Research On Forecasting Method Of Emergency Materials For Meteorological Disasters In Power Grid

Meteorological disasters in power grids have occurred frequently in recent years,causing huge economic losses and casualties to society.In August 2019,Typhoon"Lichma" and 2018 Typhoon "Mangosteen" invaded many eastern provinces such as Guangdong,Fujian,Zhejiang,Jiangsu,and Shandong,causing large-scale power outages and damage to a large number of power equipment,which poses a huge challenge to rush repair work.Emergency management of power grid emergencies has become an important research topic that academia is concerned about.Power grid emergencies have the characteristics of suddenness,complexity,and uncertainty,so traditional emergency management methods are not suitable for handling power grid emergencies.Most of the current researches focus on the construction of emergency management systems,countermeasures,and emergency technologies,and there is less research on the prediction of emergency personnel and emergency resources in power grid emergencies.Based on the premise of meteorological disasters,this paper studies the impact of various scenario characteristics on the emergency demand of power grid emergencies,and uses case reasoning as a guide to comprehensively use theories of case reasoning,regression analysis,principal component analysis,and neural networks to construct power grid emergencies.Emergency materials prediction model for emergency decision.First,this article studies the application of case-based reasoning in emergency decision-making for power grid emergencies.It analyzes the information on the meteorological hazards of the power grid at the time of the emergencies,integrates the level of equipment disaster resistance and terrain factors,and bases on the material needs of historical emergencies In this paper,the application of case-based reasoning method in power grid emergency materials prediction is studied.This paper proposes the application of Deep Confidence Network to the key joint of case adaptation in case inference,and constructs the input-output association relationship model.Secondly,it analyzes the impact of various characteristic factors on the demand for emergency supplies,and adopts different processing methods according to the characteristics of different factors to suit the case characteristics and model input.The regression analysis method is used to obtain the meteorological factors with cumulative effects to obtain data with higher correlation with material needs.The principal component analysis method is used to transform the equipment status data into data describing the vulnerability level of the equipment.Terrain factors were quantified using terrain fluctuation parameters,standard deviation of elevation,and local full curvature to improve the correlation between feature factors and material demand.Thirdly,the structure principle and characteristics of the deep confidence network are studied,and the feasibility of its application to electric power emergency material prediction is demonstrated.According to the characteristic factors processed by the above methods,a deep confidence network-based emergency material prediction model for power grid accidents is established.The unsupervised self-learning and supervised parameter adjustment of the neural network are used to establish the relationship between the characteristic factors and the material requirements to achieve disaster Material forecast in case of unclear prior or accident information.Under the premise of inaccurate disaster information,a prediction network of a combination of accident scale and characteristic factors is proposed,which not only provides a reference for the current material demand,but also reflects the material demand of the accident scale change,which improves the application of the method in the actual field.Finally,taking the historical case data of power grid emergency repair as an example,the relevant simulation examples are simulated on the MATLAB simulation software platform.The simulation results show that the case adaptation method based on the deep confidence network under the meteorological disaster conditions proposed in this paper can accurately and comprehensively establish the correlation between the characteristic factors,the scale of the accident and the power grid emergency supplies,providing a scientific reference for the power grid emergency.