Construction method of reservoir capacity siltation loss rate prediction model
Through the combination of random forest model and binary split decision tree, a model that can more accurately predict the reservoir capacity silt loss rate on a global scale was constructed, which solved the prediction deviation problem under the influence of factor interaction in the existing technology, and improved the prediction Accuracy and accuracy.
Patent Information
- Application Number
- CN202411182767.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-27
- Publication Date
- 2024-12-03
AI Technical Summary
When the existing reservoir capacity silt loss rate prediction model takes into account global scale and accuracy, it is difficult to comprehensively consider the dynamic interaction of factors such as runoff, basin area, reservoir capacity and latitude and longitude, resulting in deviations from the prediction results from the actual situation.
The random forest model is used to construct the reservoir capacity siltation loss rate prediction model. By obtaining the data of reservoir characteristic variables, recursive region division and binary split decision tree construction are carried out, and the decision tree prediction results are integrated to improve the prediction accuracy.
It achieves higher prediction accuracy and more accurate prediction of reservoir capacity silt loss rate on a global scale, which can better simulate the dynamic interaction between factors such as reservoir runoff, basin area, reservoir capacity, and latitude and longitude, and improve the accuracy of prediction results. .
Smart Images

Figure CN119067259A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the fields of earth science and artificial intelligence technology, and in particular to a method for constructing a reservoir volume sedimentation loss rate prediction model. Background Art
[0002] The reservoir sedimentation rate (RSR) of global reservoirs is significantly location-specific. Traditional reservoir sedimentation rate prediction models can be divided into three categories based on their scope of application:
[0003] 1. Detailed models designed for individual reservoirs have complex structures and numerous parameters. Although they offer high prediction accuracy for specific reservoirs, they are difficult to apply widely.
[0004] 2. The model is applicable to the watershed scale and can accurately assess the sedimentation loss rate within the watershed through appropriate parameter settings;
[0005] 3. The global-scale model has relatively simple parameters and can be used to roughly calculate the sedimentation loss rate of global reservoirs, but the accuracy is relatively low.
[0006] However, the above methods still have limitations in terms of accuracy when obtaining RSR data with high site specificity at the global scale:
[0007] (1) Existing methods struggle to balance global scale and accuracy. Pursuing global model applicability often requires sacrificing location-specific accuracy. Conversely, improving location-specific accuracy makes it difficult to cover all reservoirs globally.
[0008] (2) Existing methods make it difficult to comprehensively consider the impact of the dynamic interactions among factors such as runoff, basin area, storage capacity, and longitude and latitude of reservoirs around the world on RSR, resulting in deviations between the predicted RSR and the actual situation. Summary of the Invention
[0009] The present disclosure provides a method for constructing a reservoir capacity sedimentation loss rate prediction model, which fully considers the impact of reservoir environmental variables, geographical variables and the reservoir's own properties on the reservoir capacity sedimentation loss rate, so that the constructed reservoir capacity sedimentation loss rate prediction model can more accurately reflect the actual global reservoir capacity sedimentation loss rate.
[0010] In a first aspect, the present disclosure provides a method for constructing a reservoir volume sedimentation loss rate prediction model, comprising:
[0011] Acquire several sets of sample data to construct a training data set; each set of sample data includes data values of reservoir characteristic variables and measured values of reservoir sedimentation loss rate. The reservoir characteristic variables include the maximum reservoir capacity, the upstream basin area of the reservoir, the average runoff flow of the reservoir dam, the longitude and latitude of the reservoir dam, and characteristic variables of the empirical model of suspended sediment capture efficiency;
[0012] Randomly sampling the training data set to form a plurality of training sample subsets;
[0013] Recursively partition each training sample subset into regions and construct a binary split decision tree;
[0014] The constructed binary split decision trees are integrated to obtain a random forest model, and the average value of the prediction results of all decision trees in the random forest model is taken as the reservoir volume sedimentation loss rate prediction value of the reservoir volume sedimentation loss rate prediction model.
[0015] In some embodiments, constructing a binary split decision tree comprises the following steps:
[0016] The training sample subset is divided into two regions to obtain two secondary data sets and the optimal reservoir sedimentation loss rate corresponding to each secondary data set;
[0017] Determine whether the regional division cut-off conditions are met;
[0018] If the regional division cutoff condition is not met, the secondary dataset is regionalized again until the regional division cutoff condition is met;
[0019] If the regional division cutoff condition is met, the secondary data sets obtained by regional division and the optimal reservoir sedimentation loss rate corresponding to each secondary data set form a binary split decision tree.
