An air communication information display method, a storage medium and an electronic device

Abstract

The present application relates to the technical field of aviation communication auxiliary, particularly relates to an aviation communication information display method, a storage medium and an electronic device.The technical scheme is as follows: when the pilot switches to a new radio frequency, the current radio frequency is obtained in real time, and the communication between the controller and other flights occurs; for each communication, the spatial distance risk, the call sign similarity risk, the pressure microphone risk and the content matching degree are obtained, and the correlation score is obtained through the weighting formula; the communication with the standard score is fixed on the top of the display interface in a highlighted style, so that it is not covered by new information. The present application solves the problem that the pilot cannot quickly identify key information from a large number of real-time communications after switching frequencies through the intelligent screening mechanism of multi-dimensional risk fusion, and the fixed display design ensures that important information is continuously visible, significantly improving the flight situation awareness and safety.

Description

Technical Field

[0001] This invention relates to the field of aviation communication auxiliary technology, and in particular to an aviation communication information display method, storage medium, and electronic device. Background Technology

[0002] Aviation radio communication is the core method for pilots and air traffic controllers to exchange instructions and information, and its accuracy and real-time performance are directly related to flight safety. In actual flight missions, especially in international flights or non-English speaking environments, pilots often face the challenge of insufficient English listening skills. Foreign controllers may have regional accents, speak quickly, or have non-standard pronunciation, further increasing the difficulty for pilots to understand the content of the conversation. To assist pilots in obtaining critical information, aviation communication assistance systems have emerged, which convert controllers' speech into text and present it in a visual interface.

[0003] In existing technologies, such auxiliary systems typically display all calls within the same radio frequency sequentially on the screen. As the number of calls increases, earlier information is pushed to the top, retaining only the most recent few messages. However, in aviation communication scenarios, the same frequency often includes conversations between controllers and multiple aircraft, much of which is less relevant to the pilot's own aircraft. When a pilot switches to a new frequency, a large number of real-time conversations between controllers and other flights occur, potentially containing crucial information relevant to the pilot's own aircraft. This could include emergencies reported by other aircraft, such as engine fires or landing gear malfunctions, or information closely related to the pilot's safety due to proximity, similar call signs, or microphone interference. The scrolling display method in existing technologies causes this critical information to be quickly overwhelmed, making it difficult for pilots to identify the truly important information in a state of information overload, thus reducing their contextual awareness.

[0004] Therefore, how to intelligently filter and highlight key information related to the pilot from the large number of calls generated in real time after the pilot switches frequencies has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] To address one of the aforementioned technical problems, the present invention adopts the following technical solution:

[0006] According to one aspect of the present invention, an aviation communication information display method is provided, characterized by comprising the following steps:

[0007] After the pilot switches to the new radio frequency, multiple communications occurring on the current radio frequency are acquired in real time; these communications are between the controller and other flights.

[0008] For each call, obtain the relevance score between the call and the local machine; the relevance score between the i-th call and the local machine satisfies the following relationship:

[0009] ;

[0010] in, For spatial distance risk, its value is negatively correlated with the real-time distance between the aircraft and other flights obtained based on the ADS-B receiver;

[0011] For call sign similarity risk, its value is positively correlated with the string similarity between the flight call sign mentioned in the call and the call sign of the user;

[0012] To mitigate the risk of microphone suppression, voice activity detection and voiceprint segmentation are used to determine if microphone suppression occurs due to the superposition of voices from different flights. If so, ,otherwise ;

[0013] Content matching accuracy is determined based on whether the caller ID is mentioned in the call, whether it contains an emergency indicator, and whether it matches a preset security event template.

[0014] α, β, μ, δ are preset weight coefficients and satisfy α + β + μ + δ = 1;

[0015] The correlation score is compared with a preset threshold.

[0016] When the correlation score is greater than or equal to the preset threshold, the corresponding call is highlighted and fixed in the top area of ​​the display interface so that it is not overwritten by subsequent new call information.

[0017] According to a second aspect of the present invention, a non-transitory computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the above-described method for displaying aviation communication information.

[0018] According to a third aspect of the present invention, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described method for displaying aviation communication information.

[0019] This invention has at least one of the following beneficial effects:

[0020] First, this invention acquires real-time conversations occurring after frequency switching and calculates a relevance score for each conversation to the pilot based on spatial distance risk, call sign similarity risk, microphone suppression risk, and content matching degree, achieving intelligent filtering and multi-dimensional correlation utilization of real-time conversation information. In aviation communications, pilots switching radio frequencies signifies entering a new control area. Other aircraft on the same frequency are located nearby, and events such as in-flight emergencies and equipment malfunctions reflected in their conversations are spatially highly correlated with the current flight and may pose a potential impact on the pilot. This invention addresses this by using spatial distance risk to acquire the real-time distance between the pilot and other flights via an ADS-B receiver, assigning higher weight to conversations from nearby flights; by using call sign similarity risk to identify situations where the flight call sign mentioned in the conversation is highly similar to the pilot's own call sign, preventing the risk of mishearing instructions due to similar call signs; by using microphone suppression risk to detect the overlapping of instructions caused by the superposition of voices from different flights, avoiding misunderstandings caused by communication interference; and by using content matching degree to comprehensively evaluate whether the pilot's own call sign is mentioned in the conversation, whether it contains emergency signs, and whether it matches a safety event template, achieving comprehensive coverage of the concept of "relevance to the pilot." Relying solely on call signs can only identify direct commands and cannot capture indirect connections such as other aircraft mentioning abnormalities or reporting nearby dangers; relying solely on keywords may lead to misjudgments due to interference from everyday language. This invention, through weighted fusion of these four factors, enables the system to accurately locate truly noteworthy content from real-time conversations, solving the problem of pilots being unable to obtain crucial information in a timely manner due to information overload after switching frequencies.

[0021] Secondly, this invention compares relevance scores with preset thresholds and highlights calls exceeding the threshold at the top of the display interface, preventing them from being overwritten by subsequent new calls. This fundamentally changes the shortcoming of traditional scrolling display modes where crucial information is fleeting. In existing technologies, even if the system filters out important information, it is quickly pushed off the screen as new calls flood in, requiring pilots to manually search for it again, which can easily lead to information loss under high-pressure flight missions. This invention, through a "fixed display" mechanism, ensures that important calls highly relevant to the aircraft remain prominently displayed for an extended period, maintaining visibility regardless of subsequent new calls. This feature allows pilots to grasp real-time background information related to aircraft safety the moment they enter a new frequency and refer to it throughout the flight, significantly improving situational awareness and flight safety. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 A flowchart illustrating an aviation communication information display method provided in an embodiment of the present invention.

