Radar scanning mode discrimination method based on multiple features and convex hull anomaly detection

Abstract

The invention discloses a radar scanning mode discrimination method based on multiple features and convex hull anomaly detection, which comprises the following steps: firstly, based on a radar radiation source pattern TOA-PA sequence intercepted by a reconnaissance receiver, extracting three features with remarkable discrimination by utilizing the difference characteristic of pulse amplitude: slope change point number, symmetry ratio and amplitude continuity; and constructing a three-dimensional feature vector. And then, a convex hull algorithm is adopted to train and recognize a judgment area, and an anomaly detection framework is embedded, so that accurate classification of mechanical scanning and phased array scanning radars is realized.

Description

Technical Field

[0001] This invention belongs to the field of radar technology, specifically relating to a radar scanning mode discrimination method based on multi-feature and convex hull anomaly detection. Background Technology

[0002] Radar source identification is a core component of modern electronic reconnaissance and intelligence analysis. Its core task is to accurately distinguish the identity, operational status, and threat level of various radiation sources from intercepted pulse streams. Among these, the scan type, as a key attribute of a radar radiation source, directly reflects the radar's tactical intentions and operational modes, including search, tracking, and guidance, and is a crucial basis for high-level situational awareness and intent interpretation. Therefore, rapid, accurate, and robust identification of radar scanning methods is of paramount importance for gaining battlefield initiative and achieving precise electronic countermeasures in complex electromagnetic environments.

[0003] Radar scanning types can be broadly categorized into mechanical scanning and electronic scanning. Electronic reconnaissance receivers acquire pulse amplitude information by intercepting signals from radar radiation sources, and the changing characteristics of this pulse amplitude reflect the radar's scanning mode. However, in real electromagnetic environments, radar radiation source signals are often affected by non-ideal factors such as pulse loss and noise pulses, posing a significant challenge to scanning type identification. Due to the differences in amplitude envelopes generated by the antenna pattern scanning process and spatial location, and considering the regularity of beam scanning for the same radiation source, the Time of Arrival (TOA) and Pulse Amplitude (PA) parameters are typically introduced to characterize the antenna scanning information of the radar radiation source. Existing radar radiation source scanning type identification methods mainly involve two steps: first, extracting features from the TOA-PA of the radiation source, such as main beam flatness and discretized spectrum characteristics; second, using different classifiers to identify the radar radiation source scanning type. Traditional classifiers largely rely on domain prior knowledge and expert experience to set thresholds, which easily introduces human error. When parameter ranges overlap, scanning modes switch frequently, or factors such as noise and pulse loss affect the identification accuracy, it significantly decreases. Currently, deep learning models such as residual networks, long short-term memory networks, convolutional neural networks, and graph neural networks have been applied to scan type recognition. However, these methods still face problems such as dependence on large-scale data, insufficient generalization ability, and weak interpretability in complex electromagnetic environments. Therefore, there is an urgent need for a high-accuracy scan type recognition method based on signal feature analysis that can be used under non-ideal observation conditions. Summary of the Invention

[0004] To overcome the shortcomings of existing technologies, this invention provides a radar scanning mode discrimination method based on multi-feature and convex hull anomaly detection. First, based on the radar radiation source pattern TOA-PA sequence intercepted by a reconnaissance receiver, three features with significant discriminative power are extracted using the differences in pulse amplitude: the number of slope change points, the symmetry ratio, and amplitude continuity, and a three-dimensional feature vector is constructed. Subsequently, a convex hull algorithm is used to train the identification and decision region, and an anomaly detection framework is embedded to achieve accurate classification of mechanically scanned and phased array scanned radars.

[0005] The technical solution adopted by this invention to solve its technical problem is as follows: Step 1: Generating the radar radiation source pulse amplitude sequence; Step 2: Scan feature extraction; Step 3: Feature detection and convex hull construction.

[0006] Preferably, step 1 specifically comprises: First, establish a model of the radar pulse power received by the reconnaissance receiver; Assume the radar is in t A pulse signal was emitted at a distance of [time value missing] from the reconnaissance aircraft. R The pulse amplitude PA received by the reconnaissance aircraft is expressed by the received power as follows: (1) in, It is the radar transmission power. To increase the receiving antenna gain of the reconnaissance aircraft It is the radar transmitting antenna gain. It is the wavelength of the transmitted pulse signal. It is the propagation loss factor. yes The radar transmitting antenna beam is always in the direction of the reconnaissance aircraft. Normalized pattern function values ​​on, and These are the heading angle and pitch angle of the reconnaissance aircraft, respectively. Antenna radiation pattern in azimuth angle The modeling is approximated by the Sinc function form: (2) in M For the number of array elements, d The distance between array elements.

