A Respiratory Rate Detection Scheme Based on Improved Adaptive Kalman Filtering and Gait Matching

To achieve the measurement of respiratory rate that adapts to complex environments, while meeting the requirements of portability, quick deployment, and low cost, this study designed and implemented a respiratory rate detection scheme based on wearable motion sensors. Firstly, the detection solution uses a covariance matrix that adapts to environmental noise for Kalman filtering, and is combined with an improved moving window method that is weighted with attenuated memory index, aiming to improve the robustness and stability of filtering in complex environments. Secondly, with the intension of calculating the respiratory rate based on the discrete respiratory waveform curve, this study proposes a variable interval difference algorithm based on step size similarity matching, which aims to reduce the influence of spikes and small oscillations in the respiratory waveform curve, and adapt to respiratory rate calculation in different scenarios. The experimental results demonstrate that the error of the measured respiratory rate in this study is below 3%, and the root mean square error of respiratory rate detection is less than 1, which confirms the practical value of the proposed method and provides new technical reference for portable respiratory rate detection.