Machine learning to predict part consumption using flight demographics
The present disclosure provides for using machine learning to evaluate flight demographics and predict part consumption. Historical aircraft part consumption data indicating prior consumption of aircraft parts is accessed, and historical aircraft flight demographics associated with the historical aircraft part consumption data are determined. A machine learning model is trained based on the historical aircraft part consumption data and the historical aircraft flight demographics, and the machine learning model is deployed to predict future aircraft part consumption.
1 . A computer program product, the computer program product comprising a non-transitory computer-readable storage medium having computer-readable program code embodied therewith, the computer-readable program code executable by one or more computer processors to perform an operation, the operation comprising:
forecasting, for at least a first aircraft type of a plurality of aircraft types, a number of aircraft of the first aircraft type that will be in service during a future window of time;
forecasting future flight demographics during the future window of time based at least in part on the forecasted number of aircraft;
determining an aircraft configuration for the first aircraft type, the aircraft configuration indicating a number of parts of a first part type on each aircraft of the first aircraft type;
generating predicted part consumption of the first part type during the future window of time by processing the future flight demographics and the aircraft configuration using a machine learning model, wherein the machine learning model is a neural network, and refining the machine learning model comprises using backpropagation to iteratively refine at least one of weights or biases of the neural network;
wherein the predicted part consumption represents a predicted number of parts of the first part type that will be consumed during the future window of time by the aircraft of the first aircraft type;
facilitating reconfiguration of a spare part system based on the predicted part consumption, wherein the one or more computer processors are configured to facilitate the reconfiguration of the spare part system by controlling a number of the first part type that is at least one of constructed, ordered, stored, or shipped based on the predicted number of parts of the first part type that will be consumed during the future window of time;
monitoring actual part consumption of the first part type during the future window of time;
and refining the machine learning model based on a comparison of the actual part consumption to the predicted part consumption.
2 . The computer program product of claim 1 , wherein the future flight demographics comprise predicted time in flight of the aircraft of the first aircraft type during the future window of time.
3 . The computer program product of claim 1 , wherein forecasting the future flight demographics comprises forecasting one or more exogenous indicators.
4 . The computer program product of claim 3 , wherein the one or more exogenous indicators comprise one or more of:
gross domestic product (GDP) growth;
crude oil price; or
stock price of one or more airlines.
5 . The computer program product of claim 1 , wherein forecasting the number of aircraft of the first aircraft type that will be in service during the future window of time comprises:
determining a current number of active aircraft of the first aircraft type currently in service;
determining a number of expected deliveries of new aircraft of the first aircraft type during the future window of time; and
determining a number of retirements of existing aircraft of the first aircraft type during the future window of time.
6 . A method, comprising:
forecasting, for at least a first aircraft type of a plurality of aircraft types and via one or more processors, a number of aircraft of the first aircraft type that will be in service during a future window of time;
forecasting, via the one or more processors, future flight demographics during the future window of time based at least in part on the forecasted number of aircraft;
determining, via the one or more processors, an aircraft configuration for the first aircraft type, the aircraft configuration indicating a number of parts of a first part type on each aircraft of the first aircraft type;
generating predicted part consumption of the first part type during the future window of time by processing the future flight demographics and the aircraft configuration using a machine learning model, wherein the machine learning model is a neural network, and refining the machine learning model comprises using backpropagation to iteratively refine at least one of weights or biases of the neural network;
wherein the predicted part consumption represents a predicted number of parts of the first part type that will be consumed during the future window of time by the aircraft of the first aircraft type;
facilitating reconfiguration of a spare part system, via the one or more processors, based on the predicted part consumption, wherein facilitating the reconfiguration of the spare part system comprises controlling a number of the first part type that is at least one of constructed, ordered,
stored, or shipped based on the predicted number of parts of the first part type that will be consumed during the future window of time;
monitoring actual part consumption of the first part type during the future window of time;
refining the machine learning model based on a comparison of the actual part consumption to the predicted part consumption.
7 . The method of claim 6 , wherein forecasting the future flight demographics comprises predicted time in flight of the aircraft of the first aircraft type during the future window of time.
