CN117349797A - Aircraft fault detection method and system based on artificial intelligence - Google Patents
Aircraft fault detection method and system based on artificial intelligence Download PDFInfo
- Publication number
- CN117349797A CN117349797A CN202311641840.1A CN202311641840A CN117349797A CN 117349797 A CN117349797 A CN 117349797A CN 202311641840 A CN202311641840 A CN 202311641840A CN 117349797 A CN117349797 A CN 117349797A
- Authority
- CN
- China
- Prior art keywords
- model
- data
- fault detection
- parameter
- gate
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 50
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 21
- 230000006870 function Effects 0.000 claims abstract description 54
- 230000007246 mechanism Effects 0.000 claims abstract description 17
- 230000007787 long-term memory Effects 0.000 claims abstract description 13
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 10
- 238000000034 method Methods 0.000 claims description 37
- 230000008569 process Effects 0.000 claims description 23
- 238000007781 pre-processing Methods 0.000 claims description 18
- 230000004913 activation Effects 0.000 claims description 6
- 230000008859 change Effects 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 230000015654 memory Effects 0.000 claims description 3
- 238000012887 quadratic function Methods 0.000 claims description 3
- 230000006403 short-term memory Effects 0.000 claims description 3
- 238000005457 optimization Methods 0.000 abstract description 7
- 238000010276 construction Methods 0.000 abstract description 4
- 230000007774 longterm Effects 0.000 abstract description 4
- 230000008034 disappearance Effects 0.000 description 4
- 238000004880 explosion Methods 0.000 description 4
- 238000010845 search algorithm Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/27—Regression, e.g. linear or logistic regression
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/082—Learning methods modifying the architecture, e.g. adding, deleting or silencing nodes or connections
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Evolutionary Biology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an aircraft fault detection method and system based on artificial intelligence. The invention belongs to the technical field of fault detection, in particular to an aircraft fault detection method and system based on artificial intelligence, wherein the scheme is based on local extremum searching and constructing upper and lower limit curves to filter noise, and a decomposition result is ensured based on monotonic function judgment; capturing a long-term dependency relationship of data by adopting a long-term memory network predictor model based on construction, and predicting result probability distribution by adopting a Gaussian regression model based on construction; the search accuracy and the global optimization capacity of the algorithm are improved based on the definition of a global optimal solution search mechanism and a movement strategy, the search coverage rate is improved based on the division of subspaces, and the search dimension is reduced.
Description
Technical Field
The invention relates to the technical field of fault detection, in particular to an aircraft fault detection method and system based on artificial intelligence.
Background
An aircraft fault detection method based on artificial intelligence is a technology utilizing machine learning and data analysis, and can detect possible faults or abnormal conditions of an aircraft by monitoring and analyzing aircraft sensor data. However, the original signal has noise interference and inaccurate decomposed signal; the general fault detection model has the problems of gradient disappearance and gradient explosion and weak robustness, so that the model performance is low; the search algorithm has the problems of blind selection and trapping in local optimum due to a single search mechanism.
Disclosure of Invention
Aiming at the situation, in order to overcome the defects of the prior art, the invention provides an aircraft fault detection method and system based on artificial intelligence, and aims at solving the problems that noise interference exists in an original signal and a decomposed signal is inaccurate; aiming at the problems of low model performance caused by gradient disappearance and gradient explosion and weak robustness of a general fault detection model, the scheme adopts a long-term dependency relationship of data captured based on a long-term memory network predictor model constructed, so that the whole model is easier to train and optimize, and the model performance is improved by more flexibly adapting to various uncertainties and improving the model robustness based on probability distribution of a Gaussian regression sub model prediction result; aiming at the problems of blind selection and local optimization caused by a single search mechanism in a search algorithm, the scheme improves the search precision and global optimization capacity of the algorithm based on defining a global optimal solution search mechanism and a movement strategy, accelerates the convergence rate of the algorithm, improves the search coverage rate based on dividing subspaces and reduces the search dimension, thereby improving the operation efficiency of the algorithm.
The technical scheme adopted by the invention is as follows: the invention provides an aircraft fault detection method based on artificial intelligence, which comprises the following steps:
step S1: collecting data;
step S2: data preprocessing, namely constructing an upper limit curve and a lower limit curve based on adjacent values in comparison fluctuation and connecting local extreme points by using spline lines, and judging a decomposition signal based on a monotonic function so as to construct a sample data set;
step S3: establishing a hybrid aircraft fault detection model, and establishing the hybrid aircraft fault detection model based on the long-short-term memory network prediction sub-model and the Gaussian process regression sub-model;
step S4: model parameter searching, namely updating parameter positions based on a defined global optimal solution searching mechanism and a movement strategy, and judging a searching result based on a dividing subspace, a local threshold value, a global threshold value and the maximum iteration number;
step S5: and (5) running in real time.
