Fine feature extraction-oriented IFF signal quality evaluation classification method and device

Abstract

The invention discloses an IFF signal quality evaluation and classification method and device for fine feature extraction, and belongs to the field of digital signal processing, and the method comprises the steps: S1, outputting an initial evaluation result through a computer device after the initial evaluation of signal quality is completed, and providing a support for the subsequent fine evaluation; and S2, setting a signal quality fine evaluation program in the computer device, executing a signal quality fine evaluation process when the signal quality fine evaluation program is loaded and operated by the processor, and finally outputting an evaluation classification result by utilizing the computer device. The method has the main beneficial effects that a novel IFF signal quality evaluation and classification method is provided, the digital signal processing difficulty can be reduced, the burden of an equipment operator is relieved, the method is used as a basis for subsequent digital signal processing, and the problem that the processing effects of IFF signal fine feature extraction, target classification and the like are greatly influenced by the signal quality is well solved.

Description

Technical Field

[0001] The present invention relates to the field of digital signal processing, and more specifically, to an IFF signal quality assessment classification method and device for fine feature extraction. Background Art

[0002] The quality judgment of the existing IFF radiation source S-mode reply signal is mainly used for data bit decoding on the one hand; on the other hand, it is also used more for the evaluation and analysis of the quality of the data after interpretation, which belongs to the information level, mainly referring to the accuracy, integrity and coverage evaluation of the information after interpretation.

[0003] At present, the fine feature extraction methods of IFF signals are mainly divided into two categories: manual feature extraction and machine feature extraction. Both methods usually use simple statistics, signal-to-noise ratio filtering and other methods to pre-process the IFF signals before subsequent fine feature extraction. There is no targeted research on the IFF signal quality assessment and classification methods, and it is impossible to determine the appropriate signal with quality classification labels according to the type of fine features. In the actual electromagnetic environment, there are often non-ideal conditions such as low signal-to-noise ratio, interference overlapping signals (multipath, false or other radiation source signals, etc.), which makes the signal collected by the actual equipment unable to reach the ideal state. When the detected IFF signal is affected by one or more of the above factors, it is very easy to affect the accuracy of the signal fine feature extraction and the effect of target classification, and it also increases the difficulty of subsequent digital signal processing and the burden on equipment operators. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art and provide an IFF signal quality assessment and classification method and device for fine feature extraction. For the S-mode response signal of the IFF radiation source target, a computer device is used to evaluate and classify its signal quality, and multiple categories of typical assessment and classification results can be automatically output, providing technical support for selecting signals of different subdivided quality categories, especially providing technical support for the extraction of different fine features and target classification.

[0005] The object of the present invention is achieved through the following solutions:

[0006] An IFF signal quality assessment classification method for fine feature extraction includes the following steps:

[0007] S1, setting a signal quality preliminary evaluation program in a computer device, and executing a signal quality preliminary evaluation process when the signal quality preliminary evaluation program is loaded and run by a processor, for completing the signal quality preliminary evaluation, and outputting the preliminary evaluation result by using the computer device to provide support for subsequent fine evaluation;

[0008] S2, setting a signal quality precision evaluation program in a computer device, executing a signal quality precision evaluation process when the signal quality precision evaluation program is loaded and run by a processor, and finally outputting an evaluation classification result by using the computer device.

[0009] Furthermore, in step S1, the outputted preliminary evaluation result includes the signal quality preliminary evaluation type and the reference noise ratio.

[0010] Further, in step S1, the initial signal quality assessment types include: unknown abnormality type, serious missing type, complete with interference signal in the front segment, complete with interference signal in the back segment, complete with three segments of normal amplitude signal, incomplete with missing or interference signal in the front segment, incomplete with missing or interference signal in the back segment, incomplete with three segments of normal amplitude signal.

[0011] Further, in step S2, the evaluation classification results include: Class I represents complete high-quality signals; Class II represents incomplete high-quality signals; Class III represents aliased signals containing various types of interference; Class IV represents signals with poor quality or acquisition measurement errors.

[0012] Furthermore, the signal quality initial assessment process also includes the following sub-steps: if an unknown abnormality class or a serious missing class appears, these two classes are directly judged as Class IV signals with poor quality or acquisition measurement errors.

[0013] Furthermore, the signal quality initial assessment process specifically includes the following sub-steps:

[0014] S11, initial judgment of signal integrity: for the S-mode response signal of the IFF signal, according to the characteristics of the detected S-mode response signal, the signal length Len is used to preliminarily judge whether the signal is complete, and then the two types of preliminary classification signals, complete and incomplete, are processed separately;

[0015] S12, front and rear noise floor estimation: take partial signals from the front and rear to estimate the noise floor, the length Len_n taken is the first set value of the total length Len of the detection signal, and then calculate the variance varL and varR of the taken front and rear partial signals as the front and rear noise floor estimation values;

[0016] S13, for the initial judgment of the complete signal segment estimation, output the initial evaluation type corresponding to the complete signal: make a preliminary estimate of the signal quality, divide the signal into three segments, Sig1, Sig2, and Sig3, and calculate the means Mean1, Mean2, and Mean3 of the three segments, as well as the variances Var1, Var 2, and Var 3 respectively; then compare the variances to obtain the maximum value, and use this value as the signal estimated amplitude value Var, and then compare Var with varL and varR, and judge and output the following situations in turn:

[0017] Case 1: If Var is less than varL or varR, then the signal is directly judged as an unknown abnormality class, the output quality assessment type is Class IV, and the assessment ends;

[0018] Case 2: If it does not meet case 1, and Var is less than the second set value multiple of varL, the initial cutoff is complete and there is an interference signal in the front section;

[0019] Case 3: If both cases 1 and 2 are not met, and Var is less than the third set value multiple of varR, the initial cutoff is complete and there is an interference signal in the subsequent section;

[0020] Case 4: If it does not meet the conditions 1, 2, and 3 at the same time, it is initially judged as a complete signal with three normal amplitudes;

[0021] If it meets one of the conditions 2, 3, and 4, the reference signal-to-noise ratio Qi of the signal is calculated. The calculation process is as follows: first compare the noise floor values ​​of the front and back segments, take the smaller value as the signal noise floor estimation value VarN, and then calculate it using the following formula: Qi = 10*log10((Var-VarN) / VarN);

[0022] For the initially judged incomplete signal, an estimate is made, and the initial assessment type corresponding to the incomplete signal is output: if the signal length Len is less than the set value of the leading frame length, it is directly judged as a serious missing class, and the output quality assessment type is Class IV, and the assessment ends; otherwise, the signal quality is judged again, and the signal is estimated in segments. The signal is evenly divided into three segments, Sig1, Sig2, and Sig3, and the means Mean1, Mean2, and Mean3 of the three segments, as well as the variances Var1, Var 2, and Var 3 are calculated respectively; then the variances are compared to obtain the maximum value, and this value is used as the signal estimated amplitude value Var, and then Var is compared with varL and varR, and the following situations are judged and output in turn:

[0023] Case 5: If Var is less than varL or varR, it is directly judged as an unknown abnormality class, the output quality assessment type is Class IV, and the assessment ends;

[0024] Case 6: If it does not meet case 5, then determine the size of Var1, Var2, and Var3. When Var1 is the largest, it is initially determined to be incomplete and the front section is missing or there is an interference signal.

[0025] Case 7: If it does not meet the conditions 5 and 6 at the same time, then determine the size of Var1, Var2, and Var3. When Var3 is the largest, it is initially determined to be incomplete and the latter part is missing or there is an interference signal.

[0026] Case 8: If it does not meet the conditions 5, 6, and 7 at the same time, it is initially judged as an incomplete signal with three normal amplitudes;

[0027] If one of the situations 6, 7, and 8 is met, the reference signal-to-noise ratio Qi of the signal is calculated. The calculation process is as follows: first compare the noise floor values ​​of the front and rear segments, take the smaller value as the signal noise floor estimate VarN, and then use the following formula to calculate: Qi = 10*log10((Var-VarN) / VarN).

[0028] Further, in step S12, the first setting value can be adjusted according to an actual signal detection method, and the first setting value includes 0.3%.

[0029] Further, in step S12, the second setting value multiple in case 2 is 2 to 4, and the third setting value multiple in case 3 is 2 to 4; in step S13, the setting value is 0.8us.

[0030] Furthermore, in step S2, the signal quality precision evaluation process specifically includes the following sub-steps:

[0031] S21, low signal-to-noise ratio signal processing: further process the initial evaluation signal, first judge the reference signal-to-noise ratio Qi, if Qi<Lim_q, Lim_q is the set threshold, then the signal is judged as a low signal-to-noise ratio signal, the output quality evaluation type is Class IV, and the evaluation ends; when Qi≥Lim_q, enter the subsequent steps;

[0032] S22, the overall process of precise evaluation and classification: signal envelope acquisition, sub-pulse separation, leading frame search, data block interference signal judgment, high-quality signal evaluation and confirmation.

[0033] Furthermore, in step S22, the signal envelope obtaining, sub-pulse separation, leading frame search, data block interference signal determination, high-quality signal evaluation and confirmation specifically include the following sub-steps:

[0034] S221, obtain the signal envelope: perform FFT on the signal and find the maximum spectrum, retain the spectrum signals with BW_thd bandwidths on the left and right of the maximum spectrum for windowing, set the remaining points to zero, and then inverse transform to obtain the envelope signal, and perform normalization. The obtained envelope sequence is expressed as Envelope = (xelp1, xelp2, ..., xelp n , …, xelp Len ), n is the number of sampling points, ranging from 1 to Len;

[0035] S222, separate the signal data block sub-pulses: calculate the total number of rising edges and falling edges Tnum in the signal, record and form a position sequence Pos=(xp1, xp2, ..., xp j , …, xp Tnum) , where j is 1, 2, ..., Tnum;

[0036] S223, search for the leading frame: use the fixed pulse width of the four leading sub-pulses and the fixed arrival time difference between the sub-pulses to search; search backward from the first sub-pulse, and when the four sub-pulses that meet the leading frame conditions are found, the search ends, and the leading frame pulse position label sequence is output. If the label sequence is 1 to 4, the leading does not contain interference, and the next step is continued. Otherwise, it contains interference, and the quality assessment type is output as Class III, and the assessment ends; if the leading frame pulse is not found, the position label is set to invalid, and the next step is continued;

[0037] S224, determine the data block interference signal: determine the signal data block, according to the signal characteristics, the two ideal values ​​Pw1 and Pw2 of the sub-pulse width Pw are 0.5us and 1us respectively, and the judgment of the normal pulse width leaves a margin Tlc, and the value of the margin includes 0.2us; when the margin value is 0.2us, the sub-pulse width Pw<0.3us or Pw>1.2us, it is judged as an abnormal pulse width, skip and continue to judge the next sub-pulse, count the number of sub-pulses with normal pulse width TopNum, and at the same time, the normal sub-pulse must meet the condition that the amplitude Pa of more than half of the sampling points is greater than the threshold Lim_p, and record the rising and falling edge position sequences IndexTop and IndexBottom; if TopNum is less than the number of sub-pulses with the minimum normal pulse width TopN_min or greater than the number of sub-pulses with the maximum normal pulse width TopN_max, it is directly judged as containing interference, the output quality evaluation type is Class III, and the evaluation ends; otherwise, continue to the next step;

[0038] The following method is used to calculate TopN_min and TopN_max: according to the signal interference-free data block length Len_db and the ideal pulse width of the sub-pulse Pw1, Pw2, combined with the signal characteristics, when there is no interference signal, the number of sub-pulses with the minimum normal pulse width is: TopN_min = (Len_db + Pw2) / (Pw2*2), the number of sub-pulses with the maximum normal pulse width is: TopN_max = (Len_db + Pw1) / (Pw1*2), wherein the value of the initial judgment complete signal Len_db is determined according to the inherent length of the S long mode or S short mode data block; the initial judgment incomplete signal Len_db is obtained by further calculation, specifically including two situations: the calculation formula for the signal containing the leading frame is: Len_db = Len-0.8-Len_L-Len_R, the calculation formula for the signal without searching the leading frame is: Len_db = Len-Len_L-Len_R, wherein Len_L and Len_R are the noise lengths before and after the signal respectively;

[0039] S225, High-quality Signal Evaluation and Confirmation: To evaluate a signal as a high-quality signal, the following two conditions must be met simultaneously: First, the starting part of the signal must be a preamble frame, that is, the preamble frame pulse position tag sequence obtained by searching in step S223 is from 1 to 4; Second, it is confirmed by statistically calculating the effective data length. The lengths of the parts with signals recorded are accumulated and then compared with the total signal length to obtain the proportion of effective data Radio. The defined calculation formula is: If Radio ≥ 0.4, it indicates a high-quality valid signal. Combining the initial evaluation as complete or incomplete, the output quality evaluation type is type Ⅰ or Ⅱ. Otherwise, this data is of unknown abnormal type, and the output quality evaluation type is type Ⅳ, and the evaluation ends.

[0040] Furthermore, in step S222, the method for judging rising and falling edges is: taking Lim1 as the threshold, successively taking two adjacent points xelp n 、xelp (n+1) to judge. The condition for judging the rising edge is: xelp n < Lim1 and xelp (n+1) >= Lim1. The condition for judging the falling edge is: xelp n >= Lim1 and xelp (n+1) < Lim1.

[0041] An IFF signal quality evaluation and classification device for fine feature extraction includes a processor and a memory. A computer program is stored in the memory, and when the computer program is loaded by the processor, it executes the method described in any one of the above.

[0042] The beneficial effects of the present invention include:

[0043] The signal quality evaluation and classification method proposed by the present invention can generally achieve accurate evaluation and classification of the quality of IFF radiation source target S-mode response signals, and has good support for the extraction of various subsequent fine features in the time-frequency domain and the classification of IFF radiation source targets, and can significantly improve the effectiveness of feature extraction and the accuracy of target classification. Specifically, the present invention makes full use of the characteristics in dimensions such as time domain and frequency domain under various actual situations after signal detection, and details the evaluation and classification of different quality types of IFF signals, and preferably solves the problem that the processing effects such as fine feature extraction and target classification of IFF signals are greatly affected by signal quality.

[0044] In the existing radiation source target classification algorithm, frequency domain features are often extracted as typical fine features. Whether the signal has interference has a great influence on the quality of such feature extraction. Therefore, judging whether the expected signal contains interference is one of the purposes of the quality assessment classification of the present invention. In combination with the signal characteristics, the judgment of interference mainly uses information such as signal length, sub-pulse pulse width PW, amplitude PA and timing. At this time, quality assessment types I and II are available; in addition, typical time domain waveforms are also often used as fine features for target classification. Therefore, the judgment of signal integrity is also the purpose of the quality assessment classification of the present invention. However, since the S-mode signal data block carries coding information, it cannot be directly used as a fine feature of the radiation source. The leading frame coding method is fixed and is often used to extract fine features. Therefore, the present invention combines this information to judge whether the leading frame of the previous section is complete and contains interference, and outputs the corresponding evaluation result. At this time, quality assessment type I is available. At the same time, the present invention also combines the situation that the expected signal detection is often incomplete in practical applications, and compares the complete signal to carry out a detailed scheme design in steps. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0046] Figure 1 It is the secondary evaluation process of the method of the embodiment of the present invention;

[0047] Figure 2 This is the signal initial evaluation process of the method according to the embodiment of the present invention;

[0048] Figure 3 This is the complete signal processing flow of the initial judgment method of the embodiment of the present invention;

[0049] Figure 4 The following is a flow chart of the incomplete signal processing method according to the embodiment of the present invention;

[0050] Figure 5 The signal quality precision evaluation process of the method according to the embodiment of the present invention is shown in FIG. DETAILED DESCRIPTION

[0051] All features disclosed in all embodiments in this specification, or steps in all methods or processes implicitly disclosed, except for mutually exclusive features and / or steps, can be combined and / or expanded or replaced in any manner.

[0052] The specific implementation process of the present invention is as follows:

[0053] The present invention provides an IFF signal quality assessment and classification method and device for fine feature extraction. In a preferred embodiment, based on signal detection, it particularly relates to an IFF radiation source target S-mode reply signal quality assessment and classification method for fine feature extraction, which specifically includes the following inventive concepts: starting from the needs of IFF signal fine feature extraction and target classification, the signal quality is assessed and classified, aiming to make full use of the characteristics of the time domain, frequency domain and other dimensions in various actual situations after signal detection, and to evaluate and classify the different quality types of IFF signals in detail, and further better solve the technical problem that the processing effects of IFF signal fine feature extraction and target classification are greatly affected by signal quality.

[0054] Specifically, the present invention comprehensively and qualitatively describes the quality assessment types of signals from multiple dimensions such as signal integrity, the presence or absence of interference, and the signal-to-noise ratio. Four types of quality assessment classification types are specifically considered, namely: Category I represents complete high-quality signals, that is, signals with complete leading frames and data blocks, high signal-to-noise ratio and no interference; Category II represents incomplete high-quality signals, that is, signals with incomplete front or back segments but high signal-to-noise ratio and no interference; Category III represents aliased signals containing various types of interference; Category IV represents signals with poor quality or acquisition and measurement errors, such as severe omissions, noise, low signal-to-noise ratio, unknown anomalies and the like.

[0055] In a further embodiment, it should be noted that the S-mode response signal is uniformly referred to as a signal in the following. The method of the present invention comprises the following steps:

[0056] S1, setting a signal quality initial assessment program in a computer device, and executing a signal quality initial assessment process when the signal quality initial assessment program is loaded and run by a processor, specifically, including:

[0057] S11, initial judgment of signal integrity: based on the detected signal characteristics, firstly judge whether the signal is complete based on the signal length Len, and then process the two types of preliminary classified signals, complete and incomplete, separately;

[0058] S12, front and rear noise floor estimation: take part of the signal from the front and rear to estimate the noise floor, the length Len_n taken is 0.3% of the total length Len of the detection signal (the specific value can be adjusted according to the actual signal detection method), and then calculate the variance varL and varR of the taken front and rear partial signals as the front and rear noise floor estimation values;

[0059] S13, for the initial judgment of the complete signal segment estimation, output the initial evaluation type corresponding to the complete signal: make a preliminary estimate of the signal quality, divide the signal into three segments, Sig1, Sig2, and Sig3, and calculate the mean Mean1, Mean2, Mean3 and variance Var1, Var 2, Var 3 of the three segments respectively. 3. Then, the variance comparison is performed to obtain the maximum value, and the value is used as the signal estimated amplitude value Var. Var is then compared with varL and varR, and the following situations are judged and output in turn: Situation 1: If Var is less than varL or varR, then in this case, the signal is directly judged as an unknown abnormality class, and the output quality assessment type is Class IV, and the assessment ends; Situation 2: If it does not meet Situation 1, and Var is less than 4 times varL, the initial judgment is complete and there is an interference signal in the front section; Situation 3: If it does not meet Situations 1 and 2 at the same time, and Var is less than 4 times varR, the initial judgment is complete and there is an interference signal in the back section; Situation 4: If it does not meet Situations 1, 2, and 3 at the same time, it is initially judged to be a complete signal with normal amplitudes in three sections; If it meets one of Situations 2, 3, and 4, the signal reference signal-to-noise ratio Qi is calculated, and the calculation process is as follows: First, compare the front and rear noise floor values, take the smaller value as the noise floor estimation value VarN, and then use the following formula to calculate: Qi = 10*log10((Var-VarN) / VarN);

[0060] For the initially judged incomplete signal, an estimate is made, and the initial evaluation type corresponding to the incomplete signal is output: if the signal length Len is less than the set value of the leading frame length (0.8us), it is directly judged as a serious missing class, and the output quality evaluation type is Class IV, and the evaluation ends; otherwise, the signal quality is judged again, and the signal is estimated in segments. The signal is evenly divided into three segments, namely Sig1, Sig2, and Sig3, and the means Mean1, Mean2, and Mean3 and variances Var1, Var2, and Var3 of the three segments are calculated respectively. Then, the variances are compared to obtain the maximum value, and this value is used as the signal estimated amplitude value Var, and Var is compared with varL and varR. The following situations are judged and output in sequence: Situation 5: When the maximum value Var is less than varL or varR, it is directly judged as an unknown abnormality class, and the output quality assessment type is Class IV, and the assessment ends; Situation 6: If it does not meet Situation 5, then the size of Var1, Var2, and Var3 is judged. When Var1 is the largest, it is initially judged as incomplete and the front segment is missing or there is an interference signal; Situation 7: If it does not meet Situation 5 and 6 at the same time, then the size of Var1, Var2, and Var3 is judged. When Var3 is the largest, it is initially judged as incomplete and the rear segment is missing or there is an interference signal; Situation 8: If it does not meet Situation 5, 6, and 7 at the same time, it is initially judged as incomplete and the three-segment amplitude is normal. After the initial judgment, the reference signal-to-noise ratio Qi is calculated. The reference signal-to-noise ratio calculation method is the same as the initial judgment of the complete signal processing step.

