基于专家知识与BP神经网络的架空导地线状态评价研究

Research on evaluation of aerial guideline status based on expert knowledge and BP neural network

目前输电设备运维过程存在着规程过于繁杂,规定的参量较为冗余等问题。提出了一种导地线状态评价方法,该方法通过关联分析从规程中提取出日常运维过程中广泛应用的关键参数作为运行状态的评价指标,并采用灰色关联确定对设备运行状态具有强关联性的外界因素作为影响因子,结合运维专家经验建立导地线状态评价专家系统,提高状态评价的准确性与可靠性。采用历史缺陷样本与专家系统评价样本作为训练样本,通过Levenberg-Marquardt算法进行BP神经网络的训练,建立融入专家经验的导地线智能状态评价模型。通过运行实例分析表明本文训练得到的状态评价模型能够准确地实现对导地线的状态评价,具有相当可靠性。

At present,there are problems in the operation and maintenance process of transmission equipment,such as too complicated procedures and redundant parameters.A method for evaluating the state of a wire is proposed.The method extracts the key parameters widely used in the daily operation and maintenance process from the rules as the evaluation index of the running state,and uses the gray correlation to determine the strong correlation with the operating state of the equipment.As an influencing factor,the external factors are combined with the experience of operation and maintenance experts to establish a wire state evaluation expert system to improve the accuracy and reliability of the state evaluation.Historical defect samples and expert system evaluation samples are used as training samples.The BP neural network is trained by Levenberg-Marquardt algorithm to establish a wire intelligent state evaluation model incorporating expert experience.The running example analysis shows that the state evaluation model obtained in this paper can accurately evaluate the state of the grounding line and has considerable reliability.