A conical scan based antenna tracking alignment method and system
By constructing real-time residual mean, intertemporal jump degree and elite stability, the interference of radar cross-section scintillation on the Tianying optimization algorithm is quantified. By adopting signal quality scoring decision, the problem of the Tianying optimization algorithm's tracking accuracy decreases under random radar cross-section scintillation is solved, and continuous accurate tracking and beam alignment are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHAANXI TURN ELECTRONICS TECH
- Filing Date
- 2026-05-25
- Publication Date
- 2026-08-04
AI Technical Summary
When the target radar cross-section randomly flickers, the fixed phase switching mechanism and fixed strategy activation probability of the Tianying optimization algorithm cannot respond to the non-stationary distortion of the error signal, resulting in a decrease in tracking accuracy.
By constructing real-time residual mean, intertemporal jump degree and elite stability, the interference degree of radar cross-section scintillation on the fitness evaluation of the Skyhawk optimization algorithm is quantified. The decision basis is fused with signal quality score, and the adaptive switching threshold and discrimination threshold are updated with historical statistics to cut off the elite pollution effect and achieve continuous and accurate tracking and beam alignment.
In scenarios where random scintillation of the radar cross-section coexists with high-speed target maneuvering, the antenna tracking system achieves continuous and accurate tracking and beam alignment, reducing the risk of pseudo-extreme convergence and improving tracking accuracy and stability.
Smart Images
Figure CN122291942B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of radio direction finding technology. More specifically, this application relates to an antenna tracking and alignment method and system based on conical scanning. Background Technology
[0002] With the continuous expansion of application scenarios such as low-orbit satellite constellation networking, UAV formation collaboration, and shipborne mobile communication, the real-time and accurate tracking and alignment of antennas with highly dynamic moving targets has become a core technical requirement for wireless communication and telemetry and control systems. The high-speed maneuvering of targets, the coexistence of carrier vibration and complex channel noise from multiple sources place high demands on the parameter adaptability of the tracking system.
[0003] Chinese patent application CN121643883A discloses a vehicle-mounted satellite dynamic tracking device and method. The method includes: calculating the theoretical alignment angle of the antenna, driving the antenna to rotate to that angle for coarse alignment, and initiating a frame search when no valid satellite signal is acquired; adaptively switching between primary and secondary tracking modes based on real-time vehicle speed, and dynamically adjusting servo drive parameters based on the processed tracking deviation to drive the antenna to track the satellite in real time; performing real-time compensation for the antenna's tracking angle; and predicting the current pointing angle of the antenna when a satellite is lost, and performing a conical scan centered on this predicted angle to reacquire the satellite signal. The method employs an adaptive hybrid tracking mode switching, multi-dimensional error real-time compensation, and a rapid recovery coordination mechanism combining inertial navigation prediction and conical scanning to achieve dynamic satellite tracking.
[0004] The aforementioned method still has shortcomings in the conical scanning tracking stage. For example, the target radar cross-section is affected by attitude flipping and multipath scattering interference, causing random and violent fluctuations within a timescale shorter than the scanning period. Random components unrelated to the scanning phase are superimposed in the amplitude envelope, resulting in a continuous mix of random pseudo-errors in the demodulated pointing error. The filtering gain, error weighting coefficient, and control parameters of the tracking system are difficult to precisely tune in advance. To address this, this application uses the Skyhawk optimization algorithm to adaptively adjust the aforementioned multidimensional parameters online. However, the Skyhawk optimization algorithm uses a rigid switching between exploration and development based on a fixed iteration ratio threshold. Each hunting strategy is randomly activated with a fixed probability, making it impossible to perceive the instantaneous reliability of the error signal throughout the process. When radar cross-section flicker causes fitness surface distortion, the algorithm cannot dynamically adjust the stage switching timing, nor can it prioritize the activation of large-step strategies to jump away from the pseudo-extreme attraction domain, ultimately converging to a false pointing extremum, leading to a decrease in tracking accuracy. Summary of the Invention
[0005] This application provides an antenna tracking and alignment method and system based on conical scanning, which aims to solve the problem that the fixed-stage switching mechanism and fixed-strategy activation probability of the Tianying optimization algorithm cannot respond to the non-stationary distortion of the error signal when the target radar cross-section is randomly flickering, causing the algorithm to converge to a false pointing extremum.
