A deep learning method and device for evaluating changes in water level and flow in the middle and lower reaches of a river basin

Abstract

The present invention discloses a deep learning method and device for evaluating changes in water level and flow in the middle and lower reaches of a river basin. The method includes the collection and preprocessing of hydrological section data; flow restoration calculation and evolution of the upstream control station of the river basin; construction of a proxy model for responding to water level and flow changes, based on deep learning, constructing a proxy model for responding to upstream reservoir regulation and storage and water level and flow changes in the middle and lower reaches of the river basin; data set division and proxy model testing and application, dividing time series data into training set, validation set, and test set, and finally obtaining the flow change value of the control station in the middle and lower reaches of the river basin affected by upstream reservoir regulation and storage. The present invention combines measured water conservancy data and flood evolution models, and constructs a proxy model for responding to upstream reservoir regulation and storage and water level and flow changes in the middle and lower reaches of the river basin based on deep learning. It can achieve immediate response to the downstream hydrological situation in different regulation and storage scenarios of the upstream reservoir of the river basin, meeting the real-time and rapid response requirements when evaluating the scheduling strategy.

Description

Technical Field

[0001] The present invention relates to the field of hydrology and water resources, and more specifically, to a deep learning method and device for evaluating changes in water level and flow in the downstream of a river basin. Background Art

[0002] Amid global climate change and increasingly scarce water resources, reservoirs, as core components of water conservancy projects, play an irreplaceable role in flood control, power generation, irrigation, water supply, and ecological restoration. However, with the construction and operation of large-scale reservoir clusters, particularly mega-reservoirs consisting of multiple series-parallel reservoirs, these clusters, as key control projects in the upper reaches of river basins, have directly altered the hydrological regimes in the middle and lower reaches of the basin.

[0003] While traditional research methods, such as hydrodynamic models based on river flood evolution, have made significant progress in simulating hydrodynamic processes, they are complex and time-consuming to calibrate, limiting their ability to rapidly assess the impact of varying storage conditions in upstream reservoirs on the hydrological regime in the middle and lower reaches of a river basin. Given the complex regulation system of multiple series and parallel reservoirs and the hydrological response characteristics of the complex river network within a river basin, traditional methods struggle to quickly and accurately capture and simulate changes in the hydrological regime in the middle and lower reaches of the basin, hindering the evaluation of different scheduling options for cascade reservoirs in the upper reaches of the basin.

[0004] Therefore, there is an urgent need for a faster and more accurate assessment method to explore the response law of the upstream reservoir storage and the changes in water level and flow in the middle and lower reaches of the basin. The limitations of the current method are mainly reflected in two aspects: (1) Insufficient model timeliness and flexibility: Although the traditional hydrodynamic model based on the river flood evolution process has made significant progress, its model calibration process is complicated and relies on expert experience for parameter adjustment. The model operation time is long, which makes it difficult to meet the real-time and rapid response requirements when quickly assessing the impact of different storage conditions of the upstream reservoir on the hydrological situation in the middle and lower reaches of the basin. Therefore, when responding to emergency scheduling decisions or evaluating rapidly changing hydrological situations, the timeliness and flexibility of the traditional model are limited. (2) Lack of efficient data fusion and processing capabilities: When evaluating the response law of the upstream reservoir storage and the changes in water level and flow in the middle and lower reaches of the basin, it is necessary to comprehensively consider multiple data sources, including but not limited to measured water conservancy data, meteorological data, flood evolution models, etc. However, the traditional method has deficiencies in data fusion and processing, making it difficult to seamlessly integrate and deeply apply multi-source heterogeneous data. Summary of the Invention

[0005] In view of the shortcomings of existing technologies, it is necessary to propose a method that can accurately predict the changes in water level and flow in the middle and lower reaches of the basin, and further propose an evaluation method that takes into account the response law of the regulation and storage of reservoirs in the upstream of the basin and the changes in water level and flow in the middle and lower reaches. In this way, when analyzing the relationship between water level and flow at hydrological stations, it is possible to quickly judge the impact of different regulation and storage conditions of reservoirs in the upstream of the basin on the hydrological situation in the middle and lower reaches.

[0006] To achieve the above objectives, the first aspect of the present invention provides a deep learning method for evaluating changes in water level and flow in the downstream of a river basin, comprising:

[0007] Step S1: Collect and preprocess hydrological section data;

[0008] Step S2: Using a flow restoration calculation method, the measured flow at the upstream control station of the basin is restored to the natural flow, and the storage change of the upstream reservoir of the basin is further calculated. The measured flow and natural flow at the upstream control station of the basin are evolved to the mid- and downstream control stations of the basin through a one-dimensional hydrodynamic model to obtain the flow change value of the mid- and downstream control stations affected by the storage and regulation of the upstream reservoir;

[0009] Step S3: Based on deep learning methods, a proxy model is constructed to represent the response of upstream reservoir regulation and water level and flow changes in the middle and lower reaches of the basin. The input of the model is the natural flow of the upper reaches of the basin mainstream, the measured flow of other tributaries, the measured water level of the control station in the middle and lower reaches of the basin considering time lag, and the change in regulation of the upstream reservoir. The output of the model is the flow change value of the control station in the middle and lower reaches of the basin affected by the regulation of the upstream reservoir.

