THI millimeter wave cloud radar clear sky echo filtering method, device and equipment

Abstract

The present application relates to a method, device and equipment for filtering clear sky echoes of THI millimeter wave cloud radar, which uses the dual-modal features of spatiotemporal physical information to mark clear sky echoes or meteorological echoes for sample data of each channel in a sample data set, and constructs a corresponding fuzzy logic membership function according to the sample data of each channel after marking, and uses the corresponding fuzzy logic membership function to obtain the membership of clear sky echoes and meteorological echoes from the measured base data of different channels in the original THI data, and inputs the membership of each channel into a trained classification recognition model to identify the clear sky echoes and meteorological echoes in the original THI data, so as to filter out the clear sky echoes. This method can be used to accurately filter the clear sky echoes.

Description

Technical Field

[0001] The present application relates to the field of radar technology, and in particular to a method, device and equipment for filtering clear sky echoes of a THI millimeter-wave cloud radar. Background Art

[0002] Millimeter-wave cloud radar has the ability to detect the microphysical structure inside the cloud, such as particle size, liquid water content, and water droplet spectrum distribution, so it is widely used in meteorological observations. However, in actual observations, the observation results of millimeter-wave cloud radar are often contaminated by clear sky echoes, which greatly interferes with the identification of meteorological targets. There are two main scattering mechanisms of clear sky echoes: one is particle scattering, which is a point-like or thin-line echo, mainly caused by pollen, insects, birds, etc. The other is Bragg scattering, which is a layered echo, mainly caused by turbulent clumps in the atmosphere or uneven atmospheric refractive index. Generally speaking, compared with meteorological echoes, the clear sky echo distribution height is lower (generally below 3km), its linear depolarization ratio is larger, the reflectivity and signal-to-noise ratio are smaller, and the velocity spectrum distribution is more scattered. In this case, the clear sky and meteorological echoes are easier to distinguish. However, for the overlap of clear sky and low cloud echoes, and the reflectivity and other parameter values ​​of the clear sky echo are close to those of the low cloud echo, the two types of echoes are not easy to distinguish.

[0003] However, although the existing methods can effectively filter out clear sky echoes caused by pollen, insects, birds, etc., they do not consider the clear sky echoes caused by turbulent clumps in the atmosphere or uneven atmospheric refractive index, and have limitations in filtering range, low efficiency, low safety, or waste of resources. There are also some automatic recognition algorithms that recognize images pixel by pixel, ignoring the overall image features such as the connectivity of meteorological targets, which makes it easy to misjudge meteorological targets in edge areas, affecting the accuracy of echo type prediction. Summary of the invention

[0004] Based on this, it is necessary to provide a THI millimeter-wave cloud radar clear sky echo filtering method, device and equipment that can accurately identify clear sky echoes and meteorological echoes in response to the above technical problems.

[0005] A method for filtering clear sky echoes of a THI millimeter-wave cloud radar, the method comprising:

[0006] Acquire a sample data set, where the sample data set is multi-channel original THI data obtained by detecting the sky with a millimeter-wave cloud radar;

[0007] For the sample data of each channel in the sample data set, the clear sky echo or the weather echo is marked by using the dual-modal features of the spatiotemporal physical information, and the corresponding fuzzy logic membership function is constructed according to the sample data of each channel after marking;

[0008] Obtaining the measured original THI data, and using the corresponding fuzzy logic membership function to obtain the membership of the clear sky echo and the weather echo for the base data of different channels in the original THI data;

[0009] The membership degree of each channel is input into the trained classification recognition model to identify the clear sky echo and the weather echo in the original THI data, so as to filter out the clear sky echo.

[0010] In one embodiment, the original THI data includes base data of five channels, including reflectivity base data, velocity base data, spectrum width base data, signal-to-noise ratio base data and linear depolarization base data.

