Deep learning system for real time maximum sound pressure level prediction

A method for predicting sequential LAmax sound pressure levels generated by an aircraft, the method comprising:

• receiving, during a first portion of a flight by the aircraft, continuous parameter logging data from sensors comprising noise sensors including a first set of the sequential LAmax sound pressure levels recorded by a first consecutive set of microphones in the noise sensors along a flight path during the first portion of the flight;
• feature engineering, in real time in a controller on the aircraft, the continuous parameter logging data into selected parameters comprising stable parameters within a time instance comprising a smallest slant distance between the aircraft and a noise sensor in the noise sensors;
• transforming, in real time in a controller on the aircraft, the selected parameters into a training dataset for a deep learning model;
• training, using the training dataset in real time in a controller on the aircraft, the deep learning model for predicting sequential LAmax sound pressure levels generated by the aircraft over a second portion of the flight;
• and
• subsequently, using the deep learning model, in real time in a controller on the aircraft, predicting, with a mean error of less than 2 dBA with a standard deviation of 0.2 dBA, a second set of the sequential LAmax sound pressure levels that will be recorded by a second consecutive set of the microphones in the noise sensors along the second portion of the flight path during the flight of the aircraft.

A deep learning system that comprises a computer system that comprises a controller configured to in real time:

• receive, in a first portion of a flight by an aircraft, continuous parameter logging data from at least:
• aircraft, environmental, and noise, sensors including a first set of sequential LAmax sound pressure levels recorded by a first consecutive set of microphones in the noise sensors along a flight path during a first portion of a flight of the aircraft;
• reduce the continuous parameter logging data into selected parameters comprising stable parameters within a time instance comprising a smallest slant distance between the aircraft and a noise sensor in the noise sensors;
• transform the selected parameters into a training dataset for a set of deep learning models;
• train the set of deep learning models to predict sequential LAmax sound pressure levels generated by the aircraft on a second portion of the flight;
• and
• predict, with a mean error of less than 2 dBA with a standard deviation of 0.2 dBA, a second set of the sequential LAmax sound pressure levels that will be recorded by a second consecutive set of the microphones along the flight path during the flight of the aircraft, wherein the second set of the sequential LAmax sound pressure levels is predicted using the set of deep learning models after training the set of deep learning models using the training dataset.

A computer program product configured to predict in real time sequential LAmax sound pressure levels, wherein the computer program product comprising a computer readable storage medium that comprises program instructions embodied therewith, the program instructions configured for execution by a computer system to cause the computer system to:

• receive, during a first portion of a flight of an aircraft, continuous parameter logging data from sensors that comprise noise sensors;
• feature engineering the continuous parameter logging data into selected parameters comprising stable parameters within a time instance comprising a smallest slant distance between the aircraft and a noise sensor in the noise sensors;
• transform the selected parameters into a training dataset for a deep learning model;
• train, based upon the training dataset, a set of deep learning models to predict the sequential LAmax sound pressure levels generated by the aircraft on a second portion of the flight;
• and
• predict, with a mean error of less than 2 dBA with a standard deviation of 0.2 dBA, a second set of the sequential LAmax sound pressure levels that will be recorded by a second consecutive set of noise sensors along the second portion of the flight during the flight of the aircraft, wherein a prediction of the second set of the sequential LAmax sound pressure levels is a product of the set of deep learning models trained with the training dataset.