Risk assessment method and storage medium

Abstract

The invention provides a risk assessment method and a storage medium, relates to the technical field of aviation, and is used for assessing the risk of aircraft navigation. The method comprises the following steps: acquiring aviation data of an aircraft; feature extraction is carried out on the aviation data to obtain aviation risk features, the aviation risk features comprise an over-limit risk feature, an aircraft state control feature and an environment risk feature, and the over-limit risk feature is used for reflecting the degree that aircraft operation parameters of the aircraft approach or exceed a preset navigation safety threshold value; the aircraft state control feature is used for reflecting the aircraft navigation control stability degree of a pilot, and the environment risk feature is used for reflecting the risk degree of the external environment of the aircraft. Based on the aviation risk characteristics, an aviation risk assessment result is determined, and the aviation risk assessment result is used for indicating the risk degree of the safety accident of the aircraft.

Description

Technical Field

[0001] This invention relates to the field of aviation technology, and in particular to a risk assessment method and storage medium. Background Technology

[0002] With the development of air transport, the route network continues to expand and the flight density continues to increase, which gradually increases the probability of risks during aircraft flights.

[0003] Therefore, how to assess the risks of aircraft flight in order to reduce the occurrence of safety accidents has become an urgent technical problem to be solved. Summary of the Invention

[0004] The purpose of this application is to provide a risk assessment method and storage medium, which aims to solve the problem of how to assess the risks of aircraft flight.

[0005] To achieve the above objectives, in a first aspect, this application provides a risk assessment method, which includes: acquiring aircraft aeronautical data; extracting features from the aeronautical data to obtain aeronautical risk features, including: exceedance risk features, aircraft state control features, and environmental risk features. Exceedance risk features reflect the degree to which aircraft operating parameters approach or exceed preset flight safety thresholds; aircraft state control features reflect the stability of pilot control of the aircraft; and environmental risk features reflect the degree of risk of the aircraft's external environment. Based on the aeronautical risk features, an aeronautical risk assessment result is determined, which indicates the degree of risk of an aircraft safety accident.

[0006] Based on the above technical solution, by extracting features from aviation data, aviation risk characteristics can be obtained, and characteristic data affecting aircraft flight safety can be identified. Subsequently, aviation risk assessment results can be determined based on these aviation risk characteristics. Thus, accurate assessment of the risks of aircraft flight can be achieved through over-limit risk characteristics, aircraft condition control characteristics, and environmental risk characteristics.

[0007] In some embodiments, the aviation data includes: aircraft status text data, which includes at least one of the following: pilot report text, aircraft maintenance record text, or weather report, wherein the pilot report text describes the pilot's flying conditions during the flight, and the weather report describes the weather conditions during the flight. The aviation risk features also include: risk event features, which indicate risk events in the aircraft status text data.

[0008] In some embodiments, risk event characteristics are determined as follows: Keyword extraction is performed on aircraft status text data to obtain risk event keywords. These risk event keywords are compared with preset risk event keywords to obtain keyword matching results for the aircraft status text data. The keyword matching results indicate whether the risk event keywords match the preset risk event keywords. Based on the risk event matching results and the risk event keywords, risk event characteristics are determined.

[0009] In some embodiments, the aviation data further includes: Quick Access Recorder (QAR) data and flight environment data, the flight environment data representing the external environmental conditions of the aircraft's location during flight. Feature extraction is performed on the aviation data to obtain aviation risk features, including: feature extraction from the QAR data to obtain over-limit risk features and aircraft state control features; and feature extraction from the flight environment data to obtain environmental risk features.

[0010] In some embodiments, QAR data includes: aircraft operation data corresponding to multiple exceedance risk indicators. The aircraft operation data represents the changes in aircraft operating parameters during flight. Feature extraction is performed on the QAR data to obtain exceedance risk features, including: for target aircraft operation data, determining multiple sub-operation data corresponding to the target aircraft operation data based on multiple flight stages corresponding to the target exceedance risk indicators. The target aircraft operation data is the aircraft operation data corresponding to any one of the multiple exceedance risk indicators. The target exceedance risk indicator is any one of the multiple exceedance risk indicators. Based on a preset flight safety threshold corresponding to the flight stage to which the sub-operation data belongs, and with each sub-operation data, determining the sub-exceedance risk feature corresponding to the sub-operation data. Determining the exceedance risk feature based on the sub-exceedance risk feature corresponding to each of the multiple sub-operation data.

[0011] In some embodiments, the preset flight safety threshold is obtained by: acquiring an initial safety threshold corresponding to the aircraft's operating parameters; adjusting the initial safety threshold based on airport data and / or flight meteorological data to obtain the preset flight safety threshold, where the airport data reflects the flight difficulty of the aircraft's takeoff or landing airport, and the flight meteorological data reflects the weather conditions during the aircraft's flight.

[0012] In some embodiments, the aviation risk assessment results are determined based on aviation risk characteristics, including: Aviation risk features are input into the embedding layer of the risk assessment model. The embedding layer performs a linear transformation on the aviation risk features to obtain a high-dimensional embedding vector. This high-dimensional embedding vector is then positionally encoded by the risk assessment model's positional encoding module, resulting in an encoded high-dimensional embedding vector. The multi-head self-attention layer of the risk assessment model processes the encoded high-dimensional embedding vector to obtain an attention weight matrix and a value matrix. These matrices are then weighted to obtain an enhanced feature vector. The feedforward network layer of the risk assessment model performs a non-linear correlation on the enhanced feature vector to obtain a global feature vector. Finally, the output layer of the risk assessment model processes the global feature vector to obtain the aviation risk assessment result.

[0013] In some embodiments, the risk assessment method further includes: extracting weights from the attention weight matrix to obtain multiple attention weights, which are used to indicate the degree of influence of the flight data corresponding to the attention weight on the aviation risk assessment result; and performing aggregate analysis based on aviation risk features and multiple attention weights to obtain multiple target risk features corresponding to the aviation risk assessment result and the risk contribution rate corresponding to each target risk feature, wherein the target risk feature is an aviation risk feature whose corresponding risk contribution rate is greater than or equal to a preset contribution rate threshold.

