Pipeline fluid concentration detection method and device based on millimeter wave radar

Abstract

The invention discloses a pipeline fluid concentration detection method and device based on a millimeter wave radar, and the method comprises the steps: transmitting a linear frequency modulation signal to a pipeline fluid at a specific distance through the millimeter wave radar, and carrying out the data acquisition of a reflected intermediate frequency signal. And performing fast Fourier transform processing on the obtained intermediate frequency signal information to obtain position information of the liquid and the reflecting surface. And feature extraction is performed on the expanded radar data through a pre-trained specific liquid concentration identification neural network, and attention mechanism processing is performed on liquid and reflecting surface information at a feature level, so that refined full-range liquid concentration detection is realized. The liquid concentration condition in the pipeline fluid can be accurately reflected. Meanwhile, the information is transmitted to the management terminal through a specific data protocol.

Description

Technical Field

[0001] The present invention relates to the field of pipeline fluid detection, and in particular to a pipeline fluid concentration detection method and device based on millimeter wave radar. Background Art

[0002] With the continuous development of industrial technology, the demand for liquid detection technology in the fields of chemical, petroleum and pharmaceutical is increasing. Liquid detection in these fields usually involves the precise measurement and monitoring of fluid concentration. Traditional methods for detecting liquid concentration include chemical analysis, conductivity measurement and optical measurement. These methods rely on the photoelectric properties and various chemical properties of the liquid, and usually require the use of precision instruments such as Raman analyzers. However, these professional instruments are expensive and complicated to operate, which limits their widespread application in large-scale production. In addition, traditional methods are usually offline detection, which has lags and is difficult to adjust in real time during the production process. This detection method lacks the ability to make real-time judgments at multiple points in system engineering, and usually only performs intermittent detection at some points, which affects the efficiency and accuracy of production monitoring.

[0003] In recent years, with the continuous development of wireless sensing technology, detecting liquid characteristics by analyzing the changes in wireless signals after penetrating liquids has become a fast, convenient, and low-cost liquid detection solution. In particular, its non-contact measurement characteristics have attracted the attention of more and more researchers. The application of wireless sensing technology in liquid detection overcomes many limitations of traditional methods, can perform rapid detection without contacting the liquid, and reduces the cost and complexity of detection.

[0004] However, current wireless sensing technology mainly focuses on the identification of liquid types, and there is little research on the detection of liquid concentration. In addition, existing methods still face challenges in terms of recognition accuracy and recognition robustness. For example, the offset of the liquid position will lead to unstable recognition results. These problems limit the application of wireless sensing technology in liquid concentration detection. Summary of the invention

[0005] The present invention aims to address the deficiencies of the prior art and to propose a method and device for detecting pipeline fluid concentration based on millimeter wave radar.

[0006] The objective of the present invention is achieved through the following technical solution: a pipeline fluid concentration detection method based on millimeter wave radar, the method comprising the following steps:

[0007] S1. Use millimeter-wave radar in a closed environment to transmit linear frequency modulation signals to both ends of the pipeline mirror to be inspected, receive the reflected signal to obtain the original data, and extract the strong correlation radar information between the liquid and the reflection surface according to the original data;

[0008] S2. Construct a concentration dataset of the full range of pipeline fluids; classify them according to the minimum concentration required for detection and generate corresponding true labels;

[0009] S3. Use the concentration data set to train a neural network that recognizes the full range of concentrations. The neural network extracts one-dimensional features from strongly correlated radar information, extracts high-dimensional features from the original data through a complex domain multilayer perceptron, uses an attention mechanism to fuse features, and decodes the fused features to obtain concentration prediction results.

[0010] S4. Input the original data and the strongly correlated radar information into a neural network trained to identify the full range of concentrations to obtain concentration prediction results.

[0011] Furthermore, the use of millimeter-wave radar in a closed environment to transmit linear frequency modulation signals to both ends of the pipeline mirror to be inspected specifically includes: installing a box-type structure at both ends of the pipeline liquid mirror to transmit the linear frequency modulation signal, the millimeter-wave radar, the corresponding power adapter and the embedded device are installed in the box-type structure, the transceiver antenna of the millimeter-wave radar is exposed in the field of view of the mirror, and the other parts are isolated by metal plates.

