River channel safety early warning method and system based on BIM and multi-model information fusion

Abstract

The invention discloses a river channel safety early warning method and system based on BIM and multi-model information fusion, and the method comprises the steps: obtaining the terrain environment data and historical flow data of an early warning target river channel, and building a rainfall flow model based on the BIM technology; performing regional division on the upstream of a target river channel according to the rainfall flow model, and setting a flow alarm threshold value of an early warning section; acquiring continuous flow data of the early warning section of the upstream site of the target river channel as a training data set and a test data set; completing the calibration of the deep neural network model and the Elman neural network model; performing multi-model information fusion on the two models by using a random forest algorithm based on a Stacking algorithm as a framework to complete construction of a traffic prediction combination model; and inputting real-time data of an upstream station into the flow prediction combination model to generate flow prediction data of a future early warning section, and importing the flow prediction data into the rainfall flow model to realize visualization and generate early warning information.

Description

Technical Field

[0001] The present invention relates to the technical field of river emergency warning, and in particular to a river safety warning method and system based on BIM and multi-model information fusion. Background Art

[0002] As an important part of the urban ecosystem, rivers not only carry important functions such as urban drainage and flood control, but their health also directly affects the quality of life of urban residents and the sustainable development of cities. With the acceleration of urbanization, river management and protection are facing more and more challenges, such as erosion, siltation, and extreme weather events caused by climate change, which may lead to drastic changes in river water levels and flow rates, and then cause natural disasters such as floods.

[0003] In the existing technology, the river safety early warning method mainly relies on traditional monitoring and forecasting means, such as setting up water level monitoring stations to monitor the changes in river water levels in real time, or using weather forecasts to predict possible rainfall. These methods can provide early warning information for river safety to a certain extent, but there are often problems such as delayed warning time and insufficient warning accuracy. In addition, the existing technology mostly adopts a single monitoring or forecasting model, lacks comprehensive analysis and utilization of multi-source data, and it is difficult to achieve a comprehensive assessment of the safety status of the river.

[0004] The shortcomings of existing technologies are mainly manifested in the following aspects: first, the real-time and accuracy of early warnings are insufficient to meet the needs of the rapidly changing river environment; second, early warning methods mostly rely on experience and judgment, lack of scientific data-driven analysis, resulting in low reliability of early warning results; third, existing early warning systems often only focus on a single factor, such as water level or rainfall, ignoring the complexity of the river environment and the impact of multiple factors. The existence of these problems limits the application effect of river safety early warning systems and increases the difficulty of river management and protection. Summary of the invention

[0005] In view of the above defects or improvement needs of the prior art, the purpose of the present invention is to provide a river safety early warning method based on BIM and multi-model information fusion. The present invention adopts a multi-model information fusion method in combination with BIM technology, and uses a data-driven approach for simulation and early warning. Compared with the process-driven model, it does not need to determine complex physical parameters, and has a better fitting effect on nonlinear functions. It can use more complex upstream site data to predict the downstream early warning section flow, which can effectively improve the early warning accuracy and realize the visualization of early warning information; and the present invention uses a variety of neural network models and performs information fusion through the Stacking algorithm framework to complete the construction of a flow prediction combined model. Compared with a single model, the combined model combines the advantages of each model, makes up for the shortcomings of a single model, and effectively improves the accuracy of the early warning.

[0006] To achieve the above, according to one aspect of the present invention, a river safety early warning method based on BIM and multi-model information fusion is proposed, comprising:

[0007] S100, obtaining terrain and environmental data of the target river channel for early warning and historical flow data, and establishing a precipitation flow model based on the terrain and environmental data of the target river channel and the historical flow data based on BIM technology;

[0008] S200, dividing the upstream of the target river into regions according to the precipitation flow model, and setting a flow alarm threshold of the early warning section;

[0009] S300, obtaining continuous flow data of the warning section of the upstream station of the target river as a training data set and a test data set;

[0010] S400, selecting historical flow data of the warning section of the upstream station of the target river as candidate factors, and screening the candidate factors using the grey correlation method to obtain prediction factors;

[0011] S500, inputting the prediction factors into the deep neural network model and the Elman neural network model, and using the test data set to perform tests to complete model calibration;

[0012] S600, using the random forest algorithm based on the Stacking algorithm as a framework to fuse the multi-model information of the two models and complete the construction of the traffic prediction combination model;

[0013] S700, input the real-time data of the upstream site into the flow prediction combination model to generate flow prediction data for the future warning section, and import the flow prediction data into the precipitation flow model to realize visualization and generate warning information.

