由於環境的多變,塑膠射出的產品變化脈動非常頻繁,一般台灣製造業所面臨的問題為:產品成本難以預估、交期急迫、生產計劃經常變更……等問題,在傳統的塑膠射出成型產業中也會面臨相同的問題,塑膠射出成型業者要掌握市場趨勢並有效回應,加強對顧客的服務,以改善過去產業、產品及市場過度集中的情況。在經營策略上,如何將產品研發管理轉變為企業競爭力,以適時做有效之成本預估,快速回應市場是重要的一環。 塑膠射出產業是由許多不同stage的生產製造所構成,強調分工,以塑膠射出產品的製程來說,首先必須由產品研發設計人員設計產品,然後委由模具廠製造模具,然後再由專業的射出成型廠或廠內的射出設備進行塑膠射出製程,由組裝廠組裝、封裝,最後再進行半成品的檢核,才有完成品提供銷售。由於塑膠射出產品的過程繁瑣,如何提供顧客更快速的回應如新產品的設計、原料的採購、外包、品質控制、新產品的價格等就格外重要。 本研究主要目的在運用類神經網路結合粒子群演算法,解決塑膠射出成型產業之新產品成本預估的問題,研究中運用粒子群演算法解決類神經網路參數設定的問題,找出最佳的網路參數,以下簡稱為粒子群類神經網路(Particle Swarm Optimization-Back-Propagation,PSO-BP ),PSO-BP的建置就是要精確的預估塑膠射出成型產業的產品成本,改善整個成本預估的正確性,適時提供顧客正確迅速的資訊。
Injection plastic products are subject to frequent variability due to the changeable environment, normally, manufacturers in Taiwan are faced with: unpredictable product cost, urgent delivery and variable production planning, etc. Traditional plastic injection moldings are also confronted with likewise problems, so plastic injection molding practitioners have to quickly capture market trends so as to make effective response, strengthen customer services, and improve the traditional over concentration in market, production and product. In terms of business strategies, it is an important process to convert product research and development management into competence for timely and effective cost estimation in rapid response to the market. With a stress on labor division, plastic injection is composed of multi-stage manufacturing, take injection plastic product manufacturing process for example, firstly a product design has to be drawn by the product research and design staff, then, molds will be provided by mould manufacturers for specialized injection plants or injection equipments within the factory to carry out the plastic injection process, next, assembly and packaging will be done by assemblers, and finally semi-finished products will be inspected to provide finished products for sale. Making rapid response to customer needs in terms of new product design, material purchase, outsourcing, quality control, new product pricing seems especially important due to the complex procedures for injection plastic product. This study focuses on the integration of neutral network and particle swarm algorithm to solve the problems concerned with cost estimation for new products for the plastic injection molding industry, where particle swarm optimization was used to solve the problems concerned with parameter setting for the neural network in order find optimum network parameters. The utilization of Particle Swarm Optimization-Back-Propagation (PSO-BP) was aimed at precise estimation of production cost for plastic injection molding, improvement on overall production estimation accuracy, and offering correct and prompt information to customers. PSO-BP was used in this study to strengthen plastic cost control to enhance customer services, make quick response, reduce production cost and enhance enterprise competence.