[0014] Secondly, this application provides a risk assessment device, which includes an acquisition module and a processing module.

[0015] The acquisition module is used to acquire aircraft aeronautical data. The processing module is used to extract features from the aeronautical data to obtain aeronautical risk features, including: exceedance risk features, aircraft status control features, and environmental risk features. Exceedance risk features reflect the degree to which aircraft operating parameters approach or exceed preset flight safety thresholds; aircraft status control features reflect the stability of pilot control of the aircraft; and environmental risk features reflect the degree of risk from the external environment. The processing module also determines the aeronautical risk assessment result based on the aeronautical risk features, which indicates the degree of risk of an aircraft safety incident.

[0016] In some embodiments, the aviation data includes: aircraft status text data, which includes at least one of the following: pilot report text, aircraft maintenance record text, or weather report, wherein the pilot report text describes the pilot's flying conditions during flight, and the weather report describes the weather conditions during flight. The aviation risk features also include: risk event features, which indicate risk events in the aircraft status text data.

[0017] In some embodiments, the processing module is configured to extract keywords from the aircraft status text data to obtain risk event keywords. The processing module is further configured to compare the risk event keywords with preset risk event keywords to obtain keyword matching results for the aircraft status text data. These keyword matching results indicate whether the risk event keywords match the preset risk event keywords. The processing module is also configured to determine risk event characteristics based on the risk event matching results and the risk event keywords.

[0018] In some embodiments, the aviation data further includes: Quick Access Recorder (QAR) data and flight environment data, wherein the flight environment data represents the external environmental conditions of the aircraft's location during flight. A processing module is used to extract features from the QAR data to obtain over-limit risk features and aircraft state control features. The processing module is also used to extract features from the flight environment data to obtain environmental risk features.

[0019] In some embodiments, QAR data includes: aircraft operation data corresponding to multiple exceedance risk indicators. The aircraft operation data represents the changes in aircraft operating parameters during flight. The processing module is used to determine multiple sub-operation data corresponding to the target aircraft operation data, based on multiple flight stages corresponding to the target exceedance risk indicators. The target aircraft operation data is the aircraft operation data corresponding to any one of the multiple exceedance risk indicators. The target exceedance risk indicator is any one of the multiple exceedance risk indicators. The processing module is further used to determine a sub-exceedance risk characteristic corresponding to each sub-operation data, based on a preset flight safety threshold corresponding to the flight stage to which the sub-operation data belongs, and the sub-operation data itself. The processing module is further used to determine an exceedance risk characteristic based on the sub-exceedance risk characteristic corresponding to each of the multiple sub-operation data.

[0020] In some embodiments, the acquisition module is used to acquire the initial safety threshold corresponding to the aircraft operating parameters. The processing module is used to adjust the initial safety threshold based on airport data and / or flight meteorological data to obtain a preset flight safety threshold. The airport data is used to reflect the flight difficulty of the aircraft's takeoff airport or landing airport, and the flight meteorological data is used to reflect the weather conditions during the aircraft's flight.

[0021] In some embodiments, the processing module is configured to input aviation risk features into the embedding layer of the risk assessment model, perform a linear transformation on the aviation risk features through the embedding layer to obtain a high-dimensional embedding vector, and then perform position encoding on the high-dimensional embedding vector through the position encoding module of the risk assessment model to obtain an encoded high-dimensional embedding vector. The processing module is further configured to perform data processing on the encoded high-dimensional embedding vector through the multi-head self-attention layer of the risk assessment model to obtain an attention weight matrix and a value matrix, and then perform weighted processing on the attention weight matrix and the value matrix to obtain an enhanced feature vector. The processing module is further configured to perform nonlinear correlation on the enhanced feature vector through the feedforward network layer of the risk assessment model to obtain a global feature vector. The processing module is further configured to perform data processing on the global feature vector through the output layer of the risk assessment model to obtain the aviation risk assessment result.

[0022] In some embodiments, the processing module is configured to extract weights from the attention weight matrix to obtain multiple attention weights, which indicate the degree of influence of the flight data corresponding to the attention weight on the aviation risk assessment result. The processing module is also configured to perform aggregation analysis based on aviation risk features and multiple attention weights to obtain multiple target risk features corresponding to the aviation risk assessment result and the risk contribution rate corresponding to each target risk feature. The target risk features are aviation risk features whose corresponding risk contribution rate is greater than or equal to a preset contribution rate threshold.

[0023] Thirdly, this application provides a risk assessment apparatus, comprising a processor and a memory. The processor and memory are coupled. The memory stores one or more programs, which include computer-executable instructions. When the risk assessment apparatus is running, the processor executes the computer-executable instructions stored in the memory to implement the methods described in the first aspect and any possible implementation thereof.

[0024] Fourthly, this application provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the methods described in the first aspect and any possible implementation thereof.

[0025] Fifthly, this application provides a computer program product containing instructions that, when executed by a computer, cause the computer to perform the methods described in the first aspect and any possible implementation thereof.

[0026] The technical problems that the risk assessment device, computer storage medium, or computer program product can solve and the technical effects it can achieve can be found in the technical problems and effects solved in the first aspect above, and will not be repeated here. Attached Figure Description

[0027] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A flowchart illustrating a risk assessment method provided in an embodiment of this application; Figure 2 This is a schematic diagram of the structure of a risk assessment device provided in an embodiment of this application; Figure 3 This is a schematic diagram of another risk assessment device provided in an embodiment of this application; Figure 4 A conceptual partial view of a computer program product provided for an embodiment of this application. Detailed Implementation

[0028] 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. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0029] In the description of this application, it should be understood that the terms "upper," "lower," "left," "right," "front," "rear," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or relative positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this application and for simplification, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application. Unless otherwise specified, the above-mentioned orientational descriptions can be flexibly set in practical applications, provided that the relative positional relationships shown in the accompanying drawings are satisfied.

[0030] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0031] In embodiments of this application, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, article, or apparatus that includes that element.

