A method, device and storage medium for decoding cabin sound time signal

Abstract

The present invention discloses a method, device and storage medium for decoding a cockpit voice time signal, the method comprising: obtaining cockpit voice data of a target aircraft and preprocessing it; performing short-time Fourier transform on the preprocessed cockpit voice data to obtain its time-frequency spectrum; selecting first spectrum data corresponding to all candidate cockpit voice times from the time-frequency spectrum; decoding each first spectrum data to obtain a corresponding candidate binary sequence; selecting a candidate binary subsequence from each candidate binary sequence based on a preset rule that binary symbols represent cockpit voice time signals; converting all candidate binary subsequences based on the preset rule to obtain all candidate cockpit voice times; correcting all candidate cockpit voice times to obtain all target cockpit voice times. The present invention can accurately obtain and decode cockpit voice time signals under the condition of frequency loss, so as to achieve precise timing of CVR data, thereby improving the accuracy and reliability of the data.

Description

Technical Field

[0001] The present invention relates to the field of cabin sound signal decoding, and in particular to a method, device and storage medium for decoding a cabin sound time signal. Background Art

[0002] In the civil aviation industry, the Cockpit Voice Recorder (CVR) is a key device used to record the sounds in the cockpit, including pilots' conversations, warning sounds, and possible mechanical failure sounds. These records are crucial for investigating the cause of an aircraft accident when one occurs.

[0003] At present, the commonly used method for decoding civil aviation cabin sound time signals is frequency shift keying (FSK) decoding. FSK is a digital modulation and demodulation technology that transmits digital information by changing the frequency of the carrier signal. The received FSK signal is first filtered and amplified to improve the signal quality, and then the signal frequency is detected by a frequency detector, and finally the original digital information is restored based on the detected frequency. In the specific FSK decoding, the frequency is usually divided into low frequency and high frequency, representing the numbers "0" and "1" respectively. For example, a specific low frequency range may represent the number "0", while a specific high frequency range may represent the number "1". In this way, the frequency changes of the signal can be used to encode and transmit digital information. At the receiving end, the FSK decoder needs to accurately detect and identify these frequency changes to restore the original digital information.

[0004] In actual use scenarios in the civil aviation field, the sampling rate of cabin sound is 8000Hz, but in actual operation, CVR data may have frequency loss problems due to hardware failure, electromagnetic interference, signal transmission loss, etc., resulting in the lack of high-frequency information in the cabin sound data, such as high-frequency information above 4000Hz. In this case, when using FSK to decode the audio, no specific "0" and "1" will be obtained. The low-frequency "0" range is 3607±30Hz, and the high-frequency "1" range is 4193±30Hz. This means that any frequency outside the range of 3577Hz to 3637Hz is identified as "0", and any frequency within the range of 4163Hz to 4223Hz is identified as "1". When the CVR data lacks high-frequency information above 4000Hz, the system will not be able to detect the high-frequency signal marked as "1", and thus cannot perform effective FSK decoding. This defect makes the traditional FSK method infeasible for decoding civil aviation cabin sound time signals, and it is impossible to obtain accurate time information, which poses a great challenge to accident cause investigation and flight performance analysis and maintenance. Therefore, how to decode the cabin sound time signal under the condition of frequency loss is an urgent problem to be solved. Summary of the invention

[0005] The purpose of the embodiments of the present invention is to provide a method, device and storage medium for decoding the cabin sound time signal, which can accurately acquire and decode the cabin sound time signal under the condition of frequency loss, so as to achieve precise timing of CVR data, thereby improving the accuracy and reliability of the data.

[0006] A first aspect of the present invention provides a method for decoding a cabin sound time signal, comprising:

[0007] Acquire the target aircraft's cabin sound data;

[0008] Preprocessing the cabin sound data to obtain preprocessed cabin sound data;

[0009] Performing short-time Fourier transform on the preprocessed cabin sound data to obtain a time spectrum diagram thereof;

[0010] Selecting first spectrum data corresponding to all candidate cabin sound times from the time-spectrogram;

[0011] Decoding each of the first spectrum data to obtain a corresponding candidate binary sequence;

[0012] Selecting a candidate binary subsequence from each of the candidate binary sequences based on a preset rule for representing cabin sound time signals with binary symbols;

[0013] Convert all the candidate binary subsequences based on the preset rule to obtain all candidate cabin sound times;

[0014] All the candidate cabin voice times are corrected to obtain all target cabin voice times.

