Abstract: Aiming at the problems of lacking typical fault data and quantifying damage degree in the anomaly detection of equipment system, an anomaly detection method based on adaptive chaotic particle swarm optimization and support vector data description was proposed to detect gearbox fault anomaly. In this method, and the adaptive chaos theory was introduced into the conventional particle swarm optimization (PSO) algorithm,which enhanced the ability of particles to jump out of local optimal solutions and improved the global search ability of particle populations for optimal solutions. The ACPSO algorithm was used to optimize the penalty factor and kernel parameter of SVDD. The proposed method was successfully applied to identify the abnormal stateof gearbox. The results indicate that the ACPSO-SVDD anomaly detection method can not only accurately detect different types of fault anomalies, but also quantitatively analyze the degree of fault.
Key words: adaptive chaotic particle swarm optimization(ACPSO); support vector data description（SVDD）; anomaly detection; parameter optimization