A large model-based QAR data decoding method, device, equipment and medium

Abstract

The application discloses a large model-based QAR data decoding method and device, equipment and medium, relates to the field of aviation flight data processing, and comprises the following steps: preprocessing initial QAR data to generate processed QAR data, determining a target protocol version and parameter number by analyzing structural features by using a preset QAR analysis large model; generating initial flight data by analyzing the processed QAR data based on the target protocol version by using the preset QAR analysis large model, and determining a target sensor channel type and a target encoding mode; converting the initial flight data into converted flight data according to the target sensor channel type and the target encoding mode by using a target mapping function; generating standardized flight data by standardizing and naming the fields, and obtaining decoded flight data by unifying parameter units and parameter value ranges; and generating a visual flight report by sorting and aligning the decoded flight data according to a preset time sequence and a timestamp. The automatic analysis capability of QAR data is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of aviation flight data processing, and in particular to a QAR data decoding method and device based on a large model, equipment and medium. BACKGROUND

[0002] As a key flight data recording device on an aircraft, a QAR (Quick Access Recorder) can record hundreds of key flight parameters, such as attitude, speed, engine state, control input, etc. Its data format varies with random type, manufacturer and system architecture. For a long time, it has mainly relied on a static decoding method based on templates, which requires experts to manually prepare meta-information such as field meaning, offset and scaling factor.

[0003] The current QAR data decoding has the following technical bottlenecks:

[0004] Severe dependence on templates: unable to analyze without manufacturer template files, template loss or change leading to decoding failure;

[0005] Heavy rule configuration: field annotation needs to be manually maintained by experts, with large configuration workload and high error rate;

[0006] Poor generalization of parsing: large variation of fields between different models and manufacturers, lack of adaptive ability;

[0007] Lack of intelligent assistance: current decoding systems cannot perform semantic reasoning, field completion or interactive correction.

[0008] In summary, how to solve the problems of strong dependence on templates, high cost of manual rule maintenance and poor parsing adaptability of existing decoding methods is a problem to be solved at present. SUMMARY

[0009] Therefore, the purpose of the present application is to provide a QAR data decoding method, device, equipment and medium based on a large model, which can solve the problems of strong dependence on templates, high cost of manual rule maintenance and poor parsing adaptability of existing decoding methods. The specific scheme is as follows:

[0010] In a first aspect, the present application provides a QAR data decoding method based on a large model, comprising:

[0011] obtaining initial QAR data, preprocessing the initial QAR data to generate processed QAR data, and analyzing the structure characteristics of the processed QAR data using a preset QAR analysis large model to determine the target protocol version and parameter number corresponding to the processed QAR data;

[0012] resolve the data segment corresponding to the parameter number in the processed QAR data based on the preset QAR resolution large model to generate initial flight data, and determine a target sensor channel type and a target encoding mode corresponding to the parameter number;

[0013] construct a target mapping function according to the initial flight data, and convert the initial flight data into converted flight data representing actual physical quantities according to the target sensor channel type and the target encoding mode by using the target mapping function;

[0014] standardize the fields in the converted flight data to generate standardized flight data, unify the parameter units and parameter value ranges in the standardized flight data to obtain decoded flight data, sort the decoded flight data according to a preset time sequence and align the decoded flight data according to timestamps to generate a visual flight report.

[0015] Optionally, the structure characteristics of the processed QAR data are analyzed by using the preset QAR resolution large model to determine the target protocol version and the parameter number corresponding to the processed QAR data, including:

[0016] The data segment characteristics of the processed QAR data are identified based on the QAR structure mode in the preset database by using the preset QAR resolution large model;

[0017] The target protocol version corresponding to the processed QAR data is determined according to the identified data segment characteristics of the processed QAR data;

[0018] Statistical analysis and semantic matching operations are performed on the processed QAR data to obtain the parameter number corresponding to the processed QAR data.

[0019] Optionally, the target encoding mode corresponding to the parameter number is determined, including:

[0020] It is judged whether the data segment corresponding to the parameter number in the processed QAR data is a preset encoding mode according to a preset signal change mode;

[0021] The target encoding mode of the parameter number in the processed QAR data is determined according to the obtained judgment result.

