Application of fuzzy intelligence to Elebike control design

This paper proposes the application of fuzzy intelligence to an electrical power-aided bicycle, i.e. "Elebike". Fuzzy intelligence presented in this paper is grouped into two categories. One is fuzzy sensing to gripe the attempt of the Elebike rider, i.e. the expected speed and acceleration in his/her mind that nobody else knows. The other is fuzzy logic control (FLC) that supplies complementary power, i.e. electrical motor current and torque, to accompany the rider's pedal force for reaching the expected speed and acceleration. An overall system is introduced to show the concept of the complementary torque generation scheme. A dynamic model of a conventional bike including torque transmission, positioning and analysis of a servo motor and sensors is given. Fuzzy logic control (FLC) is employed to meet various conditions of riders, ground and up/down hill. Fuzzification, inference and defuzzification in FLC are used in order to generate motor current and torque as well by assigning motor voltage. A two-level inference consisting of top level and bottom level inference is developed. The top level inference judges if the motion status is acceleration, constant speed or deceleration and places weighting on the dominant bottom level inference. Eventually, what to be tuned and how to tune the fuzzy logic control (FLC) are schemed by means of neural network learning for improving the performance of response and smoothness.