[0020] In some embodiments, the region division includes the following steps:
[0021] Determine the optimal segmentation variables and optimal segmentation points for the target data set to be divided;
[0022] The target data set is segmented using the optimal segmentation variables and optimal segmentation points of the target data set to obtain two secondary data sets of the target data set.
[0023] In some embodiments, the region division cutoff condition includes:
[0024] The total number of layers of regional division of the training sample subset reaches the preset number of layers.
[0025] In some embodiments, the information entropy corresponding to the data set of each divided node in the decision tree is calculated, and the node path with the fastest information entropy decrease rate is used as the optimal path of the decision tree, and the reservoir volume sedimentation loss rate prediction value corresponding to the optimal path of the decision tree is used as the final reservoir volume sedimentation loss rate prediction value of the decision tree.
[0026] In some embodiments, the process of determining the optimal splitting variable and the optimal splitting point includes:
[0027] Determine the segmentation error when each reservoir characteristic variable corresponding to the target data set is used as the segmentation variable;
[0028] The reservoir characteristic variable corresponding to the minimum segmentation error is used as the optimal segmentation variable of the target data set, and the data value corresponding to the optimal segmentation variable is used as the optimal segmentation point of the target data set.
[0029] In some embodiments, the segmentation error when each reservoir characteristic variable is used as a segmentation variable is determined using the following steps:
[0030] The target reservoir characteristic variable of the target data set is used as the segmentation variable, and the data value corresponding to the target reservoir characteristic variable is used as the segmentation point. The target reservoir characteristic variable is the reservoir characteristic variable used as the segmentation variable this time.
[0031] Comparing each data value corresponding to the characteristic variable of the target reservoir in the target data set with the split point;
[0032] The sample data whose data value corresponding to the characteristic variable of the target reservoir is less than or equal to the split point is divided into the first secondary data set, and the sample data whose data value corresponding to the characteristic variable of the reservoir is greater than the split point is divided into the second secondary data set;
[0033] Calculating a first mean square error between an actual value of the reservoir volume sedimentation loss rate corresponding to each set of sample data in the first secondary data set and a first predicted value of the reservoir volume sedimentation loss rate, and calculating a cumulative sum of the first mean square errors corresponding to each set of sample data in the first secondary data set as a first sum value;
[0034] Calculating a second mean square error between a measured value of the reservoir volume sedimentation loss rate corresponding to each set of sample data in the second secondary data set and a second predicted value of the reservoir volume sedimentation loss rate, and calculating a cumulative sum of the second mean square errors corresponding to each set of sample data in the second secondary data set as a second sum value;
[0035] The sum of the first sum and the second sum is calculated as a segmentation error when the target reservoir characteristic variable is used as a segmentation variable.
[0036] In some embodiments, for each secondary data set, the predicted value of reservoir volume sedimentation loss with the smallest mean square error of the measured value of the reservoir volume sedimentation loss rate corresponding to each sample data in the secondary data set is used as the optimal reservoir volume sedimentation loss rate corresponding to the secondary data set.
[0037] In a second aspect, the present disclosure provides a reservoir capacity sedimentation loss rate prediction model, including a reservoir capacity sedimentation loss rate prediction model constructed by the reservoir capacity sedimentation loss rate prediction model construction method as described in the first aspect.
[0038] In a third aspect, the present disclosure provides a method for predicting reservoir volume sedimentation loss rate, comprising:
[0039] Obtain data values of reservoir characteristic variables of the target reservoir;
[0040] The data values of the reservoir characteristic variables of the target reservoir are input into the reservoir capacity sedimentation loss rate prediction model as described in the second aspect to perform prediction, thereby obtaining a predicted value of the reservoir capacity sedimentation loss rate of the target reservoir.
[0041] In a fourth aspect, a computer device comprises a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described in the first aspect or the third aspect.
[0042] In a fifth aspect, the present disclosure provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in the first aspect or the third aspect.
[0043] In a sixth aspect, the present disclosure provides a computer program product, comprising a computer program / instructions, which, when executed by a processor, implements the steps of the method described in the first aspect or the third aspect.
[0044] The present disclosure provides a method for constructing a reservoir volume sedimentation loss rate prediction model, which has the following beneficial effects:
[0045] (1) Compared with existing empirical models of suspended sediment capture efficiency such as Model 1 and Model 2, the reservoir sedimentation loss rate prediction model constructed in this paper achieves higher prediction accuracy at the global scale.
[0046] (2) This disclosure comprehensively considers the impact of the dynamic interaction between factors such as reservoir capacity, location, runoff and basin area on RSR, so that the predicted RSR is more consistent with the actual situation. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] Hereinafter, the present disclosure will be described in more detail based on embodiments and with reference to the accompanying drawings:
[0048] Figure 1 A schematic flow chart of a method for constructing a reservoir volume sedimentation loss rate prediction model provided in an embodiment of the present disclosure;
[0049] Figure 2 A schematic diagram of a process for constructing a binary split decision tree according to an embodiment of the present disclosure;
[0050] Figure 3 A schematic diagram of a process for determining segmentation errors according to an embodiment of the present disclosure;
[0051] Figure 4 A schematic diagram of a random forest model provided in an embodiment of the present disclosure.
[0052] In the drawings, like components are given like reference numerals, and the drawings are not drawn to scale. DETAILED DESCRIPTION
[0053] In order to enable those skilled in the art to better understand the technical solutions of the present disclosure, and to fully understand and implement how the present disclosure applies technical means to solve technical problems and achieve the corresponding technical effects, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, not all of the embodiments. The embodiments of the present disclosure and the various features in the embodiments can be combined with each other without conflict, and the technical solutions formed are all within the scope of protection of the present disclosure. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of the present disclosure.
[0054] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0055] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.
[0056] Globally, reservoirs supply approximately 30%-40% of surface water irrigation needs, contributed 14.3% of global electricity generation through hydropower in 2023, and serve a variety of beneficial purposes, including flood control and freshwater supply. Global water use has increased sixfold over the past 100 years and continues to grow at approximately 1% per year, driven by factors such as population growth, economic development, and shifting consumption patterns. However, the construction rate of large dams has declined sharply over the past 40 years, and this trend continues. At the same time, the water storage benefits of reservoirs have also been declining. Therefore, against the backdrop of increasing water demand and declining large dam construction, the remaining service life and sustainability management of existing dams are particularly important. The lifespan and sustainability of reservoirs are primarily affected by sediment deposition.
[0057] Sedimentation is causing reservoirs to lose storage capacity faster than anticipated. For example, in China, sedimentation has already caused a 66% loss in reservoir storage capacity. In the United States, many large reservoirs with a design life of 150-200 years have had their lifespans shortened by 50-100 years due to sedimentation. The reservoir storage rate (RSR) of reservoirs worldwide is highly site-specific. Traditional reservoir storage rate prediction models can be divided into three categories based on their scope of application:
[0058] 1. Detailed models designed for individual reservoirs have complex structures and numerous parameters. Although they offer high prediction accuracy for specific reservoirs, they are difficult to apply widely.
[0059] 2. The model is applicable to the watershed scale and can accurately assess the sedimentation loss rate within the watershed through appropriate parameter settings;
[0060] 3. The global-scale model has relatively simple parameters and can be used to roughly calculate the sedimentation loss rate of global reservoirs, but the accuracy is relatively low.
[0061] However, the above methods still have limitations in terms of accuracy when obtaining RSR data with high site specificity at the global scale:
[0062] (1) Existing methods struggle to balance global scale and accuracy. Pursuing global model applicability often requires sacrificing location-specific accuracy. Conversely, improving location-specific accuracy makes it difficult to cover all reservoirs globally.
[0063] (2) Existing methods make it difficult to comprehensively consider the impact of the dynamic interactions among factors such as runoff, basin area, storage capacity, and longitude and latitude of reservoirs around the world on RSR, resulting in deviations between the predicted RSR and the actual situation.
[0064] Example 1
[0065] Figure 1 This is a flow chart of a method for constructing a reservoir volume sedimentation loss rate prediction model provided by an embodiment of the present disclosure. Figure 1 As shown, a method for constructing a reservoir capacity sedimentation loss rate prediction model includes:
[0066] Step S1, obtain several groups of sample data and construct a training data set; each group of sample data includes data values of reservoir characteristic variables and measured values of reservoir sedimentation loss rate RSR, the reservoir characteristic variables include the maximum storage capacity C of the reservoir, the upstream basin area A of the reservoir, the average runoff flow of the reservoir dam Q, the longitude Lon of the reservoir dam, the latitude Lat of the reservoir dam, and the characteristic variables M1 and M2 of the empirical model of suspended sediment capture efficiency.
