A high-precision water level slope measurement method based on complex channel characteristic estimation technology

Abstract

The application discloses a high-precision water level oblique measurement method based on a complex channel feature estimation technology, belongs to the cross technical field of radar signal processing and intelligent hydrological monitoring, and aims to solve the technical problems of weak radar water surface echo intensity and low target feature extraction precision in an oblique measurement scene. The method takes echo complex signals collected by a bankside oblique measurement water level radar as a processing object, sequentially executes six core steps of N-way complex channelization branch filtering, sorting statistical constant false alarm rate channel detection, historical prior target channel matching, affine exponential spectrum enhancement, weighted frequency domain feature extraction and heteroscedastic adaptive smoothing, can accurately separate water surface echo signals in a strong ground object clutter background, and greatly improves the estimation precision and stability of weak target features. The application has excellent anti-interference performance, high frequency domain estimation precision, can be widely applied to water level radar monitoring systems in complex scenes such as high-flood river channels and estuary tides, and provides core algorithm support for a new generation of intelligent hydrological sensing equipment.

Description

Technical Field

[0001] This invention relates to the field of radar signal processing and intelligent hydrological monitoring, and in particular to a high-precision water level oblique measurement method based on complex channel feature estimation technology. Background Technology

[0002] Water level observation is one of the basic hydrological measurement items. Accurate water level observation data is of great significance for watershed flood control and water resource management. For example, the calibration and verification of river hydrodynamic models and flood forecasting models rely on accurate water level observation data. In recent years, as the requirements for hydrological monitoring of small and medium-sized rivers in national flood control and drought relief work have continued to increase, many watersheds and large and medium-sized reservoirs and power stations have gradually established automatic water level monitoring and reporting systems, and water level monitoring is an important component of these systems.

[0003] Currently, the following methods are mainly used for river water level monitoring: (1) Manual observation method. Although this method is simple and easy to implement, it is time-consuming, labor-intensive, and easily affected by environmental conditions. (2) Various contact water level gauges. Contact water level gauges include photoelectric coding, digital coding, and mechanical coding types. They are mainly suitable for water level stations that can build dedicated still water wells, requiring stable riverbanks, minimal riverbed erosion and sedimentation, and low sediment content. (3) Non-contact measurement methods, such as radar and laser water level gauges. This method does not require contact with the water body, but it is easily affected by floating objects. At the same time, radar water level gauges currently use vertical measurement technology, which requires the construction of a long support structure for conventional rivers, making them unsuitable for high flood measurement requirements. Laser water level gauges are also highly susceptible to adverse environmental conditions, and their application is greatly limited under the conditions of emergency high flood measurement.

[0004] In applications involving large flood cross-sections, contact-type measuring equipment is easily affected by water quality and floating debris, and is prone to damage or blockage, resulting in significant measurement errors or data loss, and also impacting the safety of testing personnel. Radar level gauges require long support structures, and assuming the distance must meet the requirements for radar illumination of the water surface, in scenarios with rapid rises and falls in water levels during large flood cross-sections, the radar beam cannot completely illuminate the water surface, leading to a sharp increase in measurement errors, or it may illuminate surrounding fixed objects, causing data distortion. Laser level gauges use a laser beam to illuminate the boundary between the water and the shore, and calculate the water level by the beam pointing angle. Measurements can be performed assuming the location is on the shore, with a simple installation structure and high safety. However, the equipment is highly susceptible to extreme weather conditions, especially during high floods and emergencies, where the performance of the measuring equipment cannot be guaranteed.

[0005] Considering the practical characteristics and difficulties of automatic monitoring of water levels at high flood cross sections, and utilizing modern technology and materials, a versatile, reliable, and easy-to-maintain oblique measurement automatic water level monitoring system was designed and developed. The system can adjust the beam illumination angle according to application requirements to measure water level elevations at different water levels, which is of great significance for monitoring natural disasters and emergency hydrological conditions during high floods.

