Research Of Exercise Patterns Recognition And Energy Expenditure Model

With the development of technology, more and more advanced devices come into being, and wearable device has become a research focus. Combined with sports exercise, wearable devices are able to help people acquire different kinds of exercise data, and get feedback in real time. Because of the decline in Chinese students’ physical fitness, wearable devices can be used to acquire students’ exercise data. After processing and analyzing, exercise patterns can be recognized and energy expenditure can be calculated, so that exercised program can be planned elaborately to improve students’ physical fitness. Based on the project from Adolescent Health Assessment and Exercise Interference Key Laboratory of Education Ministry in East China Normal University, this paper developed a data acquisition system, and proposed exercise patterns recognition algorithm and energy expenditure model.Sports data acquisition system is an important part in the Online Platform of Adolescent Health Assessment and Exercise Interference. This paper proposed an architecture for sports data acquisition system. Then, in order to fulfill the requirements in high concurrency and real-time, this paper introduces an independently developed wearable sports data collector and hardware base station. Meanwhile, this paper defined data transmission protocol and designed a multithreading structure for servers, so that the interaction between servers and database can be achieved. This design guarantees data transmission in both software and hardware aspects.Exercise patterns are related to exercise effects closely. This paper analyzes three usual exercise patterns in elementary and secondary school:walking, jumping and running. On the foundation of data acquisition system, this paper preliminarily identifies jumping through analyzing exercise duration, then recognizes approach jumping and standing long jump through signal magnitude area. Next, the algorithm distinguishes normal-pace walking and fast running on the basis of stride frequency. After that, the algorithm recognizes fast walking and slow running through signal magnitude area, body angles and average acceleration in accelerating phase. Making the best use of above methods, this paper proposes exercise patterns recognition algorithm to recognize three exercise patterns.Energy expenditure is a direct quantitative reflection of exercise effects. This paper proposes three energy expenditure models based on different exercise patterns and students’ body features. The paper analyzes the relation between energy expenditure and age, height, weight and gender. Combined with different exercise patterns, the paper proposes linear acceleration model. Through multivariable linear regression analysis, the paper also proposed linear integration model. Besides, the author also uses theorem of kinetic energy to calculate energy expenditure. All these models can be used to quantify energy expenditure.At last, the author conducts comprehensive test on the research. Through computer stimulation and field test, the system delay gradually decreases with the increase in the number of thread and request, and the system meets requirements in performance on barrier-free playground. The result of test on exercise patterns recognition algorithm shows that algorithm’s recognition correct rate is up to 93.3%. In the end, this paper analyzes linear relation between results calculated by three models proposed in this paper and the result calculated by equation proposed by Meijer et al, and the result shows that the linear acceleration model can estimate energy expenditure accurately in relative terms.