Further, in step S1, the data acquisition is to acquire historical operation data signals and corresponding operation states of the aircraft.
Further, in step S2, the data preprocessing specifically includes the following steps:
step S21: setting data set, selecting historical operation data signal as C b The method comprises the steps of carrying out a first treatment on the surface of the By at C b Searching local maximum values and minimum values in a mode of comparing adjacent values in all the fluctuation, wherein the local minimum values and the maximum values respectively represent minimum and maximum data points on a local fluctuation scale; by connecting local extreme points using spline lines, upper limit curves e are respectively constructed up And a lower limit curve e low ；
Step S22: calculating a local average value m e And calculates a local average value and C b The formula used is as follows:
；
step S23: repeating the above operation until the obtained difference becomes a monotonic function;
step S24: after the decomposition signal is ended and the n1 data signal IMF and a monotonic function are obtained, C b Consists of the following contents:
；
where j1 is the signal index, r m Is the remainder, i.e., the trend portion remaining in the decomposition process;
step S25: and constructing a data set, namely constructing the data set based on the decomposed signals, and taking the running state corresponding to the data as a label.
Further, in step S3, the building a hybrid aircraft fault detection model specifically includes the following steps:
step S31: constructing a long-term and short-term memory network predictor model, wherein the states of all gates are input X at the current moment k through S-shaped units k And the output h of the last instant k-1 k-1 Determining; the input gate decides whether new state information is received; the forget gate is responsible for forgetting the previous state in the hidden layer; the output gate determines whether the information calculated by the model is output, and the steps include:
step S311: the input gate status information and the state of the input gate are updated using the following formulas:
；
wherein p is k Representing input gate state information, tanh () represents hyperbolic tangent activation function, ω s Weights representing input gate states, u s Weights indicating the previous state of the input door b s Bias amount indicating input gate state, i k Representing the state of the input door, sigma s Representing a sigmoid activation function, w i Representing the weight of the input gate, u i Indicating the weight of the input door at the previous moment, b i Representing the offset of the input gate;
step S312: updating the states of the forgetting door and the forgetting door, wherein the following formula is used:
；
wherein f k Indicating forgetful door omega f Weight indicating forgetting gate, u f Weight indicating the moment before forgetting to leave the door, b f Representing the bias of the forgetting gate, s k Indicating the state of the forgotten door,representing convolution operations, s k-1 Representing a previous state of the forget gate;
step S313: the states of the output gate and the output gate are updated using the following formulas:
；
in the formula, o k Representing the output gate, omega o Representing the weight of the output gate, u o Indicating the weight of the output door at the previous moment, b o Indicating the offset of the output gate, h k Representing the status of the output gate;
step S32: the method for constructing the Gaussian process regression sub-model comprises the following steps of:
step S321: the GPR probability distribution is defined using the formula:
；
where f () is an objective function, l is an input, and l' is a predicted value; GPR () is a gaussian regression process; m (L) and L (L, L') are a mean function and a covariance function, respectively;
step S322: defining a kernel function k M The formula used is as follows:
；
wherein, gamma is a super parameter reflecting smoothness;is Bessel function, sigma 2 Is the noise variance in the regression process, ρ is the hyper-parameter that adjusts the distance scale,is a gamma function;
step S323: adding the kernel functions of different scales to obtain a rational quadratic function k of the popular kernel function RQ The formula used is as follows:
；
where a is the relative weight reflecting the scale change,andis a hyper-parameter that affects the axis scaling;
step S324: the total joint prior distribution of the known output y and the predicted output y' is defined using the following formula:
；
wherein I is n Is a unit matrix with a diagonal line of 1, and L (L, L) and L (L ', L') are respectively an input covariance function and a covariance function of an output value;
step S325: by calculating the condition distributionThe output is predicted using the following formula:
；
wherein T is a transpose operation;
step S326: and taking the weighted sum of the long-term memory neural network and the output of the Gaussian process regression model as a final prediction result.
Further, in step S4, the model parameter search specifically includes the following steps:
step S41: defining a globally optimal solution guide search mechanism, wherein the following formula is used:
；
wherein x is a parameter position, I and j are different individuals, it is the iteration number, r is a random number between 0 and 1, and P is a global optimal position; fl is the distance traveled by the individual; m is the individual's historical optimal position;
step S42: defining a movement strategy, and introducing a Levy flight search strategy to replace random search, wherein the following formula is adopted:
；
wherein, gamma s Is a randomly changed symbol, and takes the value ofThe method comprises the steps of carrying out a first treatment on the surface of the a is a step factor, β is a constant that controls the step change, Γ () is a gamma function;
step S43: initializing, decomposing a parameter search space, initializing a parameter position based on a subspace, and taking the model performance established based on parameters as a parameter fitness value;
step S44: updating the parameter position by using the following formula:
；
wherein, AP is a movement threshold;
step S45: judging, wherein a local threshold value, a global threshold value and a maximum iteration number are preset; if the individual fitness value of the subspace is higher than a local threshold, establishing a model based on the optimal individual of each subspace, if the performance of the model is higher than the global optimum, establishing the model at the output parameter position, and if the performance of the model is lower than the global optimum, searching the locally optimal individual with the lowest subspace; if the maximum iteration times are reached, re-dividing the subspace and initializing the parameter positions to perform parameter searching; otherwise, the search is continued.
