of data were used to train the model and verify the learning performance. Another 15 groups of data were used to verify its generalization ability. Experiments show that this method can achieve the soft measurement of moisture content of microwave cavity material, and has advantages over the neural network under small sample conditions. The multivariate nonlinear regression generalization performance of SVM was tested, the relative root mean square error was 1.06%, the average absolute relative error was 0.96% and the maximum absolute relative error was 1.16%.