[0024] Figure 2 A flowchart of a location-based aviation voice recognition method provided for another embodiment of the present invention.

[0025] Figure 3 This is a schematic diagram of an aviation communication information display interface provided in an embodiment of the present invention. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] As one possible embodiment of the present invention, such as Figure 1 As shown, an aviation communication information display method is provided, which includes the following steps. This method aims to solve the problems that pilots cannot quickly identify key information related to their own aircraft (such as emergencies reported by other aircraft) from a large number of real-time calls when switching to a new radio frequency, and the problem that existing technologies cannot dynamically enable deep semantic verification based on the characteristics of the controller's accent. Each step is explained in detail below with specific flight scenarios.

[0028] In aviation communications, when a pilot switches radio frequencies, it signifies entering a new controlled area. Other aircraft on the same frequency are located nearby, and the in-flight emergencies, equipment malfunctions, and other events reflected in their communications are highly relevant to the current flight in both time and space, potentially impacting the aircraft itself. Therefore, how to extract key information relevant to the aircraft from real-time communications is the primary technical problem this method must address.

[0029] A100: After a pilot switches to a new radio frequency, it acquires real-time information on multiple communications occurring on that frequency. These communications include those between air traffic controllers and other flights.

[0030] Once a pilot enters a new frequency, the system continuously monitors conversations between air traffic controllers and other flights on that frequency, capturing every call in real time. Because aircraft on the same frequency are all located near the current control area, the events reflected in these calls are spatially highly relevant to the pilot's location and may have a potential impact. By acquiring these messages in real time, the system can process each new call immediately, avoiding processing delays caused by information backlog.

[0031] During flight, pilots need to constantly switch radio frequencies to enter new control areas as the flight progresses. For example, when an aircraft flies from the "Beijing Approach" control area to the "Shanghai Area Control" area, the pilot needs to manually switch the frequency from 118.35MHz to 128.75MHz. The moment the switch is completed, the system automatically begins real-time capture of all communications on that frequency. Each call record includes complete text content (converted through speech recognition), speaker identification, timestamps, and other information.

[0032] A200: For each call, obtain a relevance score between the call and the device. The relevance score is calculated based on spatial distance risk, call sign similarity risk, microphone suppression risk, and content matching degree.

[0033] Specifically, the relevance score between the i-th call and the local machine satisfies the following relationship:

[0034] .

[0035] in, For spatial distance risk, its value is negatively correlated with the real-time distance between the aircraft and other flights obtained based on the ADS-B receiver.

[0036] The call sign similarity risk is positively correlated with the string similarity between the flight call sign mentioned in the call and the call sign of the user.

[0037] To mitigate the risk of microphone suppression, voice activity detection and voiceprint segmentation are used to determine if microphone suppression occurs due to the superposition of voices from different flights. If so, ,otherwise .

[0038] Content matching accuracy is determined based on whether the caller ID is mentioned during the call, whether it contains an emergency indicator, and whether it matches a preset security event template.

[0039] α, β, μ, and δ are preset weighting coefficients that satisfy α + β + μ + δ = 1; for example, they can be set to α = 0.4, β = 0.3, μ = 0.2, and δ = 0.1. This weighted summation design fully considers the characteristics of aviation communications: spatial distance risk directly reflects the proximity of other flights to the aircraft, representing the most direct physical threat; call sign similarity risk captures potential mishearing due to similar call signs; microphone suppression risk identifies command confusion caused by communication interference; and content matching comprehensively assesses the relevance of the call content itself to the aircraft (whether the aircraft is mentioned, whether it is urgent, or whether it involves a security incident). Through the integration of these four factors, a comprehensive coverage of the concept of "relevance to the aircraft" is achieved.

[0040] The technical challenge this step addresses is how to quantify the correlation between other flight calls and the aircraft's own calls from multiple risk dimensions. Relying solely on content matching (such as whether the aircraft's call sign is mentioned) can only identify directly related information. However, due to factors like proximity, similar call signs, or microphone suppression, other flights' calls may also pose a threat to the aircraft's security. Therefore, it is necessary to introduce spatial distance, call sign similarity, and microphone suppression detection to construct a multi-dimensional risk quantification system.

[0041] The following explains how each dimension is obtained.

[0042] A201a: Risk of Acquiring Spatial Distance .

[0043] The real-time distance d (in nautical miles) between this aircraft and other flights is obtained using an ADS-B receiver. Spatial distance risk. The value of is negatively correlated with distance, meaning the closer the distance, the higher the risk. This can be specifically calculated using a piecewise function.

[0044] If d≤d close (e.g. d) close (5 nautical miles), then ;

[0045] If d close <d<d far (e.g. d) far (20 nautical miles), then ;

[0046] If d≥d far ,but .

[0047] This mechanism assigns a higher risk weight to calls from flights that are closer to the aircraft, thus prioritizing flights that may approach or collide in the air.

[0048] A201b: Risks associated with obtaining call sign similarity .

[0049] Extract flight call signs mentioned in the call (e.g., "Cessna 123") and calculate string similarity with the machine's preset call signs, for example, using Levenshtein distance. Call sign similarity risk. The value of this value is positively correlated with similarity, meaning the more similar the call signs, the higher the risk. The calculation formula is:

[0050] ;

[0051] denominator The meaning is: take the maximum value between the call sign length of other flights and the call sign length of this flight.

[0052] : The length (number of characters) of the call sign of other flights mentioned in the call.

[0053] : The default string length of the call sign for this machine.

[0054] If no call sign is mentioned during the call, then This mechanism is used to prevent the risk of mishearing instructions due to similar call signs. For example, if the call sign of this flight is "Cessna 123" and the call sign of other flights is "Cessna 1231", the similarity is high and should be noted.

[0055] A201c: Risk of wheat stalking .

[0056] Voice activity detection (VAD) and voiceprint segmentation techniques are used to determine if microphone suppression occurs during a call. Microphone suppression refers to a situation where background noise splits the same instruction audio into two segments, with each segment originating from a different pilot (i.e., two different speakers). If microphone suppression is detected, then... ,otherwise This mechanism is used to identify command aliasing caused by communication interference, preventing pilots from making mistakes due to receiving incorrect information.

