Non-meteorological noise filtering method and device based on millimeter wave cloud radar

Abstract

The invention provides a non-meteorological noise filtering method and device based on a millimeter wave cloud radar. The method comprises the following steps: obtaining original data detected by the millimeter wave cloud radar; the method comprises the following steps: analyzing radar original data, obtaining a meteorological echo and hybrid noise data distribution rule, extracting feature quantities for distinguishing noise, processing the original data based on a rolling sliding window method in combination with a judgment rule of a plurality of feature quantities, and obtaining first de-noised radar data; performing two-dimensional examination on the radar data after the first denoising in time and space, judging the difference between each data point and an adjacent observation value library by library, and obtaining radar data after the second denoising; and carrying out library-by-library convolution processing on the radar data after the second denoising, and obtaining radar data without non-meteorological noise by combining with a meteorological echo continuity principle and utilizing the difference between the current data point and the surrounding data point. Meteorological echo mixed noise is effectively filtered out, so that the technical effect that the capability of detecting weak echoes and clouds in real time by the millimeter wave cloud detection radar is improved is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of radar meteorological echo noise filtering, and in particular to a method and device for filtering non-meteorological noise based on millimeter wave cloud radar. Background Art

[0002] Cloud observation is of great significance to weather forecasting, climate model, cloud-precipitation mechanism research, etc. It is necessary to obtain cloud observation data with high temporal and spatial resolution using effective detection methods. With the continuous development of detection technology, millimeter wave radar has become an important tool for cloud observation.

[0003] Millimeter-wave cloud radar has the advantages of short wavelength, high temporal and spatial resolution, small size, not easily interfered by ground echoes, good ability to observe continuously for a long time, and different observation modes designed for different cloud observation conditions.

[0004] Generally speaking, millimeter-wave cloud radar improves its detection capability for weak echoes by reducing the signal-to-noise ratio. However, while this method improves the radar's detection capability for weak echoes, it also introduces a large amount of noise that is irrelevant to meteorological targets. This noise is often mixed with meteorological echoes such as clouds and precipitation. How to accurately extract meteorological echoes from mixed echoes is of great significance in the application field of radar meteorological echoes.

[0005] However, the inventors of this application found that the above-mentioned prior art has at least the following technical problems in the process of implementing the technical solutions in the embodiments of this application:

[0006] In the existing noise processing method based on fuzzy logic, a large amount of historical data is needed to analyze and extract the physical quantities that distinguish noise from meteorological echoes and the membership functions corresponding to each physical quantity, which results in a large workload of statistical analysis. If the data is insufficient, it is difficult to apply, and the application results are of low accuracy. The membership function parameters of each physical quantity need to be reset in different environments, and the technical problem of not being able to effectively filter out the mixed noise of meteorological echoes. Summary of the invention

[0007] The embodiment of the present invention provides a non-meteorological noise filtering method and device based on millimeter-wave cloud radar, which is used to solve the technical problems that in the noise processing method based on fuzzy logic method in the prior art, a large amount of historical data is needed to analyze and extract the physical quantity that distinguishes noise from meteorological echo and the membership function corresponding to each physical quantity, resulting in a large workload of statistical analysis, and it is difficult to apply if the data is insufficient, the accuracy of the application result is low, the membership function parameters of each physical quantity need to be reset in different environments, and the meteorological echo mixed noise cannot be effectively filtered out. The original data detected by the millimeter-wave radar is used as input, the data distribution law of meteorological echo and mixed noise is analyzed, and the principle of continuity of meteorological echo is combined to achieve effective filtering of meteorological echo mixed noise, thereby improving the technical effect of the millimeter-wave cloud radar's ability to detect weak echoes and clouds in real time.

[0008] In view of the above problems, an embodiment of the present application is proposed to provide a method and device for filtering non-meteorological noise based on millimeter-wave cloud radar.

[0009] In a first aspect, the present invention provides a non-meteorological noise filtering method based on millimeter-wave cloud radar, the method comprising: obtaining raw data detected by millimeter-wave cloud radar, wherein the radar raw data comprises radar basic data such as reflectivity factor, radial velocity, spectral width, and depolarization ratio; statistically analyzing the distribution law and characteristics of meteorological echoes and mixed noise in the radar raw data, extracting characteristic quantities for distinguishing noise, and processing the radar raw data based on a rolling sliding window method and combining judgment rules of multiple characteristic quantities to obtain a first judgment result; according to the first judgment result, removing noise points from the radar raw data to obtain first denoised radar data; based on a rolling sliding window method, A window method is used to perform a two-dimensional check in time and space on the first denoised radar data, and the difference between the radial velocity and the adjacent observation value in each range bin is judged library by library to obtain a second judgment result; according to the second judgment result, the first denoised radar data is subjected to noise removal to obtain the second denoised radar data; the second denoised radar data is subjected to library by library convolution processing, and the difference between the current and surrounding data points is judged in combination with the principle of continuity of meteorological echoes to obtain a third judgment result; according to the third judgment result, the second denoised radar data is subjected to noise removal, and the missing points are supplemented to obtain radar data with non-meteorological noise filtered out.

[0010] Preferably, the amount of the radar raw data is M pieces of vertical detection data (M≥3) or plane scanning data (radial number≥3) at different times.

[0011] Preferably, the statistical analysis of the distribution law and characteristics of the meteorological echo and the mixed noise in the radar raw data, extracting the characteristic quantity for distinguishing the noise, and processing the radar raw data based on the rolling sliding window method in combination with the judgment rules of multiple characteristic quantities to obtain the first judgment result, including:

[0012] Extracting the characteristic quantity for distinguishing noise, wherein the characteristic quantities used are: reflectivity, depolarization ratio, spectrum width, and velocity variation;

[0013] Performing rolling window processing on the radar raw data starting from the first range library, wherein the length of the first sliding window is L, and L≥3;

[0014] Calculate the speed difference between every two adjacent distance bins and calculate the average difference Vcoff average , where every two adjacent distance libraries select all data in both horizontal and vertical directions within the window coverage range;

[0015] Obtaining a first speed difference threshold and a second speed difference threshold;

[0016] Obtaining a spectral width threshold, a depolarization ratio threshold, a first reflectivity threshold, and a second reflectivity threshold;

[0017] Comparing the average difference with the first speed difference threshold and the second speed difference threshold;

[0018] If the average difference is greater than the speed difference first threshold, determining whether the spectrum width is less than a spectrum width threshold;

[0019] If the spectral width is less than the spectral width threshold, the current distance library is determined to be a noise point;

[0020] If the average difference is between the first speed difference threshold and the second speed difference threshold, and the depolarization ratio is greater than the depolarization ratio threshold, and the reflectivity is less than the first reflectivity threshold, it is determined that the current distance library is a noise point;

[0021] If the average difference is between the first speed difference threshold and the second speed difference threshold, and the depolarization ratio is an invalid value, the spectral width is less than the spectral width threshold, and the reflectivity is less than the second reflectivity threshold, the current distance library is determined to be a noise point.

