论文部分内容阅读
针对传统温升计算方法的缺陷,提出了结合模糊系统的温升计算新方法,运用新方法对阀厅金具温升进行计算.通过阀厅金具温升试验得到足量数据,将所有数据分成训练数据与测试数据,要求训练数据代表样本空间的主要特征.将基本粒子群算法与梯度下降算法结合,得到改进粒子群算法.利用训练数据训练模糊系统,所用算法依次为基本粒子群算法、梯度法、改进算法,改进算法的收敛效果最好.运用回归分析对相应温升进行计算.通过测试检测各模型可靠性,结果说明通过改进粒子群算法训练模糊系统计算温升是可行的.“,”In light of the disadvantages of the traditional method for calculating the temperature rise,the new method combined fuzzy system is used to calculate the temperature rise.Many sample data is obtained from the temperature rise of connection fitting test.The data are divided into training data and testing data;the training data can represent the whole sample space.Improved particle swarm optimization algorithm is constituted by combining gradient descent algorithm and basic particle swarm optimization algorithm.Fuzzy system is trained by training data;the algorithm is basic particle swarm optimization algorithm,gradient descent algorithm and improved algorithm successively.The convergence effect of improved algorithm is best.The corresponding temperature rise is calculated by regression analysis.Test the reliability of each model by testing.The results show that the temperature rise could be calculated through the fuzzy system that is trained by improved particle swarm optimization algorithm.