[0015] Optionally, preprocessing the cabin sound data to obtain preprocessed cabin sound data includes:

[0016] The channel of the cabin sound data is set to mono, the bit depth is set to 16 bits and the sampling rate is adjusted to 16000 Hz.

[0017] Optionally, the number of candidate cabin sound times is determined by the number of bright line areas in the time-spectrogram whose frequency intensity is greater than a preset intensity threshold.

[0018] Optionally, the selecting the first spectrum data corresponding to all candidate cabin sound times from the time-spectrogram includes:

[0019] Acquire a bright line region whose frequency intensity is greater than a preset intensity threshold from the time-frequency spectrum in the direction of the frequency axis; wherein the bright line region is parallel to the frequency axis;

[0020] Acquire row spectrum data with the largest frequency average intensity from the time-frequency spectrum diagram in the direction of the time axis; wherein the row spectrum data is parallel to the direction of the time axis;

[0021] Extend all the bright line areas to be perpendicular to the line spectrum data to obtain N vertical intersection points, where N≥1;

[0022] Based on a preset interval, N first spectrum data are selected from the row spectrum data; wherein each of the preset intervals includes a vertical intersection; and each of the first spectrum data corresponds to a candidate cabin sound time.

[0023] Optionally, decoding each of the first spectrum data to obtain a corresponding candidate binary sequence includes:

[0024] Perform threshold decoding judgment on each of the first spectrum data to obtain a corresponding candidate binary sequence; wherein the threshold decoding judgment is to judge whether the values ​​of all frequency components in each of the first spectrum data are greater than a preset decoding threshold.

[0025] Optionally, the preset rule based on binary symbols to represent the cabin sound time signal is to select a candidate binary subsequence from each of the candidate binary sequences, specifically:

[0026] Based on a preset rule for representing cabin sound time signals with binary symbols, identifying a preset header array from each of the candidate binary sequences;

[0027] Extracting K bits of data located after the corresponding preset head array from each candidate binary sequence as a tail array; wherein K≥1;

[0028] In each combination of the preset head array and the tail array, M bits of data are selected from the back to the front as the corresponding candidate binary subsequence; wherein M>K;

[0029] Optionally, the converting of all the candidate binary subsequences based on the preset rule to obtain all candidate cabin sound times is specifically:

[0030] According to the preset rule, determine the binary combination corresponding to the hour, minute and second in all the candidate binary subsequences;

[0031] Convert each of the binary combinations into decimal to obtain the corresponding candidate cabin sound time.

[0032] Optionally, the correcting all the candidate cabin voice times to obtain all target cabin voice times includes:

[0033] Compare all the candidate cabin sound time pairs, and calculate the time interval between each candidate cabin sound time pair;

[0034] Determine whether each time interval is an integer multiple of a preset time difference;

[0035] If yes, then record the corresponding candidate cabin voice time pair;

[0036] The candidate cabin sound time that appears most frequently in the recording results is used as the reference time;

[0037] All the candidate cabin sound times are corrected based on the reference time and the preset time difference to obtain all the target cabin sound times.

[0038] The second aspect of the present invention further provides a device for decoding a cabin sound time signal, which is used to implement the method for decoding a cabin sound time signal described in the first aspect. The device includes:

[0039] A cockpit sound data acquisition module, used to acquire the cockpit sound data of the target aircraft;

[0040] A cabin sound data preprocessing module, used for preprocessing the cabin sound data to obtain preprocessed cabin sound data;

[0041] A short-time Fourier transform module, used for performing a short-time Fourier transform on the pre-processed cabin sound data to obtain a time spectrum thereof;

[0042] A first spectrum data selection module is used to select first spectrum data corresponding to all candidate cabin sound times from the time-spectrum diagram;

[0043] A first spectrum data decoding module, used for decoding each of the first spectrum data to obtain a corresponding candidate binary sequence;

[0044] A candidate binary subsequence selection module, configured to select a candidate binary subsequence from each of the candidate binary sequences based on a preset rule for representing cabin sound time signals with binary symbols;

[0045] A candidate cabin sound time acquisition module, used for converting all the candidate binary subsequences based on the preset rule to obtain all candidate cabin sound times;

[0046] The candidate cabin voice time correction module is used to correct all the candidate cabin voice times to obtain all target cabin voice times.

[0047] An embodiment of the third aspect of the present invention also provides a computer-readable storage medium, which includes a stored computer program; wherein, when the computer program is running, it controls the device where the computer-readable storage medium is located to execute the method for decoding the cabin sound time signal as described in any one of the first aspects above.

