A hierarchical reservoir capacity intelligent measurement method and system

Abstract

The present invention relates to the technical field of reservoir capacity monitoring, and specifically discloses a hierarchical reservoir capacity intelligent measurement method and system, the method comprising counting remote sensing images and water level matrices corresponding to the same time as a first sample; calculating difference images of remote sensing images at different times and difference matrices of water level matrices, counting corresponding difference images and difference matrices as a second sample; training a first recognition model from remote sensing images to water level matrices according to the first sample, and training a second recognition model from difference images to difference matrices according to the second sample; and recognizing the acquired remote sensing images based on the trained first recognition model and the second recognition model. The present invention measures the reservoir area through photos, and can calculate the reservoir capacity in combination with the water level at the current time. While measuring the reservoir capacity, a neural network model is trained. When in use, the reservoir capacity can be obtained directly with the help of the trained neural network model, and the response speed is extremely fast.

Description

Technical Field

[0001] The present invention relates to the technical field of reservoir capacity monitoring, and in particular to a hierarchical reservoir capacity intelligent measurement method and system. Background Art

[0002] Reservoirs can play a role in flood control, water storage for irrigation, water supply, power generation, and fish farming. Their importance is self-evident.

[0003] While reservoirs bring convenience, they also have certain risks. Therefore, it is necessary to monitor the water level in real time. The water level monitoring process is not difficult. There are many measuring instruments in the prior art that can be used to measure the water level. However, there is no means in the prior art that can detect the reservoir capacity. In fact, the reservoir capacity is more direct data than the water level, and it is extremely convenient to use. How to provide a reservoir capacity measurement solution is the technical problem that the technical solution of the present invention wants to solve. Summary of the invention

[0004] The purpose of the present invention is to provide a hierarchical reservoir capacity intelligent measurement method and system to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A hierarchical reservoir capacity intelligent measurement method, the method comprising:

[0007] Obtaining the absolute position of a water level measuring instrument, creating a measuring instrument matrix according to the absolute position, and calculating the water level at each moment based on the measuring instrument matrix to obtain a water level matrix;

[0008] Send a monitoring request to the satellite, receive the remote sensing images fed back by the satellite, and count the remote sensing images and water level matrices corresponding to the same time as the first sample;

[0009] Calculate the difference images of the remote sensing images at different times and the difference matrix of the water level matrix, and count the corresponding difference images and difference matrices as the second sample;

[0010] Training a first recognition model from a remote sensing image to a water level matrix according to the first sample, and training a second recognition model from a difference image to a difference matrix according to the second sample;

[0011] The acquired remote sensing image is recognized based on the trained first recognition model and the second recognition model to determine the current water level matrix; wherein the use frequency of the second recognition model is greater than the use frequency of the first recognition model.

[0012] As a further solution of the present invention: the steps of obtaining the absolute position of the water level measuring instrument, creating a measuring instrument matrix according to the absolute position, and counting the water level at each moment based on the measuring instrument matrix to obtain the water level matrix include:

[0013] Obtain the absolute position of the water level measuring instrument, and calculate the maximum distance of the water level measuring instrument in the east-west direction and the maximum distance of the water level measuring instrument in the north-south direction according to the absolute position;

[0014] Select the larger value as the span of the measuring instrument, determine the matrix size corresponding to the measuring instrument according to the preset scale, and create the measuring instrument matrix;

[0015] Mark the row and column positions corresponding to the water level measuring instruments in the measuring instrument matrix;

[0016] At the predicted frequency, the water level obtained by the water level measuring instrument is received in real time, and the water level is filled to the marked row and column position;

[0017] Data fitting is performed on unlabeled row and column positions according to the water levels at the row and column positions to obtain a water level matrix; wherein the water level matrix uses the acquisition time of the water level as a label.