Further, in step S5, the real-time operation is to collect the aircraft operation data in real time and input the data into a model, and the model outputs a weighted prediction result.
The invention provides an aircraft fault detection system based on artificial intelligence, which comprises a data acquisition module, a data preprocessing module, a hybrid aircraft fault detection model building module, a model parameter searching module and a real-time operation module;
the data acquisition module is used for collecting and acquiring historical operation data signals and corresponding operation states of the aircraft and sending the data to the data preprocessing module;
the data preprocessing module receives the data sent by the data acquisition module, constructs upper and lower limit curves based on adjacent values in comparison fluctuation and local extreme points connected by spline lines, judges and decomposes signals based on a monotonic function, thereby constructing a sample data set, and sends the data to the mixed aircraft fault detection model building module;
the mixed aircraft fault detection model building module receives the data sent by the data preprocessing module, builds a long-term memory network predictor model and a Gaussian process regression sub model to build a mixed aircraft fault detection model, and sends the data to the model parameter searching module;
the model parameter searching module receives data sent by the hybrid aircraft fault detection model building module, updates parameter positions based on a defined global optimal solution searching mechanism and a movement strategy, sets a local threshold value, a global threshold value and a maximum iteration number to judge a searching result based on a division subspace, and sends the data to the real-time operation module;
the real-time operation module receives the data sent by the model parameter search module, acquires the aircraft operation data in real time, inputs the aircraft operation data into the model, and outputs a weighted prediction result.
By adopting the scheme, the beneficial effects obtained by the invention are as follows:
(1) Aiming at the problems that noise interference exists in an original signal and a decomposed signal is inaccurate, the scheme filters noise based on local extremum searching and upper and lower limit curve construction, and ensures the accuracy of a decomposed result based on monotonic function judgment.
(2) Aiming at the problems of low model performance caused by gradient disappearance and gradient explosion and weak robustness of a general fault detection model, the scheme adopts a long-term dependency relationship based on constructing long-term memory network predictive sub-models to capture data, so that the whole model is easier to train and optimize, and the model performance is improved by more flexibly adapting to various uncertainties and improving the model robustness based on constructing Gaussian regression sub-model predictive result probability distribution.
(3) Aiming at the problems of blind selection and local optimization caused by a single search mechanism in a search algorithm, the scheme improves the search precision and global optimization capacity of the algorithm based on defining a global optimal solution search mechanism and a movement strategy, accelerates the convergence rate of the algorithm, improves the search coverage rate based on dividing subspaces and reduces the search dimension, thereby improving the operation efficiency of the algorithm.
Drawings
FIG. 1 is a schematic flow chart of an aircraft fault detection method based on artificial intelligence;
FIG. 2 is a schematic diagram of an artificial intelligence based aircraft fault detection system provided by the present invention;
FIG. 3 is a flow chart of step S2;
fig. 4 is a flow chart of step S4.
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all embodiments of the invention; all other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the description of the present invention, it should be understood that the terms "upper," "lower," "front," "rear," "left," "right," "top," "bottom," "inner," "outer," and the like indicate orientation or positional relationships based on those shown in the drawings, merely to facilitate description of the invention and simplify the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus should not be construed as limiting the invention.
Referring to fig. 1, the method for detecting an aircraft fault based on artificial intelligence provided by the invention comprises the following steps:
step S1: collecting data;
step S2: data preprocessing, namely constructing an upper limit curve and a lower limit curve based on adjacent values in comparison fluctuation and connecting local extreme points by using spline lines, and judging a decomposition signal based on a monotonic function so as to construct a sample data set;
step S3: establishing a hybrid aircraft fault detection model, and establishing the hybrid aircraft fault detection model based on the long-short-term memory network prediction sub-model and the Gaussian process regression sub-model;
step S4: model parameter searching, namely updating parameter positions based on a defined global optimal solution searching mechanism and a movement strategy, and judging a searching result based on a dividing subspace, a local threshold value, a global threshold value and the maximum iteration number;
step S5: and (5) running in real time.
In the second embodiment, referring to fig. 1, the embodiment is based on the above embodiment, and in step S1, the data acquisition is to acquire historical operation data signals and corresponding operation states of the aircraft.
