A UAV signal recognition method based on quadratic phase difference method
Through the drone signal recognition method based on the secondary phase difference method, the problem of insufficient real-time and robustness of radio spectrum detection technology for high-speed drone signal detection in the prior art is solved, and the rapid and accurate UAV signal detection reduces hardware requirements and improves detection efficiency.
Patent Information
- Application Number
- CN202311769765.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-12-21
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2043-12-21
AI Technical Summary
The existing radio spectrum detection technology is not very real-time, has low algorithm robustness when detecting high-speed and burst drone signals, and has a low probability of detection of burst signals. The hardware circuit is complex, making it difficult to achieve rapid identification. .
The drone signal recognition method based on the quadratic phase difference method is adopted. By acquiring the phase difference of the target signal and performing clustering and secondary differential processing, combining feature parameters and threshold comparison, we can quickly identify the features of the drone. This method can be implemented in distributed computing chips, reducing hardware requirements and improving detection efficiency.
The detection of drone signals is realized in real time, which improves the detection probability of high-speed burst signals and the portability of algorithms, shortens the detection time, and improves the efficiency of signal frequency domain information calculation.
Smart Images

Figure CN117874489B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a signal processing algorithm neighborhood, and in particular to drone signal recognition based on a quadratic phase difference method. Background Art
[0002] In modern confrontation and key defense scenarios, the threat posed by drone confrontation units is becoming more and more serious. How to quickly identify and detect high-speed, sudden, and non-cooperative drone targets is one of the key issues faced in the above confrontation scenarios. Radio spectrum detection and identification is one of the effective technical paths. It aims to address the shortcomings of existing technologies in engineering implementation, such as low real-time performance, low algorithm robustness, and low probability of detecting burst signals. At the same time, existing radio spectrum detection technology has shortcomings in detecting high-speed illegal drones such as cross-country drones, such as low real-time performance, low probability of detecting burst signals, and relatively complex hardware circuits. Summary of the invention
[0003] Purpose of the invention: In order to overcome the deficiencies in the prior art, the present invention provides a method for identifying UAV signals based on the quadratic phase difference method, and proposes a pipelined quadratic phase difference method that can be implemented in a distributed computing chip, which can detect target signals in quasi-real time.
[0004] Technical solution: To achieve the above purpose, the present invention provides a UAV signal recognition method based on the quadratic phase difference method.
[0005] The following steps are involved:
[0006] E1. Obtain the target signal and perform preprocessing;
[0007] E2. Find the phase ψ of each element point of the digital sequence of the input signal i The phase difference Δψ between adjacent element points i ;
[0008] E3, the phase difference between adjacent elements Δψ i Perform clustering to obtain the result element Δθ j ;
[0009] E4, performing secondary difference calculation on the result elements of phase difference clustering to obtain the subscript values of the result elements that meet the conditions;
[0010] E5. The subscript value of the obtained result element and the set characteristic parameter A are calculated by a formula and compared with the characteristic threshold ΔY, and the num-flag of the result variable is calculated according to the comparison result to obtain the calculated num-flag;
[0011] E6. Compare the num-flag after each operation with the set recognition threshold K to determine whether data with drone characteristics appears.
[0012] Furthermore, the digital sequence of the input signal in step E2 is a series of continuous complex number sets with the sampling rate as the data rate, and the following subscripts 1, 2, 3, ..., i are set to identify the position of the element of the digital sequence of the input signal to be processed once; the i-th element in the digital sequence of the input signal is recorded as:
[0013] S i =a+bi
[0014] In the formula, a represents the real part of the signal, that is, the amplitude or strength of the signal; b represents the imaginary part of the signal, that is, the phase or initial phase of the signal; i in bi is not the element position, but the imaginary unit, i = (-1) 0.5 ;
[0015] The phase of the i-th element in the digital sequence of the input signal:
[0016]
[0017] Where atan is the inverse trigonometric function;
[0018] After calculating the phase of each element, the reserved register is used to cache the phase calculation result of the previous element, so that the phase difference Δψ between adjacent elements can be calculated. i :
[0019] Δψ i =ψ i+1 -ψ i
[0020] The phase difference Δψ between adjacent elements is obtained i Sort by size and store in internal registers.