[0006] In a first aspect, this application provides an antenna tracking and alignment method based on conical scanning. The alignment method includes: acquiring the amplitude envelope, azimuth and elevation angles, and scanning phase reference signal of the received antenna signal; normalizing the amplitude envelope frame-by-frame using the reference signal; dividing each frame into equal-length sub-windows; demodulating the mean amplitude of each sub-window using the reference signal to obtain a predicted value; normalizing the difference between each mean amplitude and the predicted value to obtain a real-time residual mean; storing the estimated azimuth and elevation angle errors of each scanning period into a fixed-length queue to form a sliding window; taking the absolute value of the azimuth and elevation angle error difference between adjacent periods and summing them to obtain a merged jump variable; and using the mean of each merged jump variable as the inter-period jump degree to track the target. The kinematic discrimination threshold is determined by multiplying the product of the target's maximum angular velocity and the scanning period. The number of reassessment periods is determined by the ratio of the inter-period jump degree to the kinematic discrimination threshold. The optimal parameter combination within the number of reassessment periods is frozen, and the fitness is recalculated using the new acquisition amplitude envelope. The deviation between each fitness and the mean is normalized to obtain the elite stability. The adaptive discrimination threshold is obtained based on the Laida criterion. The real-time residual mean and the inter-period jump degree are fused to obtain the signal quality score. When the score is less than or equal to the adaptive switching threshold, the Skyhawk optimization algorithm switches back to the exploration phase and expands the search range. When the elite stability is greater than the adaptive discrimination threshold, elite degradation is triggered. The optimal parameter combination is output to the antenna servo system to complete beam alignment.
[0007] By constructing a real-time residual mean through sub-window amplitude residual normalization, precise quantification of single-cycle amplitude distortion is achieved. By constructing an inter-period jump degree based on kinematic constraints, the blind spot of the amplitude domain in perceiving inter-period error jumps is supplemented. By constructing an elite stability degree to independently evaluate the inter-period credibility of elite solutions from the optimization parameter domain, the propagation path of elite contamination effect is cut off. The signal quality score integrates the above indicators into a unified decision basis. The adaptive switching threshold and adaptive discrimination threshold are both adaptively updated with historical statistics, enabling the Tianying optimization algorithm to achieve continuous and accurate tracking and beam alignment in scenarios where random scintillation of radar cross-section and high-speed target maneuvering coexist.
[0008] Furthermore, the frame normalization of the amplitude envelope using the reference signal includes: using the scanning phase reference signal as the time reference, performing time alignment on the amplitude envelope sampling sequence, and dividing it into independent frames with the scanning period as the boundary, so that each frame covers a complete scanning period; normalizing the amplitude envelope sampling sequence of the frame using the mean of all sampling points in each frame; and determining frames whose mean amplitude exceeds three times the historical mean as abnormal frames, filling them with data from the previous valid frame to complete the frame normalization.
[0009] The abnormal frame filling mechanism prevents singular sampled values from entering the calculation of subsequent real-time residual mean and inter-period jump degree in scenarios of sudden changes in target radar cross-section or instantaneous link interruption, ensuring that the statistical input of the three-layer credibility assessment system still has numerical validity during strong interference periods.
[0010] Furthermore, the step of demodulating the amplitude mean of the sub-window with the reference signal to obtain the predicted value includes: using the scanning phase reference signal as a reference, extracting the in-phase component and the quadrature component from the amplitude mean sequence of each sub-window respectively; reconstructing the sinusoidal amplitude value at the phase corresponding to each sub-window using the in-phase component and the quadrature component; and using the sinusoidal reconstructed amplitude value corresponding to each sub-window as the predicted value of that sub-window to complete the demodulation and reconstruction of the predicted value of each sub-window.
[0011] The sinusoidal prediction value is reconstructed by using in-phase and quadrature dual-component demodulation. Even when there is an overall non-stationary offset in the amplitude envelope, the scanning fundamental frequency modulation component can still be accurately separated. This makes the deviation between the mean amplitude of each sub-window and the prediction value more accurately correspond to the non-periodic distortion introduced by radar cross-section scintillation, and improves the perception accuracy of scintillation distortion by the real-time residual mean.
[0012] Furthermore, the step of storing the azimuth and elevation angle error estimates of each scanning cycle into a fixed-length queue to form a sliding window in chronological order includes: after the azimuth and elevation angle error demodulation is completed in each scanning cycle, the azimuth and elevation angle error estimates of the current cycle are rolled into the sliding window, and the record of the earliest cycle is removed, keeping the total length of the sliding window unchanged; for abnormal error estimates that exceed the antenna's paraxial range, the previous valid error estimate is used to replace them; the above operations are performed in each scanning cycle to form a sliding window for calculating the inter-period jump degree.
[0013] Furthermore, the kinematic discrimination threshold includes: in each scanning cycle, statistical analysis of each merged jump variable within the sliding window, taking the maximum value of the merged jump variable within the sliding window as the online estimate of the maximum angular velocity of the tracked target in the current cycle; and obtaining the kinematic discrimination threshold by multiplying the product of this online estimate and the scanning cycle.
[0014] By extracting the maximum value of the sliding window jump variable online and estimating the target's maximum angular velocity in real time, the refresh of the kinematic discrimination threshold is directly linked to the target's maneuvering state. This reduces the dual mismatch caused by the fixed kinematic threshold: missing reasonable error jumps when the target is highly maneuvering and misjudging normal changes as flickering interference when the target is low maneuvering.
