CN115017207A - Aircraft sensor fault association relation determining method and system - Google Patents

Abstract

The invention relates to a method and a system for determining a fault association relation of an aircraft sensor, and relates to the technical field of aircraft sensor faults. The method comprises the steps of obtaining feature data of all feature types of the sensor at each moment, wherein one feature type corresponds to one feature data at one moment; processing the feature data set of each feature type to be associated by adopting an association relationship mining method based on a neighborhood view angle to obtain a fault association relationship of the feature types to be associated of the sensor; the feature data set comprises feature data of feature types at various time instants. The method and the device can improve the accuracy of the incidence relation of the sensor, so that the result obtained when the sensor fault is diagnosed subsequently according to the incidence relation is more accurate.

Description

Aircraft sensor fault association relation determination method and system

Technical Field

The invention relates to the technical field of aircraft sensor faults, in particular to a method and a system for determining an aircraft sensor fault association relation.

Background

During the operation of the aircraft, various sensors are used for their own functions, and the aircraft is in a normal state or an optimal state by monitoring and controlling various parameters in the aircraft operation process. The sensor fault analysis is a key link for maintaining the normal operation of the aircraft, the characteristics inside the sensors or between the sensors are mutually associated and closely coupled, so that the association relationship between the fault characteristics presents the characteristics of nonlinearity and high dimension, the traditional fault association analysis method cannot solve the characteristics of high dimension, complexity of the association relationship and the like of the fault characteristics, the obtained sensor fault association relationship is inaccurate, misjudgment is easy to be generated for fault diagnosis of the sensors, and therefore, the method for identifying the association relationship with complex sensor faults from the high dimension fault characteristics fairly is developed, the accuracy of the association relationship is improved, and the method has important significance.

Disclosure of Invention

The invention aims to provide a method and a system for determining the incidence relation of aircraft sensor faults, which can improve the accuracy of the incidence relation of sensors and enable the results obtained when the sensor faults are diagnosed according to the incidence relation to be more accurate.

In order to achieve the purpose, the invention provides the following scheme:

an aircraft sensor fault correlation determination method comprising:

for any sensor in the aircraft, acquiring feature data of all feature types of the sensor at each moment, wherein one feature type corresponds to one feature data at one moment;

processing the feature data set of each feature type to be associated by adopting an association relationship mining method based on a neighborhood view angle to obtain a fault association relationship of the feature types to be associated of the sensor; the feature data set comprises feature data of feature types at various time instants.

Optionally, the processing the feature data set of each feature type to be associated by using an association mining method based on a neighborhood view to obtain the fault association relationship of the feature types to be associated of the sensor specifically includes:

for any preset neighborhood combination, under the current iteration times, respectively determining neighbor sets of the feature data of each feature type to be associated, which are acquired at the target moment, according to the feature data sets of each feature type to be associated and the preset neighborhood combination; the target time is the time corresponding to the current iteration times, and the preset neighborhood combination comprises the preset number of neighbors to the feature data of each feature type to be associated;

calculating neighborhood mutual information of the feature data of each feature type to be associated acquired at the target moment under the preset neighborhood combination according to the preset neighborhood combination and a neighbor set of the feature data of each feature type to be associated acquired at the target moment, updating the time corresponding to the next iteration time and entering the next iteration until neighborhood mutual information of the feature data of each feature type to be associated acquired at all the times under the preset neighborhood combination is obtained;

determining neighborhood mutual information of each feature type to be associated under the preset neighborhood combination according to neighborhood mutual information of feature data of each feature type to be associated, acquired at all times, under the preset neighborhood combination;

respectively carrying out normalization processing on neighborhood mutual information of each feature type to be associated under all preset neighborhood combinations to obtain normalized neighborhood mutual information;

determining the normalized neighborhood mutual information with the maximum numerical value as a maximum neighborhood coefficient;

and determining the incidence relation of the characteristic types of the sensors according to the maximum neighborhood coefficient.

