Use of knowledge intensive CAD in small electrical family appliance industry

Abstract

Many local small electrical family appliance manufacturing companies have tried to shift their business from original equipment manufacturing (OEM) to original design manufacturing (ODM) in order to get away from the price war that induced by the neighborhood underdeveloping counties including Mainland, Thailand, Malaysia, and even South America countries. However, such business migration has been largely prohibited by their incompetency in product design and development. Futhermore, the developed product knowledge is unable to retain and reuse due to the high mobility of staffing. Through literature, it was noticed that many giant and multi-national enterprises are now going, or will go, through the deployment of knowledge management (KM) technology, to enhance their new product design and development processes and shorten their new product's time to market. It was also found that, as a common believe, the development of a KM system has to involve a huge capital investment, and the process is so demanding that almost all research studies done in the area were only confined to highly complex/technology originated products including aircraft and automobile. Up to this moment, literature on application of KM for the development of simple products likes small electrical family appliance does not exist. In order to break such common believe, the research study "Use of knowledge intensive CAD in small electrical family appliance industry" that financed by the Teaching Company Scheme under the Industry Technology Fund and a local manufacturing company "General Electrical Work Corporation Limited" was set up. The project aims to: (i) investigate how the Knowledge Intensive CAD (KIC) technology can be used as a vehicle to support the deployment of knowledge management to transfer explicit knowledge (historical data) to tacit knowledge and form a knowledge database for reuse, and (ii) to evident that KM can also be deployed by less complicated product manufacturers and used as a strategic tool to enhance their new product development capability. The argument of the research project is that through the use of a proper selected Artificial Intelligence (AI)/Artificial Neural Network (ANN) algorithm and the availability of an appropriated amount of legacy data, a knowledge database can be crystallized to predict the performance of a similar or even an entirely new design/style so that a lean and agile new product development process can be obtained. The proposed KIC methodology/roadmap composes of five phases: (i) Decision and selection of a KIC application, (ii) Problem dissociation and identification of attribute characteristics/properties, (iii) Selection of Al/ANN algorithms, (iv) Knowledge capitalization, and (v) Knowledge deployment. Based upon the proposed KIC development methodology/roadmap, a prototype KIC system for the design of a plastic toaster case (heating test) was developed and evaluated. Two investigations that included: (i) prediction of dedicated style with variable sizes, and (ii) prediction of an entirely new style from existing styles with the whole and divided data set input approaches were made. It was found that the temperature predictions made by the KIC prototyping system were well within the +20C and -10C design error limits and the throughput of processing an enquiry (prediction of a toaster case surface temperature and resultant thermal strain) could be done within eight hours. It was also evidenced that the KIC prototyping system is both capable the predictions of toaster case design of similar shape and entirely new style. The estimated time for the development of an additional KIC module would be around three months with an investment around HK$43,000 that is affordable by most manufacturing companies. The results of the project study concluded that the use of KIC to aid the product development of the small electrical family appliance industry is feasible and efficient whilst the missing gap in between the deployment of the entirely CAE approach and traditional experience dependent method can be bridged.