航空学报 > 2021, Vol. 42 Issue (3): 324842-324842   doi: 10.7527/S1000-6893.2020.24842

基于GRNN算法的飞机用电设备非侵入式负荷监测方法

杨娟1, 杨占刚2   

  1. 1. 中国民航大学 工程技术训练中心, 天津 300300;
    2. 中国民航大学 电子信息与自动化学院, 天津 300300
  • 收稿日期:2020-10-07 修回日期:2020-11-05 发布日期:2020-12-18
  • 通讯作者: 杨娟 E-mail:haishi_yj11@126.com
  • 基金资助:
    中央高校基本科研业务费(3122020029)

Non-intrusive load monitoring method for aircraft electrical equipment based on GRNN algorithm

YANG Juan1, YANG Zhan'gang2   

  1. 1. China Engineering Technical Training Center, Civil Aviation University of China, Tianjin 300300, China;
    2. College of Electronic Information and Automation, Civil Aviation University of China, Tianjin 300300, China
  • Received:2020-10-07 Revised:2020-11-05 Published:2020-12-18
  • Supported by:
    Fundamental Research Funds for the Central Universities (3122020029)

摘要: 随着飞机的多电化和全电化,机上电气检测及其负荷管理至关重要,然而飞机上为诊断所设置的传感器数量要求越少越好,非侵入式负荷监测(NILM)方法无需分散进入负荷内部,仅检测汇流条级别电力参数可完成负荷识别。选择稳态电流谐波参数为负荷印记,采集某型飞机交流主汇流条上用电设备真实电流波形,提取1~19次谐波含量建立特征库,用广义回归神经网络(GRNN)算法辨识负荷类别,设置适当样本数和扩展速度以有效提高识别准确度。实验表明:GRNN算法较之BP神经网络算法和SVM算法识别准确度更高,计算速度更适于飞机电气系统负荷监测和管理。将非侵入式负荷监测方法引入飞机供电系统分析,将为飞机电气管理、故障诊断和预测等进一步研究提供有效参考。

关键词: 飞机用电设备, 非侵入式负荷监测, GRNN, 稳态电流谐波, 负荷识别

Abstract: With the development of the more and all electrical aircraft, onboard electrical equipment testing and load management has become more and more important. However, it is better to have fewer sensors set for onboard diagnosis. The Non-Intrusive Load Monitoring (NILM) method does not need to enter the internal load, and can accurately identify the load data by only detecting the bus load parameters. The steady-state current harmonic parameters are selected as the load signature, and the actual current waveform of the electrical equipment on the AC main bus of a certain type of aircraft is collected. The 1 st-19th harmonic contents are extracted to establish the feature database. The load of typical aircraft electrical equipment is identified by using the Generalized Regression Neural Network (GRNN) algorithm. The number of the samples and value of spread parameter are set appropriately to improve identification accuracy. Experimental results show that the GRNN algorithm is more accurate than the BP neural network algorithm and SVM algorithm in identification of electrical equipment on the bus, and is more applicable for management and monitoring of the aircraft electrical system due to better calculation speed. Application of the non-intrusive load monitoring method for analysis of the aircraft electrical power system can provide effective reference for management and fault diagnosis and prediction of aircraft electrical equipment.

Key words: aircraft electrical equipment, non-intrusive load monitoring, GRNN, steady-state current harmonic, load recognition

中图分类号: