Algorithm and system for avoiding water surface floating objects and improving monitoring precision based on radar water level monitoring

Abstract

The invention relates to the technical field of intelligent sensing systems, in particular to an algorithm and system for avoiding water surface floating objects and improving monitoring precision based on radar water level monitoring, and the method comprises the following steps: collecting radar echoes, constructing spatial-temporal characteristics based on waveform asymmetry and coherent attenuation gradient, and calculating the spatial-temporal characteristics; according to the method, the central moments of the front edge and the rear edge of the echo are calculated, the waveform asymmetry is quantified, the dynamic stability is analyzed in combination with the attenuation gradient of the multi-time-lag coherence coefficient, the floating object interference is recognized from the double dimensions of the spatial form and the time memorability, and the actual water level is inverted. An effective scattering trailing interval is positioned, nonlinear approximation operation is performed on trailing data in combination with an index physical model, a shielded water surface energy distribution curve is reconstructed, a derivative zero point is solved, distance measurement deviation caused by floating objects is corrected, the real water level height is effectively restored, and the monitoring anti-interference capability and the data accuracy in a complex hydrological environment are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent sensing system technology, and in particular to an algorithm and system for improving monitoring accuracy by avoiding floating objects on the water surface based on radar water level monitoring. Background Technology

[0002] The field of intelligent sensing system technology involves the use of advanced sensors to acquire multi-dimensional environmental information and to achieve accurate perception and intelligent identification of target states through signal processing and algorithm analysis. The core aspects include high-precision acquisition of environmental data, intelligent filtering of interference signals, extraction and calculation of target features, and fusion processing of sensor data. By analyzing the feature differences generated by the interaction between sensor signals and the measured environmental medium, target detection and attribute identification in complex environments can be achieved. Among them, the traditional radar water level monitoring system, as a typical non-contact sensing application, refers to a device that uses radar sensors to emit microwave signals vertically towards the water surface and receive reflected echoes to sense changes in water level. When conducting hydrological sensing for rivers or reservoirs, a detection beam is emitted through an antenna. When there are floating objects such as aquatic plants, garbage, or floating ice in the sensing path, microwave energy will be directly reflected on the surface of these solid floating objects and captured by the sensor. Existing systems usually calculate the distance based on the time of electromagnetic wave propagation (ToF) and simply select the earliest arriving first wave signal or the peak signal with the largest amplitude in the echo sequence as the effective sensing data, and then estimate the current water level.

[0003] Traditional radar water level monitoring focuses on extracting the strongest peak or first wave signal of the echo sequence as the basis for ranging. When encountering floating objects that obstruct the water surface, microwave energy is reflected on the solid surface to form a high-intensity echo. Existing logic cannot distinguish the physical difference between the reflection from the solid medium and the liquid surface, which leads to the incorrect locking of the floating object reflection signal as the effective water level signal. This results in the measured distance being less than the actual water surface distance and causes the water level data to be artificially high. In scenarios where floating objects are present, the single peak detection mechanism is difficult to eliminate interference, causing the monitoring data to deviate significantly from the true value and failing to guarantee the accuracy of hydrological data in complex environments. Summary of the Invention

[0004] To achieve the above objectives, the present invention adopts the following technical solution: an algorithm for improving monitoring accuracy by avoiding floating objects on the water surface based on radar water level monitoring, comprising the following steps:

[0005] S1: Acquire radar echo signals and perform time-frequency conversion, output amplitude sequence and complex sequence, locate the main peak position of amplitude sequence and extract the current main peak complex signal, truncate amplitude sequence based on main peak position, construct leading edge sampling set and trailing edge sampling set, and output radar signal dataset;

[0006] S2: Call the radar signal dataset, calculate the higher-order central moments and lower-order central moments of the front and rear edge sampling sets, and generate the leading-edge kurtosis coefficient and trailing-edge kurtosis coefficient. Obtain the asymmetry index based on the coefficient difference, compare the asymmetry index with the preset medium scattering difference benchmark, and generate a spatial morphology identifier.

[0007] S3: Call the radar signal dataset, store the current main peak complex signal into the first-in-first-out queue, calculate the cross-correlation modulus, generate a multi-time-delay coherence coefficient sequence, analyze the changing trend of the sequence with the increase of time interval and calculate the coherence decrease rate, determine the dynamic characteristics of the target, and generate a time stability indicator.

[0008] S4: Analyze the spatial morphology identifier and the temporal stability identifier, filter interference frames, locate the tail start point based on the attenuation ratio of the amplitude sequence, extract the segment of the amplitude sequence from the tail start point to the noise floor, and construct an effective scattering tail dataset.

[0009] S5: Construct an exponential function model, input the effective scattering tail dataset into the nonlinear approximation logic for iterative calculation, minimize the residual by adjusting the model parameters, and output the distance coordinates when the model derivative is zero as the water level height data.

[0010] As a further aspect of the present invention, the radar signal dataset includes a time-domain amplitude value sequence arranged by distance, the corresponding complex signal components of the main peak, and the distance cell index of the main peak. The spatial morphology identifier includes the kurtosis difference state of the waveform's leading and trailing edges, the threshold comparison result of the asymmetry index, and the physical scattering attribute category of the echo signal. The time stability identifier includes the coherence attenuation level of the signal under multiple time delays, the long-range memory characteristic state of the target, and the classification result of steady-state water body and transient interference. The effective scattering tail dataset includes the location index of the tail starting point, continuous amplitude sample values ​​from the starting point to the noise floor, and the energy benchmark value of the background noise. The water level height data specifically refers to the coordinates of the extreme point of the model after parameter optimization, the actual peak position obscured by floating objects, and the vertical distance value of radar ranging.

[0011] As a further aspect of the present invention, the step of acquiring the radar signal dataset specifically includes:

[0012] S101: Acquires radar analog echo signals and performs analog-to-digital conversion, generates digital sampling sequences, performs discrete Fourier transform on the digital sampling sequences, analyzes the energy distribution and phase characteristics in the frequency domain, separates the real and imaginary parts of the frequency domain signal, constructs a complex sequence arranged by range units, calculates the modulus of the complex sequence, generates an amplitude sequence, integrates the amplitude sequence and the complex sequence, and outputs frequency domain transformed signal data.

[0013] S102: Call the frequency domain transformed signal data, traverse the energy values ​​of each distance unit in the amplitude sequence and perform numerical comparison, lock the maximum energy point, obtain the distance axis coordinate index and mark it as the main peak position, locate the corresponding signal unit in the complex sequence based on the main peak position, extract the real part and imaginary part data of the target unit and mark it as the current main peak complex signal, and generate the main peak feature parameters.

[0014] S103: Based on the main peak characteristic parameters, the amplitude sequence is truncated and divided with the main peak position as the boundary. The amplitude values ​​in the interval in front of the main peak position are selected and a leading edge sampling set is constructed. The amplitude values ​​in the interval behind the main peak position are selected and a trailing edge sampling set is constructed to establish a radar signal dataset.

[0015] As a further aspect of the present invention, the step of obtaining the spatial morphology identifier specifically includes:

[0016] S201: Call the radar signal dataset, extract the leading edge sampling set and the trailing edge sampling set, calculate the fourth central moment and the second central moment of the numerical distribution in the two sets respectively, perform a division operation of the fourth central moment divided by the square of the second central moment for each set, obtain the ratio value, and establish the leading edge kurtosis coefficient and trailing edge kurtosis coefficient of the corresponding set.

[0017] S202: Based on the leading edge kurtosis coefficient and the trailing edge kurtosis coefficient, perform a difference operation to calculate the numerical difference between the leading edge kurtosis coefficient and the trailing edge kurtosis coefficient, quantify the degree of asymmetry between the leading and trailing edges of the waveform, and generate an asymmetry index.

[0018] S203: Call the asymmetry index to obtain the preset medium scattering difference benchmark, compare the asymmetry index with the medium scattering difference benchmark, determine the physical scattering properties of the echo corresponding to each distance unit, including specular reflection and diffuse reflection trailing characteristics, and generate spatial morphology identifiers.

[0019] As a further aspect of the present invention, the process of obtaining a preset medium scattering difference benchmark, comparing the asymmetry index with the medium scattering difference benchmark, and determining the physical scattering properties of the echo corresponding to each distance unit specifically comprises:

[0020] A pure water sample library containing several standard specular reflection signals and an interference medium sample library containing several diffuse reflection interference signals are constructed. Central moment calculation is performed on each data sample in the pure water sample library and the interference medium sample library to extract the sample asymmetry index.

[0021] Based on Gaussian statistical distribution logic, the asymmetry index of the two sets of samples is fitted respectively, and the first probability density distribution function corresponding to the pure water sample library and the second probability density distribution function corresponding to the interference medium sample library are calculated.

[0022] Traverse the numerical coordinate axes and search for the numerical intersection point where the function values ​​of the first probability density distribution function and the second probability density distribution function are equal, and set the horizontal axis value corresponding to the numerical intersection point as the medium scattering difference benchmark;

[0023] The asymmetry index of the target under test is called and its numerical value is compared with the medium scattering difference benchmark.

[0024] If the asymmetry index is not greater than the medium scattering difference benchmark, it indicates that the difference in steepness between the leading and trailing edges of the echo signal is in a steady state range. The physical scattering property of the echo corresponding to the distance unit is determined to be the specular reflection feature, and the spatial morphology identifier representing the non-interference state is generated.

[0025] If the asymmetry index is greater than the medium scattering difference benchmark, it indicates that the energy distribution of the echo signal is distorted and trailed due to medium scattering. The physical scattering attribute of the echo corresponding to the distance unit is determined to be the diffuse reflection trailing feature, and the spatial morphology identifier characterizing the interference state is generated.

[0026] As a further aspect of the present invention, the step of obtaining the time stability indicator specifically includes:

[0027] S301: Call the radar signal dataset, extract the current main peak complex signal, store it in a first-in-first-out queue container containing data from multiple time points, calculate the cross-correlation modulus with the current main peak complex signal for the stored signal at each time delay node in the queue container, quantify the correlation between the current signal and past signals, and establish a multi-time delay coherence coefficient sequence.

[0028] S302: Call the multi-delay coherent coefficient sequence, perform first-order difference operation on the coefficient values ​​in the sequence in the order of increasing time delay interval, calculate the numerical decrease between adjacent time delay nodes, perform ratio calculation on the numerical decrease based on the time delay interval, determine the attenuation rate, quantify the memory strength of the echo phase in the time dimension, and generate a long-range coherent attenuation gradient.

[0029] S303: Call the long-range coherent attenuation gradient, obtain the preset time correlation benchmark, compare the long-range coherent attenuation gradient with the time correlation benchmark, determine the dynamic characteristics of the target, and generate a time stability indicator.

[0030] As a further aspect of the present invention, the process of obtaining a preset time correlation benchmark, comparing the long-range coherent attenuation gradient with the time correlation benchmark, and determining the dynamic characteristics of the target specifically includes:

[0031] The pulse repetition frequency parameters of the radar system are obtained, and combined with the hydrodynamic characteristics of gravity waves on the water surface, a theoretical characteristic decoherence time is set to characterize the stability of the water echo phase maintenance.

[0032] A standard exponential decay mathematical model is constructed, and the process by which the cross-correlation modulus decays from the initial normalized peak value to the reciprocal of the natural logarithm base is defined as the theoretical decoherence process. The theoretical decay slope per unit time corresponding to the theoretical decoherence process is calculated, and the theoretical decay slope is set as the time correlation benchmark.

[0033] The long-range coherent attenuation gradient of the target under test is invoked, and the long-range coherent attenuation gradient is numerically compared with the time correlation benchmark.

[0034] If the long-range coherent attenuation gradient is less than the time correlation benchmark, it indicates that the coherent attenuation rate of the echo signal is within a reasonable range of the physical properties of the water body, and the signal phase remains relatively stable over time. The dynamic characteristics of the target are determined to be the steady-state water body characteristics, and the time stability identifier characterizing the effective observation target is generated.

[0035] If the long-range coherent attenuation gradient is not less than the time correlation benchmark, it indicates that the echo signal undergoes rapid decoherence in a short period of time, and the signal phase is affected by random disturbances, resulting in transient abrupt changes. The dynamic characteristics of the target are determined to be the transient interference characteristics, and the time stability identifier characterizing the floating object interference is generated.

[0036] As a further aspect of the present invention, the steps for obtaining the effective scattering tail dataset are as follows:

[0037] S401: Call the spatial morphology identifier and the temporal stability identifier, perform logical association operation on the two sets of identifiers, evaluate the target features, identify whether there is a floating object occlusion state in the current detection period, filter out the signal frames with occlusion, extract the corresponding amplitude sequence and main peak position, and generate interference scene signal data to be corrected.

[0038] S402: Call the interference scene signal data to be corrected, obtain the energy value of the main peak position in the amplitude sequence as a reference, traverse the sampling data of the distance unit after the main peak position, calculate the attenuation ratio of the energy value of each unit relative to the reference, search for the position where the energy intensity drops to the preset attenuation ratio and mark it as the tail start point, and output the tail start positioning index.

[0039] S403: Call the trailing start positioning index and the interference scene signal data to be corrected, locate the distribution position of the background noise base in the amplitude sequence, perform sequence truncation operation with the trailing start positioning index as the starting point and the background noise base position as the ending point, extract continuous amplitude value segments in the interval, remove outlier noise data in the segments, and construct an effective scattering trailing dataset.

[0040] As a further aspect of the present invention, the process of evaluating target features and identifying whether there is a floating object obstruction state within the current detection period specifically includes:

[0041] A multi-dimensional feature joint verification mechanism based on Boolean logic is constructed. The spatial morphology identifier and the temporal stability identifier are extracted. If the spatial morphology identifier is characterized as the diffuse reflection trailing feature and the temporal stability identifier is characterized as the transient interference feature, it is determined that the detected target conforms to the interference property in both the spatial scattering dimension and the temporal phase dimension, and it is determined that there is a floating object occlusion state in the current detection period.

[0042] The process of reducing the search energy intensity to a preset attenuation ratio and marking it as the tailing start point is as follows:

[0043] Based on the physical characteristics of the half-power beamwidth of the radar antenna pattern, the signal node corresponding to the moment when the radar main beam energy is attenuated from the peak to half is defined as the physical boundary between the main echo and the multipath trail, and the value 0.5 is set as the preset attenuation ratio.

[0044] Using the main peak position in the amplitude sequence as the starting anchor point, a one-way traversal operation is performed on the backscattered data along the direction of increasing distance index to calculate the energy ratio between the square of the amplitude value of the current sampling unit and the square of the amplitude value of the main peak position.

[0045] The energy ratio is compared with the preset attenuation ratio. When the energy ratio is not greater than the preset attenuation ratio for the first time, it is determined that the signal strength has attenuated to outside the main lobe range and entered the scattering tail region. The traversal operation is terminated and the distance axis coordinate of the current sampling unit is marked as the tail start positioning index.

[0046] As a further aspect of the present invention, the step of obtaining the water level height data specifically includes:

[0047] S501: Establish an exponential decay mathematical relationship with distance as the independent variable and energy as the dependent variable. Set the amplitude scaling factor, peak position coordinates and medium decay rate as the model parameters to be optimized. Perform initial numerical estimation and assignment for each model parameter. Map the discrete sampling points in the dataset to the coordinate system of the mathematical relationship to establish an initial parameterized exponential decay model.

[0048] S502: Call the initial parameterized exponential decay model and the effective scattering tail dataset, input the effective scattering tail dataset into the nonlinear approximation logic for iterative calculation, and minimize the error between the model curve and the dataset by adjusting the amplitude parameter and decay rate parameter in the decay model to generate the optimal fitting parameter set.

[0049] S503: Based on the optimal fitting parameter set, obtain the parameter combination in the convergent state, calculate the distance variable corresponding to the zero point of the derivative in the physical decay model after parameter adjustment, and use it as water level height data.

[0050] A system for improving monitoring accuracy by avoiding floating debris on the water surface based on radar water level monitoring includes:

[0051] The signal preprocessing module acquires radar echo signals and performs time-frequency conversion, outputs amplitude sequences and complex sequences, locates the main peak position of the amplitude sequence and extracts the current main peak complex signal, truncates the amplitude sequence based on the main peak position, constructs leading edge sampling set and trailing edge sampling set, and outputs radar signal dataset.

[0052] The waveform feature analysis module calls the radar signal dataset, calculates the higher-order central moments and lower-order central moments of the front and rear edge sampling sets, generates the leading-edge kurtosis coefficient and the trailing-edge kurtosis coefficient, obtains the asymmetry index based on the coefficient difference, compares the asymmetry index with the preset medium scattering difference benchmark, and generates a spatial morphology identifier.

[0053] The phase stability assessment module calls the radar signal dataset, stores the current main peak complex signal into a first-in-first-out queue, calculates the cross-correlation modulus, generates a multi-time-delay coherence coefficient sequence, analyzes the changing trend of the sequence with the increase of time interval and calculates the coherence decrease rate, determines the dynamic characteristics of the target, and generates a time stability indicator.

[0054] The trailing signal extraction module analyzes the spatial morphology identifier and the temporal stability identifier, filters interference frames, locates the trailing start point based on the attenuation ratio of the amplitude sequence, extracts the segment of the amplitude sequence from the trailing start point to the noise floor, and constructs an effective scattering trailing dataset.

[0055] The water level inversion calculation module constructs an exponential function model, inputs the effective scattering tail dataset into a nonlinear approximation logic for iterative calculation, minimizes the residual by adjusting the model parameters, and outputs the distance coordinates when the model derivative is zero, as the water level height data.

[0056] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0057] In this invention, by calculating the center moments of the echo's leading and trailing edges and quantifying the waveform asymmetry, and combining the dynamic stability analysis with the attenuation gradient of the multi-delay coherence coefficient, floating object interference is identified from both spatial morphology and temporal memory dimensions. The effective scattering tail interval is located, and nonlinear approximation calculations are performed on the tail data using an exponential physics model. The energy distribution curve of the obscured water surface is reconstructed, and the zero point of the derivative is solved. The ranging deviation caused by floating objects is corrected, effectively restoring the true water level height and improving the monitoring anti-interference capability and data accuracy in complex hydrological environments. Attached Figure Description

[0058] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0059] Figure 1 This is a schematic diagram of the steps of the present invention;

[0060] Figure 2 This is a detailed schematic diagram of S1 of the present invention;

[0061] Figure 3 This is a detailed schematic diagram of S2 of the present invention;

[0062] Figure 4 This is a detailed schematic diagram of S3 of the present invention;

[0063] Figure 5 This is a detailed schematic diagram of S4 of the present invention;

[0064] Figure 6 This is a detailed schematic diagram of S5 of the present invention;

[0065] Figure 7 This is a system module diagram of the present invention. Detailed Implementation

[0066] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0067] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0068] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0069] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0070] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0071] Please see Figure 1 This invention provides an algorithm for improving monitoring accuracy by avoiding floating objects on the water surface based on radar water level monitoring, comprising the following steps:

[0072] S1: Acquire radar echo signals and perform time-frequency conversion, output amplitude sequence and complex sequence, locate the main peak position of amplitude sequence and extract the current main peak complex signal, truncate amplitude sequence based on main peak position, construct leading edge sampling set and trailing edge sampling set, and output radar signal dataset;

[0073] S2: Call the radar signal dataset, calculate the higher-order central moments and lower-order central moments of the front and rear edge sampling sets, and generate the leading edge kurtosis coefficient and trailing edge kurtosis coefficient. Based on the coefficient difference, obtain the asymmetry index, compare the asymmetry index with the preset medium scattering difference benchmark, and generate spatial morphology identifier.

[0074] S3: Call the radar signal dataset, store the current main peak complex signal into the first-in-first-out queue, calculate the cross-correlation modulus, generate a multi-time-delay coherence coefficient sequence, analyze the changing trend of the sequence with the increase of time interval and calculate the coherence decrease rate, determine the dynamic characteristics of the target, and generate a time stability indicator.

[0075] S4: Analyze spatial morphology and temporal stability indicators, filter interference frames, locate the tail start point based on the attenuation ratio of the amplitude sequence, extract the segment of the amplitude sequence from the tail start point to the noise floor, and construct an effective scattering tail dataset.

[0076] S5: Construct an exponential function model, input the effective scattering tail dataset into the nonlinear approximation logic for iterative calculation, minimize the residual by adjusting the model parameters, and output the distance coordinates when the model derivative is zero as the water level height data.

[0077] The radar signal dataset includes a time-domain amplitude sequence arranged by distance, the corresponding complex signal components of the main peak, the distance cell index of the main peak, spatial morphology identifiers including the kurtosis difference state of the waveform's leading and trailing edges, threshold comparison results of the asymmetry index, and the physical scattering attribute category of the echo signal. Temporal stability identifiers include the coherence attenuation level of the signal under multiple time delays, the long-range memory characteristic state of the target, and the classification results of steady-state water bodies and transient interference. The effective scattering tail dataset includes the location index of the tail starting point, continuous amplitude sample values ​​from the starting point to the noise floor, and the energy benchmark value of the background noise. The water level height data specifically refers to the coordinates of the extreme points of the model after parameter optimization, the actual peak position obscured by floating objects, and the vertical distance value measured by radar ranging.