[0022] Optionally, the fields in the converted flight data are standardized and named to generate standardized flight data, and the parameter units and parameter value ranges in the standardized flight data are unified to obtain decoded flight data, including:

[0023] The target name meaning of the fields in the converted flight data is identified by using a pre-trained language model;

[0024] map the fields in the converted flight data that meet the preset name meaning similarity condition to corresponding standard parameter names in the preset knowledge graph based on the target name meaning, to generate standardized flight data;

[0025] identify parameter units in the standardized flight data, and convert the parameter units in the standardized flight data to preset units according to a preset parameter unit rule, to generate unit-converted flight data;

[0026] standardize parameter value ranges in the unit-converted flight data according to a preset aviation general range to obtain decoded flight data.

[0027] Optionally, the sorting the decoded flight data according to a preset time sequence and aligning the decoded flight data according to timestamps to generate a visual flight report comprises:

[0028] aligning time sequences of each parameter in the decoded flight data according to a preset time sequence by using a dynamic time warping algorithm, to obtain time-aligned flight data;

[0029] identifying the time-aligned flight data based on a flight parameter change pattern to determine a target flight phase;

[0030] aligning the decoded flight data and target exogenous information according to timestamps, and then mapping the aligned target exogenous information to corresponding decoded flight data to generate flight operation data; wherein the target exogenous information includes any one or several of cockpit voice, flight data recorder data, and flight parameter recording system documents;

[0031] generating a visual flight report according to the target flight phase and the flight operation data.

[0032] Optionally, the method further comprises:

[0033] identifying abnormal data segments in the time-aligned flight data by using a preset anomaly detection model;

[0034] filling in the abnormal data segments in the time-aligned flight data based on a preset filling strategy, and scoring the filled data for credibility, to generate a visual flight report according to the credibility score.

[0035] Optionally, the visual flight report includes any one or several of a flight situation graph, a parameter trend graph, and an alarm event graph.

[0036] In a second aspect, the present application provides a QAR data decoding device based on a large model, comprising:

[0037] A version and number determination module is configured to obtain initial QAR data, preprocess the initial QAR data to generate processed QAR data, and analyze a structure feature of the processed QAR data by using a preset QAR analysis large model to determine a target protocol version and a parameter number corresponding to the processed QAR data.

[0038] A channel and coding determination module is configured to analyze a data segment corresponding to the parameter number in the processed QAR data by using the preset QAR analysis large model based on the target protocol version to generate initial flight data, and determine a target sensor channel type and a target coding mode corresponding to the parameter number.

[0039] A data conversion module is configured to construct a target mapping function according to the initial flight data, and convert the initial flight data into converted flight data representing an actual physical quantity by using the target mapping function according to the target sensor channel type and the target coding mode.

[0040] A report generation module is configured to perform standardized naming mapping on fields in the converted flight data to generate standardized flight data, unify parameter units and parameter value ranges in the standardized flight data to obtain decoded flight data, sort the decoded flight data according to a preset time sequence and align the decoded flight data according to time stamps to generate a visual flight report.

[0041] In a third aspect, the present application provides an electronic device, comprising:

[0042] A memory configured to save a computer program;

[0043] A processor configured to execute the computer program to implement the large model-based QAR data decoding method as described above.

[0044] In a fourth aspect, the present application provides a computer readable storage medium configured to save a computer program; wherein the computer program is executed by a processor to implement the large model-based QAR data decoding method as described above.

[0045] In summary, the application first acquires initial QAR data, pre-processes the initial QAR data to generate processed QAR data, analyzes the structural features of the processed QAR data using a preset QAR analysis large model to determine the target protocol version and parameter number corresponding to the processed QAR data; based on the target protocol version, the preset QAR analysis large model is used to analyze the data segment corresponding to the parameter number in the processed QAR data to generate initial flight data, and determine the target sensor channel type and target encoding mode corresponding to the parameter number; a target mapping function is constructed according to the initial flight data, and the initial flight data is converted into converted flight data representing actual physical quantities by using the target mapping function according to the target sensor channel type and target encoding mode; the fields in the converted flight data are standardized and named to generate standardized flight data, the parameter units and parameter value ranges in the standardized flight data are unified to obtain decoded flight data, and the decoded flight data is sorted according to a preset time sequence and aligned according to a timestamp to generate a visual flight report. As can be seen from the above, the application first acquires initial QAR data and pre-processes it, then analyzes the processed data with the help of a preset QAR analysis large model to determine the corresponding target protocol version and parameter number, and then parses the initial flight data based on the protocol version, while clearly specifying the target sensor channel type and target encoding mode corresponding to the parameter number. Then, a target mapping function is constructed according to the initial flight data, and the initial flight data is converted into converted flight data representing actual physical quantities in combination with the target sensor channel and encoding mode. Subsequently, the fields of the converted data are standardized and named, the parameter units and value ranges are unified to obtain decoded flight data, and finally, the decoded flight data is sorted according to a preset time sequence and aligned according to a timestamp to generate a visual flight report. In this way, the application combines the semantic understanding, data modeling and knowledge generalization capabilities of the large model, and constructs an end-to-end decoding process with self-adaptive capability to solve the key problems of complex QAR data sources, protocol closure and signal analysis difficulty, effectively improving the automatic analysis rate, analysis accuracy and system adaptability of QAR data. BRIEF DESCRIPTION OF DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on the provided drawings.

