基于ANFIS和SOM的抽水蓄能电站主变压器健康状态监测

针对抽水蓄能电站主变压器运行工况多变、状态信息复杂,难以实现健康状态监测的问题,提出基于自适应模糊推理系统(ANFIS)和自组织映射(SOM)的主变压器健康状态监测方法。该方法充分考虑主变压器运行工况多变,健康状态特征的高维和非线性特点,采用ANFIS消除运行工况对单状态特征参数异常监测精度的影响,并在此基础上利用SOM建立多特征融合的健康状态监测模型,采用最小量化误差(MQE)作为变压器健康状态监测指标。将该方法用于某抽水蓄能电站主变压器的健康状态监测,发现该方法可提前监测到主变压器健康状态的衰退,从而实现故障的早期预警,避免严重故障发生,并为运行和检修决策提供了依据。

Health Condition Monitoring of Main Transformer in Pumped Storage Power Station Based on Adaptive Neural Fuzzy Inference System and Self-organizing Map

Aiming at the problem that the operational condition and condition information of main transformer in pumped storage power station are complex and difficult to monitor health condition, a health condition monitoring method is proposed based on adaptive neural fuzzy inference system (ANFIS) and self-organizing maps (SOM). The complexity of operational condition, high-dimension and nonlinear of state features are fully considered. By using ANIFIS, the impact of operational condition on single parameter condition monitoring is weakened. Then the multi-feature fusion health condition monitoring model is established based on SOM. In the model, minimum quantization error (MQE) is used as the index of health condition monitoring. The method is applied to monitor the health condition of a main transformer in pumped storage power station. It is proved that this method can detect the health degradation before failure, and can be used for early forecasting potential failure. The evaluation results can be used as a support for operation and maintenance decisions.