[0015] Furthermore, the recalculation of fitness with the new acquisition amplitude envelope includes: substituting the new acquisition amplitude envelope after frame normalization into the fitness function, driving the Kalman filter with the currently frozen optimal parameter combination to estimate the errors of the azimuth and elevation channels respectively; multiplying the absolute values of the tracking errors of the azimuth channel and the absolute values of the tracking errors of the elevation channel by their respective error weighting coefficients and summing them, and using the resulting weighted sum as the fitness of the current scanning period to complete the fitness recalculation under the new acquisition amplitude envelope.
[0016] Furthermore, the signal quality score includes: dividing the real-time residual mean by its maximum value within the historical sliding window to obtain the normalized residual mean index; dividing the inter-period jump degree by its maximum value within the historical sliding window to obtain the normalized jump index; summing the normalized residual mean index and the normalized jump index, and taking the negative of the summation result to obtain the signal quality score.
[0017] Furthermore, the process of switching back to the exploration phase and expanding the search range in the Skyhawk optimization algorithm includes: activating the high-altitude vertical dive strategy of the Skyhawk optimization algorithm, dynamically amplifying the search step size scaling factor of the high-altitude vertical dive strategy based on the difference between the signal quality score and the adaptive switching threshold; the larger the absolute value of the difference, the larger the search step size scaling factor; when the signal quality score is higher than or equal to the adaptive switching threshold, switching to the fine local search strategy, performing development operations within the neighborhood of the optimal parameter combination, and completing the phase switching and search range control of the Skyhawk optimization algorithm.
[0018] The search step size scaling factor dynamically amplifies as the signal quality score deteriorates. When severe radar cross-section flicker causes serious distortion of the fitness surface, the search radius is automatically expanded. After the signal quality is restored, the radius is automatically narrowed to a fine development mode, which solves the problem that the fixed step size search strategy cannot effectively escape in the pseudo-extreme attraction domain.
[0019] Furthermore, the step of outputting the optimal parameter combination to the antenna servo system to complete beam alignment includes: the optimal parameter combination containing the diagonal elements of the Kalman filter observation noise covariance matrix and error weighting coefficients; outputting the diagonal elements of the Kalman filter observation noise covariance matrix to the antenna servo system, increasing the diagonal elements of the Kalman filter observation noise covariance matrix when the signal quality score is less than or equal to the adaptive switching threshold, in order to adjust the degree of acceptance of the current observation by the Kalman filter; outputting the error weighting coefficients to the antenna servo system, adjusting the contribution ratio of the azimuth and elevation channel errors to the servo drive command based on the current signal-to-noise ratio of the two channels, and completing the full output of the optimal parameter combination and beam alignment.
[0020] In a second aspect, this application also provides an antenna tracking and alignment system based on conical scanning, including a processor, a memory, and a communication interface. The memory stores computer program instructions, which, when executed by the processor, implement the antenna tracking and alignment method based on conical scanning according to the first aspect of this application.
[0021] This application has the following technical advantages: This application constructs real-time residual mean, inter-period jump degree, and elite stability respectively to complement each other and jointly quantify the interference of radar cross-section scintillation on the fitness evaluation of the Tianying optimization algorithm. The signal quality score integrates the above indicators into a unified decision criterion. When the score is less than or equal to the adaptive switching threshold, the algorithm is forced to switch back to the exploration phase and the search step size is increased according to the signal quality. After the signal quality is restored, it automatically narrows to the fine development mode. When the elite stability is greater than the adaptive discrimination threshold, elite degradation is triggered, and the historical best non-degraded solution is temporarily replaced with the false elite, cutting off the propagation path of the elite contamination effect. This enables the Tianying optimization algorithm to achieve continuous and accurate tracking and beam alignment in scenarios where radar cross-section random scintillation and high-speed target maneuvering coexist. Attached Figure Description
[0022] Figure 1 This is a flowchart of an antenna tracking and alignment method based on conical scanning according to an embodiment of this application.
[0023] Figure 2 This is a comparison chart of the real-time residual mean and the intertemporal jump degree in response to the continuous random flashing of the target radar cross-section.
[0024] Figure 3 This is a comparison chart showing the effect of elite stability on the identification and downgrading of fake elites.
[0025] Figure 4 This is a comparison chart of the tracking error convergence between this application and the traditional Skyhawk optimization algorithm.
[0026] Figure 5 This is a structural block diagram of an antenna tracking and alignment system based on conical scanning according to an embodiment of this application. Detailed Implementation
[0027] Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0028] The first aspect of this application provides an antenna tracking and alignment method based on conical scanning. Figure 1 This is a flowchart of an antenna tracking and alignment method based on conical scanning according to an embodiment of this application. The specific implementation process of this method will be described in detail below.
[0029] S101: Acquire relevant signal data and perform preprocessing.
[0030] In this embodiment, during system operation, the system acquires the received signal amplitude envelope, real-time readings of antenna azimuth and elevation angles, and scanning phase reference signals.
[0031] The amplitude envelope is processed sequentially by the bandpass filter and envelope detection circuit at the front end of the RF receiver, and then digitally acquired by the analog-to-digital converter at a rate of 64 sampling points per scan cycle; the antenna azimuth and elevation angles are output in real time by the photoelectric encoder installed on the turntable axis; the scanning phase reference signal is synchronously read from the conical scanning drive controller to provide a phase reference for subsequent frame alignment and synchronous demodulation.