[0047] Figure 1 A flow chart of a QAR data decoding method based on a large model is disclosed in the present application.

[0048] Figure 2A specific large model-based QAR data decoding method flowchart disclosed in the present application;

[0049] Figure 3 A large model-based QAR data decoding device structure schematic diagram disclosed in the present application;

[0050] Figure 4 A structure diagram of an electronic device disclosed in the present application. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0052] At present, as a key flight data recording device on an aircraft, QAR can record hundreds of key flight parameters such as attitude, speed, engine state, control input, etc. Its data format varies with random type, manufacturer and system architecture. For a long time, it mainly relies on a static decoding method based on a template, and needs experts to manually prepare meta-information such as field meaning, offset and scaling factor. The current QAR data decoding has the following technical bottlenecks: severe dependence on templates: unable to analyze without manufacturer template files, template loss or change leading to decoding failure; heavy rule configuration: field annotation needs to be manually maintained by experts, large configuration workload and high error rate; poor generalization of analysis: large variation of fields among different models and different manufacturers, lack of adaptive ability; lack of intelligent assistance: the current decoding system cannot perform semantic reasoning, field completion or interactive correction. In order to solve the above technical problems, the present application discloses a large model-based QAR data decoding method, device, equipment and medium, which can solve the problems of strong dependence on templates, high cost of manual maintenance of rules and poor analysis adaptability of the existing decoding method.

[0053] Referring to Figure 1 The embodiments of the present application disclose a large model-based QAR data decoding method, which comprises:

[0054] Step S11, obtaining initial QAR data, preprocessing the initial QAR data to generate processed QAR data, and analyzing the structure characteristics of the processed QAR data by using a preset QAR analysis large model to determine the target protocol version and parameter number corresponding to the processed QAR data.

[0055] In this embodiment, first, the initial QAR data on the aircraft is acquired, then the initial QAR data is cleaned and standardized in format. The initial QAR data is often stored in binary or special format, containing a large number of undecoded flight parameters and signals. The initial QAR data can be first identified and unpacked in data format. By judging whether the initial QAR data is in ARINC717, ARINC429 or other standard format, the packet structure is disassembled, and the timestamp error in the initial QAR data is corrected. Then the signal frames are rearranged in time sequence, the physical layer noise is processed by using sliding window mean, median filtering and other methods, the missing frames are interpolated and filled, and the processed QAR data is obtained.

[0056] In this embodiment, after the processed QAR data is acquired, a preset QAR analysis large model is used to identify the data segment features of the processed QAR data based on the QAR structure mode in the preset database; the target protocol version corresponding to the processed QAR data is determined according to the identified data segment features of the processed QAR data; statistical analysis and semantic matching operations are performed on the processed QAR data to obtain the parameter number corresponding to the processed QAR data. It should be noted that due to the large difference in QAR protocols used by different aircraft models and manufacturers, traditional methods need to manually check protocol documents, which is low in efficiency and poor in fault tolerance. In order to determine the QAR data version on the aircraft, the preset QAR analysis large model can be used to identify the header, frame length, channel number, etc. of the data segment of the processed QAR data by comparing the structure mode in the known protocol library, so as to determine the target protocol version of the processed QAR data. At the same time, combined with statistical analysis and semantic matching, the characteristic parameters and their byte bit offsets are identified, and the parameter number corresponding to the processed QAR data is obtained. In addition, based on the existing part of the parameter label, pseudo-labeling and model self-learning can be performed on the unlabeled data.