[0022] Preferably, the method of performing a two-dimensional inspection in time and space on the first denoised radar data based on a rolling window method, judging the difference between the radial velocity and the adjacent observation value in each range bin one by one, and obtaining a second judgment result includes:

[0023] Performing sliding window processing on the radar data after the first denoising from the first range bin, wherein the length of the second sliding window is P, where P≥3;

[0024] Get the radial velocity value v of the current distance library i,j , and the radial velocity values ​​of the surrounding N distance bins, where the value of N is determined by the length P of the second sliding window:

[0025] N=P 2 -1

[0026] Assuming P = 3, the window settings are as follows:

[0027] v i-1,j+1 v i,j+1 v i+1,j+1

[0028] v i-1,j v i,j v i+1,j

[0029] v i-1,j-1 v i,j-1 v i+1,j-1

[0030] Where i is the radial data (or horizontal time) direction,

[0031] j is the vertical distance to the reservoir direction,

[0032] v i,j is the radial velocity value of the current altitude layer to be determined;

[0033] Calculate the number of valid speed data in the surrounding N distance libraries;

[0034] Get the threshold of the number of valid speed data;

[0035] Determine whether the number of valid speed data in the surrounding N distance libraries is less than the threshold of the number of valid speed data;

[0036] If the number of valid speed data in the surrounding N distance libraries is less than the valid speed data number threshold, the current distance library is determined to be an invalid value;

[0037] If the number of valid velocity data of the surrounding N distance bins is not less than the valid velocity data threshold, calculating the effective radial velocity average value of the surrounding N distance bins;

[0038] Calculate the radial velocity value v of the current distance library i,j The difference between the effective radial velocity average value and the effective radial velocity average value, wherein the difference calculation formula is:

[0039] n≥preset threshold of valid data number

[0040] Among them, v i is the radial velocity value of the altitude layer to be judged,

[0041] v j is the effective radial velocity value adjacent to it,

[0042] n is the number of valid data;

[0043] Obtaining a radial velocity difference threshold;

[0044] Determining whether the difference is greater than a radial velocity difference threshold;

[0045] If the difference is greater than the radial velocity difference threshold, it is determined that the current database is a noise point.

[0046] Preferably, the method of performing a two-dimensional inspection in time and space on the first denoised radar data based on a rolling window method, judging the difference between the radial velocity and the adjacent observation value in each range bin one by one, and obtaining the second judgment result, further comprises:

[0047] Find the radial velocity value v of the current distance library i,j N distance bins in the vertical direction;

[0048] Find the total number of distance libraries N of the continuous valid values ​​above and below the current distance library total ;

[0049] Get the total library number threshold;

[0050] Determine whether the effective total library number is less than the total library number threshold;

[0051] If the total effective library number is less than the total library number threshold, and the radial velocity v in the adjacent radial direction of the current distance library i-1,j and v i+1,j All the distance libraries are invalid values, and the current distance library is determined to be a noise point.

[0052] Preferably, performing library-by-library convolution processing on the second denoised radar data, combining the principle of continuity of meteorological echoes, judging the difference between the current data point and the surrounding data points, and obtaining a third judgment result, includes:

[0053] Perform library-by-library convolution processing on the second denoised radar data, where the convolution kernel size is Q, Q=2N * +1, N * is a positive integer;

[0054] The median value of the area covered by the convolution kernel is used as the judgment object, and the area covered by the convolution kernel is used as the operation range;

[0055] Calculating the number of invalid values ​​in the area covered by the convolution kernel;

[0056] Get the threshold value of the number of invalid values;

[0057] Determine whether the number of invalid values ​​in the area covered by the convolution kernel is greater than the invalid value number threshold;

[0058] If the number of invalid values ​​in the area covered by the convolution kernel is greater than the invalid value quantity threshold, the middle value of the area covered by the convolution kernel is determined to be an invalid value, and the middle value of the area covered by the convolution kernel is replaced by the invalid value.

[0059] Preferably, performing library-by-library convolution processing on the second denoised radar data, combining the principle of continuity of meteorological echoes, judging the difference between the current data point and the surrounding data points, and obtaining the third judgment result, further includes:

[0060] If the number of invalid values ​​in the area covered by the convolution kernel is not greater than the invalid value number threshold, determining whether the middle value of the area covered by the convolution kernel is a valid value;

[0061] If the middle value of the area covered by the convolution kernel is a valid value, calculate the average value error between the valid middle value and the remaining valid values ​​in the area;

[0062] Determine whether the mean value error satisfies a first formula, wherein the first formula is:

[0063] |Z ij -Z ave |≥3×σ

[0064] Among them, Z ave is the mean of all valid values ​​in the area covered by the convolution kernel,

[0065] σ is the standard deviation of all effective values ​​in the area covered by the convolution kernel;

[0066] If the mean value error satisfies the first formula, the middle value of the area covered by the convolution kernel is determined to be a noise point, and the middle value of the area covered by the convolution kernel is replaced by the median of all valid values ​​of the remaining grid points in the area covered by the convolution kernel;

[0067] If the middle value of the area covered by the convolution kernel is an invalid value, the middle value of the area covered by the convolution kernel is determined to be a missing value, and the middle value of the area covered by the convolution kernel is replaced by the median value of all valid values ​​of the remaining grid points in the area covered by the convolution kernel.

[0068] In a second aspect, the present invention provides a non-meteorological noise filtering device based on millimeter wave cloud radar, the device comprising:

[0069] A first acquisition unit, wherein the first acquisition unit is used to obtain raw data detected by a millimeter-wave cloud radar, wherein the radar raw data includes radar basic data such as reflectivity factor, radial velocity, spectrum width, and depolarization ratio;

[0070] A first processing unit, the first processing unit is used to statistically analyze the distribution law and characteristics of the meteorological echo and the mixed noise in the radar raw data, extract the characteristic quantity for distinguishing the noise, and process the radar raw data based on the rolling sliding window method in combination with the judgment rules of multiple characteristic quantities to obtain a first judgment result;

[0071] a second obtaining unit, the second obtaining unit being used to remove noise points from the radar raw data according to the first judgment result, so as to obtain first denoised radar data;

[0072] A second processing unit, the second processing unit is used to perform a two-dimensional check in time and space on the first denoised radar data based on a rolling window method, and judge the difference between the radial velocity in each range bin and the adjacent observation value, to obtain a second judgment result.

[0073] a third obtaining unit, wherein the third obtaining unit is used to remove noise points from the first denoised radar data according to the second judgment result to obtain second denoised radar data;

[0074] A third processing unit, the third processing unit is used to perform library-by-library convolution processing on the second denoised radar data, and determine the difference between the current data point and the surrounding data point in combination with the continuity principle of the meteorological echo to obtain a third determination result;

[0075] A fourth obtaining unit, the fourth obtaining unit is used to remove noise points from the second denoised radar data according to the third judgment result, and to fill in the missing points to obtain radar data with non-meteorological noise filtered out.

[0076] Preferably, the amount of the radar raw data is M pieces of vertical detection data (M≥3) or plane scanning data (radial number≥3) at different times.

[0077] Preferably, the first processing unit includes:

[0078] A first extraction unit, wherein the first extraction unit is used to extract a feature quantity for distinguishing noise, wherein the feature quantities used are: reflectivity, depolarization ratio, spectrum width, and velocity variation;

[0079] A fourth processing unit, the fourth processing unit is used to perform rolling sliding window processing on the radar raw data starting from the first distance library, wherein the length of the first sliding window is L, L ≥ 3;

[0080] The first calculation unit is used to calculate the speed difference between each two adjacent distance bins and calculate the average difference Vcoff average , where every two adjacent distance libraries select all data in both horizontal and vertical directions within the window coverage range;

[0081] a fifth obtaining unit, the fifth obtaining unit being used to obtain a first speed difference threshold and a second speed difference threshold;

[0082] a sixth obtaining unit, the sixth obtaining unit being used to obtain a spectral width threshold, a depolarization ratio threshold, a first reflectivity threshold, and a second reflectivity threshold;

[0083] a first comparison unit, the first comparison unit being used to compare the average difference value with the first speed difference threshold and the second speed difference threshold;

[0084] a first judging unit, the first judging unit being configured to judge whether the spectrum width is less than a spectrum width threshold value if the average difference value is greater than the speed difference first threshold value;

[0085] A first determination unit, the first determination unit is used to determine that the current distance library is a noise point if the spectrum width is less than the spectrum width threshold;

[0086] a second determination unit, configured to determine that the current distance library is a noise point if the average difference is between the first speed difference threshold and the second speed difference threshold, the depolarization ratio is greater than the depolarization ratio threshold, and the reflectivity is less than the first reflectivity threshold;

[0087] The third determination unit is used to determine that the current distance library is a noise point if the average difference is between the first threshold value of the speed difference and the second threshold value of the speed difference, and the depolarization ratio is an invalid value, the spectral width is less than the spectral width threshold, and the reflectivity is less than the second reflectivity threshold.