[0011] In one embodiment, when marking clear sky echo or weather echo using dual-modal features combined with spatiotemporal physical information:

[0012] First, the clear sky echo or weather echo is marked according to the dual-modal characteristics of the physical characteristics and spatiotemporal characteristics of the basic data, and then the clear sky echo or weather echo is further finely marked manually according to the characteristics of the velocity basic data and the spectral width basic data.

[0013] In one embodiment, when marking clear sky echoes or weather echoes according to the dual-modal characteristics of the physical characteristics and spatiotemporal characteristics of the base data:

[0014] The base data whose linear depolarization ratio is greater than -20 dB as the clear sky echo;

[0015] The base data of the reflectivity greater than -10dB are marked as weather echoes;

[0016] The base data with a height greater than 3 km are marked as weather echoes.

[0017] In one embodiment, when marking clear sky echo or weather echo according to the characteristics of velocity base data and spectrum width base data:

[0018] Marking the base data showing velocity spectrum spurious in the velocity base data as clear sky echo;

[0019] The basic data showing good connectivity in the spectrum width basic data are marked as meteorological echoes.

[0020] In one embodiment, constructing a corresponding fuzzy logic membership function according to the labeled sample data of each channel includes:

[0021] According to the markings of clear sky echo and weather echo, statistics are made on the distribution of basic data of each channel;

[0022] According to the statistical distribution of the base data, the kernel density estimation fitting and polynomial fitting methods are used to adaptively construct the corresponding fuzzy logic membership function, wherein the Gaussian kernel function is used for polynomial fitting.

[0023] In one embodiment, the classification and recognition model adopts the DeepLabV3 neural network.

[0024] The present application also provides a THI millimeter-wave cloud radar clear sky echo filtering device, the device comprising:

[0025] A sample data acquisition module is used to acquire a sample data set, where the sample data set is multi-channel original THI data obtained by detecting the sky with a millimeter-wave cloud radar;

[0026] A fuzzy logic membership function construction module is used to mark the sample data of each channel in the sample data set as clear sky echo or meteorological echo by using the dual-modal characteristics of spatiotemporal physical information, and to construct a corresponding fuzzy logic membership function according to the sample data of each channel after marking;

[0027] The membership obtaining module is used to obtain the measured original THI data, and obtain the membership of the clear sky echo and the weather echo by using the corresponding fuzzy logic membership function of the base data of different channels in the original THI data;

[0028] The clear sky echo filtering module is used to input the membership of each channel into the trained classification recognition model, identify the clear sky echo and the meteorological echo in the original THI data, so as to filter out the clear sky echo.

[0029] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0030] Acquire a sample data set, where the sample data set is multi-channel original THI data obtained by detecting the sky with a millimeter-wave cloud radar;

[0031] For the sample data of each channel in the sample data set, the clear sky echo or the weather echo is marked by using the dual-modal features of the spatiotemporal physical information, and the corresponding fuzzy logic membership function is constructed according to the sample data of each channel after marking;

[0032] Obtaining the measured original THI data, and using the corresponding fuzzy logic membership function to obtain the membership of the clear sky echo and the weather echo for the base data of different channels in the original THI data;

[0033] The membership degree of each channel is input into the trained classification recognition model to identify the clear sky echo and the weather echo in the original THI data, so as to filter out the clear sky echo.

[0034] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:

[0035] Acquire a sample data set, where the sample data set is multi-channel original THI data obtained by detecting the sky with a millimeter-wave cloud radar;

[0036] For the sample data of each channel in the sample data set, the clear sky echo or the weather echo is marked by using the dual-modal features of the spatiotemporal physical information, and the corresponding fuzzy logic membership function is constructed according to the sample data of each channel after marking;

[0037] Obtaining the measured original THI data, and using the corresponding fuzzy logic membership function to obtain the membership of the clear sky echo and the weather echo for the base data of different channels in the original THI data;

[0038] The membership degree of each channel is input into the trained classification recognition model to identify the clear sky echo and the weather echo in the original THI data, so as to filter out the clear sky echo.

