Meteorological Vulnerability Assessment and Warning of On-load Tap Changer Based on Fuzzy-C Means-cloud Model-LSTM Network

To address the issue of severe damage caused by meteorological disasters and fault potential, a vulnerability assessment and early warning method for on load tap changer (OLTC) of transformer in the new power system was proposed based on improved cloud model and long short term memory network (LSTM). Firstly, the assessment system of disaster-causing factors was established based on the meteorological monitoring data of OLTC. According to the FCM clustering algorithm, the threshold division of the traditional cloud model was improved to obtain the objective cloud model, and consequently a combination cloud model was constructed by combining the subjective and objective cloud models. Based on the natural disaster theory, the disaster indicators to be evaluated are dynamically adjusted by considering the factors such as the geographical disaster pregnant environment, the disaster resistance ability of the equipment itself, the accumulation degree of disaster risk and the comprehensive risk processing ability facing the disaster. These modified indicators were calculated and utilized in the membership degree of the combined cloud model, aiming to acquire the disaster vulnerability level for OLTC. Finally, the LSTM neural network was applied to extract the association rules between each disaster causing factor and each disaster vulnerability, thereby facilitating meteorological disaster early warning and forming the optimal response strategy. The example results indicate that the OLTC meteorological disaster vulnerability assessment and early warning method proposed in this paper exhibits high accuracy and effectively achieves the objective of disaster prevention and reduction.