[0061] In the above steps, the initial assessment of signal quality is completed. The initial assessment type and signal reference signal-to-noise ratio output by the initial assessment of signal quality provide important support for the subsequent fine assessment, and the initial assessment results are output by computer. The output initial assessment results include the initial assessment type of signal quality and the reference signal-to-noise ratio. The initial assessment types of signal quality include: unknown abnormality type, severe missing type, complete and interference signal in the front segment, complete and interference signal in the back segment, complete and three-segment normal amplitude signal, incomplete and missing or interference signal in the front segment, incomplete and missing or interference signal in the back segment, incomplete and three-segment normal amplitude signal. Among them, the first two categories are also the final quality assessment type IV.

[0062] S2, setting a signal quality precision assessment program in a computer device, and executing a signal quality precision assessment process when the signal quality precision assessment program is loaded and run by a processor, specifically, including:

[0063] S21, low signal-to-noise ratio signal processing: further analyze the initial evaluation signal, first judge the reference signal-to-noise ratio Qi, if Qi<Lim_q, Lim_q is the set threshold (here 7dB, can be adjusted according to actual conditions), then the signal is judged as a low signal-to-noise ratio signal, the output quality evaluation type is Class IV, and the evaluation ends; when Qi≥Lim_q, enter the subsequent steps;

[0064] S22, the overall process of precise evaluation and classification: signal envelope acquisition, sub-pulse separation, leading frame search, data block interference signal judgment, high-quality signal evaluation and confirmation;

[0065] S221, obtain the signal envelope: perform FFT on the signal and find the maximum spectrum, retain the spectrum signals with BW_thd bandwidths (here 20MHz is taken, which can be adjusted according to actual conditions) around the maximum spectrum for windowing, set the remaining points to zero, and then inverse transform to obtain the envelope signal, and perform normalization. The obtained envelope sequence is expressed as Envelope = (xelp1, xelp2, ..., xelp n , …, xelp Len ), n is the number of sampling points, ranging from 1 to Len;

[0066] S222, separate the signal data block sub-pulses: calculate the total number of rising edges and falling edges Tnum in the signal, record and form a position sequence Pos=(xp1, xp2, ..., xp j , …, xp Tnum) , where j is 1, 2, ..., Tnum. The rising and falling edge judgment method is: take Lim1 = 0.5 as the threshold, and take the two adjacent points xelp in turn. n 、xelp (n+1) To judge, the rising edge condition is: xelp n<0.5 and xelp (n+1) >=0.5, the condition for judging the falling edge is: xelp n >=0.5 and xelp (n+1) <0.5;

[0067] S223, search for the leading frame: use the fixed pulse width of the four leading sub-pulses and the fixed arrival time difference between the sub-pulses to search, search from the first sub-pulse backward, and when the four sub-pulses that meet the leading frame conditions are found, the search ends, and the leading frame pulse position label sequence is output. If the label sequence is 1 to 4, the leading does not contain interference, and the next step is continued. Otherwise, it contains interference, and the quality assessment type is output as Class III, and the assessment ends; if the leading frame pulse is not found, the position label is set to invalid, and the next step is continued;

[0068] S224, determine the data block interference signal: judge the signal data block, according to the signal characteristics, the two ideal values ​​Pw1 and Pw2 of the sub-pulse width Pw are 0.5us and 1us respectively. Due to the acquisition and measurement errors in the actual acquisition, the judgment of the normal pulse width needs to leave a certain margin Tlc (here 0.2us, which can be adjusted according to the actual situation). When the margin Tlc is 0.2us, when Pw<0.3us or Pw>1.2us, it is judged as an abnormal pulse width, skip and continue to judge the next sub-pulse, and count the number of sub-pulses of normal pulse width TopNum. At the same time, the normal sub-pulse must meet the condition that the amplitude Pa of more than half of the sampling points is greater than the threshold Lim_p (here 0.7, which can be adjusted according to the actual situation), and record the rising and falling edge position sequences IndexTop and IndexBottom. If TopNum is less than TopN_min or greater than TopN_max, it can be directly judged as containing interference, and the output quality assessment type is Class III, and the assessment ends; otherwise, continue to the next step;

[0069] Among them, the method for obtaining the accurate values ​​of TopN_min and TopN_max is of great significance for the effective judgment of data block interference signals, and the calculation methods of complete signals and incomplete signals are different. Therefore, the present invention combines the inherent characteristics of the signal and the existence of a large number of non-ideal signals such as interference overlap and incomplete detection in practical applications, and adopts the following method to calculate TopN_min and TopN_max: determined according to the signal interference-free data block length Len_db and the ideal pulse widths Pw1 and Pw2 of the sub-pulses, combined with the signal characteristics, when there is no interference signal in the signal, the number of sub-pulses with the minimum normal pulse width: TopN_min = (Len_db + Pw2) / (Pw2*2) (rounded down), the number of sub-pulses with the maximum normal pulse width: TopN_max = (Len_db + Pw1) / (Pw1*2) (rounded down), wherein the value of the initial judgment complete signal Len_db is determined according to the inherent length of the S long mode or S short mode data block. For example, the initial judgment complete signal Len_db in the S long mode is directly 112, and the S short mode is 56. The Len_db of the initially judged incomplete signal needs to be obtained through further calculation. The calculation formula for the signal containing a leading frame is: Len_db=Len-0.8-Len_L-Len_R. The calculation formula for the signal without searching for a leading frame is: Len_db=Len-Len_L-Len_R, where Len_L and Len_R are the noise lengths before and after the signal respectively.

