Pulp concentration control by PID with BP neural network in the production of light weight cardboard

It is difficult for conventional optimal proportion integration differentiation (PID) controllers to obtain the optimal PID parameters to achieve the best operating position during papermaking process,because the parameters change greatly during papermaking process and the paper machine system is characterized with non-linear, time-varying and hysteresis qualities. Back propagation (BP) network can find the best PID parameters through online learning and adaptive processing. Combining BP network with PID controllers can make full use of both online learning ability of neural networks and the effectiveness of PID control. In this paper, neural network controller combining BP with PID is used for pulp concentration control in the production process of light weight cardboard. Using self-learning and adaptive functions of neural networks to make online real-time adjustment of PID parameters according to the actual working status online, the control system makes pulp concentration control in an optimal state, and ensures cardboard a uniform and stable basis weight.