[0057] A201d: Get content matching score .

[0058] Content matching is used to comprehensively evaluate the relevance of the call content to the caller, including three sub-dimensions: call sign matching, semantic relevance, and urgency. Specifically:

[0059] Call sign matching degree : Identify the source callsign and destination callsign in each call. If either the source or destination callsign exactly matches the machine's preset callsign, the callsign matching score is... If the aircraft's call sign is mentioned during the call but not directly called (e.g., mentioned in other aircraft reports), then ;otherwise .

[0060] Callsign matching score characterizes the degree to which the local machine is mentioned during a call. The system pre-stores the local machine's callsign, such as "Cessna 123". For each call, the source callsign (who is speaking) and the destination callsign (who is being addressed) are identified. If either the source or destination callsign exactly matches the local machine's callsign, for example, if the call content is "Cessna 123, contact ShanghaiControl", then the callsign matching score is high. If the caller ID is mentioned during the call but not directly called, for example, if another aircraft report states "Cessna 123 is behind me," then... If the call sign is not heard during the call, then .

[0061] Obtain the semantic relevance of this call to the local machine. ,include:

[0062] A211d: Identify whether the call contains keywords from a preset aviation safety event keyword library and obtain the weight of each keyword.

[0063] The pre-built aviation safety incident keyword database contains keywords related to flight safety, such as "engine fire," "landing gear," "emergency," "traffic," and "wind shear." Each keyword corresponds to a pre-defined weight α. k The weight values ​​are preset based on the urgency of the event represented by the word and its potential impact on the machine. For example, the weight of "engine fire" is set to 0.9, and that of "landing gear" is set to 0.6.

[0064] A222d: Identify whether the call matches a preset aviation safety event template.

[0065] Pre-set aviation safety event templates are used to identify specific safety scenarios, such as "other aircraft reporting nearby malfunctions," "an aircraft encountering wind shear," and "runway incursion warnings." The reason for using event templates is that keywords alone may not capture the complete semantic context. For example, "smoke from an aircraft ahead" may not contain specific keywords but is still an emergency. Template matching can more accurately identify potentially relevant external events. Template matching can employ natural language processing techniques, such as dependency parsing or semantic similarity calculation. When the semantic similarity between the call content and the template exceeds a threshold, a match is determined, and an indicator value is displayed. Otherwise, it is 0.

[0066] A213d: Based on the keyword weight and whether it matches the event template, obtain the semantic relevance. The semantic relevance satisfies the following relationship:

[0067] ;

[0068] in, For the i-th call, The weight of keyword k This is the sum of all weights in the keyword database, used to normalize keyword scores. The indicator value for whether the event template matches. A preset template weight coefficient (e.g., 0.3) is used to balance the contributions of keyword matching and template matching. This formula considers both the occurrence of specific keywords and the recognition of the overall context, making the assessment of semantic relevance more comprehensive and accurate.

[0069] urgency : Identify whether each call contains a preset emergency indicator, which includes "Mayday" or "Pan-Pan"; if so, the level of urgency is determined. ;otherwise Urgency level, as an independent dimension, can highlight high-risk information that requires immediate attention.

[0070] Content matching We obtain the result by weighted summation of the three sub-dimensions mentioned above:

[0071] ;

[0072] , , Preset weighting coefficients and satisfying .

[0073] For example, it can be set to This weighted summation design fully considers the characteristics of aviation communications: call sign matching directly reflects the explicit association between the call and the aircraft, serving as the most direct evidence; semantic relevance captures implicit associations and potential threats, providing an important supplement to explicit associations; and urgency assigns higher weight to high-risk information, ensuring that emergencies are prioritized for identification. By integrating these three factors, a comprehensive coverage of the concept of "association with the aircraft" is achieved.

[0074] A300: Compares the relevance score with a preset threshold.

[0075] Set a preset threshold T high(For example, 0.6) is used to determine the importance of the call. This threshold can be adjusted according to the actual application scenario and pilot preferences.

[0076] A400: When the relevance score is greater than or equal to the preset threshold, the corresponding call will be highlighted and fixed in the top area of ​​the display interface to prevent it from being replaced or covered by subsequent new call information.

[0077] This step addresses the problem of critical information being easily buried in the traditional scrolling display mode. Even if the system filters out highly relevant information, if it is displayed only in chronological order, this information will quickly be pushed off the screen as new calls keep coming in, making it impossible for pilots to maintain focus. The "fixed display" mechanism ensures that highly relevant information always resides in the most prominent position on the interface, allowing pilots to access it at any time without manual searching, greatly improving the continuity and reliability of context awareness.

[0078] For calls with scores exceeding the threshold, the system classifies them as highly relevant information and highlights them on the display interface (e.g., with a green background or a star icon). Figure 3 The dialogue content is displayed in green and pinned to the top area. This "pinned display" mechanism ensures that this critical information remains within the pilot's field of vision and is not overwritten by subsequent new calls. For example, if another aircraft reports "Mayday, engine fire" and that flight is close to the pilot, its relevance score is high, and it will be pinned to the top of the screen, allowing newly arrived pilots to immediately know that there is an emergency in that area.

[0079] A500: When the relevance score is less than the preset threshold, the corresponding call is displayed in a normal or weakened style.

[0080] For information with low relevance, the system uses a regular style (e.g., black font) or a weaker style (e.g., gray font). Figure 3 The conversation content (displayed in medium gray) is displayed normally in chronological order and scrolls upwards as new calls arrive. A filtering mode, "Show only highly relevant information," can be set when necessary to further streamline the interface.

[0081] A600: When switching to a new radio frequency, acquire a preset number of communication pairs corresponding to the new radio frequency. Each communication pair contains controller instructions and the corresponding pilot repeats them.

[0082] The technical challenge addressed in this step and subsequent steps is how to assess the accent difficulty of the current air traffic controller in order to dynamically enable deep semantic verification. Since air traffic controller accents and expression habits are relatively stable, even if there are no preceding flights close to the current flight's arrival time, the real-time acquired call pairs still provide a sufficient statistical sample. Therefore, to ensure statistical validity, the most recent N call pairs (e.g., 20) from the real-time dialogue are extracted first; if the number is insufficient, the actual number is used for statistical analysis, thereby providing an objective evaluation of the controller's comprehensibility.

