Dual-polarization rainfall radar signal processing method and system

Abstract

The invention provides a dual-polarization rainfall radar signal processing method and a dual-polarization rainfall radar signal processing system, relates to the field of weather radars, and solves the problems of poor expansibility, insufficient flexibility and poor adaptivity caused by adopting FPGA and DSP hardware processing boards to process rainfall signals in a Doppler mode in the prior art. The method specifically comprises the following steps: acquiring a dual-polarization echo signal, preprocessing the dual-polarization echo signal, and performing orthogonal demodulation and digital processing to obtain dual-polarization channel IQ data; performing multi-thread parallel processing on the dual-polarization channel IQ data to generate meteorological product data and a processing state data packet; performing grading processing according to the state data packet to generate a control instruction; and dynamically adjusting the multi-thread parallel processing process according to the control instruction. According to the method and the device, the limitation of hardware equipment on a signal processing process can be reduced by generating the IQ data of the dual-polarization channel, and the multi-thread parallel processing process can be dynamically adjusted through the control instruction, so that the flexibility and the adaptivity of signal processing are improved.

Description

Technical Field

[0001] This invention belongs to the field of weather radar and relates to radar signal processing technology, specifically a dual-polarization rainfall radar signal processing method and system. Background Technology

[0002] Rain gauge radar is a core piece of equipment for monitoring regional precipitation and providing early warnings of floods. The performance of its signal processing system directly determines the accuracy and reliability of the final meteorological products (such as intensity, velocity, and precipitation rate). Traditional rain gauge radar signal processing systems are mostly based on Doppler rain gauge radar implemented using FPGA (Field Programmable Gate Array) and DSP (Digital Signal Processor) hardware processing boards. While this architecture is highly efficient in handling deterministic tasks, it also has significant limitations: First, the hardware resources (such as logic units and memory) are fixed and limited, making it difficult to expand the algorithm's functionality and integrate more complex processing flows such as dual-polarization parameter calculation; second, the deep coupling between the algorithm and hardware results in long cycles, high costs, and insufficient flexibility for later updates, optimizations, or reconstruction of processing flows for different scenarios.

[0003] This application provides a dual-polarization rain radar signal processing method and system, which solves the technical problems of poor scalability, insufficient flexibility, and inability to adaptively adjust according to the scenario in the existing fixed processing architecture.

[0004] To solve the above-mentioned technical problems, this application adopts the following technical solution: Firstly, a dual-polarization rain radar signal processing method is provided, including: The system acquires dual-polarization echo signals received by the radar antenna and preprocesses them. The preprocessed dual-polarization echo signals undergo orthogonal demodulation and analog-to-digital conversion to obtain dual-polarization channel IQ data (horizontal polarization H channel and vertical polarization V channel). The dual-polarization channel IQ data is then processed in a multi-threaded parallel manner to generate a processing status data packet and the final meteorological product. The processing status data packet contains the processing results of each step in the multi-threaded parallel processing. The processing status data packet is then processed hierarchically to generate control commands. The multi-threaded parallel processing process is dynamically adjusted according to the control commands.

[0005] Based on the above technical solution, the dual-polarization rain radar signal processing method provided in this application replaces the traditional FPGA / DSP hardware-based process with a fully software-based parallel processing architecture. This achieves general-purpose, logic-driven optimization at the system level, solving the core problems of poor scalability and difficulty in algorithm updates inherent in traditional architectures. This method not only allows for the flexible integration of complex algorithm modules but also enables comparison and matching between real-time output processing status data packets and scene rules and quality thresholds in a preset knowledge base. This allows for dynamic adjustment of the processing flow and resource scheduling at the logical level, laying the foundation for the high real-time performance and strong adaptability of this method.

[0006] In conjunction with the first aspect above, in one possible implementation, the method for acquiring the dual-polarization channel IQ data includes: The synchronous intermediate frequency signal is obtained by demodulating the dual-polarized echo signal after preprocessing the local oscillator signal using the same clock source. The synchronized intermediate frequency signal is mixed with two orthogonal local oscillator signals from the same source to obtain the I / Q analog signals of the H channel and V channel respectively; where I is the in-phase component and Q is the quadrature component, and the local oscillator signals from the same source include: a local oscillator signal in phase with the main local oscillator and a local oscillator signal orthogonal to the main local oscillator. The I / Q analog signals of the H channel and the I / Q analog signals of the V channel are converted from analog to digital on the same clock edge to obtain the dual-polarized channel IQ data of the horizontally polarized H channel and the vertically polarized V channel.

[0007] In conjunction with the first aspect above, in one possible implementation, the multi-threaded parallel processing includes: pulse compression processing, coherent accumulation processing, filtering processing, parameter calculation processing, quality control processing, and parameter configuration processing; The pulse compression process is used to obtain a one-dimensional range image sequence from the dual-polarized channel IQ data through matched filtering. The coherent accumulation process is used to coherently superimpose one-dimensional range image sequences to obtain a range Doppler data matrix. The filtering process is used to filter the range Doppler data matrix to obtain the meteorological signal spectrum after filtering out clutter components. The parameter calculation and processing are used to calculate the spectral distance, estimate the autocorrelation, and calculate the dual polarization parameters of the meteorological signal spectrum after filtering out clutter components to obtain meteorological base data. The meteorological base data includes: reflectivity factor, radial velocity, spectral width, differential reflectivity, differential phase constant, correlation coefficient, signal-to-noise ratio, quality factor SQI, and quality factor CCOR. The quality control process is used to remove meteorological base data by distance averaging and extract the quality factor SQI and quality factor CCOR control output parameters to obtain quality-controlled meteorological product data. The control output parameters include: setting a decision threshold based on the quality factor SQI and quality factor CCOR values ​​to remove invalid values ​​below the threshold in the meteorological base data, and using the quality factor SQI and quality factor CCOR as weighting factors to perform a weighted average calculation on the meteorological base data after removing invalid values ​​to obtain the quality-controlled meteorological product data.

[0008] The parameter configuration processing is used to correct noise in the quality-controlled meteorological product data using actual radar measurement parameters and to calibrate the quality-controlled meteorological product data to obtain the final meteorological product. The results of pulse compression processing, coherent accumulation processing, filtering processing, parameter calculation processing, and quality control processing are integrated into a processing status data packet.

[0009] In conjunction with the first aspect above, in one possible implementation, the processing status data packet is processed hierarchically to generate control instructions; the hierarchical processing includes: Feature parameters are extracted and calculated from the processed status data packets. These feature parameters include: signal-to-noise ratio (SNR), average correlation coefficient (ρhv), reflectance gradient, data intensity, power spectrum, and quality factor. The feature parameters are compared and matched with scene rules and quality thresholds in a preset knowledge base; Based on the matching results, control instructions containing specific adjustment actions are generated; the control instructions include: instruction type, parameter set, and conditions or range under which the instruction takes effect.

[0010] In conjunction with the first aspect above, in one possible implementation, the control command dynamically adjusts the multi-threaded parallel processing process, including: selecting a filtering algorithm, setting a quality factor and switching to the corresponding rainfall detection mode, and adjusting the parameter configuration according to the corresponding rainfall detection mode; the filtering algorithm includes: clutter map construction, adaptive spectral filtering algorithm, and IIR filtering algorithm.

[0011] In conjunction with the first aspect above, in one possible implementation, the quality factor includes: SQI quality factor and CCOR quality factor; The SQI quality factor is calculated using the following formula: calculate; It is the signal-to-noise ratio, and W is the spectral width. It is a function of spectral width; for a pure sinusoidal signal with a spectral width of zero, yes The function.

[0012] In conjunction with the first aspect mentioned above, in one possible implementation, the CCOR quality factor is calculated as follows: According to calculation Value, and preset Threshold comparison; When calculated Greater than or equal to the preset At the threshold; through the calculation formula calculate Quality factor; among which, It's about data strength. It is the power spectrum; When calculated Less than preset At the threshold; through the calculation formula calculate Quality factor; where C is clutter power and S is signal power. , N is the noise power. It is a statistic obtained by performing a first-order autocorrelation operation on a dual-polarization echo signal.

[0013] In conjunction with the first aspect above, in one possible implementation, the switching to the corresponding rainfall detection mode includes: selecting one from the predefined rainfall detection modes based on the result of comparing and matching the feature parameters with the scene rules and quality thresholds in the preset knowledge base, and generating the corresponding control command; The predefined rainfall detection modes include: Conventional precipitation monitoring model, severe convection monitoring model, quantitative estimation model for heavy rainfall, and weak precipitation observation model; The rainfall detection mode is used to associate different sets of parameters; the set of parameters includes: pulse repetition frequency, beam scanning strategy, coherent accumulation pulse number, filtering algorithm preset for the selected rainfall detection mode, and quality factor threshold.