[0067] Specifically, the data values of the reservoir characteristic variables can be obtained from multiple databases, such as the Global Reservoir and Dam Database (GRanD). The measured value of the reservoir capacity sedimentation loss rate can be obtained from multiple data sources, such as the Reservoir Sedimentation Survey Information System database (RESIS).
[0068] The suspended sediment capture efficiency empirical model includes a first suspended sediment capture efficiency empirical model Model 1 shown in the following formula 1 and a second suspended sediment capture efficiency empirical model Model 2 shown in the following formula 2, as shown below:
[0069]
[0070] SSC=β0+β1log 10 Q (3)
[0071] RSR=SSC*TE (4)
[0072] Where TE is the suspended sediment capture efficiency (%); SSC is the suspended sediment flux (kg / m 3 ); β0 and β1 are empirical coefficients; C is the maximum storage capacity of the reservoir (km 3 ); Q is the average runoff of the reservoir dam (m 3 / s), upstream watershed area of reservoir A (km 2 ), D is a constant with a value range between 0.046 and 1, and the value of this model is 0.1.
[0073] In formula 1, Named as the characteristic variable M1 of the first suspended sediment capture efficiency empirical model; replace Named as the characteristic variable M2 of the second suspended sediment capture efficiency empirical model.
[0074] Step S2: Randomly sample the training data set to form several training sample subsets.
[0075] Step S3, recursively partition each training sample subset into regions and construct a binary split decision tree. Specifically, Figure 2 As shown, step S3 includes:
[0076] Step S31, dividing the training sample subset into regions to obtain two secondary data sets and the optimal reservoir volume sedimentation loss rate corresponding to each secondary data set;
[0077] Step S32: determine whether the area division cutoff condition is met.
[0078] Specifically, the region division cutoff condition includes: the total number of layers of region division of the training sample subset reaches a preset number of layers, that is, the total number of layers of the decision tree growing downward reaches a preset number of layers.
[0079] Step S33: If the region division cutoff condition is not met, the secondary data set is divided into regions again until the region division cutoff condition is met;
[0080] In step S34, if the regional division cutoff condition is met, a binary split decision tree is formed based on the secondary data sets obtained from the regional division and the optimal reservoir volume siltation loss rate corresponding to each secondary data set. Specifically, for each secondary data set, the predicted reservoir volume siltation loss rate with the minimum mean square error (MSE) with the measured reservoir volume siltation loss rate corresponding to each sample data in the secondary data set is used as the optimal reservoir volume siltation loss rate for that secondary data set.
[0081] Step S4: Integrate the constructed binary split decision trees to obtain a random forest model, and take the average of the prediction results of all decision trees in the random forest model as the reservoir volume sedimentation loss rate prediction value output by the reservoir volume sedimentation loss rate prediction model. The prediction result of the decision tree is the final reservoir volume sedimentation loss rate prediction value of the decision tree.
[0082] In one embodiment, the region division includes the following steps:
[0083] Determining optimal segmentation variables and optimal segmentation points of a target data set, wherein the target data set is a data set to be divided into regions;
[0084] The target data set is segmented using the optimal segmentation variables and optimal segmentation points of the target data set, that is, a binary split of a decision tree is performed to obtain two secondary data sets of the target data set.
[0085] The process of determining the optimal splitting variables and the optimal splitting points includes:
[0086] Determine the segmentation error when each reservoir characteristic variable corresponding to the target data set is used as the segmentation variable;
[0087] The reservoir characteristic variable corresponding to the minimum segmentation error is used as the optimal segmentation variable of the target data set, and the data value corresponding to the optimal segmentation variable is used as the optimal segmentation point of the target data set.
[0088] Among them, for each reservoir characteristic variable as the segmentation variable, the segmentation error is as follows: Figure 3 As shown, the following steps are used to determine:
[0089] The target reservoir characteristic variable of the target data set is used as the segmentation variable, and the data value corresponding to the target reservoir characteristic variable is used as the segmentation point. The target reservoir characteristic variable is the reservoir characteristic variable used as the segmentation variable this time.
[0090] Comparing each data value corresponding to the characteristic variable of the target reservoir in the target data set with the split point;
[0091] The sample data whose data value corresponding to the characteristic variable of the target reservoir is less than or equal to the split point is divided into the first secondary data set, and the sample data whose data value corresponding to the characteristic variable of the reservoir is greater than the split point is divided into the second secondary data set;
[0092] Calculating a first mean square error between an actual measured value of the reservoir volume sedimentation loss rate corresponding to each set of sample data in the first secondary data set and a preset first reservoir volume sedimentation loss rate prediction value, and calculating a cumulative sum of the first mean square errors corresponding to each set of sample data in the first secondary data set as a first sum value;
[0093] Calculating a second mean square error between an actual measured value of the reservoir volume sedimentation loss rate corresponding to each set of sample data in the second secondary data set and a preset second reservoir volume sedimentation loss rate prediction value, and calculating a cumulative sum of the second mean square errors corresponding to each set of sample data in the second secondary data set as a second sum value;
[0094] The sum of the first sum and the second sum is calculated as a segmentation error when the target reservoir characteristic variable is used as a segmentation variable.
