ABNORMALITY DIAGNOSIS SYSTEM AND METHOD

摘要

PROBLEM TO BE SOLVED: To predict a health state of machine plants accurately.SOLUTION: A health management system (abnormality diagnosis system) 1A comprises: a time series data acquisition part 11 for acquiring sensor data and environment data of multi-dimension from a machine plant; a first identification part 133a and a second identification part 133b for calculating a multi-dimensional vector in which the machine plant becomes an abnormal state from a normal state by a statistical method using the time series data as learning data; a state prediction part 23 for predicting a machine state or a health state using the multi-dimensional vector; a start point change detection part 31 for detecting whether a start point position of the multi-dimensional vector changes by a predetermined value or more or not, about the time series data acquired in a period before and after a period when maintenance work was performed; a sensor data extraction part 21 for extracting sensor data contributing to change of the start point position of the multi-dimensional vector; and an output part 15 for outputting the extracted sensor data as an index indicating the health state.