System and method for generating an aircraft fault prediction classifier

A system comprising:

• a processor;
• and
• a memory coupled to the processor and storing instructions executable by the processor to perform operations comprising:
• receiving input data including a plurality of feature vectors, the input data including sensor data associated with a first aircraft;
• receiving fault data indicating timing of a plurality of faults, the fault data distinct from the input data;
• labeling each feature vector of the plurality of feature vectors based on a temporal proximity of the feature vector to occurrence of a fault of the plurality of faults as indicated by the fault data, wherein feature vectors that are within a threshold temporal proximity to the occurrence of the fault are labeled with a first label value, and wherein feature vectors that are not within the threshold temporal proximity to the occurrence of the fault are labeled with a second label value;
• identifying a subset of the plurality of feature vectors and a second subset of feature vectors based on corresponding labels, wherein the subset includes feature vectors associated with labels that indicate the first label value, and wherein the second subset includes feature vectors associated with labels that indicate the second label value;
• determining, for each feature vector of the subset of the plurality of feature vectors, a probability that a label associated with the feature vector is correct, wherein probabilities that an associated label is correct are not determined for the second subset of the plurality of feature vectors, and wherein the second subset of the plurality of feature vectors is not considered in determining probabilities for the subset of the plurality of feature vectors;
• reassigning labels of one or more feature vectors of the subset, the one or more feature vectors having a probability that fails to satisfy a probability threshold;
• after reassigning the labels of the one or more feature vectors, training an aircraft fault prediction classifier using supervised training data including the plurality of feature vectors and the labels associated with the plurality of feature vectors, the aircraft fault prediction classifier configured to predict occurrence of a second fault of a second aircraft using second sensor data of the second aircraft, wherein the aircraft fault prediction classifier comprises a random forest classifier;
• and
• training a probability predictor comprising a random forest regression predictor distinct from the random forest classifier;
• and
• generating, by the probability predictor, a confidence score associated with a probability that the label associated with the feature vector is correct for each vector of the subset of the plurality of feature vectors based on an average of outputs from multiple regression decision trees.

A non-transitory computer-readable storage device storing instructions that, when executed by a processor, cause the processor to perform operations comprising:

• receiving input data including a plurality of feature vectors, the input data including sensor data associated with a first aircraft;
• receiving fault data indicating timing of a plurality of faults, the fault data distinct from the input data;
• labeling each feature vector of the plurality of feature vectors based on a temporal proximity of the feature vector to occurrence of a fault of the plurality of faults as indicated by the fault data, wherein feature vectors that are within a threshold temporal proximity to the occurrence of the fault are labeled with a first label value, and wherein feature vectors that are not within the threshold temporal proximity to the occurrence of the fault are labeled with a second label value;
• identifying a subset of the plurality of feature vectors and a second subset of feature vectors based on corresponding labels, wherein the subset includes feature vectors associated with labels that indicate the first label value, and wherein the second subset includes feature vectors associated with labels that indicate the second label value;
• determining, for each feature vector of the subset of the plurality of feature vectors, a probability that a label associated with the feature vector is correct, wherein probabilities that an associated label is correct are not determined for the second subset of the plurality of feature vectors, and wherein the second subset of the plurality of feature vectors is not considered in determining probabilities for the subset of the plurality of feature vectors;
• reassigning labels of one or more feature vectors of the subset, the one or more feature vectors having a probability that fails to satisfy a probability threshold;
• after reassigning the labels of the one or more feature vectors, training an aircraft fault prediction classifier using supervised training data including the plurality of feature vectors and the labels associated with the plurality of feature vectors, the aircraft fault prediction classifier configured to predict occurrence of a second fault of a second aircraft using second sensor data of the second aircraft, wherein the aircraft fault prediction classifier comprises a random forest classifier;
• and
• training a probability predictor comprising a random forest regression predictor distinct from the random forest classifier;
• and
• generating, by the probability predictor, a confidence score associated with a probability that the label associated with the feature vector is correct for each vector of the subset of the plurality of feature vectors based on an average of outputs from multiple regression decision trees.