[0078] Please see Figure 2 The specific steps for obtaining the radar signal dataset are as follows:

[0079] S101: Acquires radar analog echo signals and performs analog-to-digital conversion, generates digital sampling sequences, performs discrete Fourier transform on the digital sampling sequences, analyzes the energy distribution and phase characteristics in the frequency domain, separates the real and imaginary parts of the frequency domain signal, constructs a complex sequence arranged by range units, calculates the modulus of the complex sequence, generates an amplitude sequence, integrates the amplitude sequence and the complex sequence, and outputs frequency domain transformed signal data.

[0080] In the signal acquisition link of the radar water level monitoring system, the front-end radio frequency unit is first activated to transmit a frequency-modulated continuous wave (FMCW) signal, where the carrier frequency is set to [value missing]. The scan bandwidth is set to According to the formula Calculations show that the corresponding theoretical distance resolution is (Right now The target echo signal captured by the receiving antenna is demodulated by a mixer and output as an intermediate frequency analog signal. This analog signal is then converted to an analog-to-digital converter (ADC) for further processing. The sampling rate is discretized to generate a length of [value missing]. A point-domain digital sampling sequence. Subsequently, a digital signal processor (DSP) applies a Hanning window to this sequence to suppress spectral sidelobe leakage and performs... The point-discrete Fourier transform (FFT) maps a signal from the time domain to the frequency domain. The processed frequency domain data is then separated into a sequence of real parts. With imaginary part sequence ,in This is the index for the frequency unit (range gate), with a value range of [value range missing]. to For each frequency unit, the processor uses the formula... Calculate its magnitude and generate the corresponding amplitude sequence. and will and Combined storage as a complex sequence For example, in a certain sampling, the first... The real part values ​​at each distance gate are collected as follows: Volt, imaginary part value acquisition is Volt, then the calculated amplitude value at that point is This step ultimately outputs frequency-domain transformed signal data containing complete frequency domain information.

[0081] S102: Call the frequency domain transformed signal data, traverse the energy values ​​of each distance unit in the amplitude sequence and perform numerical comparison, lock the maximum energy point, obtain the distance axis coordinate index and mark it as the main peak position, locate the corresponding signal unit in the complex sequence based on the main peak position, extract the real and imaginary part data of the target unit and mark it as the current main peak complex signal, and generate the main peak characteristic parameters.

[0082] The processor calls the frequency domain transformed signal data and processes the amplitude sequence. Execute the global maximum value search algorithm. Initialize the system's maximum energy value. and corresponding index Then iterate through the indexes. from to In each iteration, if the current amplitude value... Greater than Then update Set the current amplitude value and update. For the current index Suppose that during an actual monitoring session, the system locks the index. The amplitude value at that point is the global maximum value. Volt, then The location of the main wave crest is marked, which physically corresponds to the line-of-sight distance from the radar to the reflector. The system then uses this index... Accessing complex sequences Extract the complex data at that position. This serves as the current dominant complex signal. Simultaneously, the system extracts the signal-to-noise ratio (SNR) parameter of the dominant peak; if the dominant peak energy is lower than the background noise floor... The above are then marked as valid signals, and the extracted main peak position index, peak energy, and complex components are encapsulated as main peak characteristic parameters.

[0083] S103: Based on the characteristic parameters of the main wave peak, the amplitude sequence is truncated and divided with the position of the main wave peak as the boundary. The amplitude values ​​in the interval before the position of the main wave peak are selected and the leading edge sampling set is constructed. The amplitude values ​​in the interval after the position of the main wave peak are selected and the trailing edge sampling set is constructed to establish a radar signal dataset.

[0084] Based on the characteristic parameters of the main peak, the system performs a data window truncation operation centered on the main peak. The sampling window width is set. for Each distance unit, i.e., the intercepted range covers the area in front of the main wave peak. After each point One point. The processor is at the main peak position. Based on this, extract the index range. within Construct a frontier sampling set from the amplitude data points, and simultaneously extract the index range. within A trailing edge sampling set is constructed from several amplitude data points. As shown in Table 1 below, the data in the leading edge sampling set exhibits a steep upward trend, while the data in the trailing edge sampling set shows an asymmetric tailing characteristic due to the diffuse reflection from floating objects. These two sampling sets, along with the main peak position and complex signals, are structured and stored to establish a radar signal dataset.

[0085] Table 1: Radar Echo Signal Amplitude Sampling Data (Partial Example)

[0086] Distance cell index ( ) Amplitude value (V) Regional affiliation Data feature description 445 0.25 Frontier sampling set Low noise level, extremely low energy 448 1.80 Frontier sampling set The signal rises rapidly, with steep edges. 449 3.00 Frontier sampling set Approaching peak 450 3.50 main peak Maximum energy point (detection benchmark) 451 3.20 trailing edge sampling set Energy begins to decline 455 2.40 trailing edge sampling set Slow decay, high energy residue remains 460 1.80 trailing edge sampling set Significant trailing effect, indicating diffuse reflection characteristics

[0087] As shown in Table 1, the trailing edge data (e.g., 1.8V at index 460) shows a significant asymmetry compared to the corresponding position at the leading edge (e.g., close to 0V at index 440, not listed in the figure but conforming to physical laws), indicating that the reflecting surface has a complex non-mirror scattering structure.

[0088] Please see Figure 3 The specific steps for obtaining spatial morphological identifiers are as follows:

[0089] S201: Call the radar signal dataset, extract the leading edge sampling set and the trailing edge sampling set, calculate the fourth central moment and the second central moment of the numerical distribution in the two sets respectively, perform a division operation of the fourth central moment divided by the square of the second central moment for each set, obtain the ratio value, and establish the leading edge kurtosis coefficient and trailing edge kurtosis coefficient of the corresponding set.

[0090] The system calls the radar signal dataset and performs high-order statistical moment operations on the leading-edge and trailing-edge sampling sets respectively. This is for sets containing... A sample set of data points First, calculate its mean. With the second-order central moment (variance) Then the fourth central moments were calculated. Based on this, the system utilizes formulas Calculate the kurtosis coefficient. The calculation is performed based on the distribution characteristics of the data in Table 1: Since the rising edge of the leading edge sample set is steep and clean, its second central moment calculation result is... The fourth-order central moment is Then the leading edge kurtosis coefficient (Conforms to Gaussian distribution characteristics); however, due to the tailing effect, the energy distribution dispersion increases in the subsequent sampling set, and the second-order central moment is calculated as follows: The fourth-order central moment is The trailing kurtosis coefficient was calculated. This step quantifies the morphological features on both sides of the waveform.

[0091] S202: Based on the leading edge kurtosis coefficient and the trailing edge kurtosis coefficient, perform a difference operation to calculate the numerical difference between the leading edge kurtosis coefficient and the trailing edge kurtosis coefficient, quantify the degree of asymmetry of the waveform's leading and trailing edges, and generate an asymmetry index.