[0006] However, when using radar technology for oblique water level measurement, the radar's illumination angle is no longer perpendicular, causing a sharp attenuation of the water surface echo intensity. Echoes from nearby ground features, vegetation, and stationary targets all enter the radar system, generating distance characteristics far exceeding those of the target echo at different distances. This increases the difficulty of water level measurement, urgently requiring a high-precision water level feature extraction algorithm to achieve accurate water level feature extraction in complex environments during high floods. Based on these application requirements, and building upon oblique water level radar equipment, a high-precision oblique water level measurement method based on complex channel feature estimation technology is proposed to address the application needs of high-flood, large-section water level measurement.

[0007] In this invention, addressing the challenge of measuring water levels in large-section rivers during high floods, we conducted in-depth research on radar-related theories and, by integrating existing oblique level radar equipment, developed a high-precision oblique level measurement method based on complex channel feature estimation technology. This method utilizes echo signals from oblique level radar equipment deployed along the riverbank, employing complex channelization, affine transformation, Doppler feature estimation, and heteroscedasticity denoising techniques to effectively remove interference features in high-flood river level measurements. This enables high-precision oblique measurement of water level characteristics, thereby achieving automatic, continuous, and high-precision monitoring of water levels under high-flood, large-section conditions. This method exhibits strong anti-interference capabilities, effectively filtering out various strong ground object interference echo signals, enhancing the characteristics of water surface echo signals, and effectively improving the accuracy and stability of water level measurement data. It also significantly expands the application range and operating conditions of oblique level radar, holding significant importance for hydrological monitoring and providing valuable reference for the automatic monitoring of other hydrological elements such as high-flood disaster prevention. Summary of the Invention

[0008] The purpose of this invention is to provide a high-precision oblique water level measurement method based on complex channel feature estimation technology, which solves the technical problems of low frequency domain resolution, strong ground clutter interference, poor accuracy of weak target feature extraction, and insufficient data stability in oblique water level radar echo processing. It enables high-precision identification and feature estimation of water surface echoes under strong clutter background, and improves the measurement accuracy and environmental adaptability of oblique water level radar.

[0009] To achieve the above objectives, the present invention provides the following technical solution: A high-precision water level oblique measurement method based on complex channel feature estimation technology processes the echo complex signal collected by the oblique measurement water level radar equipment deployed on the shore. The water level feature extraction is completed through six core steps: complex channelization processing, channel-level target detection, target channel identification, affine exponential transformation, water level frequency domain feature extraction, and heteroscedasticity smoothing correction. The specific implementation principle of each step is as follows.

[0010] 1. Multichannel Processing: The broadband echo complex signal received by the radar is split into N channels by a power distribution unit. After channel-by-channel sampling delay, downsampling decimation, and split filtering, multichannel frequency domain separation is achieved through joint inverse Fourier transform, and frequency offset correction is performed, ultimately obtaining N narrowband output signals. This step splits the broadband echo into multiple narrowband sub-signals through multichannel parallel processing, improving the frequency resolution and interference suppression capability of subsequent frequency domain processing, and providing a foundation for weak target identification under strong clutter.

[0011] The specific implementation process is as follows: (1) Power allocation: The water level acquisition echo complex signal received by the inclined water level radar is divided into N echo signals with equal power through N power allocation units, where N is the preset total number of complex channel processing channels.

[0012] (2) Sampling delay: The N signals are sampled and delayed sequentially according to the channel number. The first channel is not delayed, the second channel is delayed by 1 sampling unit, and the kth channel is delayed by k-1 sampling units (k=1,2,…,N). The channel is evenly divided by delay in conjunction with subsequent sampling.

[0013] (3) Downsampling: The signal of each channel after the delay is decimated by D times to reduce the signal sampling rate, reduce the amount of computation while retaining the effective bandwidth, and improve the real-time performance of the processing.

[0014] (4) Filter design: Based on the number of channels N and the original sampling rate of the radar echo, design an M-order low-pass filter (M is an even number), determine the passband cutoff frequency and the stopband start frequency, and solve for the filter coefficients.

