摘要
电缆内部温度是电缆管理的重要参数,针对高精度数值计算和实验测量均不能快速准确预测电缆内部温度的问题,提出一种基于多物理场仿真数据构建代理模型,并基于代理模型实现电缆内部温度场快速计算的方法。以110 kV单芯高压电缆为例,首先对比拉丁超立方设计和最大最小思想优化的拉丁超立方方法构造样本空间的效果,选取最优采样方案,再通过径向基神经网络构建代理模型,以环境温度和载流量为变量构建温度数据集进行测试,进一步通过粒子群算法对代理模型进行优化,最后利用网格节点数据使温度场分布可视化。对比110 kV单芯高压电缆内部温度场计算结果,表明所提的基于代理模型的电力电缆温度场快速计算方法具有较高的精度和效率。
Accurately predicting the temperature inside cables is crucial for effective cable management.However,this process can be time-consuming through numerical calculations and experimental measurements.To address this issue,a new method for constructing surrogate model based on multi-physics field simulation data is proposed to calculate the cable's internal temperature field quickly and accurately.A single-core 110 kV high-voltage cable is taken as a study subject.Firstly,the Latin hypercube design and the Latin hypercube method optimized by the maximum-minimum idea are compared in constructing the sample space.The optimal sampling scheme is selected to construct the surrogate mode through RBF neural network,and a temperature data set is created for testing using ambient temperature and load capacity.Additionally,the surrogate model is optimized by the particle swarm algorithm,and the temperature field distribution is visualized with the grid node data.The calculation result of the internal temperature of the 110 kV single-core high-voltage cable shows that the proposed surrogate model-based fast calculation method for power cable temperature field has high accuracy and efficiency.
作者
黄莉
梁云
黄辉
孙晓艳
王珊
杨玉强
HUANG Li;LIANG Yun;HUANG Hui;SUN Xiaoyan;WANG Shan;YANG Yuqiang(State Grid Smart Grid Research Institute Co.,Ltd.,Beijing 102209,China;Electric Power Intelligent Sensing Technology of State Grid Corporation Laboratory,Beijing 102209,China;State Grid Zhejiang Electric Power Co.,Ltd.,Hangzhou 310007,China)
出处
《中国电力》
CSCD
北大核心
2024年第5期178-187,共10页
Electric Power
基金
国家电网有限公司科技项目(基于多物理场耦合分析的电力电缆感知反演模型及状态表征方法研究,5500-202258316A-2-0-QZ)。
关键词
代理模型
电力电缆
有限元仿真
快速计算
surrogate model
power cable
finite element simulation
fast calculation
