A novel modelling of glue allowance prediction for time-pressure dispensing system based on gated recurrent unit and fully connected neural network

Abstract

Aiming at the phenomenon that the glue output of the time-pressure pneumatic dispensing system decreases with the decrease of the glue allowance in the glue storage tube, this paper presents method of a glue allowance prediction for time-pressure dispensing systems. This method takes the gas pressure data sequence and the dispensing pressure value at the outlet of the solenoid valve of the time-pressure dispensing system during dispensing, and uses the deep neural network to predict the glue remaining value in the glue storage tube of the current dispensing system. Moreover, according to the nature of different input data, a network architecture combining Gated Recurrent Unit (GRU) and Fully Connected Neural Network (FCNN) is proposed, and two different neural networks are used to process temporal input data and non-temporal input data. This method solves the problem that the traditional glue dispensing system model and control method cannot obtain the glue residual value in real time. And through the measured data experiments, the algorithm is better than the traditional machine learning model in terms of root mean square error and mean absolute error performance indicators.

Keywords

• time-pressure dispensing
• gated recurrent unit (gru)
• glue allowance prediction
• deep learning
• smart manufacturing