[0088] Preferably, the second processing unit includes:

[0089] a fifth processing unit, configured to perform sliding window processing on the first denoised radar data starting from the first distance bin, wherein the length of the second sliding window is P, and P≥3;

[0090] a seventh obtaining unit, the seventh obtaining unit being used to obtain a radial velocity value of a current distance bin and radial velocity values ​​of surrounding N distance bins, wherein the value of N is determined by a length P of the second sliding window;

[0091] N=P 2 -1

[0092] Assuming P = 3, the window settings are as follows:

[0093] v i-1,j+1 v i,j+1 v i+1,j+1

[0094] v i-1,j v i,j v i+1,j

[0095] v i-1,j-1 v i,j-1 v i+1,j-1

[0096] Where i is the radial data (or horizontal time) direction,

[0097] j is the vertical distance to the reservoir direction,

[0098] v i,j is the radial velocity value of the current altitude layer to be determined;

[0099] A second calculation unit, the second calculation unit is used to calculate the number of valid speed data of the surrounding N distance libraries;

[0100] An eighth obtaining unit, the eighth obtaining unit being used to obtain a threshold value of the number of speed valid data;

[0101] A second judgment unit, the second judgment unit is used to judge whether the number of valid speed data of the surrounding N distance libraries is less than the threshold value of the number of valid speed data;

[0102] a fourth determination unit, configured to determine that the current distance library is an invalid value if the number of valid speed data in the surrounding N distance libraries is less than the valid speed data number threshold;

[0103] A third calculation unit, the third calculation unit is used to calculate the effective radial velocity average value of the surrounding N distance bins if the number of the velocity valid data of the surrounding N distance bins is not less than the velocity valid data threshold;

[0104] A fourth calculation unit, the fourth calculation unit is used to calculate the radial velocity value v of the current distance library i,j The difference between the effective radial velocity average value and the effective radial velocity average value, wherein the difference calculation formula is:

[0105] n≥preset threshold of valid data number

[0106] Among them, v i is the radial velocity value of the altitude layer to be judged,

[0107] vj is the effective radial velocity value adjacent to it,

[0108] n is the number of valid data;

[0109] a ninth obtaining unit, the ninth obtaining unit being used to obtain a radial velocity difference threshold;

[0110] a third judging unit, the third judging unit being used to judge whether the difference is greater than a radial velocity difference threshold;

[0111] A fifth determination unit, wherein the fifth determination unit is configured to determine that the current database is a noise point if the difference is greater than a radial velocity difference threshold.

[0112] Preferably, the second processing unit further includes:

[0113] A first search unit, the first search unit is used to search for the radial velocity value v of the current distance library i,j N distance bins in the vertical direction;

[0114] The second search unit is used to search for the total number of distance libraries N of the upper and lower consecutive valid values ​​of the current distance library. total ;

[0115] a tenth obtaining unit, the tenth obtaining unit being used to obtain a total library number threshold;

[0116] A fourth judgment unit, the fourth judgment unit is used to judge whether the effective total number of libraries is less than the total number of libraries threshold;

[0117] A sixth determination unit, the sixth determination unit is used to determine if the effective total library number is less than the total library number threshold, and the radial velocity v in the adjacent radial direction of the current distance library i-1,j and v i+1,j All the distance libraries are invalid values, and the current distance library is determined to be a noise point.

[0118] Preferably, the third processing unit includes:

[0119] A sixth processing unit, the sixth processing unit is used to perform library-by-library convolution processing on the second denoised radar data, wherein the convolution kernel size is Q, Q=2N * +1, N * is a positive integer;

[0120] A first delimiting unit, wherein the first delimiting unit is used to use the middle value of the convolution kernel coverage area as a judgment object and the convolution kernel coverage area as an operation range;

[0121] A fifth calculation unit, the fifth calculation unit is used to calculate the number of invalid values ​​in the area covered by the convolution kernel;

[0122] an eleventh obtaining unit, the eleventh obtaining unit being used to obtain an invalid value quantity threshold;

[0123] A fifth judgment unit, the fifth judgment unit is used to judge whether the number of invalid values ​​in the area covered by the convolution kernel is greater than the invalid value number threshold;

[0124] A first replacement unit, wherein the first replacement unit is used to determine that the middle value of the area covered by the convolution kernel is an invalid value if the number of invalid values ​​in the area covered by the convolution kernel is greater than the invalid value number threshold, and replace the middle value of the area covered by the convolution kernel with an invalid value.

[0125] Preferably, the third processing unit further includes:

[0126] A sixth judgment unit, the sixth judgment unit is used to judge whether the middle value of the area covered by the convolution kernel is a valid value if the number of invalid values ​​in the area covered by the convolution kernel is not greater than the invalid value number threshold;

[0127] A sixth calculation unit, the sixth calculation unit is used to calculate the average value error between the effective middle value and the remaining effective values ​​in the region if the middle value of the region covered by the convolution kernel is a valid value;

[0128] A seventh judgment unit, the seventh judgment unit is used to judge whether the average value error satisfies a first formula, wherein the first formula is:

[0129] |Z ij -Z ave |≥3×σ

[0130] Among them, Z ave is the mean of all valid values ​​in the area covered by the convolution kernel,

[0131] σ is the standard deviation of all effective values ​​in the area covered by the convolution kernel;

[0132] A second replacement unit, the second replacement unit is used to determine that the middle value of the area covered by the convolution kernel is a noise point if the mean value error satisfies the first formula, and replace the middle value of the area covered by the convolution kernel with the median of all valid values ​​of the remaining grid points in the area covered by the convolution kernel;

[0133] The third replacement unit is used to determine that the middle value of the area covered by the convolution kernel is a missing value if the middle value of the area covered by the convolution kernel is an invalid value, and replace the middle value of the area covered by the convolution kernel with the median value of all valid values ​​of the remaining grid points in the area covered by the convolution kernel.

[0134] In a third aspect, the present invention provides a non-meteorological noise filtering device based on millimeter-wave cloud radar, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of any method of the aforementioned first aspect are implemented.

[0135] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any method of the aforementioned first aspect.

[0136] The above one or more technical solutions in the embodiments of the present application have at least one or more of the following technical effects:

[0137] The embodiment of the present invention provides a method and device for filtering non-meteorological noise based on millimeter-wave cloud radar, the method comprising: obtaining raw data detected by millimeter-wave cloud radar, wherein the radar raw data comprises radar basic data such as reflectivity factor, radial velocity, spectral width, depolarization ratio, etc.; statistically analyzing the distribution law and characteristics of meteorological echoes and mixed noise in the radar raw data, extracting characteristic quantities for distinguishing noise, processing the radar raw data based on a rolling sliding window method and combining judgment rules of multiple characteristic quantities to obtain a first judgment result; according to the first judgment result, removing noise points from the radar raw data to obtain first denoised radar data; based on a rolling sliding window method, removing noise points from the radar raw data to obtain first denoised radar data; based on a rolling sliding window method, removing noise points from the radar raw data to obtain first denoised radar data. The sliding window method performs a two-dimensional inspection in time and space on the first denoised radar data, and judges the difference between the radial velocity and the adjacent observation value in each distance library, and obtains the second judgment result; according to the second judgment result, the first denoised radar data is subjected to noise elimination to obtain the second denoised radar data; the second denoised radar data is subjected to library-by-library convolution processing, and the difference between the current and surrounding data points is judged in combination with the principle of continuity of meteorological echoes to obtain the third judgment result; according to the third judgment result, the second denoised radar data is subjected to noise elimination, and the missing points are supplemented to obtain radar data with non-meteorological noise filtered out. It is used to solve the technical problem that in the existing noise processing method based on fuzzy logic method, a large amount of historical data is needed to analyze and extract the physical quantity that distinguishes noise from meteorological echoes and the membership function corresponding to each physical quantity, resulting in a large workload of statistical analysis, and it is difficult to apply if the data is insufficient, the accuracy of the application result is low, the membership function parameters of each physical quantity need to be reset in different environments, and the mixed noise of meteorological echoes cannot be effectively filtered out. Taking the original data of millimeter-wave radar detection as input, the data distribution law of meteorological echo and mixed noise is analyzed. Combined with the principle of meteorological echo continuity, the effective filtering of meteorological echo mixed noise is achieved, thereby improving the technical effect of millimeter-wave cloud radar's real-time detection of weak echoes and clouds.