[0057] In step S12, based on the target protocol version, the preset QAR analysis large model is used to analyze the data segment corresponding to the parameter number in the processed QAR data to generate initial flight data, and determine the target sensor channel type and target encoding mode corresponding to the parameter number.

[0058] In this embodiment, the processed QAR data is deeply analyzed by means of the preset QAR analysis large model, each data segment corresponding to the parameter number in the processed QAR data is accurately identified and extracted, and through the operation and analysis of the model, the relatively scattered and disordered data segments are converted into structured initial flight data. These initial flight data can preliminarily reflect the basic information in the flight process.

[0059] Next, since the flight data is composed of multiple sensor channels, different channel encoding methods and signal amplitude ranges vary, so it is necessary to identify the target sensor channel type in the initial flight data, such as analog signals of voltage and current, digital signals, event flags, etc. At the same time, according to the preset signal change mode, it is judged whether the data segment corresponding to the parameter number in the processed QAR data is of the preset encoding method; according to the obtained judgment result, the target encoding method of the parameter number in the processed QAR data is determined. Specifically, it can be determined whether it is BCD encoding, binary complement, IEEE754 floating point number encoding method, etc. according to the signal change mode of the processed QAR data, and the target encoding method of the processed QAR data is determined.

[0060] Step S13, constructing a target mapping function according to the initial flight data, and using the target mapping function according to the target sensor channel type and the target encoding method to convert the initial flight data into converted flight data representing actual physical quantities.

[0061] In this embodiment, after obtaining the initial flight data, the target mapping function between the channel signal and the real physical quantity is constructed by combining the parameter characteristics, sensor channel signal characteristics and encoding rules, to build a bridge between the channel signal and the real physical quantity, such as linear transformation, logarithmic relationship, etc. Common forms include linear transformation, logarithmic relationship, etc. For example, engine speed signal is often linearly transformed, and pressure related parameters are often in logarithmic form. Then, according to the target sensor channel type and the encoding method, the target mapping function is called to convert the initial flight data, and the initial flight data is converted into intuitive actual physical quantity data, i.e. converted flight data, which can be directly used in flight state evaluation, fault diagnosis and other scenes.

[0062] Step S14, standardizing and naming mapping of fields in the converted flight data to generate standardized flight data, unifying parameter units and parameter value ranges in the standardized flight data to obtain decoded flight data, and sorting and aligning the decoded flight data according to a preset time sequence to generate a visual flight report.

[0063] In this embodiment, due to the large number of flight data fields, the naming is not uniform, there are a large number of abbreviations, aliases and other cases, the field needs to be semantically normalized, the target name meaning of the field in the converted flight data can be recognized through a pre-trained language model; based on the target name meaning, the field in the converted flight data that meets the preset name meaning similarity condition is mapped to the corresponding standard parameter name in the preset knowledge graph to generate standardized flight data; the parameter unit in the standardized flight data is recognized, and the parameter unit in the standardized flight data is converted into a preset unit according to a preset parameter unit rule to generate unit-converted flight data; the parameter value range in the unit-converted flight data is standardized according to a preset aviation general range to obtain decoded flight data. Specifically, the target name meaning of the field is recognized through a pre-trained language model, such as recognizing “ALT”, “HGT”, “Altitude” as “height” uniformly. Similar fields are semantically vector clustered and automatically mapped to standard parameters in a unified knowledge graph to generate standardized flight data. The semantic disambiguation is performed by using the context of other fields, for example, according to the upper and lower fields “IAS” and “GS”, it can be judged that “SPD” represents airspeed or ground speed. It can be understood that different manufacturers and different models of QAR data may use different units or scale coefficients for the same parameter, which needs to be uniformly converted. The units such as “ft” and “m”, “knots” and “km / h” can be automatically recognized and converted to generate unit-converted flight data. The parameters are standardized according to the aviation general range, for example, the temperature is unified to -50℃~+150℃. Finally, the decoding of the initial QAR data is completed, and the decoded flight data is obtained.