[0018] As a further solution of the present invention: the step of performing data fitting on unmarked row and column positions according to the water levels at the row and column positions to obtain a water level matrix comprises:

[0019] Connect the marked row and column positions in sequence to obtain a connecting line and simultaneously calculate the value change gradient of the connecting line;

[0020] For any unmarked row and column position, calculate its distance from the connecting line and determine the weight based on the distance and the length of the connecting line;

[0021] According to the relative positions of the unmarked row and column positions on each connecting line, the predicted data is determined according to the relative positions and the gradient of value changes;

[0022] Calculate the mean of the predicted data as the fitted data, insert the fitted data into the unmarked row and column positions to obtain the water level matrix;

[0023] The calculation process of fitting data is:

[0024] Where N(i,j) is the fitting data at point (i,j), L is the total number of connecting lines, and d k is the distance between point (i, j) and the kth connecting line, Δ k is the numerical difference between the row and column positions of the two ends of the kth connecting line; B k It is the ratio of the distance between the foot of the perpendicular point (i, j) on the kth connecting line and one end of the connecting line to the length of the connecting line.

[0025] As a further solution of the present invention, the step of sending a monitoring request to a satellite, receiving remote sensing images fed back by the satellite, and counting the remote sensing images and water level matrices corresponding to the same time as the first sample includes:

[0026] Send monitoring requests to satellites and receive remote sensing images fed back by satellites;

[0027] Extract standard building parameters from remote sensing images based on ENVI software;

[0028] Color matching of all remote sensing images according to standard architectural parameters;

[0029] The remote sensing image and the water level matrix after color value matching at the same time are counted as the first sample.

[0030] As a further solution of the present invention: the step of calculating the difference images of the remote sensing images at different times and the difference matrix of the water level matrix, and counting the corresponding difference images and difference matrices as the second sample includes:

[0031] receiving the number of difference images input by a staff member, and randomly combining remote sensing images at different times based on the number;

[0032] Read the water level matrix corresponding to the combined remote sensing image and calculate the difference matrix;

[0033] The difference images and difference matrices of different combinations are counted as the second samples; wherein the difference images and the difference matrices are in a many-to-one relationship.

[0034] As a further solution of the present invention: the step of identifying the acquired remote sensing image based on the trained first recognition model and the second recognition model to determine the current water level matrix includes:

[0035] After each preset period, the acquired remote sensing image is input into the trained first recognition model, a water level matrix is ​​output, and the remote sensing image is marked as a reference image;

[0036] For each acquired remote sensing image, calculate the difference image between it and the nearest reference image, input the trained second recognition model, and output the water level matrix;

[0037] The output of the same remote sensing image in the first recognition model and the output in the second recognition model are compared to determine and display the accuracy of the second recognition model.

[0038] The technical solution of the present invention also provides a hierarchical reservoir capacity intelligent measurement system, the system comprising:

[0039] A water level statistics module is used to obtain the absolute position of the water level measuring instrument, create a measuring instrument matrix according to the absolute position, and count the water level at each moment based on the measuring instrument matrix to obtain a water level matrix;

[0040] The first sample acquisition module is used to send a monitoring request to the satellite, receive the remote sensing image fed back by the satellite, and count the remote sensing image and water level matrix corresponding to the same time as the first sample;

[0041] The second sample acquisition module is used to calculate the difference images of the remote sensing images at different times and the difference matrix of the water level matrix, and count the corresponding difference images and difference matrices as the second samples;

[0042] A model training module, used for training a first recognition model from a remote sensing image to a water level matrix according to a first sample, and training a second recognition model from a difference image to a difference matrix according to a second sample;

[0043] The model application module is used to identify the acquired remote sensing image based on the trained first recognition model and the second recognition model to determine the current water level matrix; wherein the use frequency of the second recognition model is greater than the use frequency of the first recognition model.

[0044] As a further solution of the present invention: the water level statistics module includes:

[0045] A distance query unit is used to obtain the absolute position of the water level measuring instrument, and calculate the maximum distance of the water level measuring instrument in the east-west direction and the maximum distance of the water level measuring instrument in the north-south direction according to the absolute position;

[0046] A matrix creation unit, used for selecting a larger value as the span of the measuring instrument, determining the matrix size corresponding to the measuring instrument according to a preset scale, and creating a matrix of the measuring instrument;

[0047] A position marking unit, used for marking the row and column positions corresponding to the water level measuring instruments in the measuring instrument matrix;

[0048] A matrix filling unit is used to receive the water level obtained by the water level measuring instrument in real time at the predicted frequency and fill the water level to the marked row and column positions;

[0049] The data fitting unit is used to perform data fitting on the unmarked row and column positions according to the water levels at the row and column positions to obtain a water level matrix; wherein the water level matrix uses the acquisition time of the water level as a label.

