Temporal Decision Trees In Diagnostics Systems

In this paper the problem of diagnostics in dynamic systems using artificial intelligence methods is considered. An analysis of the dynamic situation at a technical object involves two stages: building of a model (in particular case, a decision tree) and using it to classify new situations occurring at this object. Data, on the basis of which it is necessary to carry out diagnostics, can be represented by sets of time series. As a classification model in such situations, temporal decision trees can be used. The definition of the temporal decision tree is given, an algorithm for constructing such decision trees is described. A new algorithm for constructing Temporal ID3 temporary decision trees is proposed. It is characterized by the new criterion for selecting decision tree nodes. The process of faults and anomaly situations diagnostics for various methods of building decision trees is modeled and compared. The analysis of the dynamic situations diagnostics results by temporal trees constructed by different algorithms is given and followed by recommendations on the use cases of these algorithms.