【摘要】 天线等弱电通信设备工作在电力铁塔等强电环境中时，出于安全考虑需要计算天线所在位置的电磁环境。现有电磁计算方法涉及复杂变换与迭代过程，耗时长、占用内存高，工程人员不易掌握。为此，将机器学习中的随机森林回归模型应用到多场源和散射体的复杂情况中，建立了多层快速多极子算法（MLFMM）与随机森林结合的场强预测模型。基于MLFMM算法，利用电磁仿真软件FEKO获取电磁环境中大量的电磁分布数据，并进行数据挖掘和特征分析，以提高场强预测效率，降低计算复杂度。最后，利用随机森林回归模型实现场强预测。高电压环境下金属散射体附近的天线场强预测结果与数值计算方法结果吻合良好，准确率高于95%，证明了机器学习方法结合数值电磁计算方法的可靠性。更多还原

【Abstract】 When the weak current communication equipment such as antenna works in a high voltage electric environment such as power tower, the electromagnetic environment of the working position needs to be calculated for safety reasons.The existing electromagnetic calculation methods involve complex transformation and iteration process, which is time-consuming and occupies high memory, so it is difcult for engineers to master them. Therefore, the random forest regression model in machine learning is applied to the complex situation of multiple field sources and scatterers, and a field strength prediction model combining multi-layer fast multipole algorithm(MLFMM) and random forest is established. Based on MLFMM algorithm, the electromagnetic simulation software FEKO is used to obtain a large number of electromagnetic distribution data in the electromagnetic environment, and the data mining and feature analysis are carried out to improve the efciency of field strength prediction and reduce the computational complexity. Finally, the random forest regression model is used to achieve field strength prediction. The prediction results of the antenna field intensity near the metal scatterers in high voltage environment are in good agreement with the results of the numerical calculation method, and the accuracy is higher than 95%, which proves the reliability of the machine learning method combined with the numerical electromagnetic calculation method.更多还原