[0039] The above-mentioned THI millimeter-wave cloud radar clear sky echo filtering method, device and equipment use the dual-modal characteristics of spatiotemporal physical information to mark the clear sky echo or meteorological echo by using the sample data of each channel in the sample data set, and construct the corresponding fuzzy logic membership function according to the sample data of each channel after marking, and use the corresponding fuzzy logic membership function to obtain the membership of the clear sky echo and the meteorological echo from the base data of different channels in the original THI data measured, and input the membership of each channel into the trained classification recognition model to identify the clear sky echo and the meteorological echo in the original THI data, so as to achieve the filtering of the clear sky echo. This method can be used to accurately filter the clear sky echo. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a schematic diagram of the process of a clear sky echo filtering method of a THI millimeter wave cloud radar in one embodiment;

[0041] Figure 2 FIG. 1 is a schematic diagram of the marking result of the dual-modal feature marking method for combining spatiotemporal and physical information in one embodiment, wherein: Figure 2 (a) is a schematic diagram of reflectivity base data. Figure 2 (b) is a schematic diagram of velocity base data. Figure 2 (c) is a schematic diagram of spectrum width basis data. Figure 2(d) is a schematic diagram of the signal-to-noise ratio base data. Figure 2 (e) is a schematic diagram of linear depolarization basis data. Figure 2 (f) Schematic diagram of the result after labeling using the dual-modal feature labeling method combining spatiotemporal and physical information;

[0042] Figure 3 Schematic diagram of clear sky echo data distribution and fitting results in one embodiment, wherein: Figure 3 (a) is a schematic diagram of the distribution and fitting results of LDR parameters. Figure 3 (b) is a schematic diagram of the distribution and fitting results of SNR parameters. Figure 3 (c) is a schematic diagram of the distribution and fitting results of the V parameter. Figure 3 (d) is a schematic diagram of the distribution and fitting results of the W parameter. Figure 3 (e) Schematic diagram of the distribution and fitting results of the Z parameter;

[0043] Figure 4 is a schematic diagram of weather echo data distribution and fitting results in one embodiment, Figure 4 (a) is a schematic diagram of the distribution and fitting results of LDR parameters. Figure 4 (b) is a schematic diagram of the distribution and fitting results of SNR parameters. Figure 4 (c) is a schematic diagram of the distribution and fitting results of the V parameter. Figure 4 (d) is a schematic diagram of the distribution and fitting results of the W parameter. Figure 4 (e) Schematic diagram of the distribution and fitting results of the Z parameter;

[0044] Figure 5 It is a schematic diagram of the overall implementation steps of the THI millimeter-wave cloud radar clear sky echo filtering method in one embodiment;

[0045] Figure 6 The following is a schematic diagram of the results of an experiment using this method to identify clear sky and weather echo data, where: Figure 6 (a) is a schematic diagram of reflectivity base data. Figure 6 (b) is a schematic diagram of velocity base data. Figure 6 (c) is a schematic diagram of spectrum width basis data. Figure 6 (d) is a schematic diagram of the signal-to-noise ratio base data. Figure 6 (e) is a schematic diagram of linear depolarization basis data. Figure 6 (f) is a schematic diagram of the identification results using this method;

[0046] Figure 7 It is a structural block diagram of a THI millimeter-wave cloud radar clear sky echo filtering device in one embodiment;

[0047] Figure 8 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0049] The existing clear sky echo filtering methods have limitations, including not considering some factors, which limits the filtering range, and have problems such as low efficiency, poor safety, and waste of resources. When using automatic recognition algorithms to process pixels one by one, the overall characteristics of the image are ignored, and it is easy to misjudge meteorological targets in edge areas, resulting in inaccurate predictions. Figure 1 As shown, a THI millimeter-wave cloud radar clear sky echo filtering method is provided, which specifically includes the following steps:

[0050] Step S100, obtaining a sample data set, where the sample data set is multi-channel original THI data obtained by detecting the sky with a millimeter-wave cloud radar.

