Load Interval Prediction of Offshore Oil and Gas Planning Platform Based on KNN-KDE

Load forecasting of offshore oil and gas planning platform is an important basis for future distribution network planning. Accurate load forecasting helps to reduce investment and operating costs. Aiming at the problem that the planning platform cannot directly predict the load based on historical data due to the lack of historical data, a strategy based on historical data of similar platforms is proposed. Firstly, the load forecasting model of each electrical equipment on the existing platform is constructed based on the power consumption characteristics of the platform; Then, based on the equipment characteristics of the planning platform, KNN(K-nearest neighbor algorithm) is used to find similar equipment in the same type of equipment on the existing platform, and the similar equipment model and parameters are migrated to the planning platform as the load forecasting model of its own equipment; Then, aiming at the problem of the value of the parameters of the prediction model of each equipment of the platform, based on the parameter values in the similar equipment set, the KDE(kernel density estimation) is used to fit the value probability interval of the prediction model parameters, and the load interval prediction model of each equipment of the planning platform is formed;Finally, the overall load interval prediction model of the platform is formed. Taking the data of an offshore oil and gas field area as an example, the effectiveness of the proposed prediction model is verified.