[0010] Step S4: Train the proxy model for the response of the upstream reservoir storage and the mid- and downstream water level and flow changes in the basin, and use the trained model to predict the changes in the mid- and downstream water level and flow.

[0011] In one embodiment, step S1 includes:

[0012] Collect measured water level and flow data from hydrological stations upstream and downstream of the basin, including measured flow in the upper reaches of the main stream, measured flow in other tributaries, and measured flow and water level at control stations in the middle and lower reaches of the basin;

[0013] The collected data and information are cleaned to form time series data.

[0014] In one embodiment, step S2 includes:

[0015] The flow reduction calculation method is used to calculate the reservoir inflow before storage and regulation according to the water balance equation of the upstream reservoir of the basin.

[0016] Based on the reservoir inflow before storage, the evolution flow at the outlet section of the upstream control station of the basin is calculated according to the Changban confluence curve formula, and the river flood evolution from the dam site of the upstream reservoir to the control section is completed;

[0017] The evolved flow of each reservoir at the control section is superimposed with the measured flow at the control section to obtain the natural flow restored at the section of the upstream control station of the basin. Based on the difference between the natural flow and the measured flow, the storage change of the upstream reservoir of the basin is obtained.

[0018] The flow before regulation is obtained by restoring the evolution of natural flow, and the flow after regulation is obtained by the evolution of measured flow. According to the difference between the flow before regulation and the flow after regulation, the flow change value of the control station in the middle and lower reaches of the basin affected by the regulation of the reservoir upstream of the basin is obtained.

[0019] In one embodiment, a flow restoration calculation method is used to calculate the reservoir inflow before storage according to the water balance equation based on the reservoir storage capacity, including: according to the reservoir water level, storage capacity curve and outflow, the inflow is calculated based on the water balance method, specifically:

[0020]

[0021] Where: is the average inflow flow during the period, is the average outbound flow during the period, is the water loss in the reservoir, The change in reservoir water storage at the beginning and end of the time period; is the calculation period;

[0022] The formula for the long-term convergence curve is:

[0023]

[0024] Where: For the The vertical coordinate value of the confluence parameter of a river section is called the confluence parameter; is the relative time, For time; is the average confluence time of a river section; is the number of river sections it flows through; is the relative inflow termination time, is the end time of inflow, and the start time of inflow is taken as the starting time. ; is the base of natural logarithms; It is the relative time starting from the end of inflow.

[0025] In one embodiment, in step S3, when constructing an agent model for the regulation of reservoirs in the upper reaches of the basin and the response of water level and flow changes in the middle and lower reaches based on a deep learning method, the model used is a recursive neural network of the fitting type.

[0026] In one embodiment, the proxy model of the upstream reservoir regulation and the mid- and downstream water level and flow change response in step S3 constructs the mapping relationship as follows:

[0027]

[0028] in, , , , The flow changes of the control stations in the middle and lower reaches of the basin under the influence of the upstream reservoir regulation; is the natural flow at the upstream control station of the basin; is the measured discharge at the upstream control station in the basin; To regulate the storage capacity of reservoirs upstream of the basin; Measured flows for other tributaries; The measured water levels at the downstream control stations in the basin are taken into account for the time lag; It is the mapping function between the storage capacity of the upstream reservoir and the response of the water level and flow change in the middle and lower reaches of the basin; A hydrodynamic model for the evolution of flow from upstream control stations to mid- and downstream control stations in the basin; is the hydrodynamic model of the evolution of the reservoir storage variable to the upstream control station of the basin, where is the total number of reservoirs above the control station, is the reservoir number, For reservoirs The storage variable.

[0029] In one embodiment, the method further includes: using the prediction results obtained in step S4 to evaluate the response law of water level and flow changes in the middle and lower reaches of the basin under different storage scenarios of the upstream reservoir.

[0030] Based on the same inventive concept, the second aspect of the present invention provides a deep learning device for evaluating changes in water level and flow in the downstream of a river basin, comprising:

[0031] Data collection and preprocessing module, used to collect and preprocess hydrological section data;

[0032] The flow restoration calculation and evolution module is used to restore the measured flow at the upstream control station of the basin to the natural flow using the flow restoration calculation method, further calculate the storage change of the upstream reservoir of the basin, and evolve the measured flow and natural flow at the upstream control station of the basin to the mid- and downstream control stations of the basin through a one-dimensional hydrodynamic model to obtain the flow change value of the mid- and downstream control stations affected by the storage and regulation of the upstream reservoir;

[0033] The model building module is used to construct a proxy model for the response of upstream reservoir regulation and water level and flow changes in the middle and lower reaches of the basin based on deep learning methods. The input of the model is the natural flow of the upstream main stream of the basin, the measured flow of other tributaries, the measured water level of the control station in the middle and lower reaches of the basin considering time lag, and the change in regulation of the upstream reservoir. The model output is the change in flow at the control station in the middle and lower reaches of the basin affected by the regulation of the upstream reservoir.