[0064] Further, after decoding the initial QAR data, the state sequence of the complete flight process needs to be reconstructed in chronological order. The dynamic time warping algorithm is used to align the time sequences of each parameter in the decoded flight data according to the preset chronological order to obtain time-aligned flight data. The time-aligned flight data is identified based on the flight parameter change pattern to determine the target flight phase. The decoded flight data and the target external source information are aligned according to the time stamp, and then the aligned target external source information is mapped to the corresponding decoded flight data to generate flight operation data. The target external source information includes any one or several of cockpit voice, flight data recorder data, and flight parameter recording system documents. A visual flight report is generated according to the target flight phase and the flight operation data. Specifically, the dynamic time warping algorithm is used to align the time sequences of multiple parameters in the decoded flight data to ensure the continuity of the flight state. The target flight phase is identified according to the flight parameter change pattern of the decoded flight data, such as rolling, taking off, climbing, cruising, descending, and landing. Each time window can also be automatically labeled as a "climbing segment" or an "approach segment" by a rule engine or a weakly supervised model. In addition, the decoded flight data can also integrate multi-source information such as cockpit voice, flight data recorder data, and flight parameter recording system documents. The data from different systems are aligned according to a unified timeline, and then the voice command content in the CVR (Cockpit Voice Recorder) is mapped to the corresponding flight operation data points. A large model is used to jointly model and reconstruct events for video, audio, text, and flight parameters. Finally, the decoded flight data is converted into a structured report or a database format with strong readability and friendly visualization, such as generating flight situation graphs, parameter trend graphs, and warning event graphs. At the same time, multiple formats such as JSON (JavaScript Object Notation) and CSV (Comma-Separated Values) are supported, which facilitates subsequent analysis and archiving, and can provide a message queue to push the visual flight report to a flight quality system or an analysis platform.

[0065] As can be seen, the embodiment of the application first acquires initial QAR data and performs preprocessing, then analyzes the processed data by means of a preset QAR analysis large model, determines the corresponding target protocol version and parameter number, and then analyzes the initial flight data based on the protocol version, and determines the target sensor channel type and target encoding mode corresponding to the parameter number. Then, the target mapping function is constructed according to the initial flight data, and the initial flight data is converted into converted flight data representing the actual physical quantity in combination with the target sensor channel and the encoding mode. Then, the fields of the converted data are standardized and named, the parameter units and value ranges are unified to obtain decoded flight data, and finally, the decoded flight data is sorted in a preset time sequence and aligned by timestamp to generate a visual flight report. In this way, the application combines the semantic understanding, data modeling and knowledge generalization capabilities of the large model, and solves the key problems of complex QAR data sources, protocol closure and signal analysis difficulty, and constructs an end-to-end decoding process with self-adaptive capability, effectively improving the automatic analysis rate, analysis accuracy and system adaptability of QAR data.

[0066] Referring to Figure 2 As shown in the figure, in order to avoid the interference of abnormal data on QAR data analysis, the embodiment of the application discloses a QAR data decoding method based on a large model, comprising:

[0067] Step S21, identifying abnormal data segments in the time-aligned flight data by using a preset abnormal detection model.

[0068] In this embodiment, the preset abnormal detection model is used to identify abnormal data segments in the time-aligned flight data. This model integrates rules, statistical models and machine learning models to identify various abnormal data segments such as mutations, static values and drifts in the time-aligned flight data.

[0069] Step S22, filling the abnormal data segments in the time-aligned flight data based on a preset completion strategy, and scoring the filled data for credibility, to generate a visual flight report according to the credibility score.

[0070] In this embodiment, after identifying the abnormal data segments, the abnormal data segments in the time-aligned flight data are filled based on a preset completion strategy. The completion strategy includes interpolation of adjacent data, reconstruction based on physical modeling, and predictive filling based on historical models. At the same time, the filled data is scored for credibility, and finally a visual flight report is generated according to the credibility scores, providing valuable reference for subsequent related decisions.

[0071] From the above, the embodiment of the application avoids the interference of abnormal data on subsequent analysis from the source, and guarantees the basic reliability of data for analysis. At the same time, the mechanism of assigning a credibility score to the filled data adds a clear reliability identifier to the filled data, and the visual flight report generated in combination with the score also provides a more credible and intuitive reference basis for subsequent decision-making, both of which improve the accuracy and practicality of flight data processing.

