Non-stationary converter method and device for river basin flood forecasting under influence of reservoir regulation

The non-stationary converter deep learning model addresses the over-stabilization issue in data-driven hydrological models by characterizing and adapting to both stationary and non-stationary flood sequence characteristics, improving flood forecasting accuracy and adaptability in changing environments.

CN119090137APending Publication Date: 2024-12-06WUHAN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411183631.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-27
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

Data-driven hydrological models often reduce the non-stationarity of flood sequences by smoothing data, leading to over-stabilization and reduced adaptability and generalization ability in changing environments, which is contrary to the significance of basin flood forecasting under reservoir regulation.

Method used

A non-stationary converter deep learning model is constructed using a normalization-inverse normalization module and an inverse stationary attention mechanism to characterize and capture the stationary and non-stationary characteristics of flood sequences, combined with double regularization and the Adam optimization algorithm to calibrate model parameters, and an encoder and decoder to adapt to changing environments.

Benefits of technology

The model improves the accuracy and adaptability of flood forecasting, extending the forecast period and enhancing the generalization ability of deep learning models in changing environments by utilizing both stationary and non-stationary flood sequence characteristics.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119090137A_ABST
    Figure CN119090137A_ABST
Patent Text Reader

Abstract

The invention discloses a non-stationary converter method and device for watershed flood forecasting under the influence of reservoir regulation, and the method comprises the steps: considering the runoff and confluence duration of a watershed, constructing a non-stationary converter deep learning model, matching a watershed flood forecasting architecture under the influence of reservoir regulation, representing the stationary features of a flood sequence through a standardization-inverse standardization module, and carrying out the deep learning of the non-stationary converter. Capturing non-stationary features of the flood sequence under the influence of reservoir regulation by adopting an inverse stationary attention mechanism module; double regularization of an objective function is considered, and parameters of the deep learning model of the non-stationary converter are calibrated by adopting a sub-when optimization algorithm; and an interpretability method is applied to analyze the contribution degree of the stationary features and the non-stationary features to the basin flood forecasting precision, and the adaptability and generalization ability of the deep learning model in a changing environment are improved. According to the invention, the accuracy of basin flood forecasting is improved, and the flood forecasting period is prolonged.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of basin hydrological models, and in particular to a non-stationary converter method and device for basin flood forecasting under the influence of reservoir regulation. Background Art

[0002] The data-driven hydrological model treats the studied basin or region as a dynamic system, using input (generally upstream rainfall and upstream main and tributary water) and output (generally basin or regional control section flow) data, such as artificial neural network to establish a mathematical relationship between input and output, and then infer new output from new input. This type of model has been widely used in the fields of water conservancy project planning and design, flood forecasting, water resources development and utilization, and provides theoretical support for solving various engineering hydrological problems and improving people's understanding of hydrological laws.

[0003] With the development of machine learning, data-driven hydrological models have made great progress. However, such models often reduce the non-stationarity of the original sequence by smoothing the data. This practice is contrary to the significance of basin flood forecasting for emergency forecasting under the influence of a changing environment. It ignores the universal existence of non-stationary flood sequences under a changing environment, and ultimately leads to over-stabilization of modeling and forecasting, which reduces the adaptability of such models to a changing environment and the generalization ability of the model. Summary of the invention

[0004] In order to overcome the deficiencies of the above-mentioned prior art, the present invention provides a non-stationary converter method and device for basin flood forecasting under the influence of reservoir regulation, which can improve the accuracy of basin flood forecasting and extend the flood forecast period, and enhance the adaptability of hydrological models to changing environments and the generalization ability of models.

[0005] According to one aspect of the present invention, there is provided a non-stationary converter method for flood forecasting in a river basin under the influence of reservoir regulation, comprising:

[0006] Obtain source data to be forecasted;

[0007] Input the source data to be predicted into the trained non-stationary converter deep learning model, and output the predicted flow; wherein the training of the non-stationary converter deep learning model includes:

[0008] Considering the duration of runoff generation and convergence in the basin, a non-stationary converter deep learning model is constructed. The non-stationary converter deep learning model matches the basin flood forecasting framework under the influence of reservoir regulation, and uses a normalization-inverse normalization module to characterize the stationary characteristics of the flood sequence, and uses an inverse stationary attention mechanism module to capture the non-stationary characteristics of the flood sequence under the influence of reservoir regulation;

[0009] According to the non-stationary converter deep learning model, the double regularization of the objective function is considered, the Adam optimization algorithm is used to calibrate the model parameters, and the trained non-stationary converter deep learning model is output.

[0010] As a further technical solution, the method also includes: applying interpretable methods to analyze the contribution of stationary features and non-stationary features to the accuracy of basin flood forecasting, and improving the adaptability and generalization ability of deep learning models under changing environments.

[0011] As a further technical solution, the basin flood forecasting architecture under the influence of reservoir regulation that matches the non-stationary converter deep learning model includes an encoder and a decoder, wherein the encoder aims to add the non-stationary features of the flood sequence to the attention mechanism, and the decoder aims to bring the non-stationary features caused by reservoir regulation into the calculation after stabilization through a classification function, thereby adapting the non-stationary converter deep learning model.

[0012] According to one aspect of the present invention, there is provided a non-stationary converter method for flood forecasting in a river basin under the influence of reservoir regulation, comprising:

[0013] The first main module is used to obtain source data to be forecasted;

[0014] The second main module is used to input the source data to be predicted into the trained non-stationary converter deep learning model and output the predicted flow rate; wherein the training of the non-stationary converter deep learning model includes:

[0015] Considering the duration of runoff generation and convergence in the basin, a non-stationary converter deep learning model is constructed. The non-stationary converter deep learning model matches the basin flood forecasting framework under the influence of reservoir regulation, and uses a normalization-inverse normalization module to characterize the stationary characteristics of the flood sequence, and uses an inverse stationary attention mechanism module to capture the non-stationary characteristics of the flood sequence under the influence of reservoir regulation;

[0016] According to the non-stationary converter deep learning model, the double regularization of the objective function is considered, the Adam optimization algorithm is used to calibrate the model parameters, and the trained non-stationary converter deep learning model is output.

[0017] As a further technical solution, it also includes: a third main module, which is used to apply interpretable methods to analyze the contribution of stationary features and non-stationary features to the accuracy of basin flood forecasting, and to improve the adaptability and generalization ability of deep learning models under changing environments.

[0018] As a further technical solution, the basin flood forecasting architecture under the influence of reservoir regulation that matches the non-stationary converter deep learning model includes an encoder and a decoder, wherein the encoder aims to add the non-stationary characteristics of the flood sequence to the attention mechanism, and the decoder aims to bring the non-stationary characteristics such as mutations, trends, and cycles caused by reservoir regulation into the calculation after stabilization through a classification function, thereby adapting the non-stationary converter deep learning model.

[0019] According to one aspect of the present invention, there is provided an execution device, comprising a processor and a memory, wherein the processor is coupled to the memory; the memory is used to store a program; and the processor is used to execute the program in the memory, so that the execution device executes the described method.

[0020] According to one aspect of the present specification, a non-transitory computer-readable storage medium is provided, wherein the non-transitory computer-readable storage medium stores computer instructions, wherein the computer instructions enable the computer to execute the method described above.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] The present invention constructs a non-stationary converter model for basin flood forecasting under the influence of reservoir regulation, which can not only adapt to the basin flood forecasting framework with different non-stationary characteristics, but also make full use of the stationary and non-stationary characteristics of the basin flood series under the changing environment; not only combines the objective function double regularization and Adam optimization algorithm to calibrate the model parameters, but also adopts the adjoint model based on Shapley value to analyze the contribution of stationary characteristics and non-stationary characteristics to the accuracy of basin flood forecasting, overcomes the technical bottleneck of over-stationary learning of machine learning models, and improves the adaptability and generalization ability of deep learning models under changing environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, a brief introduction is given below to the drawings used in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0024] Figure 1 A flow chart of a non-stationary converter method for basin flood forecasting under the influence of reservoir regulation provided by an exemplary embodiment is shown.

[0025] Figure 2 A schematic diagram of the training process of a non-stationary converter model for basin flood forecasting under the influence of reservoir regulation provided by an exemplary embodiment is shown.

[0026] Figure 3A schematic diagram of a post-interpretation flow of a non-stationary converter model provided by an exemplary embodiment is shown.

[0027] Figure 4 A logical schematic diagram of a non-stationary converter method for basin flood forecasting under reservoir regulation provided by an exemplary embodiment is shown.

[0028] Figure 5 A logical schematic diagram of an inverse stationary attention mechanism in a non-stationary converter model provided by an exemplary embodiment is shown.

[0029] Figure 6 An architectural diagram of a non-stationary converter model for basin flood forecasting under reservoir regulation provided by an exemplary embodiment is shown.

[0030] Figure 7 A block diagram of a non-stationary converter device for basin flood forecasting under the influence of reservoir regulation provided by an exemplary embodiment is shown.

[0031] Figure 8 A block diagram of a non-stationary converter model for basin flood forecasting under the influence of reservoir regulation provided by an exemplary embodiment is shown.

[0032] Fig. 9 A block diagram of an execution device provided by an exemplary embodiment is shown. DETAILED DESCRIPTION

[0033] The terms "including" and "having" and any variations thereof in the specification and claims of the present invention and the above-mentioned drawings are intended to cover non-exclusive inclusions, for example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to the steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products or apparatuses.

[0034] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices. The flowcharts shown in the accompanying drawings are merely exemplary and do not necessarily include all the contents and operations / steps, nor do they necessarily have to be executed in the order described. For example, some operations / steps may be further decomposed, while some operations / steps may be combined or partially combined, so the actual execution order may change according to the actual situation.

[0035] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. In addition, the technical features in the various embodiments or single embodiments provided by the present invention are arbitrarily combined with each other to form a new technical solution. This combination is not restricted by the sequence of steps and / or the structural composition mode, but must be based on the ability of ordinary technicians in this field to achieve. When the combination of technical solutions is contradictory or cannot be achieved, it should be considered that this combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0036] Please refer to Figure 1 ,in Figure 1 A flow chart showing a non-stationary converter method for basin flood forecasting under reservoir regulation provided by an exemplary embodiment, the method comprising:

[0037] Step 1: Obtain the source data to be forecasted, including numerically predicted rainfall, measured rainfall, and measured flow at hydrological stations.

[0038] Step 2: Input the source data to be predicted into the trained non-stationary converter deep learning model and output the predicted flow.

[0039] Among them, the training of the non-stationary converter deep learning model includes: considering the duration of basin runoff generation and convergence, constructing a non-stationary converter deep learning model, the non-stationary converter deep learning model matches the basin flood forecasting framework under the influence of reservoir regulation, and uses a normalization-inverse normalization module to characterize the stationary characteristics of the flood sequence, and uses an inverse stationary attention mechanism module to capture the non-stationary characteristics of the flood sequence under the influence of reservoir regulation; considering the double regularization of the objective function, using the Adam optimization algorithm to calibrate the model parameters, and outputting the trained non-stationary converter deep learning model.

[0040] Please refer to Figure 2 , Figure 2 A schematic diagram of the training process of a non-stationary converter model for basin flood forecasting under the influence of reservoir regulation provided by an embodiment is shown.

[0041] The training process is implemented through steps 2.1 to 2.2.

[0042] Please refer to Figure 3 , Figure 3A schematic diagram of a post-interpretation process of a non-stationary converter model provided by an exemplary embodiment is shown. After the model training is completed, the contribution is analyzed through step 2.3 to improve the adaptability and generalization ability of the deep learning model in a changing environment.

[0043] Please refer to Figure 4 , Figure 4 The following is a logical principle diagram of a non-stationary converter method for flood forecasting in a river basin under the influence of reservoir regulation provided by an embodiment. The source data for forecasting is calculated through the forecasting logic to output the final forecast flow.

[0044] In step 2.1, the standardization-denormalization module is used to characterize the stationary information of the flood series. For each input series, the mean and variance of the input series are used to convert it into a Gaussian distribution with a mean of 0 and a variance of 1, thereby eliminating the differences in the statistics of time series in different time windows. The calculation formula is

[0045]

[0046]

[0047]

[0048] Where: x t is the observation value of the input sequence at time t; T is the number of samples in the input sequence; μ x is the average value of the input sequence; and σ x are the variance and standard deviation of the input sequence respectively; x' t is the normalized input sequence.

[0049]

[0050] Where: y' t is the model output obtained using the standardized input sequence; is the forecast obtained by inverse normalization using the mean and standard deviation.

[0051] The calculation formula for capturing the non-stationary information of flood sequences under the influence of reservoir regulation using the inverse stationary attention mechanism module is as follows.

[0052] Calculation formula of self-attention module in stationary converter

[0053]

[0054] Where: z is the output of the stationary converter; G, K and V are three sets of characteristics of the stationary sequence; k is the number of model input factors.

[0055] Self-attention calculation formula in non-stationary transformer

[0056]

[0057] Where: z' is the output of the non-stationary converter; G', K' and V' are three sets of characteristics of the non-stationary sequence (such as trend, variation, and period); τ and Δ are two parameters of the non-stationary converter.

[0058] The computational steps to add non-stationary features to the self-attention module are:

[0059] ①Assume that G is obtained by linear mapping of the input sequence;

[0060] ② Assuming that each batch of input sequences has the same expectation and variance, the standardized sequence features

[0061]

[0062]

[0063] Where: μ G and μ K are the expected values ​​of features G and K respectively.

[0064] ③Calculate the product of two sets of features

[0065]

[0066] 4④ Convert the self-attention module in the smooth converter to

[0067]

[0068] 5⑤ Considering the second assumption that each batch of input sequences has the same expectation and variance, the last two terms of the formula in step ④ are constants and can be deleted, resulting in a simplified formula

[0069]

[0070] ⑥ Three variables σ related to non-stationary characteristics x , K, μ G , where σ x 2 Denoted as τ (τ ≥ 0), Kμ G Denoted as Δ, a multi-layer perceptron is used to introduce non-stationary features into the stationary self-attention module, and we can get

[0071] log(τ)=MLP(σ x ,x t )

[0072] Δ=MLP(μ x ,x t )

[0073] Where: MLP(·,·) is a multi-layer perceptron.

[0074] In step 2.2, the non-stationary converter model parameter calibration is performed by the following formula: The double regularization formula of the objective function is

[0075]

[0076]

[0077]

[0078] Where: L 1 and L 2 The first and second regularized objective functions are adopted respectively; MSE is the mean square error; Q(t) is the measured flow of the hydrological station; is the predicted flow of the hydrological station; ρ and μ are L 1 and L 2 Regularization coefficient; ω i is the i-th weight parameter of the hidden layer; M is the number of neurons in the hidden layer.

[0079] L 1 Regularization is more robust to outliers and is beneficial for non-stationary feature extraction, while L 2 Regularization is more sensitive to outliers, which is beneficial to parameter optimization and easier to converge to the global optimal solution. In order to balance the efficiency of feature extraction and parameter optimization, a double regularization objective function is used to calibrate the model parameters. Considering that the use of a double regularization objective function to calibrate the model parameters is prone to the curse of dimensionality, the Adam optimization algorithm is used to learn in batches (equal to the number of batches in step 1) to calibrate the parameters of the non-stationary transformer deep learning model to improve the efficiency of parameter optimization.

[0080] In step 2.3, the contribution of stationary and non-stationary information to the accuracy of flood forecasting in the basin is performed by the following model, using Shapley values ​​to describe the adjoint model of the interpretation

[0081]

[0082] Where: f j ∈{0,1} indicates whether the jth input factor can be observed, 0 means it cannot be observed, and 1 means it can be observed; N is the number of input factors; ψ j is the Shapley value of the jth input factor; ψ 0 is a constant that explains the model, and its value is equal to the predicted mean of all training samples; f is a set that describes whether the input factor can be observed; g(f) is the adjoint model function of the input factor.

[0083] Shapley value ψ j It can be a positive value, a negative value, or 0, representing that the j-th input factor is positively correlated, negatively correlated, or uncorrelated with the output variable, respectively. The larger its absolute value, the greater the contribution of the corresponding input factor to the accuracy of multi-step flood forecasting.

[0084] Figure 6 The architecture diagram of a non-stationary converter model for flood forecasting in a river basin under the influence of reservoir regulation provided by an exemplary embodiment is shown. Using the trained model for multi-step flood forecasting can improve the accuracy of flood forecasting in the river basin and extend the flood forecast period, and enhance the adaptability and generalization ability of deep learning models in changing environments.

[0085] The implementation basis of each embodiment of the present invention is to implement programmed processing through a device with a processor function. Therefore, in engineering practice, the technical solutions and functions of each embodiment of the present invention are encapsulated into various modules. Based on this reality, on the basis of the above embodiments, an embodiment of the present invention provides a non-stationary converter device for flood forecasting in a river basin under the influence of reservoir regulation, which is used to execute the flood forecasting in a river basin under the influence of reservoir regulation in the above method embodiment.

[0086] Please refer to Figure 7 , which shows a block diagram of the non-stationary converter device for basin flood forecasting under the influence of reservoir regulation provided in the present application, the device includes: a first main module 102, used to obtain source data to be forecasted; a second main module 104, used to input the source data to be forecasted into a trained non-stationary converter deep learning model, and output a forecast flow.

[0087] Please refer to Figure 8 , which shows a block diagram of the non-stationary converter model for basin flood forecasting under the influence of reservoir regulation provided by the present application, the model includes:

[0088] The first training module 202 considers the duration of runoff generation and convergence in the basin, constructs a non-stationary converter deep learning model, the non-stationary converter deep learning model matches the basin flood forecasting framework under the influence of reservoir regulation, and uses a normalization-inverse normalization module to characterize the stationary characteristics of the flood sequence, and uses an inverse stationary attention mechanism module to capture the non-stationary characteristics of the flood sequence under the influence of reservoir regulation;

[0089] The second training module 204 considers double regularization of the objective function and uses the Adam optimization algorithm to calibrate the model parameters.

[0090] The non-stationary converter device for flood forecasting in a river basin under the influence of reservoir regulation provided by the embodiment of the present invention aims at reducing the non-stationarity of the original sequence by data stabilization. This practice is contrary to the significance of flood forecasting in a river basin under the influence of a changing environment for emergency forecasting, and ignores the universal existence of non-stationary flood sequences in a changing environment, which ultimately leads to over-stabilization of modeling and forecasting, reducing the adaptability of such models to a changing environment and the current status of the model generalization ability. Figure 5 Several modules in it can not only adapt to the basin flood forecasting architecture with different non-stationary characteristics, but also make full use of the stationary and non-stationary characteristics of basin flood sequences in a changing environment, overcome the technical bottleneck of over-stationary learning of machine learning models, and improve the adaptability and generalization ability of deep learning models in a changing environment.

[0091] It should be noted that the device embodiments provided by the present invention are used to implement the methods in the above method embodiments as well as the methods in other method embodiments provided by the present invention. The only difference is that the corresponding functional modules are set, and the principle is basically the same as the principle of the above device embodiments provided by the present invention. As long as the technical personnel in this field refer to the specific technical solutions in other method embodiments on the basis of the above device embodiments, obtain the corresponding technical means and the technical solutions composed of these technical means by combining the technical features, and on the premise of ensuring the practicality of the technical solutions, they can improve the equipment in the above device embodiments to obtain the corresponding device class embodiments, which are used to implement the methods in other method class embodiments. For example:

[0092] Based on the contents of the above device embodiments, as a preferred embodiment, the non-stationary converter device for flood forecasting in a river basin under the influence of reservoir regulation provided in the embodiments of the present invention further includes:

[0093] The third main module applies interpretable methods to analyze the contribution of stationary and non-stationary features to the accuracy of basin flood forecasting, and improve the adaptability and generalization ability of deep learning models under changing environments.

[0094] Based on the content of the above-mentioned device embodiment, as a preferred embodiment, in the non-stationary converter device for basin flood forecasting under the influence of reservoir regulation provided in the embodiment of the present invention, the matching basin flood forecasting architecture under the influence of reservoir regulation of the non-stationary converter deep learning model includes an encoder and a decoder, wherein the encoder is intended to add the non-stationary features of the flood sequence to the attention mechanism, and the decoder is intended to bring the non-stationary features such as mutations, trends, and cycles caused by reservoir regulation into the calculation after stabilization through a classification function, and adapt the non-stationary converter deep learning model.

[0095] Based on the contents of the above-mentioned device embodiments, as a preferred embodiment, a non-stationary converter device for flood forecasting in a river basin under the influence of reservoir regulation is provided in the embodiments of the present invention, and the training of the non-stationary converter deep learning model also includes: for the stationary characteristics of the flood sequence, a normalization-inverse normalization module is used to characterize it; for the non-stationary characteristics of the flood sequence, an inverse stationary attention mechanism module is used to capture it.

[0096] Since the above methods have been discussed in detail, the specific implementation of the above modules will not be repeated here.

[0097] Next, an execution device provided by an embodiment of the present invention is introduced. Fig. 9 , Fig. 9 A schematic diagram of a structure of an execution device provided in an embodiment of the present invention, the execution device 300 can be specifically manifested as an autonomous driving vehicle, a mobile phone, a tablet, a laptop computer, a desktop computer, a monitoring data processing device, etc., which is not limited here. Figure 1 The functions of the execution device in the corresponding embodiment. Specifically, the execution device 300 includes: a receiver 301, a transmitter 302, a processor 303 and a memory 304 (wherein the number of the processor 303 in the execution device 300 can be one or more, Fig. 9 In the example of FIG. 3 , the processor 303 may include an application processor 3031 and a communication processor 3032. In some embodiments of the present application, the receiver 301, the transmitter 302, the processor 303 and the memory 304 may be connected via a bus or other means.

[0098] The memory 304 may include a read-only memory and a random access memory, and provides instructions and data to the processor 303. A portion of the memory 304 may also include a non-volatile random access memory (NVRAM). The memory 304 stores processor and operation instructions, executable modules or data structures, or subsets thereof, or extended sets thereof, wherein the operation instructions may include various operation instructions for implementing various operations.

[0099] The processor 303 controls the operation of the execution device. In a specific application, the various components of the execution device are coupled together through a bus system, wherein the bus system includes not only a data bus but also a power bus, a control bus, and a status signal bus, etc. However, for the sake of clarity, various buses are referred to as bus systems in the figure.

[0100] The method disclosed in the above embodiment of the present invention can be applied to the processor 303, or implemented by the processor 303. The processor 303 can be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit in the processor 303 or the instructions in the form of software. The above processor 303 can be a general processor, a digital signal processor (digital signal processing, DSP), a microprocessor or a microcontroller, and can further include an application specific integrated circuit (application specific integrated circuit, ASIC), a field programmable gate array (field-programmable gate array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The processor 303 can implement or execute the methods, steps and logic block diagrams disclosed in the embodiment of the present invention. The general processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in conjunction with the embodiment of the present invention can be directly embodied as a hardware decoding processor to be executed, or a combination of hardware and software modules in the decoding processor can be executed. The software module may be located in a storage medium mature in the art, such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 304, and the processor 303 reads the information in the memory 304 and completes the steps of the above method in combination with its hardware.

[0101] The receiver 301 can be used to receive input digital or character information and generate signal input related to the relevant settings and function control of the execution device. The transmitter 302 can be used to output digital or character information through the first interface; the transmitter 302 can also be used to send instructions to the disk group through the first interface to modify the data in the disk group; the transmitter 302 can also include a display device such as a display screen.

[0102] In the embodiment of the present invention, the processor 303 is used to execute Figure 1 The specific manner in which the application processor 3031 in the processor 303 performs the above steps is the same as that in the present application. Figure 1 The corresponding method embodiments are based on the same concept, and the technical effects they bring are the same as those in this application. Figure 1 The corresponding method embodiments are the same. For specific contents, please refer to the description in the method embodiments shown above in this application, which will not be repeated here.

[0103] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0104] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0105] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0106] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0107] In summary, the present invention considers the duration of runoff generation and convergence in the basin, constructs a non-stationary converter deep learning model, matches the basin flood forecasting framework under the influence of reservoir regulation, adopts a normalization-inverse normalization module to characterize the stationary information of the flood sequence, and adopts an inverse stationary attention mechanism module to capture the non-stationary information of the flood sequence under the influence of reservoir regulation; considers the double regularization of the objective function, and adopts the Adam optimization algorithm to calibrate the parameters of the non-stationary converter deep learning model; applies an interpretable method to analyze the contribution of stationary information and non-stationary information to the accuracy of basin flood forecasting, which not only improves the accuracy of basin flood forecasting, but also extends the flood forecast period, thereby improving the adaptability and generalization ability of the deep learning model in a changing environment.

[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the technical solutions of the embodiments of the present invention.

Claims

1. A non-stationary converter method for flood forecasting in a river basin under the influence of reservoir regulation, characterized in that include: Obtain source data to be forecasted; Input the source data to be predicted into the trained non-stationary converter deep learning model, and output the predicted flow; wherein the training of the non-stationary converter deep learning model includes: Considering the duration of runoff generation and convergence in the basin, a non-stationary converter deep learning model is constructed. The non-stationary converter deep learning model matches the basin flood forecasting framework under the influence of reservoir regulation, and uses a normalization-inverse normalization module to characterize the stationary characteristics of the flood sequence, and uses an inverse stationary attention mechanism module to capture the non-stationary characteristics of the flood sequence under the influence of reservoir regulation; According to the non-stationary converter deep learning model, the double regularization of the objective function is considered, the Adam optimization algorithm is used to calibrate the model parameters, and the trained non-stationary converter deep learning model is output.

2. The non-stationary converter method for flood forecasting in a river basin under the influence of reservoir regulation according to claim 1 is characterized in that: The method also includes: applying an interpretable method to analyze the contribution of stationary features and non-stationary features to the accuracy of basin flood forecasting, and improving the adaptability and generalization ability of deep learning models under changing environments.

3. The non-stationary converter method for flood forecasting in a river basin under the influence of reservoir regulation according to claim 1, characterized in that: The basin flood forecasting architecture under the influence of reservoir regulation that matches the non-stationary converter deep learning model includes an encoder and a decoder, wherein the encoder is intended to add the non-stationary features of the flood sequence to the attention mechanism, and the decoder is intended to bring the non-stationary features caused by reservoir regulation into the calculation after stabilization through a classification function, thereby adapting the non-stationary converter deep learning model.

4. A non-stationary converter device for flood forecasting in a river basin under the influence of reservoir regulation, characterized in that: include: The first main module is used to obtain source data to be forecasted; The second main module is used to input the source data to be predicted into the trained non-stationary converter deep learning model and output the predicted flow rate; wherein the training of the non-stationary converter deep learning model includes: Considering the duration of runoff generation and convergence in the basin, a non-stationary converter deep learning model is constructed. The non-stationary converter deep learning model matches the basin flood forecasting framework under the influence of reservoir regulation, and uses a normalization-inverse normalization module to characterize the stationary characteristics of the flood sequence, and uses an inverse stationary attention mechanism module to capture the non-stationary characteristics of the flood sequence under the influence of reservoir regulation; According to the non-stationary converter deep learning model, the double regularization of the objective function is considered, the Adam optimization algorithm is used to calibrate the model parameters, and the trained non-stationary converter deep learning model is output.

5. The non-stationary converter device for flood forecasting in a river basin under the influence of reservoir regulation according to claim 4, characterized in that: Also includes: The third main module is used to apply interpretable methods to analyze the contribution of stationary and non-stationary features to the accuracy of basin flood forecasting, and to improve the adaptability and generalization ability of deep learning models under changing environments.

6. The non-stationary converter device for flood forecasting in a river basin under the influence of reservoir regulation according to claim 4, characterized in that: The basin flood forecasting architecture under the influence of reservoir regulation that matches the non-stationary converter deep learning model includes an encoder and a decoder, wherein the encoder is intended to add the non-stationary features of the flood sequence to the attention mechanism, and the decoder is intended to bring the non-stationary features caused by reservoir regulation into the calculation after stabilization through a classification function, thereby adapting the non-stationary converter deep learning model.

7. An execution device, characterized in that: It comprises a processor and a memory, wherein the processor is coupled to the memory; the memory is used to store programs; the processor is used to execute the programs in the memory, so that the execution device executes the method according to any one of claims 1 to 3.

8. A non-transitory computer readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions cause the computer to execute the method of any one of claims 1 to 3.