[0138] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0139] Figure 1 It is a flowchart of a non-meteorological noise filtering method based on millimeter-wave cloud radar in an embodiment of the present invention;

[0140] Figure 2 This is a schematic structural diagram of a non-meteorological noise filtering device based on millimeter-wave cloud radar in an embodiment of the present invention;

[0141] Figure 3 It is another structural schematic diagram of a non-meteorological noise filtering device based on millimeter-wave cloud radar in an embodiment of the present invention;

[0142] Figure 4 This is a comparison diagram before and after non-meteorological noise is filtered out in an embodiment of the present invention;

[0143] Figure 5 This is another comparison diagram before and after non-meteorological noise is filtered out in an embodiment of the present invention;

[0144] Explanation of the figure numbers: first obtaining unit 1, first processing unit 2, second obtaining unit 3, second processing unit 4, third obtaining unit 5, third processing unit 6, fourth obtaining unit 7, bus 300, receiver 301, processor 302, transmitter 303, memory 304, bus interface 306. DETAILED DESCRIPTION

[0145] The embodiment of the present invention provides a non-meteorological noise filtering method and device based on millimeter-wave cloud radar, which solves the technical problems in the noise processing method based on fuzzy logic method in the prior art, that is, it is necessary to use a large amount of historical data to analyze and extract the physical quantity that distinguishes noise from meteorological echoes and the membership function corresponding to each physical quantity, thereby resulting in a large workload of statistical analysis, and it is difficult to apply if the data is insufficient, the accuracy of the application results is low, the membership function parameters of each physical quantity need to be reset in different environments, and the mixed noise of meteorological echoes cannot be effectively filtered out.

[0146] The overall idea of ​​the technical solution provided by the present invention is as follows: obtaining raw data detected by a millimeter-wave cloud radar, wherein the raw radar data includes radar basic data such as reflectivity factor, radial velocity, spectral width, and depolarization ratio; statistically analyzing the distribution law and characteristics of meteorological echoes and mixed noise in the raw radar data, extracting characteristic quantities for distinguishing noise, and processing the raw radar data based on a rolling window method and a judgment rule of multiple characteristic quantities to obtain a first judgment result; according to the first judgment result, removing noise points from the raw radar data to obtain first denoised radar data; based on the rolling window method, the first denoised radar data is processed. The radar data after de-noising is subjected to a two-dimensional inspection in time and space, and the difference between the radial velocity and the adjacent observation value in each distance library is judged library by library to obtain the second judgment result; according to the second judgment result, the radar data after the first de-noising is subjected to noise removal to obtain the second de-noising radar data; the radar data after the second de-noising is subjected to library by library convolution processing, and the difference between the current and surrounding data points is judged in combination with the principle of continuity of meteorological echoes to obtain the third judgment result; according to the third judgment result, the radar data after the second de-noising is subjected to noise removal, and the missing points are supplemented to obtain the radar data with non-meteorological noise filtered out. The original data detected by the millimeter-wave radar is used as input, and the data distribution law of meteorological echoes and mixed noises is analyzed. In combination with the principle of continuity of meteorological echoes, the mixed noise of meteorological echoes is effectively filtered out, thereby improving the technical effect of the millimeter-wave cloud radar's ability to detect weak echoes and clouds in real time.

[0147] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations on the technical solution of the present application. In the absence of conflict, the embodiments of the present application and the technical features in the embodiments can be combined with each other.

[0148] The term "and / or" in this article is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0149] Embodiment 1

[0150] Figure 1 FIG. 1 is a flow chart of a non-meteorological noise filtering method based on millimeter wave cloud radar in an embodiment of the present invention. Figure 1 As shown, an embodiment of the present invention provides a non-meteorological noise filtering method based on millimeter wave cloud radar, the method comprising:

[0151] Step s1: obtaining raw data detected by millimeter-wave cloud radar, wherein the radar raw data includes radar basic data such as reflectivity factor, radial velocity, spectrum width, depolarization ratio, etc.;

[0152] Step s2: Statistically analyzing the distribution law and characteristics of the weather echo and the mixed noise in the radar raw data, extracting the characteristic quantity for distinguishing the noise, and processing the radar raw data based on the rolling sliding window method and combining the judgment rules of multiple characteristic quantities to obtain a first judgment result;

[0153] Step s3: According to the first judgment result, noise points are removed from the radar raw data to obtain first denoised radar data;

[0154] Furthermore, the data volume of the radar raw data is M pieces of vertical detection data (M≥3) or plane scanning data (radial number≥3) at different times.

[0155] Specifically, the embodiment of the present application uses the raw data detected by the millimeter-wave cloud radar as input, wherein the radar raw data includes radar basic data such as reflectivity factor, radial velocity, spectral width, depolarization ratio, and the like, and the data volume of the radar raw data is M vertical detection data (M≥3) or plane scanning data (radial number≥3) at different times; secondly, a statistical analysis is performed on the distribution law of the radar raw data, and characteristic quantities for noise point identification are selected according to the distribution law of the raw data: reflectivity, depolarization ratio, spectral width, and velocity change; then, a rolling window method is used to filter out the meteorological echo mixed noise from the acquired radar raw data in combination with the judgment rules of multiple characteristic quantities.

[0156] In step s2, the radar raw data is processed based on the rolling window method in combination with the judgment rules of multiple feature quantities to obtain a first judgment result, including:

[0157] Performing rolling window processing on the radar raw data starting from the first range library, wherein the length of the first sliding window is L, and L≥3;

[0158] Calculate the speed difference between every two adjacent distance bins and calculate the average difference Vcoff average , where every two adjacent distance libraries select all data in both horizontal and vertical directions within the window coverage range;

[0159] Obtaining a first speed difference threshold and a second speed difference threshold;

[0160] Obtaining a spectral width threshold, a depolarization ratio threshold, a first reflectivity threshold, and a second reflectivity threshold;

[0161] Comparing the average difference with the first speed difference threshold and the second speed difference threshold;

[0162] If the average difference is greater than the speed difference first threshold, determining whether the spectrum width is less than a spectrum width threshold;

[0163] If the spectral width is less than the spectral width threshold, the current distance library is determined to be a noise point;

[0164] If the average difference is between the first speed difference threshold and the second speed difference threshold, and the depolarization ratio is greater than the depolarization ratio threshold, and the reflectivity is less than the first reflectivity threshold, it is determined that the current distance library is a noise point;

[0165] If the average difference is between the first speed difference threshold and the second speed difference threshold, and the depolarization ratio is an invalid value, the spectral width is less than the spectral width threshold, and the reflectivity is less than the second reflectivity threshold, the current distance library is determined to be a noise point.

[0166] Step s4: Based on the rolling window method, a two-dimensional check is performed on the first denoised radar data in time and space, and the difference between the radial velocity and the adjacent observation value in each range bin is determined bin by bin to obtain a second judgment result;

[0167] Step s5: according to the second judgment result, removing noise points from the first denoised radar data to obtain second denoised radar data;

[0168] Specifically, the first denoised radar data is processed library by library using a rolling window method, radial velocity data in the first denoised radar data is selected for processing, a two-dimensional check is performed on the radial velocity data in time and space, and the difference between the radial velocity in each distance library and the observation values ​​at adjacent moments and adjacent altitude layers is determined. If the difference exceeds a preset threshold, the data at the current position is determined to be suspicious or invalid.