[0032] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0033] Quick access recorder (QAR) data is a core data source for aircraft safety management. Traditional methods determine whether there is a safety risk by setting critical values ​​for key aircraft operating parameters (such as bank angle exceeding limits and pitch angle exceeding limits) and monitoring the aircraft operating parameters in QAR data in real time to see if they exceed safe ranges.

[0034] However, this method is relatively simple, only identifying safety risks when aircraft operating parameters exceed fixed thresholds, ignoring parameters that, while not exceeding limits, still exhibit risk tendencies. This overlooks operational issues, leading to an incomplete assessment of aircraft safety risks. Furthermore, it cannot dynamically adjust risk assessment standards based on different aircraft types, weather conditions, and airport environments, resulting in inaccurate assessments. Additionally, this method isolates QAR data from textual data such as pilot reports and maintenance records, lacking cross-validation and fusion of multi-source information, making it difficult to identify coupled risks in complex systems. Moreover, this method relies on expert experience and is insensitive to novel and unknown risk patterns, typically relying on post-event analysis, resulting in insufficient risk warning capabilities. Finally, this method often uses the number of exceedances as a rough measure of risk level, lacking a comprehensive, horizontally comparable quantitative indicator to accurately measure safety levels.

[0035] To address the aforementioned technical problems, this application provides a risk assessment method. In this method, by extracting features from aviation data, aviation risk features are obtained, and characteristic data affecting aircraft flight safety can be identified. Subsequently, based on these aviation risk features, an aviation risk assessment result can be determined. Thus, an accurate assessment of the risks of aircraft flight can be achieved through over-limit risk features, aircraft condition control features, and environmental risk features.

[0036] The risk assessment method provided in this application can be applied to a risk assessment device, specifically to the processor of that device. The risk assessment device can be an electronic device such as a personal computer (PC), laptop computer, mobile device, tablet computer, or laptop computer; this application does not limit the specific form of the electronic device. Alternatively, the risk assessment device can also be a server, or a server cluster consisting of multiple servers; in some implementations, the server cluster can be a distributed cluster server. This application does not impose any limitations in this regard.

[0037] like Figure 1 As shown, this application provides a risk assessment method, which includes: S101. Obtain the aircraft's aeronautical data.

[0038] In this embodiment of the application, the aviation data includes: QAR data, aircraft status text data, and flight environment data. The flight environment data is used to represent the external environmental conditions of the aircraft's location during flight.

[0039] Optionally, the aircraft status text data includes at least one of the following: pilot report text, aircraft maintenance record text, or weather report, wherein the pilot report text is used to indicate the pilot's flying status during the flight, and the weather report is used to indicate the weather conditions during the flight.

[0040] Among these, pilot reports reflect risk events encountered by pilots during flight, such as pilot fatigue, adverse weather conditions, and mental stress. Aircraft maintenance records reflect past risk events, such as oil pump failure. Weather reports reflect weather conditions during flight, thus identifying corresponding weather-related risk events. Therefore, aircraft status text data accurately reflects the aircraft's risk situation.

[0041] Optionally, the flight environment data includes: airport data, flight meteorological data, and flight mission attribute data. Airport data is used to reflect the flight difficulty of the aircraft's takeoff airport or landing airport, and flight meteorological data is used to reflect the weather conditions during the flight.

[0042] Specifically, airport data includes at least one of the following airport parameters: airport altitude, runway length, runway width, runway slope, or information on obstacles around the runway. Navigation meteorological data includes at least one of the following meteorological parameters: wind direction, wind speed, rate of change of wind speed, cloud top height, cloud cover, turbulence intensity, icing intensity, precipitation intensity, fog concentration, or surface pressure. Flight mission attribute data includes at least one of the following mission parameters: aircraft flight number, pilot number, aircraft type, or aircraft registration number.

[0043] In one possible implementation, the aircraft's aeronautical data is pre-stored in a database of the risk assessment device, and the aircraft's aeronautical data can be retrieved from the database by establishing a connection with it.

[0044] S102. Extract features from aviation data to obtain aviation risk features.

[0045] In this embodiment of the application, aviation risk features include: over-limit risk features, aircraft state control features, environmental risk features, and risk event features. Over-limit risk features are used to indicate the degree to which the aircraft's operating parameters approach or exceed a preset flight safety threshold. Aircraft state control features are used to indicate the degree of stability of the pilot's control over the aircraft's flight. Environmental risk features are used to indicate the degree of risk of the aircraft's external environment. Risk event features are used to indicate risk events in the aircraft state text data.

[0046] In one possible implementation, feature extraction can be performed on the QAR data to obtain over-limit risk features and aircraft state control features. Subsequently, feature extraction can be performed on the flight environment data to obtain environmental risk features.

[0047] QAR data reflects changes in aircraft operating parameters during flight, as well as pilot maneuvers and the aircraft's corresponding attitude responses. Therefore, by comparing aircraft operating parameters in QAR data with corresponding limits, the extent to which these parameters approach or exceed preset flight safety thresholds can be determined, thus identifying exceedance risk characteristics. Statistical analysis of flight parameters related to pilot maneuvers in QAR data determines the stability of pilot control, thereby identifying aircraft state control characteristics. Flight environment data reflects the external environment in which the aircraft is located during flight. Therefore, feature extraction from flight environment data reveals the risk level of the external environment, thus determining the environmental risk characteristics.

[0048] Therefore, feature extraction from QAR data can accurately obtain over-limit risk characteristics and aircraft state control characteristics, while feature extraction from flight environment data can accurately obtain environmental risk characteristics.

[0049] In this embodiment, the QAR data includes aircraft operation data corresponding to multiple exceedance risk indicators. The aircraft operation data is used to represent the changes in aircraft operating parameters during flight.

