Interval Type-2 Fuzzy Sets for Self-Organising Fuzzy Logic based Control with On-line PSO Optimisation

The Self-Organising Fuzzy Logic Controller (SOFLC) which is a modified and extended version of the traditional Fuzzy logic controller (FLC) was introduced to enable the controller to modify its structure based on its evaluation to the performance of the system under control. Due to design difficulties, the performance index (PI) table which is responsible for the corrections of the low-level control 'adaptable' has been left practically unchanged in most applications since the original SOFLC scheme was introduced in 1979. The self-organising mechanism included in the SOFLC enables the controller to effectively control non-linear, mathematically ill-understood, and uncertain systems. However, SOFLC-based systems can also suffer from drawbacks such as high memory storage requirement and high computational burden especially when the scheme is applied to multivariable systems. In this paper, interval type-2 fuzzy sets rather than the traditional type-1 fuzzy sets are used to enhance the capabilities of the SOFLC scheme which is also given an additional degree of freedom by using a dynamic on-line Particle Swarm Optimisation-based performance index table. When compared with type-1 SOFLC scheme, the simulation results show that the proposed algorithm is more robust when it was tested on a non-linear muscle relation process in the presence of noise and parameter changes.