A Machine Learning Approach for Energy-Efficient Intelligent Transportation Scheduling Problem in a Real-World Dynamic Circumstances
Jianhui Mou,Kaizhou Gao,Peiyong Duan,Junqing Li,Akhil Garg,Rohit Sharma
DOI: https://doi.org/10.1109/tits.2022.3183215
IF: 8.5
2022-01-01
IEEE Transactions on Intelligent Transportation Systems
Abstract:This paper provides a novel intelligent scheduling strategy for a real-world transportation dynamic scheduling case from an engine workshop of general motor company (GMEW), which is a key production line throughout the manufacturing process. In order to reduce the carbon emission in the scheduling process and make up for ignoring the energy consumption of each part in the scheduling when optimizing the carbon emission of the workshop and the factory. This paper first formulates a fuzzy random chance-constrained programming model of inverse scheduling problem (ISP) with energy consumption. A multi-strategy parallel genetic algorithm based on machine learning (RL-MSPGA) is proposed, which uses machine learning to improve the genetic algorithm. First, the parallel idea is developed to accelerate the process of evolution of genetic algorithm, and the initial population is divided into clusters by $k$ -means clustering algorithm. Second, similar individuals are evenly distributed to different sub-populations to ensure the diversity and uniformity of sub-populations. Third, in the process of evolution, the sub-populations communicate with each other, and extend the excellent individuals to replace the poor ones in other populations, so as to improve the overall quality of the population. Fourth, the self-learning of the crossover probability is realized by the self-learning of the self-sensing environment, which makes the crossover probability adapt to the evolutionary process according to experience. Finally, the real instance is used to validate the different algorithms. It can effectively adjust the completion time and the proportion of energy consumption, thus providing the possibility for the production of energy-saving enterprises. This implies that the suggested model is reasonable and the provided algorithm can effectively solve the inverse shop scheduling problem.
engineering, electrical & electronic,transportation science & technology, civil