A method and system for aircraft failure early warning based on flight data

Abstract

The application provides a kind of aircraft failure early warning method and system based on flight data, it is related to aircraft failure early warning technical field.In the present application, first, the aircraft external data semantic vector and the aircraft internal data semantic vector corresponding to the flight data to be processed are mined out;Second, the pre-generated flight data noise vector is subjected to semantic strengthening operation based on the aircraft external data semantic vector, and the flight data strengthening vector is output;Then, according to the aircraft internal data semantic vector, the aircraft internal depth vector is mined out;Finally, according to the flight data strengthening vector, the aircraft internal depth vector is subjected to associated mining operation, and the target flight data semantic vector is output, and the aircraft failure early warning information is analyzed according to the target flight data semantic vector.Based on the above, the problem that the reliability of aircraft failure early warning in the prior art is relatively low can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of aircraft fault early warning, in particular to an aircraft fault early warning method and system based on flight data. BACKGROUND

[0002] In the prior art, the aircraft fault early warning based on aircraft flight data generally includes a data collection stage and a data analysis stage. In the data collection stage, various data of the aircraft during flight, such as environmental data such as weather conditions or state data of the aircraft components, are generally collected. Then, in the data analysis stage, the collected data can be compared with the threshold value set according to experience, so as to determine whether there is a fault. However, the inventors have found that since the threshold value is set according to experience, its reliability is not high, and the separate comparison of various data also ignores the correlation between the data, which also leads to relatively low reliability. SUMMARY

[0003] Therefore, the purpose of the present application is to provide an aircraft fault early warning method and system based on flight data to improve the problem of relatively low reliability of aircraft fault early warning in the prior art.

[0004] To achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0005] An aircraft fault early warning method based on flight data, comprising:

[0006] mining an aircraft external data semantic vector corresponding to the to-be-processed flight data, and mining an aircraft internal data semantic vector corresponding to the to-be-processed flight data, wherein the to-be-processed flight data includes aircraft internal data and aircraft external data, the aircraft internal data is used to reflect the state of each aircraft component of a target aircraft during flight, and the aircraft external data is used to reflect the state of the environment in which the target aircraft is located during flight;

[0007] loading the aircraft external data semantic vector, so that an aircraft fault prediction network obtains the aircraft external data semantic vector, and using a semantic mining unit included in the aircraft fault prediction network, performing semantic strengthening operation on a pre-generated flight data noise vector based on the aircraft external data semantic vector, and outputting a flight data strengthening vector corresponding to the semantic mining unit;

[0008] mining an aircraft internal deep vector corresponding to the semantic mining unit according to the aircraft internal data semantic vector;

[0009] According to the flight data strengthening vector corresponding to the semantic mining unit, the aircraft internal depth vector is subjected to a correlation mining operation, and a target flight data semantic vector is output, and according to the target flight data semantic vector, aircraft fault early warning information corresponding to the to-be-processed flight data is analyzed, wherein the aircraft fault early warning information is used to reflect whether the target aircraft has a fault.

[0010] In a preferred selection of the present application, in the aircraft fault early warning method based on flight data, the step of loading the aircraft external data semantic vector so that the aircraft fault prediction network obtains the aircraft external data semantic vector, and using a semantic mining unit included in the aircraft fault prediction network, performing semantic strengthening operation on the pre-generated flight data noise vector based on the aircraft external data semantic vector to output the flight data strengthening vector corresponding to the semantic mining unit, comprises:

[0011] The aircraft external data semantic vector is loaded so that the aircraft fault prediction network obtains the aircraft external data semantic vector, and a semantic mining unit included in the aircraft fault prediction network is used to mine a flight data noise mining vector from a pre-generated flight data noise vector;

[0012] According to the flight data noise mining vector, a corresponding flight data mapping vector is formed by mapping;

[0013] According to the aircraft external data semantic vector, a flight first mapping vector and a flight second mapping vector are respectively formed by mapping;

[0014] According to the flight data mapping vector, the flight first mapping vector and the flight second mapping vector, a focus fusion operation is performed, and according to the output vector of the focus fusion operation, the flight data strengthening vector corresponding to the semantic mining unit is obtained.

[0015] In a preferred selection of the present application, in the aircraft fault early warning method based on flight data, the step of loading the aircraft external data semantic vector so that the aircraft fault prediction network obtains the aircraft external data semantic vector, and using a semantic mining unit included in the aircraft fault prediction network, performing semantic strengthening operation on the pre-generated flight data noise vector based on the aircraft external data semantic vector to output the flight data strengthening vector corresponding to the semantic mining unit, comprises:

[0016] A target parameter is determined, and a corresponding number of initial parameters are determined based on the target parameter, wherein the initial parameters are multiple;

[0017] determine a corresponding initial flight data noise vector according to the initial parameter, wherein each vector parameter in the initial flight data noise vector is determined by a random number generator based on the initial parameter;

[0018] obtain a flight data noise vector based on each determined initial flight data noise vector.

[0019] In a preferred selection of the present application, in the above-mentioned aircraft fault early warning method based on flight data, the step of mining an aircraft internal deep vector corresponding to the semantic mining unit according to the aircraft internal data semantic vector comprises:

[0020] determining a flight data noise vector of a semantic mining unit included in the aircraft fault prediction network;

[0021] performing convolution mining operation on the aircraft internal data semantic vector, and performing vector fusion operation on the aircraft internal data semantic vector after convolution mining operation and the flight data noise vector, and outputting the corresponding aircraft internal data semantic vector after vector fusion operation;

[0022] loading the aircraft internal data semantic vector after vector fusion operation, so that the vector mining unit corresponding to the semantic mining unit obtains the aircraft internal data semantic vector after vector fusion operation and performs corresponding processing, and performing convolution mining operation on the output vector of the vector mining unit, and outputting the corresponding aircraft internal deep vector.

