基于竞争双深度Q学习的智能电表隐私保护与成本管理

智能电表能够实时采集、计算、存储和传输电力数据,对智能电网的运转起着关键性的作用.配备储能设备的智能家居是智能电表的一种重要的应用场景,它的发展面临隐私数据泄露隐患和高用电成本2个问题,需要研究两者的权衡优化策略.系统模型考虑了2种不同类型的储电设备,并建立了电表数据泄露和用电成本量化的权衡模型.考虑到传统深度强化学习存在过度估计和收敛慢的缺陷,提出一种基于竞争双深度Q学习的储能电器功率分配方法,实现了性能优化的目标.仿真结果表明,对比传统的深度Q学习和双深度Q学习方法,所提方法在隐私保护和成本控制2方面能获得更好的性能.

Privacy protection and cost management of smart meters based on dueling double deep Q-learning

Smart meters are able to collect, compute, store and transmit power data in time, which plays a critical role in the operation of smart grid. Smart home with energy storage devices is an important application of smart meters.Its development faces two problems:hidden danger of privacy data leakage and high power cost. It is necessary to study the trade-off optimization strategy between the two problems. For this, the system model considered two types of chargestorage devices, and established a trade-off model between meter data leakage and power consumption cost quantification. Then, considering the flaws of over-estimation and slow convergence of traditional deep Q-learning, this paper proposed a power allocation method for energy storage devices based on the dueling double deep Q-learning in order to achieve the target of performance optimization. The simulation results revealed that the proposed method achieved better performance in terms of privacy protection and cost management compared with traditional deep Q-learning and double deep Q-learning methods.