Use this URL to cite or link to this record in EThOS:
Title: Repetitive control for FES-based tremor suppression
Author: Copur, Engin H.
ISNI:       0000 0004 7224 9729
Awarding Body: University of Southampton
Current Institution: University of Southampton
Date of Award: 2017
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Tremor is an involuntary, rhythmic movement of one or more body parts which prevents patients from performing activities of daily living (ADL), thereby greatly diminishing quality of life. Although there are various methods to suppress tremor, each is limited by either cost, ability to produce sufficient tremor suppression (TS) or interference with voluntary action. Functional electrical stimulation (FES) has attracted significant research attention as a novel approach for TS, and has substantial potential advantage in terms of cost, convenience and size. Results from previous studies have demonstrated that FES controlled by classical control methods is unable to produce satisfactory TS, while also producing intense interference with voluntary action. Thus their performance is inherently limited by the drawbacks of conventional feedback design. Since repetitive control (RC) is able to reject a periodic disturbance completely and tremor can be regarded as a periodic disturbance, RC may have more potential for TS than other classical feedback control techniques. This thesis thus provides the first thorough implementation and assessment of RC for TS by means of FES in order to establish its feasibility and performance advantages compared to the conventional filter techniques (FT) that have so far been employed. Due to the importance of the wrist joint in ADL, wrist movement in flexion/extension is targeted within this programme of research. First, simulations are performed using a previously validated linear model of tremulous wrist dynamics to evaluate the feasibility of applying linear RC for TS, and results show that RC is able to produce complete tremor suppression but incurs a severe interference with voluntary action. Since this linear model does not account for the critical properties of the dynamics and RC requires a reliable knowledge of the plant dynamics to guarantee the stability of RC and improve its performance, this thesis then proposes a nonlinear model structure with an identification procedure that maintains the balance between the accuracy and the needs of the clinical domain. Then a linearising control approach is undertaken to enable linear RC design and a mechanism is proposed to preserve the voluntary action which is restricted by applying RC. Analysis confirms that the proposed mechanism can significantly reduce the undesirable interference of RC with voluntary action. Finally, two separate experimental studies are conducted using a test platform. The former is performed to fix the order of the model components in advance to reduce the computational effort of the identification procedure. The response of the wrist joint to FES signals applied to the flexion/extension muscles of each participant is collected. Using these input/output data sets, different models are identified for a number of different orders. Then the fitting accuracies of all models with the same order are averaged and an order is selected with maintaining the balance between model accuracy and simplicity. The second experimental study is to confirm the higher potential of the proposed RC compared with FT for FES-based TS. A mechanical system is introduced to the existing test platform to induce tremor artificially since the participants are unimpaired. The participants are asked to complete three consecutive tests to examine the effect of induced tremor on the ability of patients to perform voluntary task, the capability of FES to suppress the induced tremor and the effect of FES on the voluntary motion. The results indicate that the proposed RC approach with the developed model identification procedure can suppress tremor more effectively than FT and leads to a minimal interference with voluntary motion. This research therefore has established the feasibility of RC for TS by validating the model accuracy, the identification procedure, and the control performance. It has therefore potential for future exploration to provide more effective solutions for patients.
Supervisor: Freeman, Christopher Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available