Ships short-term power load prediction based on ProbSparse self-attention
Aiming at the problems of poor real-time perform-ance,small storage data and low quality of ship power load data prediction,a new short-term load prediction method,by combining data interpolation,wavelet threshold denoising and ProbSparse self-attention was proposed.Firstly,in data pre-processing stage,the database was expanded by using interpo-lation to complement the missing data,thereby meet the mod-el training requirements without affecting the characteristics and trends of the original data,at the same time,as consider-ing the noise disturbance in the original ship power load data,in order to reduce its impact on the model prediction effect,a new wavelet threshold denoising method was adopted to process the original signals to improve the quality of the data.Secondly,the ProbSparse self-attention was introduced into the prediction model to effectively capture the dependency re-lationships and important features in time-series power data,reduce memory resource occupation,and model complexity to meet the real-time requirements of ship power load prediction,and achieve dual optimization of prediction accuracy and effi-ciency.Compared with traditional Transformer model,ARIMA and LSTM models,the proposed model reduces root mean square error and average absolute percentage error by an aver-age of 13.1％and 18.6％,respectively,and improves effi-ciency by an average of over 24.0％,which indicates that the proposed method has significant advantages in the accuracy and efficiency of ship power load data prediction models.
ship power loadshort-term predictiondata pre-processingProbSparse self-attention mechanism
万方数据
维普
