Integrating expert systems with dynamic programming in generation expansion planning

Interactive software developed for integrating engineering experience and judgment from the planning department with a powerful mathematical optimization method is described. The excessive size of the state space generated by conventional multidimensional dynamic programming is reduced to manageable proportions by rule-based procedures for implementing windows in state space and controls in policy state. Project frames describing generation options and state frames describing future conditions of the system are established and manipulated by rules. Dynamic programming simultaneously tracks a feasible set of suboptimal scenarios. The program is interactive and is written in Prolog with numerically intensive portions in C