Embodiment three, referring to fig. 1 and 3, based on the above embodiment, in step S2, the data preprocessing specifically includes the following steps:
step S21: setting data set, selecting historical operation data signal as C b The method comprises the steps of carrying out a first treatment on the surface of the By at C b Searching local maximum values and minimum values in a mode of comparing adjacent values in all the fluctuation, wherein the local minimum values and the maximum values respectively represent minimum and maximum data points on a local fluctuation scale; connecting local extreme points by using spline lines to respectively constructLimit curve e up And a lower limit curve e low ；
Step S22: calculating a local average value m e And calculates a local average value and C b The formula used is as follows:
；
step S23: repeating the above operation until the obtained difference becomes a monotonic function;
step S24: after the decomposition signal is ended and the n1 data signal IMF and a monotonic function are obtained, C b Consists of the following contents:
；
where j1 is the signal index, r m Is the remainder, i.e., the trend portion remaining in the decomposition process;
step S25: and constructing a data set, namely constructing the data set based on the decomposed signals, and taking the running state corresponding to the data as a label.
By executing the above operation, aiming at the problems that noise interference exists in an original signal and a decomposed signal is inaccurate, the scheme is used for filtering noise based on local extremum searching and constructing upper and lower limit curves, and ensuring the accuracy of a decomposed result based on monotonic function judgment.
Fourth embodiment, referring to fig. 1, the method for establishing a hybrid aircraft fault detection model in step S3 specifically includes the following steps:
step S31: constructing a long-term and short-term memory network predictor model, wherein the states of all gates are input X at the current moment k through S-shaped units k And the output h of the last instant k-1 k-1 Determining; the input gate decides whether new state information is received; the forget gate is responsible for forgetting the previous state in the hidden layer; the output gate determines whether the information calculated by the model is output, and the steps include:
step S311: the input gate status information and the state of the input gate are updated using the following formulas:
；
wherein p is k Representing input gate state information, tanh () represents hyperbolic tangent activation function, ω s Weights representing input gate states, u s Weights indicating the previous state of the input door b s Bias amount indicating input gate state, i k Representing the state of the input door, sigma s Representing a sigmoid activation function, w i Representing the weight of the input gate, u i Indicating the weight of the input door at the previous moment, b i Representing the offset of the input gate;
step S312: updating the states of the forgetting door and the forgetting door, wherein the following formula is used:
；
wherein f k Indicating forgetful door omega f Weight indicating forgetting gate, u f Weight indicating the moment before forgetting to leave the door, b f Representing the bias of the forgetting gate, s k Indicating the state of the forgotten door,representing convolution operations, s k-1 Representing a previous state of the forget gate;
step S313: the states of the output gate and the output gate are updated using the following formulas:
；
in the formula, o k Representing the output gate, omega o Representing the weight of the output gate, u o Indicating the weight of the output door at the previous moment, b o Indicating the offset of the output gate, h k Representing the status of the output gate;
step S32: the method for constructing the Gaussian process regression sub-model comprises the following steps of:
step S321: the GPR probability distribution is defined using the formula:
；
where f () is an objective function, l is an input, and l' is a predicted value; GPR () is a gaussian regression process; m (L) and L (L, L') are a mean function and a covariance function, respectively;
step S322: defining a kernel function k M The formula used is as follows:
；
wherein, gamma is a super parameter reflecting smoothness;is Bessel function, sigma 2 Is the noise variance in the regression process, ρ is the hyper-parameter that adjusts the distance scale,is a gamma function;
step S323: adding the kernel functions of different scales to obtain a rational quadratic function k of the popular kernel function RQ The formula used is as follows:
；
where a is the relative weight reflecting the scale change,andis a hyper-parameter that affects the axis scaling;
step S324: the total joint prior distribution of the known output y and the predicted output y' is defined using the following formula:
；
wherein I is n Is a unit matrix with a diagonal line of 1, and L (L, L) and L (L ', L') are respectively an input covariance function and a covariance function of an output value;
step S325: by calculating the condition distributionThe output is predicted using the following formula:
；
wherein T is a transpose operation;
step S326: and taking the weighted sum of the long-term memory neural network and the output of the Gaussian process regression model as a final prediction result.
By executing the operation, aiming at the problems of low model performance caused by gradient disappearance and gradient explosion of a general fault detection model and weak robustness, the scheme adopts the long-term dependency relationship of data captured based on the long-term memory network predictor model, so that the whole model is easier to train and optimize, the probability distribution of the prediction result based on the Gaussian regression model is constructed, various uncertainties are more flexibly adapted, the model robustness is improved, and the model performance is improved.