[0012] Furthermore, the obtaining of the original data includes performing mixing processing on the received reflected signal through a mixer to generate the original data of the intermediate frequency signal seat radar.

[0013] Furthermore, the extraction of strongly correlated radar information reflected by the liquid and the reflecting surface includes: performing fast Fourier transform on the original data to obtain a frequency domain signal, sorting the frequency domain signal intensities within a set distance range, and obtaining three peak signals as the radar signal that first reaches the mirror surface, the radar signal that first penetrates the mirror, and the radar signal that first reaches the reflecting surface; according to the subscript of the signal location and the distance resolution of the radar, the positions of the three target points are calculated, and the distance from the millimeter wave radar to the reflecting surface is used as a reference to divide the entire radar information into multiple sections, and the three signals are stacked in the same dimension in these sections to obtain strongly correlated radar information reflected by the liquid and the reflecting surface.

[0014] Furthermore, the neural network for identifying the full range of concentrations uses modules in a deep complex network model as basic modules in the network, including: a data expansion layer, a feature extraction layer, an attention mechanism, and a decoding part;

[0015] The data expansion layer expands the radar data in the original data from two dimensions to three dimensions by duplication, so that the original data changes from the dimension of number of antennas × number of samples to the dimension of number of antennas × number of samples × number of samples;

[0016] The feature extraction layer includes a multi-layer perceptron to extract high-dimensional features from the original radar data, and also includes a one-dimensional perceptron to extract information from the strongly correlated radar information reflected by the liquid and the reflective surface;

[0017] In the attention mechanism, the high-dimensional feature part of the original radar data is divided into H×1×C and 1×W×C parts, which are respectively combined with the reflection of the reflecting surface to form a strong correlation radar information feature, and the multi-head attention mechanism operation is performed to perform dimension mapping, map to the required number of attention heads, and then broadcast the obtained results, expand the two results to the same dimension and then add them;

[0018] The decoding part includes multiple complex fully connected layers to reduce the dimension of the result to one dimension, and then convert the result of the complex domain to the real domain by taking the modulus, and finally obtain the probability of the corresponding concentration gradient through the softmax function.

[0019] Furthermore, the multi-layer perceptron model consists of a number of 2-dimensional 3*3 complex convolution modules, a complex activation function module, a complex pooling module and a complex batch normalization module.

[0020] Furthermore, the complex activation function module adopts the CReLU function, and the complex batch normalization module performs a 0 mean 1 variance operation on the input:

[0021]

[0022] Where x is the input vector and V is the correlation matrix, where the correlation matrix V is calculated as follows

[0023]

[0024] The final BN has two parameters γ and β, so the final batch normalization result is as follows

[0025]

[0026] Furthermore, in the training process, the cross entropy function is selected as the loss function, and its calculation method is:

[0027]

[0028] Where N is the number of samples, C is the number of categories obtained based on the minimum concentration resolution, and x y is the probability that the i-th sample belongs to the true category y, x c is the probability of the i-th sample data category c.

[0029] On the other hand, the present invention specification also provides a pipeline fluid concentration detection device based on millimeter wave radar, including a memory and one or more processors, wherein the memory stores executable code, and when the processor executes the executable code, the pipeline fluid concentration detection method based on millimeter wave radar is implemented.

[0030] On the other hand, the present invention specification also provides a computer-readable storage medium on which a program is stored. When the program is executed by a processor, the method for detecting pipeline fluid concentration based on millimeter wave radar is implemented.

[0031] Beneficial effects of the present invention: The pipeline fluid concentration detection method and system based on millimeter wave radar described in the present invention, by designing a fully enclosed metal material external plug-in detection environment system, the system is externally mounted on both ends of the liquid pipeline to be detected to form a closed metal environment, and the millimeter wave radar sends a linear frequency modulation signal in the closed metal environment. After the signal penetrates the liquid and reaches the metal reflective surface, it is reflected multiple times, and the data of the millimeter wave penetrating the liquid and reaching the reflective surface multiple times is obtained, which greatly enriches the proportion of valid data in the received data, and improves the accuracy and robustness of the neural network by superimposing the data of multiple penetrations at the data level. At the same time, the data dimension is expanded by replication, and the two-dimensional data is converted into three-dimensional data, so that the network can perform more complex training.

[0032] The neural network model established by the present invention for pipeline fluid concentration detection mainly realizes efficient and reasonable use of radar intermediate frequency complex data by reusing the complex modules in the deep complex network. The perceptron model composed of complex modules is used as a feature extraction module to extract data from the original data, and the data of millimeter waves penetrating the liquid multiple times and reaching the reflecting surface are trained with the high-dimensional features of the original data through the parameter matrix to perform attention mechanism training, so as to realize long-distance serialization features and correlation fusion. The above method realizes the refined step-by-step recognition of liquid concentration in the whole range, and can recognize data in the concentration range of 0.5%. At the same time, the concentration that is not put into the network training can also be distinguished as the corresponding step concentration, realizing a more stable and smooth concentration recognition.

[0033] The pipeline fluid concentration detection method and system based on millimeter wave radar of the present invention can be easily and efficiently installed outside the required detection liquid pipeline sight glass through the combination of millimeter wave radar, embedded system and external system, and is easy to disassemble and assemble. The cumbersome process and complex training process of manual detection are also avoided through online detection. It is easy to use and has a wide range of applications, especially for high-risk liquid concentration detection, avoiding potential safety hazards. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1A flow chart of the main steps of a pipeline fluid concentration detection method based on millimeter wave radar provided in an embodiment of the present invention;

[0035] Figure 2 This is a schematic diagram of the fully enclosed metal texture detection environment system of the present invention;

[0036] Figure 3 This is a schematic diagram of the radar signal echo in liquids of different concentrations, taking 1% and 10% salt water as an example, where there are eight echo curves representing the dimensional data of the transmitting antenna x the receiving antenna;

[0037] Figure 4 This is a schematic diagram of the neural network structure of the present invention;

[0038] Figure 5 A schematic diagram of the attention mechanism module for multi-feature fusion adopted by the neural network of the present invention;

[0039] Figure 6 It is a schematic diagram of the accuracy of the present invention at a saline concentration interval of 0.5%;

[0040] Figure 7 This is a schematic diagram of the accuracy of the present invention when the 1% concentration interval is used as the training sample and the 0%-1% random concentration interval is used as the test sample, wherein the accuracy is calculated by rounding the test set concentration;

[0041] Figure 8 A schematic diagram of a pipeline fluid concentration detection device based on millimeter wave radar provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0042] The specific implementation modes of the present invention are further described in detail below with reference to the accompanying drawings.

[0043] like Figure 1 As shown, the present invention provides a pipeline fluid concentration detection method and device based on millimeter wave radar.

[0044] In order to make the above-mentioned purposes, features and advantages of the present invention more clearly understood, the specific implementation of the present invention is described in detail below in conjunction with the accompanying drawings. Here, the exemplary implementation of the present invention is only an application case of the present invention, but is not intended to limit the present invention.

[0045] First, a linear frequency modulation signal is transmitted toward the direction of the sight glass through multiple antennas of the millimeter-wave radar. In the embodiment of the present invention, a special structure is used to transmit and receive the signal. Two symmetrical box structures are tightly connected at both ends of the pipeline liquid sight glass to be detected by screws, thereby ensuring the formation of a closed detection environment. The millimeter-wave radar, the corresponding power adapter and the embedded device are installed in the box on one side of the system, and a position for the millimeter-wave radar cabin is reserved. The other side of the box is a metal plate as a millimeter-wave reflector. The transceiver antenna of the millimeter-wave radar is exposed in the field of view of the sight glass, while the other parts are isolated by metal plates to achieve reflection of the millimeter waves. One side of the embedded board is coated with insulating material to avoid short circuit. In order to ensure the corrosion resistance and durability of the system, the metal material is 316 austenitic stainless steel, which is widely used in the chemical industry. This stainless steel material not only has excellent corrosion resistance, but also can be used for a long time in harsh industrial environments to ensure the stability and reliability of the detection system. Its specific embodiment can be seen. Figure 2 .