[0023] In a preferred selection of the present application, in the above-mentioned aircraft fault early warning method based on flight data, the step of loading the aircraft internal data semantic vector after vector fusion operation, so that the vector mining unit corresponding to the semantic mining unit obtains the aircraft internal data semantic vector after vector fusion operation and performs corresponding processing, and performing convolution mining operation on the output vector of the vector mining unit, and outputting the corresponding aircraft internal deep vector comprises:

[0024] loading the aircraft internal data semantic vector after vector fusion operation, so that the vector mining unit corresponding to the semantic mining unit obtains the aircraft internal data semantic vector after vector fusion operation, and loading a predetermined reference flight data semantic vector, so that the vector mining unit corresponding to the semantic mining unit obtains the reference flight data semantic vector, wherein the reference flight data semantic vector is determined based on each training flight data semantic vector mined by the semantic mining unit and the vector mining unit in the process of training optimization, and each training flight data semantic vector is obtained by performing corresponding semantic strengthening operation and association mining operation on corresponding training flight data after training optimization is completed;

[0025] According to the reference flight data semantic vector, the vector fusion operation is performed on the aircraft internal data semantic vector after the vector fusion operation, and an associated vector of the aircraft internal data is output;

[0026] The associated vector of the aircraft internal data is taken as an output vector of the vector mining unit, and a convolution mining operation is performed, and a corresponding aircraft internal depth vector is output.

[0027] In the preferred selection of the present application, in the above-mentioned aircraft fault early warning method based on flight data, the step of performing an associated mining operation on the aircraft internal data semantic vector after the vector fusion operation according to the reference flight data semantic vector, and outputting a corresponding aircraft internal data associated vector, comprises:

[0028] Determine the association parameter between the reference flight data semantic vector and the aircraft internal data semantic vector after the vector fusion operation, and based on the association parameter, perform an associated adjustment on the aircraft internal data semantic vector after the vector fusion operation, and output an aircraft internal data adjustment vector;

[0029] The vector fusion operation is performed on the aircraft internal data semantic vector after the vector fusion operation and the aircraft internal data adjustment vector, and a corresponding aircraft internal data associated vector is output.

[0030] In the preferred selection of the present application, in the above-mentioned aircraft fault early warning method based on flight data, the step of performing an associated mining operation on the aircraft internal depth vector according to the flight data strengthening vector corresponding to the semantic mining unit, outputting a target flight data semantic vector, and analyzing the aircraft fault early warning information corresponding to the to-be-processed flight data according to the target flight data semantic vector, comprises:

[0031] Determine the association parameter between the flight data strengthening vector corresponding to the semantic mining unit and the aircraft internal depth vector, and based on the association parameter, perform an associated adjustment on the aircraft internal depth vector, and output a corresponding adjusted aircraft internal depth vector;

[0032] The vector fusion operation is performed on the adjusted aircraft internal depth vector and the aircraft internal depth vector, and a corresponding target flight data semantic vector is output.

[0033] A full connection flight data semantic vector corresponding to the target flight data semantic vector is output by performing a full connection operation on the target flight data semantic vector, and the aircraft fault early warning information corresponding to the to-be-processed flight data is analyzed according to the full connection flight data semantic vector.

[0034] The present application also provides an aircraft fault early warning system based on flight data, comprising:

[0035] a flight data semantic mining module, configured to mine an aircraft external data semantic vector corresponding to to-be-processed flight data, and mine an aircraft internal data semantic vector corresponding to the to-be-processed flight data, wherein the to-be-processed flight data comprises aircraft internal data and aircraft external data, the aircraft internal data is used to reflect the state of each aircraft component of a target aircraft during flight, and the aircraft external data is used to reflect the state of an environment in which the target aircraft is located during flight;

[0036] a vector semantic reinforcement module, configured to load the aircraft external data semantic vector, so that an aircraft fault prediction network obtains the aircraft external data semantic vector, and perform a semantic reinforcement operation on a pre-generated flight data noise vector based on the aircraft external data semantic vector by using a semantic mining unit included in the aircraft fault prediction network, and output a flight data reinforcement vector corresponding to the semantic mining unit;

[0037] a vector deep mining module, configured to mine an aircraft internal deep vector corresponding to the semantic mining unit according to the aircraft internal data semantic vector;

[0038] an aircraft fault analysis module, configured to perform an associated mining operation on the aircraft internal deep vector according to the flight data reinforcement vector corresponding to the semantic mining unit, output a target flight data semantic vector, and analyze aircraft fault warning information corresponding to the to-be-processed flight data according to the target flight data semantic vector, wherein the aircraft fault warning information is used to reflect whether the target aircraft has a fault.

[0039] In a preferred selection of the present application, in the aircraft fault warning system based on flight data, the vector semantic reinforcement module is specifically configured to:

[0040] load the aircraft external data semantic vector, so that the aircraft fault prediction network obtains the aircraft external data semantic vector, and mine a flight data noise mining vector from a pre-generated flight data noise vector by using a semantic mining unit included in the aircraft fault prediction network;

[0041] map the flight data noise mining vector to form a corresponding flight data mapping vector;

[0042] map the aircraft external data semantic vector to form a first flight mapping vector and a second flight mapping vector, respectively;

[0043] perform a focused fusion operation according to the flight data mapping vector, the flight first mapping vector and the flight second mapping vector, and obtain a flight data reinforced vector corresponding to the semantic mining unit according to an output vector of the focused fusion operation.

[0044] In a preferred selection of the present application, in the above-mentioned aircraft fault early warning system based on flight data, the vector deep mining module is specifically used for:

[0045] determining a flight data noise vector of a semantic mining unit included in the aircraft fault prediction network;

[0046] performing a convolution mining operation on the aircraft internal data semantic vector, and performing a vector fusion operation on the aircraft internal data semantic vector after the convolution mining operation and the flight data noise vector, to output an aircraft internal data semantic vector after the vector fusion operation;

[0047] loading the aircraft internal data semantic vector after the vector fusion operation, so that a vector mining unit corresponding to the semantic mining unit obtains the aircraft internal data semantic vector after the vector fusion operation and performs corresponding processing, and performing a convolution mining operation on an output vector of the vector mining unit to output an aircraft internal deep vector.

