Iterative Learning Control of Functional Electrical Stimulation Based on Joint Muscle Model

Abstract

Functional electrical stimulation (FES) is an effective treatment for the rehabilitation of stroke patients with hemiplegia. At present, it is challenging to accurately control the functional electrical stimulation during rehabilitation as various parameters of electrical stimulation are difficult to determine, and the stimulation response is easily affected by interferences. To improve the control accuracy for trajectory tracking during repetitive training and to compensate external interference, in this paper we take the knee joint as an example designed a functional electrical stimulation system based on adaptive network-based fuzzy inference system (ANFIS) and iterative learning control (ILC). Firstly, an adaptive fuzzy neural inference system was used to establish the joint muscle model, and a PID-type iterative learning controller was used to achieve the adjustment of functional electrical stimulation parameters. The maximum error of the ANFIS-based muscle model was 1.64Nm and the root means square error was 0.4327Nm. The maximum angle error of the actual knee motion compared with the expected angle was 22.76°, and the root means square error was 6.7413° after 10 iterations. Therefore, the system realizes the control of the pulse width of functional electrical stimulation in rehabilitation training, so that patients can carry out rehabilitation training according to the expected trajectory, which provides convenience for the rehabilitation training of stroke hemiplegia patients.