Abstract: For the past years, China's energy structure has been significantly optimized, and the installed capacity of renewable energy generation has steadily expanded. Meanwhile, the uncertainty of renewable energy output poses challenges to the operation and scheduling of the power system. Thus, effectively describing theuncertainty of renewable energy output and optimizing the complementary operation of multi-energy power systems are currently a research hotspot. The complementary operation of multi-energy power systems can be divided into short-term optimization scheduling and medium to long-term optimization scheduling on a time scale. Wherein, when load and renewable energy time series curves are used in long-term scheduling models, the computational burden is significant for larger systems, and may even lead to the problem of “NP hard”.
Besides, establishing the renewable portfolio standard (RPS) and the tradable green certificate (TGC) trading system is an important way for China to promote the development of renewable energy and energy system reform. RPS requires a certain proportion of renewable energy generation in the regional power grid, and there is renewable energy electricity that is comparable to the quota ratio that can be traded between different regions. TGC is a transaction voucher issued by the National Renewable Energy Information Management Center, and the power grid company can obtain 1 TGC for every 1MWh of renewable energy purchased. At present, some scholars have studied scheduling of power systems a day ahead under the implementation of RPS. However, since China's power market currently mainly adopts the form of medium and long-term contracts, it is necessary to study the impact of RPS on medium and long-term scheduling.
Therefore, this paper analyzes the impact of RPS implementation on the scheduling model and constructs an annual scheduling model for a multi-energy power system considering renewable energy quotas based on the typical scenario method. The transaction cost of TGC is added to the optimization objective of the annual scheduling model, and the RPS constraint is added to the constraint conditions of the model. Herein, the load-wind power correlation scenarios in the annual scheduling model are generated by an improved k-means algorithm. The specific content is as follows.
Firstly, an annual scheduling model for wind-water-thermal multi-energy power systems considering RPS is constructed. The optimization objective is to reduce the purchase cost of wind power, hydropower, and thermal power, as well as the transaction cost of TGC. The constraints include system power balance constraints, wind power plant output constraints, thermal power unit operation constraints, as well as hydroelectric unit operation constraints, TGC transaction constraints, etc.
Secondly, an improved k-means algorithm that can automatically determine the number of clusters is proposed. By clustering the power curves obtained by concatenating the historical data curves of electricity load and wind power, typical scenarios considering load wind power correlation can be obtained. This is done because both the electricity load and wind power output have the characteristics of daily periodicity and seasonality. For the same region, there are correlations between load and wind power outputs due to the simultaneous influence of weather, temperature, and other factors. Therefore, it is necessary to concatenate the load of the same day in historical data with wind power data into a data dashed line, and then cluster the newly concatenated data curve to obtain a typical load wind power scenario that considers correlation.
Finally, this paper takes the annual scheduling of a power grid company as an example to verify the rationality and effectiveness of the constructed annual scheduling model, and analyzes the impact of renewable energy quota system on wind curtailment rate and electricity purchase cost. In this example, the power grid company needs to purchase electricity from 8 thermal power units, 6 hydroelectric units, and 1 wind power plant.
The simulation results show that the multi typical scenario has a better characterization effect on annual electricity consumption than a single typical day, with an average accuracy improvement of 12.7%. The typical scenario method, which considers the correlation between load and wind power, can more accurately reflect the actual situation of annual load and wind power, and the average scenario compliance rate increases by 20.25%. When the price of a TGC is higher than the price difference between wind power and thermal power, implementing RPS can increase the consumption of wind power, which has important reference value for determining the price of TGCs.

Key words: wind-hydro-thermal power system, renewable portfolio standard, annual dispatch mode

摘要： 通过分析配额制的实施对购电策略的影响,建立考虑可再生能源配额的风水火多能源电力系统的年度调度模型。为减小计算负担,将月度的负荷与风电用数个典型场景及其占比来表示。提出能够自动选择最佳聚类数的改进k-means算法,并应用于典型场景生成过程中。将同一天的负荷与风电的历史数据曲线拼接为一条功率曲线,并基于改进k-means算法进行聚类,得到考虑负荷与风电时序相关性的典型场景。仿真结果表明:(1)所提典型场景生成方法对年度电量刻画的精度平均提高了12.7%,由其所生成典型场景的平均符合率提高了20.25%,所提方法具有更好的年度电量刻画效果和更高的平均场景符合率;(2)当绿色证书的价格高于风电与火电价格差值时,实施配额制后能够提高风电的消纳量,这对绿色证书价格的确定具有重要参考价值。

关键词: 风水火多能源系统, 可再生能源配额制, 年度调度模型