[0014] In conjunction with the first aspect above, in one possible implementation, the adjustment parameter configuration includes: When calculated Less than preset At the threshold, the quality-controlled meteorological product data is corrected for deviation based on the noise power N measured by actual radar. Gain and phase compensation corrections are applied to the relative intensity and phase deviations between the horizontally polarized H channel and the vertically polarized V channel. The deviation correction based on noise power includes: The corrected signal power estimate is obtained by subtracting the noise power from the autocorrelation estimate. Through calculation formula The bias signal-to-noise ratio values ​​of the horizontally polarized H channel and the vertically polarized V channel were calculated respectively. and ; The differential reflectivity ZDR is calculated according to the formula. The calculation and correction were obtained. The correlation coefficient According to the calculation formula The result was obtained through calculation and correction.

[0015] In a second aspect, an electronic device is provided, comprising: a communication unit and a processing unit; the communication unit is configured to acquire a dual-polarization echo signal received by a radar antenna; the processing unit is configured to preprocess, orthogonally demodulate and digitize the signal to obtain synchronous IQ data of horizontal and vertical dual-polarization channels, perform software-based parallel processing on the synchronous IQ data to generate meteorological product data, and send the meteorological product data to a monitoring terminal; and the processing unit is configured to generate and execute control commands based on feedback from the monitoring terminal to dynamically adjust the parameters or modes of the software-based parallel processing.

[0016] Thirdly, this application provides a processing apparatus, comprising: a processor and a storage medium; the storage medium includes instructions, and the processor is configured to execute the instructions to implement the methods described in the first aspect and any possible implementation thereof. The electronic apparatus may be an electronic device or a chip within an electronic device.

[0017] Fourthly, this application provides a dual-polarization rainfall radar signal processing system, comprising: a digital transceiver module, a software-based signal processing module, and a monitoring terminal module; wherein, the digital transceiver module is used to acquire dual-polarization echo signals received by the radar antenna and preprocess the dual-polarization echo signals; the software-based signal processing module is used to perform quadrature demodulation and digitization processing to obtain dual-polarization channel IQ data and to perform software-based parallel processing on the dual-polarization channel IQ data to generate meteorological product data; the monitoring terminal module is used to analyze the meteorological product data, generate control commands, and feed them back to the software-based parallel processing steps.

[0018] Fifthly, this application provides a computer-readable storage medium storing instructions that, when executed on a processing device, cause the processing device to perform the methods described in the first aspect and any possible implementation thereof.

[0019] In a sixth aspect, this application provides a computer program product containing instructions that, when executed on a processing device, cause the processing device to perform the methods described in the first aspect and any possible implementation thereof.

[0020] This application provides a dual-polarization rainfall radar signal processing method and system, which implements scenario-based and technology-driven optimizations at the technical level. During data processing, ensuring the synchronous acquisition and alignment of IQ data from the dual-polarization channels provides a foundation for subsequent precise analysis. For different detection scenarios, the system can dynamically switch preset rainfall detection modes and call matching filter algorithm combinations and quality factor thresholds. In the core signal processing stage, a correction model based on actual noise power is used to correct deviations in key polarization quantities such as differential reflectivity and correlation coefficient under low signal-to-noise ratio conditions. The synergistic effect of these specific technical measures enables the system to implement precise technical adjustments for specific scenario-based needs such as clutter suppression, weak signal extraction, and improved particle classification accuracy, thereby directly and significantly improving the accuracy and reliability of the final meteorological product data under different actual weather conditions.

[0021] It should be understood that the descriptions of technical features, technical solutions, beneficial effects, or similar language in this application do not imply that all features and advantages can be achieved in any single embodiment. Rather, it is understood that the description of a feature or beneficial effect means that a specific technical feature, technical solution, or beneficial effect is included in at least one embodiment. Therefore, the descriptions of technical features, technical solutions, or beneficial effects in this specification do not necessarily refer to the same embodiment. Furthermore, the technical features, technical solutions, and beneficial effects described in this embodiment can be combined in any suitable manner. Those skilled in the art will understand that embodiments can be implemented without one or more specific technical features, technical solutions, or beneficial effects of a particular embodiment. In other embodiments, additional technical features and beneficial effects may be identified in specific embodiments that do not embody all embodiments. Attached Figure Description

[0022] Figure 1 A system architecture diagram of a dual-polarization rain radar signal processing system provided in this application embodiment; Figure 2 A flowchart illustrating a dual-polarization rain radar signal processing method provided in this application embodiment; Figure 3 This is a schematic diagram of the process for obtaining and processing status data packets and final meteorological products provided in an embodiment of this application; Figure 4 A flowchart illustrating the method for obtaining control commands provided in an embodiment of this application; Figure 5 A schematic flowchart of the closed-loop feedback control method provided in the embodiments of this application; Detailed Implementation

[0023] In the description of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B. The "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone. Furthermore, "at least one" means one or more, and "multiple" means two or more. The terms "first," "second," etc., do not limit the quantity or order of execution, and "first," "second," etc., do not necessarily imply differences.

[0024] It should be noted that, in this application, the terms "exemplary" or "for example" are used to indicate that something is being described as an example, illustration, or illustration. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0025] The dual-polarization rain radar signal processing method provided in this application embodiment can be applied to, for example... Figure 1 In the dual-polarization rain radar signal processing system 100 shown, such as Figure 1 As shown, the communication system includes: a digital transceiver system 101, a signal processing system 102, and a monitoring terminal 103.

[0026] The digital transceiver system is used to acquire the dual-polarization echo signal received by the radar antenna and perform preprocessing, quadrature demodulation, and digitization to obtain synchronous IQ data for the horizontal and vertical dual-polarization channels.

[0027] The signal processing system is used to perform software-based parallel processing of synchronous IQ data to generate processing status data packets and final meteorological products.

[0028] The monitoring terminal is used to generate control instructions after hierarchical processing of the processing status data packets; and to dynamically adjust the multi-threaded parallel processing process according to the control instructions.

[0029] To address the technical problems of poor scalability, insufficient flexibility, and inability to adaptively adjust to different scenarios in existing fixed processing architectures, this application provides a dual-polarization rain radar signal processing method. The method includes: acquiring dual-polarization echo signals received by a radar antenna and preprocessing the dual-polarization echo signals; performing orthogonal demodulation and analog-to-digital conversion on the preprocessed dual-polarization echo signals to obtain dual-polarization channel IQ data (horizontal polarization H channel and vertical polarization V channel); performing multi-threaded parallel processing on the dual-polarization channel IQ data to generate a processing status data package and a final meteorological product; the processing status data package includes the processing results of each step in the multi-threaded parallel processing; generating control commands after hierarchical processing of the processing status data package; and dynamically adjusting the multi-threaded parallel processing process according to the control commands. Based on this, the dual-polarization radar signal processing architecture can be dynamically adjusted, achieving scenario-adaptive performance through status feedback and command control, thus overcoming the limitations of traditional fixed architectures.

[0030] like Figure 2 As shown in the embodiment of this application, a dual-polarization rain radar signal processing method includes: S201. Acquire the dual-polarization echo signal received by the radar antenna and preprocess the dual-polarization echo signal.

[0031] The dual-polarization echo signal is obtained by the radar transmitter alternately or simultaneously emitting electromagnetic wave pulses with horizontal polarization H and vertical polarization V, which are then scattered back to the radar receiver after encountering precipitation particles. The dual-polarization radar echo contains intensity, phase, and the amplitude ratio, phase difference, and correlation coefficient between the H and V channel echoes.

[0032] In some implementations, the above preprocessing process includes: pre-selection filtering, low-noise amplification, and down-conversion. The dual-polarized echo signal is pre-filtered by inputting it into a bandpass filter to remove interference outside the operating frequency band, resulting in a clean dual-polarized echo signal. A low-noise amplifier with a noise figure below 3.0 dB amplifies the signal power of the clean dual-polarized echo signal, while simultaneously introducing its own noise, outputting an amplified radio frequency (RF) signal. This amplified RF signal is then multiplied by the local oscillator signal generated internally by the radar in a mixer, yielding a sum-frequency component and an intermediate frequency (IF) component. Finally, a low-pass filter removes the sum-frequency component, resulting in the IF signal.