[0050] As a further solution of the present invention: the first sample acquisition module includes:

[0051] An image receiving unit is used to send monitoring requests to the satellite and receive remote sensing images fed back by the satellite;

[0052] Standard extraction unit, used to extract standard building parameters from remote sensing images based on ENVI software;

[0053] A color value matching unit is used to match the color values ​​of all remote sensing images according to standard building parameters;

[0054] The sample construction unit is used to count the remote sensing image and the water level matrix after color value matching at the same time as the first sample.

[0055] As a further solution of the present invention: the second sample acquisition module includes:

[0056] An image combining unit, used for receiving the number of difference images input by a staff member, and randomly combining remote sensing images at different times based on the number;

[0057] A difference calculation unit, used for reading the water level matrix corresponding to the combined remote sensing image and calculating the difference matrix;

[0058] The statistical unit is used to count the difference images and difference matrices of different combinations as the second sample; wherein the difference images and the difference matrices are in a many-to-one relationship.

[0059] Compared with the prior art, the beneficial effects of the present invention are: the present invention measures the reservoir area through photos, and can calculate the reservoir capacity in combination with the current water level. While measuring the reservoir capacity, the reservoir photo is used as input, and the reservoir capacity is used as output to train the neural network model. When in use, the reservoir capacity can be obtained directly with the help of the trained neural network model, and the response speed is extremely fast. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention.

[0061] Figure 1 This is a flowchart of the hierarchical reservoir capacity intelligent measurement method.

[0062] Figure 2 This is a structural block diagram of the hierarchical reservoir capacity intelligent measurement system. DETAILED DESCRIPTION

[0063] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0064] Figure 1The flowchart of the hierarchical reservoir capacity intelligent measurement method is shown in FIG. 1 . In an embodiment of the present invention, a hierarchical reservoir capacity intelligent measurement method is provided, and the method includes:

[0065] Step S100: obtaining the absolute position of the water level measuring instrument, creating a measuring instrument matrix according to the absolute position, and calculating the water level at each moment based on the measuring instrument matrix to obtain a water level matrix;

[0066] There are many types of water level measuring instruments, including radar water level gauges, ultrasonic water level gauges, pressure water level gauges and electronic water gauges. Their principles are different, but they have the same function, which is to obtain water levels. There are more than one water level measuring instruments installed in the reservoir. The more water level measuring instruments there are, the more points there are to measure and the more realistic the measurement results are. A measuring instrument matrix is ​​created based on the absolute position of the water level measuring instrument to count the data collected by each measuring instrument and obtain a water level matrix. One moment corresponds to one water level matrix.

[0067] Step S200: sending a monitoring request to the satellite, receiving the remote sensing image fed back by the satellite, and counting the remote sensing image and water level matrix corresponding to the same time as the first sample;

[0068] Send a monitoring request to the satellite to obtain remote sensing images of the reservoir area. The remote sensing images have their own time tags, which are used to indicate when the remote sensing images were taken. Query the water level matrix corresponding to that moment, and take the remote sensing images and water level matrix at the same moment as a sample.

[0069] It is worth mentioning that the process of sending monitoring requests to satellites can make use of existing map services and obtain remote sensing images with the help of map service providers. The acquisition process only requires registering an account, which is very simple.

[0070] Step S300: Calculate the difference images of the remote sensing images at different times and the difference matrix of the water level matrix, and count the corresponding difference images and difference matrices as the second sample;

[0071] Furthermore, by randomly selecting different moments, the remote sensing images and water level matrices corresponding to the two moments can be read. The difference images can be obtained by subtracting the remote sensing images, and the difference matrices can be obtained by subtracting the water level matrices. The corresponding difference images and difference matrices are statistically counted as the second sample.

[0072] In actual application, the reservoir is in a stable state most of the time, which makes the difference between each pixel in the difference image at different times very small, that is, the amount of data in the difference image is extremely small, which is a significant simplification of the remote sensing image. Based on the same principle, the difference matrix is ​​also similar; because the amount of data in the second sample is much smaller than that in the first sample.