[0072] Referring to Figure 3 As shown in the drawings, the embodiment of the application discloses a QAR data decoding device based on a large model, comprising:

[0073] A version and number determination module 11 is configured to obtain initial QAR data, preprocess the initial QAR data to generate processed QAR data, analyze the structural features of the processed QAR data by using a preset QAR analysis large model to determine the target protocol version and parameter number corresponding to the processed QAR data;

[0074] A channel and encoding determination module 12 is configured to analyze the data segment corresponding to the parameter number in the processed QAR data by using the preset QAR analysis large model based on the target protocol version to generate initial flight data, and determine the target sensor channel type and target encoding mode corresponding to the parameter number;

[0075] A data conversion module 13 is configured to construct a target mapping function according to the initial flight data, and convert the initial flight data into converted flight data representing actual physical quantities by using the target mapping function according to the target sensor channel type and target encoding mode;

[0076] A report generation module 14 is configured to perform standardized naming mapping on the fields in the converted flight data to generate standardized flight data, unify the parameter units and parameter value ranges in the standardized flight data to obtain decoded flight data, sort the decoded flight data according to a preset time sequence and align the time stamps to generate a visual flight report.

[0077] From the above, the application first acquires initial QAR data and pre-processes, then analyzes the processed data with the aid of a preset QAR analysis large model, determines the corresponding target protocol version and parameter number, and then analyzes the initial flight data based on the protocol version, and determines the target sensor channel type and target encoding mode corresponding to the parameter number. Then, according to the initial flight data, a target mapping function is constructed, and the initial flight data is converted into converted flight data representing the actual physical quantity by combining the target sensor channel and the encoding mode. Then, the fields of the converted data are standardized and named, the parameter units and value ranges are unified to obtain decoded flight data, and finally, the decoded flight data is sorted and aligned according to the preset time sequence and timestamp, and a visual flight report is generated. In this way, the application combines the semantic understanding, data modeling and knowledge generalization ability of the large model, and solves the key problems of complex QAR data sources, protocol closure and signal analysis difficulty, and constructs an end-to-end decoding process with self-adaptive ability, effectively improving the automatic analysis rate, analysis accuracy and system adaptability of QAR data.

[0078] In some specific embodiments, the version and number determination module 11 comprises:

[0079] A feature recognition unit is configured to recognize the data segment features of the processed QAR data based on the QAR structure mode in the preset database by using the preset QAR analysis large model.

[0080] A version determination unit is configured to determine the target protocol version corresponding to the processed QAR data according to the recognized data segment features of the processed QAR data.

[0081] A number acquisition unit is configured to perform statistical analysis and semantic matching operations on the processed QAR data to obtain the parameter number corresponding to the processed QAR data.

[0082] In some specific embodiments, the version and number determination module 11 comprises:

[0083] A data segment judgment unit is configured to determine whether the data segment corresponding to the parameter number in the processed QAR data is a preset encoding mode according to a preset signal change mode.

[0084] An encoding mode determination unit is configured to determine the target encoding mode of the parameter number in the processed QAR data according to the obtained determination result.

[0085] In some specific embodiments, the report generation module 14 comprises:

[0086] A name meaning recognition unit is configured to recognize the target name meaning of the fields in the converted flight data by using a pre-trained language model.

[0087] The first data generation unit is configured to map fields in the converted flight data that meet preset name meaning similarity conditions to corresponding standard parameter names in a preset knowledge graph based on the target name meaning, so as to generate standardized flight data.

[0088] The second data generation unit is configured to identify parameter units in the standardized flight data, and convert the parameter units in the standardized flight data into preset units according to a preset parameter unit rule, so as to generate unit-converted flight data.

[0089] The third data generation unit is configured to perform standardization processing on parameter value ranges in the unit-converted flight data according to a preset aviation general range, so as to obtain decoded flight data.

[0090] In some specific embodiments, the report generation module 14 comprises:

[0091] The data acquisition unit is configured to align time sequences of each parameter in the decoded flight data according to a preset time sequence by using a dynamic time warping algorithm, so as to obtain time-aligned flight data.

[0092] The flight phase determination unit is configured to identify the time-aligned flight data based on a flight parameter change pattern, so as to determine a target flight phase.

[0093] The flight operation data generation unit is configured to align the decoded flight data and target exogenous information according to timestamps, and then map the aligned target exogenous information to corresponding decoded flight data, so as to generate flight operation data. The target exogenous information includes any one or several of aircraft cabin voice, flight data recorder data, and flight parameter recording system documents.

[0094] The flight report generation unit is configured to generate a visual flight report according to the target flight phase and the flight operation data.

[0095] In some specific embodiments, the large model-based QAR data decoding device can further comprise:

[0096] The abnormal data segment identification module is configured to identify abnormal data segments in the time-aligned flight data by using a preset anomaly detection model.