[0046] The reflected signal is received by multiple receiving antennas. The received signal is mixed by a mixer to generate an intermediate frequency signal as the raw data of the reflected data, that is, the reflected time domain signal. In order to show that the reflected time domain signal shows different results for liquids of different concentrations, Figure 3 You can see the echo diagram of the radar signal in liquids of different concentrations. Figure 3 The left side shows the echo data of 1% salt water concentration. Figure 3 The right side shows the echo data of 10% concentration salt water, in which eight echo curves represent the dimensional data of the transmitting antenna x the receiving antenna. It can be seen that the echo data reflects different signals under different concentrations of salt water, which is also the basis for the present invention.

[0047] Perform a fast Fourier transform (FFT) on the time domain signal to obtain a frequency domain signal. The different peak areas in the frequency domain signal are the locations where the radar detects the object. In actual operation, three peak signals are concerned, namely the radar signal that reaches the mirror surface for the first time, the radar signal that penetrates the mirror for the first time, and the radar signal that reaches the reflecting surface for the first time. The specific steps are as follows: first, sort the frequency domain signal strength within the set distance range, and select the positions of the three values ​​with the strongest signal strength. These positions are the three signal positions of concern, among which the strongest signal is generally the reflecting surface signal. According to the subscript of the signal location and the distance resolution of the radar, the positions of the three target points are calculated. . Taking the distance from the millimeter wave radar to the reflecting surface as a reference, the entire radar information is divided into multiple segments. In these segments, the radar signals that reach the mirror surface multiple times, the radar signals that penetrate the mirror multiple times, and the radar signals that reach the reflecting surface multiple times are stacked in the same dimension to obtain the main liquid information data. Among them, due to the influence of the data resolution of the millimeter-wave radar, the position may not be accurate. During operation, it is often necessary to stack 1 to 2 more sampling point data to improve accuracy and robustness.

[0048] The above method is used to construct a concentration dataset of the full range of pipeline fluids. Classification is performed according to the minimum concentration required for detection, and the corresponding true labels are generated. When constructing the dataset, the radar data of a single chirp is used as the original input data. The original data file is saved when constructing the sample set. The specific liquid data information of the radar signals that reach the mirror surface multiple times, the radar signals that penetrate the mirror multiple times, and the radar signals that reach the reflecting surface multiple times are loaded by parsing during data loading.

[0049] Considering that the original data of a single chirp can only constitute two-dimensional data of (number of transmitting antennas × number of receiving antennas) × number of samples, only one-dimensional convolution operation can be performed in the neural network, making it difficult to perform deeper feature mining. Therefore, after loading the radar data, the original data needs to be expanded. By copying, the original radar data is expanded into three-dimensional data of (number of transmitting antennas × number of receiving antennas) × number of samples × number of samples. This expansion method allows a general two-dimensional convolution operation to be performed in the neural network, thereby breaking through the limitation of network depth and enhancing the network's fitting and learning capabilities.

[0050] For the neural network model, its basic architecture mainly uses the basic structure in the deep complex network to process the complex data of the millimeter wave radar, so as to efficiently and reasonably utilize the complex data reflected by the radar. Figure 4 Specifically, for the convolution operation, assuming that for a learnable convolution kernel W = A + iB, the input vector is h = x + iy, the convolution operation is expressed as the following formula:

[0051] W*h=(A*xB*y)+i(B*xA*y)

[0052] Where A and B are the real and imaginary matrices of the complex convolution kernel, respectively, and x and y are the real and imaginary parts of the input vector, respectively.

[0053] The activation function adopts the CReLU function, that is

[0054]

[0055] in represents the real part, Represents the imaginary part.

[0056] For batch normalization in the complex domain, the following operation is performed to perform a 0 mean and 1 variance operation on the input:

[0057]

[0058] Where x is the input vector, E[x] is the expectation of the input batch data. In practice, the mean is used instead of the expectation. V is the correlation matrix, where the correlation matrix V is calculated as follows

[0059]

[0060] The final BN has two parameters γ and β, where γ is the scaling factor and β is the offset, which enable the network to recover and adjust the normalized data distribution. Therefore, the final batch normalization result is as follows

[0061]

[0062] In addition, the complex pooling operation is similar to the activation function, and the linear fully connected layer is similar to the convolutional layer, which will not be repeated here.

