Optimization of Large-Scale Dynamic Systems with Special Reference to Energy Resources

这项研究的总体目标是开发技术来制定和解决(优化)复杂系统的大规模模型，并鼓励它们的应用。最优化研究包括求解大规模、多阶段(阶梯、动态)线性规划模型的技术，其扩展以考虑不确定性，以及某些类型的非线性约束(大规模但稀疏)系统的求解，例如在平衡分析中出现的那些。这项研究包括用于评估资源稀缺或通过现代化、创新和计算机化改变技术的影响的复杂模型的方法论，以及用于复杂工程系统(如分配能源的系统)的长期规划、调度和控制的模型。关于建模的研究包括：如何建立大型复杂系统的模型；如何更好地表达动态模型中的结束条件；如何更紧凑地重新表述模型；如何使用高速采样方法(如并行处理器)来模拟随机事件，以制定表示对不确定性的对冲的约束；以及如何更好地理解、维护和更新用于指定模型的大量数据库和大量约束。这项研究的总体目标是开发用于制定和优化复杂系统的大规模模型的技术。将开发的优化技术将在计算机上实施，所产生的程序将可用于各种大型优化问题。

The overall objective of this research is to develop techniques for formulating and solving (optimizing) large-scale models of complex systems and to encourage their application. The research on optimization includes techniques for solving large-scale, multi-stage (staircase, dynamic) linear-programming models, their extension to take into account uncertainty, and the solving of certain classes of nonlinearly constrained (large-scale but sparse) systems such as those that arise in equilibrium analysis. The research includes methodology for complex models used to assess the effects of resource scarcity or changes technology through modernization, innovation, and computerization, and also models used for long-term planning, scheduling and control of complex engineering systems such as those that distribute energy. The research on modeling includes ways to formulate models of large complex systems; better ways to express end condition in dynamic models; ways to reformulate models more compactly; ways to use high-speed sampling methods (such as parallel processors) to simulate random events in order to formulate constraints that represent a hedge against uncertainty; and better ways to understand, maintain and update the extensive data bases and large number of constraints used to specify models. The overall objective of this research is to develop techniques for formulating and optimizing large-scale models of complex systems. The optimization techniques which will be developed will be implemented on the computer and the resulting programs will be available for use on a wide variety of large optimization problems.