[0048] Compared with the prior art, the embodiments of the present invention provide a method, device and storage medium for decoding a cockpit sound time signal, the method comprising: obtaining cockpit sound data of a target aircraft; preprocessing the cockpit sound data to obtain preprocessed cockpit sound data; performing short-time Fourier transform on the preprocessed cockpit sound data to obtain its time-frequency spectrum; selecting first spectrum data corresponding to all candidate cockpit sound times from the time-frequency spectrum; decoding each first spectrum data to obtain a corresponding candidate binary sequence; selecting a candidate binary subsequence from each candidate binary sequence based on a preset rule that binary symbols represent cockpit sound time signals; converting all candidate binary subsequences based on the preset rule to obtain all candidate cockpit sound times; correcting all candidate cockpit sound times to obtain all target cockpit sound times. The present invention can accurately obtain and decode cockpit sound time signals under the condition of frequency loss to achieve precise timing of CVR data, thereby improving the accuracy and reliability of data, which is of great value to aviation safety and aircraft maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a flowchart of an embodiment of a method for decoding cabin sound time signal provided by the present invention;

[0050] Figure 2 is an example diagram of an embodiment of the time-frequency spectrum diagram provided by the present invention;

[0051] Figure 3 is an example diagram of an embodiment of the line spectrum data provided by the present invention;

[0052] Figure 4 is an example diagram of an embodiment of the first spectrum data provided by the present invention;

[0053] Figure 5 is a flow chart of another embodiment of a method for decoding a cabin sound time signal provided by the present invention;

[0054] Figure 6 It is a structural schematic diagram of an embodiment of a decoding device for cabin sound time signal provided by the present invention. DETAILED DESCRIPTION

[0055] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this technical field without creative work are within the scope of protection of the present invention.

[0056] See also Figure 1, is a flow chart of an embodiment of a method for decoding a cabin sound time signal provided by the present invention.

[0057] The embodiment of the present invention provides a method for decoding a cabin sound time signal, comprising steps S1 to S8:

[0058] Step S1: Acquire the cabin sound data of the target aircraft;

[0059] Step S2: preprocessing the cabin sound data to obtain preprocessed cabin sound data;

[0060] Step S3: performing short-time Fourier transform on the pre-processed cabin sound data to obtain a time spectrum thereof;

[0061] Step S4: selecting first spectrum data corresponding to all candidate cabin sound times from the time-spectrogram;

[0062] Step S5: Decode each of the first spectrum data to obtain a corresponding candidate binary sequence;

[0063] Step S6: based on the preset rule of using binary symbols to represent cabin sound time signals, selecting a candidate binary subsequence from each of the candidate binary sequences;

[0064] Step S7: converting all the candidate binary subsequences based on the preset rule to obtain all candidate cabin sound times;

[0065] Step S8: Correct all the candidate cabin voice times to obtain all target cabin voice times.

[0066] It should be noted that the frequency loss condition in the embodiment of the present invention includes the loss or reduction of any frequency component in the signal. Under the original FSK signal modulation, the short-time Fourier transform (STFT) is introduced to perform spectrum analysis on the cabin sound time signal to deal with the problem of frequency loss. For example, when cabin sound data is recorded in a complex environment, some frequency components may not be clearly identified in the entire signal when the cabin sound data is analyzed using the standard Fourier transform, and thus the correct FSK decoding cannot be performed; while in the time-frequency spectrum obtained by STFT analysis, the change of the signal frequency over time can be clearly seen. By observing the time-frequency spectrum, a specific frequency component (a key row of data in the instant frequency spectrum, which contains the first spectrum data corresponding to all candidate cabin sound times) can be found, and then all candidate cabin sound times are obtained according to the preset rule that the row of data and the binary symbol represent the cabin sound time signal. The embodiment of the present invention can effectively decode the cabin sound time signal under the condition of partial frequency loss, and realize accurate timing of the cabin sound data, thereby improving the accuracy and reliability of the data.

[0067] In an optional embodiment, the preprocessing of the cabin voice data in step S2 to obtain the preprocessed cabin voice data includes:

[0068] The channel of the cabin sound data is set to mono, the bit depth is set to 16 bits and the sampling rate is adjusted to 16000 Hz.

[0069] In the specific implementation, the cabin sound data is preprocessed / parameters are adjusted using the Adobe Audition tool or Java or Python programming languages. In order to achieve more accurate decoding and timing, we set the cabin sound data to mono, with a bit depth of 16 and a sampling rate of 16000Hz. Such settings can effectively reduce the complexity of the data while maintaining the quality and accuracy of the data, and facilitate subsequent processing.