[0095] In some embodiments, after step S34, it also includes: calculating the information entropy corresponding to the data set of each divided node in the decision tree, taking the node path with the fastest information entropy decrease rate as the optimal path of the decision tree, and taking the reservoir volume sedimentation loss rate prediction value corresponding to the optimal path of the decision tree as the final reservoir volume sedimentation loss rate prediction value of the decision tree.
[0096] For example, several sets of sample data are obtained to construct a training data set. Random sampling is performed on the training data set to generate k training sample subsets. Recursive region division is performed on the k training sample subsets to construct Figure 4 The decision tree for the binary split shown.
[0097] The decision tree construction process includes: partitioning the training sample subset into regions to obtain two secondary datasets and the optimal reservoir siltation loss rate corresponding to each secondary dataset. Each regional partitioning step is equivalent to a binary split of the training sample subset. For each target dataset to be partitioned, each regional partitioning step produces two secondary datasets. A determination is then made as to whether the regional partitioning cutoff condition is met: that is, whether the total number of regional partitioning levels in the training sample subset reaches a preset 17 levels. If not, regional partitioning of the secondary datasets is continued until the cutoff condition is met. If the cutoff condition is met, regional partitioning is stopped. The secondary datasets obtained from the regional partitioning and the optimal reservoir siltation loss rate corresponding to each secondary dataset form a binary split decision tree. The information entropy corresponding to the datasets at each partitioning node in the decision tree is calculated. The node path with the fastest information entropy decrease is selected as the optimal path of the decision tree. The predicted reservoir siltation loss rate corresponding to the optimal path of the decision tree is used as the final predicted reservoir siltation loss rate of the decision tree, i.e., the prediction result of the decision tree.
[0098] Among them, the regional division process is illustrated by an example. It is assumed that the target data set includes sample data X1 and sample data X2, where X1 = (C = C1, A = A1, Q = Q1, Lon = Lon1, Lat = Lat1, M1 = M11, M2 = M21, RSR1), X2 = (C = C2, A = A2, Q = Q2, Lon = Lon2, Lat = Lat2, M1 = M12, M2 = M22, RSR2), C, A, Q, Lon, Lat, M1, M2 are the seven reservoir characteristic variables corresponding to the target data set, RSR1 represents the reservoir capacity siltation loss rate corresponding to the sample data X1, and RSR2 represents the reservoir capacity siltation loss rate corresponding to the sample data X2. Taking C as the splitting variable as an example, and the data value corresponding to C as C1 as the splitting point, the sample data in the target data set with C less than or equal to C1 are divided into the first secondary data set under the reservoir characteristic variable C, and the sample data with C greater than or equal to C1 are divided into the second secondary data set under the reservoir characteristic variable C.
[0099] Then use Formula 5 to determine the optimal split variable:
[0100]
[0101] Among them, K represents the segmentation error, j represents the segmentation variable, s represents the segmentation point corresponding to the segmentation variable j, R1 and R2 represent the secondary data sets obtained by segmenting the target data set using the segmentation point s, and x i is the i-th group of sample data in the secondary data set; y i is x i The measured value of RSR; R1 and R2 represent the first secondary data set and the second secondary data set respectively obtained by the cutting point s; v1 and v2 represent the predicted value of the first reservoir volume sedimentation loss rate corresponding to R1 and the predicted value of the second reservoir volume sedimentation loss rate corresponding to R2 respectively.
[0102] Traverse the segmentation variable j and segmentation point s, and select the segmentation variable j that makes formula (5) reach the minimum segmentation error best and the split point s best .