[0015] (5) Coefficient allocation: The coefficients of the M-order one-dimensional filter are converted from serial to parallel and split into N coefficient sequences, which are then allocated to each channel in sequence.

[0016] (6) Channel filtering: Each channel performs a convolution operation between the extracted signal and the corresponding filter coefficients to complete low-pass filtering and filter out out-of-band noise and aliasing components.

[0017] (7) Joint inverse Fourier transform: Perform N-point inverse Fourier transform on the signal samples of N channels at the same time to complete the channelized frequency domain separation.

[0018] (8) Frequency correction: Multiply the output signal of each channel with the corresponding frequency domain transformation coefficient point by point to correct the frequency offset introduced by the channelization process and obtain the final output signal of N channels.

[0019] 2. Channel-level target detection: The output signals of N channels are transformed in the frequency domain. A constant false alarm rate (CFAR) detection method based on sorting statistics is used to calculate a globally unified frequency domain detection threshold, initially screen out channels with target echoes, eliminate pure noise channels, and compress the amount of data for subsequent processing.

[0020] The specific implementation process is as follows: (1) Frequency domain transformation: Perform fast Fourier transform on the output time domain signal of each channel to obtain the corresponding frequency domain amplitude characteristics.

[0021] (2) Peak extraction and sorting: Extract the maximum value of the frequency domain amplitude of each channel to obtain N peak values, and sort them in descending order of amplitude.

[0022] (3) Threshold calculation: Take the median amplitude value after sorting as the benchmark, and add 5dB margin as the global detection threshold. This threshold can adapt to the background noise level and maintain constant false alarm characteristics.

[0023] (4) Channel decision: If there is a channel in the spectrum whose amplitude exceeds the detection threshold, it is determined to be a channel with target echo, and the corresponding echo signal is output.

[0024] 3. Target Channel Identification: Combining prior information from historical water level data, the water level elevation value is converted into the characteristic frequency of the water surface echo. Among the initially screened target channels, the channel that is closest to the characteristic frequency of the water surface echo is selected as the real water surface target channel, and the remaining target channels are identified as interference channels such as ground objects and vegetation and are eliminated, thus achieving channel-level separation between the target and clutter.

[0025] 4. Affine Exponential Transform: Perform a Fourier transform on the echo signal of the target channel to obtain the spectral amplitude sequence. Perform a nonlinear transformation through affine exponential mapping to amplify the amplitude difference of the spectrum, enhance the spectral characteristics of the weak water surface echo, suppress background noise and weak interference components, and improve the signal-to-noise ratio of subsequent feature extraction.

[0026] The specific implementation process is as follows: (1) Spectrum calculation: Perform fast Fourier transform on the single-frame echo signal of the target channel to obtain the spectrum amplitude sequence.

[0027] (2) Peak value determination: Calculate the maximum value of the spectral amplitude sequence.

[0028] (3) Exponential mapping: Based on the set affine transformation base, perform an exponential transformation on the normalized spectrum amplitude to obtain the exponential amplitude spectrum.

[0029] 5. Water level frequency domain feature extraction: Based on the enhanced exponential amplitude spectrum, the weighted mean and weighted variance of the frequency are calculated by weighted statistical methods. An effective frequency domain range is set to remove outlier interference data, and the weighted mean is recalculated as the frequency domain value corresponding to the water level feature, providing a high-precision frequency domain basis for subsequent water level elevation calculation.

[0030] The specific implementation process is as follows: (1) Weighted average calculation: The weighted average of the corresponding frequency is calculated with the index amplitude as the weight.

[0031] (2) Weighted variance calculation: The weighted variance of the frequency is calculated using the exponential amplitude value as the weight.

[0032] (3) Setting the effective range: The upper and lower boundaries of the effective data are set with the weighted mean as the center and the weighted variance as the center.

[0033] (4) Valid data recalculation: Extract the valid frequency domain data within the boundary, remove outliers, and recalculate the weighted average value as the current water level characteristic frequency domain value.