[0014] Furthermore, the target river terrain environment data in step S100 includes the roughness, water depth, river section length and cross-sectional data of the target river.

[0015] Furthermore, in step S300, the training data set is longer than the test data set.

[0016] Furthermore, step S400 includes: selecting the flow data of the previous 1 to 12 days of the warning section of the upstream station during the training period and the flow data of the same period of the previous year as alternative factors, selecting the warning section inspection period data as a reference object, using grey correlation analysis to output the correlation between each alternative factor and the reference object, and selecting multiple alternative factors as forecasting factors according to the correlation from high to low.

[0017] Furthermore, step S400 includes: adjusting the number of prediction factors according to the degree of correlation to avoid using prediction factors with low correlation; if the correlations are all low, reselecting candidate factors for screening.

[0018] Further, step S600 includes:

[0019] S601, the training data set is divided into three parts, two of which are used alternately as training data to input into the deep neural network model for simulation prediction, and the third fold data is obtained. After three cross-checks, three groups of data are obtained, which are recorded as the first data set; the Elman neural network model is also simulated and predicted, which is recorded as the second data set;

[0020] S602, using the same training data for the two models to simulate and predict the test period data set, and also after three cross-validations, three sets of data are obtained respectively, which are recorded as the third data set and the fourth data set;

[0021] S603, stacking the first data set and the second data set vertically in the order of the original data to obtain two sets of data, denoted as A1 and A2; taking the average of the third data set and the fourth data set to obtain two sets of data, denoted as B1 and B2;

[0022] S604. Integrate the meta-model based on the random forest algorithm model, use A1 and A2 as the training period data of the combination model, and use B1 and B2 as the test period data of the combination model. Determine the relevant parameters through grid search and adjust them on this basis to meet the set certainty coefficient requirements, and complete the construction of the traffic prediction combination model.

[0023] According to a second aspect of the present invention, a river safety early warning system based on BIM and multi-model information fusion is provided, comprising:

[0024] The first module is used to obtain the terrain and environmental data of the warning target river and the historical flow data, and establish a precipitation flow model based on the terrain and environmental data of the target river and the historical flow data based on the BIM technology; divide the upstream of the target river into regions according to the precipitation flow model, and set the flow alarm threshold of the warning section; obtain the continuous flow data of the warning section of the upstream station of the target river as the training data set and the test data set;

[0025] The second module is used to select the historical flow data of the warning section of the upstream station of the target river as the candidate factor, and use the grey correlation method to screen the candidate factors to obtain the prediction factor; input the prediction factor into the deep neural network model and the Elman neural network model, and use the test data set to test, compare the certainty coefficient, adjust the number of nodes in each layer, the learning rate, the number of iterations and the expected error threshold to achieve the required accuracy and complete the model calibration; use the random forest algorithm based on the Stacking algorithm as the framework to perform multi-model information fusion on the two models to complete the construction of the flow prediction combination model;

[0026] The third module is used to input the real-time data of the upstream site into the flow prediction combination model to generate flow prediction data for the future warning section, and import the flow prediction data into the precipitation flow combination model to realize visualization and generate warning information.

[0027] According to a third aspect of the present invention, there is provided an electronic device, comprising:

[0028] At least one processor, at least one memory and a communication interface; wherein,

[0029] The processor, memory and communication interface communicate with each other;

[0030] The memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute any step of the above-mentioned river safety early warning method based on BIM and multi-model information fusion.

[0031] According to a fourth aspect of the present invention, a non-transitory computer-readable storage medium is provided, characterized in that the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions enable the computer to execute any step in the above-mentioned river safety warning method based on BIM and multi-model information fusion.

[0032] Beneficial effects of the present invention:

[0033] 1. The present invention adopts a multi-model information fusion method combined with BIM technology, and uses a data-driven approach for simulation and early warning. Compared with the process-driven model, it does not need to determine complex physical parameters, and has a better fitting effect on nonlinear functions. It can use more complex upstream site data to predict the downstream early warning section flow, which can effectively improve the early warning accuracy and realize the visualization of early warning information.

[0034] 2. The present invention adopts a multi-model information fusion method to realize the screening of the most relevant prediction factors, uses a variety of neural network models and performs information fusion through the Stacking algorithm framework to complete the construction of a traffic prediction combination model. Compared with a single model, the combined model integrates the advantages of each model, makes up for the shortcomings of a single model, and effectively improves the accuracy of the early warning.