[0083] In addition to the relevance assessment of individual calls described above, this embodiment also analyzes the overall speech comprehensibility of the frequency controller from a more macroscopic perspective. While acquiring the call, the system extracts multiple complete "call pairs," each consisting of a controller instruction and its corresponding pilot repetition. For example, a typical call pair is: controller instruction "Cessna 123, descend to three thousand," pilot repetition "Descend to three thousand, Cessna 123."

[0084] A610: Based on multiple conversation pairs, obtain the controller's comprehension difficulty level.

[0085] The comprehension difficulty coefficient D is used to quantify the degree of difficulty that the current controller's accent, speech rate, and other pronunciation characteristics pose to the pilot's understanding. The comprehension difficulty coefficient satisfies the following relationship:

[0086] .

[0087] in, To understand the difficulty level, The average delay time, calculated from statistics across multiple communication pairs, characterizes the average response speed of pilots to instructions. The time interval between the issuance of an instruction and the pilot's first repetition in each communication pair is statistically analyzed, and its average value is calculated. A longer delay time may mean that the pilot needs more time to understand the instruction, reflecting insufficient clarity of the instruction. This is the preset maximum delay time, used to normalize the delay time; for example, it can be set to 5 seconds.

[0088] The average request repetition rate, calculated from statistics across multiple communication pairs, characterizes the average frequency at which instructions are requested to be repeated. It involves calculating the average of the indication values ​​(0 or 1) for whether pilots issued repeat requests (such as "say again") in each communication pair. A high repetition rate directly indicates that the controller's instructions are difficult to understand.

[0089] The average recognition confidence score, calculated from multiple call pairs, characterizes the average level of confidence the speech recognition system has in responding to commands. It is calculated by statistically analyzing the confidence score (0 to 1) provided by the speech recognition engine for each command. Low confidence means the speech recognition system also struggles to accurately capture the command content, indirectly reflecting non-standard pronunciation by the controller.

[0090] The average consistency error rate, calculated from multiple communication pairs, characterizes the average frequency of discrepancies between pilot repetition and instructions. It is calculated based on the indicator value (0 or 1) indicating whether the instruction and repetition match in each communication pair. A high error rate indicates that pilots frequently mishear or misrepeat instructions, providing direct evidence of comprehension difficulties.

[0091] Preset weighting coefficients and satisfying For example, it can be set to 0.3, 0.3, 0.2, 0.2.

[0092] Through the above formula, the system integrates multiple dimensions of behavioral characteristics into a quantitative comprehension difficulty coefficient, which objectively reflects the comprehensibility level of the controller.

[0093] The delay time is the time interval between the issuance of the instruction in each call pair and the pilot's first repetition. The request repetition rate is calculated based on the indication value of whether the pilot issued a repetition request in each call pair. The recognition confidence is calculated by the confidence value provided by the speech recognition engine for each instruction. The consistency error rate is calculated based on the indication value of whether the comparison result between the instruction and the repetition in each call pair is consistent.

[0094] A620: When the difficulty level exceeds the preset difficulty threshold, the deep semantic verification function is activated.

[0095] Set a preset difficulty threshold T D (e.g., 0.5). When D≥T D If the system determines that the current controller's accent is too strong or the instructions are unclear, there is a high risk of misunderstanding in subsequent communications with the pilot. In this case, the system automatically activates deep semantic verification to perform more rigorous analysis and verification of subsequent communications between the controller and the pilot, ensuring that the pilot correctly understands the instructions.

[0096] Deep semantic validation features include:

[0097] A621: Perform semantic analysis on subsequent communications between the controller and the local machine to obtain the operation type, operation object, and operation value, forming an operation triplet.

[0098] The deep semantic verification function first performs structured parsing on each sentence associated with the local machine. Using natural language processing technology, it extracts core operational elements from the text, forming operation triples (operation type, operation object, operation value). For example, for the instruction "descend to three thousand", the parsed operation triple is {DESCEND,altitude, 3000}; for the instruction "turn left heading zero nine zero", the parsed result is {TURN,heading, 090}. This structured representation converts natural language into a machine-understandable form, laying the foundation for subsequent automated verification.

[0099] A622: Compare the operation triplet with the preset operation parameter range corresponding to the current flight phase. If it exceeds the range, a reasonableness alarm will be triggered.

[0100] Based on the aircraft's real-time position information (GPS coordinates, altitude) and flight plan, the current flight phase is determined, such as takeoff, climb, cruise, approach, and landing. Each flight phase has pre-configured preset operating parameter ranges; for example, the altitude during the approach phase should be below 5000 feet, and the speed should be between 150 and 250 knots. The parameter values ​​in the parsed operating triplets are compared with the reasonable range for the corresponding phase. If they exceed the range (e.g., receiving "descend to 10,000 feet" during the approach phase), a reasonableness alarm is triggered, displaying a red warning bar "Abnormal Command Altitude" at the top of the interface, and highlighting the abnormal parameters to remind the pilot to verify.

[0101] A623: Identifies consecutive control commands and pilot repetitions, obtains the corresponding operation triplets and performs structured comparison. If they are inconsistent, a consistency alarm is triggered.

[0102] The current flight phase is determined based on the aircraft's real-time position information, and the preset operating parameter range is associated with the current flight phase.

[0103] Continuously monitor and identify consecutive "control instructions - pilot repetition" pairs. Extract the operation triplet T of each control instruction. c The pilot repeats the operation triple T p Perform a structured comparison. If T c With T pIf everything matches perfectly (operation type, object, and value all match), the operation passes; if there are discrepancies (e.g., the instruction is {DESCEND, altitude, 3000}, but the repetition is {DESCEND, altitude, 5000}), a consistency alarm is immediately triggered, highlighting the incorrect parameter in red on the interface, accompanied by a voice prompt: "Repetition inconsistent, please check!" This mechanism can automatically detect when pilots mishear or misrepeat, correcting the error before it becomes an operation.

[0104] A700: Receives voice messages on the current radio frequency in real time and performs voice recognition to obtain the real-time call text.

[0105] This step is performed in parallel with call processing. The system continuously monitors the current frequency's voice, performs real-time speech recognition, and converts every word spoken by the controller and pilot into text, forming a real-time call text stream.

[0106] A710: Performs semantic analysis on real-time call text to obtain its real-time relevance score.

[0107] For each real-time call text, the same method as A200 to A300 is used, namely, calculating its real-time relevance score based on spatial distance risk, call sign similarity risk, microphone suppression risk, and content matching degree. This allows the system to dynamically determine whether each newly generated call is associated with the local machine.