[0073] Step S400: training a first recognition model from a remote sensing image to a water level matrix according to a first sample, and training a second recognition model from a difference image to a difference matrix according to a second sample;

[0074] The first recognition model is the mapping relationship from remote sensing image to water level matrix. The sample data volume is large, the training process and application process are time-consuming, the recognition speed is slow, and the recognition accuracy is high. The second recognition model is the mapping relationship from difference image to difference matrix. The sample data volume is small, the training process and application process are time-consuming, the recognition speed is fast, and the recognition accuracy is low.

[0075] Step S500: Recognize the acquired remote sensing image based on the trained first recognition model and the second recognition model to determine the current water level matrix; wherein the use frequency of the second recognition model is greater than the use frequency of the first recognition model;

[0076] The first recognition model and the second recognition model together form a gradient recognition scheme. Under normal circumstances, the second recognition model is used to ensure the feedback speed and realize the real-time monitoring function of reservoir capacity. In order to ensure accuracy, the remote sensing images are recognized with higher accuracy at regular intervals to ensure accuracy.

[0077] A specific application scheme is that the first recognition model regularly recognizes the remote sensing image to obtain the recognition result, and then uses it as a benchmark. Every time a new remote sensing image is acquired, the difference between the base and the benchmark is calculated and input into the second recognition model to obtain the difference matrix. The difference matrix and the recognition result are combined to obtain the result corresponding to the current remote sensing image. The second recognition model is extremely efficient and ensures real-time performance.

[0078] It should be noted that in the reservoir capacity identification process based on remote sensing images, the time relationship is extremely important, and it is necessary to ensure that the remote sensing image and the water level measurement results are at the same time (the time difference is extremely small). An example is as follows:

[0079] The water level detection equipment detects that the current water depth of the reservoir is 15 meters. A reservoir photo of the reservoir is obtained within the effective time period (almost the same time), and the reservoir capacity is calculated based on 15 meters. The calculated reservoir capacity is used as the label of the reservoir photo. Then, according to the rise or fall of the water level, the above process is repeated continuously, and finally multiple reservoir capacities and reservoir photos are obtained. The neural network model between reservoir photos and reservoir capacity is trained. When the reservoir capacity needs to be calculated, the neural network model can be directly applied, which is extremely convenient.

[0080] In addition, regarding the sample construction process, the main process of this application is:

[0081] First select a water level and record the time of that water level. When the water level rises or sinks, record the time of another water level. Similarly, you can get the corresponding time of different water levels. At this time, query the remote sensing image by time, measure the area in the remote sensing image with the help of calculus, and multiply it by the current water level to get the reservoir capacity. In this process, it is necessary to ensure that the water level and the remote sensing image correspond in time.

[0082] As a preferred embodiment of the technical solution of the present invention, step S100 is specifically defined, and the steps of obtaining the absolute position of the water level measuring instrument, creating a measuring instrument matrix according to the absolute position, and counting the water level at each moment based on the measuring instrument matrix to obtain the water level matrix include:

[0083] Obtain the absolute position of the water level measuring instrument, and calculate the maximum distance of the water level measuring instrument in the east-west direction and the maximum distance of the water level measuring instrument in the north-south direction according to the absolute position;

[0084] Select the larger value as the span of the measuring instrument, determine the matrix size corresponding to the measuring instrument according to the preset scale, and create the measuring instrument matrix;

[0085] Mark the row and column positions corresponding to the water level measuring instruments in the measuring instrument matrix;

[0086] At the predicted frequency, the water level obtained by the water level measuring instrument is received in real time, and the water level is filled to the marked row and column position;

[0087] Data fitting is performed on unlabeled row and column positions according to the water levels at the row and column positions to obtain a water level matrix; wherein the water level matrix uses the acquisition time of the water level as a label.

[0088] The above content provides a specific water level matrix acquisition solution. First, the absolute position of the water level measuring instrument is obtained, the maximum distance in the two vertical directions is obtained, and then the larger distance is selected from the maximum distance. The matrix size is calculated according to the preset scale, and the measuring instrument matrix is ​​determined by the matrix size. In the process of determining the measuring instrument matrix, the row and column positions corresponding to the water level measuring instrument are synchronously marked; then, when the water level measuring instrument monitors the water level, the water level is filled to the marked row and column positions.

[0089] In the above content, the water level measuring instrument is limited, and the corresponding row and column positions are only a part of the water level matrix. Regarding the data at other row and column positions, this application fills them in by fitting.