[0048] The aircraft fault early warning method and system based on flight data provided by the present application first mine an aircraft external data semantic vector and an aircraft internal data semantic vector corresponding to to-be-processed flight data. Second, a flight data noise vector generated in advance is subjected to semantic reinforcement operation based on the aircraft external data semantic vector, to output a flight data reinforced vector. Then, an aircraft internal deep vector is mined according to the aircraft internal data semantic vector. Finally, a target flight data semantic vector is output by performing associated mining operation on the aircraft internal deep vector according to the flight data reinforced vector, and aircraft fault early warning information is analyzed according to the target flight data semantic vector. Based on the above, on the one hand, the flight data reinforced vector carrying semantic information of aircraft external data and the aircraft internal deep vector carrying semantic information of aircraft internal data are subjected to associated mining operation, so that the semantic representation ability of the obtained target flight data semantic vector can be better, and therefore the reliability of the aircraft fault early warning information analyzed based on the target flight data semantic vector is higher. On the other hand, since the flight data reinforced vector not only carries semantic information of aircraft external data, but also fuses the flight data noise vector, it can better simulate these real situations (some noise and interference may be ignored in the collected aircraft external data), thereby improving its performance in the actual environment, so that the representation ability of the aircraft data reinforced vector can be further improved, thereby improving the relatively low reliability of the aircraft fault early warning in the prior art. Attached Figure Description

[0049] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings.

[0050] Figure 1 This is a structural block diagram of an electronic device provided in an embodiment of this application.

[0051] Figure 2 This is a flowchart illustrating the aircraft fault early warning method based on flight data provided in an embodiment of this application.

[0052] Figure 3 This is a block diagram of an aircraft fault early warning system based on flight data provided in an embodiment of this application. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0054] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0055] like Figure 1 As shown in the illustration, this application provides an electronic device. The electronic device may include a memory, a processor, and an aircraft fault warning system based on flight data.

[0056] In detail, the memory and the processor are electrically connected directly or indirectly to enable data transmission or interaction. For example, the memory and the processor can be electrically connected via one or more communication buses or signal lines. The flight data-based aircraft fault warning system includes at least one software functional module stored in the memory in the form of software or firmware. The processor is used to execute executable computer programs stored in the memory, such as the software functional modules and computer programs included in the flight data-based aircraft fault warning system, to implement the flight data-based aircraft fault warning method provided in this application embodiment.

[0057] Optionally, the memory can be, but is not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), etc. Moreover, the processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), a system on chip (SoC), etc.; and can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component.

[0058] It can be understood that Figure 1 The structure shown is only schematic, and the electronic device can further include more or fewer components than those shown, or have a different configuration of components than those shown, such as further including a communication unit for information interaction with other devices (such as various sensors, etc.). Figure 1 It can be understood that Figure 1 The structure shown is only schematic, and the electronic device can further include more or fewer components than those shown, or have a different configuration of components than those shown, such as further including a communication unit for information interaction with other devices (such as various sensors, etc.).

[0059] In combination with Figure 2 The embodiments of the present application also provide a flight data based aircraft failure early warning method applicable to the above-mentioned electronic device. The method steps defined by the flow related to the flight data based aircraft failure early warning method can be implemented by the electronic device.

[0060] The specific flow shown in Figure 2 will be described in detail below.

[0061] Step S110: mining out an aircraft external data semantic vector corresponding to the to-be-processed flight data, and mining out an aircraft internal data semantic vector corresponding to the to-be-processed flight data.

[0062] In the embodiment of the present application, the electronic device can mine the aircraft external data semantic vector corresponding to the to-be-processed flight data, and mine the aircraft internal data semantic vector corresponding to the to-be-processed flight data. Wherein, the to-be-processed flight data includes aircraft internal data and aircraft external data, the aircraft internal data is used to reflect the state of each aircraft component of the target aircraft in the flight process (such as the data sequence formed by collecting the temperature, stress, etc. of each aircraft component), and the aircraft external data is used to reflect the state of the environment where the target aircraft is in the flight process (such as the data sequence formed by collecting the temperature, humidity, pressure, etc. of the environment where the target aircraft is). In addition, since the aircraft internal data and the aircraft external data can be data sequences, after embedding processing by a word embedding model to obtain a word embedding vector, the position of each word in the data sequence can also be embedded to obtain a corresponding position embedding vector. Then, each word embedding vector and each position embedding vector can be fused by splicing, etc. to form a corresponding semantic vector, i.e. to form a corresponding aircraft external data semantic vector and an aircraft internal data semantic vector. And the word embedding model can be trained and optimized together with the aircraft fault prediction network in the following text to form.

[0063] In step S120, the aircraft external data semantic vector is loaded, so that the aircraft fault prediction network obtains the aircraft external data semantic vector, and a semantic mining unit included in the aircraft fault prediction network performs semantic strengthening operation on the pre-generated flight data noise vector based on the aircraft external data semantic vector, and outputs the flight data strengthening vector corresponding to the semantic mining unit.

[0064] In the embodiment of the present application, after the corresponding aircraft external data semantic vector is mined, the electronic device can load the aircraft external data semantic vector, so that the aircraft fault prediction network (a kind of neural network) obtains the aircraft external data semantic vector, and a semantic mining unit included in the aircraft fault prediction network performs semantic strengthening operation on the pre-generated flight data noise vector based on the aircraft external data semantic vector, and outputs the flight data strengthening vector corresponding to the semantic mining unit. In this way, the flight data strengthening vector can carry semantic information in the aircraft external data semantic vector, and also increase noise semantics to improve the characterization of the flight data strengthening vector to the real external environment.

[0065] In step S130, according to the aircraft internal data semantic vector, an aircraft internal deep vector corresponding to the semantic mining unit is mined.

[0066] In the embodiment of the present application, after the corresponding aircraft internal data semantic vector is mined, the electronic device can mine the aircraft internal deep vector corresponding to the semantic mining unit according to the aircraft internal data semantic vector. That is, the aircraft internal data semantic vector can be deeply mined to obtain deep semantic information, that is, the aircraft internal deep vector, so that the semantic of the aircraft internal deep vector can be more rich and general.

[0067] In step S140, the aircraft internal deep vector is associatedly mined according to the flight data strengthening vector corresponding to the semantic mining unit, a target flight data semantic vector is output, and the aircraft fault warning information corresponding to the to-be-processed flight data is analyzed according to the target flight data semantic vector.