[0070] In an optional embodiment, the number of candidate cabin sound times is determined by the number of bright line areas in the time-spectrogram whose frequency intensity is greater than a preset intensity threshold.

[0071] It should be noted that the short-time Fourier transform combines the characteristics of the Fourier transform and the window function to analyze the frequency content of a signal that changes over time. The Fourier transform converts the signal from the time domain to the frequency domain, providing a global view of the frequency components of the signal, but cannot provide information about how the frequency components change over time. This can be a problem when analyzing non-stationary signals, such as audio signals or electromagnetic signals. To solve this problem, the STFT divides the signal into shorter time periods and performs a Fourier transform on each segment, so that the frequency content of the signal in each small time period can be seen, and a local view of frequency and time can be obtained. The calculation of the STFT involves a window function that weights the signal in each small time period to ensure that the beginning and end of each signal segment are smooth to reduce spectral leakage in the Fourier transform.

[0072] The embodiment of the present invention performs STFT on the cabin sound data to obtain its complete time-frequency spectrum. Figure 2 As shown, it is an example diagram of an embodiment of the time-frequency spectrum diagram provided by the present invention, which shows a comprehensive view of the frequency distribution of cabin sound data; wherein, the bright lines in the white frame are the bright line areas of the embodiment of the present invention. Through this complete time-frequency spectrum diagram, the number of cabin sound time signals contained in the cabin sound data can be determined (the same as the number of bright lines in the vertical direction), and each bright line area represents a line area in the time-frequency spectrum diagram whose frequency intensity is greater than a preset intensity threshold.

[0073] In an optional embodiment, the step S4 described above of selecting the first spectrum data corresponding to all candidate cabin sound times from the time-spectrogram includes:

[0074] Acquire a bright line region whose frequency intensity is greater than a preset intensity threshold from the time-frequency spectrum in the direction of the frequency axis; wherein the bright line region is parallel to the frequency axis;

[0075] Acquire row spectrum data with the largest frequency average intensity from the time-frequency spectrum diagram in the direction of the time axis; wherein the row spectrum data is parallel to the direction of the time axis;

[0076] Extend all the bright line areas to be perpendicular to the line spectrum data to obtain N vertical intersection points, where N≥1;

[0077] Based on a preset interval, N first spectrum data are selected from the row spectrum data; wherein each of the preset intervals includes a vertical intersection; and each of the first spectrum data corresponds to a candidate cabin sound time.

[0078] In specific implementation, N bright line areas in the vertical direction / frequency axis direction are determined from the time-spectrum graph; and the row spectrum data with the largest average frequency intensity in the horizontal direction / time axis direction is determined from the time-spectrum graph (that is, the row data is the clearest, that is, the row with the highest brightness in the entire time-spectrum graph). Figure 3 The figure shown is an example diagram of an embodiment of the row spectrum data provided by the present invention, and the row spectrum data is obtained by horizontally slicing from the complete time-frequency spectrum diagram and plotted as a bar graph for observation. The row spectrum data includes the first spectrum data corresponding to all candidate cabin sound times, and the specific steps of obtaining N first spectrum data from the row spectrum data are as follows: each bright line area is extended to be perpendicular to the row spectrum data to obtain N vertical intersections, and a preset interval data including each vertical intersection is intercepted on the row spectrum data as the first spectrum data, and the first spectrum data includes a complete cabin sound time signal. As shown in FIG. Figure 4 FIG. 1 is an exemplary diagram of an embodiment of the first spectrum data provided by the present invention. Figure 4 The position of the bright line area Figure 3 The line spectrum data shown is sliced ​​longitudinally to obtain (i.e. Figure 4 The columnar data in is Figure 3 The column data from columns 2330 to 2430 in the figure provide a more detailed data analysis view to help understand the detailed characteristics of the cabin sound data.

[0079] In an optional embodiment, the decoding of each of the first spectrum data in the above step S5 to obtain a corresponding candidate binary sequence includes:

[0080] Perform threshold decoding judgment on each of the first spectrum data to obtain a corresponding candidate binary sequence; wherein the threshold decoding judgment is to judge whether the values ​​of all frequency components in each of the first spectrum data are greater than a preset decoding threshold.