[0021] Furthermore, the phase difference Δψ between adjacent elements in E3 is i Perform clustering processing; set the phase difference deviation threshold ΔP, start from the phase difference Δψ1 between the first adjacent elements, and calculate the difference Δψ between the phase difference between the first adjacent element and the phase difference between all subsequent adjacent elements i -Δψ1; and the difference Δψ i -Δψ1 divided by Δψ1 is compared with the phase difference deviation threshold ΔP. The specific comparison is as follows:
[0022]
[0023] Then it will be from Δψ1 to Δψ i Multiply all elements of by (i-1):
[0024] Δψ1×(i-1),
[0025] Δψ2×(i-1),
[0026] ...,
[0027] Δψ i ×(i-1);
[0028] From Δψ i+1 The elements that start from it are not changed;
[0029]
[0030] Then it will be from Δψ1 to Δψ i All elements of the are unchanged.
[0031] Furthermore, the phase difference Δψ between the i+1th adjacent elements is i+1 First, find the difference Δψ between the phase difference between the i+1th adjacent element and the phase difference between all subsequent adjacent elements k -Δψ i+1 ; and the difference Δψ is obtained k -Δψ i+1 Divide by Δψ i+1 Compared with the phase difference deviation threshold ΔP, the specific comparison is as follows:
[0032]
[0033] Then from Δψ i+1 to Δψ k Multiply all elements of by (k-i+1):
[0034] Δψ i+1 ×(k-i+1),
[0035] Δψ i+2 ×(k-i+1),
[0036] ...,
[0037] Δψ k ×(k-i+1);
[0038]
[0039] Then from Δψ i+1 to Δψ k All elements of remain unchanged;
[0040] And so on, until the last element of the phase difference between all adjacent elements is calculated; after the above operation, the result element of phase difference clustering is obtained, and is recorded as Δθ1, Δθ2, Δθ3, ... Δθ j , ...ΔθM .
[0041] Furthermore, in step E4, the result elements of the phase difference clustering are subjected to secondary difference calculation; starting from Δθ1, the difference between the next result element and the previous result element is calculated in sequence:
[0042] Δθ2-Δθ1=Δθ 21
[0043] Δθ3-Δθ2=Δθ 32
[0044] …
[0045] Δθ j+1 -Δθ j =Δθ (j+1)j
[0046] Set the phase secondary difference threshold ΔP", and calculate the difference Δθ 21 , Δθ 32 , Δθ 43 , ...Δθ j(j-1) The absolute value of is compared with the phase secondary difference threshold ΔP”:
[0047] |Δθ (j+1)j |>ΔP”
[0048] The difference Δθ that exceeds the phase secondary difference threshold ΔP” (j+1)j The subscript values j+1 and j are recorded;
[0049] |Δθ (j+1)j |≤ΔP”
[0050] Then the difference Δθ (j+1)j Discard and do not record the index values j+1 and j;
[0051] At the same time, there may be multiple positions exceeding the phase secondary difference threshold ΔP", and only the next subscript value j is retained for each subscript value j+1 and j that meet the conditions; and it is cached in the over-threshold secondary difference subscript register, recorded as j1, j2, j3, ...j k .
[0052] Furthermore, in the step E5, a characteristic parameter A is set, and the characteristic parameter A is related to the signal characteristics of the drone itself and parameters such as the sampling rate; the subscript values cached in the threshold secondary difference subscript register are subtracted before and after, and the absolute value is obtained, which is recorded as:
[0053] abs(j k -j k-1 )=|j k -j k-1 |
[0054] The difference between the above calculation result and the characteristic parameter A is divided by the characteristic parameter A and then compared with the characteristic threshold ΔY. The specific comparison is as follows:
[0055]
[0056] At the beginning, an identification result variable num-flag with an initial value of 0 is preset. If the above formula is satisfied, 1 is accumulated on the identification result variable num-flag: num-flag = num-flag + 1;
[0057]
[0058] Then the num-flag identifying the result variable does not change;
[0059] Get the num-flag after operation.