[0050] Optionally, the over-limit risk characteristics are determined as follows: For target aircraft operational data, multiple sub-operational data points corresponding to the target aircraft operational data can be determined based on multiple flight stages corresponding to the target over-limit risk indicators. The target aircraft operational data is the aircraft operational data corresponding to any one of the multiple over-limit risk indicators. The target over-limit risk indicator is any one of the multiple over-limit risk indicators. Then, based on a preset flight safety threshold corresponding to the flight stage to which the sub-operational data belongs, and with each sub-operational data point, the sub-over-limit risk characteristics corresponding to the sub-operational data can be determined. Finally, the over-limit risk characteristics can be determined based on the sub-over-limit risk characteristics corresponding to each of the multiple sub-operational data points.

[0051] This application does not limit the number of risk indicators exceeding the limit; in practical applications, they can be set according to needs. For example, the number of risk indicators exceeding the limit can be 1, 2, 4, 5, or 10.

[0052] For example, the over-limit risk indicator may include the number of times airspeed deviation exceeds the limit during takeoff. The aircraft operational data 1 corresponding to the number of times airspeed deviation exceeds the limit during takeoff is the aircraft's operating airspeed. The sub-operational data 1 corresponding to this operating airspeed includes: at 14 seconds into the takeoff phase, the aircraft's actual airspeed is 85 knots (kt), and the target airspeed is 88 kt; at 15 seconds into the takeoff phase, the aircraft's actual airspeed is 78 kt, and the target airspeed is 90 kt; at 16 seconds into the takeoff phase, the aircraft's actual airspeed is 82 kt, and the target airspeed is 95 kt; at 17 seconds into the takeoff phase, the aircraft's actual airspeed is 92 kt, and the target airspeed is 100 kt. The airspeed difference between the actual and target airspeed at 14 seconds is 3 kt; at 15 seconds, the difference is 12 kt; at 16 seconds, the difference is 13 kt; and at 17 seconds, the difference is 8 kt. The airspeed difference threshold for the takeoff phase of sub-operation data 1 is 10 kt. Therefore, the sub-over-limit risk characteristic 1 corresponding to sub-operation data 1 is determined as follows: number of over-limit times is 2, maximum over-limit value is 13 kt, over-limit duration is 2 seconds, number of near-over-limit times is 1, and maximum near-over-limit value is 8 kt.

[0053] Therefore, extracting features from QAR data according to the flight phase is more consistent with the actual operation of the aircraft, making the extracted over-limit risk features more suitable for the aircraft and resulting in a more accurate risk assessment.

[0054] In one possible design, aircraft state control characteristics can reflect the pilot's finesse in maneuvering the aircraft. Even if the aircraft is not necessarily prone to failure, this characteristic can still be used to some extent to quantify the degree of potential failure risk caused by differences in pilot operation.

[0055] Optionally, the QAR data also includes: aircraft operational data corresponding to multiple aircraft state control indicators. The aircraft state control characteristics are determined as follows: For the aircraft operational data to be processed, multiple sub-operational data corresponding to the aircraft operational data to be processed can be determined based on multiple flight phases corresponding to the target aircraft state control indicator. The aircraft operational data to be processed is the aircraft operational data corresponding to any one of the multiple aircraft state control indicators. The target aircraft state control indicator is any one of the multiple aircraft state control indicators. Then, data processing can be performed on each sub-operational data to determine the sub-aircraft state control characteristics corresponding to the sub-operational data. Then, the aircraft state control characteristics can be determined based on the sub-aircraft state control characteristics corresponding to each of the multiple sub-operational data.

[0056] For example, aircraft state control characteristics include at least one of the following: control input frequency, deviation correction magnitude, deviation correction delay, and energy stability characteristics. Control input frequency is the number of effective operations performed by the pilot on the main control system per unit time. Deviation correction magnitude is the amount of control applied by the pilot when the aircraft deviates from the target state (e.g., glide path, heading, speed). Deviation correction delay is the time interval between the occurrence of the deviation and the pilot initiating correction. Energy stability characteristics reflect the pilot's ability to maintain energy balance during flight. Energy stability characteristics include at least one of the following: the standard deviation of the total energy time series, the root mean square of the energy change rate, or the number of energy overshoots per unit time.

[0057] For example, aircraft status control indicators may include the roll axis control input frequency during the approach phase. The aircraft operation data 2 corresponding to the roll axis control input frequency during the approach phase is the aircraft's roll axis control input value. The sub-operation data 2 corresponding to this roll axis control input value includes: a roll axis control input value of 0 at 0 seconds of the approach phase; a roll axis control input value of +0.12 at 8 seconds of the approach phase; a roll axis control input value of -0.08 at 15 seconds of the approach phase; a roll axis control input value of +0.06 at 22 seconds of the approach phase; a roll axis control input value of -0.10 at 31 seconds of the approach phase; a roll axis control input value of +0.07 at 40 seconds of the approach phase; and a roll axis control input value of -0.09 at 55 seconds of the approach phase. A roll axis control input value that is not 0 is considered a valid control event. If the number of valid control events within one minute is determined to be 6, then the corresponding aircraft state control feature 1 for sub-operation data 2 is determined to be: roll axis control input frequency of 6 times per minute, which is a relatively high frequency. This aircraft state control feature 1 will be marked as roll axis control instability during subsequent model processing.

[0058] Optionally, text extraction can be performed on the flight environment data based on the trained large language model to obtain environmental risk features. For example, environmental risk features may include at least one of the following: flight phase, aircraft weight, aircraft configuration, airport difficulty, or weather conditions.

[0059] Optionally, risk event characteristics can be determined as follows: Keywords can be extracted from aircraft status text data to obtain risk event keywords. Then, the risk event keywords can be compared with preset risk event keywords to obtain keyword matching results for the aircraft status text data. These matching results indicate whether the risk event keywords match the preset risk event keywords. Finally, risk event characteristics can be determined based on the matching results and the risk event keywords.

[0060] Alternatively, risk event features can be obtained by extracting aircraft status text data based on a trained large language model. The trained large language model can perform sentiment analysis, event extraction, and entity linking on pilot reports and maintenance records, extracting features implicit in the text such as pilot operational risks, pilot emotional states, and equipment hazards, and converting them into numerical vectors.

[0061] In this way, by comparing the risk event keywords with the preset risk event keywords, it can be determined whether the risk event keywords match the preset risk event keywords, thereby accurately identifying the risk events in the aircraft status text data and obtaining the risk event characteristics.

