近年来，随着碳达峰碳中和成为全球趋势，能源系统正在经历结构调整和能源转型，分布式资源在电力系统中的占比快速提升。相比于传统可调节机组，分布式资源个体规模小、所有权分散、随机性强等特点导致其难以直接参与电网调控。针对这一问题，虚拟电厂作为连接分布式资源与大电网的中间媒介，成为近年来的研究热点。也正是因为其上述特殊定位，虚拟电厂面对着来自批发电力市场、分布式资源与其他虚拟电厂的多维不确定性，并需就分布式资源聚合、市场竞价等具有不同时间尺度的问题做出决策。本文就多维不确定性下虚拟电厂竞争性资源聚合与报价策略问题开展了系统研究，提出了虚拟电厂考虑风险的面向批发市场竞价与面向分布式资源聚合的行为模型与相应策略，主要工作包括：首先，针对虚拟电厂在新型电力系统中的角色定位，分析了虚拟电厂面临的多源不确定性来源与决策时序的相关性，提出了虚拟电厂与批发电力市场和分布式资源互动的决策框架。其次，针对虚拟电厂参与上层电网中批发市场的竞价问题，提出可计及不同风险偏好的虚拟电厂参与日前能量与备用市场的内部资源优化和竞价策略。基于现行电力市场竞价规则，通过条件风险价值和机会约束刻画虚拟电厂风险偏好，提出了风险调整下边际成本的虚拟电厂的竞价曲线生成方法。然后，针对虚拟电厂与下层分布式资源的聚合问题，提出静态博弈下计及有限分布式资源及其出力特性的虚拟电厂竞争性定价策略。分析了各类型分布式资源的出力特性和时空相关性，以及日前和实时阶段的不确定性对虚拟电厂聚合行为的影响，提出了在分布式资源有限的情况下多个虚拟电厂的竞争性定价问题，构建了对应的博弈模型并分析求解纳什均衡。最后，针对虚拟电厂与下层分布式资源的长期资源配置问题，提出重复博弈下计及有限分布式资源长期价值与市场竞争的虚拟电厂长期资源优化配置模型和定价策略。基于分布式资源的转移成本，定义了考虑物理特征与出力特性的分布式资源长期价值，设计了虚拟电厂长期竞争性聚合模型，提出深度强化学习和虚拟对手算法的虚拟电厂长期竞争性定价策略。通过以上研究，本文提出了计及多维不确定性的虚拟电厂面对上层电网与下层分布式资源的决策模型和相应策略，所提方法符合虚拟电厂运营需求，可提升虚拟电厂的经济性，促进分布式可再生能源消纳，助力实现碳中和。

In recent years, driven by the urgent need for peak carbon emission and carbon neutralization worldwide, the energy system is currently undergoing structural reformation and energy transition whereas the ratio of distributed energy resources (DERs) in the power system grows rapidly. Compared with traditional generators, system operators often face difficulties when they try to dispatch DERs directly. This is mainly caused by characteristics of DERs as they are small in individual capacities, owned by different entities and possess uncertainty. Targeting this problem, virtual power plants (VPPs), as intermediate between DERs and the bulk power system, have caught the eyes of researchers worldwide. Because the VPP plays a special role in the power system, it faces multi-dimensional uncertainties from the wholesale market, DERs and other VPPs. VPPs also need to make decisions for problems covering multiple time scales, such as determining the aggregation method of distributed resources or bidding strategies in wholesale markets. Systematic research is conducted in this dissertation to discuss the competitive resource aggregation and bidding method of VPPs under multi-dimensional uncertainties. Optimization models and corresponding strategies for the bidding problem in the wholesale market and the aggregation problem with DERs considering correlations among different layers in the hierarchical power system are presented. The main research content is summarized below:First, targeting the role of VPPs in the new power system, the sources for multi-dimensional uncertainties faced by VPPs and their correlations under different time scales are analyzed, the decision-making framework and the risk-adjusted optimization models for interacting with the bulk power system and DERs are proposed.Second, bidding and dispatching strategies of VPPs in the day-ahead energy and reserve markets that reflect heterogeneous risk attitudes of VPPs are proposed for coordinating with the upper-layer grid. Based on current electricity market mechanisms, the calculation method for bidding curves that precisely capture risk-adjusted marginal generation values of VPPs for the day-ahead energy market and reserve market is proposed. Risk attitudes of virtual power plants are characterized by conditional value-at-risk indices and chance constraints. Numerical tests verify the importance of individualized risk management for VPPs. Meanwhile, the proposed method ensures the maximal reward of VPPs for economically dispatching internal resources and bidding in energy and reserve markets.Third, in the aggregation stage, a competitive pricing strategy for VPPs is proposed that takes limited DERs with various characteristics into account. Based on the coupling relationship between markets of different time scales, output characteristics and spatial-temporal correlations of different DERs are analyzed, with focus paid on how uncertainties in the day-ahead and real-time stages influence the aggregation behavior of VPPs. The competitive pricing problem of VPPs with DERs as limited resources is proposed while the corresponding game model is constructed and the Nash equilibrium is analyzed.Finally, the aggregation problem is extended to consider the dynamic competition among VPPs. A long-term resource allocation model of VPPs along with related pricing strategies that consider long-term values of limited DERs and market competition is proposed. Based on switching costs of DERs between different VPPs, the long-term value of a DER to the VPP that reflects the DER’s physical characteristic and the previous contract status is defined. The long-term aggregation model for VPPs is thus designed and solved based on the deep deterministic policy gradient algorithm and the proposed fictitious adversary method. Based on the above research content, decision models and corresponding strategies of VPPs facing multi-dimensional uncertainties for their interaction with the upper layer bulk power system and DERs at the distributed side. The proposed methods fit the operational and commercial requirements of VPPs and shall enhance the efficiency of VPPs, promote the accommodation of renewable energy at the distributed side and therefore help achieving the goal of carbon neutralization.