[0063] The above basic modules form a complex domain multi-layer perceptron model. Each multi-layer perceptron model consists of a 3*3 convolution layer, an activation function, an average pooling layer, and a regularization layer. The feature extraction layer is composed of multiple multi-layer perceptron models to extract high-dimensional features from the radar's raw data. For the above stacked radar information with strong correlation reflected by liquid and reflective surface, a one-dimensional perceptron is used to extract features.

[0064] At the attention mechanism level, the high-dimensional features of the original radar data and the strongly correlated radar information features composed of liquid and reflective surface reflections are first dimensionally mapped to the required number of attention heads through learnable parameter matrices and vectors, where the high-dimensional features of the original radar data are used as the query part, the strongly correlated radar information features composed of reflective surface reflections are used as the key part, and the value is still formed by the high-dimensional features of the original radar data through the parameter matrix. At the attention mechanism calculation level, since the high-dimensional feature part of the original radar data is three-dimensional data and the strongly correlated radar information features composed of reflective surface reflections are two-dimensional data, in order to match the dimensions and reduce the amount of calculation, the high-dimensional feature part of the original radar data is divided into H×1×C and 1×W×C parts, and the multi-head attention mechanism operation is performed with the strongly correlated radar information features composed of reflective surface reflections, and then the results are broadcasted, and the two results are expanded to the same dimension and then added. The specific implementation can be seen. Figure 5 Schematic diagram of the attention mechanism.

[0065] The decoding operation of the neural network mainly expands the dimension through the above-mentioned multi-layer perceptron, and then reduces the dimension of the result to one dimension through multiple complex fully connected layers, and then converts the result of the complex domain to the real domain by taking the modulus, and then obtains the probability of the corresponding concentration gradient through the softmax function. During the training process, the loss can be transferred through this probability. During the prediction process, the concentration with the highest probability is selected as the actual liquid concentration under the corresponding concentration gradient.

[0066] During the training process, the cross entropy function is selected as the loss function, and its calculation method is:

[0067]

[0068] Where N is the number of samples, C is the number of categories obtained based on the minimum concentration resolution, and x y is the probability that the i-th sample belongs to the true category y, x c is the probability of the i-th sample data category c.

[0069] Also, Adam is used as the optimizer during training. The learning rate is initialized to 0.001, eps is set to 1e-8 to prevent division by zero, and weight_decay is set to 0.0001 to achieve L2 regularization. The model parameters are decayed at each update to prevent overfitting.

[0070] During the training phase, the number of epochs should be no less than 20 to allow the network to converge to the optimal value while ensuring its accurate performance. In addition, the average test loss on the test set should be less than 0.01, which can generally be used in actual scenarios.

[0071] After the training is completed, select the model and export it to onnx mode. Deploy the model in an embedded system to achieve online real-time detection of pipeline fluid concentration through automatic configuration, and transmit the predicted liquid concentration back to the control center through serial communication via the modbus protocol.

[0072] In order to prove the feasibility of the full concentration step detection described in the present invention, a salt water concentration test experiment with a concentration interval of 0.5% was first carried out. The accuracy of the confusion matrix of the results is as follows: Figure 6 As shown in the figure, the horizontal axis is the predicted concentration and the vertical axis is the actual concentration. In addition, we used 1% salt water concentration as a step training and 0%-1% random concentration interval as a test. The results are shown in the figure. Figure 7 As shown, it can be seen that without adding sample concentration gradient, the network can basically classify it into the correct concentration range, basically realizing the detection of the full range of liquid concentration. For the convenience of display, the horizontal and vertical axes are not marked with concentration, but only with concentration step-by-step marking, with 0% as the starting concentration and 0.5% or 1% as the step-by-step increase.

[0073] Corresponding to the aforementioned embodiment of a pipeline fluid concentration detection method based on millimeter wave radar, the present invention also provides an embodiment of a pipeline fluid concentration detection device based on millimeter wave radar.

