Research On Two-sided Assembly Line Balancing Problem Based On Hybrid Colonial Competitive Algorithm

The two-sided assembly line balancing problem(TALBP) is mainly used in the cars, trucks, excavating machinery, loading machinery and other large-sized high-volume machinery. In contrast with the traditional one-sided assembly line, the two-sided assembly line can provide the following advantages: shorter line length, higher equipment utilization, less transportation of raw materials, less workers' movement cost, higher labor productivity. The assembly line can be classified into deterministic type and stochastic type based on the task time. It can also be categorized into single model, mixed models and multiple models according to the product models.The TALBP is a typical NP-hard problem, which has the characteristics of computational complexity, multiple constraints and multiple objectives. However, in the real-life production process, the widespread constraints include the elementary constraints and the additional constraints. The elementary constraints contain the operating directional constraints, precedence constraints, cycle time constraints and occurrence constraints. Besides the elementary constraints, the additional constraints include the positional constraints, zoning constraints(positive and negative zoning) and synchronous constraints. The common optimization goals include the optimal number of stations and mated-stations within certain cycle time, optimal cycle time under deterministic number of stations, maximal line efficiency, minimal smooth index and minimal the total cost of unit product. For the multi-objective optimization problem, the multi-objective weighted integration method and the fast non-dominated sorting method based on the Pareto concept are adopted.In this study, mathematical formulation of balancing multi-objective two-sided assembly line with multiple constraints is established. According to the characteristics of the problem, a novel hybrid colonial competitive algorithm(HCCA) is proposed, in which the empire initialization, empire assimilation, combined population method, empire update and colonial competition are designed. In the HCCA, the late acceptance hill-climbing is integrated to improve the results efficiently. Based on the HCCA, the single-objective HCCA is mainly used in the deterministic TALBP-I, stochastic single-model problem and mixed-model TALBP with one goal. The experimental results, compared with the ones computed by other algorithm and in current literature, validate the effectiveness of the proposed algorithm. The single-objective HCCA is adopted for the cost-oriented mixed-model TALBP in the real-life assembly of the passenger vehicle. The optimal task allocation scheme with lower cost is proposed. The multi-objective HCCA based on the Pareto concept optimize the time-oriented and cost-oriented problem simultaneously, such as deterministic single-model problem, the assembly of the engineering machinery chassis, stochastic single-model problem and deterministic mix-model problem. The better Pareto-optimal curve is obtained compared with the fast elitism non-dominated sorting genetic algorithm(NSGA-II).At last, the summary and outlook of the research are concluded from the algorithm and problem.