Cluster industrial robot failure diagnosis method based on outlier excavation

The invention relates to a fault diagnosis method for trunked industrial robots based on outliers mining, belonging to the filed of fault diagnosis of electromechanical equipment. The method comprises the following steps of: firstly collecting data of original operating state of the trunked industrial robots and carrying out preprocessing operation of classifying and the like; then using a cluster analysis method to carry out analysis by taking a plurality of robots as a group, so as to lead a plurality of equipment to carry out classification according to operational state; based on clustering, utilizing an outliers mining method to calculate outlier factors of each industrial robot and then obtaining outlier degree thereof; separating outliers according to outlier degree and further determining that whether individual industrial robot represented by the outlier occurs fault or not; and judging the specific parts of faults of the robots according to the types of abnormal operation parameters and obtaining fault diagnosis results. By utilizing the fault diagnosis results, the method can implement targeted and predictive maintenance, thus avoiding the occurrence of faults of equipment, improving the reliability of equipment and guaranteeing reliable operation of trunked operational robots.