[0092] Based on the frontier kurtosis coefficient With trailing kurtosis coefficient The system performs numerical difference operations to quantize the asymmetry of the waveform. The calculation formula is set as follows: Substituting the values ​​from the previous example, the leading edge kurtosis coefficient is: The trailing edge kurtosis coefficient is The asymmetry index is then calculated as follows: This value directly reflects the difference in energy distribution on both sides of the main wave peak of the echo signal. In an ideal specular reflection scenario, the front and rear edges are basically symmetrical, and this exponent approaches... In the diffuse reflection scenario of floating objects, the trailing edge energy is dispersed due to multipath effects, causing the trailing edge kurtosis to deviate significantly from the leading edge kurtosis, resulting in a substantial increase in the exponent. The system will calculate the resulting values. As the asymmetric index of the current range cell echo.

[0093] S203: Call the asymmetry index, obtain the preset medium scattering difference benchmark, compare the asymmetry index with the medium scattering difference benchmark, determine the physical scattering properties of the echo corresponding to each distance unit, including specular reflection and diffuse reflection trailing characteristics, and generate spatial morphology identifiers.

[0094] The asymmetric index is invoked, and the system loads a pre-trained benchmark for medium scattering differences. The process of obtaining this benchmark is as follows: Constructing a system containing... A sample library of pure water body echo data and A sample library of known floating object interference echo data was established. The asymmetry index was calculated for pure water samples, and its value followed a mean of [value missing]. Standard deviation is The statistical distribution of the interference medium sample is calculated, and the asymmetry index is obtained, the value of which follows the mean. Standard deviation is The statistical distribution of the probability density function (PDF) is calculated. The system iterates through the two PDFs, solves for their numerical intersection points, and calculates the horizontal axis value corresponding to the intersection points. Therefore, Set as the benchmark for medium scattering difference. Returning to the real-time processing flow, the system will use the currently calculated asymmetry index. Compared with the benchmark value A comparison was performed. Because... The system determines that the echo signal of the current distance cell has diffuse reflection tail characteristics, that is, the energy distribution has produced significant distortion. Based on this, the system generates a spatial morphological identifier “Diffusive_Tail” to represent the interference state.

[0095] Please see Figure 4 The specific steps for obtaining the time stability indicator are as follows:

[0096] S301: Call the radar signal dataset, extract the current main peak complex signal, store it in a first-in-first-out queue container containing data from multiple time points, calculate the cross-correlation modulus with the current main peak complex signal for the stored signal at each time delay node in the queue container, quantify the correlation between the current signal and past signals, and establish a multi-time delay coherence coefficient sequence.

[0097] The system maintains a depth of [missing information - likely a data set] when accessing the radar signal dataset. First-In-First-Out (FIFO) queue container, used to store consecutive... Each monitoring time (sampling time interval) The main peak complex signal. When the new main peak complex signal... When data enters the queue, the oldest data is removed. The system processes each historical signal in the queue. (in ), calculate its relationship with the current signal The normalized cross-correlation modulus. The calculation formula is as follows: ,in This indicates conjugate complex number operations. Assuming a scenario of a floating object tumbling, the signal phase changes drastically; the previous moment ( The signal is The calculated cross-correlation modulus value decreased to ; and (Lag) At that time, the cross-correlation modulus had decreased to By traversing the queue, the system establishes a list containing... A multi-delay coherence coefficient sequence of data points .

[0098] S302: Call the multi-delay coherent coefficient sequence, perform first-order difference operation on the coefficient values ​​in the sequence in ascending order of time delay interval, calculate the numerical decrease between adjacent time delay nodes, perform ratio calculation on the numerical decrease based on the time delay interval, determine the attenuation rate, quantify the strength of the memory of the echo phase in the time dimension, and generate a long-range coherent attenuation gradient.

[0099] The system calls a multi-time-delay coherence coefficient sequence and calculates the decay slope of the sequence with respect to time delay. The time delay interval is then used. The system processes sequences Perform a first-order difference operation. For example, in the time lag from to Within the interval, the correlation coefficient is from (Autocorrelation) decreased to The total decrease was The system uses the least squares method to process the sequence points. A linear fit is performed, and the absolute value of the slope of the fitted line is extracted as the decay rate. In this example, the slope value obtained from the fit is... (That is, a decrease of 4.0 units per second). This gradient value directly reflects the micro-motion characteristics of the target surface: the floating object is randomly tumbling due to the turbulence of the water flow, resulting in rapid phase decoherence, which manifests as an extremely high decay gradient.

[0100] S303: Call the long-range coherent decay gradient, obtain the preset time correlation benchmark, compare the long-range coherent decay gradient with the time correlation benchmark, determine the dynamic characteristics of the target, and generate a time stability indicator.

[0101] The system invokes the long-range coherent attenuation gradient and loads a time-correlation benchmark calculated based on the physical characteristics of the water surface. The benchmark setting process is as follows: Given the radar pulse repetition frequency and the characteristic decoherence time constant of the water surface gravity wave... Approximately ( Constructing a standard exponential decay model. The cross-correlation modulus value will be changed from decay to The process is defined as a theoretical decoherence process. The average decay slope of this process is calculated to be approximately... (Note: This is a conservative estimate; the actual steady-state water level is usually lower.) To strictly distinguish disturbances, the system sets a more stringent baseline value. In real-time monitoring, the long-range coherent attenuation gradient obtained in step S302 is used. Compared with the benchmark A comparison was performed. Because... This indicates that the signal underwent rapid decoherence in a short period of time, and the signal phase was affected by random disturbances, resulting in transient changes. The dynamic characteristics of the target were determined to be transient interference characteristics, and a time stability identifier “Transient_Interference” was generated to characterize the floating object interference.

[0102] Please see Figure 5 The specific steps for obtaining the effective scattering tail dataset are as follows:

[0103] S401: Call the spatial morphology identifier and the temporal stability identifier, perform logical association operation on the two sets of identifiers, evaluate the target features, identify whether there is a floating object occlusion state in the current detection period, filter out the signal frames with occlusion, extract the corresponding amplitude sequence and main peak position, and generate interference scene signal data to be corrected.

[0104] The system invokes spatial morphology and temporal stability identifiers, performing a Boolean AND operation to identify floating object occlusion. The logical rule is defined as follows: the system outputs a true value (True) if and only if the spatial morphology identifier is "Diffusive_Tail" and the temporal stability identifier is "Transient_Interference," confirming that the main wave peak is occluded by a floating object within the current detection period. As shown in Table 2 below, this joint determination mechanism can effectively distinguish between different scenarios.

[0105] Table 2: Truth Table for Joint Decision-Making Based on Multidimensional Features

[0106] Spatial morphology identifier Time stability indicator Judgment result Explanation of physical meaning Specular (mirror) Stable (steady state) normal water surface Clean water surface, directly use main peak distance measurement Specular (mirror) Transient sudden disturbance Occasional splashes or electronic noise, keep the previous frame. Diffusive Stable (steady state) Static clutter Fixed aquatic plants or shoreline extensions Diffusive Transient Floating objects obstruct Floating debris / icebergs require tailing correction.