[0033] It should be noted that the frequency of the local oscillator signal must be extremely stable and precise. If the local oscillator frequency drifts due to radar equipment, it will directly cause the difference frequency output to deviate from the expected intermediate frequency, potentially causing the signal to fall outside the passband of the subsequent intermediate frequency filter, resulting in signal attenuation or distortion. Therefore, in practical system design, it is essential to include a local oscillator frequency stabilization design and a possible automatic frequency control loop.

[0034] For example, consider an X-band dual-polarization weather radar with an operating radio frequency of 9.41 GHz. The pre-selection filter can be a bandpass filter with a center frequency of 9.41 GHz and a bandwidth of 60 MHz. The low-noise amplifier can have a gain of 30 dB and a noise figure of 2.0 dB. The local oscillator frequency is set to 9.35 GHz. After multiplication by the mixer and filtering out the sum frequency, the resulting difference frequency signal is 60 MHz. This 60 MHz signal is the intermediate frequency signal fed into the subsequent quadrature demodulator. These parameters are for illustrative purposes only; in actual design, they can be adjusted according to the radar's band (C, S, X band) and system specifications.

[0035] S202. The preprocessed dual-polarization echo signal is quadrature demodulated and analog-to-digital converted to obtain dual-polarization channel IQ data of horizontal polarization H channel and vertical polarization V channel.

[0036] Quadrature demodulation is used to extract the in-phase I and quadrature Q components of the baseband from the intermediate frequency signal; analog-to-digital conversion converts the in-phase I and quadrature Q components into digital signals; the final dual-polarization channel IQ data are two time-aligned complex sequences, which are the fundamental source for calculating all meteorological base data and dual polarization parameters.

[0037] In some implementations, the quadrature demodulation and analog-to-digital conversion processing is achieved through an integrated digital receiver. The specific process includes: using the same highly stable clock source to drive the RF front-end and local oscillator, and employing the same local oscillator signal source to supply the H and V down-conversion links respectively via a power divider. This ensures that the two signals have a consistent phase reference during the RF-to-IF conversion process, fundamentally guaranteeing the accuracy of subsequent polarization phase measurements. The pre-processed, frequency-uniformed IF H and V analog signals are then input into two structurally identical quadrature demodulation channels. Within each channel, the intermediate frequency signal is split into two paths: one path is mixed with the original local oscillator signal to obtain the in-phase I component, and the other path is mixed with the local oscillator signal after a 90-degree phase shift to obtain the quadrature Q component. Subsequently, the I and Q analog signals pass through an anti-aliasing low-pass filter to remove high-frequency spurious components. Then, a high-speed analog-to-digital converter with multi-channel synchronous sampling capability digitizes the signals under the control of the same sampling clock edge to obtain dual-polarized channel IQ data with horizontally polarized H channel and vertically polarized V channel that are completely synchronized and aligned on the timestamp.

[0038] It should be noted that the shared clock source, shared local oscillator signal, and synchronous sampling emphasized in the above implementation method are the core of ensuring the accuracy of dual polarization measurement. This eliminates random phase errors introduced by device differences and timing jitter between the two receiving channels, making it possible to calculate key polarization quantities such as differential phase. The implementation of this step is not limited to a specific intermediate frequency or ADC resolution; its core lies in generating a dual-channel digital IQ sequence with strict time and phase alignment.

[0039] For example, the intermediate frequency (IF) signal output from the preprocessing is 60MHz. The local oscillator (LO) signal is also 60MHz. In the quadrature demodulator of the H channel, the IF signal is mixed with the original LO signal, and after low-pass filtering, the in-phase component I_h is obtained; it is mixed with the LO signal after a 90° phase shift to obtain the quadrature component Q_h. The V channel undergoes the same processing to obtain I_v and Q_v. A four-channel synchronous ADC synchronously samples and quantizes the four signals I_h, Q_h, I_v, and Q_v at a rate of 100MS / s under a unified clock. For each sampling time n, a pair of complex IQ data is obtained: H[n] = I_h + jQ_h and V[n] = I_v + jQ_v, where j is the imaginary unit. The {H[n], V[n]} sequence is the time-strictly aligned dual-polarization channel IQ data. The parameters such as 60MHz intermediate frequency and 100MS / s sampling rate mentioned here are only for illustrating the working principle and can be changed according to requirements in actual design.

[0040] S203. Perform multi-threaded parallel processing on the dual-polarized channel IQ data to generate processing status data packets and final meteorological products.

[0041] in, Multithreaded parallel processing includes: pulse compression processing, coherent accumulation processing, filtering processing, parameter calculation processing, quality control processing, and parameter configuration processing. The multithreaded parallelism includes: data-level parallelism and pipelined parallelism.

[0042] The processing status data packet refers to a structured data container used to summarize and transmit all key intermediate results from dual-polarization channel IQ data to the final meteorological product during the process of radar completing a signal acquisition and processing in the direction of a beam.

[0043] Final meteorological products refer to data generated after a complete processing chain that can be directly used for meteorological analysis and forecasting, including calibrated and quality-controlled reflectance factors, radial velocity, spectral width, differential reflectance, correlation coefficients, and inverted quantitative precipitation estimates.

[0044] The multi-threaded parallel processing is executed through a dynamic thread pool management and pipeline scheduling architecture. Specifically, a thread pool is created during system initialization, and the entire signal processing flow is modeled as a directed acyclic graph consisting of six sequentially dependent processing stages: pulse compression, coherent accumulation, filtering, parameter calculation, quality control, and parameter configuration. Each processing stage is instantiated as one or more parallelizable task units; the thread allocation and pipeline mapping mechanism is as follows:

[0045] Data-level parallelism: For processing stages where pulse compression, coherent accumulation, and parameter calculation can be performed independently on a large number of range cells, the system divides all range cells acquired in each radar scan into several consecutive data blocks, with each block consisting of several range cells. Worker threads in the thread pool are dynamically allocated these data blocks, and each thread independently and concurrently completes all calculations for that stage on one data block. This means that at the same time, multiple threads may be performing the same pulse compression operation simultaneously, but processing data from different range cell blocks.

[0046] Pipeline Parallelism: The six processing stages described above are arranged in the order of pulse compression, coherent accumulation, filtering, parameter calculation, quality control, and parameter configuration to form a pipeline. Once a data block completes the pulse compression stage, it can be immediately released to the downstream coherent accumulation stage thread for processing, while the upstream pulse compression thread has already begun processing the next data block. This design achieves task-level pipeline parallelism, significantly improving data throughput and hardware resource utilization.

[0047] After processing its assigned data block, each worker thread not only passes the result to the downstream processing stage, but also writes the output of this stage as a record into the shared processing status data packet. The processing status data packet maintains a log sorted by processing timestamp for each distance unit, ensuring that data integrity and causality are maintained even in high-concurrency environments.

[0048] It's important to note that the data block partitioning strategy aims to optimize memory access efficiency and thread load balancing. In one example, the data block size can be set to 256 distance units to reduce cache invalidation. The thread pool manager dynamically allocates data blocks to the earliest idle worker threads based on the number of pending data blocks and the idle state of the thread pool, ensuring that all processing cores are effectively utilized and preventing some threads from becoming idle.

[0049] The processing status data packet can be instantiated in memory as a thread-safe hash table or concurrent dictionary. Its key is a unique identifier for the distance unit (such as a distance library number), and its value is an ordered list or structure indexed by processing stage. Each worker thread, upon writing, updates the field belonging to its current processing stage in the value corresponding to the key of the distance unit it is responsible for through atomic operations or locking mechanisms. This structure ensures that all intermediate results are aggregated by distance unit, and concurrent writes do not overwrite each other. The "log sorted by processing timestamp" can be implemented by appending an atomically incrementing sequence number to each record.

[0050] In one possible implementation of the embodiments of this application, combined with Figure 2 ,like Figure 3 As shown, the above S203 can be specifically implemented through the following S301 to S306, which are explained in detail below: S301, Pulse compression processing; The pulse compression process compresses the dual-channel IQ data in the range dimension by using a matched filtering algorithm on the data blocks allocated by each thread. Specifically, it convolves the IQ sequence of each received pulse with the coefficients of the complex conjugate matched filter of the transmitted waveform. The output is a one-dimensional range image sequence, in which the echo energy of each range cell is concentrated into a narrower time window, significantly improving the range signal-to-noise ratio and resolution. This result is written into the pulse compression output field of the processing status data packet.

