基于深度置信网络状态最优反馈的智能发电控制策略

针对自动发电控制问题中的状态信息测取和最优发电控制策略的难题,结合深度置信网络与最优控制理论,提出了一种应用于自动发电控制领域的深度置信网络状态最优反馈算法。首先,设计了一种适用于发电控制问题的全状态最优反馈控制策略,并引入了深度置信网络学习全状态最优反馈控制特性,以深度置信网络的非线性表达能力弥补传统线性控制的不足,进而实现了非全状态信息反馈下的近似最优发电控制,有效解决了最优发电控制问题中的状态信息测取难题,同时提升了自动发电控制性能。在IEEE标准两区域系统的算例仿真结果表明,所提算法可以实现由自区域频率偏差与传输功率偏差组合反馈下的系统近似最优发电控制,验证了所提方法的有效性、可行性和强鲁棒性。

Smart Generation Control Strategy Based on State Optimal Feedback of Deep Belief Network

According to the theory of deep belief network (DBN) and optimal control, this paper proposes a DBN state optimal feedback algorithm, which is applied to the field of automatic generation control. Firstly, a full-state optimal feedback control strategy suitable for power generation control problem is designed, and the deep-belief network is introduced to learn the characteristics of full-state optimal feedback control. The nonlinear expression ability of deep-belief network is used to make up for the deficiency of traditional linear control, thus the approximate optimal power generation control with incomplete state information feedback is realized, which effectively solves the problem of state information measurement in optimal generation control, and improves the performance of automatic generation control. The simulation results show that the the DBN state optimal feedback algorithm is able to realize the approximate optimal generation control under the combined feedback of self-area frequency deviation and transmission power deviation. The effectiveness, feasibility and strong robustness of the proposed algorithm are verified.