[0107] In the current example, the system determines that there is floating object occlusion due to the generation of "Diffusive_Tail" and "Transient_Interference". Subsequently, the system extracts the amplitude sequence data and the main peak position index of the current frame. The data is marked as interference scene signal data to be corrected, and the subsequent trailing repair process is started.

[0108] S402: Call the signal data to be corrected in the interference scene, obtain the energy value of the main peak position in the amplitude sequence as a reference, traverse the sampling data of the distance unit after the main peak position, calculate the attenuation ratio of the energy value of each unit relative to the reference, search for the position where the energy intensity drops to the preset attenuation ratio and mark it as the tail start point, and output the tail start positioning index.

[0109] To retrieve the signal data to be corrected in the interference scenario, the system needs to locate the boundary between the floating object echo and the actual water surface signal. Given that the half-power beamwidth (HPBW) of the radar antenna determines the main lobe energy distribution, the system sets a preset attenuation ratio of... (i.e., the -3dB point). The system starts from the main peak position. Start in the direction of increasing distance index (i.e.) Scan the amplitude sequence point by point. At each position Calculate the energy ratio Known Energy is Assuming in At this point, the amplitude value is Its energy is At this point, the ratio is .because The system determines the signal in The area has attenuated beyond the main beam range and entered the scattering tail region caused by the interaction between the floating object and the water surface. The system immediately stops scanning and indexes... The output is the starting position index of the trail.

[0110] S403: Call the trailing start location index and the interference scene signal data to be corrected, locate the distribution position of the background noise base in the amplitude sequence, perform sequence truncation operation with the trailing start location index as the starting point and the background noise base position as the ending point, extract continuous amplitude value segments within the interval, remove outlier noise data in the segments, and construct an effective scattering trailing dataset.

[0111] Call the tail start positioning index Based on the interference scene and the signal data to be corrected, the system further determines the termination point of the valid data. The system analyzes the histogram distribution of the amplitude sequence, identifying the mode of the low-energy region as the background noise floor level, and measures the value as... The system starts from the tailing point. Continue searching backwards until the amplitude value first falls below (Set as noise floor) (double safety margin), assuming this position is The system extracts an index range from the amplitude sequence. to The system executes an outlier cleaning algorithm within this interval to remove outliers that deviate from the local mean by more than [a certain value]. Isolated noise points of multiple standard deviations are ultimately used to construct a dataset containing This dataset contains an effective scattering tail data set from a sample point. The direct interference components of the main wave peak have been removed, retaining only the physically attenuated tail data carrying water level information.

[0112] Please see Figure 6 The specific steps for obtaining water level data are as follows:

[0113] S501: Establish an exponential decay mathematical relationship with distance as the independent variable and energy as the dependent variable. Set the amplitude scaling factor, peak position coordinates and medium decay rate as the model parameters to be optimized. Perform initial numerical estimation and assignment for each model parameter. Map the discrete sampling points in the dataset to the coordinate system of the mathematical relationship to establish an initial parameterized exponential decay model.

[0114] Based on the effective scattering tail dataset, the system constructs an exponential physical decay model for inverting the true water level. The mathematical relationship is defined as follows: ,in For energy, For distance index, This is the amplitude scaling factor. For the dielectric decay rate, The true peak water level is the location to be determined. This represents the residual noise floor constant. The system first performs initial parameter estimation: assigning the maximum energy value from the dataset to... The starting distance of the dataset Assign to The initial value is estimated based on the average descent slope of the data. for The initial parameter set is set as follows: The system will effectively scatter the tail data in the dataset. discrete sampling points Map the model to its coordinate system to establish an initial parameterized exponential decay model.

[0115] S502: Call the initial parameterized exponential decay model and the effective scattering tail dataset, input the effective scattering tail dataset into the nonlinear approximation logic for iterative calculation, and minimize the error between the model curve and the dataset by adjusting the amplitude parameter and decay rate parameter in the decay model to generate the optimal set of fitting parameters.

[0116] Using the initial parameterized exponential decay model and the effective scattering tail dataset, the system performs nonlinear approximation iterative calculations using the Levenberg-Marquardt (LM) algorithm. In each iteration step, the system substitutes the distance coordinates of the sampling points into the current model to generate theoretical energy predictions and calculates the mean square error (MSE) between the predicted and measured values. As shown in Table 3 below, the parameters are continuously adjusted during the iteration process to minimize the error.

[0117] Table 3: Parameter Variation Table of Nonlinear Approximation Iteration Process

[0118] Number of iterations Amplitude factor Attenuation rate Position parameters Mean Squared Error (MSE) state 1 5.000 0.800 450.00 0.45000 initial 5 5.850 0.780 455.20 0.08200 Optimization in progress 10 6.150 0.755 458.10 0.00530 Optimization in progress 15 6.200 0.750 458.50 0.00012 convergence

[0119] go through After the last iteration, MSE converges to And the rate of change of parameters is lower than the preset tolerance. The iteration terminates. The system outputs the set of optimal fitting parameters in the convergent state. .

[0120] S503: Based on the optimal fitting parameter set, obtain the parameter combination in the convergent state, calculate the distance variable corresponding to the zero point of the derivative in the physical decay model after parameter adjustment, and use it as water level height data;

[0121] Based on the optimal set of fitted parameters, the system analyzes the final water level height data. In the optimized model described above, the parameters... This refers to the distance variable corresponding to the zero derivative of the energy distribution in the physical model (i.e., the theoretical geometric vertex). In a scenario obstructed by floating objects, the primary peak location detected by the original radar is... (Corresponding to the surface of the floating object), while the actual water surface position obtained through tail inversion is (Located below the floating object). The system indexes this distance. Convert to physical distance: Subsequently, the system combined the radar installation height (set to...) ), calculate water level height The results demonstrate that, through model inversion based on tail data, the system successfully corrected the approximately [missing information - likely referring to an error or incomplete sentence] caused by floating objects. ( The ranging error was reduced, and high-precision water level data reflecting the true water surface was output.

[0122] Please see Figure 7 A system for improving monitoring accuracy by avoiding floating debris on the water surface based on radar water level monitoring includes:

[0123] The signal preprocessing module acquires radar echo signals and performs time-frequency conversion, outputs amplitude sequences and complex sequences, locates the main peak position of the amplitude sequence and extracts the current main peak complex signal, truncates the amplitude sequence based on the main peak position, constructs leading edge sampling set and trailing edge sampling set, and outputs radar signal dataset.

[0124] The waveform feature analysis module calls the radar signal dataset, calculates the higher-order and lower-order central moments of the front and rear edge sampling sets, and generates the leading-edge kurtosis coefficient and trailing-edge kurtosis coefficient. Based on the coefficient difference, it obtains the asymmetry index, compares the asymmetry index with the preset medium scattering difference benchmark, and generates a spatial morphology identifier.

[0125] The phase stability assessment module calls the radar signal dataset, stores the current main peak complex signal into the first-in-first-out queue, calculates the cross-correlation modulus, generates a multi-time-delay coherence coefficient sequence, analyzes the changing trend of the sequence with the increase of time interval and calculates the coherence decrease rate, determines the dynamic characteristics of the target, and generates a time stability indicator.

