Integrating Improved Genetic Algorithm and Network Simulation-Based Procedure to Optimize Risk Interaction Network

Risk interaction network (RIN), a system comprised of risk (nodes) and triggering relations (edges) between risks, changes the likelihood of occurring and also the impact of a risk. An issue of optimizing RIN, allowing for the risk response actions of weakening risk interactions, departs from the conventional project risk management focusing on independent risks. However, not all the potential response actions have been considered in the literature, partially because this can be overwhelming to the existing heuristics. In response, a simulation optimization model concerning each action is first constructed along with improving genetic algorithm (GA) within a novel operator-proportion mapping crossover (PMC). PMC solves the problem of gene value definition in risk interaction context, capable of maintaining more genetic information and ensuring the generated offsprings’ legality. Besides, by integrating RIN structure analysis and the RIN simulation model, this work customizes a novel network simulation-based procedure (NS procedure) for solution generation to increase the matching degree between differing chromosome segments. i.e., two kinds of response actions. The comparative analysis in a series of computational experiments demonstrates the effectiveness and efficiency of PMC-GA with the NS procedure, especially in solving complex and large-scale problems