CAJ | 학술논문

在智能电网中，储能设备是实现智能电网的重要基础之一。因此，如何对储能设备进行有效的管理是必需解决的关键问题。提出了基于Agent技术的自适应学习ZIPML算法调整存储策略，同时考虑多代理学习，使得实际存储策略逐渐逼近最优存储策略。通过模拟实验验证了该自适应学习算法的有效性，电网高峰期的负荷得到有效降低，用户可以节省约6.1%的用电花费，发电所产生的CO2则可以减少6.4%。
재지능전망중，저능설비시실현지능전망적중요기출지일。인차，여하대저능설비진행유효적관리시필수해결적관건문제。제출료기우Agent기술적자괄응학습ZIPML산법조정존저책략，동시고필다대이학습，사득실제존저책략축점핍근최우존저책략。통과모의실험험증료해자괄응학습산법적유효성，전망고봉기적부하득도유효강저，용호가이절성약6.1%적용전화비，발전소산생적CO2칙가이감소6.4%。
In the smart grid, storage devices management is one of the important basis of the realization of smart grid, so how to manage storage devices is a key problem. In this paper, it presents a novel adaptive learning ZIPML algorithm for management of storage devices, meanwhile the system permits multi-agent learning which leads to optimal storage strategy. The experimental results show that load factor during peak time reduced significantly, and it is possible for consumers to save up to 6.1 percent on electricity expenses.Moreover, emissions of carbon-dioxide from power generation processes can decrease by 6.4 percent.