[0081] It should be noted that the embodiment of the present invention decodes the first spectrum data by means of threshold decoding judgment, that is, judging whether the values ​​of all frequency components in each of the first spectrum data are greater than a preset decoding threshold; if greater than or equal to the preset decoding threshold, the value is decoded as 1; if less than the preset decoding threshold, the value is decoded as 0. Figure 4 As shown, the binary sequence of “00000000 10110110 10101100 10010110” is read from the 22-bit to 54-bit values; wherein the binary sequence is a candidate binary subsequence.

[0082] In an optional embodiment, the preset rule of representing the cabin sound time signal based on binary symbols in the above step S6 is to select a candidate binary subsequence from each of the candidate binary sequences, specifically:

[0083] Based on a preset rule for representing cabin sound time signals with binary symbols, identifying a preset header array from each of the candidate binary sequences;

[0084] Extracting K bits of data located after the corresponding preset head array from each candidate binary sequence as a tail array; wherein K≥1;

[0085] In each combination of the preset head array and the tail array, M bits of data are selected from the back to the front as the corresponding candidate binary subsequence; wherein M>K;

[0086] It should be noted that in the implementation of the present invention, the preset rules for binary symbols to represent the cabin sound time signal are: (1) a cabin sound time signal is a sequence composed of 32-bit binary symbols; (2) in the 32-bit binary sequence corresponding to each cabin sound time signal, the 28th to 24th bits represent hours, the 23rd to 18th bits represent minutes, and the 17th to 12th bits represent seconds; (3) in the 32-bit binary sequence corresponding to each cabin sound time signal, the first bit is the 9-bit "000000001", and the next 23 bits are any combination of 0 and 1.

[0087] Take the extraction of a candidate binary subsequence for characterizing a specific candidate cabin sound time signal from a candidate binary sequence as an example to explain: In some cases, the beginning of each time data may not be "000000001"; therefore, when a preset header array (5 or more consecutive 0s followed by a 1) is identified in the candidate binary sequence (greater than 32 bits), it is considered that a candidate binary subsequence (32 bits) corresponding to a candidate cabin sound time signal is found; then, the last 23 bits are automatically identified; finally, the 32-bit candidate binary subsequence is obtained in reverse order from the tail. This can effectively solve the problem of sequence omission caused by irregular data format in some cases, ensuring that each candidate binary subsequence corresponding to the cabin sound time signal can be accurately captured.

[0088] In an optional embodiment, the above step S7 converts all the candidate binary subsequences based on the preset rule to obtain all the candidate cabin voice time, specifically:

[0089] According to the preset rule, determine the binary combination corresponding to the hour, minute and second in all the candidate binary subsequences;

[0090] Convert each of the binary combinations into decimal to obtain the corresponding candidate cabin sound time.

[0091] It should be noted that after obtaining the candidate binary subsequence (32 bits) corresponding to each cabin sound time signal, the specific time corresponding to each subsequence is found out according to the preset rules.

[0092] For example, if the candidate binary subsequence obtained by parsing is "00000000 101101101010110010010110", in this sequence, the binary representations of hours (28th to 24th bit), minutes (23rd to 18th bit) and seconds (17th to 12th bit) can be extracted, that is, '10010' is the binary representation of hours, which is 22 hours after conversion to decimal; '011010' is the binary representation of minutes, which is 32 minutes after conversion to decimal; '101101' is the binary representation of seconds, which is 55 seconds after conversion to decimal. Therefore, the candidate cabin voice time represented by this candidate binary subsequence is "22:32:55". Similarly, each 32-bit candidate binary subsequence is mapped to a specific time point based on the same means.

[0093] In an optional embodiment, the above step S8 corrects all the candidate cabin voice times to obtain all target cabin voice times, including:

[0094] Compare all the candidate cabin sound time pairs, and calculate the time interval between each candidate cabin sound time pair;

[0095] Determine whether each time interval is an integer multiple of a preset time difference;

[0096] If yes, then record the corresponding candidate cabin voice time pair;

[0097] The candidate cabin sound time that appears most frequently in the recording results is used as the reference time;

[0098] All the candidate cabin sound times are corrected based on the reference time and the preset time difference to obtain all the target cabin sound times.