[0060] Furthermore, in the step E6, a recognition threshold K is set, and the num-flag after each operation is compared with the recognition threshold K. When the value of num-flag after each operation exceeds the recognition threshold K, it indicates that data that meets the characteristics of a drone has appeared; when the value of num-flag after each operation does not exceed the recognition threshold K, it indicates that data that does not meet the characteristics of a drone has appeared.
[0061] Furthermore, the number N of stored original data is used as a key element of loop control, and the size of N is determined according to the computing power of the processor used, the memory, and the size of the external memory; after N elements have been operated on by the algorithm, the next operation of N elements can be received to implement pipeline processing of the signal; but when the accumulated elements exceed 2, the loop of the entire algorithm can be started, and there is no need to wait until N elements are accumulated before starting the algorithm.
[0062] Beneficial effects: The UAV signal recognition method based on the quadratic phase difference method of the present invention can realize quasi-real-time detection of the target signal spectrum characteristics, has low requirements for hardware storage resources, has a high probability of detecting high-speed burst signals, and has strong algorithm portability; it can effectively shorten the time for detecting UAVs, and can also be applied to the extraction of target signal parameters in the sorting and identification of communication signals for electronic countermeasures, thereby improving the efficiency of calculating and obtaining relevant signal frequency domain information.
[0063] The present invention takes pipelined secondary phase difference as its main technical feature, performs two phase difference operations on the bursty UAV communication signal after sampling, and simultaneously performs multiple threshold comparisons, optimizes the intermediate data and processing flow, and improves the detection probability and resource utilization. BRIEF DESCRIPTION OF THE DRAWINGS
[0064] Attached Figure 1Flowchart of drone signal recognition. DETAILED DESCRIPTION
[0065] The present invention will be further described below in conjunction with the accompanying drawings.
[0066] As attached Figure 1 As shown, a method for identifying UAV signals based on a quadratic phase difference method is characterized by:
[0067] The following steps are involved:
[0068] E1. Obtain the target signal and perform preprocessing;
[0069] E2. Find the phase ψ of each element point of the digital sequence of the input signal i The phase difference Δψ between adjacent element points i ;
[0070] E3, the phase difference between adjacent elements Δψ i Perform clustering to obtain the result element Δθ j ;
[0071] E4, performing secondary difference calculation on the result elements of phase difference clustering to obtain the subscript values of the result elements that meet the conditions;
[0072] E5. The subscript value of the obtained result element and the set characteristic parameter A are calculated by a formula and compared with the characteristic threshold ΔY, and the num-flag of the result variable is calculated according to the comparison result to obtain the calculated num-flag;
[0073] E6. Compare the num-flag after each operation with the set recognition threshold K to determine whether data with drone characteristics appears.
[0074] The digital sequence of the input signal in step E2 is a series of continuous complex sets with the sampling rate as the data rate. The following subscripts 1, 2, 3, ..., i are set to identify the position of the element of the digital sequence of the input signal to be processed once; the i-th element in the digital sequence of the input signal is recorded as:
[0075] S i =a+bi
[0076] In the formula, a represents the real part of the signal, that is, the amplitude or strength of the signal; b represents the imaginary part of the signal, that is, the phase or initial phase of the signal; i in bi is not the element position, but the imaginary unit, i = (-1) 0.5 ;
[0077] The phase of the i-th element in the digital sequence of the input signal:
[0078]
[0079] Where atan is the inverse trigonometric function;
[0080] After calculating the phase of each element, the reserved register is used to cache the phase calculation result of the previous element, so that the phase difference Δψ between adjacent elements can be calculated. i :
[0081] Δψ i =ψ i+1 -ψ i
[0082] The phase difference Δψ between adjacent elements is obtained i Sort by size and store in internal registers, such as RAM; the phase difference Δψ between adjacent elements of the cache i The number is determined based on the specific drone signal characteristics.