[0051] For example, assume the radar transmit pulse is of width The linear frequency modulated pulse, its uncompressed theoretical distance resolution Meters, where c is the speed of light. After pulse compression processing using matched filtering, the effective pulse width is compressed to [value missing]. This improves the actual distance resolution to Meters. Meanwhile, because the matched filter can coherently accumulate the signal energy across the entire pulse width, while noise is incoherent, this processing can bring approximately [a certain amount of energy] to the distance unit. The signal-to-noise ratio processing gain is improved. For an echo from a precipitation target at 10 km, after this step, it will appear as a spike with a significantly higher amplitude than the background noise in the one-dimensional range image sequence of the corresponding range cell in the processed state data packet, laying a high-quality data foundation for subsequent velocity analysis and polarization calculation.

[0052] S302, Coherent accumulation processing; For each range cell, the coherent accumulation processing thread performs a Fast Fourier Transform (FFT) along the time dimension on the one-dimensional range image sequence output from the pulse compression processing. This operation converts the time-domain signal to the Doppler frequency domain, yielding a range-Doppler data matrix. The horizontal axis of this matrix represents the range cell, and the vertical axis represents the Doppler channel, corresponding to the radial velocity. Each cell's value is a complex number, containing both the power and phase information of the signal at that range-velocity cell. This range-Doppler data matrix forms the basis for subsequent velocity information extraction and clutter filtering, and is written into the coherent accumulation output field of the processing status data packet.

[0053] It should be noted that coherent accumulation processing is used to transform information in the time dimension to the Doppler frequency dimension, thereby explicitly separating targets with different radial velocities from the data for the first time. The range-Doppler data matrix is ​​the basis for subsequent extraction of average radial velocity and spectral width, and its clear spectral structure is also a prerequisite for the effective operation of all subsequent clutter filtering algorithms.

[0054] Before performing a Fast Fourier Transform (FFT), a window function is typically applied to the one-dimensional range image sequence to suppress spectral leakage and improve the accuracy of velocity estimation. The number of points in the FFT is generally equal to or greater than the number of pulses within a coherent processing interval, which determines the number of Doppler velocity channels and the velocity resolution.

[0055] For example, the radar transmits 64 coherent pulses in a beam direction. For a specific range cell, pulse compression processing outputs one-dimensional complex values ​​(I / Q data) of the 64 consecutive pulses in that cell. The coherent accumulation processing thread applies a 64-point FFT to these 64 values, outputting complex results for 64 Doppler channels. These 64 channels correspond to the velocity range from the maximum negative velocity to the maximum positive velocity. Given a radar wavelength λ = 0.05 meters and a pulse repetition frequency (PRF) = 1000 Hz, the velocity resolution is λ. PRF / (2 64)≈0.39m / s, and the complex number of each channel simultaneously contains the signal power and phase of that velocity component.

[0056] S303, Filtering; In the filtering process, before receiving control commands, a preset combination of clutter map filtering and IIR filter filtering is used to suppress clutter in the generated range-Doppler data matrix. After receiving control commands, the control commands include a filtering configuration field, which explicitly specifies the filtering algorithms to be used in this process, their execution order, and the key parameters of each algorithm. Based on the control commands, a combination of three filtering methods—clutter map filtering, IIR filter filtering, and adaptive spectral filtering—is selected. Clutter filtering involves maintaining a background power clutter map corresponding to the radar spatial grid. This clutter map is generated by statistically analyzing the average echo power of each grid cell observed by the radar under clear-sky or precipitation-free conditions over an 8- to 24-hour meteorological cycle. The meteorological cycle aims to ensure the capture of typical stable clutter states at the site under different temperatures and humidity levels, while avoiding contamination of the statistical sample by short-term weather events. The map is dynamically updated in a background task mode using a moving average based on new clear-sky observation data to adapt to seasonal changes or minor environmental variations.

[0057] When processing real-time data, the instantaneous signal power of each range-azimuth cell in the current range-Doppler data matrix in the velocity range near zero velocity is calculated. The background clutter power statistics stored in the grid cell corresponding to the same location in the clutter map. Compare them.

[0058] Decision threshold , where Δ is the offset, set by the system noise level. If If the signal of the current unit is determined to be mainly contributed by static or quasi-static ground clutter, the power value of the signal will be set to zero to suppress it.

[0059] The IIR filter filters the signal, with its stopband center located at zero frequency and its notch width automatically configured according to the current radar pulse repetition frequency. The IIR filter performs recursive calculations on the Doppler spectrum of each range cell to accurately filter out residual clutter components within a velocity range of zero frequency and its vicinity.

[0060] After receiving the control command, the system combines three filtering methods—clutter graph filtering, IIR filter filtering, and adaptive spectral filtering—based on the command content. Adaptive spectral filtering: For each range cell in the range-Doppler data matrix to be processed, the complex spectral data of that range cell across the entire Doppler dimension is extracted. The squared modulus of this spectrum is calculated to obtain the instantaneous power spectrum of that range cell. ,in The Doppler frequency is represented. A three-point moving average is applied to the instantaneous power spectrum to suppress random noise and enhance spectral characteristics. A clutter region centered at the zero Doppler frequency is identified. The power values ​​of several adjacent spectral lines near zero frequency are calculated, and this clutter region is considered the core area where strong static clutter may exist. The search proceeds outward from the left and right sides of this core area. The power value of each spectral line is checked sequentially along the power spectrum towards both negative and positive frequencies. The search objective is to find the inflection point where the power value changes from a decreasing trend to a stable or increasing trend. Specific criteria may include: power value... The power of the preceding spectral line Exceeding a relative threshold set based on the global noise level. The frequencies corresponding to the two inflection points. and The left and right boundaries of the current range cell clutter spectrum are adaptively determined; the power spectrum is... In the interval [ All spectral line power values ​​within the clutter interval are set to zero, thus directly filtering out the clutter component in the frequency domain. To maintain spectral continuity and recover as much of the weak meteorological signal as possible from the removed area, the removed area is interpolated and filled using the meteorological signal power on both sides of the clutter interval. Linear interpolation is typically used, i.e., using boundary points... and The power values ​​of each spectral line within the interval are estimated by connecting the lines. For complex spectra, a similar interpolation is performed on the amplitude.

[0061] After the above processing, a new power spectrum is obtained in which clutter components of the range cell are suppressed and meteorological signals are preserved and restored. After performing the above steps in parallel on all range cells, adaptive spectral filtering of the entire range-Doppler data matrix is ​​completed.

[0062] After filtering, the output is a meteorological signal spectrum with suppressed clutter components, which is then written into the processing status data packet.

[0063] S304, Parameter Calculation and Processing; In this process, parametric calculation and processing are performed on the filtered meteorological signal spectrum with suppressed clutter components. The zero-moment reflectivity factor is obtained by calculating the zero-moment integral of the meteorological signal power spectrum for each range cell, as shown in the calculation formula. We obtain, where, This represents the signal power of the k-th Doppler channel, where M is the total number of Doppler channels. It is a comprehensive calibration coefficient that includes factors such as radar constants and range corrections.

[0064] The radial velocity is obtained by calculating the first-order moment, from the formula. Calculated, λ is the frequency corresponding to the k-th Doppler channel, and λ is the radar wavelength.

[0065] The spectral width is obtained by calculating the second-order moment, and then by using the formula... Calculated, The average frequency is calculated from the first moment.

[0066] Simultaneously, calculating the signal-to-noise ratio (SNR) using the morphological characteristics of the spectrum is fundamental to assessing data reliability, as demonstrated by the calculation formula. The calculation results show that, Let N be the total signal power, and N be the noise power of the receiving channel as measured by the system.

[0067] The Quality Index (SQI) is a composite index that integrates signal-to-noise ratio (SNR) and velocity spectral width information. Its value ranges from 0 to 1; the closer it is to 1, the purer and more reliable the signal. It is calculated using the formula... Calculated results; It is the signal-to-noise ratio, and W is the spectral width. It is a function of spectral width; for a pure sinusoidal signal with a spectral width of zero, yes The function.

[0068] The quality factor CCOR is a correlation coefficient used to measure the coherence between adjacent pulse-echo sequences, reflecting the temporal consistency of the data; it is calculated based on... Value, and preset Comparison was performed with a threshold of 10dB. When calculated Greater than or equal to the preset At the threshold; through the calculation formula calculate Quality factor; among which, It's about data strength. It is the power spectrum; When calculated Less than preset At the threshold; through the calculation formula calculate Quality factor; where C is clutter power and S is signal power. , N represents noise power. It is a statistic obtained by performing a first-order autocorrelation operation on a dual-polarization echo signal.