[0126] The trailing signal extraction module analyzes spatial morphology and temporal stability indicators, filters interference frames, locates the trailing start point based on the attenuation ratio of the amplitude sequence, extracts the segment of the amplitude sequence from the trailing start point to the noise floor, and constructs an effective scattering trailing dataset.

[0127] The water level inversion calculation module constructs an exponential function model, inputs the effective scattering tail dataset into a nonlinear approximation logic for iterative calculation, minimizes the residual by adjusting the model parameters, and outputs the distance coordinates when the model derivative is zero, as the water level height data.

[0128] 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. An algorithm for improving monitoring accuracy by avoiding floating debris on the water surface based on radar water level monitoring, characterized in that, Includes the following steps: S1: Acquire radar echo signals and perform time-frequency conversion, output amplitude sequence and complex sequence, locate the main peak position of amplitude sequence and extract the current main peak complex signal, truncate amplitude sequence based on main peak position, construct leading edge sampling set and trailing edge sampling set, and output radar signal dataset; S2: Call the radar signal dataset, calculate the higher-order central moments and lower-order central moments of the front and rear edge sampling sets, and generate the leading-edge kurtosis coefficient and trailing-edge kurtosis coefficient. Obtain the asymmetry index based on the coefficient difference, compare the asymmetry index with the preset medium scattering difference benchmark, and generate a spatial morphology identifier. S3: Call the radar signal dataset, store the current main peak complex signal into the first-in-first-out queue, calculate the cross-correlation modulus, generate a multi-time-delay coherence coefficient sequence, analyze the changing trend of the sequence with the increase of time interval and calculate the coherence decrease rate, determine the dynamic characteristics of the target, and generate a time stability indicator. S4: Analyze the spatial morphology identifier and the temporal stability identifier, filter interference frames, locate the trailing start point based on the attenuation ratio of the amplitude sequence, extract the segment of the amplitude sequence from the trailing start point to the noise floor, and construct an effective scattering trailing dataset. S5: Construct an exponential function model, input the effective scattering tail dataset into the nonlinear approximation logic for iterative calculation, minimize the residual by adjusting the model parameters, and output the distance coordinates when the model derivative is zero as the water level height data.

2. The algorithm for improving monitoring accuracy by avoiding floating objects on the water surface based on radar water level monitoring according to claim 1, characterized in that, The radar signal dataset includes a time-domain amplitude value sequence arranged by distance, the corresponding complex signal components of the main peak, and the distance cell index of the main peak. The spatial morphology identifier includes the kurtosis difference state of the waveform's leading and trailing edges, the threshold comparison result of the asymmetry index, and the physical scattering attribute category of the echo signal. The temporal stability identifier includes the coherence attenuation level of the signal under multiple time delays, the long-range memory characteristic state of the target, and the classification result of steady-state water body and transient interference. The effective scattering tail dataset includes the location index of the tail starting point, continuous amplitude sample values ​​from the starting point to the noise floor, and the energy benchmark value of the background noise. The water level height data specifically refers to the coordinates of the extreme points of the model after parameter optimization, the actual peak position obscured by floating objects, and the vertical distance value of radar ranging.

3. The algorithm for improving monitoring accuracy by avoiding floating objects on the water surface based on radar water level monitoring according to claim 1, characterized in that, The specific steps for acquiring the radar signal dataset are as follows: S101: Acquires radar analog echo signals and performs analog-to-digital conversion, generates digital sampling sequences, performs discrete Fourier transform on the digital sampling sequences, analyzes the energy distribution and phase characteristics in the frequency domain, separates the real and imaginary parts of the frequency domain signal, constructs a complex sequence arranged by range units, calculates the modulus of the complex sequence, generates an amplitude sequence, integrates the amplitude sequence and the complex sequence, and outputs frequency domain transformed signal data. S102: Call the frequency domain transformed signal data, traverse the energy values ​​of each distance unit in the amplitude sequence and perform numerical comparison, lock the maximum energy point, obtain the distance axis coordinate index and mark it as the main peak position, locate the corresponding signal unit in the complex sequence based on the main peak position, extract the real part and imaginary part data of the target unit and mark it as the current main peak complex signal, and generate the main peak feature parameters. S103: Based on the main peak characteristic parameters, the amplitude sequence is truncated and divided with the main peak position as the boundary. The amplitude values ​​in the interval in front of the main peak position are selected and a leading edge sampling set is constructed. The amplitude values ​​in the interval behind the main peak position are selected and a trailing edge sampling set is constructed to establish a radar signal dataset.

4. The algorithm for improving monitoring accuracy by avoiding floating debris on the water surface based on radar water level monitoring according to claim 3, characterized in that, The specific steps for obtaining the spatial morphology identifier are as follows: S201: Call the radar signal dataset, extract the leading edge sampling set and the trailing edge sampling set, calculate the fourth central moment and the second central moment of the numerical distribution in the two sets respectively, perform a division operation of the fourth central moment divided by the square of the second central moment for each set, obtain the ratio value, and establish the leading edge kurtosis coefficient and trailing edge kurtosis coefficient of the corresponding set. S202: Based on the leading edge kurtosis coefficient and the trailing edge kurtosis coefficient, perform a difference operation to calculate the numerical difference between the leading edge kurtosis coefficient and the trailing edge kurtosis coefficient, quantify the degree of asymmetry between the leading and trailing edges of the waveform, and generate an asymmetry index. S203: Call the asymmetry index to obtain the preset medium scattering difference benchmark, compare the asymmetry index with the medium scattering difference benchmark, determine the physical scattering properties of the echo corresponding to each distance unit, including specular reflection and diffuse reflection trailing characteristics, and generate spatial morphology identifiers.

5. The algorithm for improving monitoring accuracy by avoiding floating debris on the water surface based on radar water level monitoring according to claim 4, characterized in that, The process of obtaining a preset medium scattering difference benchmark, comparing the asymmetry index with the medium scattering difference benchmark, and determining the physical scattering properties of the echo corresponding to each range unit is as follows: A pure water sample library containing several standard specular reflection signals and an interference medium sample library containing several diffuse reflection interference signals are constructed. Central moment calculation is performed on each data sample in the pure water sample library and the interference medium sample library to extract the sample asymmetry index. Based on Gaussian statistical distribution logic, the asymmetry index of the two sets of samples is fitted respectively, and the first probability density distribution function corresponding to the pure water sample library and the second probability density distribution function corresponding to the interference medium sample library are calculated. Traverse the numerical coordinate axes and search for the numerical intersection point where the function values ​​of the first probability density distribution function and the second probability density distribution function are equal, and set the horizontal axis value corresponding to the numerical intersection point as the medium scattering difference benchmark; The asymmetry index of the target under test is called and its numerical value is compared with the medium scattering difference benchmark. If the asymmetry index is not greater than the medium scattering difference benchmark, it indicates that the difference in steepness between the leading and trailing edges of the echo signal is in a steady state range. The physical scattering property of the echo corresponding to the distance unit is determined to be the specular reflection feature, and the spatial morphology identifier representing the non-interference state is generated. If the asymmetry index is greater than the medium scattering difference benchmark, it indicates that the energy distribution of the echo signal is distorted and trailed due to medium scattering. The physical scattering attribute of the echo corresponding to the distance unit is determined to be the diffuse reflection trailing feature, and the spatial morphology identifier characterizing the interference state is generated.

