Research On Kinetic Energy Harvesting And Human Activity Assessment Approach Based On Wearable Device

With the improvement of people's living standards,people's daily health monitoring needs become more and more urgent,and human activity assessment based on wearable devices has become a new hotspot in pattern recognition and machine learning.However,limited battery power has become a bottleneck in the application of wearable devices in the field of health monitoring.How to fully utilize the rechargeable energy to achieve sustainable health monitoring while satisfying human activity monitoring requirements is a problem that needs to be solved urgently.In order to overcome the problem of excessively high energy consumption in the health monitoring system based on traditional inertial sensors,this paper studies the human activity assessment method based on rechargeable kinetic energy.Human activity assessment mainly includes two major parts,i.e.human activity recognition and caloric expenditure estimation.The essence is to use the correlation between the harvested kinetic energy signals and the human activity to realize activity classification and calorie estimation.This paper takes harvested kinetic energy as the research object and aims to achieve long-term and continuous monitoring of human activities.It studies the human activity recognition method and the calorie regression prediction method.The main research contents are summarized as follows:1.This paper implements a wearable device based on piezoelectric energy harvester and tri-accelerometer,and builds a human activity dataset collected from kinetic energy harvester and accelerometer.The dataset contains 9 different activities from ten participants.The samples are collected from three different parts of body positions,i.e.leg,waist,and wrist,using a single sensor node in the free-living conditions.2.This paper proposes an activity recognition method based on human kinetic energy and acceleration(SRC-EA).This method utilizes the multimode information generated by human motion to improve the accuracy of the system recognition and decrease the system power consumption by reducing the sampling frequency of the accelerometer.The experimental results show that the average accuracies have achieved 2.5% and 39.44% higher with SRCEA activity recognition method and reduced system power consumption by 60%.3.This paper proposes a caloric expenditure estimation method based on human kinetic energy(CEE-KEH).This method utilizes the voltage output signal generated by human motion to first classify the activity intensity of different actions,and then selects a specific regression model based on the results of classification.Then the caloric expenditure estimation based on activity-specific models is implemented.The experimental results show that the CEE-KEH method can accurately estimate the caloric expenditure of different actions and reduce system power consumption by 66.36%.