[0069] Differential reflectance was calculated from the meteorological signal spectra with suppressed clutter components in both the horizontally polarized H-channel and the vertically polarized V-channel. Differential phase shift constant and correlation coefficient Among them, differential reflectivity From the calculation formula Calculated, and Reflectivity factors calculated for the horizontally polarized H-channel and the vertically polarized V-channel, respectively; differential phase shift constant. It is the accumulation of the phase difference between the H and V polarization components of an electromagnetic wave along a two-way propagation path. It is calculated by taking the average phase of the cross spectrum of the complex spectra of the H and V channels for each distance. This is the cumulative value of the phase with distance, and it needs to be de-folded to eliminate phase ambiguity; correlation coefficient From the calculation formula ,in and These are the complex spectral values ​​of the horizontally polarized H channel and the vertically polarized V channel on the k-th Doppler channel, respectively. The summation range is all Doppler channels k on the same distance library, and this calculation should be performed in the main signal spectrum range to avoid noise lowering the correlation coefficient value.

[0070] All of the above parameters together constitute the meteorological base data, which is then written into the processing status data packet.

[0071] It should be noted that the reflectivity calibration coefficient This comprehensive coefficient includes not only radar constants, but also range attenuation corrections, system gain calibration values, and nonlinearity correction factors. The reflectivity calibration coefficient values ​​are determined through absolute calibration experiments before the radar leaves the factory and can be periodically verified during operational use using internal calibration signals.

[0072] The noise power N is not a calculated value, but rather the system noise floor power measured in real time by the receiver during periods without transmitted pulses or when the signal is pointed towards clear skies. It serves as the benchmark for signal-to-noise ratio calculation.

[0073] S305, Quality Control Processing; The quality control process uses the quality factor SQI and quality factor CCOR as core criteria to optimize the meteorological baseline data. A basic threshold is set. Where: reflectivity factor , Differential reflectivity , Correlation coefficient , Effective estimation of radial velocity and spectral width must satisfy... ,in It is a positive coefficient set according to parameter characteristics and application scenarios. A decision threshold is set based on the quality factor SQI and quality factor CCOR values ​​to remove invalid values ​​below the threshold from the meteorological baseline data. This dynamic threshold is not a fixed value. For each meteorological parameter, its effective threshold is... Based on the basic threshold Determined together with the quality factor, and calculated by the formula The calculation yielded the results. For the valid data points retained after dynamic thresholding, to further improve their estimation accuracy and spatial continuity, a quality-weighted fusion process was performed for each valid data point. Assign a confidence weight This weight is directly derived from the quality factor that determines its reliability. Data points with higher signal-to-noise ratios and better data coherence have greater influence in subsequent data fusion.

[0074] In the spatial domain, a weighted average is calculated for the same meteorological parameter. For the target location, the optimized parameter value... From the calculation formula The result of the calculation. Among them, The neighborhood set Ω represents the set of neighboring data points participating in the fusion. This neighborhood set Ω encompasses all valid data points within a 3x3 grid window extending in both the distance and azimuth directions, centered on the target data point. This operation is equivalent to a quality-weighted smoothing filter, which suppresses random noise while maximizing the retention of high-reliability information, thus improving data reliability and consistency. After completing the above steps, the quality-controlled meteorological product data is output and written to the processing status data packet.

[0075] It should be noted that the basic threshold and positive coefficient... The setting is not absolutely unique, but rather a recommended value based on industry-standard practices, the measured noise performance of this radar system, or statistical analysis of a large amount of historical data.

[0076] The core advantage of dynamic thresholding algorithms lies in their adaptability. Higher SQI values ​​result in a more lenient dynamic threshold, helping to preserve weak, true meteorological signals. Lower SQI values ​​result in a stricter dynamic threshold, with all data points below the corresponding threshold being marked as invalid. This allows for more aggressive filtering of noise and unreliable estimates.

[0077] For example, the reflectivity factor is measured on a certain distance library. SQI=0.6, set , =5.0, then =0-5 (1-0.6) = -2.0 dBZ. Since Z(-1.5) > The point (-2.0) was preserved. However, if a fixed threshold of 0 dBZ were used, this weak signal would be incorrectly rejected. This demonstrates the advantage of dynamic thresholding in preserving weak precipitation signals.

[0078] S306, Parameter Configuration Processing; The parameter configuration process utilizes real-time measured system parameters to perform final correction on the quality-controlled meteorological product data. This step includes: A known signal is transmitted through an internal calibration loop. The gain and phase differences between the two channels are measured and compensated. The radar system periodically measures the amplitude gain difference ΔG and phase deviation ΔΦ between the vertically polarized H channel and the vertically polarized V channel; these values ​​are stored as a calibration matrix. During processing, all complex data for the vertically polarized V channel are multiplied by a compensation factor. This allows for the alignment of amplitude and phase between the two channels in the algorithm.

[0079] The bias correction based on the measured signal-to-noise ratio is achieved by subtracting the measured noise power from the signal power in the autocorrelation estimation. and Obtain the corrected signal power estimate Through calculation formula The deviation signal-to-noise ratio values ​​of the horizontally polarized H channel and the vertically polarized V channel were calculated. and Differential reflectance (ZDR) is calculated according to the formula. The correlation coefficient was calculated and corrected. According to the calculation formula The result was obtained through calculation and correction.

[0080] It should be noted that the measured parameter noise power used for correction in the above parameter configuration processing is... and The gain difference ΔG and phase deviation ΔΦ are both derived from real-time measurements of the radar hardware system and provided to the processing unit through an independent monitoring link, ensuring the accuracy and timeliness of the correction.

[0081] Based on the above technical solutions, the scalability and processing efficiency of the system are fundamentally improved, and real-time, intelligent adaptive optimization of the processing process is realized. Through continuous quality control and parameter correction, the accuracy and reliability of the final meteorological products are significantly improved.

[0082] S204. The processing status data packet contains the processing results of each step in the multi-threaded parallel processing. The processing status data packet is then processed hierarchically to generate control instructions.

[0083] The hierarchical processing refers to an intelligent process of progressively extracting information, making decisions, and forming executable instructions from the raw intermediate data carried by the processing status data packets. This process includes at least three logical layers: a feature extraction layer, which calculates key feature parameters reflecting signal quality and weather scenarios from the data packets; an intelligent decision-making layer, which matches and analyzes the feature parameters with rules and templates in a preset knowledge base; and an instruction generation layer, which synthesizes structured control instruction data packets based on the matching results for dynamically adjusting the forward signal processing flow.

[0084] In one possible implementation of the embodiments of this application, combined with Figure 2 ,like Figure 4 As shown, the above S204 can be specifically implemented through the following S401 to S403, which are explained in detail below: S401, Feature extraction: Calculate key feature parameters reflecting signal quality and weather scene from data packets; The monitoring terminal reads the processing status data packets in real time, parses them, and calculates key characteristic parameters. These parameters include quality characteristic parameters, scene characteristic parameters, and system status parameters. Quality characteristic parameters cover the average signal-to-noise ratio (SNR) of the entire scan area or a specific sector, the average signal quality index (SQI), the average correlation coefficient (ρhv), and the spatial distribution statistics of these parameters. Scene characteristic parameters include the vertical gradient of the reflectivity factor, the spatial structural characteristics of reflectivity-radial velocity coupling, the vertical profile of the differential reflectivity (ZDR), and the sharply decreasing region of the correlation coefficient (ρhv). System status parameters include the time consumption statistics for each processing stage in the data packet and the key thresholds used in the processing.

[0085] S402, Intelligent decision-making: Matching and analyzing the feature parameters with rules and templates in a preset knowledge base; The calculated set of feature parameters is input into a preset knowledge base for comparison and matching. The preset knowledge base consists of the following three sets of rules: Quality assessment rules: Define quantitative criteria for classifying data quality levels. The rules include: if the average signal-to-noise ratio is greater than 20 dB and the average SQI is greater than 0.85, the quality level is excellent. The quality level is good if either the average signal-to-noise ratio is between 10 and 20 dB or the average SQI is between 0.7 and 0.85. A quality level is considered poor if either the average signal-to-noise ratio is below 10 dB or the average SQI is below 0.7. Regardless of the level, regions with a correlation coefficient ρhv below 0.85 are marked as having questionable coherence, indicating that subsequent applications should use data from these regions with caution.

[0086] Among them, the scene recognition rule, the criterion for recognizing and triggering the conventional precipitation monitoring mode is: the reflectivity factor has a gentle gradient change in space, the radial velocity field is uniformly and orderly distributed, and the average correlation coefficient ρhv is consistently greater than 0.95.

[0087] The criteria for identifying and triggering a strong convection monitoring mode are: the presence of a strong echo cell with a core reflectivity factor value exceeding 45 dB, a significant increase in the vertical cumulative liquid water content, and the detection of strong radial velocity shear or persistent mesocyclonic vortex features within the cell.

