ASPECT-f

Advanced control algorithms for programmable logic controllers

Fuzzy-Gain Scheduling Controller with On-line Learning and Tuning

The controller is intended for control of highly nonlinear processes, where the process dynamics vary depending on the operating point. For example, a process exhibits very fast dynamics in a “high" region and slow dynamics in a "low" region. A standard PID controller tuned for good performance in the "high" region performs sluggishly in the "low" region. The mentioned operating point or region is specified by a so-called scheduling variable that may be selected either as the current process output value, the controller output signal, or an external signal.
The controller could be effectively used on processes:

• Heating exchangers
• Hydraulic plants
• Level control
• Chemical and bioreactors
• Neutralization
• Control of pH

FEATURES:

• Control system, intended for closed loop process control;
• Designed for efficient control of 'demanding' processes (nonlinear, slowly time variant, with dead time);
• Better performance than the conventional PID controller;
• Scheduling or switching of controller parameters;
• Automatic tuning of controller parameters using a nonlinear model of the process;
• Nonlinear process model: fuzzy set of local linear models;
• On-line learning of the nonlinear model by analyzing process signals during experiments or regular closed loop operation;
• On-line control performance monitoring;
• Control system implemented on a standard Mitsubishi PLC platform;
• No expensive hardware required;
• Operator interacion via standard industrial HMI similarly to conventional PID controllers;
• Connection to SCADA is possible;
• Simple commissioning procedure;
• Applied and tested on several types of industrial processes;

Application example: pressure difference control (nonlinear, time-variant process)