Optionally, the processing the feature data set of each feature type to be associated by using the association relationship mining method based on the neighborhood view to obtain the fault association relationship of the feature types to be associated with the sensor further includes:

and cleaning the characteristic data of all the characteristic types of the sensor at each moment to obtain the characteristic data of all the characteristic types of the cleaned sensor at each moment.

Optionally, before the processing the feature data set of each feature type to be associated by using the association relationship mining method based on the neighborhood view to obtain the fault association relationship of the feature type to be associated with the sensor, the method further includes:

and classifying the acquired feature data of all the feature types of the sensor at each moment by adopting a clustering algorithm to obtain a feature data set of each feature type.

Optionally, the determining, according to the feature data set of each feature type to be associated and the preset neighborhood combination, a neighbor set of feature data of each feature type to be associated, which is obtained at a target time, specifically includes:

for any one feature type to be associated, calculating the distance between each feature data in the target feature data set and the target data; the target characteristic data set is a characteristic data set of the characteristic type to be associated; the target data are characteristic data which are acquired by the target characteristic data at the target moment in a centralized manner;

and determining a neighbor set of the target data according to the distance between each feature data in the target feature data set and the target data and the preset neighborhood combination.

An aircraft sensor fault correlation determination system comprising:

the acquisition module is used for acquiring the characteristic data of all the characteristic types of the sensor at each moment for any sensor in the aircraft, wherein one characteristic type corresponds to one characteristic data at one moment;

the fault incidence relation determining module is used for processing the feature data sets of the feature types to be associated by adopting an incidence relation mining method based on a neighborhood view angle to obtain the fault incidence relation of the feature types to be associated of the sensor; the feature data set comprises feature data of feature types at various time instants.

Optionally, the fault association relationship determining module specifically includes:

the neighbor determining unit is used for respectively determining neighbor sets of the feature data of the feature types to be associated, which are acquired at a target moment, according to the feature data sets of the feature types to be associated and the preset neighbor combinations under the current iteration times for any preset neighbor combinations; the target time is the time corresponding to the current iteration times, and the preset neighborhood combination comprises the preset number of neighbors to the feature data of each feature type to be associated;

a time neighborhood mutual information determining unit, configured to calculate neighborhood mutual information of the feature data of each feature type to be associated acquired at the target time under the preset neighborhood combination according to the preset neighborhood combination and a neighbor set of the feature data of each feature type to be associated acquired at the target time, update a time corresponding to a next iteration number and enter a next iteration until neighborhood mutual information of the feature data of each feature type to be associated acquired at all times under the preset neighborhood combination is obtained;

a neighborhood mutual information determining unit, configured to determine neighborhood mutual information of each feature type to be associated in the preset neighborhood combination according to neighborhood mutual information of feature data of each feature type to be associated, acquired at all times, in the preset neighborhood combination;

the normalization unit is used for respectively carrying out normalization processing on the neighborhood mutual information of each feature type to be correlated under all preset neighborhood combinations to obtain normalized neighborhood mutual information;

the maximum neighborhood coefficient determining unit is used for determining the normalized neighborhood mutual information with the maximum numerical value as a maximum neighborhood coefficient;

and the fault incidence relation determining unit is used for determining the incidence relation of the characteristic types of the sensors according to the maximum neighborhood coefficient.

Optionally, the system for determining the aircraft sensor fault association relationship further includes:

and the cleaning module is used for cleaning the characteristic data of all the characteristic types of the sensor at each moment to obtain the characteristic data of all the characteristic types of the cleaned sensor at each moment.

Optionally, the aircraft sensor fault association relationship determining system further includes:

and the characteristic data set determining module is used for classifying the acquired characteristic data of all the characteristic types of the sensor at each moment by adopting a clustering algorithm to obtain a characteristic data set of each characteristic type.

Optionally, the neighbor determining unit specifically includes:

the distance calculation subunit is used for calculating the distance between each feature data in the target feature data set and the target data for any feature type to be associated; the target characteristic data set is a characteristic data set of the characteristic type to be associated; the target data are characteristic data which are acquired by the target characteristic data at the target moment in a centralized manner;

and the neighbor set calculating subunit is used for determining the neighbor set of the target data according to the distance between each feature data in the target feature data set and the target data and the preset neighborhood combination.

