Modeling and control of FES assisted standing-up in paraplegia.

摘要

The author has developed a new approach to the design of control systems for Functional Electrical Stimulation (FES) assisted standing-up in paraplegia. The approach is based on machine learning rather than traditional analytical modeling.; The main objective of the study was to investigate the model-free design methods, as it is difficult to develop analytical models of the human body. The results showed that nonlinear FES controllers could be tailored to the needs of an individual patient. The controllers continually adapt to the physical changes and disturbances and are coordinated to the voluntary efforts of the subject.; A sagittal plane model of standing-up was developed including closed-chain configuration and a fuzzy control model of the voluntary arm forces. The model was used in the rest of the thesis for evaluation and comparison of the FES controllers of the knee and hip joints during standing-up.; A comparative study showed that the simple closed-loop controllers outperformed the open-loop controllers but there was room for considerable improvements using nonlinear and adaptive control systems.; Genetic Algorithms (GA) were used to optimize the parameters of the Fuzzy Logic Controllers (FLC) and gain-scheduling PID controllers and produced controllers with superior performance compared to the open-loop and manually designed closed-loop controllers. The voluntary arm forces and the trajectory error were used as the optimization criteria.; Reinforcement Learning (RL) was used for online tuning of the FLC controllers to minimize the arm forces and the knee terminal velocity. The combination of RL and FLC performed satisfactorily in rejecting the disturbances and adapting to the slow changes in system dynamics.; Finally, the GA was used to generate optimal trajectories tailored to the needs and limitations of the people with impairments such as paraplegia and functionally impaired elderly.; It is expected that these developments in FES control will lead to significant improvements in the clinical applications of FES. The resulting motions will be smoother, safer, and more energy efficient. There won't be a need for continual adjustments by a clinician and the FES controller will become an integral part of the patient's motion control system.