# Fuzzy control method sets new standard

Groundbreaking techniques of model-based fuzzy control may become the standard solution for controlling complex processes, say researchers at the University of Strathclyde’s chemical engineering department. ‘Fuzzy logic provides a sound mathematical framework to describe imprecision and uncertainty,’ explained Craig Edgar, a University of Strathclyde researcher. Conventional fuzzy control was pioneered at Queen Mary College, London, […]

Groundbreaking techniques of model-based fuzzy control may become the standard solution for controlling complex processes, say researchers at the University of Strathclyde’s chemical engineering department.

‘Fuzzy logic provides a sound mathematical framework to describe imprecision and uncertainty,’ explained Craig Edgar, a University of Strathclyde researcher.

Conventional fuzzy control was pioneered at Queen Mary College, London, in the early 1970s. But mathematical complexity meant that its development was held back by poor computing power.

The breakthrough made at Strathclyde, using modern computer power, has been to develop fuzzy logic models of complex processes, potentially making automation possible for the first time.

‘Fuzzy set theory is similar to traditional Boolean theory but with one fundamental difference,’ said Edgar. ‘In Boolean theory a value either belongs to a set or it does not. In fuzzy theory a value may have partial belonging to a fuzzy set.’

Partial membership is termed ‘the degree of belonging’ and is expressed in a range of 0 to 1, with 0 having no belonging, and 1 complete belonging.

This means that temperature measurement might be represented by the fuzzy sets of ‘very cold’, ‘cold’, ‘tepid’, ‘warm’, ‘hot’ and ‘very hot’.

Edgar said: ‘When controlling a process, an operator tends to think of temperature as quite high , so a valve may have to be open a little.

‘Fuzzy logic allows an automatic controller to think like a human operator.’

The fuzzy logic model corresponds to a human operators’ ‘model’ or mental picture of the process and what will happen in any set of actions or conditions.

This allows controllers to handle non-linear and dead-time processes. Feed-forward control can be added to the controller by including one or more disturbances into the process model.