Fifth embodiment referring to fig. 1 and 4, the embodiment is based on the above embodiment, and in step S4, the model parameter search specifically includes the following steps:
step S41: defining a globally optimal solution guide search mechanism, wherein the following formula is used:
；
wherein x is a parameter position, I and j are different individuals, it is the iteration number, r is a random number between 0 and 1, and P is a global optimal position; fl is the distance traveled by the individual; m is the individual's historical optimal position;
step S42: defining a movement strategy, and introducing a Levy flight search strategy to replace random search, wherein the following formula is adopted:
；
wherein, gamma s Is a randomly changed symbol, and takes the value ofThe method comprises the steps of carrying out a first treatment on the surface of the a is a step factor, β is a constant that controls the step change, Γ () is a gamma function;
step S43: initializing, decomposing a parameter search space, initializing a parameter position based on a subspace, and taking the model performance established based on parameters as a parameter fitness value;
step S44: updating the parameter position by using the following formula:
；
wherein, AP is a movement threshold;
step S45: judging, wherein a local threshold value, a global threshold value and a maximum iteration number are preset; if the individual fitness value of the subspace is higher than a local threshold, establishing a model based on the optimal individual of each subspace, if the performance of the model is higher than the global optimum, establishing the model at the output parameter position, and if the performance of the model is lower than the global optimum, searching the locally optimal individual with the lowest subspace; if the maximum iteration times are reached, re-dividing the subspace and initializing the parameter positions to perform parameter searching; otherwise, the search is continued.
By executing the operation, aiming at the problems of blind selection and local optimization caused by a single search mechanism in a search algorithm, the scheme improves the search precision and global optimization capacity of the algorithm based on defining a global optimal solution search mechanism and a movement strategy, accelerates the convergence rate of the algorithm, improves the search coverage rate based on dividing subspaces and reduces the search dimension, thereby improving the operation efficiency of the algorithm.
In a sixth embodiment, referring to fig. 1, the embodiment is based on the above embodiment, and in step S5, the real-time operation is to collect the aircraft operation data in real time and input the data into the model, and the model outputs the weighted prediction result.
An embodiment seven, referring to fig. 2, based on the above embodiment, the aircraft fault detection system based on artificial intelligence provided by the invention includes a data acquisition module, a data preprocessing module, a hybrid aircraft fault detection model building module, a model parameter searching module and a real-time operation module;
the data acquisition module is used for collecting and acquiring historical operation data signals and corresponding operation states of the aircraft and sending the data to the data preprocessing module;
the data preprocessing module receives the data sent by the data acquisition module, constructs upper and lower limit curves based on adjacent values in comparison fluctuation and local extreme points connected by spline lines, judges and decomposes signals based on a monotonic function, thereby constructing a sample data set, and sends the data to the mixed aircraft fault detection model building module;
the mixed aircraft fault detection model building module receives the data sent by the data preprocessing module, builds a long-term memory network predictor model and a Gaussian process regression sub model to build a mixed aircraft fault detection model, and sends the data to the model parameter searching module;
the model parameter searching module receives data sent by the hybrid aircraft fault detection model building module, updates parameter positions based on a defined global optimal solution searching mechanism and a movement strategy, sets a local threshold value, a global threshold value and a maximum iteration number to judge a searching result based on a division subspace, and sends the data to the real-time operation module;
the real-time operation module receives the data sent by the model parameter search module, acquires the aircraft operation data in real time, inputs the aircraft operation data into the model, and outputs a weighted prediction result.
It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
The invention and its embodiments have been described above with no limitation, and the actual construction is not limited to the embodiments of the invention as shown in the drawings. In summary, if one of ordinary skill in the art is informed by this disclosure, a structural manner and an embodiment similar to the technical solution should not be creatively devised without departing from the gist of the present invention.
Claims (7)
1. The aircraft fault detection method based on artificial intelligence is characterized by comprising the following steps of: the method comprises the following steps:
step S1: collecting data;
step S2: data preprocessing, namely constructing an upper limit curve and a lower limit curve based on adjacent values in comparison fluctuation and connecting local extreme points by using spline lines, and judging a decomposition signal based on a monotonic function so as to construct a sample data set;
step S3: establishing a hybrid aircraft fault detection model, and establishing the hybrid aircraft fault detection model based on the long-short-term memory network prediction sub-model and the Gaussian process regression sub-model;
step S4: model parameter searching, namely updating parameter positions based on a defined global optimal solution searching mechanism and a movement strategy, and judging a searching result based on a dividing subspace, a local threshold value, a global threshold value and the maximum iteration number;
step S5: and (5) running in real time.
