Research On Workshop Facilities Layout Problem Based On Estimation Of Distribution Algorithm

As the world economy faces the threat of economic depression with the obstruction in the resurgence process, the global and domestic markets suffer an extended downturn. With the pressure from expanding domestic demand and the export struggle, the competitions among domestic manufacturing enterprises get hotter than ever. To the modern manufacturing sector, how to reduce product cost is the crucial link in the competition to win. It is indicated by the practice that reasonable workshop facility layout with effectively internal logistics can greatly reduce the production cost. How to optimize and redesign the workshop facilities layout has become the focus of academic world and enterprises increasingly, also it is the original intention and significance of this dissertation.This dissertation takes the new workshop layout project commissioned by Guangdong D Company as the productive practice background, sponsored by NSFC. Considering the characteristic of workshop facilities layout problem is an NP-hard problem, a hybrid Estimation of Distribution Algorithm (named EDASDQA) is proposed. Through actual demand analysis, the multi-row layout model of the new workshop is built, and the optimization solutions of the model are given based on the EDASDQA. Furthermore, a set of systematic and practical method is introduced to carry out the workshop detailed layout procedurally. This dissertation is roughly composed by three parts:(1) An Estimation of Distribution Algorithm with Simplex Method and Diagonal Quadratic Approximation local strategies (EDASDQA) is proposed for solving the complex optimization problem. In the EDASDQA, the initial candidate solutions are generated by Tournament Selection, the new solutions are sampled from a fixed-width marginal histogram probability model which is built on the global statistical information extracted from the current population. The simplex method and diagonal quadratic approximation are employed as local search strategies. Through the experimental results on classic test functions, it is demonstrated that EDASDQA performs well in terms of the solution quality and the computational cost.(2) A Relative Position (RP) encoding strategy combined with EDASDQA is given for the complex workshop facilities layout problem. The purpose of the relative position encoding is to overcome the properties of traditional encoding strategy, such as existing invalid codes or falling the inconsistence between the coding space and the solution space of facilities layout problem. As the extension of classical facilities layout problem, loop layout is taken into account for confirming the effectiveness of with EDASDQA with RP encoding. The empirical studies on the Nearchou’s loop layout benchmark problems illustrate that the proposed method achieves better solutions and higher success ratio in the comparison to other heuristic approaches.(3) On the basis of the above hybrid EDA, the project of the new workshop layout in Guangdong D Company is given to elaborate how to implement the detailed workshop layout. The multi-row opertion area layout(MOAL) model is explored through the analysis of the overall plan, facilities details, product process, material flow and operation units’position relationships in the new workshop. Then the hybrid EDA is applied to numerically solve the MOAL model, and an available detailed workshop layout methodology is develped in view of practice applications. Finally, the detailed layout scheme of the new workshop is presented with a scheme evaluation.In theory, this dissertation forms an EDASDQA based complex workshop facilities layout method, as well as a detailed workshop layout methodology in engineering. It is expected that the research would provide technical support for practical workshop facilities layout, and be applied to the whole factory and other types of layout.