[0070] S225, high-quality signal evaluation and confirmation: To be evaluated as a high-quality signal, the following two points must be met at the same time. First, the starting part of the signal must be a leading frame, that is, the leading frame pulse position label sequence obtained by step 3) is 1 to 4; second, it is confirmed by counting the effective data length, accumulating the length of the signal part recorded above, and then comparing it with the total signal length to obtain the effective data ratio Radio, and the calculation formula is defined as: If Radio ≥ 0.4, it means that it is a high-quality valid signal. Combined with the initial evaluation, the complete or incomplete output quality evaluation type is Class I or Class II. Otherwise, the data is an unknown abnormal type, the output quality evaluation type is Class IV, and the evaluation ends.

[0071] It is uniformly explained that the specific values ​​of the above parameters in the present invention are recommended values ​​and can be appropriately adjusted according to actual conditions.

[0072] The technical effects of the present invention are verified as follows:

[0073] On the one hand, using the civil aviation IFF target long S mode reply signal data, the signal quality preliminary evaluation and signal quality fine evaluation processes are set up in the computer device to verify the proposed method. Figures 1 to 5 Process shown.

[0074] Using IFF signal data of multiple civil aviation targets, a total of 4000 IFF signal data of four different signal quality categories were selected, and the number of sample IFF signals of each category was 1000. The data were input into a computer device running the method of the present invention. When the computer device was running, the method of the present invention performed batch evaluation and classification on the signal quality, recorded the evaluation and classification accuracy, and the overall classification accuracy was 0.948.

[0075] Table 1

[0076] Evaluating classification results Category I Category II Grade III Level 4 Correct number of classifications 969 1000 874 946 Total number of categories 1000 1000 1000 1000 Classification accuracy 0.969 1 0.874 0.946

[0077] On the other hand, the long S-mode reply signal data of 8 civil aviation targets were selected for training and target classification. 2000 signal samples were selected for each target, and the training and testing ratio was 8:2. By comparing the classification effects of the training test sample set 1 (original unprocessed signals) and the training test sample set 2 (signals processed by the method of the present invention) after training, the effect of the proposed method on improving the classification of IFF radiation source targets was verified. The typical signal fine features selected for target classification are spectral features. Since the main factors affecting the accuracy of spectral feature extraction are the signal-to-noise ratio and whether there is interference, the proposed method is used. Figures 1 to 5 After the process shown, the signals with quality assessment types of Class I and Class II are selected as training test set 2. The classification accuracy test results are recorded.

[0078] Table 2

[0079] Target number 1 2 3 4 5 6 7 8 average value illustrate Correct rate 1 0.932 0.351 0.798 0.504 0.948 0.716 0.959 0.888 0.762 Sample Set 1 Correct rate 2 0.95 0.859 0.9 0.874 0.994 0.929 0.966 0.997 0.934 Sample Set 2

[0080] The above two types of verification results show that the method proposed in the present invention can realize the accurate evaluation and classification of the quality of the S-mode reply signal of the IFF radiation source target, and can greatly improve the effectiveness of the subsequent fine feature extraction and the accuracy of target classification.

[0081] The units involved in the embodiments of the present invention may be implemented by software or hardware, and the units described may also be arranged in a processor. The names of these units do not, in some cases, limit the units themselves.

[0082] According to one aspect of an embodiment of the present invention, a computer program product or a computer program is provided, the computer program product or the computer program includes a computer instruction, and the computer instruction is stored in a computer-readable storage medium. A processor of a computer device reads the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the method provided in the above various optional implementations.

[0083] As another aspect, an embodiment of the present invention further provides a computer-readable medium, which may be included in the electronic device described in the above embodiment; or may exist independently without being assembled into the electronic device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by an electronic device, the electronic device implements the method described in the above embodiment.

Claims

1. An IFF signal quality assessment and classification method for fine feature extraction, characterized in that: The following steps are involved: S1, setting a signal quality preliminary evaluation program in a computer device, and executing a signal quality preliminary evaluation process when the signal quality preliminary evaluation program is loaded and run by a processor, for completing the signal quality preliminary evaluation, and outputting the preliminary evaluation result by using the computer device to provide support for subsequent fine evaluation; S2, setting a signal quality precision evaluation program in a computer device, executing a signal quality precision evaluation process when the signal quality precision evaluation program is loaded and run by a processor, and finally outputting an evaluation classification result by using the computer device.

2. The IFF signal quality assessment classification method for fine feature extraction according to claim 1 is characterized in that: In step S1, the output preliminary evaluation result includes the signal quality preliminary evaluation type and the reference signal-to-noise ratio.

