Online Prediction Model For Icing Load On Power Transmission Lines Based On Field Data-Driven

With the rapid development of power systems,the research of power grid disasters has become an important research field to ensure the safe and stable operation of power grids,especially in the analysis and prediction of ice disasters.Due to the continuous development of on-line monitoring technology of the power grid,a large amount of filed micro-meteorological information and icing load data during the icing process on the transmission line are recorded and saved in real time.In order to reduce the possible damage caused by the ice disasters to the power grid,a field data-driven online icing-load prediction model that provides prediction-based early warnings can guide the decision-making processes regarding the maintenance and deicing of power transformation and transmission systems,thus alleviating losses in power grids caused by the icing of transmission lines.Current machine learning models for the prediction of icing loads on transmission lines are afflicted by the following issues:insufficient prediction accuracy,high randomity in the selection of the kernel functions and model parameters,and a lack of generalizability.To address these issues,this paper proposed a field data-driven online prediction model for icing loads on transmission lines.The following research results have been achieved:(1)Based on the characteristics of dynamics,nonlinearity,uncertainty and abruptness of ice load time series,we combined with ensemble empirical mode decomposition,phase space reconstruction and support vector machine regression algorithm to proposed a multi-scale analysis icing load prediction model.Throughout the example verification,the model can decompose the original ice-covered load time series signal into a relatively stable waveform,reduce the non-stationary and sensitivity of the original icing load time series,so as to achieve the purpose of tracking icing load change and its law.(2)Aiming at the shortcomings of traditional single-time series forecasting model,such as poor prediction accuracy and randomness of parameter selection,an icing load forecasting model based on principal component analysis,genetic algorithm and least squares support vector machine(PCA-GA-LSSVM)was proposed,which has been applied to the actual transmission line ice load forecasting,and the results show that principal component analysis could extract effective information as the input of the model for eliminate the correlation of micro-meteorology.Genetic algorithm-optimized least squares support vector machines can enhance the learning and training capabilities of predictive models.(3)Aiming at the problem that the existing ice load forecasting model has insufficient generalization ability and no online learning ability,an accurate online support vector machine prediction model optimized by particle swarm optimization(PSO-AOSVR)is proposed and applied to the actual ice load forecasting.The results show that the good predictive performance and effectiveness of our proposed icing load forecasting model.and the PSO algorithm is useful to solving the AOSVR parameter optimization problem,which can help solve the issues caused by incomplete data.At the same time,the online learning ability of the algorithm can update the model to adapt the new real-time micro-meteorological data.