[0034] The training and testing module is used to train the proxy model for the response of reservoir storage in the upper reaches of the basin and water level and flow changes in the middle and lower reaches, and use the trained model to predict water level and flow changes in the middle and lower reaches of the basin.

[0035] In one embodiment, the model building module is further used to determine the mapping relationship between the model input and output variables, wherein the model input is the natural flow of the upstream of the main stream of the basin, the measured flow of other tributaries, the measured water level of the downstream control station of the basin taking into account the time lag, and the change in the storage of the upstream reservoir of the basin; the model output is the change in the flow of the downstream control station of the basin affected by the storage of the upstream reservoir;

[0036] Based on the same inventive concept, the third aspect of the present invention provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, it implements the deep learning method for evaluating changes in downstream water level and flow in a basin as described in the first aspect.

[0037] Based on the same inventive concept, the fourth aspect of the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the program, it implements the deep learning method for evaluating changes in downstream water level and flow in a river basin as described in the first aspect.

[0038] Compared with the prior art, the advantages and beneficial technical effects of the present invention are as follows:

[0039] 1. The core of this invention is to use deep learning technology, combined with measured water conservancy data and flood evolution models, to accurately simulate the regulation and storage function of the upstream reservoir group in the basin, accurately predict the regulation and storage of the upstream reservoirs and the changes in water levels and flows in the middle and lower reaches of the basin, and evaluate their impact on the hydrological situation of major flood control nodes in the middle and lower reaches of the basin;

[0040] 2. This invention uses deep learning to construct a proxy model for the regulation of upstream reservoirs and the response to changes in water levels and flows in the middle and lower reaches of the basin. This model can achieve immediate responses to the hydrological situation in the middle and lower reaches of the basin under different regulation scenarios of upstream reservoirs, meeting the real-time and rapid response requirements when evaluating scheduling strategies.

[0041] 3. The present invention can be used to integrate and apply multi-source heterogeneous data. By combining real-time monitoring data of water conditions and water level and flow data of key control stations simulated by hydrodynamic models, it can achieve efficient integration and full utilization of diverse data. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use 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.

[0043] Figure 1 is a flow chart of a deep learning method for evaluating changes in water level and flow in the downstream of a river basin, provided by an embodiment of the present invention;

[0044] Figure 2 This is a comparison chart of simulated response flow changes and actual response flow changes of downstream control stations in the basin provided by an embodiment of the present invention;

[0045] Figure 3 2 is a schematic structural diagram of a deep learning device for evaluating changes in water level and flow in the downstream of a river basin in an embodiment of the present invention. DETAILED DESCRIPTION

[0046] The present invention discloses a deep learning method and device for evaluating changes in water level and flow in the middle and lower reaches of a river basin. The method includes: collecting and preprocessing hydrological section data; restoring the flow of the upstream control station of the river basin and its evolution, restoring the measured flow of the upstream and middle and lower reaches of the river basin to the natural flow, and calculating the storage change of the upstream reservoir of the river basin and the flow change value of the middle and lower reaches of the river basin affected by the storage of the upstream reservoir by a one-dimensional hydrodynamic model; constructing a proxy model for the response of the upstream reservoir storage and the middle and lower reaches of the river basin to water level and flow changes based on deep learning; dividing the data set and testing the proxy model, dividing the time series data into a training set, a validation set, and a test set, and finally obtaining the flow change value of the middle and lower reaches of the river basin affected by the storage of the upstream reservoir.

[0047] This invention combines measured water conservancy data and flood evolution models, and constructs an agent model based on deep learning for the regulation of upstream reservoirs and the response of water level and flow changes in the middle and lower reaches of the basin. It can achieve instant response to the hydrological situation in the middle and lower reaches of the basin under different regulation scenarios of upstream reservoirs, and meet the real-time and rapid response requirements when evaluating scheduling strategies.

[0048] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0049] Example 1

[0050] This embodiment discloses a deep learning method for evaluating changes in water level and flow in the downstream of a basin. Figure 1 ,include:

[0051] Step S1: Collect and preprocess hydrological section data;

[0052] Step S2: Using a flow restoration calculation method, the measured flow at the upstream control station of the basin is restored to the natural flow, and the storage change of the upstream reservoir of the basin is further calculated. The measured flow and natural flow at the upstream control station of the basin are evolved to the mid- and downstream control stations of the basin through a one-dimensional hydrodynamic model to obtain the flow change value of the mid- and downstream control stations affected by the storage and regulation of the upstream reservoir;

[0053] Step S3: Based on deep learning methods, a proxy model is constructed to represent the response of upstream reservoir regulation and water level and flow changes in the middle and lower reaches of the basin. The input of the model is the natural flow of the upper reaches of the basin mainstream, the measured flow of other tributaries, the measured water level of the control station in the middle and lower reaches of the basin considering time lag, and the change in regulation of the upstream reservoir. The output of the model is the flow change value of the control station in the middle and lower reaches of the basin affected by the regulation of the upstream reservoir.