[0083] The phase difference Δψ between adjacent elements in E3 i Perform clustering processing; set the phase difference deviation threshold ΔP, start from the phase difference Δψ1 between the first adjacent elements, and calculate the difference Δψ between the phase difference between the first adjacent element and the phase difference between all subsequent adjacent elements i -Δψ1; and the difference Δψ i -Δψ1 divided by Δψ1 is compared with the phase difference deviation threshold ΔP. The specific comparison is as follows:
[0084]
[0085] Then it will be from Δψ1 to Δψ i Multiply all elements of by (i-1):
[0086] Δψ1×(i-1),
[0087] Δψ2×(i-1),
[0088] ...,
[0089] Δψ i ×(i-1);
[0090] From Δψ i+1 The elements that start from it are not changed;
[0091]
[0092] Then it will be from Δψ1 to Δψ i All elements of the are unchanged.
[0093] Then, from the phase difference Δψ between the i+1th adjacent elements i+1Start, that is, if in the above formula, the difference between the i=5th element and the first element divided by Δψ1 is less than the phase difference deviation threshold ΔP, then i=5, i+1=6; calculate the difference Δψ between the phase difference between the i+1th adjacent element and the phase difference between all subsequent adjacent elements k -Δψ i+1 ; and the difference Δψ is obtained k -Δψ i+1 Divide by Δψ i+1 Compared with the phase difference deviation threshold ΔP, the phase difference Δψ between adjacent elements k is the last phase difference, the specific comparison is as follows:
[0094]
[0095] Then from Δψ i+1 to Δψ k Multiply all elements of by (k-i+1):
[0096] Δψ i+1 ×(k-i+1),
[0097] Δψ i+2 ×(k-i+1),
[0098] ...,
[0099] Δψ k ×(k-i+1);
[0100]
[0101] Then from Δψ i+1 to Δψ k All elements of remain unchanged;
[0102] And so on, until the last element of the phase difference between all adjacent elements is calculated; after the above operation, the result element of phase difference clustering is obtained, and is recorded as Δθ1, Δθ2, Δθ3, ... Δθ j , ...Δθ M .
[0103] In the step E4, the result elements of the phase difference clustering are subjected to secondary difference calculation; starting from Δθ1, the difference between the next result element and the previous result element is calculated in sequence:
[0104] Δθ2-Δθ1=Δθ 21
[0105] Δθ3-Δθ2=Δθ 32
[0106] …
[0107] Δθj+1 -Δθ j =Δθ (j+1)j
[0108] Set the phase secondary difference threshold ΔP", and calculate the difference Δθ 21 , Δθ 32 , Δθ 43 , ...Δθ j(j-1) The absolute value of is compared with the phase secondary difference threshold ΔP”:
[0109] |Δθ (j+1)j |>ΔP”
[0110] The difference Δθ that exceeds the phase secondary difference threshold ΔP” (j+1)j The subscript values j+1 and j are recorded;
[0111] |Δθ (j+1)j |≤ΔP”
[0112] Then the difference Δθ (j+1)j Discard and do not record the index values j+1 and j;
[0113] At the same time, there may be multiple positions exceeding the phase secondary difference threshold ΔP", and only the next subscript value j is retained for each subscript value j+1 and j that meet the conditions; and it is cached in the over-threshold secondary difference subscript register, recorded as j1, j2, j3, ...j k .