[0068] In the embodiment of the present application, after the flight data strengthening vector and the aircraft internal deep vector are obtained, the electronic device can perform associated mining operation on the aircraft internal deep vector according to the flight data strengthening vector corresponding to the semantic mining unit, output a target flight data semantic vector (that is, the flight data strengthening vector is fused into the aircraft internal deep vector, so that the semantic representation ability of the obtained target flight data semantic vector is better), and analyze the aircraft fault warning information corresponding to the to-be-processed flight data according to the target flight data semantic vector. The aircraft fault warning information is used to reflect whether the target aircraft has a fault (in some other embodiments, when there is a fault, the corresponding fault type can also be included, such as the risk of cracks in the wing, such as engine failure, etc.).

[0069] Based on the above, since the flight data strengthening vector carrying the semantic information of the aircraft external data and the aircraft internal deep vector carrying the semantic information of the aircraft internal data are associatedly mined, the semantic representation ability of the obtained target flight data semantic vector can be better, therefore, the reliability of the aircraft fault warning information analyzed based on the target flight data semantic vector is also higher. On the other hand, since the flight data strengthening vector not only carries the semantic information of the aircraft external data, but also fuses the flight data noise vector, the real situation (the collected aircraft external data can ignore some noise and interference) can be better simulated, thereby improving its performance in the actual environment, so that the representation ability of the aircraft data strengthening vector can be further improved, so that the reliability of the analyzed aircraft fault warning information is improved, thereby improving the problem that the reliability of the aircraft fault warning in the prior art is relatively low.

[0070] It should be noted that the specific manner of performing the semantic reinforcement operation on the pre-generated flight data noise vector based on the aircraft external data semantic vector is not limited, and can be selected according to actual needs. For example, in an alternative embodiment, in order to enable the aircraft external data semantic vector and the aircraft data noise vector to be sufficiently fused, the above step S120 can further include steps S121, S122, S123 and S124, and the specific contents of each step are as follows.

[0071] In step S121, the aircraft external data semantic vector is loaded, so that the aircraft fault prediction network obtains the aircraft external data semantic vector, and a semantic mining unit included in the aircraft fault prediction network is used to mine a flight data noise mining vector according to the pre-generated flight data noise vector.

[0072] In the embodiment of the present application, the aircraft external data semantic vector can be loaded so that the aircraft fault prediction network obtains the aircraft external data semantic vector. In this way, subsequent processing can be performed using the aircraft fault prediction network, for example, first, a semantic mining unit included in the aircraft fault prediction network can be used to mine a flight data noise mining vector according to the pre-generated flight data noise vector. Specifically, the flight data noise vector can be directly used as the corresponding flight data noise mining vector, or the flight data noise vector can be subjected to self-attention processing to obtain the corresponding flight data noise mining vector.

[0073] In step S122, a corresponding flight data mapping vector is formed according to the flight data noise mining vector.

[0074] In the embodiment of the present application, after obtaining the flight data noise mining vector, a corresponding flight data mapping vector can be formed according to the flight data noise mining vector. For example, the flight data noise mining vector can be directly used as the corresponding flight data mapping vector, or the flight data noise mining vector can be multiplied by a first mapping matrix in the semantic mining unit to obtain the corresponding flight data mapping vector.

[0075] In step S123, a flight first mapping vector and a flight second mapping vector are respectively formed according to the aircraft external data semantic vector.

[0076] In the embodiments of the present application, the aircraft external data semantic vector can be respectively mapped to form a flight first mapping vector and a flight second mapping vector according to the aircraft external data semantic vector. For example, the aircraft external data semantic vector can be directly taken as the corresponding flight first mapping vector and flight second mapping vector, or the aircraft external data semantic vector can be multiplied by the second mapping matrix and the third mapping matrix in the semantic mining unit respectively to obtain the flight first mapping vector and the flight second mapping vector corresponding to the aircraft external data semantic vector.

[0077] In step S124, a focus fusion operation is performed according to the flight data mapping vector, the flight first mapping vector and the flight second mapping vector, and a flight data strengthening vector corresponding to the semantic mining unit is obtained according to the output vector of the focus fusion operation.

[0078] In the embodiments of the present application, the focus fusion operation can be performed according to the flight data mapping vector, the flight first mapping vector and the flight second mapping vector, and the flight data strengthening vector corresponding to the semantic mining unit can be obtained according to the output vector of the focus fusion operation. For example, the flight data mapping vector and the flight first mapping vector can be multiplied to obtain a corresponding correlation parameter, and then the flight second mapping vector can be weighted based on the correlation parameter to obtain the output vector of the focus fusion operation. Then, the output vector of the focus fusion operation and the aircraft external data semantic vector can be superimposed to obtain the flight data strengthening vector corresponding to the semantic mining unit.

[0079] It should be further explained that, in order to ensure that the fusion based on the flight data noise vector can make the flight data strengthening vector formed have better generalization, the step S124 can further include a step of generating the flight data noise vector, which can include the following contents:

[0080] Firstly, a target parameter can be determined, and a corresponding number of initial parameters can be determined based on the target parameter, wherein the initial parameters are multiple, that is, the target parameter can be an integer greater than 1, such as 1, 2, 3, 4, 5, 6, 7, 8, etc.

[0081] Secondly, a corresponding initial flight data noise vector can be determined according to the initial parameters, wherein each vector parameter in the initial flight data noise vector is determined by a random number generator based on the initial parameters, that is, the initial parameters can be taken as the random seed of the random number generator, and then a series of random numbers corresponding to the corresponding probability distribution, such as normal distribution, etc. can be generated to simulate the corresponding probability distribution, and the series of random numbers can constitute the corresponding initial flight data noise vector.

[0082] Finally, the flight data noise vector can be obtained based on the determined initial flight data noise vectors. For example, the initial flight data noise vectors can be averaged to obtain a corresponding mean vector as the corresponding flight data noise vector. Alternatively, each of the initial flight data noise vectors can be used as a flight data noise vector, and then steps S121-S124 can be performed based on each flight data noise vector to obtain a flight data reinforcement vector corresponding to each flight data noise vector. Then, the flight data reinforcement vectors corresponding to each flight data noise vector can be superimposed or averaged to obtain the flight data reinforcement vector corresponding to the semantic mining unit.

