Abstract: This paper discusses both the applied path of operations research on intelligence and the practice driven academic path, based on the development, implementation and maintenance of a referred decision support system that has been successfully deployed to automobile outbound logistics. The system is so far a pioneering intelligent dispatching system for automobile logistics company in China, and the corresponding thoughts, theories, methodologies and technologies demonstrate the key value of the discipline of operations research in the promotion of intelligent applications and the significance of practice in stimulating academic development, and so forth provide referred systematic approach for tackling bottleneck problems. This paper proposes a "Three Stages and Seven Steps" framework for the application of operations research on intelligent research and development. Under the framework, the paper firstly addresses the characteristics of intelligent application related to operational research, and particularly addresses intelligent scheduling decision requirements of automotive logistics and its developing trends and bottlenecks. Secondly, the paper discusses the roles of systematic model and related model building methods, and further identifying the scientific problems occurring in automotive outbound logistics by analyzing its decisional factors, objectives and constraints. Moreover, the new scientific problems so called "pattern constrained bin-packing" are proposed with computational intractability, solvability and key scientific properties. Furthermore, the paper establishes mixed integer linear programming models for practical and theoretical problems, respectively, and develops branching and bound algorithm. In addition, the paper also addresses the technologies and methodologies for time-space decomposition and rolling solutions of large-scale problems, and further proposes the production testing based on real data and stress testing method for the applications of operations research, and shows the results and testing analysis for outbound automobile logistics scheduling. In addition, this paper proposes a distributed, multi-view, multi-system integration intelligent scheduling decision support system, which is deeply integrated with automobile transportation management system and warehouse management system, driven by optimization algorithm engine. Finally, we introduce detail system implementations with deployment, promotions and maintenance, and briefly address related practice-driven scientific research outputs and future directions.

Key words: intellectualize, decision support system, automotive logistics, vehicle outbound logistics, scheduling, pattern constrained bin-packing