[0034] 6. Heteroscedasticity smoothing correction: A prediction model is built based on historical water level sequences to identify and replace abnormally fluctuating measurements; then, the water level sequence is filtered and smoothed using adaptive smoothing parameters to reduce random data fluctuations, suppress heteroscedasticity noise, and improve the stability and continuity of measurement results.

[0035] The specific implementation process is as follows: (1) Parameter preset: set the smoothing order L and the smoothing factor α.

[0036] (2) Predicted value calculation: Read L historical water level data and calculate the predicted water level at the current moment by moving average.

[0037] (3) Outlier identification and replacement: Calculate the difference between the measured value and the predicted value. If the difference exceeds the threshold, it is identified as an outlier and replaced with the predicted value.

[0038] (4) Smoothing parameter calculation: Based on the smoothing factor and historical volatility variance, the adaptive smoothing parameter is calculated.

[0039] (5) Smoothing output: The final corrected water level measurement value is obtained through weighted filtering.

[0040] Compared with the prior art, the present invention has the following significant advantages: 1. High frequency domain resolution and strong clutter separation capability.

[0041] This invention employs an N-channel complex channelization processing architecture to split broadband echoes into multiple narrowband sub-signals, significantly improving frequency domain resolution. It can finely distinguish water surface echoes from clutter of nearby ground objects in the frequency domain. Combined with channel-level target detection and prior matching, it can directly eliminate strong ground object interference channels in the channel dimension, effectively solving the core problem of strong clutter submerging weak water surface echoes in oblique measurement scenarios, and significantly improving the anti-interference capability of radar systems.

[0042] 2. High-precision feature estimation improves measurement accuracy.

[0043] This invention enhances spectral characteristics through affine exponential transformation nonlinearity, and combines a frequency domain estimation algorithm with weighted statistics and effective range screening. Compared with the traditional peak detection method, it effectively reduces the impact of picket fence effect and noise interference, and significantly improves the frequency domain estimation accuracy. With the addition of heteroscedastic adaptive smoothing, it can effectively eliminate abnormal jump values ​​and suppress random data jitter. Finally, the water level measurement accuracy can reach the centimeter level, meeting the high precision requirements of hydrological monitoring.

[0044] 3. It has strong adaptability to the environment and can adapt to complex scenarios.

[0045] This invention employs a sorted statistical constant false alarm rate (CFAR) detection and adaptive smoothing parameters, which can automatically adjust the processing parameters according to the background clutter intensity and data fluctuations. No manual threshold configuration is required. It can work stably in complex environments such as high flood levels with many floating objects, complex bank slopes, and heavy rain and dense fog. Its adaptability and robustness are significantly better than traditional fixed parameter algorithms.

[0046] 4. The project is easy to implement and has low deployment costs.

[0047] This method is based on digital signal processing throughout the entire process, without requiring any changes to the radar hardware structure. It can be directly embedded into the signal processing unit of existing oblique measurement water level radars. The equipment maintains the oblique mounting mode on the bank, eliminating the need to build a support structure across the river. This results in low installation and maintenance costs and can be widely applied to various monitoring scenarios such as small and medium-sized rivers, reservoirs, and estuary tide levels. It has significant engineering value for upgrading the next generation of intelligent hydrological sensing equipment. Attached Figure Description

[0048] Figure 1 The flowchart illustrates the high-precision water level oblique measurement method based on complex channel feature estimation technology provided in this embodiment of the invention.

[0049] Figure 2 The flowchart illustrates the implementation of the high-precision water level oblique measurement method based on complex channel feature estimation technology provided in this embodiment of the invention.

[0050] Figure 3 This is the real part and spectral characteristics of the received echo complex signal of the present invention.

[0051] Figure 4 The frequency response of the filter designed for this invention is described.

[0052] Figure 5 This invention provides the time-frequency characteristics of the output signals of each channel after multichannel processing.

[0053] Figure 6 The corresponding complex channel output time-frequency characteristics provided in the embodiments of the present invention.