[0169] In step s4, the first denoised radar data is subjected to a two-dimensional inspection in time and space based on the rolling window method, and the difference between the radial velocity and the adjacent observation value in each range bin is determined bin by bin, to obtain a second judgment result, including:

[0170] Performing sliding window processing on the radar data after the first denoising from the first range bin, wherein the length of the second sliding window is P, where P≥3;

[0171] Get the radial velocity value v of the current distance library i,j , and the radial velocity values ​​of the surrounding N range bins, wherein the value of N is determined by the length P of the second sliding window;

[0172] N=P 2 -1

[0173] Assuming P = 3, the window settings are as follows:

[0174] v i-1,j+1 v i,j+1 v i+1,j+1

[0175] v i-1,j v i,j v i+1,j

[0176] v i-1,j-1 v i,j-1 v i+1,j-1

[0177] Where i is the radial data (or horizontal time) direction,

[0178] j is the vertical distance to the reservoir direction,

[0179] v i,j is the radial velocity value of the current altitude layer to be determined;

[0180] Calculate the number of valid speed data in the surrounding N distance libraries;

[0181] Get the threshold of the number of valid speed data;

[0182] Determine whether the number of valid speed data in the surrounding N distance libraries is less than the threshold of the number of valid speed data;

[0183] If the number of valid speed data in the surrounding N distance libraries is less than the valid speed data number threshold, the current distance library is determined to be an invalid value;

[0184] If the number of valid velocity data of the surrounding N distance bins is not less than the valid velocity data threshold, calculating the effective radial velocity average value of the surrounding N distance bins;

[0185] Calculate the radial velocity value v of the current distance library i,j The difference between the effective radial velocity average value and the effective radial velocity average value, wherein the difference calculation formula is:

[0186] n≥preset threshold of valid data number

[0187] Among them, v i is the radial velocity value of the altitude layer to be judged,

[0188] v j is the effective radial velocity value adjacent to it,

[0189] n is the number of valid data;

[0190] Obtaining a radial velocity difference threshold;

[0191] Determining whether the difference is greater than a radial velocity difference threshold;

[0192] If the difference is greater than the radial velocity difference threshold, it is determined that the current database is a noise point.

[0193] In step s4, the method of performing a two-dimensional inspection in time and space on the first denoised radar data based on the rolling window method, judging the difference between the radial velocity and the adjacent observation value in each range bin one by one, and obtaining a second judgment result, further includes:

[0194] Find the radial velocity value v of the current distance library i,j N distance bins in the vertical direction;

[0195] Find the total number of distance libraries N of the continuous valid values ​​above and below the current distance library total ;

[0196] Get the total library number threshold;

[0197] Determine whether the effective total library number is less than the total library number threshold;

[0198] If the total effective library number is less than the total library number threshold, and the radial velocity v in the adjacent radial direction of the current distance library i-1,j and v i+1,j All the distance libraries are invalid values, and the current distance library is determined to be a noise point.

[0199] Step s6: performing library-by-library convolution processing on the radar data after the second denoising, combining the continuity principle of meteorological echoes, and using the difference between the echo at the judgment point and the surrounding echoes to obtain a third judgment result;

[0200] Step s7: According to the third judgment result, the second denoised radar data is subjected to noise elimination, and the missing points are supplemented to obtain the millimeter-wave cloud radar data with non-meteorological noise filtered out.

[0201] Specifically, the second denoised radar data is processed by a library-by-library convolution operation method. According to the continuity principle of meteorological echo changes, the echo of the point to be judged is compared with the surrounding echoes, and the difference between the current and surrounding data is judged to determine whether it is a noise point, and the missing points are supplemented to a certain extent, thereby achieving effective filtering of meteorological echo mixed noise, and further improving the technical effect of the millimeter-wave cloud radar's ability to detect weak echoes and clouds in real time.

[0202] In step s6, the second denoised radar data is subjected to library-by-library convolution processing, and the difference between the echo at the judgment point and the surrounding echoes is used in combination with the continuity principle of the meteorological echo to obtain a third judgment result, including:

[0203] Perform library-by-library convolution processing on the second denoised radar data, where the convolution kernel size is Q, Q=2N * +1, N * is a positive integer;

[0204] The median value of the area covered by the convolution kernel is used as the judgment object, and the area covered by the convolution kernel is used as the operation range;

[0205] Calculating the number of invalid values ​​in the area covered by the convolution kernel;

[0206] Get the threshold value of the number of invalid values;

[0207] Determine whether the number of invalid values ​​in the area covered by the convolution kernel is greater than the invalid value number threshold;

[0208] If the number of invalid values ​​in the area covered by the convolution kernel is greater than the invalid value quantity threshold, the middle value of the area covered by the convolution kernel is determined to be an invalid value, and the middle value of the area covered by the convolution kernel is replaced by the invalid value.

[0209] In step s6, the second denoised radar data is subjected to library-by-library convolution processing, and the difference between the echo at the judgment point and the surrounding echoes is used to obtain a third judgment result in combination with the continuity principle of the meteorological echo, and further includes:

[0210] If the number of invalid values ​​in the area covered by the convolution kernel is not greater than the invalid value number threshold, determining whether the middle value of the area covered by the convolution kernel is a valid value;

[0211] If the middle value of the area covered by the convolution kernel is a valid value, calculate the average value error between the valid middle value and the remaining valid values ​​in the area;

[0212] Determine whether the mean value error satisfies a first formula, wherein the first formula is:

[0213] |Z ij -Z ave |≥3×σ

[0214] Among them, Z ave is the mean of all valid values ​​in the area covered by the convolution kernel,

[0215] σ is the standard deviation of all effective values ​​in the area covered by the convolution kernel;

[0216] If the mean value error satisfies the first formula, the middle value of the area covered by the convolution kernel is determined to be a noise point, and the middle value of the area covered by the convolution kernel is replaced by the median of all valid values ​​of the remaining grid points in the area covered by the convolution kernel;

[0217] If the middle value of the area covered by the convolution kernel is not a valid value, the middle value of the area covered by the convolution kernel is determined to be a missing value, and the middle value of the area covered by the convolution kernel is replaced by the median value of all valid values ​​of the remaining grid points in the area covered by the convolution kernel.

[0218] In summary, the embodiment of the present application obtains the data distribution law of meteorological echo and noise by analyzing the detection data of the millimeter wave radar and using statistical methods, and selects multiple physical quantities based on the data distribution law and the principle of continuity of meteorological echoes, and divides them into four modules. Each module removes noise according to different logical rules to filter out mixed noise, thereby improving the ability of the millimeter wave cloud radar to detect weak echoes and clouds in real time. It should be noted here that the last three modules can be used independently to filter out noise and meteorological echoes; corresponding modules can also be selected for joint application to filter out noise according to different situations.

[0219] Compared with other processing methods, this method has the following advantages: First, this method does not require a large amount of historical data, but only a small amount of historical data for analysis and statistics, and analyzes the threshold of each physical quantity that can distinguish between noise and weather echoes; second, compared with the noise point filtering method based on fuzzy logic, this method does not require a large amount of statistical analysis to determine the membership function; third, this method only needs to fine-tune a small number of parameters according to actual conditions in different local environments, while the method based on fuzzy logic may need to reset the corresponding membership function and weight of each physical quantity in different environments, and there are more parameters. Fourth, this method is relatively simple, has high execution efficiency, is easy to implement and apply in engineering, and can effectively remove the noise of mixed weather echoes, thereby improving the radar's ability to detect weak weather echoes.

[0220] The embodiments of the present application can be applied not only to millimeter-wave cloud detection radars, but also to different radars of various bands and systems. It is only necessary to obtain input data that meets the corresponding requirements and adjust some parameter thresholds in the scheme according to different types of radars.