[0035] Additional aspects and advantages of the present application will be partially given in the following description, which will become apparent from the following description, or will be understood through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0037] Figure 1 It is a flow chart of a river safety early warning method based on BIM and multi-model information fusion in an embodiment of the present invention;

[0038] Figure 2 It is a specific flow chart of step S600 in a river safety early warning method based on BIM and multi-model information fusion in an embodiment of the present invention;

[0039] Figure 3 Schematic diagram of K-fold cross-validation in step S600 in an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0041] Those skilled in the art will appreciate that, unless otherwise stated, the singular forms "a", "an", "said" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0042] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as the general understanding of those skilled in the art to which this application belongs. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and will not be interpreted with an idealized or overly formal meaning unless specifically defined as in the embodiments of this application.

[0043] The purpose of the present invention is to provide a river safety early warning method based on BIM and multi-model information fusion. The present invention adopts a multi-model information fusion method combined with BIM technology, and uses a data-driven approach to perform simulation and early warning. Compared with process-driven models, it does not need to determine complex physical parameters, and has a better fitting effect on nonlinear functions. It can use more complex upstream site data to predict the flow of downstream early warning sections, which can effectively improve the accuracy of early warning and realize the visualization of early warning information; and the present invention uses a variety of neural network models and performs information fusion through the Stacking algorithm framework to complete the construction of a flow prediction combined model. Compared with a single model, the combined model combines the advantages of each model, makes up for the shortcomings of a single model, and effectively improves the accuracy of early warning.

[0044] Embodiment 1:

[0045] like Figure 1 As shown, an embodiment of the present invention provides a river safety early warning method based on BIM and multi-model information fusion, including:

[0046] S100, obtaining terrain and environmental data of the target river channel for early warning and historical flow data, and establishing a precipitation flow model based on the terrain and environmental data of the target river channel and the historical flow data based on BIM technology;

[0047] Specifically, high-precision measurement equipment, such as the LiDAR (Light Detection and Ranging) system carried by drones, is used to collect river terrain data, including but not limited to river roughness, water depth, river section length, cross-section and other basic information. In addition, historical flow data is obtained through cooperation with local water conservancy departments. Then, using BIM software, combined with the collected terrain environment data and historical flow data, a three-dimensional river model is established. This model can simulate the dynamics of water flow under different rainfall conditions.

[0048] S200, dividing the upstream of the target river into regions according to the precipitation flow model, and setting a flow alarm threshold of the early warning section;

[0049] S300, obtaining continuous flow data of the warning section of the upstream station of the target river as a training data set and a test data set;

[0050] Specifically, this embodiment divides the continuous flow data of the warning section of the upstream station of the target river into a data set according to a ratio of 7:1, with the majority being the training period data set and the smaller portion being the test period data set.

[0051] Furthermore, the above ratio can be adjusted according to the length of continuous data, but the training period data should be longer than the test period data.

[0052] S400, selecting historical flow data of the warning section of the upstream station of the target river as candidate factors, and screening the candidate factors using the grey correlation method to obtain prediction factors;

[0053] Specifically, this embodiment takes the ten-day period as an example, selects data of several dimensions of the upstream station of the warning target river section as candidate prediction factors, takes the flow of the warning section of the upstream station 1 to 12 ten days before the training period and the flow of the same period of the previous year as candidate prediction factors, and the warning section inspection period data as reference objects, and uses gray correlation analysis to output the correlation between each candidate factor and the reference object. Select the top 7 as prediction factors according to the correlation from high to low.

[0054] Furthermore, the number of predictors can be adjusted according to the degree of correlation to avoid using predictors with low correlation. The recommended correlation is > 0.85. If the correlation is low, it is necessary to reselect alternative predictors for screening.

[0055] S500, inputting the prediction factors into the deep neural network model and the Elman neural network model, and using the test data set for testing, comparing the certainty coefficient, adjusting the number of nodes in each layer, the learning rate, the number of iterations and the expected error threshold, and completing the model calibration to achieve the required accuracy;

[0056] Specifically, in this embodiment, the number of neurons in the input layer is set according to the number of prediction factors, and the input layer has 7 neurons. The output data is the flow data of the warning section, so the output layer has 1 neuron. The hidden layer is 2 layers, and the number of nodes, learning rate, number of iterations and expected error threshold of each layer are adjusted to meet the set certainty coefficient requirements and complete the model calibration.