[0051] Step S110, the sample data of each channel in the sample data set is marked as clear sky echo or weather echo using the characteristics of the combination of spatiotemporal physical information, and a corresponding fuzzy logic membership function is constructed based on the marked sample data of each channel.

[0052] Step S120, obtaining the measured original THI data, and using the corresponding fuzzy logic membership function to obtain the membership of the clear sky echo and the weather echo for the base data of different channels in the original THI data.

[0053] Step S130, inputting the membership degree of each channel into the trained classification recognition model, identifying the clear sky echo and the weather echo in the original THI data, so as to filter out the clear sky echo.

[0054] In this application, a model-data dual-driven non-meteorological target intelligent recognition architecture is proposed. After labeling the multi-channel original THI data through a dual-modal feature labeling method that combines motion and spatiotemporal physical information, the five parameter distributions of clear sky echoes and meteorological echoes are statistically analyzed based on the manual labels. A data-driven adaptive membership function construction method is then proposed, which uses the statistical distribution of the parameters to construct the corresponding fuzzy logic membership function, so that the original data is input into the fuzzy logic to obtain the corresponding membership. The neural network design method is combined with model-driven knowledge to fuse the multi-parameter membership output of the fuzzy logic, and finally obtain accurate target classification and recognition results.

[0055] Furthermore, this method avoids the problem of two types of echoes being difficult to distinguish in a single feature dimension through a dual-modal feature labeling method that combines spatiotemporal physical information. It avoids the shortcomings of traditional triangle, trapezoid and other membership functions that are too idealized through a data-driven adaptive membership function construction method, and more accurately describes the relationship between actual physical quantities and fuzzy concepts, thereby reducing model errors and improving recognition performance. The model-driven knowledge utilization neural network design method avoids the problem of insufficient utilization of prior physical knowledge in traditional neural networks, takes the membership of fuzzy logic as input, and fully utilizes prior knowledge about data distribution to obtain high recognition accuracy and high generalization results for the classification of meteorological targets and clear sky echoes.

[0056] In step S100, the original THI data is obtained by detecting the sky using a millimeter-wave cloud radar in a THI scanning mode. The original THI data includes base data of five channels, including reflectivity base data, velocity base data, spectrum width base data, signal-to-noise ratio base data, and linear depolarization base data.

[0057] In step S110, when the clear sky echo or the weather echo is marked by using the bimodal characteristics of the combined spatiotemporal physical information: the clear sky echo or the weather echo is first marked according to the bimodal characteristics of the physical characteristics of the base data and the spatiotemporal characteristics, and then the clear sky echo or the weather echo is marked according to the characteristics of the velocity base data and the spectral width base data.

[0058] Furthermore, when marking clear sky echoes or weather echoes according to the dual-mode characteristics of the physical characteristics and spatiotemporal characteristics of the base data, firstly, according to the physical characteristics of the base data, the base data with a linear depolarization ratio greater than -20dB is marked as a clear sky echo, and the base data with a reflectivity greater than -10dB is marked as a weather echo. Then, considering the spatiotemporal characteristics of the base data, the base data with a height greater than 3Km is marked as a weather echo.

[0059] Furthermore, based on the above judgment, the labels are manually annotated according to the characteristics of the velocity base data and the spectrum width base data to finely mark the clear sky echo or the weather echo. The base data with scattered velocity spectrum in the velocity base data is marked as clear sky echo, and the base data with good connectivity in the spectrum width base data is marked as weather echo. Figure 2 shown.

[0060] In this embodiment, constructing a corresponding fuzzy logic membership function based on the sample data of each channel after labeling includes: performing statistics on the distribution of basic data of each channel according to the labels of clear sky echoes and meteorological echoes, and adaptively constructing a corresponding fuzzy logic membership function using kernel density estimation fitting and polynomial fitting methods based on the statistical distribution of the basic data.