[0088] The criteria for identifying and triggering the quantitative estimation model of heavy rainfall are as follows: in low elevation angle observations, the reflectivity factor continuously exceeds 35dB and has a wide spatial coverage, and the differential propagation phase shift ΦDP shows a stable and significant increasing trend with distance, indicating that there is signal attenuation and phase shift accumulation caused by heavy precipitation.

[0089] The criteria for identifying and triggering a weak precipitation observation pattern are: the average reflectivity factor of a large area is less than 15 dB but higher than the noise level, the average correlation coefficient ρhv is greater than 0.97, and the velocity spectrum is narrow, which is consistent with the characteristics of uniform weak precipitation.

[0090] The control strategy rules establish a mapping relationship from quality levels and identified scenarios to specific control commands. The core mapping rules include: When the quality assessment is poor, a strategy focused on improving the signal-to-noise ratio is triggered, generating instructions to increase the number of coherent accumulation pulses or switch to a high-sensitivity processing mode.

[0091] When the scene recognition module determines that it matches the characteristics of a strong convection monitoring mode, it triggers a strategy focused on improving spatiotemporal resolution and velocity detection accuracy, generating instructions to increase the pulse repetition frequency, switch to a fast scanning strategy, and enable adaptive spectral filtering.

[0092] When the scene recognition module determines that it matches the characteristics of a weak precipitation observation mode, it triggers a strategy centered on maximizing signal detection capability, generating instructions that significantly increase coherent accumulation time and employ a conservative filtering combination that minimizes signal impairment.

[0093] S403, Instruction generation: Based on the matching results, a structured control instruction data packet is synthesized for dynamically adjusting the forward signal processing flow. Based on the results of intelligent decision-making, a structured control instruction data packet is generated. This data packet is a machine-readable command entity defined internally by the system, including: instruction type, parameter set, and conditions or scope under which the instruction takes effect; wherein, The instruction type clearly defines the macroscopic purpose of the instruction and uses predefined enumerated values. Instruction types include: mode switching, used to switch between different rainfall detection modes; filter reconstruction, used to dynamically adjust the combination of filtering algorithms in the signal processing chain; parameter reconfiguration, used to adjust one or more specific signal processing parameters; and quality control, used to modify the judgment threshold in the quality control stage.

[0094] The parameter set is a structured data body tightly bound to the instruction type, containing specific key-value pairs, and used to carry all the executable information of the instruction. Different instruction types have different, fixed structures for their parameter sets: When the instruction type is mode switching, the parameter set is the complete parameter configuration group associated with the selected rainfall detection mode; the complete parameter configuration group includes: Conventional precipitation monitoring mode: suitable for general survey monitoring of large-scale stratiform cloud precipitation. The associated parameter set is as follows: pulse repetition frequency set to 800 Hz; beam scanning strategy to complete volumetric scans at 9 preset elevation angles, with a beam dwell time of 50 milliseconds at each elevation angle; coherent accumulation pulse number set to 64; preset filtering algorithm combination is "clutter map filtering" followed by "IIR filtering"; key quality factor thresholds used in the quality control process are a signal quality index (SQI) greater than 0.7 and a correlation coefficient (ρhv) greater than 0.85.

[0095] Strong Convection Monitoring Mode: Suitable for refined structural observation and early warning of convective storms. The associated parameter set is as follows: pulse repetition frequency set to 1200 Hz; beam scanning strategy employing a denser 14-layer elevation angle, and shortening the single elevation angle dwell time to 30 milliseconds for identified storm areas; coherent accumulation pulse number set to 32; preset filtering algorithm combination of "clutter map filtering" followed by "adaptive spectrum filtering"; key quality factor thresholds used in the quality control process are a signal quality index (SQI) greater than 0.6 and a correlation coefficient (ρhv) greater than 0.75.

[0096] Quantitative Rainfall Estimation Mode: Suitable for high-precision quantitative precipitation estimation in flood warning. Its associated parameter set includes: a pulse repetition frequency set to 600 Hz to extend the maximum detection range; a beam scanning strategy focusing on a low elevation angle layer of 0.5 to 4.0 degrees for refined sector scanning; an increase in the number of coherent accumulation pulses to 128 to optimize the signal-to-noise ratio and phase measurement stability; a preset filtering algorithm combination of "clutter map filtering" followed by "narrow notch IIR filtering"; and key quality factor thresholds used in the quality control process: a signal quality index (SQI) greater than 0.8 and a correlation coefficient (ρhv) greater than 0.90.

[0097] Weak Precipitation Observation Mode: Suitable for monitoring weak weather processes such as drizzle and light snow. The associated parameter set is as follows: pulse repetition frequency set to 1000 Hz; beam scanning strategy reducing the number of elevation angles to 5 layers and extending the dwell time at a single elevation angle to 100 milliseconds; coherent accumulation pulse number significantly increased to 256; the preset filtering algorithm is only "clutter map filtering"; the key quality factor thresholds used in the quality control process are a signal quality index (SQI) greater than 0.5 and a correlation coefficient (ρhv) greater than 0.95.

[0098] When the instruction type is filter reconstruction, the parameter set specifically includes: primary filtering algorithm (its value is one of "clutter map filtering", "IIR filtering", "adaptive spectrum filtering"), secondary filtering algorithm (can be empty or one of the above algorithms), IIR notch width (a value in "meters / second", such as 0.5), and whether to enable spectrum repair (Boolean value).

[0099] When the instruction type is parameter reconfiguration, the parameter set specifically includes: coherent accumulation pulse number (positive integer) and pulse repetition frequency (a value in "Hertz").

[0100] When the instruction type is quality control, the parameter set specifically includes: the signal quality index (SQI) threshold (a floating-point number between 0 and 1), the correlation coefficient (ρhv) threshold (a floating-point number between 0 and 1), and the effective range of differential reflectivity (ZDR) (a positive and negative symmetrical range expressed in "decibels", such as [-1.0, 5.0]).

[0101] Conditions or scope for command effectiveness: Precisely defines the spatiotemporal constraints and timing of command effectiveness in a structured manner. Its standard format includes three fields: effective scope, defining the geographic sector of action by the start / end azimuth angle (degrees) and start / end distance (kilometers); effective time base, defining whether the command takes effect "immediately" or at the beginning of the "next scan cycle"; and duration, defining the number of radar volume scans for which the command remains effective.

[0102] It should be noted that, in addition to the mean, the system also calculates the standard deviation, maximum value, minimum value, and spatial gradient of the spatial distribution statistics of the feature parameters. The spatial distribution statistics of the quality feature parameters, including the standard deviation of the average signal-to-noise ratio in the azimuth and distance dimensions, are used to assess the uniformity of data quality.

[0103] The criteria for determining the region of sharp ρhv decrease are: in the vertical direction, the difference in ρhv between two adjacent distances exceeds 0.1, and the ρhv value after the decrease is lower than 0.85.

[0104] The strong radial velocity shear is quantified by calculating the gradient of the velocity field in the horizontal direction. When the gradient value exceeds 15 meters per second per kilometer, it is determined that there is strong velocity shear. This threshold is based on statistical analysis of a large amount of observation data of severe convective storms and can effectively identify strong wind shear areas related to severe weather such as mesocyclones and gust fronts.

[0105] The intelligent decision-making steps are executed in the order of quality assessment, scene identification, and control strategy mapping. When the scene identification result conflicts with the strategy recommended by the quality assessment, the system prioritizes the strategy that improves data quality, and then activates the targeted scene mode after the quality has improved.

[0106] The preset knowledge base exists in the form of a JSON-formatted configuration file, allowing users to add, delete, or modify rule entries based on local climate characteristics or new observation needs without modifying the core processing code.

[0107] For example, feature extraction of a sector of records with an azimuth angle of 90° to 120° and a distance of 15-50 km showed the following quality characteristics: average signal-to-noise ratio of 18.5 dB, average signal quality index of 0.72, average correlation coefficient of 0.88, spatial standard deviation of signal-to-noise ratio of 5.2 dB, and maximum reflectivity factor of 48 dBZ, located at (azimuth 105°, distance 30 km, elevation 2.0°). Strong mesocyclonic features (shear value up to 25 m / s / km) were detected near this strong echo core. The differential reflectivity (ZDR) showed a columnar feature up to 3.5 dB, indicating the presence of large water droplets or hail, and a surge in vertically accumulated liquid water content (VIL) to 55 kg / m². The average processing time for the coherent accumulation phase in this sector increased by 20% compared to other areas.

