Method for diagnosing a state of wear of an aircraft parking brake
The invention relates to a method of diagnosing the state of wear of an actuator including at least one electric motor. The method comprises the following steps: recording a signal coming from the motor during activation and deactivation of the actuator, the signal including a first maximum during activation and a second maximum during deactivation; executing a windowing algorithm on the signal in order to obtain a standardized dataset including the first and second maximums; using at least one classification method for classifying into at least two classes in order to establish a score for the dataset in each of the classes of the classification method; and evaluating the state of wear of the actuator from the scores of the dataset.
1. A method of diagnosing the state of wear of an actuator including at least one electric motor, the method comprising the steps of:
recording a signal coming from the motor during at least one activation and/or deactivation of the actuator, the signal including a first maximum during activation and a second maximum during deactivation;
executing a windowing algorithm on the signal in order to obtain a standardized dataset including the first and second maximums;
using at least one classification method for classifying into at least two classes in order to establish a score for the dataset in each of the classes of the classification method; and
evaluating the state of wear of the actuator from the scores of the dataset.
2. The method according to claim 1 , wherein the actuator has two electric motors.
3. The method according to claim 1 , wherein the recorded signal is the power supply current of the electric motor.
4. The method according to claim 1 , wherein the windowing algorithm comprises the steps of:
sampling the signal at a determined sampling frequency;
searching for the first and second maximums in the signal; and
defining a first time window around the first maximum and a second time window around the second maximum.
5. The method according to claim 4 , wherein the first and second time windows have time dimensions that are identical.
6. The method according to claim 4 , wherein a weighting factor is applied to the signal.
7. A The method according to claim 1 , wherein the classification method uses the support-vector machine technique.
8. A The method according to claim 1 , wherein the classification method uses the neural network technique.
9. A The method according to claim 1 , wherein the classification method uses the random forest technique.
10. The method according to claim 1 , wherein training of the classification method makes use of a first group of datasets distributed among the classes of said method in order to cover the classification associated with each dataset.
11. The method according to claim 1 , wherein parameters of the classification method are set by using a second group of datasets in order to improve the level of reliability of the scores established by said method.
12. The method according to claim 1 , wherein the classification method is arranged to perform classification into four classes.
13. The method according to claim 12 , wherein the four classes of the classification method are: “new”, “run-in”, “worn”, and “failed”.
14. Diagnosis apparatus for performing the method according to claim 1 , the apparatus comprising recording means for recording the signal and a calculation unit for executing the windowing algorithm and for performing the classification method.
15. The apparatus according to claim 14 , including a connection device suitable for interposing between an electricity network of an aircraft and the parking brake in order to control the electric motor and/or record the signal from the electric motor of the parking brake actuator.