2. The artificial intelligence based aircraft fault detection method of claim 1, wherein: in step S4, the model parameter search specifically includes the following steps:
step S41: defining a globally optimal solution guide search mechanism, wherein the following formula is used:
；
wherein x is a parameter position, I and j are different individuals, it is the iteration number, r is a random number between 0 and 1, and P is a global optimal position; fl is the distance traveled by the individual; m is the individual's historical optimal position;
step S42: defining a movement strategy, and introducing a Levy flight search strategy to replace random search, wherein the following formula is adopted:
；
wherein, gamma s Is a randomly changed symbol, and takes the value ofThe method comprises the steps of carrying out a first treatment on the surface of the a is a step factor, β is a constant that controls the step change, Γ () is a gamma function;
step S43: initializing, decomposing a parameter search space, initializing a parameter position based on a subspace, and taking the model performance established based on parameters as a parameter fitness value;
step S44: updating the parameter position by using the following formula:
；
wherein, AP is a movement threshold;
step S45: judging, wherein a local threshold value, a global threshold value and a maximum iteration number are preset; if the individual fitness value of the subspace is higher than a local threshold, establishing a model based on the optimal individual of each subspace, if the performance of the model is higher than the global optimum, establishing the model at the output parameter position, and if the performance of the model is lower than the global optimum, searching the locally optimal individual with the lowest subspace; if the maximum iteration times are reached, re-dividing the subspace and initializing the parameter positions to perform parameter searching; otherwise, the search is continued.
3. The artificial intelligence based aircraft fault detection method of claim 1, wherein: in step S3, the building a hybrid aircraft fault detection model specifically includes the following steps:
step S31: constructing a long-term and short-term memory network predictor model, wherein the states of all gates are input X at the current moment k through S-shaped units k And the output h of the last instant k-1 k-1 Determining; the input gate decides whether new state information is received; the forget gate is responsible for forgetting the previous state in the hidden layer; the output gate determines whether the information calculated by the model is output, and the steps include:
step S311: the input gate status information and the state of the input gate are updated using the following formulas:
；
wherein p is k Representing input gate state information, tanh () represents hyperbolic tangent activation function, ω s Weights representing input gate states, u s Weights indicating the previous state of the input door b s Bias amount indicating input gate state, i k Representing the state of the input door, sigma s Representing a sigmoid activation function, w i Representing the weight of the input gate, u i Indicating the weight of the input door at the previous moment, b i Representing the offset of the input gate;
step S312: updating the states of the forgetting door and the forgetting door, wherein the following formula is used:
；
wherein f k Indicating forgetful door omega f Weight indicating forgetting gate, u f Weight indicating the moment before forgetting to leave the door, b f Representing the bias of the forgetting gate, s k Indicating the state of the forgotten door,representing convolution operations, s k-1 Representing a previous state of the forget gate;
step S313: the states of the output gate and the output gate are updated using the following formulas:
；
in the formula, o k Representing the output gate, omega o Representing the weight of the output gate, u o Indicating the weight of the output door at the previous moment, b o Indicating the offset of the output gate, h k Representing the status of the output gate;
step S32: the method for constructing the Gaussian process regression sub-model comprises the following steps of:
step S321: the GPR probability distribution is defined using the formula:
；
where f () is an objective function, l is an input, and l' is a predicted value; GPR () is a gaussian regression process; m (L) and L (L, L') are a mean function and a covariance function, respectively;
step S322: defining a kernel function k M The formula used is as follows:
；
wherein, gamma is a super parameter reflecting smoothness;is Bessel function, sigma 2 Is the noise variance in the regression process, ρ is the hyper-parameter that adjusts the distance scale, ++>Is a gamma function;
step S323: adding the kernel functions of different scales to obtain a rational quadratic function k of the popular kernel function RQ The formula used is as follows:
；
where a is the relative weight reflecting the scale change,and->Is a hyper-parameter that affects the axis scaling;
step S324: the total joint prior distribution of the known output y and the predicted output y' is defined using the following formula:
；
wherein I is n Is a unit matrix with a diagonal line of 1, and L (L, L) and L (L ', L') are respectively an input covariance function and a covariance function of an output value;
step S325: by calculating the condition distributionThe output is predicted using the following formula:
；
wherein T is a transpose operation;
step S326: and taking the weighted sum of the long-term memory neural network and the output of the Gaussian process regression model as a final prediction result.
4. The artificial intelligence based aircraft fault detection method of claim 1, wherein: in step S2, the data preprocessing specifically includes the following steps:
step S21: setting data set, selecting historical operation data signal as C b The method comprises the steps of carrying out a first treatment on the surface of the By at C b Searching local maximum values and minimum values in a mode of comparing adjacent values in all the fluctuation, wherein the local minimum values and the maximum values respectively represent minimum and maximum data points on a local fluctuation scale; by connecting local extreme points using spline lines, upper limit curves e are respectively constructed up And a lower limit curve e low ；
Step S22: calculating a local average value m e And calculates a local average value and C b The formula used is as follows:
；
step S23: repeating the above operation until the obtained difference becomes a monotonic function;
step S24: after the decomposition signal is ended and the n1 data signal IMF and a monotonic function are obtained, C b Consists of the following contents:
；
where j1 is the signal index, r m Is the remainder, i.e., the trend portion remaining in the decomposition process;
step S25: and constructing a data set, namely constructing the data set based on the decomposed signals, and taking the running state corresponding to the data as a label.