[0099] It should be noted that after obtaining all the candidate cabin voice time, it is also necessary to consider the noise interference that may exist in the cabin voice data, so that each candidate cabin voice time may not be accurate / true. These noises may affect the time decoding of some candidate binary subsequences and cause errors. Therefore, when analyzing and using these candidate cabin voice time data, you must be vigilant about possible errors or inaccuracies. Figure 5 Shown is a flow chart of another embodiment of the method for decoding the cabin sound time signal provided by the present invention. Based on this, the embodiment of the present invention first determines the time of the first subsequence and ensures that the time of the first subsequence is true and accurate. Then, using the information that the time difference between every two adjacent subsequences is 4 seconds, the true and accurate time of the subsequent subsequences is inferred. This information is formulated by rules and can also be obtained by analyzing the spectrum graph. In other words, first ensure that the time of the first subsequence is accurate, and then use this "anchor point" combined with the fixed time difference between the subsequences to deduce the time of all subsequent subsequences. This method can effectively reduce the impact of noise interference on time decoding, thereby improving the accuracy of data analysis.

[0100] During specific implementation, first, based on known conditions (i.e., the preset time difference between each adjacent subsequence is 4 seconds), it is inferred that the time interval between any two subsequences should be 4w seconds, where w is the distance between the two subsequences. Based on this, the times of all subsequences are compared pairwise to construct a time pair list with a time difference of 4w seconds. In this large list, the time with the highest frequency of occurrence is found as the reference time. This reference time can represent the time of the mth subsequence. Then, the time of the first subsequence is inferred from this reference time. Specifically, the time of the first subsequence should be the time of the mth subsequence minus 4(m-1) seconds, which ensures the authenticity and accuracy of the time of the first subsequence. Similarly, through the preset time difference condition that the time interval between each adjacent subsequence is 4 seconds, the true and accurate time of all subsequences is derived, and the precise timing of the cabin sound data is completed. The embodiment of the present invention can effectively offset noise interference and improve the accuracy of time resolution.

[0101] See also Figure 6 , is a structural diagram of an embodiment of a decoding device for cabin sound time signal provided by the present invention.

[0102] An embodiment of the present invention provides a decoding device for a cabin sound time signal, which is used to implement the decoding method for a cabin sound time signal described in any of the above embodiments, and the device includes:

[0103] A cockpit sound data acquisition module 11 is used to acquire the cockpit sound data of the target aircraft;

[0104] A cabin sound data preprocessing module 12, used for preprocessing the cabin sound data to obtain preprocessed cabin sound data;

[0105] A short-time Fourier transform module 13 is used to perform a short-time Fourier transform on the pre-processed cabin sound data to obtain a time spectrum thereof;

[0106] A first spectrum data selection module 14 is used to select first spectrum data corresponding to all candidate cabin sound times from the time-spectrum diagram;

[0107] A first spectrum data decoding module 15, configured to decode each of the first spectrum data to obtain a corresponding candidate binary sequence;

[0108] A candidate binary subsequence selection module 16 is used to select a candidate binary subsequence from each of the candidate binary sequences based on a preset rule for representing cabin sound time signals with binary symbols;

[0109] A candidate cabin sound time acquisition module 17 is used to convert all the candidate binary subsequences based on the preset rule to obtain all candidate cabin sound times;

[0110] The candidate cabin voice time correction module 18 is used to correct all the candidate cabin voice times to obtain all target cabin voice times.

[0111] Optionally, the cabin sound data preprocessing module 12 includes:

[0112] A parameter adjustment unit is used to set the channel of the cabin sound data to mono, the bit depth to 16 bits and the sampling rate to 16000 Hz.

[0113] Optionally, the number of candidate cabin sound times is determined by the number of bright line areas in the time-spectrogram whose frequency intensity is greater than a preset intensity threshold.

[0114] Optionally, the first spectrum data selection module 14 includes:

[0115] A bright line region acquisition unit, used to acquire a bright line region whose frequency intensity is greater than a preset intensity threshold from the time-frequency spectrum in the direction of the frequency axis; wherein the bright line region is parallel to the frequency axis;

[0116] A line spectrum data acquisition unit, used to acquire line spectrum data with the largest frequency average intensity from the time-spectrum diagram in the direction of the time axis; wherein the line spectrum data is parallel to the direction of the time axis;

[0117] A vertical intersection determination unit, used to extend all the bright line areas to be perpendicular to the line spectrum data to obtain N vertical intersections; wherein N≥1;

[0118] The first spectrum data selection unit is used to select N first spectrum data from the row spectrum data based on a preset interval; wherein each of the preset intervals includes a vertical intersection; and each of the first spectrum data corresponds to a candidate cabin sound time.

[0119] Optionally, the first spectrum data decoding module 15 includes:

[0120] The threshold decoding unit is used to perform threshold decoding judgment on each of the first spectrum data to obtain a corresponding candidate binary sequence; wherein the threshold decoding judgment is to judge whether the values ​​of all frequency components in each of the first spectrum data are greater than a preset decoding threshold.

