| CPC G05B 23/0283 (2013.01) [G05B 23/0254 (2013.01); G06F 18/2148 (2023.01); G06N 20/00 (2019.01)] | 20 Claims |
|
1. A computer-implemented method of predicting an event horizon, comprising:
maintaining, by a processor, condition data indicating a plurality of conditions occurring on one or more physical systems at a plurality of points in time;
receiving time series data comprising one or more sets of signal data from one or more sensors attached to a particular physical system of the one or more physical systems over a specific period of time;
creating, by the processor, an input feature vector from the time series data,
the input feature vector representing a given condition of the plurality of conditions occurring at a specific time during the specific period of time;
generating, using the input feature vector and a particular trained machine learning model of a plurality of trained machine learning models, a forecast value that indicates an amount of time from the specific time to an occurrence of a particular target condition of the plurality of conditions on the particular physical system,
the particular target condition being different from the given condition,
each trained machine learning model of the plurality of trained machine learning models corresponding to a distinct target condition of the plurality of conditions,
the particular trained machine learning model being trained on a certain training dataset comprising feature data and label data,
the feature data comprising a plurality of feature vectors,
each feature vector of the plurality of feature vectors representing a current condition of the plurality of conditions occurring on a certain physical system at a certain time within a range of time,
the label data comprising a plurality of time values;
causing, based on the forecast value, an action to be executed on the particular physical system.
|