[0054] Step S4: Train the proxy model for the response of the upstream reservoir storage and the mid- and downstream water level and flow changes in the basin, and use the trained model to predict the changes in the mid- and downstream water level and flow.

[0055] In one embodiment, step S1 includes:

[0056] Collect measured water level and flow data from hydrological stations upstream and downstream of the basin, including measured flow in the upper reaches of the main stream, measured flow in other tributaries, and measured flow and water level at control stations in the middle and lower reaches of the basin;

[0057] The collected data and information are cleaned to form time series data.

[0058] During the specific implementation process, the collected data and information are cleaned, including processing missing values ​​and outliers.

[0059] In one embodiment, step S2 includes:

[0060] The flow reduction calculation method is used to calculate the reservoir inflow before storage and regulation according to the water balance equation of the upstream reservoir of the basin.

[0061] Based on the reservoir inflow before storage, the evolution flow at the outlet section of the upstream control station of the basin is calculated according to the Changban confluence curve formula, and the river flood evolution from the dam site of the upstream reservoir to the control section is completed;

[0062] The evolved flow of each reservoir at the control section is superimposed with the measured flow at the control section to obtain the natural flow restored at the section of the upstream control station of the basin. Based on the difference between the natural flow and the measured flow, the storage change of the upstream reservoir of the basin is obtained.

[0063] The flow before regulation is obtained by restoring the evolution of natural flow, and the flow after regulation is obtained by the evolution of measured flow. According to the difference between the flow before regulation and the flow after regulation, the flow change value of the control station in the middle and lower reaches of the basin affected by the regulation of the reservoir upstream of the basin is obtained.

[0064] Specifically, in order to maintain the water balance of floods in each zone, the annual storage variables of each reservoir upstream of the hydrological station are calculated to the hydrological station and the annual restored flow of the hydrological station is calculated in an overlay manner.

[0065] During this implementation process, the Yangtze River Basin and the Three Gorges and its upstream reservoirs are taken as examples to illustrate the evolution of the flow restoration calculation method and the one-dimensional hydrodynamic model. The flow restoration calculation method is only carried out for the upstream control station of the Yangtze River mainstream. The storage capacity of the Three Gorges Reservoir and the reservoir upstream of the Three Gorges are respectively calculated according to the water balance equation to calculate the reservoir inflow flow before storage, and then the evolved flow at the outlet section of the upstream control station of the Yangtze River mainstream is calculated according to the Changban confluence curve formula, thereby completing the river flood evolution from the dam site to the control section.

[0066] By superimposing the evolved flow at each reservoir's control section with the measured flow at that section, we obtain the restored natural flow at the upstream control station section of the Yangtze River mainstream. Furthermore, by subtracting the measured flow from this natural flow, we obtain the change in storage capacity at the Three Gorges Dam and its upstream reservoirs.

[0067] Then, the restored natural flow and measured flow of the above-mentioned control station section in the upper reaches of the Yangtze River were respectively evolved to the control station in the middle and lower reaches of the Yangtze River based on the one-dimensional hydrodynamic model. The flow change value of the control station in the middle and lower reaches of the Yangtze River affected by the regulation of the Three Gorges and its upstream reservoirs was obtained. That is, the flow before regulation was obtained by restoring the natural flow evolution, and the flow after regulation was obtained by the measured flow evolution. Subtracting the former from the latter can obtain the flow change value affected by the regulation of the reservoir group.

[0068] In one embodiment, a flow restoration calculation method is used to calculate the reservoir inflow before storage and regulation using the water balance equation based on the storage capacity of the upstream reservoir of the basin, including: according to the reservoir water level, storage capacity curve and outflow, the inflow is calculated based on the water balance method, specifically:

[0069]

[0070] Where: The average inflow flow during the period, unit: m 3 / s; : Average outflow during the period, which is obtained by adding the power generation flow, idling flow, ship lock flow and gate discharge flow, in m 3 / s; : Reservoir water loss, including water surface evaporation, reservoir leakage loss, etc., unit: m 3 ; : Change in reservoir water storage at the beginning and end of the period, unit: m 3 ; : calculation period, unit s;

[0071] The formula for the long-term convergence curve is:

[0072]

[0073] Where: For the The vertical coordinate value of the confluence parameter of a river section is called the confluence parameter; is the relative time, For time; is the average confluence time of a river section; is the number of river sections it flows through; is the relative inflow termination time, is the end time of inflow, and the start time of inflow is taken as the starting time. ; is the base of natural logarithms; It is the relative time starting from the end of inflow.

[0074] Specifically, in this implementation method, above the control station upstream of the Yangtze River mainstream, each reservoir needs to evolve the storage capacity, which is calculated as follows: according to the reservoir water level, storage capacity curve and outflow, the inflow is calculated based on the water balance method.

