Design of a heart rate controller for treadmill exercise using a recurrent fuzzy neural network

Area of Science:

• Biomedical Engineering
• Control Systems
• Artificial Intelligence

Background:

• Computer-controlled treadmills require precise speed and incline adjustments.
• Minimizing user heart rate deviations from a target profile is crucial for effective exercise.
• Existing control systems may exhibit fluctuations and non-linearities.

Purpose of the Study:

• To develop a recurrent fuzzy neural network heart rate controller (RFNNHRC) for automated treadmill control.
• To minimize user heart rate deviations from a pre-set profile.
• To enhance the robustness and stability of treadmill exercise control systems.

Main Methods:

• The RFNNHRC integrates fuzzy reasoning for uncertainty and a recurrent neural network for learning system nonlinearities.
• Treadmill speed and incline are controlled via servo motors actuated by the RFNNHRC.
• An on-line learning algorithm and adjustable parameters were implemented using a dsPIC 30F4011 DSP for robust performance.

Main Results:

• The RFNNHRC demonstrated superior heart rate tracking compared to proportional integral control, with reduced fluctuations.
• Treadmill speed and incline adjustments were smoother under the proposed control scheme.
• Experiments confirmed improved heart rate tracking performance and exercise control.

Conclusions:

• The developed RFNNHRC scheme provides effective heart rate control during treadmill exercise.
• Novel RFNNHRC structure and stability analyses, including Lyapunov function tuning, ensure system stability.
• The RFNNHRC scheme offers superior performance for maintaining target heart rates during varied exercise intensities.