| 引用本文: | 余涛,张水平.基于5要素试错更新算法SARSA(λ)的自动发电控制[J].控制理论与应用,2013,30(10):1246~1251.[点击复制] |
| YU Tao,ZHANG Shui-ping.Automatic control of electricity generation based on 5-component update learning algorithm SARSA(λ)[J].Control Theory and Technology,2013,30(10):1246~1251.[点击复制] |
|
| 基于5要素试错更新算法SARSA(λ)的自动发电控制 |
| Automatic control of electricity generation based on 5-component update learning algorithm SARSA(λ) |
| 摘要点击 2528 全文点击 2189 投稿时间：2012-07-11 修订日期：2013-06-04 |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| DOI编号 10.7641/CTA.2013.20761 |
| 2013,30(10):1246-1251 |
| 中文关键词 SARSA(λ)算法 自动发电控制 强化学习 控制性能标准(CPS) |
| 英文关键词 SARSA(λ) algorithm automatic generation control (AGC) reinforcement learning control performance standards (CPS) |
| 基金项目 国家高技术研究发展计划(“863”计划)资助项目(2012AA050209); 国家自然科学基金资助项目(51177051); 中央高校基本业务费重点资助项目(2012ZZ0020). |
|
| 中文摘要 |
| 本文提出了一种基于5要素试错更新算法SARSA(λ)强化学习的随机最优自动发电控制方法. 该方法不依赖任何系统模型和先验知识并通过试错机理寻求最优控制策略. 以控制性能标准(control performance standards, CPS)和区域控制偏差(areal control error, ACE)瞬时滚动值为基础设计了即时奖励函数, 有效提高了该方法的收敛速度和控制效果, 并在算法中融入了资格迹以解决二次调频过程的延时问题. 本文所提出的控制方法在进行状态空间搜索时, 能有效摆脱避免搜索较大扰动状态, 以此获得更佳的控制效果. 标准两区域和南方电网仿真模型研究表明, 本算法能给系统提供更加安全的控制策略, 具有比Q(λ)算法更好的控制性能, 有效提高CPS考核的合格率. |
| 英文摘要 |
| This paper presents a stochastic optimal control for electricity generation, based on the 5-component update algorithm SARSA(λ) in compliance with the control performance standard control performance standards (CPS). This method doesn’t need any knowledge about the system model but uses the trial-and-error method to find the most desirable policy. Instantaneous values of CPS1 and area control error (ACE) are used to construct the reward function which improves the convergence rate and the control performance. Eligibility trace is introduced to deal with the long time-delay problem in the secondary control of frequency. In searching the state space, the proposed method can effectively get rid of the disturbed states, thus obtaining better control results. Simulation studies based on two-area interconnected systems and China southern power grid demonstrate that SARSA(λ) controller has better control performance than that of Q(λ) controller, thus improving the stability and robustness of interconnected power systems. |