[0075] The Changban confluence curve formula is a river confluence formula that describes the unit inflow of a certain duration, which is injected from a certain section of the river and regulated by several river sections, and the outflow process generated at the outflow section. The specific formula is as follows. In the river flood routing, the Changban confluence parameters divide the calculation period into equal The river section with the best value is convenient for calculating the location of the inflow point of the tributary (interval) in the river section, and the calculation result is more consistent with the objective situation; 、 The optimal method is used to determine the flow rate Compared with the measured flow The following three indicators are used to measure the good performance:

[0076] Flood process deviation : ;

[0077] Maximum peak flow deviation : ;

[0078] Peak time deviation : .

[0079] Where: The first calculation of the long-distance confluence curve formula The flow rate of the outflow section of the river, in m 3 / s; For the Measured flow rate at the outflow section of the river, unit: m 3 / s; is the number of river sections it flows through; is the calculated value of the maximum flood peak flow, in m 3 / s; is the measured value of the maximum flood peak flow, in m 3 / s; To calculate the time of occurrence of the maximum flood peak flow; is the time when the maximum flood peak flow is measured.

[0080] In one embodiment, in step S3, when constructing an agent model for the regulation of reservoirs in the upper reaches of the basin and the response of water level and flow changes in the middle and lower reaches based on a deep learning method, the model used is a recursive neural network of the fitting type.

[0081] In the specific implementation process, the fitting type of recurrent neural networks includes long short-term memory network (LSTM), gated recurrent unit (GRU), bidirectional long short-term memory network (Bi-LSTM), deep long short-term memory network (Deep LSTM), Transformer, etc.

[0082] In one embodiment, the proxy model of the upstream reservoir regulation and the mid- and downstream water level and flow change response in step S3 constructs the mapping relationship as follows:

[0083]

[0084] in, , , ,

[0085] The flow change of the downstream control station in the basin under the influence of the upstream reservoir, unit is m 3 / s; is the natural flow at the upstream control station of the basin, in m 3 / s; is the measured flow rate at the upstream control station of the basin, in m 3 / s; is the storage capacity of the reservoir upstream of the basin, unit: m 3 / s; is the measured flow of other tributaries, unit is m 3 / s; The measured water level at the downstream control station in the basin taking into account the time lag, in meters; It is the mapping function between the storage capacity of the upstream reservoir and the response of the water level and flow change in the middle and lower reaches of the basin; A hydrodynamic model for the evolution of flow from upstream control stations to mid- and downstream control stations in the basin; is the hydrodynamic model of the evolution of the storage variable of the upstream reservoir to the upstream control station of the basin, where: The total number of reservoirs above the control station, including the Three Gorges and its upstream reservoirs, is the reservoir number, For reservoirs Storage variable, unit m 3 .

[0086] In one embodiment, the method further includes: using the prediction results obtained in step S4 to evaluate the response law of water level and flow changes in the middle and lower reaches of the basin under different storage scenarios of the upstream reservoir.

[0087] During the specific implementation process, the data set is divided and the proxy model is tested and applied. The input and output data sets formed are normalized in time series. The time series data are divided into training set, validation set and test set. The hyperparameters of the deep learning model are optimized based on the validation set combined with the hyperparameter calibration method. The model after hyperparameter calibration is applied to the training set, and the model is trained within the number of iterations. Then, the test set is used to test the model. At this time, the hyperparameters of the deep learning model and the back propagation of the network parameters are no longer adjusted. Finally, the flow change values ​​of the control stations in the middle and lower reaches of the basin affected by the storage and regulation of the upstream reservoir are obtained, and the response law of the water level and flow changes in the middle and lower reaches of the basin under different storage and regulation scenarios of the upstream reservoir is evaluated.

[0088] The model loss function uses the root mean square error between the simulated response flow changes and the actual response flow changes of the control stations in the middle and lower reaches of the basin; the Nash efficiency coefficient NSE and root mean square error RMSE are used to evaluate the proxy model of the upstream reservoir regulation and the middle and lower reaches of the basin water level and flow change response.

[0089] In order to facilitate those skilled in the art to better understand the technical solution of the present invention, specific embodiments of the present invention (taking the Yangtze River Basin and the Three Gorges Reservoir as examples) are given as follows:

[0090] The embodiment of the present invention specifically provides a deep learning method and device for evaluating changes in water level and flow in the downstream of a river basin, including the following steps:

[0091] (1) Collect and analyze the measured water level and flow data of the flow sections of various hydrological stations in the upper and lower reaches of the Yangtze River from 2008 to 2021, including the measured flow of the upper reaches of the Yangtze River, the measured flow of other tributaries, and the measured flow and water level of the control stations in the middle and lower reaches of the Yangtze River. Specifically, they are the measured flow of the upper reaches of the Yangtze River (Yichang Station), the measured flow of other tributaries (taking the measured flow of Gaobazhou Station and the measured flow of Dongting Sishui as examples), the measured flow of the control stations in the middle and lower reaches of the Yangtze River (taking Luoshan Station as an example), and the measured water level of the control stations in the middle and lower reaches of the Yangtze River (taking Lianhuatang as an example); process the missing values ​​and outliers of the above data and form time series data;