[0097] The visual flight report generation module is configured to fill in the abnormal data segments in the time-aligned flight data based on a preset completion strategy, and perform credibility scoring on the filled data, so as to generate a visual flight report according to the credibility score.

[0098] In some specific embodiments, the visualized flight report includes any one or more of a flight situation map, a parameter trend map, and an alert event map.

[0099] Further, the embodiment of the present application further discloses an electronic device, Figure 4 is an electronic device 20 structure diagram shown according to an exemplary embodiment, the contents in the figure cannot be considered as any limitation on the use range of the present application.

[0100] Figure 4 An electronic device 20 structure diagram is provided for the embodiment of the present application. The electronic device 20 can specifically include at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25 and a communication bus 26. Among them, the memory 22 is used to store a computer program, the computer program is loaded and executed by the processor 21, to realize the related steps in the QAR data decoding method based on large model disclosed in any of the preceding embodiments. In addition, the electronic device 20 in the embodiment can be an electronic computer.

[0101] In the embodiment, the power supply 23 is used to provide working voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and the external device, and the communication protocol followed is any communication protocol applicable to the technical solution of the present application, which is not limited here; the input / output interface 25 is used to obtain external input data or output data to the outside world, and the specific interface type can be selected according to the specific application needs, which is not limited here.

[0102] In addition, the memory 22 as the carrier of resource storage can be a read-only memory, a random access memory, a magnetic disk or an optical disk, etc., and the resources stored thereon can include an operating system 221, a computer program 222, etc., and the storage mode can be temporary storage or permanent storage.

[0103] Among them, the operating system 221 is used to manage and control each hardware device on the electronic device 20 and the computer program 222, which can be Windows Server, Netware, Unix, Linux, etc. In addition to the computer program capable of completing the QAR data decoding method based on large model executed by the electronic device 20 disclosed in any of the preceding embodiments, the computer program 222 can further include a computer program capable of completing other specific work.

[0104] Further, the application also discloses a computer readable storage medium for storing a computer program, wherein the computer program is executed by a processor to realize the foregoing disclosed QAR data decoding method based on a large model. For specific steps of the method, refer to the corresponding content disclosed in the foregoing embodiments, which will not be repeated here.

[0105] The various embodiments are described in the specification by progressive stages, and each embodiment focuses on the difference from other embodiments. For the same or similar parts between the embodiments, refer to each other. For the device disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and refer to the method part for the relevant part.

[0106] The skilled person can further realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized by electronic hardware, computer software or a combination of both. In order to clearly show the interchangeability of hardware and software, the components and steps of the examples have been described in the above description. Whether the functions are realized by hardware or software depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the application.

[0107] The steps of the method or algorithm described in combination with the embodiments disclosed herein can be directly implemented by hardware, a software module executed by a processor, or a combination of both. The software module can be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0108] Finally, it should be noted that in this document, relationship terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the element defined by the statement "including a" does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0109] The technical solutions provided by the present application are described in detail above, and the principles and implementation manners of the present application are described by using specific examples. The above description of the examples is only used to help understand the method of the present application and its core idea; meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation manners and application ranges will be changed, and the above description of the content of the specification should not be understood as a limitation on the present application.

Claims

1. A QAR data decoding method based on a large model, characterized in that, include: Acquire initial QAR data, preprocess the initial QAR data to generate processed QAR data, and use a preset QAR parsing large model to analyze the structural features of the processed QAR data to determine the target protocol version and parameter number corresponding to the processed QAR data. Based on the target protocol version, the preset QAR parsing model is used to parse the data segments corresponding to the parameter numbers in the processed QAR data to generate initial flight data, and to determine the target sensor channel type and target encoding method corresponding to the parameter numbers. A target mapping function is constructed based on the initial flight data. The initial flight data is then converted into transformed flight data representing actual physical quantities using the target sensor channel type and target encoding method. The fields in the transformed flight data are standardized and named to generate standardized flight data. The parameter units and parameter value ranges in the standardized flight data are unified to obtain decoded flight data. The decoded flight data is sorted according to a preset time order and aligned by timestamps to generate a visual flight report. The step of using a preset QAR parsing large model to analyze the structural characteristics of the processed QAR data to determine the target protocol version and parameter number corresponding to the processed QAR data includes: The data segment features of the processed QAR data are identified using a preset QAR parsing large model based on the QAR structure pattern in a preset database. The target protocol version corresponding to the processed QAR data is determined based on the identified data segment characteristics of the processed QAR data; Perform statistical analysis and semantic matching on the processed QAR data to obtain the parameter number corresponding to the processed QAR data; Determining the target encoding method corresponding to the parameter number includes: Based on the preset signal change pattern, determine whether the data segment corresponding to the parameter number in the processed QAR data is a preset encoding method; Based on the obtained judgment result, determine the target encoding method for the parameter number in the processed QAR data.