According to the specific embodiment provided by the invention, the invention discloses the following technical effects: the method includes the steps that characteristic data of all characteristic types of a sensor at each moment are obtained, and one characteristic type corresponds to one characteristic data at one moment; processing the feature data set of each feature type to be associated by adopting an association relation mining method based on a neighborhood view angle to obtain a fault association relation of the feature types to be associated of the sensor; the feature data set comprises feature data of feature types at all times, and the fault association relation of the sensor is determined by adopting an association relation mining method based on a neighborhood view angle, so that the accuracy of the association relation of the sensor can be improved, and the result obtained when the sensor fault is diagnosed according to the association relation subsequently is more accurate.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.

Fig. 1 is a flowchart of a method for determining an aircraft sensor fault association relationship according to an embodiment of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.

The embodiment of the invention discloses a method for determining a fault association relation of an aircraft sensor, which comprises the following specific steps as shown in figure 1:

step 101: for any one sensor in the aircraft, acquiring feature data of all feature types of the sensor at each moment, wherein one feature type corresponds to one feature data at one moment. Aircraft sensors include humidity sensor data, ground contact sensor data, temperature sensor data, angular rate sensor data, airspeed sensor data, stall sensor data, and static-pressure-hole sensor data, among others.

Step 102: processing the feature data set of each feature type to be associated by adopting an association relation mining method based on a neighborhood view angle to obtain a fault association relation of the feature types to be associated of the sensor; the feature data set comprises feature data of feature types at various time instants.

In practical applications, step 102 specifically includes:

step 201: for any one predetermined neighborhood combination (k) _x ，k _y ) Respectively determining neighbor sets of the feature data of each feature type to be associated, which are acquired at a target moment, according to the feature data set of each feature type to be associated and the preset neighborhood combination under the current iteration times; the target time is the time corresponding to the current iteration times, and the preset neighborhood combination comprises the number of neighbors preset for the feature data of each feature type to be associated.

In practical application, the determining, according to the feature data set of each feature type to be associated and the preset neighborhood combination, a neighbor set of feature data of each feature type to be associated, which is acquired at a target time, specifically includes:

for any one feature type to be associated, calculating the distance between each feature data in the target feature data set and the target data; the target characteristic data set is a characteristic data set of the characteristic type to be associated; the target data is characteristic data acquired by the target characteristic data set at the target moment.

And determining a neighbor set of the target data according to the distance between each feature data in the target feature data set and the target data and the preset neighborhood combination. For example, when there are two feature types to be associated, set as X and Y to obtain a feature data set S of X _X ＝{X ₁ ,...,X _n Feature data set S of } and Y _Y ＝{Y ₁ ,...,Y _n }，X _n Denotes S _x N-th characteristic data, Y _n Denotes S _Y The nth characteristic data are shown as the data acquired at the same time, and the neighbor sets are respectively shown as

And

also called neighborhood particles, the number of neighbors is k respectively _x And k _y The fault sample set of any characteristic type is S _X ＝{(X ₁ ,Y ₁ ),...,(X _n ,Y _n ) From the joint distribution (X, Y).

Step 202: calculating neighborhood mutual information of the feature data of each feature type to be associated acquired at the target moment under the preset neighborhood combination according to the preset neighborhood combination and the neighbor set of the feature data of each feature type to be associated acquired at the target moment, updating the time corresponding to the next iteration time and entering the next iteration until the feature data of each feature type to be associated acquired at all the times are acquired at the preset neighbor setNeighborhood mutual information under domain combination. Specifically, X obtained at the target time is calculated according to formula (1) _i And Y _i In (k) _x ，k _y ) Neighborhood mutual information under neighborhood combination

Wherein k is _x And k _y Are each X _i ，Y _i The number of the corresponding neighbors is,

and

are each X _i ，Y _i And (X) _i ，Y _i ) The neighborhood of grains.

