Research On Planning And Control Methods For UGNL Based AGV System

As a modern integrated intelligent material handling system, automatic guided vehicle(AGV) system has a wide application prospect. In this paper, the key technologies of unidirectional guided-path network layout(UGNL) based AGV system are studied, including guided-path network design, transport task assignment and AGV dispatching, traffic management, et al. Main innovation achievements are as following:First, the advantages and disadvantages of present unidirectional guided-path network design models are analyzed. Afterwards, a new unidirectional guided-path network design model of AGV system for flexible job shop is established as well as an improved bi-group collaborative evolutionary genetic algorithm for the model. In this algorithm, there are two chromosome codes for unidirectional guided-path network and feasible processing/transport order. Two chromosome individuals respectively constitute path network sub population and process sequence sub population. Corresponding crossover operator and mutation operator are respectively designed for two sub populations. Two sub populations respectively evolve independently and concurrently. Only when individuals in a sub population need to be evaluated, it interacts with the other sub population. In the evolution process of path network sub population, neighborhood search operation is adopted to improve the rate of convergence of the algorithm. To maintain population diversity, niche technology and elitism retention strategy are comprehensively applied. Finally, the superiority of the designed model and the algorithm is proved by experimental comparison.Second, for the problem of task assignment and AGV dispatching for flexible job shop, a flexible job shop model comprehensively considering processing subsystem and transport subsystem is established by dividing the problem into two sub problems – work station allocation and AGV task dispatching, an online dynamic dispatching method is also proposed. With minimization of maximum load of machine tool, imbalance rate of load of machine tool and the load of transport subsystem as objectives, an online dynamic allocation method including the method of exhaustion and niche genetic algorithm is proposed to address work station allocation sub problem. In the niche genetic algorithm, neighborhood search is adopted to improve the rate of convergence of the algorithm, and niche elimination operation based on Hamming distance is applied to maintain population diversity. To solve the sub problem of AGV task dispatching, a real-time multi-attribute task dispatching method is proposed. To ensure the efficiency of processing subsystem, a transport task urgency evaluation method based on input-output buffer state is designed, and deadlock avoidance strategy is applied to avoid deadlock of the system caused by inadequate surplus capacity of input-output buffer; To ensure the efficiency of transport subsystem, a task dispatching method with minimization of unload travel distance is put forward. Finally, the simulation experiment validates the dynamic dispatching method.Then, a collision prevention method to avoid AGV collision for UGNL based AGV system is advanced in this paper. An AGV system operating state model based on directed graph is built with a cycle-deadlock search method for this model. The concept of cycle-deadlock critical state including single-loop and multi-cycle-deadlock critical state is defined, and corresponding deadlock control strategy is designed in allusion to two deadlock critical states. The proposed model and method are proved valid by simulation.Based on theoretical research mentioned above, some experiments are done by the AGV system of laboratory. Firstly, AGV system experimental platform including multiple high-level control dispatch units and vehicle-mounted planner is developed. High-level control dispatch units mainly include global map manager unit, monitor unit, manual command issuing unit, task dispatching unit and traffic management unit. Vehicle-mounted planner based on ARM embedded system is integrated in each AGV to realize autonomous navigation function based on unidirectional guided-path network. The theories and methods advanced in this paper are verified through the experimental platform developed. Furthermore, these theories and methods are applied to an AGV system for the electricity meters intensified verification system in certain province, which proves the feasibility and effectiveness of the proposed theories in further.Finally, the main research results are summarized. The difficult problems and valuable directions for further researching are pointed out, which could be solved and improved in future studies.