[0221] The embodiment of the present application currently selects corresponding physical quantities for processing based on two-dimensional data. When the radar has volume scanning type data, the same principle can be used to analyze the distribution law of three-dimensional directional data and select corresponding physical quantities to identify noise points and weather echoes.

[0222] Embodiment 2

[0223] Based on the same inventive concept as the non-meteorological noise filtering method based on millimeter-wave cloud radar in the aforementioned embodiment, the present invention also provides a non-meteorological noise filtering device based on millimeter-wave cloud radar, such as Figure 2 As shown, the device comprises:

[0224] A first obtaining unit 1, wherein the first obtaining unit 1 is used to obtain raw data detected by a millimeter-wave cloud radar, wherein the radar raw data includes radar basic data such as reflectivity factor, radial velocity, spectrum width, and depolarization ratio;

[0225] A first processing unit 2, the first processing unit 2 is used to statistically analyze the distribution law and characteristics of the meteorological echo and the mixed noise in the radar raw data, extract the characteristic quantity for distinguishing the noise, and process the radar raw data based on the rolling sliding window method in combination with the judgment rules of multiple characteristic quantities to obtain a first judgment result;

[0226] A second obtaining unit 3, wherein the second obtaining unit 3 is used to remove noise points from the radar raw data according to the first judgment result to obtain first denoised radar data;

[0227] A second processing unit 4, the second processing unit 4 is used to perform a two-dimensional check in time and space on the first denoised radar data based on a rolling window method, and judge the difference between the radial velocity in each range bin and the adjacent observation value, to obtain a second judgment result;

[0228] A third obtaining unit 5, wherein the third obtaining unit 5 is used to remove noise points from the first denoised radar data according to the second judgment result to obtain second denoised radar data;

[0229] A third processing unit 6, the third processing unit 6 is used to perform library-by-library convolution processing on the second denoised radar data, and determine the difference between the current data point and the surrounding data point in combination with the continuity principle of the meteorological echo to obtain a third determination result;

[0230] The fourth obtaining unit 7 is used to remove noise points from the second denoised radar data according to the third judgment result, and to fill in the missing points to obtain radar data with non-meteorological noise filtered out.

[0231] Preferably, the amount of the radar raw data is M pieces of vertical detection data (M≥3) or plane scanning data (radial number≥3) at different times.

[0232] Preferably, the first processing unit 2 includes:

[0233] A first extraction unit, wherein the first extraction unit is used to extract a feature quantity for distinguishing noise, wherein the feature quantities used are: reflectivity, depolarization ratio, spectrum width, and velocity variation;

[0234] A fourth processing unit, the fourth processing unit is used to perform rolling sliding window processing on the radar raw data starting from the first distance library, wherein the length of the first sliding window is L, L ≥ 3;

[0235] The first calculation unit is used to calculate the speed difference between each two adjacent distance bins and calculate the average difference Vcoff average , where every two adjacent distance libraries select all data in both horizontal and vertical directions within the window coverage range;

[0236] a fifth obtaining unit, the fifth obtaining unit being used to obtain a first speed difference threshold and a second speed difference threshold;

[0237] a sixth obtaining unit, the sixth obtaining unit being used to obtain a spectral width threshold, a depolarization ratio threshold, a first reflectivity threshold, and a second reflectivity threshold;

[0238] a first comparison unit, the first comparison unit being used to compare the average difference value with the first speed difference threshold and the second speed difference threshold;

[0239] a first judging unit, the first judging unit being configured to judge whether the spectrum width is less than a spectrum width threshold value if the average difference value is greater than the speed difference first threshold value;

[0240] A first determination unit, the first determination unit is used to determine that the current distance library is a noise point if the spectrum width is less than the spectrum width threshold;

[0241] a second determination unit, configured to determine that the current distance library is a noise point if the average difference is between the first speed difference threshold and the second speed difference threshold, the depolarization ratio is greater than the depolarization ratio threshold, and the reflectivity is less than the first reflectivity threshold;

[0242] The third determination unit is used to determine that the current distance library is a noise point if the average difference is between the first threshold value of the speed difference and the second threshold value of the speed difference, and the depolarization ratio is an invalid value, the spectral width is less than the spectral width threshold, and the reflectivity is less than the second reflectivity threshold.

[0243] Preferably, the second processing unit 4 includes:

[0244] a fifth processing unit, configured to perform sliding window processing on the first denoised radar data starting from the first distance bin, wherein the length of the second sliding window is P, and P≥3;

[0245] a seventh obtaining unit, the seventh obtaining unit being used to obtain a radial velocity value of a current distance bin and radial velocity values ​​of surrounding N distance bins, wherein the value of N is determined by a length P of the second sliding window;

[0246] N=P 2 -1

[0247] Assuming P = 3, the window settings are as follows:

[0248] v i-1,j+1 v i,j+1 v i+1,j+1

[0249] v i-1,j v i,j v i+1,j

[0250] v i-1,j-1 v i,j-1 v i+1,j-1

[0251] Where i is the radial data (or horizontal time) direction,

[0252] j is the vertical distance to the reservoir direction,

[0253] v i,j is the radial velocity value of the current altitude layer to be determined;

[0254] A second calculation unit, the second calculation unit is used to calculate the number of valid speed data of the surrounding N distance libraries;

[0255] An eighth obtaining unit, the eighth obtaining unit being used to obtain a threshold value of the number of speed valid data;

[0256] A second judgment unit, the second judgment unit is used to judge whether the number of valid speed data of the surrounding N distance libraries is less than the threshold value of the number of valid speed data;

[0257] a fourth determination unit, configured to determine that the current distance library is an invalid value if the number of valid speed data in the surrounding N distance libraries is less than the valid speed data number threshold;

[0258] A third calculation unit, the third calculation unit is used to calculate the effective radial velocity average value of the surrounding N distance bins if the number of the velocity valid data of the surrounding N distance bins is not less than the velocity valid data threshold;

[0259] A fourth calculation unit, the fourth calculation unit is used to calculate the difference between the radial velocity value of the current distance library and the effective radial velocity average value, wherein the difference calculation formula is:

[0260] n≥preset threshold of valid data number

[0261] Among them, v i is the radial velocity value of the altitude layer to be judged,

[0262] v j is the effective radial velocity value adjacent to it,

[0263] n is the number of valid data;

[0264] a ninth obtaining unit, the ninth obtaining unit being used to obtain a radial velocity difference threshold;

[0265] a third judging unit, the third judging unit being used to judge whether the difference is greater than a radial velocity difference threshold;

[0266] A fifth determination unit, wherein the fifth determination unit is configured to determine that the current database is a noise point if the difference is greater than a radial velocity difference threshold.

[0267] Preferably, the second processing unit 4 further includes:

[0268] A first search unit, the first search unit is used to search N distance libraries in the vertical direction for the radial velocity value of the current distance library;

[0269] The second search unit is used to search for the total number of distance libraries N of the upper and lower consecutive valid values ​​of the current distance library. total ;

[0270] a tenth obtaining unit, the tenth obtaining unit being used to obtain a total library number threshold;

[0271] A fourth judgment unit, the fourth judgment unit is used to judge whether the effective total number of libraries is less than the total number of libraries threshold;

[0272] A sixth determination unit, the sixth determination unit is used to determine if the effective total library number is less than the total library number threshold, and the radial velocity v in the adjacent radial direction of the current distance library i-1,j and v i+1,j All the distance libraries are invalid values, and the current distance library is determined to be a noise point.