[0007] Preferably, step 2 specifically comprises: The scanning modes of mechanically scanned radar and phased array scanned radar are distinguished by utilizing the morphological characteristics of pulse amplitude. The following three features are extracted: Feature 1: Number of slope variation points (NSCP); Define pulse amplitude sequence Its first difference is divided into The number of inflection points, i.e., the number of times the sign changes, is counted as the number of slope change points: (3) in, For symbolic functions, N This represents the total number of pulses. n The pulse number; Indicates the first n The amplitude of each pulse; Feature 2: Symmetry ratio SR; Divide the amplitude sequence into left and right regions centered at the midpoint, and calculate the integral area of ​​the amplitudes on both the left and right sides respectively. and The symmetry ratio is defined as: (4) in, , ; Feature 3: Amplitude continuity AC; The smoothness of a pulse sequence is measured by the standard deviation of the amplitude difference between adjacent pulses. (5) Preferably, step 3 specifically comprises: Let the feature sample set extracted from the mechanical scanning pattern be... ,in They represent the first i The number of inflection points, symmetry ratio, and amplitude continuity characteristics of each sample are analyzed; the three-dimensional convex hull of the sample set is constructed as follows: (6) in, C It is the smallest convex polyhedron containing all mechanically scanned sample points; Indicates the first i Convex combination coefficients corresponding to each sample point; After constructing the convex hull C In space, if the new sample to be tested satisfies If the result is positive, it is determined to be mechanical scanning; otherwise, it is phased array scanning.

[0008] An electronic device includes a processor and a memory; the memory stores a computer program, and the processor executes the computer program stored in the memory to enable the electronic device to perform the radar scanning discrimination method described above.

[0009] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the aforementioned radar scanning method for discrimination.

[0010] A chip includes a processor for calling and running a computer program from a memory, causing a device equipped with the chip to perform the radar scanning method discrimination method described above.

[0011] A computer program product includes a computer storage medium storing a computer program, the computer program including instructions executable by at least one processor, which, when executed by the at least one processor, implement the aforementioned radar scanning method discrimination method.

[0012] The beneficial effects of this invention are as follows: 1. Establish time-domain TOA-PA sequence models for the antenna patterns of mechanically scanned and phased array scanned radars; 2. Extract the three-dimensional decoupled features of slope change points, symmetry ratio, and amplitude continuity from the TOA-PA sequence to achieve effective characterization of the scanning features; 3. A recognition decision region is constructed based on the convex hull algorithm, and an anomaly detection framework is embedded to achieve robust scan type recognition under pulse loss and stray pulse interference conditions. Attached Figure Description

[0013] Figure 1 Flowchart of a radar scanning method for discriminative analysis based on multi-feature and convex hull anomaly detection.

[0014] Figure 2. Pulse amplitude characteristics of mechanically scanned radar under different parameters.

[0015] Figure 3 shows the pulse amplitude characteristics of the mechanical scanning pattern under non-ideal conditions. Figure 3(a) shows noise interference, Figure 3(b) shows pulse loss, and Figure 3(c) shows measurement error.

[0016] Figure 4 shows the characteristic distribution of mechanical scanning and phased array scanning. Figure 4(a) shows the number of inflection points, Figure 4(b) shows the symmetry, and Figure 4(c) shows the continuity.

[0017] Figure 5 Three-dimensional spatial distribution and convex hull construction of mechanical scanning and phased array features.

[0018] Figure 6 shows the detection results based on different feature combinations: Figure 6(a) number of inflection points + symmetry, Figure 6(b) number of inflection points + continuity, Figure 6(c) symmetry + continuity, and Figure 6(d) number of inflection points + symmetry + continuity. Detailed Implementation

[0019] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0020] This invention proposes a radar scanning mode discrimination method based on multi-feature and convex hull anomaly detection. The method first uses the radar radiation source pattern TOA-PA sequence intercepted by a reconnaissance receiver to extract three highly discriminative features based on the differences in pulse amplitude: the number of slope change points, the symmetry ratio, and amplitude continuity, constructing a three-dimensional feature vector. Subsequently, a convex hull algorithm is used to train the identification and decision region, and an anomaly detection framework is embedded to achieve accurate classification of mechanically scanned and phased array scanned radars. This method maintains high classification accuracy even under pulse loss and noise interference conditions.

[0021] Example: Step 1: Generating the radar radiation source pulse amplitude sequence; To characterize the amplitude characteristics of the received signal, a model of the radar pulse power received by the reconnaissance receiver is first established. It is assumed that the radar... t A pulse signal was emitted at a distance of [time value missing] from the reconnaissance aircraft. R The pulse amplitude PA received by the reconnaissance aircraft can be expressed by the received power as follows: (1) in, It is the radar transmission power. To increase the receiving antenna gain of the reconnaissance aircraft It is the radar transmitting antenna gain. It is the wavelength of the transmitted pulse signal. It is the propagation loss factor. yes The radar transmitting antenna beam is always in the direction of the reconnaissance aircraft. Normalized pattern function values ​​on, and These are the azimuth and pitch angles of the reconnaissance aircraft, respectively.

