A novel neural fuzzy controller (NFC) for dynamic braking resistor is presented to improve the improve the transient stability of power systems. The primitive control scheme is developed by the fuzzy logic theory according to the angle and speed on the generator. The fuzzy controller is of decentralized nature since only local measurements are employed as the input signals. Then the fuzzy rules xpressed in linguistic variables are transformed into a training set of input-output patterns. A multi-layer neural network(MNN) is applied to learn the desired control laws using the error back propagation algorithm , the neural network technique is advantageously used to refine and generalize the control rules of fuzzy system. Besides, The rule-matching time of the inference engine in the traditional fuzzy logic system can be avoided. To demonstrate the effectiveness of the proposed control scheme, a comparison to a variable structure controller (VSC) is analyzed. Numerical results show that the proposed neural fuzzy controller can provide fairly good damping characteristic over wider ranges of operating conditions , and can yield better dynamic performance than the VSC . Significantly, with the evolving symbiosis of these new techniques, the neural fuzzy controller has shown to be more adaptive and much less sensitive to changes in parameters.