[0108] A720: Determine the display style of real-time call text based on the comparison results between the real-time relevance score and the preset threshold.

[0109] Compare the real-time relevance score with the preset threshold T high The comparison determines how the live call will be displayed.

[0110] A730: Dynamically displays real-time call text on the display interface using a preset display style.

[0111] Calls with real-time relevance scores exceeding the threshold (such as instructions from air traffic controllers to the pilot or emergency reports from other nearby flights) are dynamically displayed in a highlighted style; calls with low relevance are displayed in a regular scrolling style. This allows pilots to clearly distinguish between critical and secondary information throughout the flight, maintaining a high level of situational awareness.

[0112] Through the above steps, the embodiments described above achieve intelligent filtering, dynamic display, and deep semantic verification of aviation communication information, significantly improving the efficiency of pilots in obtaining key information and the accuracy of command comprehension. However, the effectiveness of the above functions is highly dependent on the accuracy of speech recognition; if there is a deviation in the underlying recognition, all subsequent analysis and verification will be meaningless.

[0113] To this end, the present invention also provides a location-based aviation speech recognition method. This method incorporates the real-time location information of the aircraft to perform context-aware constraint correction on the speech recognition results, which can significantly improve the recognition accuracy of professional terms such as waypoints and runway numbers. In practical applications, the constraint correction mechanism of this method can be used as a preprocessing step. After obtaining the high-precision call text in the real-time speech recognition stage (i.e., the aforementioned step A700), it is then sent to the semantic analysis and display module of this embodiment, thereby forming a complete technical closed loop from the bottom-level recognition to the upper-level application. The specific implementation of this speech recognition method is described below.

[0114] like Figure 2 As shown, the method includes the following steps. This method aims to address the safety hazard of pilots misunderstanding instructions due to low accuracy in recognizing technical terms (such as waypoints and runway numbers) caused by a lack of location context information in aviation radio communications. Each step is explained in detail below with reference to a specific flight scenario.

[0115] S100: Based on the aircraft's real-time location information, determine the current preset geographical area and associate it with the corresponding preset location dictionary.

[0116] Specifically, S100 includes:

[0117] S101: Based on the real-time location information, combined with the built-in aeronautical chart or flight plan, determine the current control sector, airport terminal area or airway section where the aircraft is located.

[0118] S102: Select the dictionary corresponding to the determined region from multiple pre-configured regional location dictionaries as the preset location dictionary.

[0119] In air traffic control communications, instructions issued by air traffic controllers are often closely related to the aircraft's current geographical location. For example, during the approach phase, the waypoints mentioned in the instructions are invariably waypoints near that airport. However, general speech recognition models lack a concept of location, treating terms from all regions equally. This leads to a situation where, even when an aircraft is in the Beijing area, the model might still misidentify "BIDIB" as a similarly pronounced waypoint in the Shanghai area. This step aims to address this issue by introducing real-time location information, dynamically narrowing the recognition scope to the terminology set of the current region.

[0120] In practice, real-time location information, including latitude, longitude, and altitude, can be obtained from the onboard GPS. Combined with a built-in digital aeronautical chart or flight plan database, the system can determine the aircraft's current location within a pre-defined geographical area. This area can be classified according to airspace control, such as "Beijing Approach," "Shanghai Area Control Sector 01," or "Guangzhou Terminal Area." The system pre-configures a pre-defined location thesaurus for each such geographical area, containing all possible professional location keywords within that area, including waypoint names, navigation beacon names, runway numbers, and airport codes. For example, the "Beijing Approach" thesaurus includes waypoints such as "BIDIB," "ELKUR," and "VYK," and runway numbers such as "18L" and "36R." Once the current area is determined, the system automatically associates it with the corresponding thesaurus, providing contextual constraints for subsequent identification. This "area-aware" pre-configuration allows subsequent steps to be matched only within the local thesaurus, significantly narrowing the candidate range and reducing the possibility of terminology confusion at the source.

[0121] S200: Receives the voice at the current radio frequency and performs preliminary voice recognition to obtain a preliminary text sequence.

[0122] This step is the foundational processing for speech recognition. During flight, the aircraft listens to the air traffic control frequencies of the current area via radio equipment, capturing continuous speech signals. These signals are then fed into a general-purpose speech recognition engine (e.g., an end-to-end model based on deep neural networks) for preliminary recognition, outputting the corresponding text sequence. For example, when the controller issues the instruction "Cessna 123, proceed direct to BIKMA," the preliminary recognition might yield "Cessna one two threeproceed direct to BIKMA."

[0123] S300: Identify a preset position guide word in the preliminary text sequence, and locate candidate position words based on the preset position guide word.

[0124] In aviation communications, certain words often foreshadow location-related information that will follow, such as "to," "direct," "at," "runway," and "via." These are called location guide words. However, these words also frequently appear in everyday English. Indiscriminately correcting all words following these guide words can lead to misinterpretations of non-critical information (for example, "restroom" in a casual conversation might be mistakenly interpreted as a waypoint). The purpose of this step is to accurately identify key points that may contain specialized location terminology, preparing for the subsequent selection of candidate location words.

[0125] Specifically, the system pre-establishes a preset location-guided word library, which not only contains the guide words themselves but also records the positional relationships corresponding to each guide word (for example, the word immediately following "direct to" is likely a waypoint, and the word immediately following "runway" is a runway number). In the initial text sequence, these guide words are identified through string matching or natural language processing techniques. For example, the guide word "direct to" is identified in "direct to BIKMA". Then, based on the guide word and its preset positional relationships, words with specific positional relationships to the guide word are extracted from the text sequence as candidate position words. For example, for "direct to", the candidate position word is the first word immediately following it, "BIKMA"; for "runway27L", the candidate position word is "27L". This step marks a word in the original text as a "potential technical term," providing a clear target for the next step of verification. Without this step, subsequent word library verification would require processing the entire text, which is inefficient and prone to misjudgment.

[0126] S400: Determine whether the candidate position word exists in the preset position word library.

[0127] This step is the core verification process. The candidate location term (e.g., "BIKMA") is compared with the pre-defined location term library for the current area associated with S100. If "BIKMA" exists in the library, it indicates that the preliminary identification result is consistent with the knowledge of the current area, and the term can be initially accepted. However, considering that the identification confidence may be low, further judgment can be made based on the confidence level (see the refinement in S500). If "BIKMA" does not exist in the library, it indicates that the preliminary identification is likely incorrect, because in this geographical area, it is unlikely that the controller would mention an irrelevant waypoint. In this case, a correction mechanism needs to be triggered.

