Contents I. Introduction
Sports and fitness have become increasingly common lifestyles in modern society. For athletes or fitness enthusiasts, monitoring the health status and abnormalities of the exercise system is very important [1]–[2]. However, traditional sports system monitoring methods are often conducted through medical examinations or physical fitness tests. Although these methods can provide certain reference information, they require professional doctors or physical fitness coaches to evaluate and cannot be monitored in real time. With the rapid development of the Internet of Things, sensor technology, cloud computing, and artificial intelligence technology, intelligent motion system monitoring has become possible [3]–[4]. By embedding sensors in equipment such as sports equipment, clothing, and bracelets, a large amount of exercise data can be collected, such as heart rate, step count, speed, and posture. These data can not only assist athletes or fitness enthusiasts in self-monitoring and evaluation, but also provide a large amount of data support for researchers to promote research and development in the field of sports. However, with the increasing collection of data, how to quickly and accurately monitor and identify abnormal data of motion systems has become a challenge [5]–[6]. Traditional anomaly detection methods often require specialized design for specific types of anomalies, and require a large amount of manual intervention and data annotation. Anomaly detection methods based on artificial intelligence algorithms can quickly extract abnormal features and patterns from massive data through automatic learning and optimization, achieving effective monitoring and recognition of motion system abnormal data [7]–[8]. This article aimed to explore methods for artificial intelligence algorithms in motion systems, the capabilities of detecting abnormalities in motion, and promote the development of intelligent motion monitoring in this field.