[0083] For step S130, the specific way of mining the aircraft interior depth vector corresponding to the semantic mining unit is not limited and can be selected according to actual needs. For example, in an alternative embodiment, in order to make the generalization of the aircraft interior depth vector stronger, the above step S130 can further include steps S131, S132 and S133, and the specific contents of each step are as follows.

[0084] Step S131 determines the flight data noise vector of the semantic mining unit included in the aircraft fault prediction network.

[0085] In the embodiments of the present application, the flight data noise vector of the semantic mining unit included in the aircraft fault prediction network can be determined first. The flight data noise vector can be different from the flight data noise vector in step S120, such as being determined based on different initial parameters or probability distributions, so that different noises can be introduced for the aircraft interior data and the aircraft exterior data, which is more likely to match the actual situation and can avoid the semantic representation ability from being reduced due to the introduction of the same noise.

[0086] Step S132 performs convolution mining operation on the aircraft interior data semantic vector, and performs vector fusion operation on the aircraft interior data semantic vector after convolution mining operation and the flight data noise vector, and outputs the corresponding aircraft interior data semantic vector after vector fusion operation.

[0087] In the embodiments of the present application, the aircraft interior data semantic vector can be subjected to convolution mining operation, and the aircraft interior data semantic vector after convolution mining operation and the flight data noise vector can be subjected to vector fusion operation (for example, superimposition of vectors or the like) to output the corresponding aircraft interior data semantic vector after vector fusion operation.

[0088] Step S133, load the aircraft interior data semantic vector after the vector fusion operation, so that the vector mining unit corresponding to the semantic mining unit obtains the aircraft interior data semantic vector after the vector fusion operation and performs corresponding processing, and perform convolution mining operation on the output vector of the vector mining unit, and output the corresponding aircraft interior depth vector.

[0089] In the embodiments of the present application, the aircraft interior data semantic vector after the vector fusion operation can be loaded, so that the vector mining unit corresponding to the semantic mining unit obtains the aircraft interior data semantic vector after the vector fusion operation and performs corresponding processing (such as depth mining), and the output vector of the vector mining unit is subjected to convolution mining operation, and the corresponding aircraft interior depth vector is output.

[0090] It can be understood that in the above step S133, the specific way of depth mining by the vector mining unit is not limited, and can be selected according to actual needs. For example, in an alternative embodiment, in order to make the mined aircraft interior depth vector have better semantic representation ability, the above step S133 can further include steps S133a, S133b and S133c, and the specific contents of each step are as follows.

[0091] Step S133a, load the aircraft interior data semantic vector after the vector fusion operation, so that the vector mining unit corresponding to the semantic mining unit obtains the aircraft interior data semantic vector after the vector fusion operation, and load the predetermined reference flight data semantic vector, so that the vector mining unit corresponding to the semantic mining unit obtains the reference flight data semantic vector.

[0092] In the embodiment of the present application, the aircraft internal data semantic vector after the vector fusion operation can be loaded, so that the vector mining unit corresponding to the semantic mining unit obtains the aircraft internal data semantic vector after the vector fusion operation, and a predetermined reference flight data semantic vector is loaded, so that the vector mining unit corresponding to the semantic mining unit obtains the reference flight data semantic vector. The reference flight data semantic vector is determined based on each training flight data semantic vector (corresponding to the target flight data semantic vector in step S140, such as the mean vector of each training flight data semantic vector) mined by the semantic mining unit and the vector mining unit in the training optimization process. Each training flight data semantic vector is obtained by performing corresponding semantic strengthening operation and correlation mining operation on the corresponding training flight data after the training optimization is completed. It should be noted that in the training process, in the process of the first round of iteration, the reference flight data semantic vector can not be used, and in step S133b, the vector fusion operation after the aircraft internal data semantic vector can be directly processed by self-attention to obtain the corresponding aircraft internal data correlation vector. In each round of iteration after the first round of iteration, the mean value of each training flight data semantic vector mined in the previous round of iteration is calculated to obtain the corresponding reference flight data semantic vector of the current round of iteration.

[0093] In step S133b, the aircraft internal data semantic vector after the vector fusion operation is subjected to correlation mining operation according to the reference flight data semantic vector, and the corresponding aircraft internal data correlation vector is output.

[0094] In the embodiment of the present application, the aircraft internal data semantic vector after the vector fusion operation is subjected to correlation mining operation according to the reference flight data semantic vector, and the corresponding aircraft internal data correlation vector is output. That is, the semantic information carried by the reference flight data semantic vector can be fused into the aircraft internal data semantic vector after the vector fusion operation. Based on this, the problem of excessive loss of real semantic of aircraft internal data caused by introduction of noise can be avoided.

[0095] In step S133c, the aircraft internal data correlation vector is taken as the output vector of the vector mining unit, and convolution mining operation is performed to output the corresponding aircraft internal depth vector.

[0096] In the embodiment of the present application, the aircraft internal data correlation vector can be taken as the output vector of the vector mining unit, and convolution mining operation (the parameters of convolution operation are not specifically limited here) is performed to output the corresponding aircraft internal depth vector.

[0097] It can be understood that the specific manner of performing the correlation mining operation on the aircraft internal data semantic vector after the vector fusion operation in step S133b is not limited, for example, in an alternative embodiment, step S133b can further include the following contents:

[0098] Firstly, the correlation parameter between the reference flight data semantic vector and the aircraft internal data semantic vector after the vector fusion operation can be determined (such as multiplying the reference flight data semantic vector and the aircraft internal data semantic vector after the vector fusion operation), and based on the correlation parameter, the correlation adjustment is performed on the aircraft internal data semantic vector after the vector fusion operation (such as weighting the aircraft internal data semantic vector after the vector fusion operation based on the correlation parameter), and an aircraft internal data adjustment vector is output.