[0057] S601, the training data set is divided into three parts, two of which are used alternately as training data to input into the deep neural network model for simulation prediction, and the third fold data is obtained. After three cross-checks, three groups of data are obtained, which are recorded as the first data set; the Elman neural network model is also simulated and predicted, which is recorded as the second data set;

[0058] S602, using the same training data for the two models to simulate and predict the test period data set, and also after three cross-validations, three sets of data are obtained respectively, which are recorded as the third data set and the fourth data set;

[0059] S603, stacking the first data set and the second data set vertically in the order of the original data to obtain two sets of data, denoted as A1 and A2; taking the average of the third data set and the fourth data set to obtain two sets of data, denoted as B1 and B2;

[0060] S604. Integrate the meta-model based on the random forest algorithm model, use A1 and A2 as the training period data of the combination model, and use B1 and B2 as the test period data of the combination model. Determine the relevant parameters through grid search and adjust them on this basis to meet the set certainty coefficient requirements, and complete the construction of the traffic prediction combination model.

[0061] Specifically, it is divided into two stages. In the first stage, Figure 3 As shown in the figure, the training data set is divided into three parts, and two of them are used alternately as training data to input the base model (i.e., DNN and Elman neural network model) for simulation prediction to obtain the third fold data. After three cross-checks, three sets of training data are obtained, which are recorded as a1, a1, and a3. At the same time, the same training data set is used to simulate and predict the test data set. After three cross-checks, three sets of test data are obtained, which are recorded as b1, b2, and b3. The three sets of training data using the DNN and Elman neural network models are stacked vertically in the order of the original data, recorded as A1 and A2 respectively, and the average of the three sets of test data is recorded as B1 and B2. This data is the input data of the second-stage meta-model.

[0062] In the second stage, the random forest algorithm model is used to integrate the meta-model. A1 and A2 are used as training data, and B1 and B2 are used as test data. The relevant parameters are determined through grid search and adjusted on this basis to meet the set certainty coefficient requirements and complete the model calibration.

[0063] Furthermore, the coefficient of certainty is used to verify whether the model parameters are accurate, and the recommended coefficient of certainty is greater than 0.9.

[0064] S700, input the real-time data of the upstream site into the flow prediction combination model to generate flow prediction data for the future warning section, and import the flow prediction data into the precipitation flow combination model to realize visualization and generate warning information.

[0065] Embodiment 2:

[0066] The embodiment of the present invention provides a river safety early warning system based on BIM and multi-model information fusion, including:

[0067] The first module is used to obtain the terrain and environmental data of the warning target river and the historical flow data, and establish a precipitation flow model based on the terrain and environmental data of the target river and the historical flow data based on the BIM technology; divide the upstream of the target river into regions according to the precipitation flow model, and set the flow alarm threshold of the warning section; obtain the continuous flow data of the warning section of the upstream station of the target river as the training data set and the test data set;

[0068] The second module is used to select the historical flow data of the warning section of the upstream station of the target river as the candidate factor, and use the grey correlation method to screen the candidate factors to obtain the prediction factor; input the prediction factor into the deep neural network model and the Elman neural network model, and use the test data set to test, compare the certainty coefficient, adjust the number of nodes in each layer, the learning rate, the number of iterations and the expected error threshold to achieve the required accuracy and complete the model calibration; use the random forest algorithm based on the Stacking algorithm as the framework to perform multi-model information fusion on the two models to complete the construction of the flow prediction combination model;

[0069] The third module is used to input the real-time data of the upstream site into the flow prediction combination model to generate flow prediction data for the future warning section, and import the flow prediction data into the precipitation flow combination model to realize visualization and generate warning information.

[0070] Example 3

[0071] An embodiment of the present invention provides an electronic device, including:

[0072] At least one processor, at least one memory and a communication interface; wherein,

[0073] The processor, memory and communication interface communicate with each other;

[0074] The memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute any step of the above-mentioned river safety early warning method based on BIM and multi-model information fusion.

[0075] Embodiment 4:

[0076] An embodiment of the present invention provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions enable the computer to execute any step in the above-mentioned river safety warning method based on BIM and multi-model information fusion.