5. The artificial intelligence based aircraft fault detection method of claim 1, wherein: in step S1, the data acquisition is to acquire historical operation data signals and corresponding operation states of the aircraft.
6. The artificial intelligence based aircraft fault detection method of claim 1, wherein: in step S5, the real-time operation is to collect the aircraft operation data in real time and input the data into a model, and the model outputs a weighted prediction result.
7. An artificial intelligence based aircraft fault detection system for implementing an artificial intelligence based aircraft fault detection method as claimed in any one of claims 1 to 6, wherein: the system comprises a data acquisition module, a data preprocessing module, a hybrid aircraft fault detection model building module, a model parameter searching module and a real-time operation module;
the data acquisition module is used for collecting and acquiring historical operation data signals and corresponding operation states of the aircraft and sending the data to the data preprocessing module;
the data preprocessing module receives the data sent by the data acquisition module, constructs upper and lower limit curves based on adjacent values in comparison fluctuation and local extreme points connected by spline lines, judges and decomposes signals based on a monotonic function, thereby constructing a sample data set, and sends the data to the mixed aircraft fault detection model building module;
the mixed aircraft fault detection model building module receives the data sent by the data preprocessing module, builds a long-term memory network predictor model and a Gaussian process regression sub model to build a mixed aircraft fault detection model, and sends the data to the model parameter searching module;
the model parameter searching module receives data sent by the hybrid aircraft fault detection model building module, updates parameter positions based on a defined global optimal solution searching mechanism and a movement strategy, sets a local threshold value, a global threshold value and a maximum iteration number to judge a searching result based on a division subspace, and sends the data to the real-time operation module;
the real-time operation module receives the data sent by the model parameter search module, acquires the aircraft operation data in real time, inputs the aircraft operation data into the model, and outputs a weighted prediction result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311641840.1A CN117349797B (en) | 2023-12-04 | 2023-12-04 | Aircraft fault detection method and system based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311641840.1A CN117349797B (en) | 2023-12-04 | 2023-12-04 | Aircraft fault detection method and system based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117349797A true CN117349797A (en) | 2024-01-05 |
| CN117349797B CN117349797B (en) | 2024-02-06 |
Family
ID=89356040
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311641840.1A Active CN117349797B (en) | 2023-12-04 | 2023-12-04 | Aircraft fault detection method and system based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117349797B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117584136A (en) * | 2024-01-18 | 2024-02-23 | 泰山学院 | Robot fault detection method and system based on artificial intelligence |
| CN117647367A (en) * | 2024-01-29 | 2024-03-05 | 四川航空股份有限公司 | Machine learning-based method and system for positioning leakage points of aircraft fuel tank |
| CN117851922A (en) * | 2024-03-08 | 2024-04-09 | 山东新泽仪器有限公司 | Operation monitoring method of air concentration monitoring instrument |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100166295A1 (en) * | 2008-12-31 | 2010-07-01 | Industrial Technology Research Institute | Method and system for searching for global minimum |
| CN102831325A (en) * | 2012-09-04 | 2012-12-19 | 北京航空航天大学 | Method for predicting bearing fault based on Gaussian process regression |
| CN103778422A (en) * | 2012-10-17 | 2014-05-07 | 柯耐克斯公司 | System and method for selecting and displaying segmentation parameters for optical character recognition |
| WO2020253950A1 (en) * | 2019-06-18 | 2020-12-24 | Framatome Gmbh | Monitoring method, predicting method, monitoring system and computer program |
| WO2021232655A1 (en) * | 2020-05-20 | 2021-11-25 | 国网河北省电力有限公司电力科学研究院 | Method for evaluating mechanical state of high-voltage shunt reactor on the basis of vibration feature |
| US20220235676A1 (en) * | 2019-06-20 | 2022-07-28 | Tata Consultancy Services Limited | Method and system for optimization of combination cycle gas turbine operation |
| CN114969995A (en) * | 2021-11-22 | 2022-08-30 | 昆明理工大学 | Rolling bearing early fault intelligent diagnosis method based on improved sparrow search and acoustic emission |
| CN115561005A (en) * | 2022-10-25 | 2023-01-03 | 淮阴工学院 | Chemical process fault diagnosis method based on EEMD decomposition and lightweight neural network |
| CN116952583A (en) * | 2023-03-27 | 2023-10-27 | 哈尔滨市滨奇科技发展有限公司 | Rolling bearing fault diagnosis method of digital twin driving deep transfer learning model |
-
2023