[0103] After determining the optimal segmentation variable j best and the optimal split point s best After that, the region can be divided by formulas (6) and (7) to obtain the secondary data set and the optimal reservoir volume sedimentation loss rate corresponding to the secondary data set:
[0104]
[0105] where Θ m is to use the optimal segmentation variable j best and the optimal split point s best The optimal reservoir sedimentation loss rate corresponding to the mth secondary data set obtained by dividing the area;
[0106] The two secondary data sets are divided into regions again, and each time a region is divided, a node is generated until the given random parameter set P reaches the maximum depth. In this embodiment, the depth is set to 17. Generate a decision tree with binary split
[0107]
[0108] Where P is a given set of random parameters, I is the conditional probability; N is the subset of training samples based on the conditional probability. The number of regions in which the input space is divided, that is, the number of secondary datasets.
[0109] Information entropy is a quantitative indicator of the information content of a system. The decision tree model uses information entropy to evaluate the secondary data set x of RSR. i The average amount of information, information entropy is calculated as follows:
[0110]
[0111] Among them, x represents the secondary data set divided by the decision tree through the same node; p(x i ) represents the occurrence of the i-th reservoir characteristic variable x in the secondary dataset. i The probability of the data value. n represents the training sample subset The number of secondary data sets obtained by partitioning.
[0112] The node with the fastest information entropy decrease rate in each layer of the decision tree is selected as the optimal path of the decision tree, and the predicted value of the reservoir sedimentation loss rate corresponding to the optimal path of the decision tree is used as the predicted value of the reservoir sedimentation loss rate of the decision tree. Figure 4 In each decision tree, the node path with darker color is the node path with the fastest information entropy decrease rate, and the predicted value of reservoir capacity sedimentation loss rate corresponding to this path is used as the predicted value of reservoir capacity sedimentation loss rate of the decision tree.
[0113] The resulting decision trees are combined to form a summary regression estimate, as shown in Equation 10:
[0114]
[0115] Among them, E U Refers to the given random parameter set P and data set The expectation of ; U is the output of the independent and identically distributed random variables.
[0116] Each decision tree built through the above steps becomes an expert at predicting the RSR of a particular reservoir type. Reservoir types are categorized based on geographic variables, environmental variables, intrinsic attribute variables, and empirical model variables. These decision trees form a random forest model. Given new input data, the random forest model analyzes it from multiple perspectives. Ultimately, the expert group votes collectively, and the average of the group's votes is used as the final predicted RSR.
[0117] The random forest model consists of multiple decision trees, each constructed from a randomly sampled subset of training samples. Unlike traditional methods, random forests not only sample training samples but also randomly sample the components of the feature vectors. When training the decision trees, only a subset of the sampled feature components are used as candidate features at each split. This mechanism enables random forests to fully account for the variability of individual reservoir variables, effectively simulating the impact of dynamic interactions among factors such as runoff, catchment area, reservoir capacity, and latitude and longitude on reservoir sedimentation loss rates across reservoirs worldwide, making the predicted RSR more consistent with actual conditions.
[0118] The method for constructing a reservoir volume sedimentation loss rate prediction model provided in this embodiment has the following beneficial effects:
[0119] (1) Compared with existing empirical models of suspended sediment capture efficiency, such as Model 1 and Model 2, the reservoir sedimentation loss rate prediction model constructed in this embodiment achieves higher prediction accuracy at the global scale.
[0120] (2) The constructed reservoir sedimentation loss rate prediction model can better simulate the impact of the dynamic interaction between factors such as runoff, basin area, reservoir capacity, longitude and latitude of various reservoirs around the world on RSR, so that the predicted RSR is more consistent with the actual situation.
[0121] Example 2
[0122] On the basis of the above embodiment, this embodiment provides a reservoir capacity sedimentation loss rate prediction model constructed by the reservoir capacity sedimentation loss rate prediction model construction method described in the above embodiment.
[0123] The reservoir sedimentation loss rate prediction model provided in this embodiment can better simulate the impact of the dynamic interaction between factors such as runoff, basin area, reservoir capacity, longitude and latitude of various reservoirs around the world on RSR, thereby making the predicted RSR more consistent with the actual situation.
[0124] Example 3
[0125] Based on the above embodiment, this embodiment provides a method for predicting reservoir volume sedimentation loss rate, including:
[0126] Obtain data values of reservoir characteristic variables of the target reservoir;
[0127] The data values of the reservoir characteristic variables of the target reservoir are input into the reservoir capacity sedimentation loss rate prediction model as described in the above embodiment to perform prediction, thereby obtaining a predicted value of the reservoir capacity sedimentation loss rate of the target reservoir.
[0128] This embodiment provides a method for predicting reservoir volume sedimentation loss rate, which has the following beneficial effects:
[0129] (1) Compared with the existing empirical models of suspended sediment capture efficiency such as Model 1 and Model 2, this embodiment uses the reservoir sedimentation loss rate prediction model constructed in the above embodiment to perform predictions, achieving higher prediction accuracy on a global scale.