[0121] Optionally, the candidate binary subsequence selection module 16 is specifically:

[0122] A head array identification unit, used to identify a preset head array from each of the candidate binary sequences based on a preset rule for representing cabin sound time signals with binary symbols;

[0123] A tail array extraction unit, used to extract K bits of data located after the corresponding preset head array from each candidate binary sequence as a tail array; wherein K≥1;

[0124] A candidate binary subsequence selection unit is used to select M bits of data from back to front in each combination of a preset head array and a tail array as a corresponding candidate binary subsequence; wherein M>K;

[0125] Optionally, the candidate cabin voice time acquisition module 17 is specifically:

[0126] A time binary combination determination unit, configured to determine the binary combinations corresponding to the hours, minutes and seconds in all the candidate binary subsequences according to the preset rule;

[0127] The binary combination conversion unit is used to convert each of the binary combinations into decimal to obtain the corresponding candidate cabin sound time.

[0128] Optionally, the candidate cabin sound time correction module 18 includes:

[0129] A time interval calculation unit, used for comparing all the candidate cabin voice time pairs and calculating the time interval between each candidate cabin voice time pair;

[0130] A time interval determination unit, used to determine whether each time interval is an integer multiple of a preset time difference;

[0131] a candidate cabin voice time pair recording unit, configured to record the corresponding candidate cabin voice time pair if yes;

[0132] A reference time determination unit, configured to use the candidate cabin voice time with the highest frequency of occurrence in the recording result as the reference time;

[0133] The candidate cabin sound time correction unit is used to correct all the candidate cabin sound times based on the reference time and the preset time difference to obtain all the target cabin sound times.

[0134] It should be noted that the decoding device for the cabin sound time signal provided in the embodiment of the present invention can implement all the processes of the decoding method for the cabin sound time signal described in any of the above embodiments, and the functions of each module and unit in the device and the technical effects achieved are respectively the same as the functions and technical effects achieved by the decoding method for the cabin sound time signal described in the above embodiments, and will not be repeated here.

[0135] An embodiment of the present invention also provides a computer-readable storage medium, which includes a stored computer program; wherein, when the computer program is running, it controls the device where the computer-readable storage medium is located to execute the method for decoding the cabin sound time signal described in any of the above embodiments.

[0136] In summary, the embodiments of the present invention provide a method, device and storage medium for decoding a cockpit sound time signal, the method comprising: obtaining cockpit sound data of a target aircraft; preprocessing the cockpit sound data to obtain preprocessed cockpit sound data; performing short-time Fourier transform on the preprocessed cockpit sound data to obtain its time spectrum diagram; selecting first spectrum data corresponding to all candidate cockpit sound times from the time spectrum diagram; decoding each first spectrum data to obtain a corresponding candidate binary sequence; selecting a candidate binary subsequence from each candidate binary sequence based on a preset rule that binary symbols represent cockpit sound time signals; converting all candidate binary subsequences based on preset rules to obtain all candidate cockpit sound times; correcting all candidate cockpit sound times to obtain all target cockpit sound times. The present invention can accurately obtain and decode cockpit sound time signals under the condition of frequency loss to achieve accurate timing of CVR data, thereby improving the accuracy and reliability of data, which is of great value to aviation safety and aircraft maintenance. In addition, the present invention not only improves the reliability of data recovery, but also enhances the robustness of the system, so that it can still maintain efficient decoding performance in a complex and uncertain environment. At the same time, the present invention has the advantages of strong practicability, simple operation, wide adaptability, etc., and can be widely used in the civil aviation industry, playing an important role in improving flight safety and preventing and reducing flight accidents.