[0099] Then, the vector fusion operation can be performed on the aircraft internal data semantic vector after the vector fusion operation and the aircraft internal data adjustment vector (such as superimposition operation on the aircraft internal data semantic vector after the vector fusion operation and the aircraft internal data adjustment vector), and a corresponding aircraft internal data correlation vector is output.

[0100] For step S140, it should be noted that the specific manner of analyzing the aircraft fault early warning information corresponding to the to-be-processed flight data is not limited, and can be selected according to actual needs. For example, in an alternative embodiment, step S140 can include the following contents:

[0101] Firstly, the correlation parameter between the flight data strengthening vector corresponding to the semantic mining unit and the aircraft internal depth vector can be determined, and based on the correlation parameter, the correlation adjustment is performed on the aircraft internal depth vector, and the corresponding adjusted aircraft internal depth vector is output, as described above.

[0102] Secondly, the vector fusion operation can be performed on the adjusted aircraft internal depth vector and the aircraft internal depth vector, and the corresponding target flight data semantic vector is output, as described above.

[0103] Then, the target flight data semantic vector can be subjected to a full connection operation, and a corresponding full connection flight data semantic vector is output, and the aircraft fault early warning information corresponding to the to-be-processed flight data is analyzed according to the full connection flight data semantic vector. For example, the full connection flight data semantic vector can be processed based on an output function included in the aircraft fault prediction network to obtain a probability distribution of whether the target aircraft has a fault, and then a case with a larger probability (with a fault or without a fault) can be taken as the aircraft fault early warning information. In addition, the output function can be a classification function, which can be different according to different classification requirements, and therefore is not specifically limited in the embodiments of the present application.

[0104] In combination Figure 3 The embodiments of the present application also provide an aircraft fault early warning system based on flight data, which can be applied to the electronic device. The aircraft fault early warning system based on flight data can include a flight data semantic mining module, a vector semantic strengthening module, a vector deep mining module, and an aircraft fault analysis module.

[0105] In detail, the flight data semantic mining module can be used to mine an aircraft external data semantic vector corresponding to to-be-processed flight data and mine an aircraft internal data semantic vector corresponding to the to-be-processed flight data. The to-be-processed flight data includes aircraft internal data and aircraft external data. The aircraft internal data is used to reflect the state of each aircraft component of a target aircraft during flight, and the aircraft external data is used to reflect the state of the environment in which the target aircraft is located during flight. In the embodiments of the present application, the flight data semantic mining module can be used to perform the step S110 shown in the figure, and the related content of the flight data semantic mining module can be referred to the foregoing description of the step S110. Figure 2 The flight data semantic mining module can be used to perform the step S110 shown in the figure, and the related content of the flight data semantic mining module can be referred to the foregoing description of the step S110.

[0106] In detail, the vector semantic strengthening module can be used to load the aircraft external data semantic vector, so that the aircraft fault prediction network obtains the aircraft external data semantic vector, and a semantic strengthening operation is performed on a pre-generated flight data noise vector based on the aircraft external data semantic vector by using a semantic mining unit included in the aircraft fault prediction network, and a flight data strengthened vector corresponding to the semantic mining unit is output. In the embodiments of the present application, the vector semantic strengthening module can be used to perform the step S120 shown in the figure, and the related content of the vector semantic strengthening module can be referred to the foregoing description of the step S120. Figure 2 The vector semantic strengthening module can be used to perform the step S120 shown in the figure, and the related content of the vector semantic strengthening module can be referred to the foregoing description of the step S120.

[0107] In detail, the vector deep mining module can be configured to mine the aircraft interior deep vector corresponding to the semantic mining unit according to the aircraft interior data semantic vector. In the embodiment of the present application, the vector deep mining module can be configured to perform Figure 2 The step S130 is shown. For details of the vector deep mining module, refer to the foregoing description of the step S130.

[0108] In detail, the aircraft fault analysis module can be configured to perform a correlation mining operation on the aircraft interior deep vector according to the flight data reinforced vector corresponding to the semantic mining unit, output a target flight data semantic vector, and analyze aircraft fault warning information corresponding to the to-be-processed flight data according to the target flight data semantic vector, wherein the aircraft fault warning information is used to reflect whether the target aircraft has a fault. In the embodiment of the present application, the aircraft fault analysis module can be configured to perform Figure 2 The step S140 is shown. For details of the aircraft fault analysis module, refer to the foregoing description of the step S140.

[0109] It can be understood that, in an alternative embodiment, the vector semantic reinforcement module is specifically configured to: load the aircraft exterior data semantic vector, so that the aircraft fault prediction network obtains the aircraft exterior data semantic vector, and use a semantic mining unit included in the aircraft fault prediction network to mine a flight data noise mining vector according to a pre-generated flight data noise vector; map the flight data noise mining vector to form a corresponding flight data mapping vector; respectively map the aircraft exterior data semantic vector to form a flight first mapping vector and a flight second mapping vector; perform a focus fusion operation according to the flight data mapping vector, the flight first mapping vector, and the flight second mapping vector, and obtain the flight data reinforced vector corresponding to the semantic mining unit according to an output vector of the focus fusion operation.

[0110] It can be understood that, in an alternative embodiment, the vector deep mining module is specifically configured to: determine a flight data noise vector of a semantic mining unit included in the aircraft fault prediction network; perform a convolution mining operation on the aircraft interior data semantic vector, and perform a vector fusion operation on the aircraft interior data semantic vector after the convolution mining operation and the flight data noise vector to output a corresponding aircraft interior data semantic vector after the vector fusion operation; load the aircraft interior data semantic vector after the vector fusion operation, so that a vector mining unit corresponding to the semantic mining unit obtains the aircraft interior data semantic vector after the vector fusion operation and performs corresponding processing, and perform a convolution mining operation on an output vector of the vector mining unit to output a corresponding aircraft interior deep vector.

[0111] In the embodiments of the present application, corresponding to the above-mentioned method for early warning of aircraft failure based on flight data applied to the electronic device, a computer readable storage medium is also provided, and the computer readable storage medium stores a computer program. The computer program performs each step of the method for early warning of aircraft failure based on flight data when running.

[0112] The steps performed by the computer program when running are not described here again, and can be referred to the above description of the method for early warning of aircraft failure based on flight data.