[0114] In the step E5, the characteristic parameter A is set, and the characteristic parameter A is related to the signal characteristics of the drone itself and parameters such as the sampling rate; the subscript values cached in the threshold secondary difference subscript register are subtracted before and after, and the absolute value is obtained, which is recorded as:
[0115] abs(j k -j k-1 )=|j k -j k-1 |
[0116] The difference between the above calculation result and the characteristic parameter A is divided by the characteristic parameter A and then compared with the characteristic threshold ΔY. The specific comparison is as follows:
[0117]
[0118] At the beginning, an identification result variable num-flag with an initial value of 0 is preset. If the above formula is satisfied, 1 is accumulated on the identification result variable num-flag: num-flag = num-flag + 1;
[0119]
[0120] Then the num-flag identifying the result variable does not change;
[0121] Get the num-flag after operation.
[0122] The setting of the characteristic parameter A is based on the signal characteristics of the drone itself and parameters such as the sampling rate. When the set characteristic parameter A is less than the set characteristic threshold ΔY after calculation, 1 is accumulated on the num-flag of the identification result variable. At the same time, when it is determined in the next step that a drone signal appears, it can be determined that the drone that appears is the drone category with the set characteristic parameters, and it can quickly identify whether the drone is an enemy or ours.
[0123] In the step E6, a recognition threshold K is set, and the num-flag after each operation is compared with the recognition threshold K. When the value of num-flag after each operation exceeds the recognition threshold K, it indicates that data that meets the characteristics of a drone has appeared; when the value of num-flag after each operation does not exceed the recognition threshold K, it indicates that data that does not meet the characteristics of a drone has appeared.
[0124] The settings of the four thresholds, namely, the phase difference deviation threshold ΔP, the phase secondary difference threshold ΔP", the characteristic threshold ΔY and the identification threshold K, can be fixed thresholds or floating thresholds; only when the identification threshold among the four thresholds is met, it is determined that the drone signal is identified; the settings of the four thresholds are set according to actual needs.
[0125] The number N of stored original data is used as the key element of loop control, and the size of N is determined according to the computing power of the processor used, the memory, and the size of the external storage. After the number N elements have been operated by the algorithm, the next number N elements can be received to implement pipeline processing of the signal. However, when the accumulated elements exceed 2, the entire algorithm loop can be started to execute the entire algorithm's recognition of the signal, and there is no need to wait until N elements are accumulated before starting the algorithm.
[0126] The application of the present invention includes the field of civil communication signal processing, involving spectrum resource optimization design, cognitive radio spectrum perception, and also includes applications in the extraction of unknown signal parameters in military electronic countermeasures.
[0127] A distributed computing chip, such as FPGA and GPU, is used. Taking the above operation inside FPGA as an example, the detection time is taken as an example. Under the parameters of working clock 300MHz, single channel, fixed frequency, processing sequence length 16K points, the single processing delay cycle is no more than 10ms. If 100 times of the above operation are taken as an accumulation frame, the time to identify the drone signal will be less than 1s. The algorithm running time is reduced, and the detection result is obtained as soon as possible.
[0128] Example
[0129] Taking the image transmission signal recognition of a certain type of drone as an example, after collecting the target signal propagating in the space, the high-speed ADC chip performs discrete sampling to obtain a digital signal; then preprocessing is performed; for example, digital down-conversion, digital filtering, and extraction and downsampling operations; at the signal data input end of this embodiment, the actual data sampling rate is 32M, and the data length to be processed each time is set to 32K. According to the sampling rate, the time length for a single processing is 1ms, and each data sample point element is a 16-bit signed complex number;
[0130] Start by obtaining the phase of each element point of the digital sequence of the collected signal and the phase difference between adjacent elements; after obtaining the phase difference between adjacent elements, cluster the phase difference, in which the phase difference deviation threshold ΔP can be set to 0.05 (5%), and calculate and compare to obtain the result element; perform secondary difference on the phase difference of the result element, and compare the difference between the next element and the previous element of the result element with the secondary difference threshold ΔP", where the secondary difference threshold ΔP" can be set to 1200. When the difference between the next element and the previous element of the result element exceeds the secondary difference threshold ΔP", record the subscript value of the difference between the next element and the previous element of the result element. One difference has two subscript values, so only the next subscript value is recorded and recorded as j1, j2, j3, ... j k and cached to the over-threshold secondary differential index register;
[0131] When identifying features, the subscript values in the thresholded quadratic difference subscript register are subtracted before and after, and the absolute value is obtained. The feature parameter A is then subtracted and divided by the feature parameter A. Finally, the result is compared with the feature threshold ΔY. Here, the feature parameter A is set to 50, and the feature threshold ΔY is set to 0.1. When the result exceeds the feature threshold 0.1, 1 is added to the num-flag of the result variable. The num-flag after the operation is compared with the recognition threshold K, and the recognition threshold K is set to 70. When the num-flag after the operation exceeds the recognition threshold K, it indicates that data that meets the drone characteristics appears, that is, it indicates that a drone signal is detected, and it is a drone that meets the drone characteristic parameters set in the algorithm. At this time, it can be indicated that a certain type of drone has been detected.