[0074] See also Figure 8 An embodiment of the present invention provides a pipeline fluid concentration detection device based on millimeter wave radar, including a memory and one or more processors, wherein the memory stores executable code, and when the processor executes the executable code, it is used to implement a pipeline fluid concentration detection method based on millimeter wave radar in the above embodiment.

[0075] An embodiment of a pipeline fluid concentration detection device based on millimeter wave radar provided by the present invention can be applied to any device with data processing capabilities, and the device with data processing capabilities can be a device or apparatus such as a computer. The device embodiment can be implemented through software, or through hardware or a combination of software and hardware. Taking software implementation as an example, as a device in a logical sense, it is formed by the processor of any device with data processing capabilities in which it is located reading the corresponding computer program instructions in the non-volatile memory into the internal memory for execution. From a hardware perspective, if Figure 8 As shown in FIG. 1 , a hardware structure diagram of a pipeline fluid concentration detection device based on millimeter wave radar provided by the present invention is provided in any device with data processing capability, except Figure 8In addition to the processor, memory, network interface, and non-volatile memory shown, any device with data processing capabilities in which the apparatus in the embodiments is located may also include other hardware, generally based on the actual functions of the device with data processing capabilities, which will not be described in detail.

[0076] The implementation process of the functions and effects of each unit in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, and will not be repeated here.

[0077] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The device embodiment described above is only schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of the present invention. Ordinary technicians in this field can understand and implement it without paying creative work.

[0078] An embodiment of the present invention further provides a computer-readable storage medium having a program stored thereon. When the program is executed by a processor, a pipeline fluid concentration detection method based on millimeter wave radar in the above embodiment is implemented.

[0079] The computer-readable storage medium may be an internal storage unit of any device with data processing capability described in any of the aforementioned embodiments, such as a hard disk or a memory. The computer-readable storage medium may also be an external storage device of any device with data processing capability, such as a plug-in hard disk, a smart media card (SMC), an SD card, a flash card, etc. equipped on the device. Furthermore, the computer-readable storage medium may also include both an internal storage unit and an external storage device of any device with data processing capability. The computer-readable storage medium is used to store the computer program and other programs and data required by any device with data processing capability, and may also be used to temporarily store data that has been output or is to be output.

[0080] The present invention also provides a computer program product, including a computer program, and when the computer program is executed by a processor, the method for detecting pipeline fluid concentration based on millimeter wave radar is implemented.

[0081] Those skilled in the art will readily appreciate other embodiments of the present application after considering the description and practicing the contents disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present application. The description and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the claims.

[0082] It should be understood that the above general description and the detailed description below are only exemplary and explanatory and cannot limit the present application. The present application is not limited to the precise structure described above and shown in the drawings, and various modifications and changes can be made without departing from the scope thereof. The scope of the present application is limited only by the attached claims.

Claims

1. A pipeline fluid concentration detection method based on millimeter wave radar, characterized in that: The method comprises the following steps: S1. Use millimeter-wave radar in a closed environment to transmit linear frequency modulation signals to both ends of the pipeline mirror to be inspected, receive the reflected signal to obtain the original data, and extract the strong correlation radar information between the liquid and the reflection surface according to the original data; S2. Construct a concentration dataset of the full range of pipeline fluids; classify them according to the minimum concentration required for detection and generate corresponding true labels; S3. Use the concentration data set to train a neural network that recognizes the full range of concentrations. The neural network extracts one-dimensional features from strongly correlated radar information, extracts high-dimensional features from the original data through a complex domain multilayer perceptron, uses an attention mechanism to fuse features, and decodes the fused features to obtain concentration prediction results. S4. Input the original data and the strongly correlated radar information into a neural network trained to identify the full range of concentrations to obtain concentration prediction results.

2. The method for detecting pipeline fluid concentration based on millimeter wave radar according to claim 1, characterized in that: The method of using a millimeter-wave radar in a closed environment to transmit a linear frequency modulation signal to both ends of a pipeline mirror to be inspected specifically includes: installing a box-type structure at both ends of the pipeline liquid mirror to transmit a linear frequency modulation signal, the millimeter-wave radar, a corresponding power adapter and an embedded device are installed in the box-type structure, the transceiver antenna of the millimeter-wave radar is exposed in the field of view of the mirror, and the other parts are isolated by a metal plate.

