In this work, historical background of power generation by small hydro-power plants across the world and specially across the Iran with emphasis on small hydro-power plant utilization as recovery turbine in water tran...In this work, historical background of power generation by small hydro-power plants across the world and specially across the Iran with emphasis on small hydro-power plant utilization as recovery turbine in water transmission pipeline have been attended firstly, and then, three water transmission pipelines from Chah-nemie to Zahedan city, Shirindare dam to Bojnord city and Mokhtaran desert to Birjand city in Iran have been studied as case study samples. According to the sample pipeline characteristics and pipeline topography, reachable energy have been estimated; in the next step, pay attending to reachable energy, initial investment cost, total benefit of operating period, benefit to cost ratio and other economical parameters for small hydro-power plants in case study pipelines have been presented and generated power cost of the same amount via other resources compared to the hydro-power cost. At the end, in agreement with environmental advantages of small hydro-power plants, the methods of generated power utilization, proper solution for optimization of reachable energy in water transmission pipeline and substitution of pressure reducing valve by small hydro-power plant in the pipelines as a major solution for energy recovering in water pipelines have been proposed.展开更多
深度强化学习算法以数据为驱动,且不依赖具体模型,能有效应对虚拟电厂运营中的复杂性问题。然而,现有算法难以严格执行操作约束,在实际系统中的应用受到限制。为了克服这一问题,提出了一种基于深度强化学习的改进深度Q网络(improved dee...深度强化学习算法以数据为驱动,且不依赖具体模型,能有效应对虚拟电厂运营中的复杂性问题。然而,现有算法难以严格执行操作约束,在实际系统中的应用受到限制。为了克服这一问题,提出了一种基于深度强化学习的改进深度Q网络(improved deep Q-network,MDQN)算法。该算法将深度神经网络表达为混合整数规划公式,以确保在动作空间内严格执行所有操作约束,从而保证了所制定的调度在实际运行中的可行性。此外,还进行了敏感性分析,以灵活地调整超参数,为算法的优化提供了更大的灵活性。最后,通过对比实验验证了MDQN算法的优越性能。该算法为应对虚拟电厂运营中的复杂性问题提供了一种有效的解决方案。展开更多
文摘In this work, historical background of power generation by small hydro-power plants across the world and specially across the Iran with emphasis on small hydro-power plant utilization as recovery turbine in water transmission pipeline have been attended firstly, and then, three water transmission pipelines from Chah-nemie to Zahedan city, Shirindare dam to Bojnord city and Mokhtaran desert to Birjand city in Iran have been studied as case study samples. According to the sample pipeline characteristics and pipeline topography, reachable energy have been estimated; in the next step, pay attending to reachable energy, initial investment cost, total benefit of operating period, benefit to cost ratio and other economical parameters for small hydro-power plants in case study pipelines have been presented and generated power cost of the same amount via other resources compared to the hydro-power cost. At the end, in agreement with environmental advantages of small hydro-power plants, the methods of generated power utilization, proper solution for optimization of reachable energy in water transmission pipeline and substitution of pressure reducing valve by small hydro-power plant in the pipelines as a major solution for energy recovering in water pipelines have been proposed.
文摘深度强化学习算法以数据为驱动,且不依赖具体模型,能有效应对虚拟电厂运营中的复杂性问题。然而,现有算法难以严格执行操作约束,在实际系统中的应用受到限制。为了克服这一问题,提出了一种基于深度强化学习的改进深度Q网络(improved deep Q-network,MDQN)算法。该算法将深度神经网络表达为混合整数规划公式,以确保在动作空间内严格执行所有操作约束,从而保证了所制定的调度在实际运行中的可行性。此外,还进行了敏感性分析,以灵活地调整超参数,为算法的优化提供了更大的灵活性。最后,通过对比实验验证了MDQN算法的优越性能。该算法为应对虚拟电厂运营中的复杂性问题提供了一种有效的解决方案。