Passive Control and Optimal Regeneration in DC Motors
Vailati, Léo
0000-0002-9369-5176
:
2023-06-22
Abstract
This work presents a study of the passive control of DC motors, presents corresponding analyses and control methods, and experimentally validates the methods across multiple platforms. The proposed methods entail a model-based switching approach that constrains the motor to operate as an electric generator, and implements feedforward and feedback controllers that enable operation as a torque-controllable regenerative brake. The structure of the controller prevents the motor from adding energy to the system and additionally prevents the motor from assuming braking behaviors that surpass a maximum achievable impedance which corresponds to the shorted-leads damping behavior (beyond which energy regeneration does not occur). The passive control method has guaranteed control stability and was analytically and experimentally shown to outperform conventional techniques in terms of control robustness, vibration, and behavioral accuracy. Since the method can be implemented with standard motor control hardware, versatility to operate with conventional methods is preserved.
The passive motor control method was additionally implemented in a prototype knee prosthesis, which is an application that can benefit substantially from the energy regeneration, stability, and smoothness offered by passive motor control. Energy regeneration in particular can offer significant benefits for mobile and wearable robotic devices, including increased battery life and reduced battery weight and volume. Knee prostheses are well suited for the application of energy regeneration technology due to the net-negative energetic profile of the knee joint in walking gait at comfortable walking speeds. Within that context, this work incorporated the previously proposed passive motor control methods, informed by constrained optimization techniques, that provide maximized energy harnessing while achieving the kinematic goals of walking (e.g., toe clearance) and without requiring additional user effort. Experiments were performed with an amputee participant wearing a prototype knee prosthesis. The regeneration-optimized controller was compared to a baseline controller which had previously been developed and validated for a prototype knee prosthesis without regard to regeneration performance. Relative to the baseline controller, the regeneration-optimized controller demonstrated significant gains in regeneration (approximately double the efficiency), while retaining effectively equivalent kinematics and essentially identical mechanical energy (which serves as proxy for user effort).