2. The QAR data decoding method based on a large model according to claim 1, characterized in that, The process of standardizing and mapping the fields in the transformed flight data to generate standardized flight data, and unifying the parameter units and value ranges in the standardized flight data to obtain decoded flight data, includes: The meaning of the target names in the transformed flight data is identified by a pre-trained language model. Based on the meaning of the target name, fields in the transformed flight data that meet the preset name meaning similarity conditions are mapped to the corresponding standard parameter names in the preset knowledge graph to generate standardized flight data; Identify the parameter units in the standardized flight data, and convert the parameter units in the standardized flight data into preset units according to preset parameter unit rules to generate unit-converted flight data; The parameter value range in the unit-converted flight data is standardized according to a preset aviation general range to obtain the decoded flight data.

3. The QAR data decoding method based on a large model according to any one of claims 1 to 2, characterized in that, The step of sorting the decoded flight data according to a preset time order and aligning it by timestamps to generate a visual flight report includes: The time series of each parameter in the decoded flight data are aligned according to a preset time order using a dynamic time warping algorithm to obtain time-aligned flight data. Based on the identification of flight parameter variation patterns, the time-aligned flight data is used to determine the target flight phase; The decoded flight data and target external information are aligned by timestamp, and then the aligned target external information is mapped to the corresponding decoded flight data to generate flight operation data; wherein, the target external information includes any one or more of the following: aircraft cockpit voice, flight data recorder data, and flight parameter recording system documents; A visual flight report is generated based on the target flight phase and the flight operation data.

4. The QAR data decoding method based on a large model according to claim 3, characterized in that, Also includes: An abnormal data segment in the time-aligned flight data is identified using a preset anomaly detection model. Based on a preset completion strategy, abnormal data segments in the time-aligned flight data are filled in, and a credibility score is given to the filled data to generate a visual flight report based on the credibility score.

5. The QAR data decoding method based on a large model according to claim 1, characterized in that, The visualized flight report includes any one or more of the following: flight situation diagram, parameter trend diagram, and alarm event diagram.

6. A QAR data decoding device based on a large model, characterized in that, include: The version and number determination module is used to obtain initial QAR data, preprocess the initial QAR data to generate processed QAR data, and use a preset QAR parsing big model to analyze the structural features of the processed QAR data to determine the target protocol version and parameter number corresponding to the processed QAR data. The channel and encoding determination module is used to use the preset QAR parsing large model based on the target protocol version to parse the data segment corresponding to the parameter number in the processed QAR data to generate initial flight data, and to determine the target sensor channel type and target encoding method corresponding to the parameter number; The data conversion module is used to construct a target mapping function based on the initial flight data, and to convert the initial flight data into converted flight data representing actual physical quantities using the target sensor channel type and target encoding method according to the target mapping function. The report generation module is used to standardize the naming mapping of fields in the transformed flight data to generate standardized flight data, unify the parameter units and parameter value ranges in the standardized flight data to obtain decoded flight data, and sort the decoded flight data according to a preset time order and align it by timestamp to generate a visual flight report. The version and number determination module includes: The feature recognition unit is used to identify the data segment features of the processed QAR data based on the QAR structure pattern in the preset database using a preset QAR parsing large model; The version determination unit is used to determine the target protocol version corresponding to the processed QAR data based on the identified data segment characteristics of the processed QAR data. The number acquisition unit is used to perform statistical analysis and semantic matching operations on the processed QAR data to obtain the parameter number corresponding to the processed QAR data. The version and number determination module includes: The data segment judgment unit is used to determine whether the data segment corresponding to the parameter number in the processed QAR data is a preset encoding method according to the preset signal change pattern. The encoding method determination unit is used to determine the target encoding method of the parameter number in the processed QAR data based on the obtained judgment result.

7. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the large-model-based QAR data decoding method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, Used to store computer programs; wherein, when the computer programs are executed by a processor, they implement the large-model-based QAR data decoding method as described in any one of claims 1 to 5.

Images