基于插排外部热特征的电流反演过载预测研究OA
Research on current inversion and overload prediction based on external thermal characteristics of power strips
针对传统插排过载监测依赖电流传感器、侵入性强且成本高的问题,提出一种基于导线外部热特征的非侵入式监测方法.构建"导体芯—导线表皮—环境"双节点电热耦合模型,通过实验辨识热参数刻画负载电流与外部温度的动态关系,引入状态观测器补偿测温热滞后并推导电流反演方法;针对导线参数不确定性,设计以温度时间序列为输入的轻量级 1D 卷积神经网络,将过载判别建模为二分类问题以降低物理参数依赖.多规格导线与多负载工况实验验证表明,该方法在5~15A负载范围内动态响应迅速,在未知导线条件下过载检测准确率超90%,无需额外电流传感器即可有效监测过载风险,工程应用价值显著.
Traditional power strip overload monitoring typically relies on current sensors,which are invasive,costly,and difficult to deploy in existing devices.To address these limitations,this paper proposes a non-intrusive overload monitoring method based on the ex-ternal thermal characteristics of power strip conductors.A dual-node electro-thermal coupling model consisting of the conductor core,wire surface,and ambient environment is established to describe the dynamic relationship between load current and external temperature.The thermal parameters are identified experimentally,and a state observer is introduced to compensate for the thermal lag inherent in sur-face temperature measurements,enabling dynamic current inversion from external temperature data. To mitigate the strong dependence of analytical thermal models on unknown wire parameters,a lightweight one-dimensional convolu-tional neural network(1D-CNN)is further designed using temperature time-series inputs,and overload detection is formulated as a bina-ry classification problem.Experimental results obtained under multiple wire specifications,load profiles,and ambient conditions demon-strate that the proposed method achieves a fast dynamic response within a load range of 5-15 A and maintains an overload detection ac-curacy exceeding 90%for previously unseen wire configurations.The proposed approach enables effective overload monitoring without ad-ditional current sensors,providing a practical and low-cost solution for power strip safety monitoring.
肖林廷;李明桧
沈阳航空航天大学 电子信息工程学院,沈阳 110136沈阳航空航天大学 电子信息工程学院,沈阳 110136
信息技术与安全科学
非侵入式监测电热耦合模型状态观测器电流反演神经网络
Non-intrusive monitoringElectro-thermal coupling modelState observerCurrent inversionNeural network
《通信与信息技术》 2026 (5)
30-34,5
评论