风力发电机故障诊断系统的研究

发布时间：2018-04-09 19:13

本文选题：风力发电机　切入点：故障诊断　出处：《兰州交通大学》2014年硕士论文

【摘要】：风力发电机安装位置高，一般工作条件恶劣。出现故障不好维修，针对风力发电机组的运行状态进行有效、及时、准确的预测故障的发生和发展，避免和减少故障的发生，降低维修成本。对于风力发电机组故障率发生最高的则是发电机，所以从发电机着手来进行故障诊断研究，发电机结构的复杂性，并保证在不拆卸、不破坏设备并保证设备在正常运行的情况下，能够准确的判断工作状态、性能好坏、参数变化和故障趋势，这就需要一套功能完善、性能稳定的状态检测和故障诊断系统。由于发电机属于旋转机械，而旋转机械在高速运转时会产生振动，利用对振动信号的实时监测，，系统便可从振动信号来进行状态识别。这样一套风力发电机故障诊断系统，首先要在不妨碍机器运转的情况下，使传感器尽量靠近振源安装，从多个角度和方向来选好测点和探头位置，对运行的发电机常见故障（正常、转子不对中、轴向裂纹和油膜震荡）进行振动位移的测量，并对故障情况进行振动信号模拟。其次，对测得的振动信号进行MATLAB小波分析，利用小波包进行信噪分离，分离出无用的噪声信号，将有用的信号保留，进行小波包能量谱分解，算出每个频带所占总能量百分比，作为故障诊断中识别的特征参数。再次，将MATLAB小波包去噪分解后各频带所提取的能量特征值作为BP神经网络的输入端样本，通过隐含层的激励函数，进行神经网络训练，得到输出的故障类型，选取测试样本进行网络测试，结果达到了对测试样本的有效识别，这种BP神经网络训练对故障类型的识别是简单可行的。最后，利用Visual Basic程序对风力发电机振动检测故障诊断系统界面进行设计，可分为两大系统，即用户系统和管理员系统。用户系统可对振动信号采集、分析，提取有用信号，进行故障类型的判别，有效的输出诊断结果。管理员系统可对Access数据库进行查看、删除和打印最后结果。利用本文开发的风力发电机振动故障检测诊断系统，可简单的预测设备故障的发生和发展，有效降低了经济损失，并对下一步成熟的风力机故障诊断系统有一定的参考价值和指导借鉴作用。
[Abstract]:The installation position of wind turbine is high and the working conditions are bad.The failure is difficult to maintain, which can effectively, timely and accurately predict the occurrence and development of the fault, avoid and reduce the occurrence of the fault, and reduce the maintenance cost in view of the running state of the wind turbine.For the wind turbine, the highest failure rate is the generator, so starting from the generator to carry out the fault diagnosis study, generator structure complexity, and to ensure that no removal,Without destroying the equipment and ensuring that the equipment is in normal operation, it can accurately judge the working state, performance, parameter change and fault trend, which requires a set of state detection and fault diagnosis system with perfect function and stable performance.Since the generator belongs to the rotating machinery, the rotating machinery will produce vibration when it is running at high speed. Using the real-time monitoring of the vibration signal, the system can recognize the state from the vibration signal.In such a wind turbine fault diagnosis system, first of all, the sensor should be installed as close as possible to the vibration source without hindering the operation of the machine, and the measuring points and probe positions should be selected from multiple angles and directions.The vibration displacement of the common faults (normal, rotor misalignment, axial crack and oil film oscillation) of the running generator is measured, and the vibration signal of the fault is simulated.Secondly, MATLAB wavelet analysis is used to analyze the measured vibration signal, the signal is separated by wavelet packet, the useless noise signal is separated, the useful signal is preserved, the energy spectrum of the wavelet packet is decomposed, and the total energy percentage of each frequency band is calculated.As a characteristic parameter in fault diagnosis.Thirdly, the energy eigenvalues extracted from each frequency band of MATLAB wavelet packet de-noising are used as input samples of BP neural network. Through the excitation function of hidden layer, the neural network is trained and the output fault type is obtained.The test samples are selected for network test and the results show that the test samples can be effectively identified. The BP neural network training is simple and feasible for fault type identification.Finally, the interface of wind turbine vibration detection and fault diagnosis system is designed by using Visual Basic program, which can be divided into two systems: user system and administrator system.The user system can collect, analyze and extract useful signals for vibration signals, distinguish the types of faults, and output the diagnosis results effectively.The administrator system can view, delete and print the final results of the Access database.The wind turbine vibration fault detection and diagnosis system developed in this paper can simply predict the occurrence and development of the equipment faults and effectively reduce the economic losses.And the next mature wind turbine fault diagnosis system has a certain reference value and guidance.
【学位授予单位】：兰州交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM315

【相似文献】