[0113] In summary, the method and system for early warning of aircraft failure based on flight data provided by the present application first mines the aircraft external data semantic vector and the aircraft internal data semantic vector corresponding to the to-be-processed flight data. Second, the pre-generated flight data noise vector is subjected to semantic strengthening operation based on the aircraft external data semantic vector, and a flight data strengthened vector is output. Then, the aircraft internal deep vector is mined according to the aircraft internal data semantic vector. Finally, the aircraft internal deep vector is subjected to associated mining operation according to the flight data strengthened vector, and a target flight data semantic vector is output. In addition, the aircraft failure early warning information is analyzed according to the target flight data semantic vector. Based on the above, on the one hand, the flight data strengthened vector carrying the semantic information of the aircraft external data and the aircraft internal deep vector carrying the semantic information of the aircraft internal data are subjected to associated mining operation, so that the semantic representation ability of the obtained target flight data semantic vector can be better. Therefore, the reliability of the aircraft failure early warning information analyzed based on the target flight data semantic vector is also higher. On the other hand, since the flight data strengthened vector not only carries the semantic information of the aircraft external data, but also fuses the flight data noise vector, the real situation (the collected aircraft external data may ignore some noise and interference) can be better simulated, thereby improving its performance in the actual environment. In this way, the representation ability of the aircraft data strengthened vector can also be further improved, thereby improving the problem of relatively low reliability of aircraft failure early warning in the prior art.

[0114] In several embodiments provided by the embodiments of the present application, it should be understood that the disclosed apparatus and method can also be implemented by other manners. The apparatus and method embodiments described above are only illustrative, for example, the flowcharts and block diagrams in the drawings show the possible implementation architecture, function and operation of the apparatus, method and computer program product according to the embodiments of the present application. In this regard, each block in the flowchart or block diagram can represent a module, a program segment or a part of code, which includes one or more executable instructions for implementing the specified logic function. It should also be noted that in some alternative implementation manners, the functions noted in the blocks can also occur in different order from that noted in the drawings. For example, two consecutive blocks can actually be executed substantially in parallel, and they can also be executed in reverse order depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system for executing the specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0115] In addition, the function modules in the embodiments of the present application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0116] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, and the computer software product is stored in a storage medium, and includes a number of instructions for causing a computer device (which can be a personal computer, an electronic device, or a network device, etc.) to execute all or part of the steps of the methods described in the embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes. It should be noted that in this document, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the processes, methods, articles or devices that include a series of elements not only include those elements, but also include other elements not explicitly listed, or include elements inherent to such processes, methods, articles or devices. Without more limitations, the element defined by the statement "includes a" does not exclude the presence of another identical element in the process, method, article or device that includes the element.

[0117] The above descriptions are only the preferred embodiment of the present application, but not for limiting the present application. For those skilled in the art, the present application can have various modifications and changes. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for early warning of aircraft faults based on flight data, characterized in that, include: The semantic vectors of external aircraft data corresponding to the flight data to be processed are mined, and the semantic vectors of internal aircraft data corresponding to the flight data to be processed are mined. The flight data to be processed includes internal aircraft data and external aircraft data. The internal aircraft data is used to reflect the state of each aircraft component of the target aircraft during flight, and the external aircraft data is used to reflect the state of the environment in which the target aircraft is located during flight. The process involves loading the aircraft external data semantic vector, enabling the aircraft fault prediction network to acquire the aircraft external data semantic vector, and utilizing the semantic mining unit included in the aircraft fault prediction network to perform semantic enhancement operations on a pre-generated flight data noise vector based on the aircraft external data semantic vector, outputting the flight data enhancement vector corresponding to the semantic mining unit. This includes: loading the aircraft external data semantic vector, enabling the aircraft fault prediction network to acquire the aircraft external data semantic vector, and utilizing the semantic mining unit included in the aircraft fault prediction network to mine a flight data noise mining vector based on the pre-generated flight data noise vector; mapping the flight data noise mining vector to form a corresponding flight data mapping vector; mapping the aircraft external data semantic vector to form a first flight mapping vector and a second flight mapping vector respectively; performing a focusing fusion operation based on the flight data mapping vector, the first flight mapping vector, and the second flight mapping vector; and obtaining the flight data enhancement vector corresponding to the semantic mining unit based on the output vector of the focusing fusion operation. Based on the semantic vector of the aircraft internal data, the method for mining the aircraft internal depth vector corresponding to the semantic mining unit includes: determining the flight data noise vector of the semantic mining unit included in the aircraft fault prediction network; performing a convolution mining operation on the aircraft internal data semantic vector; performing a vector fusion operation on the convolution mining operation-derived aircraft internal data semantic vector and the flight data noise vector to output the corresponding vector fusion operation-derived aircraft internal data semantic vector; loading the vector fusion operation-derived aircraft internal data semantic vector so that the vector mining unit corresponding to the semantic mining unit obtains the vector fusion operation-derived aircraft internal data semantic vector and performs corresponding processing; and performing a convolution mining operation on the output vector of the vector mining unit to output the corresponding aircraft internal depth vector. Based on the flight data enhancement vector corresponding to the semantic mining unit, an association mining operation is performed on the aircraft internal depth vector to output a target flight data semantic vector. Furthermore, based on the target flight data semantic vector, aircraft fault warning information corresponding to the flight data to be processed is analyzed. This includes: determining the association parameter between the flight data enhancement vector corresponding to the semantic mining unit and the aircraft internal depth vector; adjusting the aircraft internal depth vector based on the association parameter to output a corresponding adjusted aircraft internal depth vector; performing a vector fusion operation on the adjusted aircraft internal depth vector and the aircraft internal depth vector to output a corresponding target flight data semantic vector; performing a full connection operation on the target flight data semantic vector to output a corresponding fully connected flight data semantic vector; and analyzing the aircraft fault warning information corresponding to the flight data to be processed based on the fully connected flight data semantic vector. The aircraft fault warning information reflects whether the target aircraft has a fault.