[0032] It should be noted that after the data collection is completed, the above data will be preprocessed sequentially, including: Using the scanning phase reference signal as the time base, the amplitude envelope sampling sequence is time-aligned and divided into independent frames with the scanning period as the boundary, ensuring that each frame covers a complete scanning period.
[0033] Amplitude normalization is performed on each frame: the mean of all sampling points in the frame is calculated and recorded as the mean of the frame before normalization; each sampling point in the frame is divided by the mean of the frame before normalization to obtain the normalized amplitude sequence.
[0034] According to the Raida criterion, frames whose mean amplitude deviates from the historical mean by more than 3 times the standard deviation are identified as abnormal frames. Data from the previous valid frame is used to replace the data in the abnormal frames to avoid extreme outliers in subsequent index calculations.
[0035] S102: Single-cycle amplitude distortion quantization based on sub-window sinusoidal residuals.
[0036] In this embodiment, under paraxial tracking conditions, the received signal amplitude varies with the scanning phase in a single-frequency sinusoidal modulation manner, and the single-cycle amplitude envelope is a smooth periodic waveform. When the target radar cross-section is affected by attitude flipping or multipath scattering interference, random and violent fluctuations occur within a timescale shorter than the scanning period. Random components unrelated to the scanning phase are superimposed in a specific phase interval of the amplitude envelope, resulting in local bulges or depressions in the waveform.
[0037] Taking a sudden drop in radar cross-section within a quarter to three-quarters of the scanning cycle as an example, the amplitude sampling values during this period are systematically low, and the envelope shows a significant residual compared to the ideal sine curve. The greater the scintillation intensity, the more significant the deviation of the amplitude envelope from the sinusoidal law. This type of distortion directly contaminates the pointing error extracted by synchronous demodulation, causing the fitness evaluation of the Tianying optimization algorithm to form a pseudo-extreme attraction domain in the parameter space.
[0038] Therefore, this application constructs an index that can quantify the degree of amplitude envelope distortion in a single cycle in real time, providing a decision-making basis for subsequent adaptive phase switching.
[0039] Specifically, the normalized amplitude sequence of each frame is evenly divided into K equal-length sub-windows, which are set to 8 in this embodiment, but can be adjusted according to actual conditions. The mean of all normalized sampling points in each sub-window is calculated to obtain the mean amplitude of K sub-windows; the center scanning phase of each sub-window is determined by the scanning phase reference signal to obtain K corresponding phase values.
[0040] Using the average amplitude and corresponding phase values of K sub-windows as input, a synchronous correlation demodulation method is employed to extract the sinusoidal fundamental frequency component, thereby reconstructing the predicted value at each sub-window. The specific steps are as follows: The average amplitude of each of the K sub-windows is multiplied by the cosine value of its corresponding phase, the average of the K products is multiplied by 2 to obtain the in-phase component; the average amplitude of each of the K sub-windows is multiplied by the sine value of its corresponding phase, the average of the K products is multiplied by 2 to obtain the quadrature component. For the k-th sub-window, its predicted value is equal to the product of the in-phase component and the cosine value of the corresponding phase of that sub-window, plus the product of the quadrature component and the sine value of the corresponding phase of that sub-window. Essentially, this method uses the sinusoidal basis function of the scanning fundamental frequency to perform least-squares fitting on the average amplitude of the K sub-windows, using the scanning phase reference signal as a benchmark to ensure phase alignment. Even when there is a gradual overall shift in the amplitude envelope, it can still accurately separate the sinusoidal modulation component, ensuring that the residual precisely corresponds to the non-periodic distortion introduced by flicker.
[0041] Calculate the difference between the mean amplitude of each sub-window and the corresponding predicted value to obtain K amplitude residuals. Since the flicker direction varies, each amplitude residual can be positive or negative. Take the absolute value of each of the K amplitude residuals and sum them to obtain the sum of the absolute values of the amplitude residuals. Divide this sum by the normalized mean of the previous frame for that frame to obtain a dimensionless ratio, which is used as the real-time mean residual of the current frame. Using the normalized mean of the previous frame as the divisor eliminates the systematic differences in the real-time mean residual across different signal power levels.
[0042] The real-time residual mean can effectively reflect the flicker intensity because: the denominator reflects the average power within the frame, and its changes come from slowly varying factors such as link loss, with a time scale greater than the scan period; the numerator corresponds to the short-term abrupt changes in the local phase interval within the frame; normalizing the denominator only removes the influence of slowly varying gain, without weakening the numerator's perception of short-term abrupt changes in flicker. Therefore, the stronger the flicker, the larger the index, and the index approaches zero when there is no flicker.
[0043] S103: Cross-cycle error jump detection based on kinematic constraints.