[0090] Specifically, regarding the fitting scheme: the step of performing data fitting on the unmarked row and column positions according to the water levels at the row and column positions to obtain a water level matrix includes:

[0091] Connect the marked row and column positions in sequence to obtain a connecting line and simultaneously calculate the value change gradient of the connecting line;

[0092] For any unmarked row and column position, calculate its distance from the connecting line and determine the weight based on the distance and the length of the connecting line;

[0093] According to the relative positions of the unmarked row and column positions on each connecting line, the predicted data is determined according to the relative positions and the gradient of value changes;

[0094] Calculate the mean of the predicted data as the fitted data, insert the fitted data into the unlabeled row and column positions to obtain the water level matrix.

[0095] In the water level matrix, the data at the marked row and column positions are obtained by the water level measuring instrument and are real data. By connecting them in pairs, we can get a connecting line. The numerical change gradient (which can be compared to the derivative) can be calculated by comparing the data at the row and column positions at both ends. Then, for each unmarked row and column position, its predicted data can be determined according to its relationship with the connecting line. This process of predicting unknown data based on existing data is called the fitting process.

[0096] Specifically, the mathematical expression of the fitting process is:

[0097] Where N(i,j) is the fitting data at point (i,j), L is the total number of connecting lines, and d k is the distance between point (i, j) and the kth connecting line, Δ k is the numerical difference between the row and column positions of the two ends of the kth connecting line, Y k The value at one end of the minuend when calculating the numerical difference; B k It is the ratio of the distance between the foot of the perpendicular point (i, j) on the kth connecting line and one end of the connecting line to the length of the connecting line.

[0098] Regarding the above fitting process, the calculation process involved is mainly the distance between the point and the line. The coordinates of the point are (i, j). The equation of the line is determined by the coordinates of the two endpoints. There is a general formula for the distance between the point and the line in mathematics, and the calculation process is very simple. In this process, there is also a general formula for the coordinates of the foot of the perpendicular, which can be directly calculated.

[0099] The closer the distance between a connecting line and point (i, j), the greater the influence of the connecting line on the fitting process, and the greater the weight. The weight is between 0 and 1, which is the value in the above content. Item; then, multiply the proportion of the perpendicular foot on the connecting line by the difference in values ​​at both ends of the connecting line, and add the value at one end to get a value that is used to represent the value of the current connecting line at point (i, j). The values ​​of all connecting lines at point (i, j) are accumulated to get the final fitting data at point (i, j).

[0100] As a preferred embodiment of the technical solution of the present invention, step S200 is specifically defined, and the steps of sending a monitoring request to a satellite, receiving remote sensing images fed back by the satellite, and counting the remote sensing images and water level matrices corresponding to the same time as the first sample include:

[0101] Send monitoring requests to satellites and receive remote sensing images fed back by satellites;

[0102] Extract standard building parameters from remote sensing images based on ENVI software;

[0103] Color matching of all remote sensing images according to standard architectural parameters;

[0104] The remote sensing image and the water level matrix after color value matching at the same time are counted as the first sample.

[0105] The processing of remote sensing images takes place in the ENVI software. The process of acquiring remote sensing images has been described in the above content. With the help of existing public map services, the acquisition difficulty is extremely low. The standard building parameters in the remote sensing images are extracted, and the standard building parameters in different remote sensing images are corrected to the same parameters and the correction process is recorded. The entire remote sensing image is corrected by the correction process, so that all remote sensing images correspond to the same environmental conditions. Among them, the same parameters corrected to the same parameters are preset parameters. The meaning of standard building parameters is the parameters of standard buildings. For example, a measuring station is regarded as a standard parameter, and the parameters of the measuring station in the remote sensing image are called standard building parameters.

[0106] It is worth mentioning that the above content is essentially a preprocessing of the remote sensing images in the sample. Correspondingly, in the actual recognition process, it is also necessary to adopt the same preprocessing scheme for new images and then apply the trained model.

[0107] As a preferred embodiment of the technical solution of the present invention, step S300 is specifically defined, and the step of calculating the difference image of the remote sensing image at different times and the difference matrix of the water level matrix, and counting the corresponding difference image and difference matrix as the second sample includes:

[0108] receiving the number of difference images input by a staff member, and randomly combining remote sensing images at different times based on the number;

[0109] Read the water level matrix corresponding to the combined remote sensing image and calculate the difference matrix;

[0110] The difference images and difference matrices of different combinations are counted as the second samples.

