Bo Xu. Application of Fuzzy Neural Network in Recognition of Athlete’s Physical State. Dynamic Systems and Applications 29 (2020) No. 5, 1755 – 1763

https://doi.org/10.46719/dsa20202956

ABSTRACT.
The action acknowledgment frameworks have been viewed as viable in following exploration zones, for example, human services and life bolster client exercises. Wearable accelerometers are given through MEMS innovation that groups physical action (PA).The cutting edge PA arrangement framework utilizes limit based strategies and machine learning (ML) calculations. Every PA can display subject and subject change, which are the primary downsides, in light of limits and AI strategies. Because of the absence of preliminary information to prepare the grouping calculations in ML, it is essential to build up a system that requires less making information for PA bunches. This system Fuzzy Neural Network (FNN)a novel customized PA acknowledgment model structure dependent on the semi-managed bunching technique, utilizing a single accelerometer to stay away from fixed limit innovation and conventional grouping strategies. The proposed method needs to calculate the (initial) centroid of PA clusters. The amount of data is limited, and the average accuracy is achieved, which has the potential to identify behavioral changes in the subject and abnormal events such as falls. Recognizing these movements, and the fuzzy associative memory system of these fuzzy rules is realized. This method is independent of the target and speed of movement.

Keywords: component; Semi-supervised clustering; Physical activity recognition; Physical activity transition model; ambient assisted living; Independent living.