3. The IFF signal quality assessment classification method for fine feature extraction according to claim 2 is characterized in that: In step S1, the initial signal quality assessment types include: unknown abnormality type, serious missing type, complete with interference signal in the front segment, complete with interference signal in the back segment, complete with three segments of normal amplitude signal, incomplete with missing or interference signal in the front segment, incomplete with missing or interference signal in the back segment, incomplete with three segments of normal amplitude signal.

4. The IFF signal quality assessment classification method for fine feature extraction according to claim 3 is characterized in that: In step S2, the evaluation classification results include: Class I represents complete high-quality signals; Class II represents incomplete high-quality signals; Class III represents mixed signals containing various types of interference; Class IV represents signals with poor quality or acquisition measurement errors.

5. The IFF signal quality assessment classification method for fine feature extraction according to claim 4 is characterized in that: The initial signal quality assessment process also includes the following sub-steps: If an unknown abnormality or severe missing signal appears, these two types will be directly judged as Class IV signals with poor quality or incorrect acquisition and measurement.

6. The IFF signal quality assessment classification method for fine feature extraction according to claim 4 is characterized in that: The signal quality initial assessment process specifically includes the following sub-steps: S11, initial judgment of signal integrity: for the S-mode response signal of the IFF signal, according to the characteristics of the detected S-mode response signal, the signal length Len is used to preliminarily judge whether the signal is complete, and then the two types of preliminary classification signals, complete and incomplete, are processed separately; S12, front and rear noise floor estimation: take partial signals from the front and rear to estimate the noise floor, the length Len_n taken is the first set value of the total length Len of the detection signal, and then calculate the variance varL and varR of the taken front and rear partial signals as the front and rear noise floor estimation values; S13, for the initial judgment of the complete signal segment estimation, output the initial evaluation type corresponding to the complete signal: make a preliminary estimate of the signal quality, divide the signal into three segments, Sig1, Sig2, and Sig3, and calculate the means Mean1, Mean2, and Mean3 of the three segments, as well as the variances Var1, Var 2, and Var 3 respectively; then compare the variances to obtain the maximum value, and use this value as the signal estimated amplitude value Var, and then compare Var with varL and varR, and judge and output the following situations in turn: Case 1: If Var is less than varL or varR, then the signal is directly judged as an unknown abnormality class, the output quality assessment type is Class IV, and the assessment ends; Case 2: If it does not meet case 1, and Var is less than the second set value multiple of varL, the initial cutoff is complete and there is an interference signal in the front section; Case 3: If both cases 1 and 2 are not met, and Var is less than the third set value multiple of varR, the initial cutoff is complete and there is an interference signal in the subsequent section; Case 4: If it does not meet the conditions 1, 2, and 3 at the same time, it is initially judged as a complete signal with three normal amplitudes; If it meets one of the conditions 2, 3, and 4, the reference signal-to-noise ratio Qi of the signal is calculated. The calculation process is as follows: first compare the noise floor values ​​of the front and back segments, take the smaller value as the signal noise floor estimation value VarN, and then calculate it using the following formula: Qi = 10*log10((Var-VarN) / VarN); For the initially judged incomplete signal, an estimate is made, and the initial assessment type corresponding to the incomplete signal is output: if the signal length Len is less than the set value of the leading frame length, it is directly judged as a serious missing class, and the output quality assessment type is Class IV, and the assessment ends; otherwise, the signal quality is judged again, and the signal is estimated in segments. The signal is evenly divided into three segments, Sig1, Sig2, and Sig3, and the means Mean1, Mean2, and Mean3 of the three segments, as well as the variances Var1, Var 2, and Var 3 are calculated respectively; then the variances are compared to obtain the maximum value, and this value is used as the signal estimated amplitude value Var, and then Var is compared with varL and varR, and the following situations are judged and output in turn: Case 5: If Var is less than varL or varR, it is directly judged as an unknown abnormality class, the output quality assessment type is Class IV, and the assessment ends; Case 6: If it does not meet case 5, then determine the size of Var1, Var2, and Var3. When Var1 is the largest, it is initially determined to be incomplete and the front section is missing or there is an interference signal. Case 7: If it does not meet the conditions 5 and 6 at the same time, then determine the size of Var1, Var2, and Var3. When Var3 is the largest, it is initially determined to be incomplete and the latter part is missing or there is an interference signal. Case 8: If it does not meet the conditions 5, 6, and 7 at the same time, it is initially judged as an incomplete signal with three normal amplitudes; If one of the situations 6, 7, and 8 is met, the reference signal-to-noise ratio Qi of the signal is calculated. The calculation process is as follows: first compare the noise floor values ​​of the front and rear segments, take the smaller value as the signal noise floor estimate VarN, and then use the following formula to calculate: Qi = 10*log10((Var-VarN) / VarN).

7. The IFF signal quality assessment and classification method for fine feature extraction according to claim 6 is characterized in that: In step S12, the first setting value can be adjusted according to an actual signal detection method, and the first setting value includes 0.3%.

8. The IFF signal quality assessment and classification method for fine feature extraction according to claim 6 is characterized in that: In step S12, the second setting value multiple in case 2 is 2 to 4, and the third setting value multiple in case 3 is 2 to 4; in step S13, the setting value is 0.8us.