Step 203: and determining neighborhood mutual information of each feature type to be associated under the preset neighborhood combination according to neighborhood mutual information of the feature data of each feature type to be associated under all the moments, wherein the neighborhood mutual information is obtained under the preset neighborhood combination. Specifically, the value at (k) is calculated according to formula (2) _x ，k _y ) Neighborhood mutual information of feature type X and feature type Y under neighborhood combination

Step 204: and respectively carrying out normalization processing on the neighborhood mutual information of each feature type to be associated under all the preset neighborhood combinations to obtain normalized neighborhood mutual information.

Step 205: and determining the normalized neighborhood mutual information with the maximum value as the maximum neighborhood coefficient. Specifically, the maximum neighborhood coefficient mnc(s) is calculated according to formula (3).

Wherein the content of the first and second substances,

is a formula for calculating normalization, NB (n) is all the cases of the neighborhood combination, k is more than or equal to 1 _x k _y ≤o(n ^α ),0＜α＜1。

Step 206: and determining the association relation of the characteristic types of the sensors according to the maximum neighborhood coefficient.

In practical application, step 206 specifically includes: when the maximum neighborhood coefficient is in a first set range, namely deviation 1, the correlation is judged, and when the maximum neighborhood coefficient is in a second set range, namely deviation 0, the correlation is judged not to be correlated.

In practical applications, step 102 further includes:

and cleaning the characteristic data of all the characteristic types of the sensor at each moment to obtain the characteristic data of all the characteristic types of the cleaned sensor at each moment.

In practical applications, before step 102, the method further includes:

and classifying the acquired feature data of all the feature types of the sensor at each moment by adopting a clustering algorithm to obtain a feature data set of each feature type.

The embodiment of the invention provides an aircraft sensor fault association relation determining system aiming at the method, which comprises the following steps:

the acquisition module is used for acquiring the characteristic data of all the characteristic types of the sensor at each moment for any sensor in the aircraft, wherein one characteristic type corresponds to one characteristic data at one moment.

The fault incidence relation determining module is used for processing the feature data sets of the feature types to be associated by adopting an incidence relation mining method based on a neighborhood view angle to obtain the fault incidence relation of the feature types to be associated of the sensor; the feature data set comprises feature data of feature types at various time instants.

As an optional implementation manner, the fault association relationship determining module specifically includes:

the neighbor determining unit is used for respectively determining neighbor sets of the feature data of each feature type to be associated, which are acquired at a target moment, according to the feature data sets of each feature type to be associated and the preset neighbor combinations for any preset neighbor combinations under the current iteration times; the target time is a time corresponding to the current iteration times, and the preset neighborhood combination comprises the number of neighbors preset for the feature data of each feature type to be associated.

And the time neighborhood mutual information determining unit is used for calculating neighborhood mutual information of the feature data of each feature type to be associated, which is acquired at the target moment, under the preset neighborhood combination according to the preset neighborhood combination and the neighbor set of the feature data of each feature type to be associated, which is acquired at the target moment, updating the time corresponding to the next iteration number and entering the next iteration until neighborhood mutual information of the feature data of each feature type to be associated, which is acquired at all the times, under the preset neighborhood combination is obtained.

And the neighbor neighborhood mutual information determining unit is used for determining the neighborhood mutual information of each feature type to be associated under the preset neighborhood combination according to the neighborhood mutual information of the feature data of each feature type to be associated, which is acquired at all times, under the preset neighborhood combination.

And the normalization unit is used for respectively carrying out normalization processing on the neighborhood mutual information of each feature type to be associated under all the preset neighborhood combinations to obtain the normalized neighborhood mutual information.

And the maximum neighborhood coefficient determining unit is used for determining the normalized neighborhood mutual information with the maximum numerical value as the maximum neighborhood coefficient.

And the fault incidence relation determining unit is used for determining the incidence relation of the characteristic types of the sensors according to the maximum neighborhood coefficient.

As an optional implementation, the aircraft sensor fault correlation determination system further includes:

and the cleaning module is used for cleaning the characteristic data of all the characteristic types of the sensor at each moment to obtain the characteristic data of all the characteristic types of the cleaned sensor at each moment.