[0130] (2) The reservoir sedimentation loss rate prediction model constructed by the above embodiment in this embodiment can better simulate the impact of the dynamic interaction between factors such as runoff, basin area, reservoir capacity, longitude and latitude of various reservoirs around the world on RSR, thereby making the predicted RSR more consistent with the actual situation.
[0131] Example 4
[0132] Based on the above embodiments, this embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described in the above embodiments.
[0133] In some implementations of this embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described in the above embodiment are implemented.
[0134] In some implementations of this embodiment, a computer program product is provided, including a computer program / instructions, which implements the steps of the method described in the above embodiment when executed by a processor.
[0135] The processor may include, but is not limited to, one or more processors or microprocessors. Each processor may be an application specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field programmable gate array (FPGA), a controller, a microcontroller, a microprocessor, or other electronic components to execute the methods in the above embodiments.
[0136] The computer-readable storage medium may be implemented by any type of volatile or non-volatile storage device or a combination thereof. The computer-readable storage medium may include, but is not limited to, for example, random access memory (RAM), read-only memory (ROM), flash memory, EPROM memory, EEPROM memory, registers, computer storage media (such as hard disks, floppy disks, solid-state drives, removable disks, CD-ROMs, DVD-ROMs, Blu-ray discs, etc.).
[0137] The computer-readable storage medium may also store at least one computer-executable program / instruction, such as a computer-readable instruction. Computer-readable storage media include, but are not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory. Computer-readable storage media may include, for example, read-only memory (ROM), a hard disk, a flash memory, etc. For example, a non-transitory computer-readable storage medium may be connected to a computing device such as a computer, and then, when the computing device executes the computer-readable instructions stored on the computer-readable storage medium, the various methods described above may be performed.
[0138] In addition, the computer device may also include (but is not limited to) a data bus, an input / output (I / O) bus, a display, and input / output devices (eg, keyboard, mouse, speaker, etc.).
[0139] The processor can communicate with external devices via an I / O bus via a wired or wireless network.
[0140] In one embodiment, the at least one computer executable instruction may also be compiled into or constitute a software product / computer program product, wherein one or more computer executable instructions are executed by a processor to perform the various functions and / or method steps in the embodiments described in the present technology.
[0141] In the embodiments provided in the present disclosure, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architectures, functions and operations of the devices, methods and computer program products according to multiple embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, program segment or part of the code, and the above-mentioned module, program segment or part of the code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, as well as the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions.
[0142] It should be noted that, in this disclosure, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element limited by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.
[0143] Although the embodiments disclosed in this disclosure are as described above, the above contents are merely embodiments adopted to facilitate understanding of the disclosure and are not intended to limit the disclosure. Any person skilled in the art of the disclosure may make any modifications and changes in the form and details of the implementation without departing from the spirit and scope of the disclosure. However, the scope of patent protection of the disclosure shall still be based on the scope defined by the attached claims.
Claims
1. A method for constructing a reservoir capacity siltation loss rate prediction model, characterized in that: include: Obtain several groups of sample data and construct a training data set; Each set of sample data includes data values of reservoir characteristic variables and measured values of reservoir siltation loss rate. The reservoir characteristic variables include the maximum reservoir capacity, the upstream basin area of the reservoir, the average runoff flow of the reservoir dam, the longitude and latitude of the reservoir dam, and the characteristic variables of the empirical model of suspended sediment capture efficiency; Randomly sampling the training data set to form a number of training sample subsets; Recursively partition each training sample subset into different regions and construct a binary split decision tree. The constructed binary split decision trees are integrated to obtain a random forest model, and the average value of the prediction results of all decision trees in the random forest model is taken as the predicted value of the reservoir capacity siltation loss rate output by the reservoir capacity siltation loss rate prediction model.
2. The method for constructing a reservoir capacity siltation loss rate prediction model according to claim 1, characterized in that: Constructing a decision tree for binary splitting involves the following steps: The training sample subsets are divided into regions to obtain two secondary data sets and the optimal reservoir siltation loss rate corresponding to each secondary data set; Determine whether the regional division cut-off condition is met; If the regional division cutoff condition is not met, the secondary data set is divided into regions again until the regional division cutoff condition is met; If the regional division cutoff condition is met, the secondary data sets obtained by the regional division and the optimal reservoir sedimentation loss rate corresponding to each secondary data set form a binary split decision tree.
3. The method for constructing a reservoir capacity sedimentation loss rate prediction model according to claim 2, characterized in that: The regional division includes the following steps: Determine the optimal segmentation variable and optimal segmentation point of the target data set to be divided; The target data set is segmented using the optimal segmentation variable and the optimal segmentation point of the target data set to obtain two secondary data sets of the target data set.
4. The method for constructing a reservoir capacity sedimentation loss rate prediction model according to claim 2, characterized in that: The regional division cut-off conditions include: The total number of layers of regional division of the training sample subset reaches the preset number of layers.
5. The method for constructing a reservoir capacity sedimentation loss rate prediction model according to claim 2, characterized in that: The information entropy corresponding to the data set of each divided node in the decision tree is calculated, and the node path with the fastest information entropy decrease rate is taken as the optimal path of the decision tree, and the reservoir capacity sedimentation loss rate prediction value corresponding to the optimal path of the decision tree is taken as the final reservoir capacity sedimentation loss rate prediction value of the decision tree.
6. The method for constructing a reservoir capacity sedimentation loss rate prediction model according to claim 3, characterized in that: The process of determining the optimal segmentation variable and the optimal segmentation point includes: Determine the segmentation error when each reservoir characteristic variable corresponding to the target data set is used as the segmentation variable; The reservoir characteristic variable corresponding to the minimum segmentation error is used as the optimal segmentation variable of the target data set, and the data value corresponding to the optimal segmentation variable is used as the optimal segmentation point of the target data set.
7. The method for constructing a reservoir capacity sedimentation loss rate prediction model according to claim 6, characterized in that: The segmentation error when each reservoir characteristic variable is used as a segmentation variable is determined using the following steps: The target reservoir characteristic variable of the target data set is used as a segmentation variable, and the data value corresponding to the target reservoir characteristic variable is used as a segmentation point, and the target reservoir characteristic variable is the reservoir characteristic variable used as the segmentation variable this time; Compare each data value corresponding to the characteristic variable of the target reservoir in the target data set with the split point; The sample data whose data value corresponding to the characteristic variable of the target reservoir is less than or equal to the split point is divided into the first secondary data set, and the sample data whose data value corresponding to the characteristic variable of the reservoir is greater than the split point is divided into the second secondary data set; Calculate the first mean square error between the measured value of the reservoir volume sedimentation loss rate corresponding to each group of sample data in the first secondary data set and the preset first reservoir volume sedimentation loss rate prediction value, and calculate the cumulative sum of the first mean square errors corresponding to each group of sample data in the first secondary data set as the first sum value; Calculate the second mean square error between the measured value of the reservoir volume sedimentation loss rate corresponding to each group of sample data in the second secondary data set and the preset second reservoir volume sedimentation loss rate predicted value, and calculate the cumulative sum of the second mean square errors corresponding to each group of sample data in the second secondary data set as the second sum value; The sum of the first sum and the second sum is calculated as a segmentation error when the target reservoir characteristic variable is used as a segmentation variable.
8. The method for constructing a reservoir capacity sedimentation loss rate prediction model according to claim 2, characterized in that: For each secondary data set, the predicted value of the reservoir sedimentation loss rate with the smallest mean square error of the measured value of the reservoir sedimentation loss rate corresponding to each sample data in the secondary data set is taken as the optimal reservoir sedimentation loss rate corresponding to the secondary data set.
9. A reservoir sedimentation loss rate prediction model, characterized in that: It comprises a reservoir capacity sedimentation loss rate prediction model constructed by the reservoir capacity sedimentation loss rate prediction model construction method described in any one of claims 1 to 8.
10. A method for predicting reservoir volume sedimentation loss rate, characterized in that: include: Obtain data values of reservoir characteristic variables of the target reservoir; The data value of the reservoir characteristic variable of the target reservoir is input into the reservoir capacity sedimentation loss rate prediction model as claimed in claim 9 for prediction, so as to obtain the predicted value of the reservoir capacity sedimentation loss rate of the target reservoir.
Citation Information
Patent Citations
Northwest inland river ecological base flow prediction method based on random forest
CN111445080A
Reservoir sediment deposition predicting and forecasting method
CN113591411A
Reservoir ecological desilting scheme decision-making method based on multi-target constraint
CN114818339A
Bayesian pattern averaging and machine learning-based reservoir storage sediment probability forecasting method
CN115062818A
Multi-dimensional risk decision rule generation method and device
CN117333024A