2. The aircraft fault early warning method based on flight data according to claim 1, characterized in that, The step of loading the semantic vector of the aircraft external data, enabling the aircraft fault prediction network to obtain the semantic vector of the aircraft external data, and using the semantic mining unit included in the aircraft fault prediction network to perform semantic enhancement operation on the pre-generated flight data noise vector based on the semantic vector of the aircraft external data, and outputting the flight data enhancement vector corresponding to the semantic mining unit, further includes: The target parameters are determined, and a corresponding number of initial parameters are determined based on the target parameters, wherein there are multiple initial parameters; The corresponding initial flight data noise vector is determined based on the initial parameters, wherein each vector parameter in the initial flight data noise vector is determined by a random number generator based on the initial parameters; Based on the determined initial flight data noise vectors, the flight data noise vector is obtained.

3. The aircraft fault early warning method based on flight data according to claim 1, characterized in that, The steps of loading the semantic vector of the aircraft interior data after the vector fusion operation, so that the vector mining unit corresponding to the semantic mining unit obtains the semantic vector of the aircraft interior data after the vector fusion operation and performs corresponding processing, and performing a convolution mining operation on the output vector of the vector mining unit to output the corresponding aircraft interior depth vector, include: The semantic vector of the aircraft internal data after the vector fusion operation is loaded, so that the vector mining unit corresponding to the semantic mining unit obtains the semantic vector of the aircraft internal data after the vector fusion operation. Also, a predetermined semantic vector of reference flight data is loaded, so that the vector mining unit corresponding to the semantic mining unit obtains the semantic vector of the reference flight data. The semantic vector of the reference flight data is determined based on the semantic vectors of each training flight data mined by the semantic mining unit and the vector mining unit during the training and optimization process. Each training flight data semantic vector is obtained after training and optimization is completed by performing corresponding semantic enhancement and association mining operations on the corresponding training flight data. Based on the reference flight data semantic vector, an association mining operation is performed on the aircraft internal data semantic vector after the vector fusion operation, and the corresponding aircraft internal data association vector is output. The internal data association vector of the aircraft is used as the output vector of the vector mining unit, and a convolution mining operation is performed to output the corresponding internal depth vector of the aircraft.

4. The aircraft fault early warning method based on flight data according to claim 3, characterized in that, The step of performing association mining on the aircraft internal data semantic vector after the vector fusion operation based on the reference flight data semantic vector, and outputting the corresponding aircraft internal data association vector, includes: Determine the association parameter between the reference flight data semantic vector and the aircraft internal data semantic vector after the vector fusion operation, and based on the association parameter, adjust the association of the aircraft internal data semantic vector after the vector fusion operation, and output the aircraft internal data adjustment vector. Perform a vector fusion operation on the semantic vector of the aircraft internal data after the vector fusion operation and the adjusted vector of the aircraft internal data to output the corresponding aircraft internal data association vector.

5. An aircraft fault early warning system based on flight data, characterized in that, include: The flight data semantic mining module is used to mine the aircraft external data semantic vector corresponding to the flight data to be processed, and to mine the aircraft internal data semantic vector corresponding to the flight data to be processed. The flight data to be processed includes aircraft internal data and aircraft external data. The aircraft internal data is used to reflect the state of each aircraft component of the target aircraft during flight, and the aircraft external data is used to reflect the state of the environment in which the target aircraft is located during flight. The vector semantic enhancement module is used to load the semantic vector of the aircraft external data, so that the aircraft fault prediction network can obtain the semantic vector of the aircraft external data, and use the semantic mining unit included in the aircraft fault prediction network to perform semantic enhancement operation on the pre-generated flight data noise vector based on the semantic vector of the aircraft external data, and output the flight data enhancement vector corresponding to the semantic mining unit. The vector depth mining module is used to mine the aircraft interior depth vector corresponding to the semantic mining unit based on the semantic vector of the aircraft interior data. The aircraft fault analysis module is used to perform correlation mining operations on the aircraft internal depth vector based on the flight data enhancement vector corresponding to the semantic mining unit, output the target flight data semantic vector, and analyze the aircraft fault warning information corresponding to the flight data to be processed based on the target flight data semantic vector, wherein the aircraft fault warning information is used to reflect whether the target aircraft has a fault. Specifically, the vector semantic enhancement module is used for: The semantic vector of the aircraft external data is loaded so that the aircraft fault prediction network can obtain the semantic vector of the aircraft external data, and the semantic mining unit included in the aircraft fault prediction network can mine the flight data noise mining vector based on the pre-generated flight data noise vector. Based on the aforementioned flight data noise mining vector mapping, a corresponding flight data mapping vector is formed; Based on the semantic vectors of the aircraft's external data, a first flight mapping vector and a second flight mapping vector are respectively mapped to form them; A focusing fusion operation is performed based on the flight data mapping vector, the first flight mapping vector, and the second flight mapping vector, and the flight data enhancement vector corresponding to the semantic mining unit is obtained based on the output vector of the focusing fusion operation. Specifically, the vector depth mining module is used for: Determine the flight data noise vector of the semantic mining unit included in the aircraft fault prediction network; The aircraft internal data semantic vector is subjected to convolution mining operation, and the aircraft internal data semantic vector after convolution mining operation and the flight data noise vector are subjected to vector fusion operation to output the corresponding aircraft internal data semantic vector after vector fusion operation. The semantic vector of the aircraft interior data after the vector fusion operation is loaded, so that the vector mining unit corresponding to the semantic mining unit obtains the semantic vector of the aircraft interior data after the vector fusion operation and performs corresponding processing, and the output vector of the vector mining unit is subjected to convolution mining operation to output the corresponding aircraft interior depth vector. Specifically, the aircraft fault analysis module is used for: The association parameters between the flight data enhancement vector corresponding to the semantic mining unit and the aircraft interior depth vector are determined. Based on these association parameters, the aircraft interior depth vector is adjusted, and the adjusted aircraft interior depth vector is output. The adjusted aircraft interior depth vector and the aircraft interior depth vector are fused together to output the target flight data semantic vector. The target flight data semantic vector is fully connected to output the fully connected flight data semantic vector. Based on the fully connected flight data semantic vector, the aircraft fault warning information corresponding to the flight data to be processed is analyzed.

Images