[0044] In this embodiment, the target's true pointing deviation is constrained by kinematic laws and is a slow variable; its change between adjacent scan cycles is limited by the target's maximum angular velocity. Within a single scan cycle... Within, the angular displacement of the tracked target is less than or equal to the maximum angular velocity and The product of the two values is the difference between adjacent cycles of the true pointing error on a single azimuth or elevation channel. Under normal conditions, this difference is less than or equal to the product of the maximum angular velocity and the scanning period. The combined jump variable is the sum of the absolute values of the differences between the azimuth and elevation channels, and its theoretical kinematic upper limit is twice the upper limit of a single channel. When the measured combined jump variable exceeds this theoretical upper limit, the difference cannot be explained by normal target maneuvers, and is therefore determined to be an error jump caused by radar cross-section scintillation. Radar cross-section scintillation randomly changes the amplitude weighting at each scanning phase, causing the current cycle error estimate to deflect randomly. After the scintillation pattern changes in the next cycle, it jumps back. In extreme cases, the difference between adjacent errors can exceed the kinematic upper limit. The real-time residual mean acts on the single-cycle amplitude domain and does not directly perceive cross-cycle error jumps. Therefore, a cross-cycle jump degree needs to be constructed separately from the multi-cycle error domain to complement it.
[0045] Specifically, the azimuth and elevation error estimates for each scanning cycle are stored sequentially in a fixed-length queue forming a sliding window. After demodulation of the azimuth and elevation errors in each scanning cycle, the estimated azimuth and elevation errors for the current cycle are added to the sliding window, and the record from the earliest cycle is removed, keeping the total length of the sliding window constant. For abnormal error estimates that exceed the antenna's paraxial operating range, the estimate of the previous effective error is used to replace them, preventing extreme values from contaminating subsequent differential statistics. Under paraxial conditions, the demodulation noise characteristics of the azimuth and elevation channel errors are approximately symmetrical, the cross-coupling effect is negligible, and the differential values of the two channels are both in degrees, allowing for direct merging and statistical analysis.
[0046] The absolute values of the azimuth and elevation angle error differences of each adjacent period within the sliding window are taken and summed to obtain the combined jump variable for that period. The mean of all combined jump variables within the sliding window is then calculated, and the resulting mean is the inter-period jump degree at the current moment, in degrees. The length of the sliding window determines the smoothness of the inter-period jump degree in relation to historical error jumps, and is generally taken as 4 to 8 scan periods.
[0047] In each scan cycle, the maximum value of the merged jump variables within the sliding window is taken as the online estimate of the maximum angular velocity of the tracked target in the current cycle. The product of this estimate and the scan cycle determines the kinematic discrimination threshold at the current moment, and it is refreshed synchronously with the scrolling update of the sliding window. This method enables the kinematic discrimination threshold to automatically rise when the target is highly maneuvering and automatically fall when it is low maneuvering, reducing the degree of missed detection during high maneuvering and false detection during low maneuvering under different maneuvering states with a fixed threshold.
[0048] When the interperiod jump degree exceeds the kinematic discrimination threshold, radar cross-section scintillation interference is detected, triggering the subsequent algorithm stage switching logic. The interperiod jump degree supplements the perception blind spot of the real-time residual mean from the multi-cycle error difference dimension, and mutually corroborates each other from the amplitude domain and error domain, jointly providing a more complete decision basis for algorithm control.
[0049] For example, suppose the maximum angular velocity of the tracked target is 100 angular velocity per second. If the scan period is 0.1 seconds, then the upper bound of the single-channel kinematics is: The kinematic discrimination threshold for merging jump variables is In normal maneuvering scenarios without radar cross-section scintillation, the target azimuth error estimation varies between adjacent scan cycles. Become The absolute value of the difference is Pitch angle error estimation is made by Become The absolute value of the difference is ; merge jump variables are less than The kinematic discrimination threshold is used to determine the error change caused by normal maneuvering.
[0050] When strong radar cross-section scintillation occurs, the scintillation causes the azimuth error estimate to randomly deflect during the Nth scan cycle. After the flashing pattern changes in the (N+1)th cycle, it jumps back to the previous state. The absolute value of the azimuth difference between adjacent periods reached The pitch channel is similar, with the absolute value of the difference reaching [a certain value]. ; merge jump variables are greater than The kinematic discrimination threshold is used to determine the presence of radar cross-section scintillation interference, triggering subsequent algorithm stage switching operations.
[0051] like Figure 2 The figure shows a comparison of the real-time residual mean and the inter-period jump degree in response to continuous random flickering of the target radar cross-section. Each time radar cross-section flickering occurs, the real-time residual mean and the inter-period jump degree jump synchronously, and the jump amplitude is positively correlated with the flicker intensity, verifying the effectiveness of both in sensing flicker interference from the amplitude domain and the error domain respectively.
[0052] S104: Elite credibility assessment based on fitness consistency across cycles.
[0053] The parameters to be optimized in the Skyhawk optimization algorithm include two types of parameters: the diagonal elements of the observation noise covariance matrix of the Kalman filter, and the error weighting coefficients of the azimuth and elevation channels. After each iteration, the algorithm selects the individual with the smallest fitness from all evaluated parameter combinations, denoted as the elite individual, as the best estimate of the current global optimum, and uses the elite individual as a benchmark to guide subsequent local fine-grained searches.