8 . The method of claim 6 , wherein forecasting the future flight demographics comprises forecasting one or more exogenous indicators.
9 . The method of claim 8 , wherein the one or more exogenous indicators comprise one or more of:
gross domestic product (GDP) growth;
crude oil price; or
stock price of one or more airlines.
10 . The method of claim 6 , wherein forecasting the number of aircraft of the first aircraft type that will be in service during the future window of time comprises:
determining a current number of active aircraft of the first aircraft type currently in service;
determining a number of expected deliveries of new aircraft of the first aircraft type during the future window of time; and
determining a number of retirements of existing aircraft of the first aircraft type during the future window of time.
11 . The method of claim 6 , further comprising:
determining, via the one or more processors and based on the future flight demographics and the aircraft configuration, forecasted part demographics of the first part type associated with the first aircraft type, the forecasted part demographics including (i) a total number of the parts of the first part type that are predicted to be installed on the aircraft of the first aircraft type during the future window of time and (ii) at least one of a total or average flight hours for the parts of the first part type predicted to be installed on the aircraft of the first aircraft type during the future window of time,
wherein generating the predicted part consumption of the first part type comprises inputting the forecasted part demographics into the machine learning model for the machine learning model to output the predicted part consumption.
12 . A system, comprising:
one or more processors; and
a memory storage device communicatively connected to the one or more processors and including program instructions, the one or more processors configured to execute the program instructions to:
forecast a number of aircraft of a first aircraft type that will be in service during a future window of time;
forecast future flight demographics during the future window of time based at least in part on the forecasted number of aircraft and future flight schedules for flights scheduled to occur within the future window of time;
determine an aircraft configuration for the first aircraft type, the aircraft configuration indicating a number of parts of a first part type on each aircraft of the first aircraft type;
generate predicted part consumption of the first part type during the future window of time by processing the future flight demographics and the aircraft configuration using a machine learning model, wherein the machine learning model is a neural network, and refining the machine learning model comprises using backpropagation to iteratively refine at least one of weights or biases of the neural network;
wherein the predicted part consumption represents a predicted number of parts of the first part type that will be consumed during the future window of time by the aircraft of the first aircraft type;
facilitate reconfiguration of a spare part system based on the predicted part consumption, wherein the one or more computer processors are configured to facilitate the reconfiguration of the spare part system by controlling a number of the first part type that is at least one of constructed, ordered, stored, or shipped based on the predicted number of parts of the first part type that will be consumed during the future window of time;
monitor actual part consumption of the first part type during the future window of time;
and refine the machine learning model based on a comparison of the actual part consumption to the predicted part consumption.
13 . The system of claim 12 , wherein the future flight demographics comprise predicted time in flight of the aircraft of the first aircraft type during the future window of time.
14 . The system of claim 12 , wherein the one or more processors are configured to forecast the future flight demographics based in part on
forecasting one or more exogenous indicators.
15 . The system of claim 14 , wherein the one or more exogenous indicators comprise one or more of:
gross domestic product (GDP) growth;
crude oil price; or
stock price of one or more airlines.
16 . The system of claim 12 , wherein the one or more processors are configured to forecast the number of aircraft of the first aircraft type that will be in service during the future window of time by:
determining a current number of active aircraft of the first aircraft type currently in service;
determining a number of expected deliveries of new aircraft of the first aircraft type during the future window of time; and
determining a number of retirements of existing aircraft of the first aircraft type during the future window of time.
17 . The system of claim 12 , wherein the one or more processors are configured to determine, based on the future flight demographics and the aircraft configuration, forecasted part demographics of the first part type associated with the first aircraft type,
wherein the forecasted part demographics include (i) a total number of the parts of the first part type that are predicted to be installed on the aircraft of the first aircraft type during the future window of time and (ii) at least one of a total or average flight hours for the parts of the first part type that are predicted to be installed on the aircraft of the first aircraft type during the future window of time.
18 . The system of claim 17 , wherein the one or more processors are configured to input the forecasted part demographics into the machine learning model for the machine learning model to output the predicted part consumption.