[0054] Figure 7 The frequency domain diagram of the target detection output signal provided in the embodiment of the present invention (red indicates the channel where the target echo signal is detected).

[0055] Figure 8 The target channel identification result is provided for the embodiments of the present invention.

[0056] Figure 9 A comparison of radar measurement parameters provided for embodiments of the present invention.

[0057] Figure 10 The measurement error of the oblique water level radar provided in the embodiment of the present invention.

[0058] Figure 11 The histogram of the distribution of oblique water level radar provided in the embodiment of the present invention. Detailed Implementation

[0059] To make the technical problems to be solved, the technical solutions, and the beneficial effects of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.

[0060] It should be noted that when a component is referred to as "fixed to" or "set on" another component, it can be directly on or indirectly on that other component. When a component is referred to as "connected to" another component, it can be directly connected to or indirectly connected to that other component.

[0061] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified. "Several" means one or more, unless otherwise explicitly specified.

[0062] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0063] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0064] Please see Figures 1-11 This embodiment applies the method of the present invention to an oblique measurement water level radar system at a tidal observation station at a river estuary, conducting long-term continuous automatic observation of river tide levels to verify the measurement accuracy and environmental adaptability of the method. The oblique measurement water level radar system is installed on a stable foundation on the riverbank, with the radar antenna at a 60° elevation angle to the horizontal plane, and the beam obliquely illuminating the river surface, eliminating the need for a supporting structure across the river. The system employs a 16-channel complex processing architecture (N=16 channels) to perform full-process processing and water level feature extraction of the radar echo complex signal.

[0065] The detailed implementation process of each component module of a high-precision water level incline measurement method based on complex channel feature estimation technology is as follows.

[0066] The oblique measurement water level radar receives the water level acquisition echo complex signal, which is then converted into N echo signals through an N-channel power distribution unit, where N is the number of channels for complex channel processing.

[0067] The signals sent to N complex channels are sequentially delayed according to the channel numbers, with the first channel being undelayed. The second channel is delayed by one sampling unit to obtain The k-th channel is delayed by k-1 sampling units to obtain .

[0068] For the signals of each channel after the delay conduct Decimation is performed by reducing the signal sampling rate to obtain the signals corresponding to each channel after decimation. .

[0069] Based on the number of channels Calculation of sampling rate for inclination level radar echo signal The filter coefficients are calculated based on the following principle.

[0070] Sampling frequency:

[0071] Passband cutoff frequency:

[0072] Stopband start frequency:

[0073] Filter order:

[0074] That is, what is to be sought 1st order filter coefficients.

[0075] Will The coefficients of the first-order filter are subjected to a serial-to-parallel transformation to obtain... The data is sequentially sent to each channel. Among them, the first... The filter coefficients input from each channel are:

[0076] According to the sent number Channel and sampled signal and the corresponding assigned filter coefficients It performs convolution operations on the corresponding channels to achieve filtering of signals from each channel.

[0077] right The signals from each channel are sequentially subjected to a joint inverse Fourier transform according to their channel numbers, corresponding to... time, The signal samples of each channel are as follows: ,in It is the first Channel Signal samples at time 1, for the above Signal samples from each channel are processed Point inverse Fourier transform yields ,in It is the kth frequency domain value of the inverse Fourier transform output, corresponding to the signal output of the kth channel.

[0078] right The signal sequence of the i-th channel is frequency transformed sequentially. For the i-th channel... The frequency domain transform coefficients of the signals in the n channels at time n are:

[0079] The corresponding inverse Fourier transform output signal at time n Multiplying the signal by the frequency domain transformation coefficients yields the output signal of the k-th channel at time n.

[0080]

[0081] Target detection is performed on the signals output from N channels. The detection threshold is calculated according to the following formula, and the channel signals with target echoes are output.

[0082] Perform a Fourier transform on the output signal of each channel to obtain the corresponding frequency domain characteristics.

[0083] Find the maximum value of the frequency domain characteristic of the signal; sort the N maximum values ​​in descending order of amplitude to obtain... .