[0062] S103. Based on the characteristics of aviation risks, determine the results of the aviation risk assessment.

[0063] The aviation risk assessment results are used to indicate the degree of risk of an aircraft safety accident. The aviation risk assessment results can be scores or risk levels; this application does not impose any restrictions on this. In practical applications, the format of the aviation risk assessment results can be set according to requirements.

[0064] In one possible implementation, aviation risk characteristics can be input into a trained risk assessment model to obtain aviation risk assessment results. This application does not impose any restrictions on the risk assessment model. For example, the risk assessment model can be a Transformer model or a multilayer perceptron model; in practical applications, an appropriate model can be selected according to the requirements.

[0065] As one implementation method, the risk assessment model includes: an embedding layer, a position encoding module, a multi-head self-attention layer, a feedforward network layer, and an output layer.

[0066] Optionally, aviation risk features can be input into the embedding layer of the risk assessment model. The embedding layer performs a linear transformation on the aviation risk features to obtain a high-dimensional embedding vector. This high-dimensional embedding vector is then positionally encoded using the risk assessment model's positional encoding module, resulting in an encoded high-dimensional embedding vector. Next, the encoded high-dimensional embedding vector is processed by the risk assessment model's multi-head self-attention layer to obtain an attention weight matrix and a value matrix. These matrices are then weighted to obtain an enhanced feature vector. Subsequently, the enhanced feature vector is non-linearly correlated using the risk assessment model's feedforward network layer to obtain a global feature vector. Finally, the global feature vector is processed by the risk assessment model's output layer to obtain the aviation risk assessment result.

[0067] In this way, by processing aviation risk characteristics through the trained risk assessment model, the operational status of aircraft can be accurately analyzed, thereby obtaining accurate aviation risk assessment results.

[0068] Based on the above technical solution, by extracting features from aviation data, aviation risk characteristics can be obtained, and characteristic data affecting aircraft flight safety can be identified. Subsequently, aviation risk assessment results can be determined based on these aviation risk characteristics. Thus, accurate assessment of the risks of aircraft flight can be achieved through over-limit risk characteristics, aircraft condition control characteristics, and environmental risk characteristics.

[0069] In some embodiments, the trained risk assessment model can be obtained by generating a training set based on historical flight data and corresponding historical risk assessment results. The training set can then be input into the risk assessment model for supervised learning to obtain the trained risk assessment model.

[0070] Among them, the historical risk assessment results are labels for historical flight data.

[0071] The risk assessment model learns the contribution of different features to the overall risk in a specific context through a self-attention mechanism, and can ultimately output a risk assessment result that represents the overall risk level of aircraft flight.

[0072] In some embodiments, the attention weight matrix can be weighted to obtain multiple attention weights, which indicate the degree of influence of the flight data corresponding to the attention weight on the aviation risk assessment result. Then, an aggregation analysis can be performed based on the aviation risk features and the multiple attention weights to obtain multiple target risk features corresponding to the aviation risk assessment result and the risk contribution rate corresponding to each target risk feature. The target risk features are aviation risk features whose corresponding risk contribution rate is greater than or equal to a preset contribution rate threshold.

[0073] In this way, the attention mechanism of the risk assessment model can be used to obtain multiple target risk features corresponding to the aviation risk assessment results and the risk contribution rate corresponding to each target risk feature, making the aviation risk assessment results interpretable.

[0074] In some embodiments, after obtaining multiple target risk features corresponding to the aviation risk assessment results and the risk contribution rate corresponding to each target risk feature, the multiple target risk features can be output sequentially in descending order based on the risk contribution rate corresponding to each target risk feature.

[0075] By calculating the mean attention weights of the last layer encoder in the risk assessment model and mapping them back to the input features (i.e., aviation risk features), the contribution of each feature in the aviation risk features to the risk index can be determined.

[0076] In some embodiments, the aviation risk assessment result is a score. After obtaining the aviation risk assessment result, the multiple target risk features corresponding to the aviation risk assessment result, and the risk contribution rate corresponding to each target risk feature, the aviation risk assessment result can be compared with a first risk level threshold and a second risk level threshold to obtain the aviation risk assessment level. Then, based on the aviation risk assessment level and a preset risk color correspondence, a target risk color code can be obtained, where the preset risk color correspondence is the correspondence between the preset risk assessment level and the preset risk color code. Finally, based on aviation data, the aviation risk assessment result, the aviation risk assessment level, the target risk color code, the multiple target risk features corresponding to the aviation risk assessment result, and the risk contribution rate corresponding to each target risk feature, an aviation risk assessment report can be obtained and rendered for display.

[0077] The threshold for the first risk level is lower than that for the second risk level. The aviation risk assessment report includes a risk trend chart and a risk attribution chart. The risk trend chart indicates changes in risk during aircraft flight, while the risk attribution chart indicates the reasons for these changes in risk.

[0078] In some embodiments, if the aviation risk assessment result is less than or equal to the first risk level threshold, the aviation risk assessment level is low risk, and the target risk color code corresponding to low risk is determined to be green, with no warning message issued. If the aviation risk assessment result is greater than the first risk level threshold and less than or equal to the second risk level threshold, the aviation risk assessment level is medium risk, and the target risk color code corresponding to medium risk is determined to be yellow, with a yellow warning message issued. If the aviation risk assessment result is greater than the second risk level threshold, the aviation risk assessment level is high risk, and the target risk color code corresponding to high risk is determined to be red, with a red warning message issued.

[0079] By establishing a risk index threshold system, different levels of risk are color-coded and graded for early warning, and a visual interface is provided to display risk index trends, distribution, and specific attribution analysis reports.

[0080] In some embodiments, the preset flight safety threshold can be obtained by acquiring an initial safety threshold corresponding to the aircraft's operating parameters. Then, the initial safety threshold can be adjusted based on airport data and / or flight meteorological data to obtain the preset flight safety threshold.

[0081] The initial safety threshold can be determined through civil aviation regulations. Alternatively, the initial safety threshold can be derived experimentally.