[0111] The creation process of the second sample is very simple, which is to select remote sensing images at two moments and obtain their corresponding water level matrices, and then calculate the difference respectively. It is worth mentioning that the method of selecting remote sensing images at two moments is sufficient. Assuming that there are N remote sensing images, there are N(N-1) / 2 combinations; in these N(N-1) / 2 combinations, two remote sensing images with the same time difference can be regarded as one category.

[0112] In order to simplify the construction process of the second sample, the staff pre-sets a number so that the number of difference images corresponding to each time difference is the same, which makes the structure of the second sample more regular.

[0113] As a preferred embodiment of the technical solution of the present invention, step S500 is specifically defined, and the step of identifying the acquired remote sensing image based on the trained first recognition model and the second recognition model to determine the current water level matrix includes:

[0114] After each preset period, the acquired remote sensing image is input into the trained first recognition model, a water level matrix is ​​output, and the remote sensing image is marked as a reference image;

[0115] For each acquired remote sensing image, calculate the difference image between it and the nearest reference image, input the trained second recognition model, and output the water level matrix;

[0116] The output of the same remote sensing image in the first recognition model and the output in the second recognition model are compared to determine and display the accuracy of the second recognition model.

[0117] Finally, the trained first recognition model and the trained second recognition model can be applied. The first recognition model can perform regular recognition to determine the reference image. Then, based on the reference image, the second recognition model can recognize the acquired remote sensing image in real time.

[0118] It is worth mentioning that while the remote sensing image is periodically input into the first recognition model, the remote sensing image is also input into the second recognition model at the same time, so that the accuracy of the second recognition model can be determined and then reported to the staff to determine whether the second recognition model needs to be updated.

[0119] Figure 2 1 is a structural block diagram of a hierarchical reservoir capacity intelligent measurement system. In an embodiment of the present invention, a hierarchical reservoir capacity intelligent measurement system is provided. The system 10 includes:

[0120] The water level statistics module 11 is used to obtain the absolute position of the water level measuring instrument, create a measuring instrument matrix according to the absolute position, and count the water level at each moment based on the measuring instrument matrix to obtain a water level matrix;

[0121] The first sample acquisition module 12 is used to send a monitoring request to the satellite, receive the remote sensing image fed back by the satellite, and count the remote sensing image and water level matrix corresponding to the same time as the first sample;

[0122] The second sample acquisition module 13 is used to calculate the difference images of the remote sensing images at different times and the difference matrix of the water level matrix, and count the corresponding difference images and difference matrices as the second samples;

[0123] A model training module 14 is used to train a first recognition model from a remote sensing image to a water level matrix according to a first sample, and to train a second recognition model from a difference image to a difference matrix according to a second sample;

[0124] The model application module 15 is used to identify the acquired remote sensing image based on the trained first recognition model and the second recognition model to determine the current water level matrix; wherein the use frequency of the second recognition model is greater than the use frequency of the first recognition model.

[0125] Furthermore, the water level statistics module 11 includes:

[0126] A distance query unit is used to obtain the absolute position of the water level measuring instrument, and calculate the maximum distance of the water level measuring instrument in the east-west direction and the maximum distance of the water level measuring instrument in the north-south direction according to the absolute position;

[0127] A matrix creation unit, used for selecting a larger value as the span of the measuring instrument, determining the matrix size corresponding to the measuring instrument according to a preset scale, and creating a matrix of the measuring instrument;

[0128] A position marking unit, used for marking the row and column positions corresponding to the water level measuring instruments in the measuring instrument matrix;

[0129] A matrix filling unit is used to receive the water level obtained by the water level measuring instrument in real time at the predicted frequency and fill the water level to the marked row and column positions;

[0130] The data fitting unit is used to perform data fitting on the unmarked row and column positions according to the water levels at the row and column positions to obtain a water level matrix; wherein the water level matrix uses the acquisition time of the water level as a label.

