Injection molds are made with the highest precision because they have to meet a variety of requirements. They are to some extent produced by greatly time- and cost- consuming procedures. So to make injection molds efficiently, processes have to be planed with estimated machining time before molds are manufactured. But in many shops the system for estimation of machining time, is not in the place to which it should be entitled. The machining time was estimated based on experience or in comparison with molds in the past by experienced workers, but either method has the problem of broad range of error.
This paper proposes a new method for estimation of machining time by neural networks. This system can improve the reliability of system as machined data of molds are added that is different from previous method.
The selection of input factors is very important for using this system for estimation of machining time by neural networks, therefore this paper recommends few general input factors and special factors for TV molds. Such input factors are divided into shape factors, machining condition factors and machining time factors.
The propriety of selected input factors is verified as the machining time of TV molds proposed by this paper is less estimated than error of actual process plan.
Using this system, processes of manufacturing molds will be more efficient by reducing range of error and consequently, improve the competitive power of molds industry.