[0108] The quality assessment showed an average signal-to-noise ratio of 18.5 dB (between 10 and 20 dB) and an average signal quality index of 0.72 (between 0.7 and 0.85). According to the rules, the quality level was rated as good. However, ρhv(0.88) > 0.85, and the coherence suspicion tag was not triggered. Scene recognition features (strong reflectivity kernel, mesocyclone, high ZDR column, high VIL) perfectly match the criteria for strong convection monitoring modes. They do not conform to the main features of other modes. The system maps control strategies and queries control strategy rules. When the scenario is strong convection, the core strategy is to improve spatiotemporal resolution and velocity detection accuracy. Therefore, the mapping result should generate instructions to switch to the strong convection monitoring mode and associate it with all the parameters of this mode. Based on the above decision, the system generates the following control instruction data package: Generate control command. Command type: "Mode Switching". Parameter set bound to "Strong Convection Monitoring Mode" configuration: pulse repetition frequency 1200Hz; perform dense volumetric scanning within a sector of azimuth 100°-130° and distance 15-40km, with a residence time of 30ms in the storm core area (azimuth 113°-117°, distance 22-28km) and 40ms in other areas; coherent accumulation pulse number 32; filtering combination: clutter map filtering followed by cascaded adaptive spectral filtering; quality factor thresholds SQI>0.6, ρhv>0.75. This command will take effect within the specified range in the next scan cycle and will continue for 3 volumetric scans.

[0109] Final result: In the next scanning cycle, when the radar antenna scans to the 90°-120° azimuth again, the signal processing unit will dynamically load the full set of parameters of the severe convective weather monitoring mode according to this instruction, use a higher PRF and faster scanning to capture storm details, and enable adaptive spectral filtering to optimize data quality, thereby achieving refined and adaptive observation of severe convective weather.

[0110] Based on the above technical solution, by summarizing the intermediate results of the entire signal processing chain into a processing status data packet and performing hierarchical intelligent analysis on it to generate dynamic control instructions, the core technical problem that the traditional hardware fixed processing architecture cannot adaptively adjust processing parameters and modes according to real-time data quality and weather scenarios is solved. This achieves a leap from fixed processes to closed-loop intelligent optimization of perception, decision-making, and execution, significantly improving the detection accuracy and efficiency of radar systems in different precipitation scenarios.

[0111] S205. Dynamically adjust the multi-threaded parallel processing process according to control instructions.

[0112] This step is the execution stage of the closed-loop feedback control mechanism. Its core function is to receive and parse the structured control instruction data packet generated in step S204, and, based on the instruction content, dynamically reconfigure the relevant parameters, algorithm combinations, and working modes in the ongoing or upcoming multi-threaded parallel processing process in real time and with high precision, thereby achieving adaptive optimization of the signal processing flow to the current weather scenario and data quality.

[0113] In one possible implementation of the embodiments of this application, combined with Figure 2 ,like Figure 5 As shown, the above S205 can be implemented by the following S501 to S504, which are explained in detail below: S501, Command Reception and Parsing; In this multi-threaded parallel processing architecture, the main control thread or a dedicated instruction distribution service continuously listens for control instruction channels from the monitoring terminal (i.e., the execution entity of step S204). When a new control instruction data packet is received, it is immediately parsed to extract key fields such as instruction type, parameter set, and effective conditions / range.

[0114] It should be noted that, in a typical embodiment, the control command channel is implemented in software such that the monitoring terminal and the signal processing unit are deployed on the same server or connected through a high-speed local area network, and UDP multicast is used for low-latency and reliable command transmission.

[0115] The parsed valid instruction parameters are immediately updated to a global, atomically accessible runtime configuration table. All worker threads read this table in a lock-free manner at the boundaries of the data blocks they are processing, thereby enabling safe and smooth dynamic switching of processing strategies and avoiding data inconsistencies or thread blocking caused by instruction updates.

[0116] S502, Command Verification and Synchronization; The main control thread verifies the legality and timeliness of the instructions, and synchronizes them to all relevant worker threads or processing stage management modules based on the instructions' "effective time base" (immediately or in the next scan cycle) and "effective range" (specific location / distance sector). For instructions that take effect in the "next scan cycle," the system stores them in a queue of pending instructions, which are then loaded uniformly after the current scan cycle ends.

[0117] S503, Execution of dynamic adjustment; Each processing stage dynamically switches or updates its internal processing logic or parameters based on the received instructions, within a specified effective scope and at a specific time. The specific adjustment method varies depending on the instruction type. For mode switching instructions: The system loads and applies the complete parameter configuration group bound to the instruction parameter set—pulse repetition frequency (PRF), scanning strategy, coherent accumulation pulse count, filter combination, and quality threshold—as a whole configuration set to the data processing pipeline within the scope of the instruction's effect.

[0118] For example, when the strong convection monitoring mode configuration is loaded, the radar RF front end switches the PRF to 1200Hz, the antenna control module adopts the corresponding fast scanning strategy, the signal processing thread pool synchronously adjusts the coherent accumulation pulse number to 32, and sets the filter combination to "clutter map + adaptive spectrum filter".

[0119] For filter reconstruction instructions: The thread in the filter processing stage will dynamically reconstruct its filter chain based on the instruction parameter set.

[0120] For example, if the instruction is {Main filter: "clutter map", Secondary filter: "adaptive spectrum", IIR width: null, Enable spectrum repair: true}, then within the scope of this instruction, the filtering process will be executed according to this new combination, and the original IIR filtering step will be skipped.

[0121] For parameter reconfiguration commands: the single or a few key parameters specified in the command will be updated to the corresponding processing module.

[0122] For example, updating the coherent accumulation pulse number parameter will directly affect the number of points in the FFT operation during the coherent accumulation processing stage; changes to the pulse repetition frequency (PRF) need to be synchronously notified to the pulse compression processing and coherent accumulation processing stages, because the coefficients in their algorithms are related to the PRF.

[0123] In response to quality control instructions: the thread in the quality control processing stage will immediately update its internal dynamic threshold elimination algorithm and judgment criteria in weight calculation based on the new quality factor threshold and parameter effective range, thereby changing the strictness of data screening and optimization.

[0124] S504, Adjustment Confirmation and Status Feedback; After each processing stage completes its dynamic adjustment, it returns a confirmation signal to the main control thread. Simultaneously, key details of the instruction (such as the effective mode and modified parameters) are recorded as metadata in the newly generated processing status data packet, thus forming a complete and traceable "decision-execution" record chain for subsequent analysis and auditing.

[0125] It should be pointed out that, Thread safety and uninterrupted adjustment: All dynamic adjustment operations are designed to be thread-safe. For data blocks currently being processed, the parameters before adjustment are typically used until completion, and the adjusted parameters only apply to newly allocated data blocks. This double-buffering mechanism ensures the continuity and stability of the processing, avoiding data inconsistencies or processing interruptions during adjustment.

[0126] Instruction Priority and Conflict Resolution: The system prioritizes the instruction with the most recent timestamp. When multiple instructions are received for the same processing area or parameter, the later instruction will overwrite the earlier one. Furthermore, as described in S204, when logical conflicts exist between instructions, such as a conflict between improving quality and increasing resolution, the adjustments performed on-site will follow the preset strategy priority in the decision-making layer of S204.

[0127] The connection with the processing status data packets: The dynamic adjustment of S205 is based on the analysis results of the processing status data packets in S204, and the new generation of processing status data packets generated by the adjusted processing flow will serve as the input for the next round of analysis and decision-making. This forms a continuous optimization closed loop of processing, analysis, decision-making, adjustment, and reprocessing.

[0128] For example, referring to the example at the end of S204, after the system generates the "strong convection monitoring mode" switching command for the "azimuth 100°-130°, distance 15-40km" sector, in step S205: The instructions are distributed to the radar antenna servo system, RF front-end, and signal processing thread pool.

[0129] Before the start of the next scan cycle, the antenna controller loads the dense scan strategy for the sector; the RF generator switches the PRF to 1200Hz.

[0130] When the scan enters the instruction sector, the signal processing master thread marks the relevant data processing task with a "strong convection mode" label.

[0131] The coherent accumulation thread in the thread pool that processes the sector data automatically performs FFT using 32 pulses; the filtering thread performs clutter suppression using a combination of "clutter map + adaptive spectral filtering"; and the quality control thread temporarily relaxes the judgment thresholds for SQI and ρhv to 0.6 and 0.75, respectively.

[0132] All these adjustments enable the system to automatically switch to an optimized processing state with higher spatiotemporal resolution, greater emphasis on velocity information fidelity, and tolerance for a certain degree of data quality degradation when facing severe convective storms, thus perfectly adapting to the observation needs of this extreme weather scenario.

