为应对可再生能源利用和用户负荷的不确定性，提出一种多时间尺度预测方法。预测过程分日前、日内滚动和实时3个阶段进行，时间尺度分别为1 h、15 min和5 min。首先，采用基于差值统计的预测方法完成气象参数的3个阶段预测；其次，在负荷预测的日前和日内阶段，建立了信号分解与机器学习相结合的回归预测模型，实时阶段建立了机器学习时间序列预测模型；接着，以测试集的预测精度指标为依据确定了日前和日内滚动阶段对典型日负荷的最佳预测方法；最后，将预测方法应用于典型日的能量预测，验证了方法的可行性。研究结果显示：3个阶段典型日气象参数预测结果的决定系数R²都在0.8以上；在日前和日内滚动阶段，多元负荷的预测任务应采用不同的信号分解方法，实时阶段负荷预测结果的R²值均超过0.9，平均绝对误差百分比(mean absolute percentage error, MAPE)接近0。

In order to cope with the uncertainty of renewable energy utilization and customer loads, a multi-timescale prediction method is proposed, where the prediction process is carried out in three phases: day-ahead, intra-day rolling and real-time, with timescales of 1 h, 15 min and 5 min, respectively. First, a prediction method based on difference statistics is used to accomplish the three stages of forecasting meteorological parameters. Second, a regression prediction model combining signal decomposition and machine learning is established for the day-ahead and intraday stages of load prediction, and a machine learning time series prediction model is established for the real-time stage. Next, the best prediction methods for typical daily loads in the day-ahead and intra-day rolling stages are determined based on the prediction accuracy metrics of the test set. Finally, the prediction method is applied to the energy forecast of typical days to verify the feasibility of the method. The results show that the determination coefficient R² of the prediction results of the meteorological parameters for a typical day in all three phases is above 0.8; in the day-ahead and intraday rolling phases, the prediction tasks of multivariate loads should be performed with different signal decomposition methods, and the R² of the load prediction results in the real-time phase is above 0.9, and the mean absolute percentage error (MAPE) is close to 0.

目的 鉴于传统配电网需求响应模型在调控负荷侧灵活可控资源方面存在局限性，特别是忽视了冰蓄冷空调这类独特且高效的资源，因此，致力于通过深度优化冰蓄冷空调这一负荷侧调控对象，以显著提升新型配电系统中可再生能源的利用率，并深入挖掘其低碳运行潜力。 方法 引入具有虚拟储能特征的冰蓄冷空调作为调控对象，面向光伏消纳和冰蓄冷空调群低碳需求响应，提出了一种新型配电系统日前-日内多时间尺度、多目标优化策略。首先，建立了冰蓄冷空调运行特性及低碳需求响应模型；其次，以供电公司低碳需求响应激励成本最小、光伏生产商利润额最大、空调用户用电费用最小为优化目标，构建了日前多目标优化模型，采用非支配排序遗传Ⅱ型算法(non-dominated sorting genetic algorithm Ⅱ，NSGA-Ⅱ)实现了模型的高效求解，并基于多维偏好分析线性规划法筛选出多目标优化模型的最优解；再次，为消除日前预测误差对模型结果的影响，进一步构建了日内滚动修正优化模型；最后，采用测试算例分析了所建模型的有效性。 结果 基于日内滚动修正后所得最优方案使得供电公司及空调用户的成本降低约20%，光伏生产商的利润额提升约3%。 结论 基于多时间尺度的多目标优化调控方法既保障了各方主体的利益，又消除了预测误差对模型结果的影响，为新型配电系统低碳运行提供了重要技术支持。

Objectives In view of the limitations of the traditional distribution network demand response model in regulating the flexible and controllable resources on the load side, especially ignoring the unique and efficient resources such as ice storage air conditioning systems. Therefore, this paper aims to deeply optimize the load-side control object such as ice storage air conditioning, so as to significantly improve the utilization rate of renewable energy in the new distribution system and deeply explore its low carbon operation potential. Methods Ice storage air conditioning with virtual energy storage characteristics is introduced as the control object. A new distribution system multi-time scale optimization strategy is proposed for photovoltaic consumption and low carbon demand response of ice storage air conditioning groups. Firstly, the operating characteristics and low carbon demand response model of ice storage air conditioning are established. Secondly, with the minimum incentive cost for low carbon demand response in power supply companies, the maximum profit for photovoltaic manufacturers, and the minimum electricity consumption cost for air conditioning users as optimization objectives, a multi-objective optimization model for the day ahead is constructed. The non-dominated sorting genetic algorithm Ⅱ (NSGA-Ⅱ) is used to efficiently solve the model, and the optimal solution for the multi-objective optimization model is selected based on the linear programming method with multi-dimensional preference analysis. Subsequently, in order to eliminate the impact of prediction errors on the model results, a daily rolling correction optimization model is further constructed. Finally, the effectiveness of the proposed model is analyzed using test cases. Results The optimal solution obtained after intra-day rolling correction results in a cost reduction of approximately 20% for both power supply companies and air conditioning users, as well as an increase in profit margins of about 3% for photovoltaic producers. Conclusions The multi-objective optimization and control method based on multiple time scales not only ensures the interests of all parties, but also eliminates the impact of prediction errors on model results, providing important technical support for low carbon operation of new distribution systems.