[0054] In radar cross-section scintillation scenarios, if the fitness of a pseudo-optimal parameter combination happens to be better than the true optimal solution within a certain strong scintillation period, this pseudo-optimal value will be incorrectly identified as an elite individual and continuously protected. Subsequent iterations will converge to this false optimal value, resulting in an elite contamination effect. The real-time residual mean and inter-period jump degree both act on the signal domain and the error domain, making it impossible to directly assess the inter-period reliability of elite individuals in the parameter space. The fitness of the true optimal solution fluctuates only slightly across periods, while the fitness of the pseudo-optimal solution will significantly deteriorate within normal periods, and the multi-period fitness sequence will oscillate dramatically.
[0055] Based on the above analysis, this application constructs an elite stability mechanism, specifically including: after determining elite individuals in each iteration, determining the number of re-evaluation periods based on the ratio of intertemporal jump degree to the kinematic discrimination threshold. When the ratio is less than or equal to 1, the target is in a low-maneuverability state, the target angular velocity changes slowly, the optimal parameter solution drift rate is low, and it can withstand a longer freeze time. The value can be taken from 4 to 6; in this embodiment, it is taken as... The ratio is 5; when the ratio is greater than 1, the target enters a high-maneuverability state, the optimal solution drifts faster, and the freeze time must be compressed. The value can be taken from 2 to 4; in this embodiment, it is taken as... The value is 3. The value is adjusted synchronously with the inter-period jump degree update of each scan cycle. During high maneuverability, the freeze time is shortened to ensure tracking continuity, while during low maneuverability, the freeze time is extended to improve assessment robustness.
[0056] Sure Afterwards, the parameter combinations of elite individuals were frozen in the subsequent... Within each scan cycle, it does not participate in iterative updates. The newly acquired amplitude envelope after S101 preprocessing in each cycle is substituted into the fitness function to obtain... The fitness reassessment value is as follows. The fitness function calculation process is as follows: The Kalman filter is driven by the diagonal elements of the observation noise covariance matrix in the current frozen parameter combination, and errors are estimated for the azimuth and elevation channels respectively. The absolute values of the tracking errors in the azimuth and elevation channels are weighted separately using the error weighting coefficients in the current frozen parameter combination, and then summed. The resulting weighted sum is the fitness for that period; a smaller value indicates a better parameter combination. Abnormal periods that occur during the period are filled with the fitness values of adjacent normal periods, and the fitness is calculated... The mean of individual weight assessment values.
[0057] calculate The difference between each fitness reassessment value and the mean is summed by taking the absolute value of each difference, and then the sum is divided by 1 / 2. The product of this and the mean fitness is the dimensionless ratio, which represents the elite stability at the current moment. The product of the fitness mean and the fitness mean is used as the divisor, which eliminates the systematic influence of the number of evaluations and the absolute magnitude of fitness on the index, so that the normalized index reflects the relative cross-period fluctuation of the fitness of elite solutions.
[0058] The adaptive threshold for elite stability employs a sliding statistical method: using a historical sliding window of 30 to 50 scan cycles, the mean and standard deviation of elite stability are continuously calculated. Based on the Laida criterion, the mean plus the standard deviation is used as the current threshold, which is updated every scan cycle. When elite stability exceeds the adaptive threshold, an elite degradation mechanism is triggered, reinstating the current elite individual to participate in global exploration. The historical best non-degraded solution is used as a temporary elite, cutting off the propagation path of false extrema. When stability is less than or equal to the adaptive threshold, the current elite individual is retained normally.
[0059] like Figure 3 As shown, this figure compares the effectiveness of elite stability in identifying and downgrading false elites, comparing the elite fitness changes of elite stability and the traditional Skyhawk optimization algorithm within the random flashing period of the radar cross-section. The traditional algorithm incorrectly retains pseudo-extreme parameters as elites, and the fitness deteriorates significantly during the flashing period; in this application, elite downgrading is triggered after the elite stability exceeds the adaptive discrimination threshold, and the temporary elite fitness remains stable, verifying the effectiveness of the elite downgrading mechanism.
[0060] S105: Execute the improved algorithm online and complete the tracking alignment.
[0061] In this embodiment, an online rolling optimization framework is adopted. The improved Skyhawk optimization algorithm completes one population iteration in each scanning cycle. Through continuous rolling iteration, it gradually approaches the optimal parameter combination, which matches the dynamic time scale of the target motion, without the need for offline pre-convergence.
[0062] After updating the real-time residual mean and interperiod jump degree in each new scan cycle, both are divided by their respective maximum values within a historical sliding window of 30 to 50 scan cycles, transforming them into dimensionless normalized quantities. The sum of the two and the negative of the sum are then used to obtain the signal quality score. The smaller the signal quality score, the worse the current signal quality.
