基因為底模糊邏輯控制集體生產規劃模式

Abstract

集體生產規劃(Aggregate Production Planning;APP)是在給定有限的規 劃幅度內，將工廠的人力及設備資源做最有效利用，以滿足市場需求的變 動。在有限資源共享的前提下，集體生產規劃需決定每一期的生產和雇用 的最佳水準。雖然，目前已有不少數學模式和啟發式(heuristic)的技巧 使用於各種特定的集體生產規劃策略中。但是，這些模式或方法必須依賴 精確的參數估計與函數關係。實際上，決策者常無法精確的估計這些參數 及給予精確的函數關係，因此規劃結果無法反應產業的實際狀況，致使管 理者對這些數學模式失去信心。因模糊邏輯不須精確的參數估計與函數關 係，因此，本研究應用模糊邏輯於集體生產規劃，並運用基因演算法( genetic algoritms;GA)來同時搜尋模糊邏輯控制(Fuzzy Logic Control; FLC)中的最佳隸屬函數和推論規則的組合，提出基因為底模糊邏輯控制集 體生產規劃模式(GA-FAPPM)，以反應生產規劃人員在實際從事規劃時所採 用的決策過程，進而改善傳統集體生產規劃技術的缺點，發展一套較能為 管理者接受且具彈性和實用性的集體生產規劃模式。 This research presents a fuzzy logic control(FLC) model ofaggregation production planning that use genetic algorithm(GA) to search the membership functions and inference fule set simultaneously. Generaly, the successful applications of fuzzylogic control depends on fuzzy rules, the structure of the rules,and membership function parameters. Usually, the sets of parametersof membership function determined subjectively. And the rule set of knowledge base is determined by experts of the domain area. It is shown in this research that a suitable set of parameters of membership functions and the rule set of knowledge base may be designed with the aid of genetic algorithm. In order to simplify the computationof the fuzzy logic control, the trapezoidal fuzzy number was adopted asthe shape of membership functions. It is shown that the genetic algorithmenable us to generate an optimal combination of the trapezoidal membershipfunctions and the rule set. The approach presented here is illustrated by using Holt's HMMS paint factory data. Comparison of results with Rinks andTurksen shows the proposed approach can be used to produce a favorable one.