[0092] (2) The measured flow at Yichang Station was restored to the natural flow according to the flow restoration calculation method, and the change in the storage and regulation of the Three Gorges Dam and its upstream reservoirs was calculated. Furthermore, the measured flow at Yichang Station and the natural flow were used to construct a flow calculation model for the middle and lower reaches of the Yangtze River using DHI MIKE11 software. Based on the river channel morphology and water level and flow measured data, the roughness was calibrated in three sections for different water levels. Finally, the flow change value of Luoshan Station affected by the storage and regulation of the Three Gorges Dam and its upstream reservoirs was obtained.

[0093] (3) Based on deep learning methods, a proxy model for the response of the Three Gorges Dam and its upstream reservoirs to the changes in water level and flow in the middle and lower reaches was constructed. The model inputs were the natural flow of the Yichang station, the upstream control station of the Yangtze River, the changes in the storage and regulation of the Three Gorges Dam and its upstream reservoirs, the measured flow of the Gaobazhou station, the measured flow of the four Dongting Lakes, and the measured water levels of Lianhuatang at the first two moments. The model output was the flow change value of the Luoshan station affected by the storage and regulation of the Three Gorges Dam and its upstream reservoirs. The research was carried out using the long short-term memory network (LSTM) model.

[0094] (4) The input and output data sets are normalized in time series, and the time series data are divided into training set, validation set, and test set. The hyperparameters of the deep learning model are optimized based on the validation set combined with the hyperparameter calibration method. The hyperparameters are the number of hidden layers, the number of hidden layer nodes, the time step, the training batch value, the number of iterations, and the learning rate. The NSE is used as the calibration function for the hyperparameters. The hyperparameters finally obtained are shown in Table 1.

[0095] Table 1 Hyperparameters of the proxy model for the Three Gorges Dam and its upstream reservoirs and the response to water level and flow changes in the middle and lower reaches

[0096]

[0097] The model, after hyperparameter calibration, was applied to the training set and trained within the specified number of iterations. The model was then tested on the test set, without adjusting the deep learning model's hyperparameters or backpropagation of the network parameters. Finally, the flow change at Luoshan Station, affected by the Three Gorges Dam and its upstream reservoirs, was determined. The simulated flow change at Luoshan Station was compared with the actual flow change, resulting in model simulation accuracy metrics for the training set, validation set, and validation set, as shown in Table 2. The accuracy metrics, NSE and RMSE, performed well in the simulation of flow change at Luoshan Station affected by the Three Gorges Dam and its upstream reservoirs, indicating that the deep learning model is able to accurately fit the flow change at Luoshan Station under the influence of the Three Gorges Dam and its upstream reservoirs.

[0098] Table 2 Accuracy of the proxy model for water level and flow change response

[0099]

[0100] (6) The simulated response flow change of Luoshan Station obtained by the water level flow change response proxy model is compared with the actual response flow change. The results are as follows: Figure 2 As shown in Figure 2, the proxy model can well simulate the flow changes at Luoshan Station under the influence of the Three Gorges Dam and its upstream reservoirs.

[0101] Example 2

[0102] Based on the same inventive concept, this embodiment discloses a deep learning device for evaluating changes in water level and flow in the downstream of a river basin. Figure 3 ,include:

[0103] The data collection and preprocessing module 101 is used to collect and preprocess the hydrological section data;

[0104] The flow restoration calculation and evolution module 102 is used to restore the measured flow at the upstream control station of the basin to the natural flow using a flow restoration calculation method, further calculate the storage change of the Three Gorges Dam and its upstream reservoirs, and evolve the measured flow and natural flow at the upstream control station of the basin to the mid- and downstream control stations of the basin using a one-dimensional hydrodynamic model to obtain the flow change value of the mid- and downstream control stations affected by the storage and regulation of the upstream reservoirs;

[0105] The model construction module 103 is used to construct a proxy model for the response of upstream reservoir storage and mid- and downstream water level and flow changes based on deep learning methods. The input of the model is the natural flow of the upstream main stream of the basin, the measured flow of other tributaries, the measured water level of the mid- and downstream control stations of the basin taking into account time lag, and the change in storage and regulation of the upstream reservoir. The model output is the change in flow at the mid- and downstream control stations of the basin affected by the storage and regulation of the upstream reservoir.

[0106] The training and testing module 104 is used to train the proxy model for the response of the upstream reservoir storage and the water level and flow changes in the middle and lower reaches of the basin, and use the trained model to predict the water level and flow changes in the middle and lower reaches of the Yangtze River.