[0063] The algorithm dynamically controls each stage based on a comparison between the signal quality score and an adaptive switching threshold. When the score is less than or equal to the adaptive switching threshold, it forcibly switches back to the exploration stage, activates the high-altitude vertical dive strategy, and dynamically amplifies the search step size scaling factor based on the absolute value of the difference between the score and the switching threshold; the larger the absolute value of the difference, the larger the search step size. When the score is higher than or equal to the adaptive switching threshold, it switches to a fine-grained local search strategy, performing development operations within the neighborhood of the current elite individual.
[0064] The adaptive switching threshold for signal quality scoring is determined as follows: a historical sliding window of 30 to 50 scan cycles is used, with a value of 30 in this embodiment; the mean and standard deviation of the signal quality score are continuously calculated, and the mean minus the standard deviation is used as the current switching threshold according to the Laida criterion. The threshold is updated every scan cycle, so that the threshold automatically tracks the baseline level of the current channel environment. That is, the threshold moves down when the overall signal deteriorates and moves up when the signal quality improves.
[0065] Each time an elite individual is updated, the comparison between the elite's stability and the adaptive discrimination threshold is checked simultaneously: if the stability is greater than the adaptive discrimination threshold, the elite is downgraded; if it is less than or equal to the adaptive discrimination threshold, the current elite individual is retained normally. Both mechanisms are executed synchronously within each scan cycle.
[0066] The improved Tianying optimization algorithm, after completing parameter optimization in each scan cycle, outputs the current elite individual parameter combination to the antenna servo control system in real time. The diagonal elements of the Kalman filter's observation noise covariance matrix are used to update the filter's observation noise model: when the signal quality score is less than or equal to the adaptive switching threshold, the diagonal elements are increased to reduce the filter's weighting of distorted observations; when the score returns to normal, the diagonal elements are appropriately decreased to improve the response sensitivity to effective observations. The error weighting coefficients are estimated based on the current signal-to-noise ratio of the azimuth and elevation channels, dynamically adjusting the contribution ratio of the two channel errors to the servo drive commands to compensate for tracking skew caused by signal-to-noise ratio asymmetry.
[0067] After completing the servo adjustment of the current cycle, the antenna continues to perform conical scanning and enters the acquisition, preprocessing, index evaluation and parameter optimization process of the next cycle. This process is repeated to achieve continuous and accurate tracking and beam alignment of high dynamic targets in the scenario of random scintillation of radar cross-section.
[0068] like Figure 4The figure shows a comparison of the tracking error convergence between the proposed algorithm and the traditional Tianying optimization algorithm. It presents a comparison of the azimuth tracking error convergence between the improved Tianying optimization algorithm and the original Tianying optimization algorithm in a scenario involving continuous random radar cross-section scintillation and target maneuvering. The proposed algorithm recovers to a smaller error within approximately two to three scan cycles after each scintillation event; the original algorithm repeatedly falls into pseudo-extremes, with a significantly higher maximum error, verifying the advantages of the adaptive control mechanism.
[0069] According to a second aspect of this application, this application also provides an antenna tracking and alignment system based on conical scanning. Figure 5 This is a structural block diagram of an antenna tracking and alignment system based on conical scanning according to an embodiment of this application. Figure 5 As shown, the system 50 includes a processor, a memory, and a communication interface. The memory stores computer program instructions, which, when executed by the processor, implement the antenna tracking and alignment method based on conical scanning according to the first aspect of this application. The system also includes other components well-known to those skilled in the art, such as a communication bus and a communication interface; their configuration and functions are known in the art and will not be described further here.
Claims
1. An antenna tracking and alignment method based on conical scanning, characterized in that, The alignment method includes: acquiring the amplitude envelope, azimuth and elevation angles, and scanning phase reference signal of the antenna received signal, and using the reference signal to frame-normalize the amplitude envelope; Each frame is divided into equal-length sub-windows. The amplitude mean of each sub-window is demodulated using a reference signal to obtain the predicted value. The difference between each amplitude mean and the predicted value is normalized to obtain the real-time residual mean. The azimuth and elevation angle error estimates for each scanning cycle are stored sequentially in a fixed-length queue to form a sliding window. The absolute values of the azimuth and elevation angle error differences between adjacent cycles are taken and summed to obtain the merged jump variable. The mean of each merged jump variable is used as the inter-period jump degree. The kinematic discrimination threshold is determined by twice the product of the maximum angular velocity of the tracked target and the scanning cycle. The inter-period jump degree and the kinematic discrimination threshold are used to determine the kinematic discrimination threshold. The ratio determines the number of reassessment cycles. The optimal parameter combination within the reassessment cycle is frozen, and the fitness is recalculated using the new acquisition amplitude envelope. The deviation between each fitness and the mean is normalized to obtain the elite stability. An adaptive discrimination threshold is obtained based on the Laida criterion. The real-time residual mean and the inter-period jump degree are fused to obtain the signal quality score. When the score is less than or equal to the adaptive switching threshold, the Skyhawk optimization algorithm switches back to the exploration phase and expands the search range. When the elite stability is greater than the adaptive discrimination threshold, elite degradation is triggered. The optimal parameter combination is output to the antenna servo system to complete beam alignment. The signal quality score includes: dividing the real-time residual mean by its maximum value within the historical sliding window to obtain the normalized residual mean index; dividing the inter-period jump degree by its maximum value within the historical sliding window to obtain the normalized jump index; summing the normalized residual mean index and the normalized jump index, and taking the negative of the summation result to obtain the signal quality score.
