Development of a Sensor Information Integrated Expert System for Optimizing Die Polishing

This paper presents a polishing expert system integrated with sensor information which can modify the polishing sequence and conditions initially set by the system using the on-site polishing status. A practical system using AE sensors is developed for the rotational polishing and the curved surface polishing. A database and knowledge base for polishing processes are established by using the results of experiments and also expert’s experience. Evaluations are performed for a die of headlight lamp by using both the sensor integrated expert system and the expert system without sensor. The results show that the sensor integrated expert system provides more optimal polishing conditions since the proposed system takes advantage of the on-line sensor information. INTRODUCTION Die polishing technology is very critical to determine quality and performance of the final products and occupies up to 30~50% of the whole die manufacturing time. Die polishing processes have not been automated as much as other die machining processes because the automation requires a great deal of experience and skill of experts and its standardization is very hard. The most difficult problem for automation is how to determine of the optimal polishing sequence and conditions. 2 Several studies on the decision making for polishing sequence and conditions using an expert system in the polishing process have been reported. 3-5 In general, these studies built up a specific database composed of experimental data and working knowledge, and showed that a set of polishing conditions could be obtained from the database. However, the off-line decision inferred by the expert system is confined to a few limited situations due to lack of on-site information, and becomes inappropriate if some on-site conditions such as polishing direction, polishing pressure, and surface status, etc. are slightly changed. In reality, experts often change their decision on polishing sequence and conditions in order to adapt the confronting circumstances, although they have prepared a set of polishing sequence supposed to be optimal in the setup stage. That is one advantage of the human polishing compared with the mechanically automated polishing. In this study, Acoustic Emission (AE) sensor is used to substitute for human finger and a sensor information integrated expert system is developed to achieve die polishing in a globally optimal way. The polishing sequence and conditions obtained from the off-line expert system can be modified according to the current polishing status which is identified by the feedback signal from the AE sensor. And the performance of the proposed system is compared with the result obtained by the expert system without sensor information. DIE POLISHING EXPERT SYSTEM Expertise of Polishing Process A skilled worker plans an optimal die polishing sequence based on his/her experience to get the required die surface condition from a pre-machined die that has lots of cusps and scratches on its surface. At first, a polishing tool of rather coarse mesh is selected and polishing is performed until the first tool is no longer effective to make the surface smoother. Then the next polishing tool of fine mesh is used and polishing is done again. Such processes are repeated until the target surface status is achieved. Figure 1 shows how to get a target surface roughness from an initial surface roughness by changing the polishing tools at some appropriate polishing intervals. The optimal interval for changing polishing tools depends on the operator’s judgement, which necessarily needs not only his/her experience and knowledge but also such sensory organisms as eye and finger. This is an important reason why it is difficult to fully automate the polishing process. Su rf ac e R ou gh ne ss Ri Number of Polishing Passes N1 N2 N3 Polishing Tool 2 Polishing Tool 1