[0082] Optionally, for each airport parameter in the airport data, an airport adjustment coefficient can be obtained based on the correspondence between the airport data and airport coefficients. The correspondence between the airport coefficients is the correspondence between preset airport parameters and preset airport adjustment coefficients. Then, the airport adjustment coefficient can be multiplied by the initial safety threshold to obtain the preset flight safety threshold.

[0083] Alternatively, for each meteorological parameter in the navigation meteorological data, a navigation meteorological adjustment coefficient can be obtained based on the correspondence between the meteorological parameter and the navigation meteorological coefficient. The correspondence between the navigation meteorological coefficient and the preset meteorological adjustment coefficient is the same as that between the preset meteorological parameter and the preset meteorological adjustment coefficient. Then, the navigation meteorological adjustment coefficient can be multiplied by the initial safety threshold to obtain the preset navigation safety threshold.

[0084] Alternatively, airport adjustment coefficients can be obtained based on the correspondence between airport data and airport coefficients. Flight weather adjustment coefficients can be obtained based on the correspondence between flight meteorological data and airport coefficients. Then, the airport adjustment coefficients and flight weather adjustment coefficients can be multiplied by the initial safety threshold to obtain the preset flight safety threshold.

[0085] Adjusting preset flight safety thresholds using airport data and / or flight weather data can make the thresholds used for feature extraction more accurate, thereby making the over-limit risk characteristics more accurate and the aircraft risk assessment results more accurate.

[0086] In this way, by adjusting the pre-set initial safety thresholds using airport data and / or aeronautical meteorological data, the risk assessment of the aircraft can be made more consistent with the actual operating conditions of the aircraft, thereby making the final risk assessment result of the aircraft more accurate.

[0087] In some embodiments, before performing S102, the QAR data can be decoded, cleaned, outlier-handled, and channel-aligned to obtain regularized time-series data.

[0088] Among them, well-organized time-series data refers to time series data with a unified structure, continuous time, and standardized format.

[0089] In some embodiments, the aircraft's aeronautical data may be aeronautical data from any one or more flights of the aircraft, thereby determining the aeronautical risk of one or more flights based on the aircraft's risk assessment results. Alternatively, it may be aeronautical data from any one or more pilots flying the aircraft, thereby determining the aeronautical risk of one or more pilots flying the aircraft based on the aircraft's risk assessment results.

[0090] This application does not limit the number of aircraft. For example, the number of aircraft can be 1, 2, 3, 4 or the entire fleet. In practical applications, the number of aircraft to be evaluated can be determined according to the needs.

[0091] When there are multiple aircraft, risk assessments can be performed on the aeronautical data of each aircraft to obtain a risk assessment result for that aircraft, thus yielding multiple risk assessment results. Alternatively, risk assessments can be performed on the aeronautical data of multiple aircraft simultaneously to obtain multiple risk assessment results for each aircraft. In this way, risk assessments for multiple aircraft can be achieved.

[0092] The following section provides a detailed description of this application using specific examples.

[0093] The system can acquire QAR data for flight XYZ, a brief report submitted by the pilot stating "minor clear-air turbulence encountered during approach, autopilot disengaged, manual landing," and crosswind weather data from the destination airport. Then, aviation risk features can be extracted based on the QAR data, brief report, and crosswind weather data. These aviation risk features can then be input into a trained risk assessment model to obtain the aviation risk assessment results. Furthermore, the system can obtain multiple target risk features corresponding to the aviation risk assessment results, and the risk contribution rate of each target risk feature.

[0094] The aviation risk characteristics include: landing vertical overload characteristics, approach lateral control characteristics, airport difficulty coefficient characteristics, meteorological condition characteristics, and risk event characteristics. Specifically, the landing vertical overload characteristic can be expressed as a landing vertical overload parameter of 1.45G, which does not exceed the landing vertical overload parameter threshold. A landing vertical overload parameter of 1.45G means that at the moment of landing touchdown, the force exerted on the aircraft fuselage in the vertical direction is 1.45 times its own weight. The corresponding landing vertical overload parameter threshold is 1.5G. Therefore, the impact force upon landing touchdown is slightly stronger than ideal, but does not exceed the limit and is considered a slight deviation within a controllable range.

[0095] The approach lateral control characteristics can be specifically represented as follows: during the approach phase, the frequency of pilot lateral control inputs is 20% higher than the average, indicating a slightly larger standard deviation of the locale deviation. The airport difficulty level characteristics can be specifically represented as a high airport difficulty level. The weather conditions characteristics can be specifically represented as a crosswind of 10 knots and turbulence in clear skies. The risk event characteristics can be specifically represented as turbulence in clear skies, autopilot disengagement, and pilot emotional state of neutral to slightly tense.

[0096] The model identifies features such as "clear-sky turbulence," "crosswinds," and "high-difficulty airports" and assigns them high weights. Meanwhile, "increased control frequency" and "slightly larger overload" are considered by the model as responses under these adverse conditions.

[0097] After comprehensive calculation by the model, the aviation risk assessment result for this flight is 65, corresponding to a medium risk level. The system then generates an attribution report: "The main contributing factors to the medium risk of this flight are: high-difficulty airport approach (30% contribution), encountering clear-air turbulence (25% contribution), crosswind conditions (20% contribution), and increased control load (15% contribution)." The system then alerted safety management personnel to the flight, but no emergency intervention was required. This report can be stored in the database as a valuable case study for future risk prediction and pilot training for flights under similar conditions.

[0098] In summary, this application, by acquiring a large amount of historical QAR data and associated text report data from airline fleets, and preprocessing and aligning multi-source heterogeneous QAR time-series data, can construct a multi-dimensional feature index system integrating standard exceedances, operational finesse, and operational context. Furthermore, by utilizing a pre-trained large language model to deeply analyze unstructured texts such as pilot reports and maintenance records, risk event features are extracted. After fusing numerical and textual features, the data is input into a large model based on the Transformer architecture. This model dynamically calculates the weights of each target risk feature through a self-attention mechanism, outputting a comprehensive and quantifiable risk index for a single flight or pilot, and enabling risk tracing and visual early warning. Thus, it achieves a leap from single-event monitoring of flight risks to comprehensive, interpretable, and quantifiable assessment.