6. The algorithm for improving monitoring accuracy by avoiding floating objects on the water surface based on radar water level monitoring according to claim 4, characterized in that, The specific steps for obtaining the time stability identifier are as follows: S301: Call the radar signal dataset, extract the current main peak complex signal, store it in a first-in-first-out queue container containing data from multiple time points, calculate the cross-correlation modulus with the current main peak complex signal for the stored signal at each time delay node in the queue container, quantify the correlation between the current signal and past signals, and establish a multi-time delay coherence coefficient sequence. S302: Call the multi-delay coherent coefficient sequence, perform first-order difference operation on the coefficient values ​​in the sequence in the order of increasing time delay interval, calculate the numerical decrease between adjacent time delay nodes, perform ratio calculation on the numerical decrease based on the time delay interval, determine the attenuation rate, quantify the memory strength of the echo phase in the time dimension, and generate a long-range coherent attenuation gradient. S303: Call the long-range coherent attenuation gradient, obtain the preset time correlation benchmark, compare the long-range coherent attenuation gradient with the time correlation benchmark, determine the dynamic characteristics of the target, and generate a time stability indicator.

7. The algorithm for improving monitoring accuracy by avoiding floating debris on the water surface based on radar water level monitoring according to claim 6, characterized in that, The specific steps for obtaining the effective scattering tail dataset are as follows: S401: Call the spatial morphology identifier and the temporal stability identifier, perform logical association operation on the two sets of identifiers, evaluate the target features, identify whether there is a floating object occlusion state in the current detection period, filter out the signal frames with occlusion, extract the corresponding amplitude sequence and main peak position, and generate interference scene signal data to be corrected. S402: Call the interference scene signal data to be corrected, obtain the energy value of the main peak position in the amplitude sequence as a reference, traverse the sampling data of the distance unit after the main peak position, calculate the attenuation ratio of the energy value of each unit relative to the reference, search for the position where the energy intensity drops to the preset attenuation ratio and mark it as the tail start point, and output the tail start positioning index. S403: Call the trailing start positioning index and the interference scene signal data to be corrected, locate the distribution position of the background noise base in the amplitude sequence, perform sequence truncation operation with the trailing start positioning index as the starting point and the background noise base position as the ending point, extract continuous amplitude value segments in the interval, remove outlier noise data in the segments, and construct an effective scattering trailing dataset.

8. The algorithm for improving monitoring accuracy by avoiding floating objects on the water surface based on radar water level monitoring according to claim 7, characterized in that, The process of evaluating target features and identifying whether there is a floating object obstructing the current detection period is specifically as follows: A multi-dimensional feature joint verification mechanism based on Boolean logic is constructed. The spatial morphology identifier and the temporal stability identifier are extracted. If the spatial morphology identifier is characterized as the diffuse reflection trailing feature and the temporal stability identifier is characterized as the transient interference feature, it is determined that the detected target conforms to the interference property in both the spatial scattering dimension and the temporal phase dimension, and it is determined that there is a floating object occlusion state in the current detection period. The process of reducing the search energy intensity to a preset attenuation ratio and marking it as the tailing start point is as follows: Based on the physical characteristics of the half-power beamwidth of the radar antenna pattern, the signal node corresponding to the moment when the radar main beam energy is attenuated from the peak to half is defined as the physical boundary between the main echo and the multipath trail, and the value 0.5 is set as the preset attenuation ratio. Using the main peak position in the amplitude sequence as the starting anchor point, a one-way traversal operation is performed on the backscattered data along the direction of increasing distance index to calculate the energy ratio between the square of the amplitude value of the current sampling unit and the square of the amplitude value of the main peak position. The energy ratio is compared with the preset attenuation ratio. When the energy ratio is not greater than the preset attenuation ratio for the first time, it is determined that the signal strength has attenuated to outside the main lobe range and entered the scattering tail region. The traversal operation is terminated and the distance axis coordinate of the current sampling unit is marked as the tail start positioning index.

9. The algorithm for improving monitoring accuracy by avoiding floating objects on the water surface based on radar water level monitoring according to claim 7, characterized in that, The specific steps for obtaining the water level data are as follows: S501: Establish an exponential decay mathematical relationship with distance as the independent variable and energy as the dependent variable. Set the amplitude scaling factor, peak position coordinates and medium decay rate as the model parameters to be optimized. Perform initial numerical estimation and assignment for each model parameter. Map the discrete sampling points in the dataset to the coordinate system of the mathematical relationship to establish an initial parameterized exponential decay model. S502: Call the initial parameterized exponential decay model and the effective scattering tail dataset, input the effective scattering tail dataset into the nonlinear approximation logic for iterative calculation, and minimize the error between the model curve and the dataset by adjusting the amplitude parameter and decay rate parameter in the decay model to generate the optimal fitting parameter set. S503: Based on the optimal fitting parameter set, obtain the parameter combination in the convergent state, calculate the distance variable corresponding to the zero point of the derivative in the physical decay model after parameter adjustment, and use it as water level height data.

10. A system for improving monitoring accuracy by avoiding floating debris on the water surface based on radar water level monitoring, characterized in that, The system is used to implement the algorithm for improving monitoring accuracy by avoiding floating objects on the water surface based on radar water level monitoring, as described in any one of claims 1-9, and the system includes: The signal preprocessing module acquires radar echo signals and performs time-frequency conversion, outputs amplitude sequences and complex sequences, locates the main peak position of the amplitude sequence and extracts the current main peak complex signal, truncates the amplitude sequence based on the main peak position, constructs leading edge sampling set and trailing edge sampling set, and outputs radar signal dataset. The waveform feature analysis module calls the radar signal dataset, calculates the higher-order central moments and lower-order central moments of the front and rear edge sampling sets, generates the leading-edge kurtosis coefficient and the trailing-edge kurtosis coefficient, obtains the asymmetry index based on the coefficient difference, compares the asymmetry index with the preset medium scattering difference benchmark, and generates a spatial morphology identifier. The phase stability assessment module calls the radar signal dataset, stores the current main peak complex signal into a first-in-first-out queue, calculates the cross-correlation modulus, generates a multi-time-delay coherence coefficient sequence, analyzes the changing trend of the sequence with the increase of time interval and calculates the coherence decrease rate, determines the dynamic characteristics of the target, and generates a time stability indicator. The trailing signal extraction module analyzes the spatial morphology identifier and the temporal stability identifier, filters interference frames, locates the trailing start point based on the attenuation ratio of the amplitude sequence, extracts the segment of the amplitude sequence from the trailing start point to the noise floor, and constructs an effective scattering trailing dataset. The water level inversion calculation module constructs an exponential function model, inputs the effective scattering tail dataset into a nonlinear approximation logic for iterative calculation, minimizes the residual by adjusting the model parameters, and outputs the distance coordinates when the model derivative is zero, as the water level height data.

Images