【摘要】 目前对于风力机齿轮箱的预警主要是以分析振动信号和温度参数为主,但前者需要安装额外的传感器,成本较高,后者有一定的时间延迟。因此提出了基于风力机运行参数和BP神经网络的风力机齿轮箱运行状态评估与预警模型,对风力机全工况下的SCADA原始数据进行清洗和归一化处理,通过参数间相关性分析建立线下预测模型;采用滑动窗口模型计算正常运行状态下的评价指标,基于小概率事件假设,获得评价指标的阈值,实现线上监测和运行状态评估。并以某风电场1.5 MW风力机齿轮箱子系统故障为例验证该方法,结果表明:该预警模型可以实现对齿轮箱子系统运行异常状态的识别和早期预警,且不需要对异常数据或相关故障的先验知识进行挖掘和训练。更多还原

【Abstract】 At present, the early warning for wind turbine gearbox mainly relies on the analysis of vibration signals and temperature. However, the former requires additional sensors, which leads to high cost, while the latter has a certain delay. Therefore, this paper proposes a wind turbine gearbox operating state evaluation and early warning model of wind turbine gearbox based on operating parameters and BP neural network. The original SCADA data of the wind turbine under the whole operation condition are cleaned and normalized, and then the off-line fault prediction model of the wind turbine is established through the correlation analysis of parameters. The sliding window model is used to calculate the evaluation index under normal operation state, the threshold value of the evaluation index of the wind turbine is obtained based on the hypothesis of small probability events, and finally, the on-line monitoring and operation state assessment are realized. Taking 1.5 MW wind turbine gearbox as the example, the results of the study show that the prediction model can realize the recognition and early warning of the abnormal state of the wind turbine gearbox, and it does not need any mining and training of the abnormal data or the prior knowledge of related faults.更多还原