[0128] S500: If it does not exist, then perform constrained secondary recognition on the speech segment corresponding to the candidate position word to obtain the corrected position word.

[0129] When a candidate positional word is not in the region lexicon, a constrained secondary recognition is triggered. This step solves the problem of not being able to automatically correct errors after the initial recognition. Specifically, in implementation:

[0130] First, S501: Extract the audio segments corresponding to candidate position words from the original speech (e.g., the pronunciation segment of "BIKMA").

[0131] Then, S502: The candidate word set is forcibly limited to all words in the preset position word library associated in S100 that match the candidate position word type (for example, if "BIKMA" is marked as a waypoint, the candidate set is only all waypoint names in the word library).

[0132] Next, S503: The speech recognition engine is used to recalculate the matching probability of each candidate word with the audio segment in the restricted word set, and the word with the highest probability is selected as the correction result. For example, in the "Beijing Approach" word set, waypoints include "BIDIB", "ELKUR", "VYK", etc. The matching results show that the audio has the highest similarity with "BIDIB" (probability 0.92), so the correction result is "BIDIB".

[0133] By employing this "forced constraint" method, the identification results are guaranteed to fall within the valid term set of the current region, thus avoiding cross-regional misidentification. Simultaneously, due to the significantly reduced candidate set, both the accuracy and speed of identification are improved.

[0134] It should be further explained that after the secondary identification, the system will obtain the confidence level of the corrected result.

[0135] S510: If the confidence level is lower than the preset discard threshold (e.g., 0.5), it is determined that the original preset position guide word may be a non-real guide word (e.g., "to" in everyday language), and the correction is abandoned. The original words in the initial text sequence are retained to avoid erroneous intervention in non-critical information. This mechanism reflects an understanding of the discourse context and prevents the system from making incorrect corrections.

[0136] Furthermore, S520: Even if the candidate positional word exists in the lexicon, if the confidence level of the initial identification is lower than a preset threshold (e.g., 0.8), the aforementioned constrained secondary identification can still be triggered to ensure the reliability of the identification result. This ensures that even if the initial identification seems correct, a review can still be performed if the confidence level is insufficient, thus improving the robustness of the system.

[0137] S600: Based on the corrected position words, output the final recognition result.

[0138] The corrected positional words replace the corresponding candidate positional words in the initial text sequence to obtain the complete corrected text. For example, replacing "BIKMA" with "BIDIB" outputs "Cessna one two three proceed direct toBIDIB". To enhance human-machine trust, the corrected positional words are presented in a visual style distinct from the uncorrected words on the display interface (such as blue font or a special background color). Pilots can click on the word to replay the original audio and judge the accuracy of the recognition themselves. In this way, the machine's auxiliary role is utilized while retaining the pilot's final judgment.

[0139] The present invention also includes a method for verifying the rationality of operation instructions, comprising the following steps:

[0140] S700: Determines the current flight phase based on real-time location information and associates it with the preset operating parameter range corresponding to the current flight phase.

[0141] S701: Identify preset operation guide words in the initial text sequence. Preset operation guide words are guide words used to indicate the subsequent occurrence of operation instruction related words.

[0142] S702: Locate candidate operation parameters based on preset operation guide words.

[0143] S703: Compare the candidate operating parameters with the preset operating parameter range.

[0144] S704: If the operation exceeds the preset operating parameter range, an operation rationality alarm will be triggered.

[0145] This method is a preferred embodiment. The system determines the current flight phase based on real-time location information, such as determining that the aircraft is in the approach phase using GPS altitude and aeronautical chart data. The system is pre-configured with preset operating parameter ranges corresponding to each flight phase, such as an approach phase altitude value below 5000 feet and a speed value between 150 and 250 knots.

[0146] After obtaining the initial text sequence, the system identifies preset operation guidance words, such as "descend," "climb," and "turn," to locate candidate operation parameters. For example, when the instruction is "descend to one zero thousand," the system identifies the guidance word "descend to," and the candidate operation parameter is "one zero thousand" (i.e., 10,000 feet). The system compares this parameter with the preset operation parameter range corresponding to the current flight phase. If it exceeds the range, an operation rationality alarm is triggered, displaying a warning bar at the top of the interface and highlighting the abnormal parameter in red, reminding the pilot to verify the instruction. This application elevates recognition from "literal translation" to "semantic understanding," providing pilots with deeper safety assurance.

[0147] This embodiment, through S100, utilizes a real-time location-associated regional lexicon to narrow the recognition scope from global terms to local terms, fundamentally eliminating the possibility of cross-regional terminology confusion, which forms the basis for all subsequent verifications. S300 accurately locates candidate words through guiding words, avoiding invalid processing of non-critical information and improving system efficiency. The "lexicon verification + constraint-based secondary recognition" of S400 and S500 constitutes a closed-loop correction mechanism: when a candidate word is not in the regional lexicon, forced re-recognition within the lexicon ensures that the output result necessarily belongs to the legitimate terminology of the current region, effectively correcting recognition errors caused by accents, noise, etc. At the same time, the confidence judgment mechanism introduced in S500 is further refined, preventing the miscorrection of low-confidence words and avoiding the omission of correct words with insufficient confidence, achieving a balance between accuracy and robustness. The extended application of S700 elevates recognition to the semantic understanding level, providing more comprehensive protection for flight safety. In summary, the method of this invention, through multi-level information fusion, significantly improves the recognition accuracy of professional terms in aviation communications and enhances the intelligence and reliability of the system.

[0148] Furthermore, in actual air traffic control communications, different controllers have different pronunciation characteristics, accents, and speaking speeds. Even among controllers within the same geographical area, individual differences can lead to fluctuations in the recognition performance of a general acoustic model. For example, some controllers may have regional accents, pronouncing "three" as something like "tree," or "five" as "fife." The general model struggles to cover all individual differences during training, resulting in decreased recognition accuracy for specific controllers. This step aims to further improve recognition accuracy by creating a personalized acoustic feature vector for each controller and incorporating this feature for adaptive adjustment in the constrained secondary recognition process.

[0149] S710: Obtain the current controller's identity identifier and obtain the preset acoustic feature vector corresponding to the identity identifier.

