An Efficient Hybrid Meta-heuristic Algorithm for Solving Capacitated Vehicle Routing Problem

Abstract

Vehicle routing inside factories is one of the hard problems that researchers try to solve for many years. When planning routes, we must think about how much vehicles can carry and how factory buildings are organized. Some factories have same type vehicles while others have different types with varying capacities. Researchers made good algorithms for this problem, but these algorithms need too much computer power. In our study, we made a new algorithm that uses adaptive memory to remember good solutions and selectively explores promising regions of the solution space. When we compare with old methods, our algorithm finds the same optimal solutions but needs about 80 percent less calculations. For testing our algorithm, we used real data from a car factory with both same type vehicles and different type vehicles. We tested five different scenarios and ran each test 30 times, performing comprehensive statistical analyses. All tests showed 100 percent success rate in finding optimal solutions with remarkable computational efficiency. Test results show us something important: We don't need to look at all possible solutions to find the best one. If we look at only promising areas, we can find best solution faster. This makes our method very useful for real factory problems because factory managers need quick solutions and don't want to use too much computer power. Our method is good at finding which solution areas are promising and focuses on these areas, so it solves problems faster with less computer resources.