针对综合能源系统中多种复杂的能量耦合关系和负荷预测误差的变化,提出了一种基于长短时记忆网络(LSTM)和多元负荷预测误差的动态时间尺度调度策略.该策略结合负荷预测的平均绝对百分比误差(MAPE)曲线与调度员所要求的误差限值,动态地选择每个月适合于各类能量的最佳时间尺度.根据日前调度计划结果,在日内每个时间尺度上对相关设备的出力进行选择性地调整以实现多能协同优化,充分发挥IES的综合优势.需要调整的设备主要是通过考虑日内调度中每个设备响应速度的差异来选择的.以某IES为算例,动态确定了每个月各能调度的时间尺度.此外,还对关键设备的出力计划进行了详细的分析,说明了该策略的必要性和优越性.“,”Considering the complex coupling of multiple energies and the varying load forecasting errors for an integrated energy system (IES), this study proposes a dynamic time-scale scheduling strategy based on long short-term memory (LSTM) and multiple load forecasting errors. This strategy dynamically selects a hybrid timescale which is suitable for a variety of energies for each month. This is obtained by combining the mean absolute percentage error (MAPE) curve of the load forecasting with the error restriction requirements of the dispatcher. Based on the day-ahead scheduling plan, the output of the partial equipment is selectively adjusted at each time-scale to realize multi-energy collaborative optimization and gives full play to the comprehensive advantages of the IES. This is achieved by considering the differences in the response speed for each piece of equipment within the intra-day scheduling. This study uses the IES as an example, and it dynamically determines the time scale of the energy monthly. In addition, this investigation presents a detailed analysis of the output plan of the key equipment to demonstrate the necessity and the advantages of the strategy.