[0150] Once an aircraft enters the current radio frequency, the system needs to identify the controller who is speaking. This can be achieved in two ways:

[0151] Method 1: Associate the current radio frequency with pre-set controller information. Each control frequency usually corresponds to a specific control position (e.g., the "Beijing Approach" frequency 118.35MHz), and the system can pre-bind the frequency with the controller's identity (e.g., job number, call sign).

[0152] Method 2: Speaker recognition of the received voice. After capturing the controller's first sentence, the system can extract its voiceprint features and compare them with a pre-established controller voiceprint database to identify the controller's identity.

[0153] Once the controller's identity is obtained, the system queries the local database to retrieve the preset acoustic feature vector corresponding to that identity. This vector is extracted and stored in advance from historical voice data and represents the controller's personalized pronunciation characteristics.

[0154] S711: Collect the controller's historical voice data.

[0155] To construct personalized acoustic feature vectors, it is necessary to accumulate the controller's historical voice data. When an aircraft first enters a control area, or when the system detects a new controller coming online, the system begins collecting voice segments from that controller. The collection process continues; for each complete control instruction received, the system stores the audio and the corresponding recognized text (or manually verified text) in the controller's dedicated dataset. Once a certain amount of data has been accumulated (e.g., 20 sentences), it can be used for feature extraction.

[0156] S712: Extract acoustic features from the historical speech data to generate an acoustic feature vector characterizing the controller's pronunciation characteristics.

[0157] Using feature extraction techniques commonly used in speaker recognition, such as x-vectors or i-vectors, fixed-dimensional acoustic feature vectors are extracted from historical speech data. Specifically, the audio of each sentence is segmented into short time frames, and low-level acoustic features such as Mel-frequency cepstral coefficients (MFCCs) are extracted. Then, a high-dimensional vector (e.g., 512-dimensional) is obtained through mapping using a deep neural network. This vector effectively represents the speaker's pronunciation habits, timbre, accent, and other individual information. Finally, the feature vectors of all historical sentences are averaged or pooled to obtain the controller's stable acoustic feature vector.

[0158] S713: The acoustic feature vector is associated with the controller's identity and stored as the preset acoustic feature vector.

[0159] The extracted feature vectors are bound to controller identification identifiers (such as employee ID and call sign) and stored in a local database for subsequent identification. When the controller reappears, the system can directly load the vector for adaptive identification.

[0160] S720: In the preliminary speech recognition or the constrained secondary recognition, the output of the general acoustic model is adaptively adjusted at least in part based on the preset acoustic feature vector to obtain a personalized recognition result.

[0161] In subsequent speech recognition processes (whether it's the initial recognition by the S200 or the constrained secondary recognition by the S500), the system loads the current controller's preset acoustic feature vector and adaptively adjusts the output of the general acoustic model. Specific implementation methods include, but are not limited to:

[0162] Feature-level adaptation: The acoustic feature vector is used as an additional input, concatenated with the acoustic features of the audio, and then fed into the model.

[0163] Model-level adaptation: By linear transformation or fine-tuning a small number of model parameters, the model output is biased towards the controller's pronunciation distribution.

[0164] Decoding-level adaptation: During the decoding process, the acoustic model score is offset and adjusted using feature vectors, for example, by increasing the probability of pronunciations that match the feature vectors.

[0165] Taking the constrained secondary recognition in the S500 as an example: when the candidate position word "BIKMA" is not in the regional word library and secondary recognition is required, the system not only limits the candidate word set to the regional word library, but also loads the current controller's personalized feature vector. When calculating the matching probability between the audio segment and the candidate word "BIDIB", since the controller has historically often pronounced "BIDIB" with a slight retroflex sound, the general model may underestimate the probability of this pronunciation. However, through adaptive adjustment of the personalized feature vector, the model will improve the score of "BIDIB" under similar pronunciations, thus selecting the correct result more accurately.

[0166] Through S710 to S720, this invention further addresses the problem of general acoustic models failing to adapt to individual pronunciation differences. First, S711 to S713 construct a personalized acoustic feature library for controllers, solidifying each controller's pronunciation characteristics as feature vectors, providing a foundation for subsequent adaptive processing. S720 incorporates these feature vectors during the recognition process, specifically adjusting the output of the general model to better reflect the actual pronunciation habits of the current controllers. This lightweight adaptive "one-person-one-model" approach avoids the high cost of retraining a complete acoustic model for each controller and significantly improves recognition accuracy under individual differences such as accent and speech rate. For example, for controllers with pronounced regional accents, words that might have been frequently misidentified (such as "three" being identified as "tree") are corrected after adaptation, further ensuring the accuracy of pilots receiving instructions. This improvement is organically combined with the S500's constrained secondary recognition, and personalized acoustic adaptation is superimposed on the basis of the word library limitation, forming a dual guarantee: on the one hand, the rationality of terminology is ensured through geographical area constraints, and on the other hand, the acoustic matching degree is improved through speaker adaptation, ultimately achieving highly robust aviation speech recognition.

[0167] Furthermore, although the steps of the method in this disclosure are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or a step may be broken down into multiple steps.

[0168] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, mobile terminal, or network device, etc.) to execute the methods according to the embodiments of this disclosure.

[0169] In an exemplary embodiment of this disclosure, an electronic device capable of implementing the above-described method is also provided.

[0170] Those skilled in the art will understand that various aspects of the present invention can be implemented as systems, methods, or program products. Therefore, various aspects of the present invention can be specifically implemented in the following forms: entirely in hardware, entirely in software (including firmware, microcode, etc.), or in a combination of hardware and software, collectively referred to herein as “circuit,” “module,” or “system.”

[0171] An electronic device according to this embodiment of the invention. The electronic device is merely an example and should not be construed as limiting the functionality or scope of the embodiments of the invention.

[0172] Electronic devices are manifested in the form of general-purpose computing devices. Components of an electronic device may include, but are not limited to: at least one processor, at least one memory, and buses connecting different system components (including memory and processor).

[0173] The memory stores program code that can be executed by a processor, causing the processor to perform the steps described in the "Exemplary Methods" section above, according to various exemplary embodiments of the present invention.

[0174] The storage may include readable media in the form of volatile storage, such as random access memory (RAM) and / or cache memory, and may further include read-only memory (ROM).

[0175] The storage may also include programs / utilities having a set (at least one) of program modules, including but not limited to: an operating system, one or more applications, other program modules, and program data, each or some combination of these examples may include an implementation of a network environment.