2. The antenna tracking and alignment method based on conical scanning according to claim 1, characterized in that, The frame normalization of amplitude envelope using a reference signal includes: using the scanning phase reference signal as a time base, aligning the amplitude envelope sampling sequence in time, and dividing it into independent frames with the scanning period as the boundary, so that each frame covers a complete scanning period; normalizing the amplitude envelope sampling sequence of each frame using the mean of all sampling points in each frame; and determining frames whose mean amplitude exceeds three times the historical mean as abnormal frames, filling them with data from the previous valid frame to complete the frame normalization.
3. The antenna tracking and alignment method based on conical scanning according to claim 1, characterized in that, The step of demodulating the amplitude mean of the sub-window with the reference signal to obtain the predicted value includes: using the scanning phase reference signal as a reference, extracting the in-phase component and the quadrature component from the amplitude mean sequence of each sub-window respectively; reconstructing the sinusoidal amplitude value at the phase corresponding to each sub-window using the in-phase component and the quadrature component; and using the sinusoidal reconstructed amplitude value corresponding to each sub-window as the predicted value of that sub-window to complete the demodulation and reconstruction of the predicted value of each sub-window.
4. The antenna tracking and alignment method based on conical scanning according to claim 1, characterized in that, The step of storing the estimated azimuth and elevation angle errors of each scanning cycle into a fixed-length queue to form a sliding window in chronological order includes: after demodulating the azimuth and elevation angle errors in each scanning cycle, rolling the estimated azimuth and elevation angle errors of the current cycle into the sliding window, and removing the record of the earliest cycle while keeping the total length of the sliding window unchanged; for abnormal error estimates that exceed the antenna's paraxial range, the previous valid error estimate is used to replace them; the above operations are performed in each scanning cycle to form a sliding window for calculating the inter-period jump degree.
5. The antenna tracking and alignment method based on conical scanning according to claim 1, characterized in that, The kinematic discrimination threshold includes: in each scanning cycle, statistical analysis of each merged jump variable within the sliding window, taking the maximum value of the merged jump variable within the sliding window as the online estimate of the maximum angular velocity of the tracked target in the current cycle; and obtaining the kinematic discrimination threshold by multiplying the product of this online estimate and the scanning cycle.
6. The antenna tracking and alignment method based on conical scanning according to claim 1, characterized in that, The fitness recalculation based on the new acquisition amplitude envelope includes: substituting the new acquisition amplitude envelope after frame normalization into the fitness function; driving the Kalman filter with the currently frozen optimal parameter combination to estimate the errors of the azimuth and elevation channels respectively; multiplying the absolute values of the tracking errors of the azimuth channel and the absolute values of the tracking errors of the elevation channel by their respective error weighting coefficients and summing them; using the resulting weighted sum as the fitness of the current scanning period to complete the fitness recalculation under the new acquisition amplitude envelope.
7. The antenna tracking and alignment method based on conical scanning according to claim 1, characterized in that, The process of switching back to the exploration phase and expanding the search range in the Skyhawk optimization algorithm includes: activating the high-altitude vertical dive strategy of the Skyhawk optimization algorithm; dynamically amplifying the search step size scaling factor of the high-altitude vertical dive strategy based on the difference between the signal quality score and the adaptive switching threshold; the larger the absolute value of the difference, the larger the search step size scaling factor; when the signal quality score is higher than or equal to the adaptive switching threshold, switching to the fine local search strategy; performing development operations within the neighborhood of the optimal parameter combination; and completing the phase switching and search range control of the Skyhawk optimization algorithm.
8. The antenna tracking and alignment method based on conical scanning according to claim 1, characterized in that, The step of outputting the optimal parameter combination to the antenna servo system to complete beam alignment includes: the optimal parameter combination containing the diagonal elements of the Kalman filter observation noise covariance matrix and error weighting coefficients; outputting the diagonal elements of the Kalman filter observation noise covariance matrix to the antenna servo system, increasing the diagonal elements of the Kalman filter observation noise covariance matrix when the signal quality score is less than or equal to the adaptive switching threshold to adjust the Kalman filter's acceptance level of the current observation; outputting the error weighting coefficients to the antenna servo system, adjusting the contribution ratio of azimuth and elevation channel errors to the servo drive command based on the current signal-to-noise ratio of the two channels, and completing the full output of the optimal parameter combination and beam alignment.
9. An antenna tracking and alignment system based on conical scanning, characterized in that, include: The processor, memory, and communication interface are provided, wherein the memory stores a computer program that, when executed by the processor, implements an antenna tracking and alignment method based on conical scanning as described in any one of claims 1 to 8.