As an optional implementation, the aircraft sensor fault correlation determination system further includes:

and the characteristic data set determining module is used for classifying the acquired characteristic data of all the characteristic types of the sensor at each moment by adopting a clustering algorithm to obtain a characteristic data set of each characteristic type.

As an optional implementation manner, the neighbor determining unit specifically includes:

the distance calculation subunit is used for calculating the distance between each feature data in the target feature data set and the target data for any feature type to be associated; the target characteristic data set is a characteristic data set of the characteristic type to be associated; the target data is characteristic data acquired by the target characteristic data set at the target moment.

And the neighbor set calculating subunit is used for determining the neighbor set of the target data according to the distance between each feature data in the target feature data set and the target data and the preset neighborhood combination.

The invention has the following technical effects:

1. the maximum neighborhood coefficient is based on a data-driven incidence relation mining method, and has better data environment adaptability compared with a measurement method based on a normal distribution hypothesis structure. The coefficient considers the local neighborhood structure of the sample, and has better robustness compared with a method based on grid division. The maximum neighborhood coefficient can fairly identify incidence relations of various forms, measures the incidence relation among fault characteristics by using the maximum neighborhood coefficient, and can accurately identify the fault type and the fault characteristics of the sensor, thereby improving the monitoring level of the sensor.

2. The method can reduce manual interference, does not need special technicians to detect, greatly improves the working efficiency, and greatly facilitates the maintenance of the technicians due to the perfect detection device.

In the present specification, the embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the description of the method part.

The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.

Claims (10)

1. An aircraft sensor fault association relationship determination method, comprising:

for any sensor in the aircraft, acquiring feature data of all feature types of the sensor at each moment, wherein one feature type corresponds to one feature data at one moment;

processing the feature data set of each feature type to be associated by adopting an association relation mining method based on a neighborhood view angle to obtain a fault association relation of the feature types to be associated of the sensor; the feature data set comprises feature data of feature types at various time instants.

2. The aircraft sensor fault association relationship determination method according to claim 1, wherein the processing of the feature data set of each feature type to be associated by using an association relationship mining method based on a neighborhood view angle to obtain the fault association relationship of the feature types to be associated with the sensor specifically comprises:

for any preset neighborhood combination, under the current iteration times, respectively determining neighbor sets of the feature data of each feature type to be associated, which are acquired at the target moment, according to the feature data sets of each feature type to be associated and the preset neighborhood combination; the target time is a time corresponding to the current iteration times, and the preset neighborhood combination comprises the number of neighbors preset for the feature data of each feature type to be associated;

calculating neighborhood mutual information of the feature data of each feature type to be associated, which is acquired at the target moment, under the preset neighborhood combination according to the preset neighborhood combination and a neighbor set of the feature data of each feature type to be associated, which is acquired at the target moment, and updating the time corresponding to the next iteration time and entering the next iteration until neighborhood mutual information of the feature data of each feature type to be associated, which is acquired at all the time, under the preset neighborhood combination is acquired;

determining neighborhood mutual information of each feature type to be associated under the preset neighborhood combination according to neighborhood mutual information of feature data of each feature type to be associated, acquired at all times, under the preset neighborhood combination;

respectively carrying out normalization processing on neighborhood mutual information of each feature type to be associated under all preset neighborhood combinations to obtain normalized neighborhood mutual information;

determining the normalized neighborhood mutual information with the maximum numerical value as a maximum neighborhood coefficient;

and determining the incidence relation of the characteristic types of the sensors according to the maximum neighborhood coefficient.

3. The aircraft sensor fault association relationship determination method according to claim 1, wherein the processing of the feature data set of each feature type to be associated by using the association relationship mining method based on the neighborhood view angle to obtain the fault association relationship of the feature type to be associated with the sensor further comprises:

and cleaning the characteristic data of all the characteristic types of the sensor at each moment to obtain the characteristic data of all the characteristic types of the cleaned sensor at each moment.

