电力系统间歇性可再生能源发电功率概率预测理论及其应用研究

The intermittency and uncertainty of renewable energy generation have introduced severe challenges to the secure and economic operation of power systems. Accurate power forecasting is one of the key factors for the efficient integration of renewable energy. Due to the chaotic nature of weather systems, the forecasting errors of intermittent renewable energy generation mainly consisting of wind power and photovoltaic (PV) generation can be inevitable to some extent. Then traditional deterministic point forecasting technologies cannot meet the requirements of power system operation. This project will study probabilistic forecasting of renewable energy generation and its applications in power systems. A hybrid artificial intelligent point prediction approach considering the temporal and spatial correlations will be developed for renewable energy generation forecasting. Then adaptive parametric and nonparametric probability distribution models will be proposed to estimate the stochastic characteristics of renewable energy generation forecasting errors. With the utilization of the high flexibility of artificial intelligence, the artificial intelligence based combined parametric probabilistic forecasting approach will be developed to overcome the limits of traditional parametric probability models. Then the artificial intelligence based adaptive nonparametric probabilistic forecasting approaches will be proposed to avoid the parametric distribution assumption. The multi-model combination based probabilistic forecasting technology will be established to integrate the merits of different traditional probabilistic forecasting approaches. Multi-period and multi-objective coordinated optimization operation model of wind power, PV generation, storage and reserve will be constructed based on probabilistic forecasting of wind power and PV generation, which can mitigate the fluctuations of wind power and PV generation and realize the optimization of reserve setting. This project will produce theoretical importance and practice value to the integration of renewable energy and the secure and economic operation of power systems.

可再生能源的间歇性和随机性给电力系统安全运行带来极大挑战，可靠的功率预测是可再生能源并网消纳的关键因素之一。由于天气系统的混沌特性，以风电和光伏为代表的间歇性可再生能源的预测误差难以消除，传统的点预测技术难以满足电力系统的运行要求。本项目将深入研究间歇性可再生能源发电功率概率预测理论及其应用，开发考虑时空分布的混合智能点预测方法，构建自适应的参数及非参数预测误差概率统计模型；在此基础上，提出基于人工智能的混合参数化概率预测方法和自适应非参数概率预测方法；结合不同类型概率预测方法的优点，提出基于多模型组合的概率预测方法，有效量化可再生能源发电功率预测的不确定性；基于风电/光伏功率概率预测，构建多时间尺度多目标的风光储及系统备用协同优化运行模型，平抑风电与光伏出力的波动，实现电力系统的备用优化。本项目的研究对可再生能源消纳、电力系统安全经济运行和风险预防具有很强的理论意义和应用价值。

以风电、光伏为代表的可再生能源的间歇性和随机性给电力系统安全经济运行带来极大挑战，可靠的功率预测是可再生能源并网消纳的关键因素之一。由于天气系统的混沌特性，以风电和光伏为代表的间歇性可再生能源的预测误差难以消除，传统的点预测技术难以满足电力系统的运行要求。本项目深入研究间歇性可再生能源发电功率概率预测理论及其应用，旨在对可再生能源发电功率预测不确定性进行有效量化，并突破基于概率预测的电力系统运行优化难题。本项目取得的研究成果包括：1）提出了考虑时空分布的混合智能点预测方法，建立了自适应的参数及非参数预测误差概率统计模型；2）提出了基于人工智能的混合参数化概率预测方法和自适应非参数概率预测方法；结合不同类型概率预测方法的优点，提出了基于多模型组合的概率预测方法，有效量化了可再生能源发电功率预测的不确定性；3）基于风电/光伏功率概率预测，构建了多时间尺度多目标的风光储及系统备用协同优化运行模型，有效平抑风电与光伏出力的波动；提出了电力系统运行备用量化的概率预测-决策一体化方法，实现了概率预测与备用决策性能的协同优化。本项目的研究对可再生能源消纳、电力系统安全经济运行和风险预防具有很强的理论意义和应用价值。.基于本项目研究成果，获国家科技进步二等奖1项、省部级一等奖3项；出版学术专著1部；发表/录用论文53篇，其中SCI刊源论文34篇（电力系统国际顶刊 IEEE PES Trans.论文25篇），EI刊源论文18篇，1篇论文入选F5000，1篇论文入选《电力系统自动化》年度优秀论文一等奖；项目负责人参加国际学术会议5次，并受邀作大会报告，项目组成员作国际会议口头报告19人次，获国际会议最佳论文奖4次；培养博士研究生4名、硕士研究生9名；2名博士生赴国际知名高校进行联合培养；项目负责人入选浙江省杰出青年基金、中国科协“青年人才托举工程”、中国电力优秀青年科技人才奖，担任IEEE Trans. Industry Applications和IEEE Systems Journal编委、IEEE ISGT Asia 2019等5个国际会议技术委员会主席。