[0133] Based on the above technical solutions, a fundamental shift from fixed processing to scenario-based adaptation has been achieved, while supporting local fine-tuning and improving monitoring efficiency under complex weather conditions; thread-safe and uninterrupted adjustment design ensures the continuity and stability of the operation steps.

[0134] In summary, this embodiment, through steps S201 to S205, fully realizes a closed-loop processing system that encompasses physical signal acquisition, intelligent product generation, and possesses self-perception, decision-making, and optimization capabilities. The processing status data packet generated in S203 acts as the system's sensory organ for perceiving its own state; the hierarchical processing in S204 is the brain that analyzes the situation and makes decisions; and the dynamic adjustment in S205 is the limb that executes those decisions. This closed-loop mechanism fundamentally breaks through the limitations of traditional fixed processing architectures, enabling adaptive processing of complex and ever-changing weather scenarios.

[0135] Although this application has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings, disclosure, and appended claims, will understand and implement other variations of the disclosed embodiments in carrying out the claimed application. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude multiple instances. A single processor or other unit can implement several functions listed in the claims. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce good results.

[0136] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of this application as defined by the appended claims, and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from the spirit and scope of this application. Thus, if such modifications and modifications of this application fall within the scope of the claims of this application and their equivalents, this application is also intended to include such modifications and modifications.

Claims

1. A dual-polarization rain radar signal processing method, characterized in that, include: Acquire the dual-polarization echo signal received by the radar antenna, and preprocess the dual-polarization echo signal; The preprocessed dual-polarization echo signal is quadrature demodulated and analog-to-digital converted to obtain dual-polarization channel IQ data of horizontal polarization H channel and vertical polarization V channel; The dual-polarized channel IQ data is processed in parallel using multiple threads to generate a processing status data packet and the final meteorological product; the processing status data packet includes the processing results of each step in the parallel processing. The processing status data packets are processed in a hierarchical manner to generate control commands; The multi-threaded parallel processing process is dynamically adjusted according to control instructions.

2. The dual-polarization rain radar signal processing method according to claim 1, characterized in that, The methods for acquiring the dual-polarization channel IQ data include: The synchronous intermediate frequency signal is obtained by demodulating the dual-polarized echo signal after preprocessing the local oscillator signal using the same clock source. The synchronized intermediate frequency signal is mixed with two orthogonal local oscillator signals from the same source to obtain the I / Q analog signals of the H channel and V channel respectively; where I is the in-phase component and Q is the quadrature component, and the local oscillator signals from the same source include: a local oscillator signal in phase with the main local oscillator and a local oscillator signal orthogonal to the main local oscillator. The I / Q analog signals of the H channel and the I / Q analog signals of the V channel are converted from analog to digital on the same clock edge to obtain the dual-polarized channel IQ data of the horizontally polarized H channel and the vertically polarized V channel.

3. The dual-polarization rain radar signal processing method according to claim 1, characterized in that, The multi-threaded parallel processing includes: pulse compression processing, coherent accumulation processing, filtering processing, parameter calculation processing, quality control processing, and parameter configuration processing; The pulse compression process is used to obtain a one-dimensional range image sequence from the dual-polarized channel IQ data through matched filtering. The coherent accumulation process is used to coherently superimpose one-dimensional range image sequences to obtain a range Doppler data matrix. The filtering process is used to filter the range Doppler data matrix to obtain the meteorological signal spectrum after filtering out clutter components. The parameter calculation and processing are used to calculate the spectral distance, estimate the autocorrelation, and calculate the dual polarization parameters of the meteorological signal spectrum after filtering out clutter components to obtain meteorological base data. The meteorological base data includes: reflectivity factor, radial velocity, spectral width, differential reflectivity, differential phase constant, correlation coefficient, signal-to-noise ratio, quality factor SQI, and quality factor CCOR. The quality control process is used to remove meteorological base data by distance averaging and extract the quality factor SQI and quality factor CCOR control output parameters to obtain quality-controlled meteorological product data. The control output parameters include: setting a decision threshold based on the quality factor SQI and quality factor CCOR values ​​to remove invalid values ​​below the threshold in the meteorological base data, and using the quality factor SQI and quality factor CCOR as weighting factors to perform a weighted average calculation on the meteorological base data after removing invalid values ​​to obtain the quality-controlled meteorological product data; The parameter configuration processing is used to correct noise in the quality-controlled meteorological product data using actual radar measurement parameters and to calibrate the quality-controlled meteorological product data to obtain the final meteorological product. The results of pulse compression processing, coherent accumulation processing, filtering processing, parameter calculation processing, and quality control processing are integrated into a processing status data packet.

4. The dual-polarization rain radar signal processing method according to claim 1, characterized in that, The processing status data packet is processed in a hierarchical manner to generate control commands; the hierarchical processing includes: Feature parameters are extracted and calculated from the processed status data packets. These feature parameters include: signal-to-noise ratio (SNR), average correlation coefficient (ρhv), reflectance gradient, data intensity, power spectrum, and quality factor. The feature parameters are compared and matched with scene rules and quality thresholds in a preset knowledge base; Based on the matching results, control instructions containing specific adjustment actions are generated; the control instructions include: instruction type, parameter set, and conditions or range under which the instruction takes effect.

5. The method according to claim 3, characterized in that, The control command dynamically adjusts the multi-threaded parallel processing process, including: selecting a filtering algorithm, setting a quality factor and switching to the corresponding rainfall detection mode, and adjusting parameter configuration according to the corresponding rainfall detection mode; the filtering algorithm includes: clutter map construction, adaptive spectral filtering algorithm, and IIR filtering algorithm.

6. The method according to claim 5, characterized in that, The quality factors include: SQI quality factor and CCOR quality factor; The SQI quality factor is calculated using the following formula: calculate; It is the signal-to-noise ratio, and W is the spectral width. It is a function of spectral width; for a pure sinusoidal signal with a spectral width of zero, yes The function.

7. The method according to claim 6, characterized in that, The CCOR quality factor is calculated as follows: According to calculation Value, and preset Threshold comparison; When calculated Greater than or equal to the preset At the threshold; through the calculation formula calculate Quality factor; among which, It's about data strength. It is the power spectrum; When calculated Less than preset At the threshold; through the calculation formula calculate Quality factor; where C is clutter power and S is signal power. , N is the noise power. It is a statistic obtained by performing a first-order autocorrelation operation on a dual-polarization echo signal.

8. The method according to claim 5, characterized in that, The step of switching to the corresponding rainfall detection mode includes: selecting one from the predefined rainfall detection modes based on the result of comparing and matching the feature parameters with the scene rules and quality thresholds in the preset knowledge base, and generating the corresponding control command; The predefined rainfall detection modes include: Conventional precipitation monitoring model, severe convection monitoring model, quantitative estimation model for heavy rainfall, and weak precipitation observation model; The rainfall detection mode is used to associate different sets of parameters; the set of parameters includes: pulse repetition frequency, beam scanning strategy, coherent accumulation pulse number, filtering algorithm preset for the selected rainfall detection mode, and quality factor threshold.

9. The method according to claim 7, characterized in that, The parameter adjustment configuration includes: When calculated Less than preset At the threshold, the quality-controlled meteorological product data is corrected for deviation based on the noise power N measured by actual radar. Gain and phase compensation corrections are applied to the relative intensity and phase deviations between the horizontally polarized H channel and the vertically polarized V channel. The deviation correction based on noise power includes: The corrected signal power estimate is obtained by subtracting the noise power from the autocorrelation estimate. Through calculation formula The bias signal-to-noise ratio values ​​of the horizontally polarized H channel and the vertically polarized V channel were calculated respectively. and ; The differential reflectivity ZDR is calculated according to the formula. The calculation and correction were obtained. The correlation coefficient According to the calculation formula The result was obtained through calculation and correction.

10. A dual-polarization rain radar signal processing system, characterized in that, include: Digital transceiver systems, software-based signal processing systems, and monitoring terminal systems; The digital transceiver system is used to acquire the dual-polarization echo signal received by the radar antenna and perform preprocessing, quadrature demodulation and digitization to obtain synchronous IQ data for the horizontal and vertical dual-polarization channels. The signal processing system is used to perform software-based parallel processing on the synchronous IQ data to generate a processing status data packet and the final meteorological product. The monitoring terminal system is used to generate control instructions after hierarchical processing of the processing status data packets; and to dynamically adjust the multi-threaded parallel processing process according to the control instructions.

Images