Skip to main content
eScholarship
Open Access Publications from the University of California

Intelligent Power Assist Algorithms for Electric Bicycles

  • Author(s): Fan, Xuan
  • Advisor(s): Tomizuka, Masayoshi
  • et al.
Abstract

This dissertation considers intelligent power-assist algorithm designs for electric bicycles. Traditional electric power-assist bicycles (EPBs) employ proportional power-assist strategy. The ratio is usually set to 1:1, which means that the motor will provide the same amount of assistive torque as the amount of the human's pedaling torque. This strategy is too rigid and does not consider the interaction between the bicycle, the human and the environment. Intelligent power-assist algorithms are needed to address such issues. In this dissertation, we focus on the uphill riding scenario, since it is the situation where the cyclist faces the most difficulties. The dynamic properties of electric bicycles will be studied and an appropriate model will be developed for intelligent power-assist algorithm design purposes. Two types of intelligent power-assist algorithms will be introduced to help the human ride uphill more easily. One is the robust disturbance observer (DOB) based power-assist algorithm, which can observe and compensate for the environmental disturbance that the bicycle system is subjected to during uphill riding. The robust DOB provides flexibility to the power assistance and within the motor's capability, it can make riding uphill feel like riding on the level ground. The other intelligent power-assist algorithm is based on repetitive control technique. The human's pedaling torque is repetitive by nature of the crankset's mechanical design. The pedaling torque reaches it local minimum and maximum twice during one complete pedal cycle. During uphill riding, the difference between the maximum torque and minimum torque is so large as to cause severe fluctuation in the torque profile, and, in turn, result in fluctuations in the velocity and acceleration profiles. We call the fluctuant human torque input "nonuniform human input" and compensate for the fluctuation with a repetitive control based power-assist algorithm. Repetitive control designs in both the time domain and the pedal-angle domain are considered. An experimental EPB system was built to verify the effectiveness of these two types of algorithms. Details of the experimental setup will be introduced. Simulation and experimental results will be shown in this dissertation.

Main Content
Current View