[0132] In this embodiment, the number of stored raw data, 32K, is used as the key element of loop control. When all elements complete all the above operations, the next 32K point data can be received and pipelined. The data does not need to be fully received before processing. According to the results of timing simulation, pipeline processing can be started after receiving 2-3 data, giving full play to the distributed computing advantages of FPGA; speeding up the operation of the algorithm, improving the efficiency of detecting signals, and identifying drones more quickly.
[0133] The above are preferred embodiments of the present invention. It should be pointed out that a person skilled in the art may make several improvements and modifications without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the protection scope of the present invention.
Claims
1. A method for identifying drone signals based on quadratic phase difference method, characterized in that: The following steps are involved: E1. Obtain target signal and perform preprocessing; E2. Find the phase ψ of each element point of the digital sequence of the input signal i The phase difference Δψ between adjacent element points i ; E3, the phase difference between adjacent elements Δψ i Perform clustering to obtain the result element Δθ j ; E4, performing secondary difference calculation on the result elements of phase difference clustering to obtain the subscript values of the result elements that meet the conditions; E5. The subscript value of the obtained result element and the set characteristic parameter A are calculated by a formula and compared with the characteristic threshold ΔY, and the num-flag of the result variable is calculated according to the comparison result to obtain the calculated num-flag; E6. Compare the num-flag after each operation with the set recognition threshold K to determine whether there is data with drone characteristics; The phase difference Δψ between adjacent elements in E3 i Perform clustering processing; Set the phase difference deviation threshold ΔP, start from the first phase difference Δψ1 between adjacent elements, and calculate the difference Δψ between the first phase difference between adjacent elements and all subsequent phase differences between adjacent elements i -Δψ1; and the difference Δψ i -Δψ1 divided by Δψ1 is compared with the phase difference deviation threshold ΔP. The specific comparison is as follows: Then it will be from Δψ1 to Δψ i Multiply all elements of by (i-1): Δψ1×(i-1), Δψ2×(i-1), ……, Δψ i ×(i-1); From Δψ i+1 The elements that start from it are not changed; Then it will be from Δψ1 to Δψ i All elements of remain unchanged; Then, from the phase difference Δψ between the i+1th adjacent elements i+1 First, find the difference Δψ between the phase difference between the i+1th adjacent element and the phase difference between all subsequent adjacent elements k -Δψ i+1 ; and the difference Δψ is obtained k -Δψ i+1 Divide by Δψ i+1 Compared with the phase difference deviation threshold ΔP, the specific comparison is as follows: Then from Δψ i+1 to Δψ k Multiply all elements of by (k-i+1): Δψ i+1 ×(k-i+1), Δψ i+2 ×(k-i+1), ……, Δψ k ×(k-i+1); Then from Δψ i+1 to Δψ k All elements of remain unchanged; And so on, until the last element of the phase difference between all adjacent elements is calculated; after the above operation, the result element of phase difference clustering is obtained, and is recorded as Δθ1, Δθ2, Δθ3, ... Δθ j , ...Δθ M .