- 2023-12-04 CN CN202311641840.1A patent/CN117349797B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100166295A1 (en) * | 2008-12-31 | 2010-07-01 | Industrial Technology Research Institute | Method and system for searching for global minimum |
| CN102831325A (en) * | 2012-09-04 | 2012-12-19 | 北京航空航天大学 | Method for predicting bearing fault based on Gaussian process regression |
| CN103778422A (en) * | 2012-10-17 | 2014-05-07 | 柯耐克斯公司 | System and method for selecting and displaying segmentation parameters for optical character recognition |
| WO2020253950A1 (en) * | 2019-06-18 | 2020-12-24 | Framatome Gmbh | Monitoring method, predicting method, monitoring system and computer program |
| US20220235676A1 (en) * | 2019-06-20 | 2022-07-28 | Tata Consultancy Services Limited | Method and system for optimization of combination cycle gas turbine operation |
| WO2021232655A1 (en) * | 2020-05-20 | 2021-11-25 | 国网河北省电力有限公司电力科学研究院 | Method for evaluating mechanical state of high-voltage shunt reactor on the basis of vibration feature |
| CN114969995A (en) * | 2021-11-22 | 2022-08-30 | 昆明理工大学 | Rolling bearing early fault intelligent diagnosis method based on improved sparrow search and acoustic emission |
| CN115561005A (en) * | 2022-10-25 | 2023-01-03 | 淮阴工学院 | Chemical process fault diagnosis method based on EEMD decomposition and lightweight neural network |
| CN116952583A (en) * | 2023-03-27 | 2023-10-27 | 哈尔滨市滨奇科技发展有限公司 | Rolling bearing fault diagnosis method of digital twin driving deep transfer learning model |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117584136A (en) * | 2024-01-18 | 2024-02-23 | 泰山学院 | Robot fault detection method and system based on artificial intelligence |
| CN117584136B (en) * | 2024-01-18 | 2024-03-29 | 泰山学院 | Robot fault detection method and system based on artificial intelligence |
| CN117647367A (en) * | 2024-01-29 | 2024-03-05 | 四川航空股份有限公司 | Machine learning-based method and system for positioning leakage points of aircraft fuel tank |
| CN117647367B (en) * | 2024-01-29 | 2024-04-16 | 四川航空股份有限公司 | Machine learning-based method and system for positioning leakage points of aircraft fuel tank |
| CN117851922A (en) * | 2024-03-08 | 2024-04-09 | 山东新泽仪器有限公司 | Operation monitoring method of air concentration monitoring instrument |
| CN117851922B (en) * | 2024-03-08 | 2024-06-07 | 山东新泽仪器有限公司 | Operation monitoring method of air concentration monitoring instrument |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117349797B (en) | 2024-02-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN117349797B (en) | Aircraft fault detection method and system based on artificial intelligence | |
| CN110738360B (en) | Method and system for predicting residual life of equipment | |
| Fan et al. | A novel machine learning method based approach for Li-ion battery prognostic and health management | |
| CN115169479A (en) | Remote monitoring method, system and storage medium for sewage treatment process | |
| CN108961460B (en) | Fault prediction method and device based on sparse ESGP (Enterprise service gateway) and multi-objective optimization | |
| Bai et al. | State of art on state estimation: Kalman filter driven by machine learning | |
| CN110991776A (en) | Method and system for realizing water level prediction based on GRU network | |
| CN111523727B (en) | Method for predicting remaining life of battery by considering recovery effect based on uncertain process | |
| CN116052254A (en) | Visual continuous emotion recognition method based on extended Kalman filtering neural network | |
| CN115147645A (en) | Membrane module membrane pollution detection method based on multi-feature information fusion | |
| CN116204779A (en) | Method, system and readable storage medium for judging operation state of energy storage salt cavern | |
| CN105303835A (en) | Short-time prediction method of road traffic flow state | |
| CN113126489B (en) | Intelligent BIT design method of heavy-duty gas turbine control system based on CNN-GRU-BINN | |
| CN113988210A (en) | Method and device for restoring distorted data of structure monitoring sensor network and storage medium | |
| CN113159395A (en) | Deep learning-based sewage treatment plant water inflow prediction method and system | |
| CN114358244B (en) | Big data intelligent detection system of pressure based on thing networking | |
| CN114386672B (en) | Environment big data Internet of things intelligent detection system | |
| CN115794805A (en) | Medium-low voltage distribution network measurement data supplementing method | |
| CN115459982A (en) | Power network false data injection attack detection method | |
| CN114720129A (en) | Rolling bearing residual life prediction method and system based on bidirectional GRU | |
| CN108446760A (en) | A kind of pipe gallery interior humidity LSTM prediction techniques and system | |
| CN113807005A (en) | Bearing residual life prediction method based on improved FPA-DBN | |
| CN113837443A (en) | Transformer substation line load prediction method based on depth BilSTM | |
| CN111221479A (en) | Method, system and storage medium for judging abnormal storage capacity variation | |
| CN112561203B (en) | Method and system for realizing water level early warning based on clustering and GRU |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |