Construction and application of the prediction model for the measurement data of fixed-point acoustic Doppler current meter

Abstract

This scheme provides the construction and application of a fixed-point acoustic Doppler flowmeter measurement data prediction model. The fixed-point acoustic Doppler flowmeter measurement data prediction model uses an improved dynamic weight convolutional neural network to extract spatial features in flow velocity measurement data, and uses a long short-term memory network to extract temporal features in flow velocity measurement data. It combines temporal features and spatial features to make accurate predictions of measurement data, and can judge abnormal data based on the predicted measurement data.

Description

Technical Field

[0001] The present application relates to the field of flow measurement, and in particular to the construction and application of a prediction model for measurement data of a fixed-point acoustic Doppler flow meter. Background Art

[0002] The horizontal acoustic Doppler current profiler (H-ADCP) is an instrument used to measure the flow velocity distribution in water bodies. It calculates the water velocity by emitting sound waves and receiving sound waves reflected from suspended particles (scatterers) in the water. However, in actual use, H-ADCP may encounter some challenges that lead to a decline in the quality of test data, including the fact that plants such as water grass in the water may interfere with the propagation path of the sound waves, causing changes in the echo signal intensity or abnormal reflection patterns, thereby affecting the accuracy of the flow velocity measurement; for example, when fish and other organisms swim in the water, they will produce additional scatterers, which will increase the noise level of the signal and make it difficult to extract the true flow velocity from the signal. In addition, the scatterers in the water may actually be aggregated or sparse due to factors such as water flow and sediment resuspension. This non-uniformity will cause the measured flow velocity value to deviate from the actual flow velocity, resulting in the presence of abnormal data or failure to measure the flow velocity value in the measurement data of the horizontal acoustic Doppler current profiler.

[0003] Currently, the common way to reduce test data anomalies is to install the H-ADCP in a location as far away from obstacles as possible to avoid interference from aquatic plants and other fixed obstacles, or to adjust the instrument's operating frequency according to the characteristics of the water body to obtain a better signal-to-noise ratio. However, these methods still cannot accurately determine whether there are abnormal data in the test data of the fixed-point (horizontal) acoustic Doppler current meter in a timely manner. Summary of the invention

[0004] The embodiment of the present application provides the construction and application of a fixed-point acoustic Doppler flow meter measurement data prediction model. A fixed-point acoustic Doppler flow meter measurement data prediction model is constructed by combining an improved convolutional network and a long short-term memory network, which can accurately predict the predicted measurement data of a fixed-point (horizontal) acoustic Doppler flow meter, and abnormal data is judged based on the predicted measurement data.

[0005] In a first aspect, an embodiment of the present application provides a method for constructing a fixed-point acoustic Doppler flow meter measurement data prediction model, comprising the following steps:

[0006] Acquire multiple groups of flow velocity measurement data sets of a fixed-point acoustic Doppler flow meter as training data, wherein each group of flow velocity measurement data includes normal flow velocity measurement data of multiple consecutive time windows, and each normal flow velocity measurement data includes multi-level layered flow velocity measurement data;

[0007] The training data is input into the constructed fixed-point acoustic Doppler flow meter measurement data prediction architecture for training until the training conditions are met to obtain a fixed-point acoustic Doppler flow meter measurement data prediction model, wherein the fixed-point acoustic Doppler flow meter measurement data prediction architecture includes an input layer, a feature filter layer, a dynamic weight convolutional neural network, a long short-term memory network and a prediction head connected in sequence, each group of flow velocity measurement data is input from the input layer to the feature filter layer to construct a multi-dimensional matrix vector, the multi-dimensional matrix vector is input into the dynamic weight convolutional neural network to dynamically adjust the weight of each layer of layered flow velocity measurement data and extract spatial features, the spatial features are input into the long short-term memory network to obtain time series features, and the time series features are input into the prediction head to predict and obtain predicted measurement data.

[0008] In the second aspect, an embodiment of the present application provides a method for predicting measurement data of a fixed-point acoustic Doppler flow meter, comprising the following steps: obtaining each group of flow velocity measurement data as the data to be measured, wherein each group includes normal flow velocity measurement data of multiple consecutive time windows, and each normal flow velocity measurement data includes multi-level stratified flow velocity measurement data; inputting each group of flow velocity measurement data into a fixed-point acoustic Doppler flow meter measurement data prediction model to output predicted measurement data.

[0009] In a third aspect, an embodiment of the present application provides a readable storage medium, in which a computer program is stored. The computer program includes a program code for controlling a process to execute a process, and the process includes a method for constructing a fixed-point acoustic Doppler flow meter measurement data prediction model.

[0010] The main contributions and innovations of the present invention are as follows:

[0011] The embodiment of the present application constructs a fixed-point acoustic Doppler flow meter measurement data prediction model that can analyze the time characteristics and spatial characteristics of the actual flow velocity measurement data of the fixed-point acoustic Doppler flow meter, and then predict the predicted measurement data at future moments based on the time series of the actual measurement data of the fixed-point acoustic Doppler flow meter. The fixed-point acoustic Doppler flow meter measurement data prediction model uses an improved dynamic weight convolutional neural network to achieve dynamic weight allocation of multi-level layered flow velocity measurement data, and then extract more accurate spatial features. The long short-term memory network in the fixed-point acoustic Doppler flow meter measurement data prediction model uses an improved activation function to reduce the impact of noise on model learning. In addition, this solution can use the predicted measurement data predicted by the fixed-point acoustic Doppler flow meter measurement data prediction model to judge abnormal data, thereby improving the efficiency and quality of flow velocity measurement in actual scenarios.

[0012] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0014] Figure 1 is a schematic diagram of the architecture of a dynamic weight convolutional neural network according to an embodiment of the present application;

[0015] Figure 2 is an image schematic diagram of an improved activation function according to an embodiment of the present application;

[0016] Figure 3 is a schematic diagram of the architecture of a fixed-point acoustic Doppler current meter measurement data prediction model according to an embodiment of the present application;

[0017] Figure 4 is a schematic diagram of actual flow rate measurement data and predicted measurement data of an embodiment;

[0018] Figure 5 is a schematic diagram of actual flow rate measurement data and predicted measurement data and a supplementary control value according to another embodiment;

[0019] Figure 6 It is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0020] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with one or more embodiments of this specification. Instead, they are merely examples of devices and methods consistent with some aspects of one or more embodiments of this specification as detailed in the appended claims.

[0021] It should be noted that: in other embodiments, the steps of the corresponding method are not necessarily performed in the order shown and described in this specification. In some other embodiments, the steps included in the method may be more or less than those described in this specification. In addition, a single step described in this specification may be decomposed into multiple steps for description in other embodiments; and multiple steps described in this specification may be combined into a single step for description in other embodiments.

[0022] Embodiment 1

[0023] The fixed-point (horizontal) acoustic Doppler current meter obtains velocity measurement data of different layers by emitting sound waves and receiving reflected signals at different depths. The velocity measurement data of each layer is a one-dimensional time series. When the velocity measurement data of multiple layers are superimposed according to the layer number, a two-dimensional time series is formed. In addition, due to the measurement principle of acoustic Doppler velocity, there is a correlation between the values ​​of the velocity measurement data of adjacent layers. This mutual connection exists between the velocity data of adjacent layers at the same time, which can be called the spatial characteristics of the velocity measurement data. At the same time, for a fixed layer, the change of its velocity measurement data in the time dimension also has certain characteristics, including the frequency of change, the amplitude of change, etc., which can be called the time characteristics of the velocity measurement data.

[0024] In view of the temporal and spatial characteristics of the flow velocity measurement data of the fixed-point (horizontal) acoustic Doppler flow meter, this scheme designs a fixed-point acoustic Doppler flow meter measurement data prediction model specifically for the fixed-point (horizontal) acoustic Doppler flow meter, which can effectively and accurately predict the normal flow velocity measurement data. Once there is a large error between the actual measured flow velocity measurement data and the predicted flow velocity measurement data, the actual measured flow velocity measurement data is judged to be abnormal data.

[0025] Specifically, this solution provides a method for constructing a fixed-point acoustic Doppler current meter measurement data prediction model, comprising the following steps:

[0026] Acquire multiple groups of flow velocity measurement data sets of a fixed-point acoustic Doppler flow meter as training data, wherein each group of flow velocity measurement data includes normal flow velocity measurement data of multiple consecutive time windows, and each normal flow velocity measurement data includes multi-level layered flow velocity measurement data;

[0027] The training data is input into the constructed fixed-point acoustic Doppler flow meter measurement data prediction architecture for training until the training conditions are met to obtain a fixed-point acoustic Doppler flow meter measurement data prediction model, wherein the fixed-point acoustic Doppler flow meter measurement data prediction architecture includes an input layer, a feature filter layer, a dynamic weight convolutional neural network, a long short-term memory network and a prediction head connected in sequence, each group of flow velocity measurement data is input from the input layer to the feature filter layer to construct a multi-dimensional matrix vector, the multi-dimensional matrix vector is input into the dynamic weight convolutional neural network to dynamically adjust the weight of each layer of layered flow velocity measurement data and extract spatial features, the spatial features are input into the long short-term memory network to obtain time series features, and the time series features are input into the prediction head to predict and obtain predicted measurement data.

[0028] The feature filtering layer in the fixed-point acoustic Doppler flowmeter measurement data prediction architecture constructed in this scheme can be used to reduce the impact of environmental noise on the current fixed-point acoustic Doppler flowmeter measurement data prediction architecture training. The dynamic weight convolutional neural network can be used to attach different dynamic weights to layered flow velocity measurement data at different levels, and then extract spatial features that can fully reflect the spatial distribution of normal flow velocity measurement data. The long short-term memory network can be used to extract temporal features from spatial features and reduce the impact of noise on training.

[0029] This solution uses multiple groups of flow velocity measurement data as training samples. Since each group of flow velocity measurement data includes normal flow velocity measurement data of multiple continuous time windows, in order to avoid the influence of the sampling process on the flow velocity measurement data, this solution ensures that the sampling frequency of the normal flow velocity measurement data in each group of flow velocity measurement data is the same, and the layered flow velocity measurement data in the normal flow velocity measurement data are arranged in time sequence, and the layered flow velocity measurement data in each normal flow velocity measurement data are aligned by layer number. In some embodiments, the training sample includes 5000 groups of flow velocity measurement data.

[0030] Specifically, a filter matrix with the same dimension as that of each group of flow velocity measurement data is set in the feature filter layer of the present scheme, wherein each group of flow velocity measurement data includes i*j layered flow velocity measurement data. Correspondingly, the filter matrix is ​​an i*j two-dimensional matrix, wherein i represents the number of time windows in each group of flow velocity measurement data, and j represents the number of levels of layered flow velocity measurement data in each normal flow velocity measurement data.

[0031] In an embodiment of the present scheme, each normal flow velocity measurement data includes 30 stratified flow velocity measurement data corresponding to layers 1-30 respectively, and each group of flow velocity measurement data includes normal flow velocity measurement data of 10 continuous time windows. At this time, each group of flow velocity measurement data includes 10*30 stratified flow velocity measurement data, and correspondingly, the filter matrix is ​​a two-dimensional matrix of 10*30.

[0032] Specifically, each group of flow velocity measurement data is input from the input layer to the feature filter layer. The input vector input to the feature filter layer is multiplied by the filter matrix to obtain a multi-dimensional matrix vector. This scheme performs sliding window fitting on the input vector through the filter matrix, which can filter out sharp incident noise such as burrs. The calculation formula is as follows;

[0033] ;

[0034]

[0035] Where A represents the filter matrix, is the input vector, is the output multidimensional matrix vector, is the parameter to be determined, which is obtained by network learning of the prediction architecture of fixed-point acoustic Doppler flow meter measurement data, i represents the number of time windows in each group of flow velocity measurement data, j represents the number of layers of layered flow velocity measurement data in each normal flow velocity measurement data, where M, O and Indicates an ellipsis.

[0036] like Figure 1 As shown, the dynamic weight convolutional neural network of the present scheme includes parallel dynamic weight branches and ordinary feature branches, wherein the ordinary feature branch includes a first convolutional layer, and the dynamic weight branch includes a second convolutional layer, an average pooling layer, a fully connected layer, and a dynamic weight allocation layer connected in sequence.

[0037] Specifically, the multi-dimensional matrix vector is input into the dynamic weight convolutional neural network and extracted through the convolution of the ordinary feature branch to obtain the ordinary feature vector. The multi-dimensional matrix vector is input into the dynamic weight convolutional neural network and processed through the dynamic weight branch to obtain the dynamic weights of the stratified flow velocity measurement data at different levels. The dynamic weights of the stratified flow velocity measurement data at different levels are used to adjust the ordinary feature vector to obtain the spatial feature, wherein the spatial feature is represented in the form of a feature map. It should be noted that the dynamic weight convolutional neural network of this scheme is trained by training the input stratified flow velocity measurement data so that the dynamic weights of the stratified flow velocity measurement data at different levels can be obtained through the dynamic weight branch processing during actual prediction. Correspondingly, the multi-dimensional matrix vector is sequentially passed through the second convolutional layer, the average pooling layer, and the fully connected layer to obtain the dynamic weights of the stratified flow velocity measurement data at different levels.

[0038] As mentioned above, for a specific fixed-point (horizontal) acoustic Doppler flow meter, the stratified flow velocity that is prone to mismeasurement or omission is specific. Therefore, the dynamic weight convolutional neural network of this scheme uses a convolution layer + global pooling layer to extract features to attach different weights to different levels of stratified flow velocity measurement data.

[0039] The LSTM network of this scheme sets the number of network layers to 5, inputs the spatial features into the LSTM network to extract the temporal features, and correspondingly, expands the feature graph corresponding to the spatial features and inputs it into the LSTM network to extract the temporal features. In addition, in order to further avoid the influence of noise on the entire fixed-point acoustic Doppler current meter measurement data prediction model, the activation function in the LSTM network is improved, and the improved activation function is as follows:

[0040] ;

[0041] in is the parameter of the input spatial feature, To satisfy the standard normal distribution of random numbers, is the standard sigmoid activation function.

[0042] The schematic diagram of the improved activation function is as follows Figure 2 As shown, it can be seen that the improved activation function of this scheme adds standard normally distributed random numbers in the saturation region, so the gradient will not disappear in the saturation region, so that the model will not fall into the local optimum due to the presence of noise during training calculation, which is more conducive to quickly converge to the global optimal point. In addition, the activation function proposed in this scheme performs a weighted summation of the classic tanh activation function and the sigmoid activation function, which is more conducive to integrating the advantages of the two and improving the training speed and final accuracy.

[0043] The prediction head of this scheme adopts a fully connected layer, and the time series features output by the long short-term memory network are input into the prediction head to predict the predicted normal flow velocity measurement data. The predicted measurement data predicted by this scheme are multi-level hierarchical flow velocity measurement data within the same time window.

[0044] The overall architecture diagram of the fixed-point acoustic Doppler flow meter measurement data prediction architecture of this scheme is as follows Figure 3 As shown, it can be seen that the fixed-point acoustic Doppler flow meter measurement data prediction architecture of this scheme includes an input layer, a feature filter layer, a dynamic weight convolutional neural network, a long short-term memory network and a prediction head connected in sequence, and MSE is used as the loss function to train the fixed-point acoustic Doppler flow meter measurement data prediction architecture and save the weights.

[0045] Embodiment 2

[0046] Based on the same concept, this solution provides an application method of a fixed-point acoustic Doppler flow meter measurement data prediction model. In other words, a fixed-point acoustic Doppler flow meter measurement data prediction method includes the following steps:

[0047] Acquire each group of flow velocity measurement data as the data to be measured, wherein each group includes normal flow velocity measurement data of a plurality of consecutive time windows, and each normal flow velocity measurement data includes multi-level layered flow velocity measurement data;

[0048] Each set of flow velocity measurement data is input into the fixed-point acoustic Doppler flow meter measurement data prediction model constructed in Example 1 to output predicted measurement data.

[0049] In addition, in some embodiments, when the threshold between the predicted measurement data and the actual flow rate measurement data actually measured in the next time window is greater than the abnormality detection threshold, it means that there is a large deviation in the actual flow rate measurement data at this time, and the predicted measurement data can be used to replace the actual flow rate measurement data of the next time window.

[0050] In other embodiments, if there is no actual flow rate measurement data in the next time window, the predicted measurement data may be used to fill the measurement gap.

[0051] like Figure 4 As shown, Figure 4 FIG. 1 is a schematic diagram of actual flow velocity measurement data and predicted measurement data of an embodiment. Taking the abnormal detection threshold value as 1 as an example, in the actual application of a fixed-point acoustic Doppler flow meter, the actual flow velocity measurement value of a set of flow velocity measurement data is as follows: Figure 4 As shown by the blue curve in ; the flow velocity measurement data is input into the fixed-point acoustic Doppler flow meter measurement data prediction model constructed by this scheme to output the corresponding predicted flow velocity measurement value as shown in Figure 4 As shown by the red curve in the figure, it can be seen that there is an abnormal flow velocity measurement value, which can be successfully detected using this method, as shown by the red circle.

[0052] Similarly, if Figure 5 As shown, Figure 5 Schematic diagram of actual flow velocity measurement data and predicted measurement data and a supplementary control value of another embodiment. In the actual application of a fixed-point acoustic Doppler flow meter, the actual flow velocity measurement value of another set of flow velocity measurement data is taken as follows: Figure 5 It can be seen that when k=25, the measurement loss phenomenon occurs. The flow velocity measurement data is input into the fixed-point acoustic Doppler flow meter measurement data prediction model constructed by this scheme to output the corresponding predicted flow velocity measurement value as follows Figure 5 As shown in the red curve in , the predicted measurement data when k=25 output by this scheme can be used to fill in the missing flow velocity measurement values. The completed flow velocity curve is as follows Figure 5 Shown as the green curve in the middle.

[0053] Embodiment 3

[0054] This embodiment also provides an electronic device, referring to Figure 6 , comprises a memory 404 and a processor 402, wherein the memory 404 stores a computer program, and the processor 402 is configured to run the computer program to execute the steps in any of the above method embodiments.

[0055] Specifically, the processor 402 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0056] Among them, the memory 404 may include a large capacity memory 404 for data or instructions. For example, but not limitation, the memory 404 may include a hard disk drive (HDD), a floppy disk drive, a solid state drive (SSD), a flash memory, an optical disk, a magneto-optical disk, a tape, or a universal serial bus (USB) drive, or a combination of two or more of these. In appropriate cases, the memory 404 may include a removable or non-removable (or fixed) medium. In appropriate cases, the memory 404 may be inside or outside the data processing device. In a specific embodiment, the memory 404 is a non-volatile memory. In a specific embodiment, the memory 404 includes a read-only memory (ROM) and a random access memory (RAM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (Programmable Read-Only Memory, PROM for short), an erasable PROM (Erasable Programmable Read-Only Memory, EPROM for short), an electrically erasable PROM (Electrically Erasable Programmable Read-Only Memory, EEPROM for short), an electrically alterable ROM (Electrically Alterable Read-Only Memory, EAROM for short) or a flash memory (FLASH) or a combination of two or more of these. In appropriate circumstances, the RAM may be a static random access memory (SRAM) or a dynamic random access memory (DRAM), wherein the DRAM may be a fast page mode dynamic random access memory 404 (FPMDRAM), an extended data output dynamic random access memory (EDODRAM), a synchronous dynamic random access memory (SDRAM), etc.

[0057] The memory 404 may be used to store or cache various data files required for processing and / or communication, as well as possible computer program instructions executed by the processor 402 .

[0058] The processor 402 reads and executes the computer program instructions stored in the memory 404 to implement any one of the methods for constructing a fixed-point acoustic Doppler current meter measurement data prediction model in the above embodiments.

[0059] Optionally, the electronic device may further include a transmission device 406 and an input / output device 408 , wherein the transmission device 406 is connected to the processor 402 , and the input / output device 408 is connected to the processor 402 .

[0060] The transmission device 406 can be used to receive or send data via a network. Specific examples of the above-mentioned network may include a wired or wireless network provided by a communication provider of the electronic device. In one example, the transmission device includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 406 can be a radio frequency (Radio Frequency, referred to as RF) module, which is used to communicate with the Internet wirelessly.

[0061] The input / output device 408 is used to input or output information. In this embodiment, the input information may be actual flow rate measurement data, etc., and the output information may be predicted measurement data, etc.

[0062] Optionally, in this embodiment, the processor 402 may be configured to perform the following steps through a computer program:

[0063] Acquire multiple groups of flow velocity measurement data sets of a fixed-point acoustic Doppler flow meter as training data, wherein each group of flow velocity measurement data includes normal flow velocity measurement data of multiple consecutive time windows, and each normal flow velocity measurement data includes multi-level layered flow velocity measurement data;

[0064] The training data is input into the constructed fixed-point acoustic Doppler flow meter measurement data prediction architecture for training until the training conditions are met to obtain a fixed-point acoustic Doppler flow meter measurement data prediction model, wherein the fixed-point acoustic Doppler flow meter measurement data prediction architecture includes an input layer, a feature filter layer, a dynamic weight convolutional neural network, a long short-term memory network and a prediction head connected in sequence, each group of flow velocity measurement data is input from the input layer to the feature filter layer to construct a multi-dimensional matrix vector, the multi-dimensional matrix vector is input into the dynamic weight convolutional neural network to dynamically adjust the weight of each layer of layered flow velocity measurement data and extract spatial features, the spatial features are input into the long short-term memory network to obtain time series features, and the time series features are input into the prediction head to predict and obtain predicted measurement data.

[0065] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0066] In general, various embodiments may be implemented in hardware or dedicated circuits, software, logic, or any combination thereof. Some aspects of the invention may be implemented in hardware, while other aspects may be implemented in firmware or software that may be executed by a controller, microprocessor, or other computing device, but the invention is not limited thereto. Although various aspects of the invention may be shown and described as block diagrams, flow charts, or using some other graphical representation, it should be understood that, as non-limiting examples, the boxes, devices, systems, techniques, or methods described herein may be implemented in hardware, software, firmware, dedicated circuits or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0067] Embodiments of the present invention can be implemented by computer software, which is executable by a data processor of a mobile device, such as in a processor entity, or implemented by hardware, or implemented by a combination of software and hardware. Computer software or programs (also referred to as program products) including software routines, applets and / or macros can be stored in any device readable data storage medium, and they include program instructions for performing specific tasks. Computer program products can include one or more computer executable components configured to perform embodiments when the program is running. One or more computer executable components can be at least one software code or a part thereof. In addition, at this point, it should be noted that any box of the logic flow in the figure can represent a program step, or interconnected logic circuits, boxes and functions, or a combination of program steps and logic circuits, boxes and functions. Software can be stored in physical media such as memory chips or storage blocks implemented in processors, magnetic media such as hard disks or floppy disks, and optical media such as, for example, DVDs and data variants thereof, CDs. Physical media are non-transient media.

[0068] Those skilled in the art should understand that the technical features of the above embodiments may be arbitrarily combined. 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.

[0069] The above 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 present application. It should be pointed out that, for a person of ordinary skill in the art, several variations 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 method for constructing a fixed-point acoustic Doppler flow meter measurement data prediction model, characterized in that: The following steps are involved: Acquire multiple groups of flow velocity measurement data sets of a fixed-point acoustic Doppler flow meter as training data, wherein each group of flow velocity measurement data includes normal flow velocity measurement data of multiple consecutive time windows, and each normal flow velocity measurement data includes multi-level layered flow velocity measurement data; The training data is input into the constructed fixed-point acoustic Doppler flowmeter measurement data prediction architecture for training until the training conditions are met to obtain a fixed-point acoustic Doppler flowmeter measurement data prediction model, wherein the fixed-point acoustic Doppler flowmeter measurement data prediction architecture includes an input layer, a feature filter layer, a dynamic weight convolutional neural network, a long short-term memory network and a prediction head connected in sequence, the dynamic weight convolutional neural network includes parallel dynamic weight branches and ordinary feature branches, wherein the ordinary feature branch includes a first convolutional layer, and the dynamic weight branch includes a second convolutional layer, an average pooling layer, a fully connected layer and a dynamic weight allocation layer connected in sequence, each group of flow velocity measurement data is input from the input layer to the feature filter layer to construct a multidimensional matrix vector, the multidimensional matrix vector is input into the dynamic weight convolutional neural network to dynamically adjust the weight of each layer of the layered flow velocity measurement data and extract the spatial features, the spatial features are input into the long short-term memory network to obtain the time series features, and the time series features are input into the prediction head to predict the predicted measurement data.

2. The method for constructing a fixed-point acoustic Doppler flow meter measurement data prediction model according to claim 1, characterized in that: In the feature filter layer, a filter matrix having the same dimension as each group of flow velocity measurement data is set.

3. The method for constructing a fixed-point acoustic Doppler flow meter measurement data prediction model according to claim 2, characterized in that: Each group of velocity measurement data is input from the input layer to the feature filter layer. The input vector input to the feature filter layer is multiplied by the filter matrix to obtain a multi-dimensional matrix vector. The calculation formula is as follows: ； ； Where A represents the filter matrix, is the input vector, is the output multidimensional matrix vector, is the parameter to be determined, which is obtained by network learning of the fixed-point acoustic Doppler flow meter measurement data prediction architecture, i represents the number of time windows in each group of flow velocity measurement data, and j represents the number of levels of stratified flow velocity measurement data in each normal flow velocity measurement data.

4. The method for constructing a fixed-point acoustic Doppler flow meter measurement data prediction model according to claim 1, characterized in that: The multidimensional matrix vector is input into the dynamic weight convolutional neural network, and the ordinary feature vector is obtained through convolution extraction of the ordinary feature branch. The multidimensional matrix vector is input into the dynamic weight convolutional neural network, and the dynamic weights of different levels are obtained through processing by the dynamic weight branch. The dynamic weights of different levels of stratified flow velocity measurement data are used to adjust the ordinary feature vector to obtain the spatial feature.

5. The method for constructing a fixed-point acoustic Doppler flow meter measurement data prediction model according to claim 1, characterized in that: The activation function in the long short-term memory network is improved. The improved activation function is as follows: ； in For input elements, To satisfy the standard normal distribution of random numbers, is the standard sigmoid activation function.

6. The method for constructing a fixed-point acoustic Doppler flow meter measurement data prediction model according to claim 1, characterized in that: The sampling frequencies of the normal flow velocity measurement data in each group of flow velocity measurement data are the same, and the layered flow velocity measurement data in the normal flow velocity measurement data are arranged in time sequence, and the layered flow velocity measurement data are aligned by layer numbers.

7. A method for predicting measurement data of a fixed-point acoustic Doppler flow meter, characterized in that: The method comprises the following steps: obtaining each group of flow velocity measurement data as the data to be measured, wherein each group comprises normal flow velocity measurement data of a plurality of consecutive time windows, and each normal flow velocity measurement data comprises multi-level layered flow velocity measurement data; Each set of flow velocity measurement data is input into a fixed-point acoustic Doppler flowmeter measurement data prediction model constructed according to the method for constructing a fixed-point acoustic Doppler flowmeter measurement data prediction model according to any one of claims 1 to 6 to output predicted measurement data.

8. The method for predicting measurement data of a fixed-point acoustic Doppler flow meter according to claim 7, characterized in that: The method comprises the following steps: if the threshold between the predicted measurement data and the actual flow velocity measurement data actually measured in the next time window is greater than the abnormality detection threshold, the actual flow velocity measurement data of the next time window is replaced by the predicted measurement data; if there is no actual flow velocity measurement data in the next time window, the actual flow velocity measurement data is filled with the predicted measurement data.

9. A readable storage medium, characterized in that: The readable storage medium stores a computer program, which includes a program code for controlling a process to execute a process, wherein the process includes a method for constructing a fixed-point acoustic Doppler current meter measurement data prediction model according to any one of claims 1 to 6.

Images