[0273] Preferably, the third processing unit 6 includes:

[0274] A sixth processing unit, the sixth processing unit is used to perform library-by-library convolution processing on the second denoised radar data, wherein the convolution kernel size is Q, Q=2N * +1, N * is a positive integer;

[0275] A first delimiting unit, wherein the first delimiting unit is used to use the middle value of the convolution kernel coverage area as a judgment object and the convolution kernel coverage area as an operation range;

[0276] A fifth calculation unit, the fifth calculation unit is used to calculate the number of invalid values ​​in the area covered by the convolution kernel;

[0277] an eleventh obtaining unit, the eleventh obtaining unit being used to obtain an invalid value quantity threshold;

[0278] A fifth judgment unit, the fifth judgment unit is used to judge whether the number of invalid values ​​in the area covered by the convolution kernel is greater than the invalid value number threshold;

[0279] A first replacement unit, wherein the first replacement unit is used to determine that the middle value of the area covered by the convolution kernel is an invalid value if the number of invalid values ​​in the area covered by the convolution kernel is greater than the invalid value number threshold, and replace the middle value of the area covered by the convolution kernel with an invalid value.

[0280] Preferably, the third processing unit 6 further includes:

[0281] A sixth judgment unit, the sixth judgment unit is used to judge whether the middle value of the area covered by the convolution kernel is a valid value if the number of invalid values ​​in the area covered by the convolution kernel is not greater than the invalid value number threshold;

[0282] A sixth calculation unit, the sixth calculation unit is used to calculate the average value error between the effective middle value and the remaining effective values ​​in the region if the middle value of the region covered by the convolution kernel is a valid value;

[0283] A seventh judgment unit, the seventh judgment unit is used to judge whether the average value error satisfies a first formula, wherein the first formula is:

[0284] |Z ij -Z ave |≥3×σ

[0285] Among them, Z ave is the mean of all valid values ​​in the area covered by the convolution kernel,

[0286] σ is the standard deviation of all effective values ​​in the area covered by the convolution kernel;

[0287] A second replacement unit, the second replacement unit is used to determine that the middle value of the area covered by the convolution kernel is a noise point if the mean value error satisfies the first formula, and replace the middle value of the area covered by the convolution kernel with the median of all valid values ​​of the remaining grid points in the area covered by the convolution kernel;

[0288] The third replacement unit is used to determine that the middle value of the area covered by the convolution kernel is a missing value if the middle value of the area covered by the convolution kernel is an invalid value, and replace the middle value of the area covered by the convolution kernel with the median value of all valid values ​​of the remaining grid points in the area covered by the convolution kernel.

[0289] The foregoing Figure 1 The various variations and specific examples of the non-meteorological noise filtering method based on millimeter-wave cloud radar in Example 1 are also applicable to a non-meteorological noise filtering device based on millimeter-wave cloud radar in this embodiment. Through the above detailed description of the non-meteorological noise filtering method based on millimeter-wave cloud radar, those skilled in the art can clearly know the implementation method of the non-meteorological noise filtering device based on millimeter-wave cloud radar in this embodiment, so for the sake of brevity of the specification, it will not be described in detail here.

[0290] Embodiment 3

[0291] Based on the same inventive concept as the non-meteorological noise filtering method based on millimeter-wave cloud radar in the aforementioned embodiment, the present invention also provides a non-meteorological noise filtering device based on millimeter-wave cloud radar, including a memory, a processor, and a computer program stored in the memory and executable on the processor, which, when executed by the processor, implements the steps of any one of the methods of the non-meteorological noise filtering method based on millimeter-wave cloud radar in the first aspect described above.

[0292] Among them, Figure 3 In the embodiment of the present invention, a bus architecture (represented by bus 300) is shown, which may include any number of interconnected buses and bridges, and bus 300 links various circuits including one or more processors represented by processor 302 and memory represented by memory 304. Bus 300 may also link various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are not further described herein. Bus interface 306 provides an interface between bus 300 and receiver 301 and transmitter 303. Receiver 301 and transmitter 303 may be the same element, namely a transceiver, which provides a unit for communicating with various other devices over a transmission medium.

[0293] The processor 302 is responsible for managing the bus 300 and general processing, while the memory 304 may be used to store data used by the processor 302 when performing operations.

[0294] Embodiment 4

[0295] Based on the same inventive concept as the non-meteorological noise filtering method based on millimeter-wave cloud radar in the aforementioned embodiment, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any method of the non-meteorological noise filtering method based on millimeter-wave cloud radar in the first aspect described above.

[0296] The above one or more technical solutions in the embodiments of the present application have at least one or more of the following technical effects:

[0297] The embodiment of the present invention provides a method and device for filtering non-meteorological noise based on millimeter-wave cloud radar, the method comprising: obtaining raw data detected by millimeter-wave cloud radar, wherein the radar raw data comprises radar basic data such as reflectivity factor, radial velocity, spectral width, depolarization ratio, etc.; statistically analyzing the distribution law and characteristics of meteorological echoes and mixed noise in the radar raw data, extracting characteristic quantities for distinguishing noise, processing the radar raw data based on a rolling sliding window method and combining judgment rules of multiple characteristic quantities to obtain a first judgment result; according to the first judgment result, removing noise points from the radar raw data to obtain first denoised radar data; based on a rolling sliding window method, removing noise points from the radar raw data to obtain first denoised radar data; based on a rolling sliding window method, removing noise points from the radar raw data to obtain first denoised radar data. The sliding window method performs a two-dimensional inspection in time and space on the first denoised radar data, and judges the difference between the radial velocity and the adjacent observation value in each distance library, and obtains the second judgment result; according to the second judgment result, the first denoised radar data is subjected to noise elimination to obtain the second denoised radar data; the second denoised radar data is subjected to library-by-library convolution processing, and the difference between the current and surrounding data points is judged in combination with the principle of continuity of meteorological echoes to obtain the third judgment result; according to the third judgment result, the second denoised radar data is subjected to noise elimination, and the missing points are supplemented to obtain radar data with non-meteorological noise filtered out. It is used to solve the technical problem that in the existing noise processing method based on fuzzy logic method, a large amount of historical data is needed to analyze and extract the physical quantity that distinguishes noise from meteorological echoes and the membership function corresponding to each physical quantity, resulting in a large workload of statistical analysis, and it is difficult to apply if the data is insufficient, the accuracy of the application result is low, the membership function parameters of each physical quantity need to be reset in different environments, and the mixed noise of meteorological echoes cannot be effectively filtered out. Taking the original data of millimeter-wave radar detection as input, the data distribution law of meteorological echo and mixed noise is analyzed. Combined with the principle of meteorological echo continuity, the effective filtering of meteorological echo mixed noise is achieved, thereby improving the technical effect of millimeter-wave cloud radar's real-time detection of weak echoes and clouds.

[0298] It should be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The present invention is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, may be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the functions in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0299] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0300] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A non-meteorological noise filtering method based on millimeter wave cloud radar, characterized in that: The method comprises: Obtaining raw data of millimeter-wave cloud radar detection, wherein the radar raw data includes radar basic data such as reflectivity factor, radial velocity, spectrum width, and depolarization ratio; Statistically analyzing the distribution law and characteristics of the meteorological echo and the mixed noise in the radar raw data, extracting the characteristic quantity for distinguishing the noise, and processing the radar raw data based on the rolling sliding window method and combining the judgment rules of multiple characteristic quantities to obtain a first judgment result; According to the first judgment result, noise points are removed from the radar raw data to obtain first denoised radar data; Based on the rolling window method, a two-dimensional inspection in time and space is performed on the first denoised radar data, and the difference between the radial velocity and the adjacent observation value in each range bin is determined bin by bin to obtain a second judgment result; According to the second judgment result, removing noise points from the first denoised radar data to obtain second denoised radar data; Performing library-by-library convolution processing on the radar data after the second denoising, combining the principle of continuity of meteorological echoes, judging the difference between the current data point and the surrounding data points, and obtaining a third judgment result; According to the third judgment result, noise points are removed from the second denoised radar data, and missing points are supplemented to obtain radar data with non-meteorological noise filtered out.

2. The method according to claim 1, characterized in that The data volume of the radar raw data is M pieces of vertical detection data (M≥3) or plane scanning data (radial number≥3) at different times.