[0107] Since the device described in Example 2 of the present invention is used to implement the deep learning method for assessing changes in water level and flow in the downstream of a watershed described in Example 1 of the present invention, the specific structure and variations of the device are readily understood by those skilled in the art based on the method described in Example 1 of the present invention, and thus will not be further described here. All devices used in the method described in Example 1 of the present invention fall within the scope of protection of the present invention.

[0108] Example 3

[0109] Based on the same inventive concept, the present invention further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method described in the first embodiment is implemented.

[0110] Since the computer-readable storage medium described in Example 3 of the present invention is used to implement the deep learning method for assessing changes in water level and flow in the downstream of a watershed in Example 1 of the present invention, the specific structure and variations of the computer-readable storage medium are readily understood by those skilled in the art based on the method described in Example 1 of the present invention, and thus will not be further described here. All computer-readable storage media used in the method of Example 1 of the present invention fall within the scope of protection of the present invention.

[0111] Example 4

[0112] The present invention also provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method described in the first embodiment when executing the program.

[0113] Since the computer device described in Example 4 of the present invention is used to implement the deep learning method for assessing changes in water level and flow in the downstream of a river basin described in Example 1 of the present invention, the specific structure and variations of the computer device are readily understood by those skilled in the art based on the method described in Example 1 of the present invention, and thus will not be further described here. All computer devices used in the method of Example 1 of the present invention fall within the scope of protection of the present invention.

[0114] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, 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 magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0115] 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 flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, 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 flowcharts and / or block diagrams. 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.

[0116] Although preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they are aware of the basic inventive concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention. Obviously, those skilled in the art may make various changes and modifications to the embodiments of the present invention without departing from the spirit and scope of the embodiments of the present invention. Thus, the present invention is intended to include such changes and modifications to the embodiments of the present invention if they fall within the scope of the claims and their equivalents.

Claims

1. A deep learning method for evaluating changes in water level and flow in the middle and lower reaches of a river basin, characterized by: include: Step S1: Collect and preprocess hydrological section data; Step S2: Using a flow restoration calculation method, the measured flow at the upstream control station of the basin is restored to the natural flow, and the storage change of the reservoir is further calculated. The measured flow and natural flow at the upstream control station of the basin are evolved to the mid- and downstream control stations of the basin through a one-dimensional hydrodynamic model to obtain the flow change value of the mid- and downstream control stations affected by the storage and regulation of the upstream reservoir; Step S3: Based on deep learning methods, a proxy model is constructed to represent the response of upstream reservoir regulation and water level and flow changes in the middle and lower reaches of the basin. The input of the model is the natural flow of the upper reaches of the basin mainstream, the measured flow of other tributaries, the measured water level of the control station in the middle and lower reaches of the basin considering time lag, and the change in regulation of the upstream reservoir. The output of the model is the flow change value of the control station in the middle and lower reaches of the basin affected by the regulation of the upstream reservoir. Step S4: training a proxy model for the response of upstream reservoir regulation and mid- and downstream water level and flow changes, and using the trained model to predict mid- and downstream water level and flow changes; Wherein, step S2 includes: The flow reduction calculation method is used to calculate the reservoir inflow before storage and regulation by using the water balance equation for the storage capacity of the Three Gorges Reservoir and the reservoir upstream of the Three Gorges. Calculate the evolution flow at the outlet section of the upstream control station according to the Changban confluence curve formula, and complete the river flood evolution from the dam site to the control section; The evolved flow of each reservoir at the control section is superimposed with the measured flow of the control section to obtain the natural flow restored at the section of the upstream control station of the basin. The storage change of the reservoir is obtained based on the difference between the natural flow and the measured flow. The flow before storage is obtained by restoring the natural flow evolution, and the flow after storage is obtained by measuring the flow evolution. Based on the difference between the flow before storage and the flow after storage, the flow change value of the control station in the middle and lower reaches of the basin affected by the storage of the upstream reservoir is obtained; The flow reduction calculation method is used to calculate the reservoir inflow before regulation based on the water balance equation for the storage capacity of the Three Gorges Reservoir and the reservoir upstream of the Three Gorges. This includes: according to the reservoir water level, storage capacity curve and outflow, the inflow is calculated based on the water balance method. Specifically, Where: is the average inflow flow during the period, is the average outbound flow during the period, is the water loss in the reservoir, The change in reservoir water storage at the beginning and end of the time period; is the calculation period; The formula for the long-term convergence curve is: Where: For the The vertical coordinate value of the confluence parameter of a river section is called the confluence parameter; is the relative time, For time; is the average confluence time of a river section; is the number of river sections it flows through; is the relative inflow termination time, is the end time of inflow, and the start time of inflow is taken as the starting time. ; is the base of natural logarithms; It is the relative time starting from the end of inflow.

2. The deep learning method for evaluating changes in water level and flow in the downstream of a river basin according to claim 1, characterized in that: Step S1 includes: Collect measured water level and flow data from hydrological stations upstream and downstream of the basin, including measured flow in the upper reaches of the main stream, measured flow in other tributaries, and measured flow and water level at control stations in the middle and lower reaches of the basin; The collected data and information are cleaned to form time series data.