[0099] This application overcomes the limitations of traditional over-limit event analysis by extracting deep features reflecting handling qualities and operational context from massive QAR data, achieving a comprehensive understanding of flight risks. Furthermore, it utilizes a large language model to parse unstructured text reports, transforming semantic information into quantitative features and fusing them with QAR data features to construct a more complete risk view. Moreover, it leverages the attention mechanism of the large model to dynamically evaluate the weights of different target risk features in different contexts, outputting a precise and comprehensive quantitative risk index. Additionally, it clearly identifies the main factors leading to an increased risk index, providing risk tracing capabilities and enhancing the interpretability and relevance of safety management. Finally, it offers multi-dimensional risk visualization and early warning capabilities, ranging from fleet size to pilots and from individual flights to long-term trends.

[0100] Furthermore, this application assesses aircraft risks from multiple dimensions, including out-of-limit events, operational processes, and operating environments. This overcomes the shortcomings of traditional methods that rely solely on out-of-limit events, resulting in deeper model insights and a more comprehensive risk assessment. Moreover, the quantitative risk assessment introduced in this application enables precise measurement and horizontal comparison of safety levels, making risk assessment more accurate. Furthermore, by analyzing subtle operational deviations and textual information, this application can identify potential risk tendencies earlier, achieving early warning, shifting the focus forward, and providing proactive insights. Additionally, this application directly targets the root causes of risks through automated risk attribution analysis, significantly improving the efficiency and accuracy of safety management personnel's investigations and decisions, providing a core tool for achieving "precise safety" management. This also enables intelligent decision-making.

[0101] The foregoing primarily describes the solutions provided in the embodiments of this application from a methodological perspective. It is understood that, in order to achieve the aforementioned functions, the risk assessment device includes corresponding hardware structures and / or software modules for executing each function. Those skilled in the art should readily recognize that, based on the risk assessment method steps described in conjunction with the embodiments disclosed in this application, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0102] This application also provides a risk assessment device. This risk assessment device can be a server, a CPU within the server, a module within the server used for risk assessment, or a client within the server used for risk assessment.

[0103] This application embodiment can divide the risk assessment device into functional modules or functional units according to the above method examples. For example, each function can be divided into its own functional modules or functional units, or two or more functions can be integrated into one processing module. The integrated modules can be implemented in hardware or in software functional modules or functional units. The module or unit division in this application embodiment is illustrative and represents only one logical functional division; other division methods may be used in actual implementation.

[0104] This application provides a risk assessment device. For example... Figure 2 As shown, the risk assessment device may include: an acquisition module 201 and a processing module 202.

[0105] Acquisition module 201 is used to acquire aeronautical data of the aircraft.

[0106] Processing module 202 is used to extract features from aviation data to obtain aviation risk features, including: exceedance risk features, aircraft state control features, and environmental risk features. Exceedance risk features reflect the degree of risk when aircraft operating parameters approach or exceed preset flight safety thresholds; aircraft state control features reflect the stability of pilot control of the aircraft; and environmental risk features reflect the degree of risk from the aircraft's external environment. Processing module 202 is also used to obtain aviation risk assessment results based on the aviation risk features and a trained risk assessment model. These results indicate the degree of risk of an aircraft accident.

[0107] Figure 3This is a schematic diagram illustrating the structure of another risk assessment apparatus according to an exemplary embodiment. The risk assessment apparatus may include a processor 302, which executes application code to implement the risk assessment method of this application.

[0108] Processor 302 may be a central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits used to control the execution of programs according to the present application.

[0109] like Figure 3 As shown, the risk assessment device may further include a memory 303. The memory 303 stores the application code that executes the scheme of this application, and its execution is controlled by the processor 302.

[0110] Memory 303 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital versatile optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. Memory 303 may exist independently and be connected to processor 302 via bus 304. Memory 303 may also be integrated with processor 302.

[0111] like Figure 3 As shown, the risk assessment device may further include a communication interface 301, wherein the communication interface 301, processor 302, and memory 303 may be coupled to each other, for example, through a bus 304. The communication interface 301 is used for information exchange with other devices, for example, supporting information exchange between the risk assessment device and other devices.

[0112] It should be pointed out that, Figure 3 The equipment structure shown does not constitute a limitation on the risk assessment device, except... Figure 3In addition to the components shown, the risk assessment device may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0113] In actual implementation, the functions implemented by processing module 202 can be provided by... Figure 3 The processor 302 shown calls the program code in memory 303 to implement this.

[0114] This application also provides a computer-readable storage medium storing instructions that, when executed by a processor of a computer device, enable the computer to perform the risk assessment method provided in the embodiments described above. For example, the computer-readable storage medium may be a memory 303 including instructions, which may be executed by a processor 302 of a computer device to complete the method. Optionally, the computer-readable storage medium may be a non-transitory computer-readable storage medium, such as a ROM, RAM, CD-ROM, magnetic tape, floppy disk, or optical data storage device.

[0115] Figure 4 A conceptual partial view of a computer program product provided in an embodiment of this application is shown schematically. The computer program product includes a computer program for executing computer processes on a computing device.

[0116] In one embodiment, the computer program product is provided using a signal bearer medium 400. The signal bearer medium 400 may include one or more program instructions that, when executed by one or more processors, can provide the above-mentioned... Figure 1 The described function or part of the function. Therefore, for example, refer to... Figure 1 In the embodiment shown, one or more features of S101-S103 can be provided by one or more instructions associated with the signal carrying medium 400. Furthermore, Figure 4 The program instructions in the document also describe example instructions.

[0117] In some examples, the signal carrying medium 400 may include a computer-readable medium 401, such as, but not limited to, a hard disk drive, a compact disc (CD), a digital video disc (DVD), a digital magnetic tape, a memory, a read-only memory (ROM), or a random access memory (RAM), etc.