[0084] Take the amplitude value in the middle of the sorted values, add a 5dB margin, and use it as the frequency domain detection threshold for the output signal of each channel. The threshold calculation principle is as follows:

[0085] Based on the detection threshold, the spectrum of the output signal of each channel is detected, and the echo signal corresponding to the channel where the target exists is output.

[0086] Based on the historical water level data from the previous moment, the data is converted to the frequency of the corresponding signal. According to the value corresponding to the frequency, the channel frequency range containing the target signal in (9) is searched. The channel closest to this frequency is used as the target channel for the corresponding water surface echo, and the remaining channels are converted into interference channels. The channel number is recorded as follows: The corresponding echo signal is .

[0087] Calculate the echo signal as Fourier transform, This represents the number of sample points corresponding to one frame of the processed signal. The corresponding spectral amplitude value is then obtained. The corresponding spectral amplitude values ​​are mapped and transformed according to the following principles.

[0088] Calculate the corresponding spectral amplitude value:

[0089] Calculate the maximum value of the amplitude:

[0090] Calculate the base of the affine transformation:

[0091] Perform exponential transformation to calculate the corresponding exponential amplitude spectrum

[0092] Based on the corresponding exponential amplitude spectrum obtained from the aforementioned exponential transformation, the corresponding water level feature data is extracted. The detailed implementation process is as follows: Calculate the weighted average of the corresponding exponential amplitude spectrum. ,in For the exponential amplitude spectrum The corresponding frequency value.

[0093] Calculate the weighted variance of the corresponding exponential amplitude spectrum:

[0094] Extract the valid data, and calculate the range of valid data values ​​based on the weighted mean and variance. The formulas for calculating the upper and lower boundaries are as follows:

[0095] Based on the boundary calculated above, data falling within the boundary range are extracted as valid data. Following the process in step A, a weighted average of the valid data is calculated and used as the corresponding frequency domain value for the water level feature, denoted as... , Indicates the current observation time.

[0096] Based on the aforementioned measurements, the initial water level value is obtained. To avoid outliers and reduce the fluctuation range of the water level measurement, heteroscedasticity processing is used to correct abnormal measurement values. Simultaneously, historical data is used to filter and smooth the current measurement value to reduce data measurement fluctuations. The detailed implementation principle is as follows: Setting the smoothing order... and smoothing factor .

[0097] Based on the set smoothing order Read historical water level measurement data of the corresponding order.

[0098] Calculate the predicted water level Calculate the current observation value With prediction Interpolation is obtained when the following conditions are met. If the current observed water level is deemed abnormal, the predicted value will be used. Replace the aforementioned measured values .

[0099] Calculate the smoothing parameters:

[0100] Calculate the final water level correction output value based on the smoothing parameters:

[0101] Based on the above calculation principles and procedures, high-precision water level feature extraction of oblique measurement water level echo complex signals can be completed, enabling accurate measurement of water level features.

[0102] The results of practical applications show that the method of the present invention has high accuracy in processing the complex signal of the oblique measurement water level echo, effectively improves the anti-interference ability of the equipment and the measurement accuracy of the water level, greatly improves the ability to measure the water level in complex water areas with high floods, and effectively improves the flood monitoring capability.

[0103] In this invention, the method has strong anti-interference performance, can effectively filter out various strong ground object interference echo signals, enhance the characteristics of water surface echo signals, effectively improve the accuracy and stability of water level measurement data, and effectively improve the application range and usage conditions of oblique measurement water level radar. It is of great significance for hydrological monitoring and has important reference value for the automatic monitoring of other hydrological elements such as flood disaster prevention.

[0104] In summary, this invention effectively solves the problems of weak echo and strong interference in oblique flood level radar under complex high-flood environments through a multi-stage sequential signal processing flow, including complex channelization processing, sorted statistical CFAR detection, prior target channel identification, affine exponent enhancement, weighted feature extraction, and heteroscedasticity smoothing. It achieves high-precision, continuous, and automatic monitoring of water levels, providing reliable technical support for high-flood hydrological monitoring and flood control emergency dispatch. It has significant engineering application value and promising prospects for promotion.

