论文部分内容阅读
根据Zr-2合金的晶粒尺寸在不同热工艺参数(变形温度、变形程度、变形速率)下的12组实测数据,应用基于粒子群算法寻找最优参数的支持向量回归方法,建立了合金晶粒尺寸的预测模型。通过与模糊神经网络模型的结果进行比较,结果表明:基于相同的试验样本,支持向量回归预测模型的平均绝对误差和平均绝对百分误差都比模糊神经网络预测模型的小,而复相关系数大。这说明,支持向量回归预测模型预测精度比模糊神经网络模型要高,是简单而精确的建模方法,可用于优化热加工参数。
According to the 12 groups of measured data of the grain size of Zr-2 alloy under different thermal parameters (deformation temperature, degree of deformation, deformation rate), a support vector regression method based on Particle Swarm Optimization Prediction model of grain size. Compared with the results of fuzzy neural network model, the results show that the average absolute error and the average absolute percentage error of support vector regression prediction model are smaller than those of fuzzy neural network based on the same test sample, and the complex correlation coefficient is large . This shows that the prediction accuracy of support vector regression prediction model is higher than the fuzzy neural network model is a simple and accurate modeling method can be used to optimize the thermal processing parameters.