Estimation of operation mode from the clamping force variation of injection molding machine (Application of big data from the injection molding machine under a production)

Abstract

This article suggests the machine operation mode can be estimated from sensors attached to the machine in the production line. An injection molding machine (IMM), which is widely operated in the plastics industry, was used in this research within an actual production line. Four strain sensors were attached to the IMM to measure the clamping force in each production cycle. Auto or semi-auto mode was estimated from the clamping force sequence. Auto mode is continuous production by the machine itself, while semi-auto mode involves intermittent human operation and includes such functions as machine start, machine finish, and trouble shooting. Approximately 340000 cycle data were obtained in 10-week production. The changing value was calculated by the binomial filter statistical technique, as well as the differencing of time series data from the clamping force in each cycle. And the relationship between changing value and operation mode is evaluated by the area under the curve (AUC) calculated from the receiver operating characteristic curve (ROC curve). As a result, we obtained that AUC value was 0.86 from 10-week production data, which means operation mode can be estimated from the changing value of clamping force in a high level. This estimating method has a low computational cost; therefore it is easy to install on a machine's board control system. We expect that this method will be used to detect the defective product, the machine malfunction or the mold failure.