[0061] Specifically, according to the statistical distribution of the data, the KDE (Kernel Density Estimation) fitting and polynomial fitting methods are used to adaptively construct the membership function of fuzzy logic. First, kernel density estimation is used. Generally speaking, the probability density function There are the following forms:

[0062] (1)

[0063] In formula (1), For any value point, For the sample set Observations, is the sample size, The bandwidth is used to control the smoothness of the kernel function. If it is too large, it will not meet the requirements of kernel density estimation. If the value is too small, the sample data points will be too few and the error will be larger. When assigning values, a suitable intermediate value should be selected. In this embodiment, the value of the bandwidth is determined by Scott's law:

[0064] (2)

[0065] In formula (2), is the number of data points, is the dimension of the data.

[0066] is a kernel function. In this embodiment, the kernel function is a Gaussian kernel function (Gaussian KernrlFunction), which is in the following form:

[0067] (3)

[0068] In actual tests, KDE fitting has a better effect, but it takes too long to run, so it can be used as a fitting option for slow processing. In fast processing, polynomial curve fitting is proposed to optimize efficiency.

[0069] Specifically, let the original data be , where the minimum value is , the maximum value is ,by and are the left and right endpoints, divided equally into intervals, and statistically analyze the distribution of the data, we get ,in Indicates The median of the intervals, Indicates The frequency of the interval.

[0070] Now use the polynomial function to Perform the fitting

[0071] (4)

[0072] In formula (4), is the highest order of the polynomial, is the coefficient to be determined, and its optimal estimate Determined by the following formula

[0073] (5)

[0074] like Figure 3 and Figure 4 As shown, the clear sky echo data distribution and fitting results, and the meteorological echo data distribution and fitting results are respectively.

[0075] Furthermore, the maximum value of the echo is normalized and the polynomial parameters are saved in a specific folder. For an element point in the sample to be processed, the fitting parameters of the clear sky echo and the meteorological echo are substituted into the calculation to obtain the membership value of each parameter of the data point.

[0076] In step S130, the classification and recognition model uses the DeepLabV3 neural network. When training the classification and recognition model, the five-channel base data is processed using the constructed fuzzy logic membership function to obtain the data's membership to the clear sky echo and the weather echo, and then the membership is input into the DeepLabV3 neural network for training, to obtain a classification and recognition model that can accurately classify the clear sky echo and the weather echo according to the membership. The trained classification and recognition model can realize pixel-level classification and recognition of the input data. When training the DeepLabV3 neural network, the existing technology is used, which will not be described in detail here.

[0077] Furthermore, according to the classification results output by the classification recognition model, the data identified as clear sky echo is filtered out, thereby achieving the filtering of clear sky echo.

[0078] like Figure 5 The figure shows the overall process flow diagram of the method.

[0079] In order to verify the effectiveness of the method proposed in this paper, experiments were conducted, and the results are as follows Figure 6 As shown. The actual measurement data of millimeter wave cloud radar is selected. Figure 6 The original base data and the result after clear sky echo filtering using this method.

[0080] Specifically, Figure 6 The first five sub-graphs from left to right and from top to bottom represent reflectivity, velocity, spectral width, signal-to-noise ratio, and linear depolarization ratio respectively. The sixth sub-graph shows the results of identification and classification, where red represents weather echoes and green represents clear sky echoes. The horizontal axes of the six sub-graphs represent time, from 8 am to 8 am the next day, and the vertical axes represent altitude, from 0 km to 15 km.

[0081] Further, such as Figure 6 As shown, in the low-altitude part, there is aliasing of clear-air echo and weather echo in the original base data. After adopting the method provided by the present invention, the clear-air echo aliased with the low-altitude weather echo can be well identified.

