Research And Development Of Cost Estimation System For Injection Mold

Mold quotation is a comprehensive reflection of technology, economy, efficiency and management of mold manufacturers, and it is also an important economic condition that determines whether manufacturers and customers could make a deal. With the fast development of the manufacturing industry of molds, quotation relied on individual experience obviously can not meet the needs of market competition and changes. It has become a common concern of manufacturers and customers to develop a set of mold quotation system which can not only reflect the production capacity of the manufacturers, but also satisfy the demands of consumers in accuracy timely.Injection mold is the research object in this paper. Firstly, it makes a detailed analysis of injection mold's cost structure and the factors that affect injection mold's cost. On this basis and combined with home and overseas research actuality, this paper has utilized methods of man-hour, BP neural network and CBR to develop a cost estimation system for injection mold, providing a potent and viable way for instant and precise injection mold quotation.In consideration of the characteristics of cost estimation for injection mold, cost estimation method based on BP neural network is proposed and a three-layer BP neural network model for cost estimation is constructed. Based on cost analysis and reference of the method of man-hour technical parameters, 10 cost factors are selected as the input of neural network. The optimization of BP network's structure is introduced by giving methods of pretreatment of sample data and determination methods of learning rate, the momentum coefficient, the number of hidden layer nodes and etc, which help accelerate convergence and improve accuracy.Case Based Reasoning is applied to estimate the cost of injection mold, and the system's CBR model for cost estimation is also constructed in this paper. Case representation is given by using the method of frame representation. Two retrieval models are established for retrieving similar cases, and one is based on BPNN, the other based on the method of NN (nearest neighbor). An algorithm of revising the weight dynamically is also proposed. The methods of interactive adaptation, rule-based adaptation and exponential smoothing adaptation are given to provide a complete set of methods for case adaptation and it realizes the integration of these adaptation methods for case adaptation. A case learning strategy is proposed to ensure the quality of new case which enters into case base and increase the efficiency of the system.Based on theoretical study mentioned above, three types of cost estimation methods have been explained respectively with three examples of injection mold.