[0022] A typical antenna radiation pattern in the azimuth angle The modeling can be approximated as a Sinc function: (2) Figure 2 shows the amplitude characteristics of a reconnaissance receiver intercepting a mechanically scanned radar under different 3dB beamwidths (BW) and main-to-side ratios (MSR). As can be seen from the figure, the amplitude of the mechanically scanned signal pulses intercepted by the receiver fluctuates periodically with antenna rotation. Meanwhile, the actual intercepted pulse sequence inevitably contains noise pulses, lost pulses, and measurement errors, which to some extent disrupt the smoothness of the ideal radiation pattern and reduce the regularity of the mechanically scanned radar amplitude sequence, posing a challenge to the identification of the scan type. Figure 3 shows the actual curves of the antenna radiation pattern pulse amplitude sequence under certain pulse loss, noise pulse, and measurement errors.

[0023] Step 2: Scan feature extraction; Mechanically scanned radar beams are driven by mechanical rotation, resulting in slow and continuous pointing changes and smooth amplitude variations between adjacent pulses. In contrast, phased array scanned radar beams are controlled by a phased array, causing rapid beam pointing jumps and discrete amplitude variations. Therefore, the morphological characteristics of the pulse amplitude can be used to distinguish between the two scanning methods. This scheme extracts the following three features: Feature 1: Number of Slope Change Points (NSCP); Mechanical scanning produces smooth pulse amplitude curves, while phased array scanning exhibits steep pulse amplitude changes during beam switching. To characterize the "bumpiness" of the curves, a pulse amplitude sequence is defined. Its first difference is divided into The number of sign changes (inflection points) is counted as the number of slope change points: (3) in, For symbolic functions, N This represents the total number of pulses. For mechanical scans with relatively smooth patterns, Smaller, while phased array scanning is subject to frequent jitter. The number of inflection points in Figure 4(a) shows that the mechanically scanned radar has fewer inflection points in its radiation pattern, while the phased array radar has a faster beam change rate between adjacent pulses and a discontinuous amplitude, resulting in more slope change points than the mechanically scanned radar.

[0024] Feature 2: Symmetry Ratio (SR); The radiation pattern of a mechanically scanned radar is relatively symmetrical on both sides, while the radiation pattern of a phased array scanned radar has no obvious regularity. Therefore, the left and right regions are divided with the midpoint of the amplitude sequence as the center, and the integral area of ​​the amplitude on the left and right sides is calculated respectively. and The symmetry ratio is defined as: (4) in, , , N The total number of pulses is shown in Figure 4(b). The symmetry of the mechanical scan tends to be close to 1, indicating that the pattern symmetry is good. The symmetry distribution of the phased array scan is more discrete, and the ratio is less than 1, indicating that the pattern symmetry of the phased array scan is not obvious.

[0025] Feature 3: Amplitude Continuity (AC); The smoothness of a pulse sequence is measured by the standard deviation of the amplitude difference between adjacent pulses. (5) Among them, the stronger the continuity of the radiation pattern, the better. AC The smaller the value, the more severe the jitter in the pattern sequence. AC The larger the amplitude continuity, the less noticeable the jitter in the mechanical scanning pulse amplitude is. Figure 4(c) shows that the amplitude continuity is consistent with the mechanical scanning pulse amplitude. AC The amplitude is relatively small, and the amplitude jitter between adjacent pulses in phased array scanning is relatively large. AC Larger.

[0026] Step 3: Feature detection and convex hull construction; Mechanically scanned radar exhibits a clear periodicity and regularity in beam pointing changes, resulting in a stable and concentrated pulse amplitude sequence distribution in the feature space. In contrast, phased array radar relies on phase shift control to achieve rapid and flexible beam switching, leading to a discrete and dispersed pulse amplitude characteristic distribution that is difficult to fully traverse using a limited number of samples. Based on this difference, this invention utilizes mechanically scanned samples to form a reliable convex hull boundary during sample construction, transforming phased array scanning identification into an anomaly detection problem outside the convex hull.

[0027] Let the feature sample set extracted from the mechanical scanning pattern be... ,in They represent the first i The number of inflection points, symmetry ratio, and amplitude continuity characteristics of each sample are analyzed. The 3D convex hull of the sample set is constructed as follows: (6) in, C It is the smallest convex polyhedron containing all mechanically scanned sample points.

[0028] After constructing the convex hull C In space, if the new sample to be tested satisfies If the result is positive, it is determined to be mechanical scanning; otherwise, it is phased array scanning.