[0118] In some implementations, the signal carrying medium 400 may include a computer recordable medium 402, such as, but not limited to, a memory, a read / write (R / W) CD, an R / W DVD, and so on.

[0119] In some implementations, the signal carrying medium 400 may include a communication medium 403, such as, but not limited to, digital and / or analog communication media (e.g., fiber optic cables, waveguides, wired communication links, wireless communication links, etc.).

[0120] The signal-bearing medium 400 can be transmitted by a wireless communication medium 403. One or more program instructions may be, for example, computer-executable instructions or logical implementation instructions.

[0121] In some examples, the risk assessment device may be configured to provide various operations, functions, or actions in response to one or more program instructions via a computer-readable medium 401, a computer-recordable medium 402, and / or a communication medium 403.

[0122] Through the above description of the embodiments, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0123] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another apparatus, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0124] The units described as separate components may or may not be physically separate. A component shown as a unit can be one or more physical units; that is, it can be located in one place or distributed in multiple different locations. Some or all of the constituent units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0125] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0126] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solution of the embodiments of this application, essentially, or the part that contributes to the prior art, or a complete or partial classification of the technical solution, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0127] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A risk assessment method, characterized in that, The method includes: Obtain aircraft aeronautical data; The aviation data is subjected to feature extraction to obtain aviation risk features, which include: over-limit risk features, aircraft state control features and environmental risk features. The over-limit risk features are used to reflect the degree to which the aircraft's operating parameters are close to or exceed the preset flight safety threshold. The aircraft state control features are used to reflect the stability of the pilot's control of the aircraft's flight. The environmental risk features are used to reflect the degree of risk of the aircraft's external environment. Based on the aviation risk characteristics, an aviation risk assessment result is determined, which is used to indicate the degree of risk of the aircraft experiencing a safety accident.

2. The method according to claim 1, characterized in that, The aviation data includes: aircraft status text data, which includes at least one of the following: pilot report text, aircraft maintenance record text, or weather report. The pilot report text is used to indicate the pilot's flying status during the flight, and the weather report is used to indicate the weather conditions during the flight. The aviation risk features also include: risk event features, which are used to indicate risk events in the aircraft status text data.

3. The method according to claim 2, characterized in that, The characteristics of the risk events are determined in the following way: Keyword extraction is performed on the aircraft status text data to obtain risk event keywords; The risk event keywords are compared with preset risk event keywords to obtain the keyword matching results of the aircraft status text data. The keyword matching results are used to indicate whether the risk event keywords match the preset risk event keywords. Based on the risk event matching results and the risk event keywords, the characteristics of the risk event are determined.

4. The method according to claim 1, characterized in that, The aviation data also includes: Quick Access Recorder (QAR) data and flight environment data, wherein the flight environment data is used to represent the external environmental conditions of the aircraft's location during flight; The process of extracting features from the aviation data to obtain aviation risk features includes: Feature extraction is performed on the QAR data to obtain the over-limit risk features and the aircraft state control features; Feature extraction is performed on the navigation environment data to obtain the environmental risk features.

5. The method according to claim 4, characterized in that, The QAR data includes: aircraft operation data corresponding to multiple over-limit risk indicators; the aircraft operation data is used to represent the changes in aircraft operation parameters during the flight of the aircraft. The step of extracting features from the QAR data to obtain the over-limit risk features includes: For the target aircraft operation data, based on multiple flight stages corresponding to the target over-limit risk indicators, multiple sub-operation data corresponding to the target aircraft operation data are determined; the target aircraft operation data is the aircraft operation data corresponding to any one of the multiple over-limit risk indicators; the target over-limit risk indicator is any one of the multiple over-limit risk indicators. Based on the preset navigation safety threshold corresponding to the navigation phase to which the sub-operation data belongs, and with each sub-operation data, the sub-excess risk characteristics corresponding to the sub-operation data are determined; The over-limit risk feature is determined based on the sub-over-limit risk feature corresponding to each of the multiple sub-running data.

6. The method according to claim 1, characterized in that, The preset navigation safety threshold is obtained in the following way: Obtain the initial safety threshold corresponding to the aircraft operating parameters; The initial safety threshold is adjusted based on airport data and / or flight weather data to obtain a preset flight safety threshold. The airport data is used to reflect the aircraft's departure airport or the difficulty of the aircraft's flight, and the flight weather data is used to reflect the weather conditions of the aircraft during flight.

7. The method according to any one of claims 1-6, characterized in that, The determination of aviation risk assessment results based on the aviation risk characteristics includes: The aviation risk features are input into the embedding layer of the risk assessment model. The aviation risk features are linearly transformed by the embedding layer to obtain a high-dimensional embedding vector. The high-dimensional embedding vector is then position-encoded by the position encoding module of the risk assessment model to obtain an encoded high-dimensional embedding vector. The encoded high-dimensional embedding vector is processed by the multi-head self-attention layer of the risk assessment model to obtain an attention weight matrix and a value matrix. The attention weight matrix and the value matrix are then weighted to obtain an enhanced feature vector. The enhanced feature vector is nonlinearly correlated through the feedforward network layer of the risk assessment model to obtain the global feature vector. The global feature vector is processed through the output layer of the risk assessment model to obtain the aviation risk assessment result.

8. The method according to claim 7, characterized in that, The method further includes: The attention weight matrix is ​​weighted to obtain multiple attention weights, which are used to indicate the degree of influence of the flight data corresponding to the attention weight on the aviation risk assessment result. Based on the aviation risk features and the multiple attention weights, an aggregation analysis is performed to obtain multiple target risk features corresponding to the aviation risk assessment results and the risk contribution rate corresponding to each of the multiple target risk features. The target risk features are aviation risk features whose corresponding risk contribution rate is greater than or equal to a preset contribution rate threshold.

9. A computer-readable storage medium storing instructions, characterized in that, When the computer executes the instruction, the computer performs the method as described in any one of claims 1-8.

10. A computer program product containing instructions, characterized in that, When the instructions are executed by the computing device, the computing device performs the method as described in any one of claims 1-8.

Images