3. The deep learning method for evaluating changes in water level and flow in the downstream of a river basin according to claim 1, characterized in that: In step S3, when constructing an agent model for the regulation of reservoirs in the upper reaches of the basin and the response of water level and flow changes in the middle and lower reaches based on the deep learning method, the model used is a recursive neural network of the fitting class.

4. The deep learning method for evaluating changes in water level and flow in the middle and lower reaches of a river basin as claimed in claim 1, characterized in that: In step S3, the proxy model for the response of upstream reservoir storage and mid- and downstream water level and flow changes is constructed with the following mapping relationship: in, , , , The flow changes of the control stations in the middle and lower reaches of the basin under the influence of the upstream reservoir regulation; is the natural flow at the upstream control station of the basin; is the measured discharge at the upstream control station in the basin; To regulate the storage capacity of reservoirs upstream of the basin; Measured flows for other tributaries; The measured water levels at the downstream control stations in the basin are taken into account for the time lag; It is the mapping function between the storage capacity of the upstream reservoir and the response of the water level and flow change in the middle and lower reaches of the basin; A hydrodynamic model for the evolution of flow from upstream control stations to mid- and downstream control stations in the basin; is the hydrodynamic model of the evolution of the reservoir storage variable to the upstream control station of the basin, where is the total number of reservoirs above the control station, is the reservoir number, For reservoirs The storage variable.

5. The deep learning method for evaluating changes in water level and flow in the downstream of a river basin as claimed in claim 1, characterized in that: The method further includes: using the prediction result obtained in step S4 to evaluate the response law of water level and flow changes in the middle and lower reaches of the basin under different storage and regulation scenarios of the upstream reservoir.

6. A deep learning device for evaluating changes in water level and flow in the middle and lower reaches of a river basin, characterized in that: include: Data collection and preprocessing module, used to collect and preprocess hydrological section data; The flow restoration calculation and evolution module is used to restore the measured flow at the upstream control station of the basin to the natural flow using the flow restoration calculation method, further calculate the change in reservoir storage, and evolve the measured flow and natural flow at the upstream control station of the basin to the mid- and downstream control stations of the basin through a one-dimensional hydrodynamic model to obtain the flow change value of the mid- and downstream control stations affected by the storage and regulation of the upstream reservoir; The model building module is used to construct a proxy model for the response of upstream reservoir regulation and water level and flow changes in the middle and lower reaches of the basin based on deep learning methods. The input of the model is the natural flow of the upstream main stream of the basin, the measured flow of other tributaries, the measured water level of the control station in the middle and lower reaches of the basin considering time lag, and the change in regulation of the upstream reservoir. The model output is the change in flow at the control station in the middle and lower reaches of the basin affected by the regulation of the upstream reservoir. The training and testing module is used to train the proxy model for the response of upstream reservoir storage and mid- and downstream water level and flow changes, and use the trained model to predict the changes in mid- and downstream water level and flow; The traffic restoration calculation and evolution module is specifically used for: The flow reduction calculation method is used to calculate the reservoir inflow before storage and regulation by using the water balance equation for the storage capacity of the Three Gorges Reservoir and the reservoir upstream of the Three Gorges. Calculate the evolution flow at the outlet section of the upstream control station according to the Changban confluence curve formula, and complete the river flood evolution from the dam site to the control section; The evolved flow of each reservoir at the control section is superimposed with the measured flow of the control section to obtain the natural flow restored at the section of the upstream control station of the basin. The storage change of the reservoir is obtained based on the difference between the natural flow and the measured flow. The flow before storage is obtained by restoring the natural flow evolution, and the flow after storage is obtained by measuring the flow evolution. Based on the difference between the flow before storage and the flow after storage, the flow change value of the control station in the middle and lower reaches of the basin affected by the storage of the upstream reservoir is obtained; The flow reduction calculation method is used to calculate the reservoir inflow before regulation based on the water balance equation for the storage capacity of the Three Gorges Reservoir and the reservoir upstream of the Three Gorges. This includes: according to the reservoir water level, storage capacity curve and outflow, the inflow is calculated based on the water balance method. Specifically, Where: is the average inflow flow during the period, is the average outbound flow during the period, is the water loss in the reservoir, The change in reservoir water storage at the beginning and end of the time period; is the calculation period; The formula for the long-term convergence curve is: Where: For the The vertical coordinate value of the confluence parameter of a river section is called the confluence parameter; is the relative time, For time; is the average confluence time of a river section; is the number of river sections it flows through; is the relative inflow termination time, is the end time of inflow, and the start time of inflow is taken as the starting time. ; is the base of natural logarithms; It is the relative time starting from the end of inflow.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the deep learning method for evaluating changes in downstream water level and flow in a river basin as claimed in any one of claims 1 to 5 is implemented.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the deep learning method for evaluating changes in downstream water level and flow in a river basin as described in any one of claims 1 to 5 is implemented.

Images