Swami, Anurag KumarYadav, Abhishek2017-12-062017-12-062016-12http://krishikosh.egranth.ac.in/handle/1/5810036997The conventional control systems perform satisfactorily as long as the design of the controller is based on the accurately known fixed dynamics of the plant or the process. But, the model that is used to represent the dynamics of the plant or the process is generally not accurate and also it is subjected to the parameters and environmental variations with time. To overcome the above problem, various strategies of artificial neural networks based nonlinear adaptive control are used. Neuron models that are generally used in the neural networks of control systems are simple units with summation type of aggregation and sigmoid type of activation functions. However, various neuron models with multiplicative type of aggregation exhibit better learning capability. Control task could be improved significantly by improving the learning capability of the neural networks of control systems. In this thesis work, some novel methods of control have been developed using different multiplicative models of artificial neurons. Different multiplicative models of neurons have been compared in terms of their usefulness in solving control problems. It has been shown that the steady-state and transient response of neural network control systems are improved by using these multiplicative models.enElectrical EngineeringThesis