[0029] Step 4: Output the recognition results based on the convex hull; During the testing phase, based on the test pulse sequence generated in step 1, different parameters such as antenna beamwidth, main-sidelobe ratio, and scanning period are randomly set to make the radiation pattern sequence exhibit richer statistical characteristics. Based on the three-dimensional features extracted in step 2, the generated test sample is input into the convex hull constructed in step 3. If the sample falls within the convex hull, the recognition method is determined to be mechanical scanning; otherwise, it is phased array scanning.

[0030] Figure 5 The distribution of 1000 sets of mechanical scanning samples and 1000 sets of phased array scanning samples in the three-dimensional feature space is presented. The mechanical scanning samples exhibit a more pronounced feature clustering distribution, forming the convex hull boundary of the feature space, while the feature distribution of the phased array scanning samples is more discrete, located outside the convex hull, thus enabling effective differentiation between the two scanning methods. Figure 6 shows the distribution diagrams of pairwise feature combinations and the combined three-dimensional features, and Table 1 shows the recognition accuracy of different feature combinations. The results show that when using three-dimensional feature combinations, a clearer boundary can be formed in the feature space, with recognition accuracy exceeding 90%, indicating that the proposed features have good discriminative power in distinguishing between mechanical scanning and phased array scanning.

[0031] Table 1 Comparison of recognition accuracy for different feature combinations

Claims

1. A radar scanning mode discrimination method based on multi-feature and convex hull anomaly detection, characterized in that, The method comprises the following steps: Step 1: generating a radar radiation source pulse amplitude sequence; Step 2: scanning feature extraction; Step 3: feature detection and convex hull construction.

2. The radar scan mode discrimination method based on multi-feature and convex hull anomaly detection according to claim 1, characterized in that, The step 1 is specifically: First, a radar pulse power model received by a reconnaissance receiver is established; Suppose the radar transmits a pulse signal at t time, and the distance between the radar and the reconnaissance aircraft is R Then the pulse amplitude PA received by the reconnaissance aircraft is represented by the received power as follows: （1） wherein is the radar transmit power, is the reconnaissance aircraft receive antenna gain, is the radar transmit antenna gain, is the transmit pulse signal wavelength, is the propagation loss factor, is is the normalized pattern function value of the radar transmit antenna beam at the direction in which the reconnaissance aircraft is located at time is the normalized pattern function value of the radar transmit antenna beam at the direction in which the reconnaissance aircraft is located at time and are the azimuth angle and the elevation angle, respectively, in which the reconnaissance aircraft is located. The antenna directional diagram in azimuth is modeled as a Sinc function form: （2） wherein M is the number of array elements, d is the array element spacing.

3. The method according to claim 2, wherein, The step 2 is specifically: The morphological features of pulse amplitudes are used to distinguish two scanning modes of mechanical scanning radars and phased array scanning radars, and the following three features are extracted: Feature 1: slope change point number NSCP; Defining pulse amplitude sequence whose first difference is Counting the number of sign changes, i.e. the number of inflection points, as the number of slope changes: （3） wherein is a symbol function, N is the total number of pulses; n is the pulse number; denotes the amplitude of the n th pulse; Feature 2: symmetry ratio SR; The left and right regions are divided with the center of the amplitude sequence as the center, and the integral areas of the amplitudes on the left and right sides are calculated respectively With The symmetry ratio is defined as: （4） wherein , ; Feature 3: amplitude continuity AC; The standard deviation of adjacent pulse amplitude differences is used to measure the smoothness of the pulse sequence: （5）。 4. The method according to claim 3, characterized in that, The step 3 is specifically: Let the feature sample set extracted from the mechanical scanning pattern be wherein respectively represent the number of inflection points, the symmetry ratio and the amplitude continuity feature of the first i sample; the three-dimensional convex hull of the sample set is constructed as follows: （6） wherein, C is the smallest convex polyhedron containing all the mechanically scanned sample points; denotes the convex combination coefficient corresponding to the i th sample point. In the constructed convex hull C In the space, if the new sample to be measured satisfies If yes, it is judged as mechanical scanning, otherwise as phased array scanning.

5. An electronic device, comprising: It comprises: A processor and a memory; The memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the electronic device executes the method as claimed in any one of claims 1 to 4.

6. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the method as claimed in any one of claims 1 to 4.

7. A chip, characterized by It comprises: A processor is used to call and run a computer program from a memory, so that a device installed with the chip executes the method as claimed in any one of claims 1 to 4.

8. A computer program product, characterised in that, The computer program product comprises a computer storage medium, the computer storage medium stores a computer program, and the computer program comprises instructions executable by at least one processor, and when the instructions are executed by the at least one processor, the method as claimed in any one of claims 1 to 4 is realized.

Images