[0105] In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0106] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A high-precision water level oblique measurement method based on complex channel feature estimation technology, characterized in that, The processing of complex echo signals acquired by the shoreline oblique water level radar involves six steps: complex channelization demultiplexing, channel-level constant false alarm rate (CFAR) target detection, target-channel matching and identification, affine exponential spectrum enhancement, water level characteristic frequency domain estimation, and heteroscedasticity adaptive smoothing. S1. Multichannel demultiplexing: The wideband echo complex signal received by the radar is divided into N channels with equal power. The channels are then subjected to channel-by-channel sampling delay, downsampling, demultiplexing low-pass filtering, joint inverse Fourier transform, and frequency offset correction to obtain the output signals of N narrowband channels. S2, Channel-level Constant False Alarm Target Detection: The output signals of N channels are transformed in the frequency domain, and the global frequency domain detection threshold is calculated by sorting and statistics method. Channel signals with target echoes are screened out and pure noise channels are eliminated. S3. Target Channel Matching and Identification: Based on the historical water level data of the previous moment, the characteristic frequency of the water surface echo is calculated. From the selected target channels, the channel with the closest center frequency is matched as the water surface target channel, and the remaining target channels are determined to be ground object interference channels and are eliminated. S4. Affine exponential spectrum enhancement: The echo signal of the water surface target channel is subjected to Fourier transform to obtain the spectrum amplitude sequence. The enhanced exponential amplitude spectrum is obtained by nonlinear transformation through affine exponential mapping, which amplifies the amplitude difference between the water surface echo and the background noise. S5. Frequency domain estimation of water level characteristics: The weighted mean and weighted variance of the frequency are calculated based on the exponential amplitude spectrum. The effective frequency domain range is set and outlier data is removed. The weighted mean of the effective data is recalculated as the high-precision frequency domain value corresponding to the water level characteristics. S6. Adaptive Heteroscedasticity Smoothing: Based on historical water level sequences, a sliding prediction model is constructed to identify and replace abnormal jump values. Then, the water level values ​​calculated in the frequency domain are filtered and smoothed through adaptive smoothing parameters to output the final stable water level measurement results.

2. The high-precision water level oblique measurement method based on complex channel feature estimation technology according to claim 1, characterized in that, The specific channelization demultiplexing process of S1 includes: S101 divides the echo complex signal received by the oblique water level radar into N echo signals with equal power through the N-channel power distribution unit, where N is the preset total number of complex channels. S102 samples and delays the N signals sequentially according to the channel number. The first channel is not delayed, the second channel is delayed by 1 sampling unit, and the kth channel is delayed by k-1 sampling units, where k=1,2,…,N; S103 performs D-fold decimation on each channel signal after the delay to reduce the signal sampling rate and obtain the downsampled discrete signal sequence of each channel. S104 Based on the total number of channels N and the original sampling frequency of the radar, design an M-order low-pass filter, where M is an even number, determine the passband cutoff frequency and the stopband start frequency, and solve for the filter coefficients. S105 performs serial-to-parallel transformation on the M-order filter coefficients, splitting them into N coefficient sequences, which are then distributed to the N channels in sequence. S106 Each channel performs a convolution operation between the extracted discrete signal and the corresponding filter coefficients to complete the low-pass filtering of each channel; S107 performs an N-point inverse Fourier transform on signal samples from N channels at the same time to obtain N frequency domain output values, thus completing the channelization separation of the broadband signal. S108 multiplies the inverse Fourier transform output signal of each channel point by point with the frequency domain transform coefficient of the corresponding channel to correct the frequency offset introduced by channelization, and obtains the final output signal of N channels.