[0077] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0078] The above description is only a partial implementation method of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A river safety early warning method based on BIM and multi-model information fusion, characterized in that: include: S100, obtaining terrain and environmental data of the target river channel for early warning and historical flow data, and establishing a precipitation flow model based on the terrain and environmental data of the target river channel and the historical flow data based on BIM technology; S200, dividing the upstream of the target river into regions according to the precipitation flow model, and setting a flow alarm threshold of the early warning section; S300, obtaining continuous flow data of the warning section of the upstream station of the target river as a training data set and a test data set; S400, selecting historical flow data of the warning section of the upstream station of the target river as candidate factors, and screening the candidate factors using the grey correlation method to obtain prediction factors; S500, inputting the prediction factors into the deep neural network model and the Elman neural network model, and using the test data set to perform tests to complete model calibration; S600, using the random forest algorithm and based on the Stacking algorithm as a framework, multi-model information fusion of the two models is performed to complete the construction of the traffic prediction combination model; S700, input the real-time data of the upstream site into the flow prediction combination model to generate the flow prediction combination data of the future warning section, and import the flow prediction data into the precipitation flow model to realize visualization and generate warning information.

2. A river safety early warning method based on BIM and multi-model information fusion according to claim 1, characterized in that: The target river terrain environment data in step S100 includes the roughness, water depth, river section length and cross-sectional data of the target river.

3. The river safety early warning method based on BIM and multi-model information fusion according to claim 1 is characterized in that: In step S300, the training data set is longer than the test data set.

4. A river safety early warning method based on BIM and multi-model information fusion according to claim 1, characterized in that: Step S400 includes: selecting the flow data of 1 to 12 days before the warning section of the upstream station and the flow data of the same period of the previous year as alternative factors, selecting the inspection period data set as a reference object, using grey correlation analysis to output the correlation between each alternative factor and the reference object, and selecting multiple alternative factors as forecasting factors according to the correlation from high to low.

5. A river safety early warning method based on BIM and multi-model information fusion according to claim 4, characterized in that: Step S400 also includes: adjusting the number of prediction factors according to the degree of correlation to avoid using prediction factors with low correlation, and if the correlations are all low, reselecting candidate factors for screening.

6. A river safety early warning method based on BIM and multi-model information fusion according to claim 4, characterized in that: The model calibration in step S500 also includes: comparing the deterministic coefficient, adjusting the number of nodes in each layer, the learning rate, the number of iterations and the expected error threshold to achieve the required accuracy of the model.

7. The river safety early warning method based on BIM and multi-model information fusion according to claim 1 is characterized in that: Step S600 includes: S601, the training data set is divided into three parts, two of which are used alternately as training data to input into the deep neural network model for simulation prediction, and the third fold data is obtained. After three cross-checks, three groups of data are obtained, which are recorded as the first data set; the Elman neural network model is also simulated and predicted, which is recorded as the second data set; S602, using the same training data for the two models to simulate and predict the test period data set, and also after three cross-validations, three sets of data are obtained respectively, which are recorded as the third data set and the fourth data set; S603, stacking the first data set and the second data set vertically in the order of the original data to obtain two sets of data, denoted as A1 and A2; taking the average of the third data set and the fourth data set to obtain two sets of data, denoted as B1 and B2; S604. Integrate the meta-model based on the random forest algorithm model, use A1 and A2 as the training period data of the combination model, and use B1 and B2 as the test period data of the combination model. Determine the relevant parameters through grid search and adjust them on this basis to meet the set certainty coefficient requirements, and complete the construction of the traffic prediction combination model.

8. A river safety early warning system based on BIM and multi-model information fusion, characterized in that: include: The first module is used to obtain the terrain and environmental data of the warning target river and the historical flow data, and establish a precipitation flow model based on the terrain and environmental data of the target river and the historical flow data based on the BIM technology; divide the upstream of the target river into regions according to the precipitation flow model, and set the flow alarm threshold of the warning section; Obtain continuous flow data of the warning section of the upstream station of the target river as training data set and test data set; The second module is used to select the historical flow data of the warning section of the upstream station of the target river as the candidate factors, and use the grey correlation method to screen the candidate factors to obtain the prediction factors; Inputting the prediction factors into the deep neural network model and the Elman neural network model, and using the test data set to perform testing to complete model calibration; The random forest algorithm is used to fuse the multi-model information of the two models based on the Stacking algorithm framework to complete the construction of the traffic prediction combination model; The third module is used to input the real-time data of the upstream site into the flow prediction combination model to generate flow prediction data for the future warning section, and import the flow prediction data into the precipitation flow combination model to realize visualization and generate warning information.

9. An electronic device, characterized in that: include: At least one processor, at least one memory and a communication interface; wherein, The processor, memory and communication interface communicate with each other; The memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute the method according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, which cause the computer to execute the method of any one of claims 1 to 7.

Images