A High-Speed Generation Method of Goal-oriented Scenarios Using Bidirectional Combination of Effect Propagation on Qualitative and Quantitative Hybrid Model

Abstract

This paper addresses a high-speed generation of goal-oriented scenarios on the qualitative and quantitative hybrid model. The goal-oriented scenarios can be generated efficiently by the bidirectional combination of the forward propagation and the backward propagation, if differences between values related to each scenario are small on a certain time step. So, the method combines the scenarios when each value can be regarded as the same from qualitative value viewpoints. Furthermore, the proposed method decides the number of both propagations to generate the required number of goal-oriented scenarios with decreasing the number of time-consuming backward propagations. Experimental results show that the computational time for generating the required number of the goal-oriented scenarios is decreased by 98% compared to the solely forward propagation.