[0176] A bus can represent one or more of several bus architectures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus that uses any of the various bus architectures.

[0177] The electronic device can also communicate with one or more external devices (e.g., keyboards, pointing devices, Bluetooth devices, etc.), one or more devices that enable a user to interact with the electronic device, and / or any device that enables the electronic device to communicate with one or more other computing devices (e.g., routers, modems, etc.). This communication can be performed via input / output (I / O) interfaces. Furthermore, the electronic device can communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter. The network adapter communicates with other modules of the electronic device via a bus. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with the electronic device, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0178] In exemplary embodiments of this disclosure, a computer-readable storage medium is also provided, on which a program product capable of implementing the methods described above is stored. In some possible embodiments, various aspects of the present invention may also be implemented as a program product comprising program code that, when the program product is run on a terminal device, causes the terminal device to perform the steps of the various exemplary embodiments of the present invention described in the "Exemplary Methods" section above.

[0179] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0180] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of sending, propagating, or transmitting programs for use by or in conjunction with an instruction execution system, apparatus, or device.

[0181] The program code contained on the readable medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0182] Program code for performing the operations of this invention can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0183] Furthermore, the accompanying drawings are merely illustrative of the processes included in the method according to exemplary embodiments of the present invention and are not intended to be limiting. It is readily understood that the processes shown in the above drawings do not indicate or limit the temporal order of these processes. Additionally, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.

[0184] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0185] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for displaying aviation communication information, characterized in that, Includes the following steps: After the pilot switches to the new radio frequency, multiple communications occurring on the current radio frequency are acquired in real time; these communications are between the controller and other flights. For each call, obtain the relevance score between the call and the local machine; the relevance score between the i-th call and the local machine satisfies the following relationship: ； in, For spatial distance risk, its value is negatively correlated with the real-time distance between the aircraft and other flights obtained based on the ADS-B receiver; For call sign similarity risk, its value is positively correlated with the string similarity between the flight call sign mentioned in the call and the call sign of the user; To mitigate the risk of microphone suppression, voice activity detection and voiceprint segmentation are used to determine if microphone suppression occurs due to the superposition of voices from different flights. If so, ,otherwise ; Content matching accuracy is determined based on whether the caller ID is mentioned in the call, whether it contains an emergency indicator, and whether it matches a preset security event template. α, β, μ, δ are preset weight coefficients and satisfy α + β + μ + δ = 1; The correlation score is compared with a preset threshold. When the correlation score is greater than or equal to the preset threshold, the corresponding call is highlighted and fixed in the top area of ​​the display interface so that it is not overwritten by subsequent new call information.

2. The method according to claim 1, characterized in that, Content matching degree Obtain it through the following methods: ； in, Let i be the content matching degree of the i-th call. Callsign matching score, used to characterize the degree to which the local machine is mentioned in the call. Semantic relevance is used to characterize the degree of relevance between the content of the call and related events on the local machine. The urgency level is used to indicate whether the call contains an emergency flag. , , Preset weighting coefficients and satisfying ; Obtain it through the following methods: Identify whether the call contains keywords from a pre-defined aviation safety event keyword database, and obtain the weight of each keyword; Identify whether the call matches a preset aviation safety event template; The content matching degree is obtained based on the weight of the keyword and whether it matches the event template. The content matching degree satisfies the following relationship: ； in, For the i-th call, for The weight of keyword k in the text. This is the sum of the weights of all keywords in the keyword database. This is an indicator value for whether the event template matches. These are the preset template weight coefficients.

3. The method according to claim 1, characterized in that, Also includes: When the correlation score is less than the preset threshold, the corresponding call is displayed in a normal style or a weakened style.

4. The method according to claim 1, characterized in that, Also includes: When switching to a new radio frequency, multiple communication pairs corresponding to the new radio frequency are acquired, each communication pair containing controller instructions and corresponding pilot repetitions; Based on the multiple call pairs, the comprehension difficulty coefficient of the controller is obtained, and the comprehension difficulty coefficient satisfies the following relationship: ； in, To understand the difficulty level, The average delay time obtained from the statistics of the multiple call pairs is used to characterize the average response speed of the pilot to commands. To preset the maximum delay time, The average request repetition rate, calculated based on the statistics of the multiple call pairs, is used to characterize the average frequency at which instructions are requested to be repeated. The average recognition confidence level, obtained statistically from the multiple call pairs, is used to characterize the average degree of confidence the speech recognition system has in the instructions. The average consistency error rate, calculated based on the statistics of the multiple call pairs, is used to characterize the average frequency of inconsistencies between the pilot's repetition and instructions. Preset weighting coefficients and satisfying ; When the difficulty level of understanding exceeds the preset difficulty threshold, the deep semantic verification function is activated.

5. The method according to claim 4, characterized in that, The delay time is the time interval between the issuance of the instruction in each call pair and the pilot's first repetition; the request repetition rate is calculated based on the indication value of whether the pilot in each call pair has issued a repetition request. The recognition confidence level is obtained by statistically analyzing the confidence values ​​provided by the speech recognition engine for each command; The consistency error rate is calculated based on the indicator value indicating whether the comparison result between the instruction and the repeat in each call pair is consistent.

6. The method according to claim 4, characterized in that, The deep semantic verification function includes: Semantic analysis is performed on subsequent calls between the controller and the local machine to obtain the operation type, operation object, and operation value, forming an operation triplet; The operation triplet is compared with the preset operation parameter range corresponding to the current flight phase. If it exceeds the range, a reasonableness alarm is triggered. The system identifies consecutive control commands and pilot recitations, obtains the corresponding operation triplets, and performs a structured comparison. If there is a discrepancy, a consistency alarm is triggered.

7. The method according to claim 6, characterized in that, The current flight phase is determined based on the aircraft's real-time position information, and the preset operating parameter range is associated with the current flight phase.

8. The method according to claim 1, characterized in that, Also includes: It receives voice messages on the current radio frequency in real time and performs speech recognition to obtain the real-time call text; Semantic analysis is performed on the real-time call text to obtain its real-time relevance score; The display style of the real-time call text is determined based on the comparison result between the real-time relevance score and the preset threshold. The real-time call text is dynamically displayed on the display interface in a preset display style.

9. A non-transitory computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements an aviation communication information display method as described in any one of claims 1 to 8.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements an aviation communication information display method as described in any one of claims 1 to 8.

Images