摘要
为了获得更高的电缆接头温度预测精度,引入了粒子群(PSO)优化算法来动态寻优标准化参数。以PSOSVM算法对电缆接头温度进行预测,生成相应的训练与测试样本。通过训练样本来计算PSO-SVM模型乘子λ及其偏差量B,再根据计算得到的B与λ处理测试样本获得模型精度与预测效果。仿真分析结果表明:采用PSO-SVM方法可以预测得到更加符合实测值的结果,获得比SVM预测方法更优的相对误差,得到的优化参数是完全有效的。大小不一样的数据样本会对预测结果精度造成明显影响,其中样本较多时可以获得相对更高的预测精度。
The particle swarm optimization(PSO)algorithm was introduced to dynamically optimize the standardized parameters in order to obtain higher prediction accuracy of cable joint temperature.The PSO-SVM algorithm was used to predict the cable joint temperature,and the corresponding training and test samples were generated.The training sample is used to calculate the multiplier of PSO-SVM model with lambda or its deviation B,and then the model accuracy and prediction effect are obtained according to the calculated B and lambda processing test sample.Simulation analysis results show that the pso-svm method can be used to predict the results that are more consistent with the measured values,the relative error is better than the SVM prediction method,and the optimized parameters obtained are completely effective.Different size of data samples will have a significant impact on the accuracy of prediction results,among which a large number of samples can obtain a relatively higher prediction accuracy.
作者
樊浩
宁博扬
何森
Fan Hao;Ning Boyang;He Sen(Skills Training Center of State Grid Hebei Electric Power Co.,Ltd.,Baoding 071051,China;State Grid Shanghai Electric Power Company,Shanghai 200072,China)
出处
《电子测量技术》
2019年第21期53-56,共4页
Electronic Measurement Technology
关键词
电力电缆
接头温度预测模型
SVM
粒子群算法
power cable
joint temperature prediction model
SVM
particle swarm optimization