[0082] In the above-mentioned THI millimeter-wave cloud radar clear sky echo filtering method, firstly, the fuzzy logic method is used to extract features from the original base data, which improves the interpretability of the intelligent non-meteorological target recognition model. Secondly, unlike the BP neural network architecture in the prior art, which ignores the overall characteristics of the THI image, this method uses the DeepLabV3 network architecture. On the basis of using the five-channel parameter data features, it comprehensively utilizes higher-level features such as the connectivity of meteorological targets. The model data dual-driven intelligent non-meteorological target recognition architecture proposed in this application not only introduces prior physical knowledge into the intelligent non-meteorological target recognition model, but also utilizes the powerful deep-level feature extraction capability of the neural network, effectively integrates the membership of multiple parameter channels, and obtains a model with high recognition accuracy and high generalization ability.

[0083] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0084] In one embodiment, Figure 7 As shown, a THI millimeter wave cloud radar clear sky echo filtering device is provided, comprising: a sample data acquisition module 200, a fuzzy logic membership function construction module 210, a membership acquisition module 220 and a clear sky echo filtering module 230, wherein:

[0085] The sample data acquisition module 200 is used to acquire a sample data set, where the sample data set is multi-channel original THI data obtained by detecting the sky with a millimeter-wave cloud radar;

[0086] A fuzzy logic membership function construction module 210 is used to mark the sample data of each channel in the sample data set as clear sky echo or weather echo by using the dual-modal features of spatiotemporal physical information, and to construct a corresponding fuzzy logic membership function according to the sample data of each channel after labeling;

[0087] The membership obtaining module 220 is used to obtain the measured original THI data, and obtain the membership of the clear sky echo and the weather echo by using the corresponding fuzzy logic membership function of the base data of different channels in the original THI data;

[0088] The clear sky echo filtering module 230 is used to input the membership of each channel into the trained classification recognition model, identify the clear sky echo and the meteorological echo in the original THI data, so as to filter out the clear sky echo.

[0089] For the specific definition of the THI millimeter-wave cloud radar clear sky echo filtering device, please refer to the definition of the THI millimeter-wave cloud radar clear sky echo filtering method above, which will not be repeated here. Each module in the above-mentioned THI millimeter-wave cloud radar clear sky echo filtering device can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0090] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 8 As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a THI millimeter-wave cloud radar clear sky echo filtering method is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0091] Those skilled in the art will understand that Figure 8 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0092] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0093] Acquire a sample data set, where the sample data set is multi-channel original THI data obtained by detecting the sky with a millimeter-wave cloud radar;

[0094] For the sample data of each channel in the sample data set, the clear sky echo or the weather echo is marked by using the dual-modal features of the spatiotemporal physical information, and the corresponding fuzzy logic membership function is constructed according to the sample data of each channel after marking;

[0095] Obtaining the measured original THI data, and using the corresponding fuzzy logic membership function to obtain the membership of the clear sky echo and the weather echo for the base data of different channels in the original THI data;

[0096] The membership degree of each channel is input into the trained classification recognition model to identify the clear sky echo and the weather echo in the original THI data, so as to filter out the clear sky echo.

[0097] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0098] Acquire a sample data set, where the sample data set is multi-channel original THI data obtained by detecting the sky with a millimeter-wave cloud radar;

[0099] For the sample data of each channel in the sample data set, the clear sky echo or the weather echo is marked by using the dual-modal features of the spatiotemporal physical information, and the corresponding fuzzy logic membership function is constructed according to the sample data of each channel after marking;

[0100] Obtaining the measured original THI data, and using the corresponding fuzzy logic membership function to obtain the membership of the clear sky echo and the weather echo for the base data of different channels in the original THI data;

[0101] The membership degree of each channel is input into the trained classification recognition model to identify the clear sky echo and the weather echo in the original THI data, so as to filter out the clear sky echo.