3. The high-precision water level oblique measurement method based on complex channel feature estimation technology according to claim 1, characterized in that, S2's channel-level constant false alarm rate (CFAR) target detection specifically includes: S201 performs a fast Fourier transform on the output time-domain signal of each channel to obtain the frequency domain amplitude spectrum of each channel; S202 extracts the maximum value of the frequency domain amplitude spectrum of each channel, resulting in N peak amplitude values, which are arranged in descending order of amplitude. S203 takes the amplitude value in the middle position after sorting as the reference value, and adds a 5dB margin as the globally unified frequency domain detection threshold; S204 compares the spectrum of each channel with the detection threshold point by point. Channels whose amplitude exceeds the detection threshold in the spectrum are determined to be channels with target echoes, and the echo signal of the corresponding channel is output.

4. The high-precision water level oblique measurement method based on complex channel feature estimation technology according to claim 1, characterized in that, S3 target channel matching identification specifically includes: The stable water level data output at the previous moment is obtained, and combined with the radar installation height, elevation angle and electromagnetic wave wavelength, the Doppler center frequency corresponding to the water surface echo is calculated. Among all the target channels screened in step two, the channel with the smallest absolute value of the difference between the channel center frequency and the Doppler center frequency is selected as the target channel corresponding to the water surface echo. The remaining target channels are determined to be interference channels of ground objects such as banks and vegetation, and their signals are eliminated.

5. The high-precision water level oblique measurement method based on complex channel feature estimation technology according to claim 1, characterized in that, The affine exponential spectrum enhancement of S4 specifically includes: S401 performs a fast Fourier transform on the single-frame echo signal of the target channel to obtain a sequence of spectral amplitude values ​​of corresponding length, the sequence length of which is consistent with the number of sample points processed in a single frame; S402 Iterate through the spectrum amplitude value sequence and find the maximum amplitude value in the sequence; S403 sets the affine transformation base, and performs exponential operations using the ratio of the spectral amplitude value to the maximum amplitude value at each frequency point as the exponent, to obtain the exponential amplitude value corresponding to each frequency point, thus forming the enhanced exponential amplitude spectrum.

6. The high-precision water level oblique measurement method based on complex channel feature estimation technology according to claim 1, characterized in that, The frequency domain estimation of the water level characteristics of S5 specifically includes: S501 uses the exponential amplitude value of each frequency point in the exponential amplitude spectrum as the weight to calculate the weighted average of the corresponding frequency values. S502 uses the exponential amplitude value of each frequency point in the exponential amplitude spectrum as the weight to calculate the weighted variance of the corresponding frequency value relative to the weighted mean. S503 uses the weighted mean as the center and combines it with the weighted variance to set the upper and lower boundaries of the effective frequency domain range; S504 extracts valid data points in the spectrum whose frequency values ​​fall within the upper and lower boundary ranges, removes outliers outside the boundaries, and recalculates the weighted average of the valid data as the frequency domain estimate of the water level characteristics at the current observation time.

7. The high-precision water level oblique measurement method based on complex channel feature estimation technology according to claim 1, characterized in that, S6's heteroscedasticity adaptive smoothing specifically includes: S601 presets the smoothing order L and the smoothing factor α, where L is a positive integer and α takes values ​​in the range of 0 < α < 1; S602 reads historical water level data from L consecutive time periods prior to the current time and calculates the predicted water level for the current time using a moving average. S603 Calculates the absolute value of the difference between the current measured water level and the predicted water level. If the absolute value of the difference exceeds the preset anomaly judgment threshold, the current observed value is judged to be abnormal, and the predicted water level is used to replace the measured value. S604 Calculates adaptive smoothing parameters based on smoothing factor α and the fluctuation variance of historical water level series; S605 performs weighted filtering on the current water level value and the smoothed output value of the previous moment based on the adaptive smoothing parameters, and outputs the final corrected water level measurement result.

8. The high-precision water level oblique measurement method based on complex channel feature estimation technology according to claim 2, characterized in that, The total number of multiple channels N is 16; the oblique water level radar is installed on the riverbank at a 60° elevation angle to the horizontal plane, and the radar beam is obliquely illuminating the water surface, which is suitable for non-contact monitoring of the tidal level of rivers with high flood cross sections and river mouths.

Images