Hybrid feature selection for performance prediction of fluid control valves

A hybrid feature selection system to generate a predictive model of valve performance in a fleet of aircraft, the hybrid feature selection system comprising:

• at least one processing unit;
• a memory storing:
• a flight database of data collected during flights of aircraft of the fleet of aircraft;
• and
• non-transitory computer readable instructions that, when executed by the at least one processing unit, cause the at least one processing unit to:
• retrieve qualification data from the flight database;
• calculate a qualification dataset by applying valve-related parameters to the qualification data;
• display the qualification dataset on a graphical user interface associated with the at least one processing unit;
• receive, from a user and via the graphical user interface, a first user selection that corresponds to a selection of the valve-related parameters, wherein the qualification data is data collected during a first series of flights of a first aircraft at least before and after a non-performance event of a valve of the first aircraft;
• retrieve verification data from the flight database;
• calculate a verification dataset by applying the selection of valve-related parameters to the verification data;
• display the verification dataset on the graphical user interface associated with the at least one processing unit;
• receive, from the user and via the graphical user interface, a second user selection that corresponds to a set of the selection of the valve-related parameters that is less than all of the valve-related parameters, wherein the verification data is data collected during a second series of flights of one of the first aircraft and a second aircraft of the fleet, during which the respective aircraft did not experience a non-performance event of a valve of the respective aircraft;
• retrieve training data from the flight database;
• calculate a training dataset by applying the set of the selection of valve-related parameters to the training data;
• train a predictive model for valve performance with the training dataset;
• and
• provide the predictive model to the user.

A method for creating a predictive model for valve performance in a fleet of aircraft, the method comprising:

• for a first aircraft of the fleet that experienced a non-performance event of a valve of the first aircraft during a first series of flights:
• calculating, by a processing unit of a computerized system, a qualification dataset of valve-related parameters from data collected during the first series of flights at least before and after the non-performance event of the valve of the first aircraft;
• and
• identifying a first set of valve-related parameters from the qualification dataset of valve-related parameters that correlate to valve non-performance;
• for a second series of flights of one of the first aircraft and a second aircraft of the fleet, during which the respective aircraft did not experience a non-performance event of a valve of the respective aircraft:
• calculating, by the processing unit, a verification dataset of the first set of valve-related parameters from data collected during the second series of flights;
• and
• identifying from the first set of valve-related parameters, a second set of valve-related parameters that contains fewer valve-related parameters than the first set of valve-related parameters;
• calculating, by the processing unit, a training dataset of the second set of the valve-related parameters from data collected during additional flights of aircraft of the fleet;
• and
• training, by the processing unit, a predictive model for valve performance with the training dataset.