2. The method for identifying drone signals based on quadratic phase difference method according to claim 1 is characterized in that: The digital sequence of the input signal in step E2 is a series of continuous complex sets with the sampling rate as the data rate. The following subscripts 1, 2, 3, ..., i are set to identify the position of the element of the digital sequence of the input signal to be processed once; the i-th element in the digital sequence of the input signal is recorded as: With i =a+bi In the formula, a represents the real part of the signal, that is, the amplitude or strength of the signal; b represents the imaginary part of the signal, that is, the phase or initial phase of the signal; i in bi is not the element position, but the imaginary unit, i = (-1) 0.5 ; The phase of the i-th element in the digital sequence of the input signal: Where atan is the inverse trigonometric function; After calculating the phase of each element, the reserved register is used to cache the phase calculation result of the previous element, so that the phase difference Δψ between adjacent elements can be calculated. i : Dp i =ψ i+1 -ψ i The phase difference Δψ between adjacent elements is obtained i Sort by size and store in internal registers.
3. The method for identifying drone signals based on quadratic phase difference method according to claim 1 is characterized in that: In the step E4, the result elements of the phase difference clustering are subjected to secondary difference calculation; starting from Δθ1, the difference between the next result element and the previous result element is calculated in sequence: Δθ2-Δθ1=Δθ 21 Δθ3-Δθ2=Δθ 32 …… Dth j+1 -Dth j =Δθ (j+1)j Set the phase secondary difference threshold ΔP", and calculate the difference Δθ 21 , Δθ 32 , Δθ 43 , ...Δθ j(j-1) The absolute value of is compared with the phase secondary difference threshold ΔP”: |Δθ (j+1)j |>ΔP” The difference Δθ that exceeds the phase secondary difference threshold ΔP” (j+1) The subscript values j+1 and j are recorded; |Dth (j+1)j |≤ΔP” Then the difference Δθ (j+1) Discard and do not record the index values j+1 and j; At the same time, there may be multiple positions exceeding the phase secondary difference threshold ΔP", and only the next subscript value j is retained for each subscript value j+1 and j that meet the conditions; and it is cached in the over-threshold secondary difference subscript register, recorded as j1, j2, j3, ...j k .
4. The method for identifying drone signals based on quadratic phase difference method according to claim 3 is characterized in that: In the step E5, the characteristic parameter A is set, and the characteristic parameter A is related to the signal characteristics and sampling rate parameters of the drone itself; the subscript values cached in the threshold secondary difference subscript register are subtracted before and after, and the absolute value is obtained, which is recorded as: abs(j k -j k-1 )=|j k -j k-1 | The difference between the above calculation result and the characteristic parameter A is divided by the characteristic parameter A and then compared with the characteristic threshold ΔY. The specific comparison is as follows: At the beginning, an identification result variable num-flag with an initial value of 0 is preset. If the above formula is satisfied, 1 is accumulated on the identification result variable num-flag: num-flag = num-flag + 1; Then the num-flag identifying the result variable does not change; Get the num-flag after operation.
5. The method for identifying drone signals based on the quadratic phase difference method according to claim 4 is characterized in that: In the step E6, a recognition threshold K is set, and the num-flag after each operation is compared with the recognition threshold K. When the value of num-flag after each operation exceeds the recognition threshold K, it indicates that data that meets the characteristics of a drone has appeared; when the value of num-flag after each operation does not exceed the recognition threshold K, it indicates that data that does not meet the characteristics of a drone has appeared.
6. The method for identifying drone signals based on quadratic phase difference method according to claim 5 is characterized in that: The number of stored raw data N is used as the key element of loop control, and the size of N is determined according to the computing power, memory and external storage size of the processor used; When N elements have been processed by the algorithm, the next N elements are processed and pipeline processing is implemented for the signal; but when the accumulated elements exceed 2, the entire algorithm loop is started, and there is no need to wait until N elements are accumulated before starting the algorithm.