[0102] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0103] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0104] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention. It should be pointed out that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A THI millimeter wave cloud radar clear sky echo filtering method, characterized in that: The method comprises: Acquire a sample data set, where the sample data set is multi-channel original THI data obtained by detecting the sky with a millimeter-wave cloud radar; For the sample data of each channel in the sample data set, the clear sky echo or the weather echo is marked by using the dual-modal features of the spatiotemporal physical information, and the corresponding fuzzy logic membership function is constructed according to the sample data of each channel after marking; Obtaining measured original THI data, and using corresponding fuzzy logic membership functions for base data of different channels in the original THI data to obtain membership of clear sky echo and weather echo; The membership degree of each channel is input into the trained classification recognition model to identify the clear sky echo and the weather echo in the original THI data, so as to filter out the clear sky echo.

2. The THI millimeter wave cloud radar clear sky echo filtering method according to claim 1 is characterized in that: The original THI data includes base data of five channels, including reflectivity base data, velocity base data, spectrum width base data, signal-to-noise ratio base data and linear depolarization base data.

3. The THI millimeter wave cloud radar clear sky echo filtering method according to claim 2 is characterized in that: When using the dual-modal features of spatiotemporal physical information to mark clear sky echoes or weather echoes: First, the clear sky echo or weather echo is marked according to the dual-modal characteristics of the physical characteristics and spatiotemporal characteristics of the basic data, and then the clear sky echo or weather echo is further refined manually according to the characteristics of the velocity basic data and the spectral width basic data.

4. The THI millimeter wave cloud radar clear sky echo filtering method according to claim 3 is characterized in that: When marking clear sky echoes or weather echoes based on the dual-modal characteristics of the base data's physical characteristics and spatiotemporal characteristics: The base data whose linear depolarization ratio is greater than -20 dB as the clear sky echo; The base data of the reflectivity greater than -10dB are marked as weather echoes; The base data with a height greater than 3 km are marked as weather echoes.

5. The THI millimeter wave cloud radar clear sky echo filtering method according to claim 4 is characterized in that: When marking clear sky echoes or weather echoes based on the characteristics of velocity-based data and spectrum-width-based data: Marking the base data showing velocity spectrum spurious in the velocity base data as clear sky echo; The basic data showing good connectivity in the spectrum width basic data are marked as meteorological echoes.

6. The THI millimeter wave cloud radar clear sky echo filtering method according to any one of claims 1 to 5, characterized in that: The step of constructing a corresponding fuzzy logic membership function according to the labeled sample data of each channel includes: According to the markings of clear sky echo and weather echo, statistics are made on the distribution of basic data of each channel; According to the statistical distribution of the base data, the kernel density estimation fitting and polynomial fitting methods are used to adaptively construct the corresponding fuzzy logic membership function, wherein the Gaussian kernel function is used for polynomial fitting.

7. The THI millimeter wave cloud radar clear sky echo filtering method according to claim 6, characterized in that: The classification and recognition model adopts the DeepLabV3 neural network.

8. A THI millimeter wave cloud radar clear sky echo filtering device, characterized in that: The device comprises: A sample data acquisition module is used to acquire a sample data set, where the sample data set is multi-channel original THI data obtained by detecting the sky with a millimeter-wave cloud radar; A fuzzy logic membership function construction module is used to mark the sample data of each channel in the sample data set as clear sky echo or meteorological echo by using the dual-modal characteristics of spatiotemporal physical information, and to construct a corresponding fuzzy logic membership function according to the sample data of each channel after marking; The membership acquisition module is used to obtain the measured original THI data, and obtain the membership of the clear sky echo and the weather echo by using the corresponding fuzzy logic membership function of the base data of different channels in the original THI data; The clear sky echo filtering module is used to input the membership of each channel into the trained classification recognition model, identify the clear sky echo and the meteorological echo in the original THI data, so as to filter out the clear sky echo.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to claim 6 or 7 are implemented.

Images