[0137] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for decoding a cabin sound time signal, characterized in that: include: Acquire the target aircraft's cabin sound data; Preprocessing the cabin sound data to obtain preprocessed cabin sound data; Performing short-time Fourier transform on the preprocessed cabin sound data to obtain a time spectrum diagram thereof; Selecting first spectrum data corresponding to all candidate cabin sound times from the time-spectrogram; Decoding each of the first spectrum data to obtain a corresponding candidate binary sequence; Selecting a candidate binary subsequence from each of the candidate binary sequences based on a preset rule for representing cabin sound time signals with binary symbols; Convert all the candidate binary subsequences based on the preset rule to obtain all candidate cabin sound times; Correcting all the candidate cabin sound times to obtain all target cabin sound times; The step of selecting first spectrum data corresponding to all candidate cabin sound times from the time-spectrogram comprises: Acquire a bright line region whose frequency intensity is greater than a preset intensity threshold from the time-frequency spectrum in the direction of the frequency axis; wherein the bright line region is parallel to the frequency axis; Acquire row spectrum data with the largest frequency average intensity from the time-frequency spectrum diagram in the direction of the time axis; wherein the row spectrum data is parallel to the direction of the time axis; Extend all the bright line areas to be perpendicular to the line spectrum data to obtain N vertical intersection points, where N≥1; Based on a preset interval, N first spectrum data are selected from the row spectrum data; wherein each of the preset intervals includes a vertical intersection; each of the first spectrum data is obtained by intercepting the preset interval data including the corresponding vertical intersection on the row spectrum data; each of the first spectrum data corresponds to a candidate cabin sound time; The decoding of each of the first spectrum data to obtain a corresponding candidate binary sequence includes: Perform threshold decoding judgment on each of the first spectrum data to obtain a corresponding candidate binary sequence; wherein the threshold decoding judgment is to judge whether the values ​​of all frequency components in each of the first spectrum data are greater than a preset decoding threshold.

2. The method for decoding the cabin sound time signal according to claim 1, characterized in that: The preprocessing of the cabin sound data to obtain the preprocessed cabin sound data includes: The channel of the cabin sound data is set to mono, the bit depth is set to 16 bits and the sampling rate is adjusted to 16000 Hz.

3. The method for decoding the cabin sound time signal according to claim 1, characterized in that: The number of candidate cabin sound times is determined by the number of bright line areas in the time-frequency spectrum diagram whose frequency intensity is greater than a preset intensity threshold.

4. The method for decoding the cabin sound time signal according to claim 1, characterized in that: The preset rule based on binary symbols to represent the cabin sound time signal, selecting a candidate binary subsequence from each of the candidate binary sequences, is specifically: Based on a preset rule for representing cabin sound time signals with binary symbols, identifying a preset header array from each of the candidate binary sequences; Extracting K bits of data located after the corresponding preset head array from each candidate binary sequence as a tail array; wherein K≥1; In each combination of a preset head array and a tail array, M bits of data are selected from back to front as corresponding candidate binary subsequences; wherein M>K.

5. The method for decoding the cabin sound time signal according to claim 1, characterized in that: The converting of all the candidate binary subsequences based on the preset rule to obtain all the candidate cabin sound times is specifically: According to the preset rule, determine the binary combination corresponding to the hour, minute and second in all the candidate binary subsequences; Convert each of the binary combinations into decimal to obtain the corresponding candidate cabin sound time.

6. The method for decoding the cabin sound time signal according to claim 1, characterized in that: The step of correcting all the candidate cabin voice times to obtain all target cabin voice times includes: Compare all the candidate cabin sound time pairs, and calculate the time interval between each candidate cabin sound time pair; Determine whether each time interval is an integer multiple of a preset time difference; If yes, then record the corresponding candidate cabin voice time pair; The candidate cabin sound time that appears most frequently in the recording results is used as the reference time; All the candidate cabin sound times are corrected based on the reference time and the preset time difference to obtain all the target cabin sound times.

7. A decoding device for cabin sound time signal, characterized in that: The device is used to implement the decoding method of the cabin sound time signal according to any one of claims 1 to 6, comprising: A cockpit sound data acquisition module, used to acquire the cockpit sound data of the target aircraft; A cabin sound data preprocessing module, used for preprocessing the cabin sound data to obtain preprocessed cabin sound data; A short-time Fourier transform module, used for performing a short-time Fourier transform on the pre-processed cabin sound data to obtain a time spectrum thereof; A first spectrum data selection module is used to select first spectrum data corresponding to all candidate cabin sound times from the time-spectrum diagram; A first spectrum data decoding module, used for decoding each of the first spectrum data to obtain a corresponding candidate binary sequence; A candidate binary subsequence selection module, configured to select a candidate binary subsequence from each of the candidate binary sequences based on a preset rule for representing cabin sound time signals with binary symbols; A candidate cabin sound time acquisition module, used for converting all the candidate binary subsequences based on the preset rule to obtain all candidate cabin sound times; The candidate cabin voice time correction module is used to correct all the candidate cabin voice times to obtain all target cabin voice times.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program; wherein, when the computer program is run, it controls the device where the computer-readable storage medium is located to execute the method for decoding the cabin sound time signal as described in any one of claims 1 to 6.

Images