架空电力线故障诊断系统的仿真数据生成算法

Simulation data generation algorithm for the fault diagnosis system of overhead power lines

现代社会的发展对电力的依赖性越来越强,为了保障线路的稳定运行,深入进行线路故障诊断的研究显得非常重要。但是由于数据缺乏一致性、采集困难等问题,模型训练、测试数据的获取一直困扰着相关研究的开展。为此,提出了一种故障数据生成算法PFGA,利用该算法可以生成大量的设备故障数据、单元故障数据及架空电力线路故障数据。由于该算法运用了云模型及多智能体系统等机制,生成的故障数据除了包含一定的先验知识外还存在一定的随机性,故障数据也很好地体现了真实故障数据的层次关联性、多样性、动态性等特点。利用构建的贝叶斯诊断模型分别利用PFGA和其他算法生成的数据进行诊断,结果表明,PFGA生成的数据更符合现实状况,相应的诊断结果也更为可靠。此外,该算法对不同线路状况的适应性也较好,研究人员可以根据实际研究情况生成所需的架空电力线路故障数据,以满足各类故障诊断研究对数据的需求。

The development of modern society is increasingly dependent on electric power. To ensure stable operation of power lines, it is extremely important to carry out the research on the fault diagnosis of overhead power lines. How- ever, due to poor data consistency and acquisition difficulty, the problems, such as model training and test data acqui- sition, have greatly hindered the relevant research work. Thus, this paper proposes a fault data generation algorithm named PFGA （ power line fault dataset generation algorithm）. The algorithm can generate a lot of equipment fault da- ta, unit fault data and the fault data of overhead power lines. Because the algorithm uses cloud model and multi-agent system, these generated fault data contain not only some priori knowledge but also some randomness;and these fault data also reflect the hierarchical correlation, diversity and dynamics of real fault data. The diagnosis model construc- ted with Dynamic Bayesian Network was used to implement the fault diagnosis on the data generated using PFGA and other algorithms. The result shows that the data generated with PFGA can better simulate the real condition and the corresponding diagnosis results are more reliable and accurate. With PFGA algorithm, researchers can generate re- quired fault data of overhead power lines based on actual studies, and fulfill the data requirements of various fault di- agnosis researches.