NONLINEAR INFERENTIAL CONTROL (REGULATION, PROCESS CONTROL, AUTOMATION)

Nonlinear Inferential Control (NLIC) is presented as a general method of controlling highly nonlinear processes. The NLIC structure uses a process model to estimate the influence of disturbances on the process, and provide control action to maintain the desired trajectory or setpoint. Very general models may be used to represent the process, and these models may: (1) be highly nonlinear; (2) include time delays (of measurement, control, etc.); (3) be of high order; (4) forced by unknown inputs (disturbances); (5) deviate significantly from the actual process.; There may or may not be (1) primary variable output measurements; (2) secondary variable measurements; (3) feed forward (disturbance) measurements.; The performance of the NLIC is related directly to the degree which the process response is represented by the model. An imperfect model will require a more sluggish control system response. The speed of response of the control system is adjusted by the user to provide the desired tradeoff between speed of response and robustness.; This thesis describes the NLIC system and the basis for its derivation. A design procedure is presented which allows the user to specify the control system based on the process model. NLIC has been applied to a laboratory heat exchanger, a simulated neutralization process, and a simulated reactor process. The results of all tests of the NLIC show that the control system is robust, and offers significant improvement over conventional methods for the control of higher nonlinear systems.; This thesis describes a general methodology for control of nonlinear systems, and illustrates the feasibility of the method. Much work remains to be done to further understand and refine the NLIC structure, and this work is a topic for future research.