9. The IFF signal quality assessment and classification method for fine feature extraction according to claim 4, characterized in that: In step S2, the signal quality precision evaluation process specifically includes the following sub-steps: S21, low signal-to-noise ratio signal processing: further process the initial evaluation signal, first judge the reference signal-to-noise ratio Qi, if Qi<Lim_q, Lim_q is the set threshold, then the signal is judged as a low signal-to-noise ratio signal, the output quality evaluation type is Class IV, and the evaluation ends; when Qi≥Lim_q, enter the subsequent steps; S22, the overall process of precise evaluation and classification: signal envelope acquisition, sub-pulse separation, leading frame search, data block interference signal judgment, high-quality signal evaluation and confirmation.

10. The IFF signal quality assessment and classification method for fine feature extraction according to claim 9, characterized in that: In step S22, the signal envelope acquisition, sub-pulse separation, leading frame search, data block interference signal judgment, high-quality signal evaluation and confirmation specifically include the following sub-steps: S221, obtain the signal envelope: perform FFT on the signal and find the maximum spectrum, retain the spectrum signals with BW_thd bandwidths on the left and right of the maximum spectrum for windowing, set the remaining points to zero, and then inverse transform to obtain the envelope signal, and perform normalization. The obtained envelope sequence is expressed as Envelope = (xelp1, xelp2, ..., xelp n , …, xelp Len ), n is the number of sampling points, ranging from 1 to Len; S222, separate the signal data block sub-pulses: calculate the total number of rising edges and falling edges Tnum in the signal, record and form a position sequence Pos=(xp1, xp2, ..., xp j , …, xp Tnum) , where j is 1, 2, ..., Tnum; S223, search for the leading frame: use the fixed pulse width of the four leading sub-pulses and the fixed arrival time difference between the sub-pulses to search; search backward from the first sub-pulse, and when the four sub-pulses that meet the leading frame conditions are found, the search ends, and the leading frame pulse position label sequence is output. If the label sequence is 1 to 4, the leading does not contain interference, and the next step is continued. Otherwise, it contains interference, and the quality assessment type is output as Class III, and the assessment ends; if the leading frame pulse is not found, the position label is set to invalid, and the next step is continued; S224, determine the data block interference signal: determine the signal data block, according to the signal characteristics, the two ideal values ​​Pw1 and Pw2 of the sub-pulse width Pw are 0.5us and 1us respectively, and the judgment of the normal pulse width leaves a margin Tlc, and the value of the margin includes 0.2us; when the margin value is 0.2us, the sub-pulse width Pw<0.3us or Pw>1.2us, it is judged as an abnormal pulse width, skip and continue to judge the next sub-pulse, count the number of sub-pulses with normal pulse width TopNum, and at the same time, the normal sub-pulse must meet the condition that the amplitude Pa of more than half of the sampling points is greater than the threshold Lim_p, and record the rising and falling edge position sequences IndexTop and IndexBottom; if TopNum is less than the number of sub-pulses with the minimum normal pulse width TopN_min or greater than the number of sub-pulses with the maximum normal pulse width TopN_max, it is directly judged as containing interference, the output quality evaluation type is Class III, and the evaluation ends; otherwise, continue to the next step; The following method is used to calculate TopN_min and TopN_max: according to the signal interference-free data block length Len_db and the ideal pulse width of the sub-pulse Pw1, Pw2, combined with the signal characteristics, when there is no interference signal, the number of sub-pulses with the minimum normal pulse width is: TopN_min = (Len_db + Pw2) / (Pw2*2), the number of sub-pulses with the maximum normal pulse width is: TopN_max = (Len_db + Pw1) / (Pw1*2), wherein the value of the initial judgment complete signal Len_db is determined according to the inherent length of the S long mode or S short mode data block; the initial judgment incomplete signal Len_db is obtained by further calculation, specifically including two situations: the calculation formula for the signal containing the leading frame is: Len_db = Len-0.8-Len_L-Len_R, the calculation formula for the signal without searching the leading frame is: Len_db = Len-Len_L-Len_R, wherein Len_L and Len_R are the noise lengths before and after the signal respectively; S225, high-quality signal evaluation and confirmation: To be evaluated as a high-quality signal, the following two points must be met at the same time: first, the starting part of the signal must be a leading frame, that is, the leading frame pulse position label sequence searched in step S223 is 1 to 4; second, it is confirmed by counting the effective data length, accumulating the length of the recorded signal part, and then comparing it with the total signal length to obtain the effective data ratio Radio, and the calculation formula is defined as: If Radio ≥ 0.4, it means that it is a high-quality valid signal. Combined with the initial evaluation, the complete or incomplete output quality evaluation type is Class I or Class II. Otherwise, the data is an unknown abnormal type, the output quality evaluation type is Class IV, and the evaluation ends.

11. The IFF signal quality assessment and classification method for fine feature extraction according to claim 10, characterized in that: In step S222, the rising and falling edge determination method is as follows: taking Lim1 as the threshold, successively taking two adjacent points xelp n , xelp (n+1) to make a judgment. The condition for judging the rising edge is: xelp n < Lim1 and xelp (n+1) >= Lim1. The condition for judging the falling edge is: xelp n >= Lim1 and xelp (n+1) < Lim1.

12. An IFF signal quality assessment and classification device for fine feature extraction, characterized in that: The method comprises a processor and a memory, wherein a computer program is stored in the memory, and when the computer program is loaded by the processor, the method according to any one of claims 1 to 11 is executed.

Images