GRNN-Based Evolving Control

Self-evolving neural networks (SENNs) have a great potential to perform better than adaptive neural networks in the control tasks with the addition of structure adaptation to the typical adaptive NN. Thus, adaptive NNs are enclosed in SENNs as shown Fig. 11.1. Adaptive controllers can only adapt their parameters while evolving controllers offer twofold of adaptation: on the structure and parameters levels. Simultaneous structure and parameter adaptations might lead to high control system accuracy particularly when dealing with systems that exhibit significant uncertainties and faults. These include the actuator faults, a commonly occurring phenomenon in dynamic systems.