3. The method according to claim 1, characterized in that The statistical analysis includes analyzing the distribution law and characteristics of the meteorological echo and the mixed noise in the radar raw data, extracting the characteristic quantity for distinguishing the noise, and processing the radar raw data based on the rolling sliding window method in combination with the judgment rules of the multiple characteristic quantities to obtain the first judgment result, including: Extracting the characteristic quantity for distinguishing noise, wherein the characteristic quantities used are: reflectivity, depolarization ratio, spectrum width, and velocity variation; Performing rolling window processing on the radar raw data starting from the first distance library, wherein the length of the first sliding window is L, and L≥3; Calculate the speed difference between every two adjacent distance bins and calculate the average difference Vcoff average , where every two adjacent distance libraries select all data in both horizontal and vertical directions within the window coverage range; Obtaining a first speed difference threshold and a second speed difference threshold; Obtaining a spectral width threshold, a depolarization ratio threshold, a first reflectivity threshold, and a second reflectivity threshold; Comparing the average difference with the first speed difference threshold and the second speed difference threshold; If the average difference is greater than the speed difference first threshold, determining whether the spectrum width is less than a spectrum width threshold; If the spectral width is less than the spectral width threshold, the current distance library is determined to be a noise point; If the average difference is between the first speed difference threshold and the second speed difference threshold, and the depolarization ratio is greater than the depolarization ratio threshold, and the reflectivity is less than the first reflectivity threshold, it is determined that the current distance library is a noise point; If the average difference is between the first speed difference threshold and the second speed difference threshold, and the depolarization ratio is an invalid value, the spectral width is smaller than the spectral width threshold, and the reflectivity is smaller than the second reflectivity threshold, the current distance library is determined to be a noise point.

4. The method according to claim 1, characterized in that The method based on the rolling window is to perform a two-dimensional inspection in time and space on the first denoised radar data, and judge the difference between the radial velocity and the adjacent observation value in each range bin one by one, to obtain a second judgment result, including: Performing sliding window processing on the radar data after the first denoising from the first range bin, wherein the length of the second sliding window is P, where P≥3; Get the radial velocity value v of the current distance library i,j , and the radial velocity values ​​of the surrounding N distance bins, where the value of N is determined by the length P of the second sliding window: N=P 2 -1 Assuming P = 3, the window settings are as follows: Where i is the radial data (or horizontal time) direction, j is the vertical distance to the reservoir direction, v i,j is the radial velocity value of the current altitude layer to be determined; Calculate the number of valid speed data in the surrounding N distance libraries; Get the threshold of the number of valid speed data; Determine whether the number of valid speed data in the surrounding N distance libraries is less than the threshold of the number of valid speed data; If the number of valid speed data in the surrounding N distance libraries is less than the valid speed data number threshold, the current distance library is determined to be an invalid value; If the number of valid velocity data of the surrounding N distance bins is not less than the valid velocity data threshold, calculating the effective radial velocity average value of the surrounding N distance bins; Calculate the radial velocity value v of the current distance library i,j The difference between the effective radial velocity average value and the effective radial velocity average value, wherein the difference calculation formula is: n≥preset threshold of valid data number Among them, v i is the radial velocity value of the altitude layer to be judged, v j is the effective radial velocity value adjacent to it, n is the number of valid data; Obtaining a radial velocity difference threshold; Determining whether the difference is greater than a radial velocity difference threshold; If the difference is greater than the radial velocity difference threshold, it is determined that the current database is a noise point.

5. The method according to claim 4, characterized in that The method of performing a two-dimensional inspection in time and space on the first denoised radar data based on the rolling window method, judging the difference between the radial velocity and the adjacent observation value in each range bin one by one, and obtaining a second judgment result, further includes: Find the radial velocity value v of the current distance library i,j N distance bins in the vertical direction; Find the total number of distance libraries N of the continuous valid values ​​above and below the current distance library total ; Get the total library number threshold; Determine whether the effective total library number is less than the total library number threshold; If the effective total library number is less than the total library number threshold, and the radial velocity v in the adjacent radial direction of the current distance library i-1,j and v i+1,j All the distance libraries are invalid values, and the current distance library is determined to be a noise point.

6. The method according to claim 1, characterized in that The performing of library-by-library convolution processing on the second denoised radar data, combining the principle of continuity of meteorological echoes, judging the difference between the current data point and the surrounding data points, and obtaining a third judgment result, includes: The second denoised radar data is subjected to library-by-library convolution processing, wherein the convolution kernel size is Q, Q=2N * +1,N * is a positive integer; The median value of the area covered by the convolution kernel is used as the judgment object, and the area covered by the convolution kernel is used as the operation range; Calculating the number of invalid values ​​in the area covered by the convolution kernel; Get the threshold value of the number of invalid values; Determine whether the number of invalid values ​​in the area covered by the convolution kernel is greater than the invalid value number threshold; If the number of invalid values ​​in the area covered by the convolution kernel is greater than the invalid value quantity threshold, the middle value of the area covered by the convolution kernel is determined to be an invalid value, and the middle value of the area covered by the convolution kernel is replaced by the invalid value.

7. The method according to claim 6, characterized in that The performing of library-by-library convolution processing on the second denoised radar data, combining the principle of continuity of meteorological echoes, judging the difference between the current data point and the surrounding data points, and obtaining a third judgment result, further includes: If the number of invalid values ​​in the area covered by the convolution kernel is not greater than the invalid value number threshold, determining whether the middle value of the area covered by the convolution kernel is a valid value; If the middle value of the area covered by the convolution kernel is a valid value, calculate the average value error between the valid middle value and the remaining valid values ​​in the area; Determine whether the mean value error satisfies a first formula, wherein the first formula is: |From ij -WITH ave |≥3×σ Among them, Z ave is the mean of all valid values ​​in the area covered by the convolution kernel, σ is the standard deviation of all effective values ​​in the area covered by the convolution kernel; If the mean value error satisfies the first formula, the middle value of the area covered by the convolution kernel is determined to be a noise point, and the middle value of the area covered by the convolution kernel is replaced by the median of all valid values ​​of the remaining grid points in the area covered by the convolution kernel; If the middle value of the area covered by the convolution kernel is an invalid value, the middle value of the area covered by the convolution kernel is determined to be a missing value, and the middle value of the area covered by the convolution kernel is replaced by the median value of all valid values ​​of the remaining grid points in the area covered by the convolution kernel.

8. A non-meteorological noise filtering device based on millimeter wave cloud radar, characterized in that: The device comprises: A first obtaining unit, wherein the first obtaining unit is used to obtain raw data detected by a millimeter-wave cloud radar, wherein the radar raw data includes radar basic data such as reflectivity factor, radial velocity, spectrum width, and depolarization ratio; A first processing unit, the first processing unit is used to statistically analyze the distribution law and characteristics of the meteorological echo and the mixed noise in the radar raw data, extract the characteristic quantity for distinguishing the noise, and process the radar raw data based on the rolling sliding window method in combination with the judgment rules of multiple characteristic quantities to obtain a first judgment result; a second obtaining unit, the second obtaining unit being used to remove noise points from the radar raw data according to the first judgment result, so as to obtain first denoised radar data; A second processing unit, the second processing unit is used to perform a two-dimensional check in time and space on the first denoised radar data based on a rolling window method, and judge the difference between the radial velocity in each range bin and the adjacent observation value, to obtain a second judgment result. a third obtaining unit, wherein the third obtaining unit is used to remove noise points from the first denoised radar data according to the second judgment result to obtain second denoised radar data; A third processing unit, the third processing unit is used to perform library-by-library convolution processing on the second denoised radar data, and judge the difference between the current data point and the surrounding data point in combination with the continuity principle of the meteorological echo to obtain a third judgment result; A fourth obtaining unit, wherein the fourth obtaining unit is used to remove noise points from the second denoised radar data according to the third judgment result, and to fill in the missing points to obtain radar data from which non-meteorological noise is filtered out.

9. A non-meteorological noise filtering device based on millimeter wave cloud radar, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images