Level and flow systems identification of an industrial processes module by LSOD method for PID controllers design

Process control in an industrial environment primarily regulates flow, level, temperature, and pressure variables. Nowadays, regulation techniques are based on Proportional-Integral-Derivative (PID) controllers and their variations. However, to ensure that the process variable achieves the desired performance, controller tuning is essential. This task is difficult for process control engineers due to the complex interaction among devices and the nonlinear dynamics involved, making it challenging to develop a mathematical model that allows for the practical application of tuning techniques. System identification is the most widely used technique in control engineering for the rapid formulation of a process model for controller tuning. The Least Squares (LS) algorithm is one of the most widely used, however, to ensure consistent estimates, it is essential to use input signals that satisfy the persistent excitation condition. Despite this, the Least Squares of Orthogonal Distances (LSOD) method has shown that this condition is not strictly necessary to obtain reliable estimates. This work presents the application of the LSOD and LS methods for the offline identification of first- and second-order level and flow systems of an industrial process module, to address the tuning problem and experimentation of PID controllers for level and flow regulation. Performance indexes show better level and flow regulation with PID controllers based on LSOD-identified models than those using the LS method. Specifically, the best performance for level regulation was achieved with the following indexes: ISE of 4.29 × 10 4 and IAE of 2432. For flow regulation, the indexes were ISE of 2.7 × 10 4 and IAE of 1625. • Identification of the level and flow systems of an industrial processes module by LSOD method without persistent excitation. • Tuning and experimentation of PID controllers through the identified models for regulating level and flow variables. • Determination and calculation of the upper and lower bounds of the objective function for the LSOD method.