面向多类型用户负荷的需求响应潜力量化评估

Quantitative Assessment of Demand Response Potential for Various Types of User Loads

随着新能源大量并网,电网局部时段“缺电”频繁出现,负荷侧需求响应(DR)潜力的准确评估成为电网调度部门制定DR策略的数据支撑,对DR实施过程中涉及的各方利益主体均有重要意义。基于非侵入负荷识别(NILM)提出一种面向居民用户多类型负荷的的需求响应潜力评估方法。首先,采用事件检测算法和卷积神经网络技术对用户负荷曲线进行监测与识别,学习和获取用户的用电习惯。然后,考量用户主观因素和市场价格等因素,建立潜力评估指标体系,量化需求侧负荷资源,挖掘零散负荷资源的利用潜力。最后,通过参考能量分解数据集(REDD)进行算例分析,结果表明本文所建模型可准确识别用户多类型负荷,所提出的评估体系可为智能电网的DR实施提供方法支撑。

With the large-scale integration of renewable energy sources and frequent occurrences of"power shortages"in specific time periods,accurate assessment of demand response(DR)potential on the load side is crucial for the grid operators to formulate DR strategies,which is of great significance for all stakeholders involved in the DR implementation process.The method for assessing demand response potential for multiple types of loads for residential users is proposed based on non-intrusive load monitoring(NILM).Firstly,event detection algorithms and convolutional neural network techniques are employed to monitor and identify user load curves,thus capture user’s electricity consumption patterns.Then,by considering subjective factors of users and market prices,a potential assessment indicator system is established to quantify demand-side load resources,explore the utilization potential of scattered load resources.Finally,case studies are conducted by using REDD database.It is shown that the proposed model can accurately identify multiple types of user loads,and the proposed evaluation system offers comprehensive support for DR programs.