4. The aircraft sensor fault association relationship determination method according to claim 1, wherein before the processing the feature data set of each feature type to be associated by using the association relationship mining method based on the neighborhood view angle to obtain the fault association relationship of the feature type to be associated with the sensor, the method further comprises:

and classifying the acquired feature data of all the feature types of the sensor at each moment by adopting a clustering algorithm to obtain a feature data set of each feature type.

5. The aircraft sensor fault association relationship determining method according to claim 2, wherein the determining, according to the feature data set of each feature type to be associated and the preset neighborhood combination, a neighbor set of feature data of each feature type to be associated, acquired at a target time, respectively, specifically includes:

for any one feature type to be associated, calculating the distance between each feature data in the target feature data set and the target data; the target characteristic data set is a characteristic data set of the characteristic type to be associated; the target data is characteristic data which is obtained by concentrating the target characteristic data at the target moment;

and determining a neighbor set of the target data according to the distance between each feature data in the target feature data set and the target data and the preset neighborhood combination.

6. An aircraft sensor fault correlation determination system, comprising:

the acquisition module is used for acquiring the characteristic data of all the characteristic types of the sensor at each moment for any sensor in the aircraft, wherein one characteristic type corresponds to one characteristic data at one moment;

the fault incidence relation determining module is used for processing the feature data sets of the feature types to be associated by adopting an incidence relation mining method based on a neighborhood view angle to obtain the fault incidence relation of the feature types to be associated of the sensor; the feature data set comprises feature data of feature types at various time instants.

7. The aircraft sensor fault association determination system of claim 6, wherein the fault association determination module specifically comprises:

the neighbor determining unit is used for respectively determining neighbor sets of the feature data of each feature type to be associated, which are acquired at a target moment, according to the feature data sets of each feature type to be associated and the preset neighbor combinations for any preset neighbor combinations under the current iteration times; the target time is the time corresponding to the current iteration times, and the preset neighborhood combination comprises the preset number of neighbors to the feature data of each feature type to be associated;

a time neighborhood mutual information determining unit, configured to calculate neighborhood mutual information of the feature data of each feature type to be associated acquired at the target time under the preset neighborhood combination according to the preset neighborhood combination and a neighbor set of the feature data of each feature type to be associated acquired at the target time, update a time corresponding to a next iteration number and enter a next iteration until neighborhood mutual information of the feature data of each feature type to be associated acquired at all times under the preset neighborhood combination is obtained;

a neighborhood mutual information determining unit, configured to determine neighborhood mutual information of each feature type to be associated in the preset neighborhood combination according to neighborhood mutual information of feature data of each feature type to be associated, acquired at all times, in the preset neighborhood combination;

the normalization unit is used for respectively carrying out normalization processing on the neighborhood mutual information of each feature type to be correlated under all preset neighborhood combinations to obtain normalized neighborhood mutual information;

the maximum neighborhood coefficient determining unit is used for determining the normalized neighborhood mutual information with the maximum numerical value as a maximum neighborhood coefficient;

and the fault incidence relation determining unit is used for determining the incidence relation of the characteristic types of the sensors according to the maximum neighborhood coefficient.

8. The aircraft sensor fault correlation determination system of claim 6, further comprising:

and the cleaning module is used for cleaning the characteristic data of all the characteristic types of the sensor at each moment to obtain the characteristic data of all the characteristic types of the cleaned sensor at each moment.

9. The aircraft sensor fault correlation determination system of claim 6, further comprising:

and the characteristic data set determining module is used for classifying the acquired characteristic data of all the characteristic types of the sensor at each moment by adopting a clustering algorithm to obtain a characteristic data set of each characteristic type.

10. The aircraft sensor fault association relationship determination system according to claim 7, wherein the neighbor determination unit specifically includes:

the distance calculation subunit is used for calculating the distance between each feature data in the target feature data set and the target data for any feature type to be associated; the target characteristic data set is a characteristic data set of the characteristic type to be associated; the target data are characteristic data which are acquired by the target characteristic data at the target moment in a centralized manner;

and the neighbor set calculating subunit is used for determining the neighbor set of the target data according to the distance between each feature data in the target feature data set and the target data and the preset neighborhood combination.

Images