基于BP神经网络的配网设备故障预测

发布时间：2018-06-27 12:41

  本文选题：BP神经网络 + 故障预测　； 参考：《广东工业大学》2017年硕士论文

【摘要】：伴随我国社会经济和科技的高速发展,电力需求量在大幅度增长,这样就促使了电力缺口越来越大,人们也对电网企业供电的稳定性与可靠性有了更高标准的要求,如此就推动了电网规模扩大、电力设备量急剧增长的进程。电网规模的扩大和设备的增多,能更好地满足人们的生产生活需求,但随之而来的是故障增多,停电范围扩大和停电时间增加等严重影响生产生活的问题。所以,保证电网正常、可靠运行,避免设备故障或少发生故障,且能在故障后能够迅速、准确地定位并排除,这对于运行维护人员是个巨大的挑战。为弥补传统维修方式的不足,人们借助计算机技术、状态监测和故障诊断技术等新锐技术,创造出了新的维修方式,就是基于状态的维修(Condition Based Maintenance,CBM),也称为视情维修。这一维修方式充分运用各种技术手段来获取设备运行时的数据,再利用故障预测和诊断技术进行综合分析,确定设备运行状态,然后预测其发展趋势以及会发生何种故障、何时发生和何地发生,实现能通过在线监测设备状态、预测即将发生的故障和制订合理的预防措施或维修策略的重大目标。基于状态的维修主要是能根据每个设备不同的运行状态来预测其劣化程度及趋势,并对设备检修做出合理、科学的维修决策,判断是否有需要对设备进行预防性维修和维修何时进行,将故障抑制在萌芽状态,所以,其维修间隔期并不固定。这对于电网企业而言,能提高企业的供电可靠性,解决电能质量低、故障停电时间多、故障停电范围大和配网系统运行的经济性低等问题,降低运行维护费用,提升设备维护和维修水平,实现精确维修,提高企业经济效益,提升企业对电网的管理水平和工作效率,提升企业“为人民服务”的形象,对公众的承诺得到兑现。而故障预测是故障诊断的重要组成部分,它通过分析历史和当前数据,筛选提取出设备故障特征值及其运行发展趋势,进而对设备未来的运行状态和可能出现的故障进行预测,确定设备运行状态级别,提早掌握设备劣化趋势,做到提早预防和修复。利用故障预测来解决设备故障问题,这样不只具有重要的理论探索价值,而且还具有广泛的工程应用意义。本论文结合基于状态的维修技术和神经网络技术,提出基于在线运行设备故障预测的模型。该模型根据故障的严重性,将风险等级划分为四个级别,分别以“Ⅰ级、Ⅱ级、Ⅲ级和Ⅳ级”来表示,这既能看出设备未来的运行状态,也有助于差异化维修的决策。通过分析处理历史数据,对故障特征值进行提取及收集,形成特征值样本集,再利用样本集来训练设计好的神经网络,调整权重,对神经网络结构设计进行优化,建立基于神经网络的故障预测模型,以达到对设备故障的预测的目的。
[Abstract]:With the rapid development of social economy and science and technology in our country, the demand for electric power is increasing by a large margin, which makes the gap of electric power more and more large, and people also have higher requirements for the stability and reliability of power supply in power grid enterprises. In this way, the expansion of the scale of the power grid, the rapid growth of power equipment process. The expansion of power grid and the increase of equipment can better meet the needs of people's production and life, but with the increase of failures, the range of power outages and the time of power outages will seriously affect the production and life. Therefore, it is a great challenge for the operation and maintenance personnel to ensure the normal and reliable operation of the power network, to avoid the equipment fault or to avoid the fault, and to be able to locate and eliminate it quickly and accurately after the failure. In order to make up for the deficiency of the traditional maintenance method, people have created a new maintenance method, which is condition based maintenance (CBM), also called condition based maintenance (CBM), with the help of computer technology, condition monitoring and fault diagnosis technology and so on. This maintenance method makes full use of all kinds of technical means to obtain the data while the equipment is running, and then uses the fault prediction and diagnosis technology to carry on the comprehensive analysis, determines the equipment running status, then predicts its development trend and what kind of malfunction will occur. When and where will happen, realize the important goal that can monitor the status of the equipment online, predict the upcoming failure and make reasonable preventive measures or maintenance strategy. Condition-based maintenance can predict the deterioration degree and trend of each equipment according to its different operating state, and make reasonable and scientific maintenance decision for equipment maintenance. Whether it is necessary to carry out preventive maintenance and when to carry out maintenance, the fault will be restrained in the embryonic state, therefore, the maintenance interval is not fixed. For power grid enterprises, it can improve the reliability of power supply, solve the problems of low power quality, more time of failure, large range of failure and low economy of distribution network operation, and reduce the cost of operation and maintenance. Improve the level of equipment maintenance and maintenance, achieve accurate maintenance, improve the economic efficiency of enterprises, improve the management level and working efficiency of enterprises to the power grid, enhance the image of enterprises "serving the people", and fulfill the promise to the public. Fault prediction is an important part of fault diagnosis. By analyzing the history and current data, it extracts the characteristic value of equipment fault and its development trend, and then predicts the future running state and possible faults of the equipment. Determine equipment operation status level, grasp equipment deterioration trend early, achieve early prevention and repair. Using fault prediction to solve the problem of equipment failure not only has important theoretical exploration value, but also has a wide range of engineering application significance. In this paper, a fault prediction model based on on-line operation equipment is proposed by combining state based maintenance technology and neural network technology. According to the severity of the fault, the model divides the risk level into four levels, which are expressed as "鈪，

本文编号：2073951