3. The method for detecting pipeline fluid concentration based on millimeter wave radar according to claim 1, characterized in that: The acquisition of raw data includes performing mixing processing on the received reflected signal through a mixer to generate raw data of the intermediate frequency signal seat radar.

4. The method for detecting pipeline fluid concentration based on millimeter wave radar according to claim 3 is characterized in that: The method for extracting the strongly correlated radar information reflected by the liquid and the reflecting surface comprises: performing fast Fourier transform on the original data to obtain a frequency domain signal, sorting the frequency domain signal strength within a set distance range, and obtaining three peak signals as the radar signal that first reaches the mirror surface, the radar signal that first penetrates the mirror, and the radar signal that first reaches the reflecting surface; calculating the positions of the three target points according to the subscript of the signal position and the distance resolution of the radar, dividing the entire radar information into multiple sections with reference to the distance from the millimeter wave radar to the reflecting surface, and stacking the three signals in the same dimension in these sections to obtain the strongly correlated radar information reflected by the liquid and the reflecting surface.

5. The method for detecting pipeline fluid concentration based on millimeter wave radar according to claim 1, characterized in that: The neural network for identifying the full range of concentrations uses modules in a deep complex network model as basic modules in the network, including: a data expansion layer, a feature extraction layer, an attention mechanism, and a decoding part; The data expansion layer expands the radar data in the original data from two dimensions to three dimensions by duplication, so that the original data changes from the dimension of number of antennas × number of samples to the dimension of number of antennas × number of samples × number of samples; The feature extraction layer includes a multi-layer perceptron to extract high-dimensional features from the original radar data, and also includes a one-dimensional perceptron to extract information from the strongly correlated radar information reflected by the liquid and the reflective surface; In the attention mechanism, the high-dimensional feature part of the original radar data is divided into H×1×C and 1×W×C parts, which are respectively combined with the reflection of the reflecting surface to form a strong correlation radar information feature, and the multi-head attention mechanism operation is performed to perform dimension mapping, map to the required number of attention heads, and then broadcast the obtained results, expand the two results to the same dimension and then add them; The decoding part includes multiple complex fully connected layers to reduce the dimension of the result to one dimension, and then convert the result of the complex domain to the real domain by taking the modulus, and finally obtain the probability of the corresponding concentration gradient through the softmax function.

6. The method for detecting pipeline fluid concentration based on millimeter wave radar according to claim 5, characterized in that: The multi-layer perceptron model consists of several 2-dimensional 3*3 complex convolution modules, complex activation function modules, complex pooling modules and complex batch normalization modules.

7. The method for detecting pipeline fluid concentration based on millimeter wave radar according to claim 6, characterized in that: The complex activation function module adopts the CReLU function, and the complex batch normalization module performs a 0 mean 1 variance operation on the input: Where x is the input vector, E[x] is the expectation of the input batch data, and V is the correlation matrix, where the correlation matrix V is calculated as follows The final BN has two parameters γ and β, where γ is the scaling factor and β is the offset. Therefore, the final batch normalization result is as follows 8. The method for detecting pipeline fluid concentration based on millimeter wave radar according to claim 1, characterized in that: The cross entropy function is selected as the loss function in the training process, and its calculation method is: Where N is the number of samples, C is the number of categories obtained based on the minimum concentration resolution, and x y is the probability that the i-th sample belongs to the true category y, x c is the probability of the i-th sample data category c.

9. A pipeline fluid concentration detection device based on millimeter wave radar, comprising a memory and one or more processors, wherein the memory stores executable code, characterized in that: When the processor executes the executable code, a pipeline fluid concentration detection method based on millimeter wave radar as described in any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium having a program stored thereon, characterized in that: When the program is executed by the processor, a pipeline fluid concentration detection method based on millimeter wave radar as described in any one of claims 1 to 8 is implemented.

Images