[0131] Specifically, the first sample acquisition module 12 includes:

[0132] An image receiving unit is used to send monitoring requests to the satellite and receive remote sensing images fed back by the satellite;

[0133] Standard extraction unit, used to extract standard building parameters from remote sensing images based on ENVI software;

[0134] A color value matching unit is used to match the color values ​​of all remote sensing images according to standard building parameters;

[0135] The sample construction unit is used to count the remote sensing image and the water level matrix after color value matching at the same time as the first sample.

[0136] Furthermore, the second sample acquisition module 13 includes:

[0137] An image combining unit, used for receiving the number of difference images input by a staff member, and randomly combining remote sensing images at different times based on the number;

[0138] A difference calculation unit, used for reading the water level matrix corresponding to the combined remote sensing image and calculating the difference matrix;

[0139] The statistical unit is used to count the difference images and difference matrices of different combinations as the second sample; wherein the difference images and the difference matrices are in a many-to-one relationship.

[0140] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A hierarchical reservoir capacity intelligent measurement method, characterized in that: The method comprises: Obtaining the absolute position of a water level measuring instrument, creating a measuring instrument matrix according to the absolute position, and calculating the water level at each moment based on the measuring instrument matrix to obtain a water level matrix; Send a monitoring request to the satellite, receive the remote sensing images fed back by the satellite, and count the remote sensing images and water level matrices corresponding to the same time as the first sample; Calculate the difference images of the remote sensing images at different times and the difference matrix of the water level matrix, and count the corresponding difference images and difference matrices as the second sample; Training a first recognition model from a remote sensing image to a water level matrix according to the first sample, and training a second recognition model from a difference image to a difference matrix according to the second sample; Based on the trained first recognition model and the second recognition model, the acquired remote sensing image is recognized to determine the current water level matrix; wherein the use frequency of the second recognition model is greater than the use frequency of the first recognition model; Regarding the sample construction process, the main process of this application is: First select a water level and record the time of that water level. When the water level rises or sinks, record the time of another water level. By analogy, you can get the corresponding time of different water levels. At this time, query the remote sensing image by time, measure the area in the remote sensing image with the help of calculus, and multiply it by the current water level to get the reservoir capacity.

2. The hierarchical reservoir capacity intelligent measurement method according to claim 1 is characterized in that: The steps of obtaining the absolute position of the water level measuring instrument, creating a measuring instrument matrix according to the absolute position, and counting the water level at each moment based on the measuring instrument matrix to obtain the water level matrix include: Obtain the absolute position of the water level measuring instrument, and calculate the maximum distance of the water level measuring instrument in the east-west direction and the maximum distance of the water level measuring instrument in the north-south direction according to the absolute position; Select the larger value as the span of the measuring instrument, determine the matrix size corresponding to the measuring instrument according to the preset scale, and create the measuring instrument matrix; Mark the row and column positions corresponding to the water level measuring instruments in the measuring instrument matrix; At the predicted frequency, the water level obtained by the water level measuring instrument is received in real time, and the water level is filled to the marked row and column position; Data fitting is performed on unlabeled row and column positions according to the water levels at the row and column positions to obtain a water level matrix; wherein the water level matrix uses the acquisition time of the water level as a label.

3. The hierarchical reservoir capacity intelligent measurement method according to claim 2 is characterized in that: The step of performing data fitting on unmarked row and column positions according to the water levels at the row and column positions to obtain a water level matrix comprises: Connect the marked row and column positions in sequence to obtain a connecting line and simultaneously calculate the value change gradient of the connecting line; For any unmarked row and column position, calculate its distance from the connecting line and determine the weight based on the distance and the length of the connecting line; According to the relative positions of the unmarked row and column positions on each connecting line, the predicted data is determined according to the relative positions and the gradient of value changes; Calculate the mean of the predicted data as the fitted data, insert the fitted data into the unmarked row and column positions to obtain the water level matrix; The calculation process of fitting data is: Where N(i,j) is the fitting data at point (i,j), L is the total number of connecting lines, and d k is the distance between point (i, j) and the kth connecting line, Δ k is the numerical difference between the row and column positions of the two ends of the kth connecting line; B k It is the ratio of the distance between the foot of the perpendicular point (i, j) on the kth connecting line and one end of the connecting line to the length of the connecting line.

4. The hierarchical reservoir capacity intelligent measurement method according to claim 1 is characterized in that: The steps of sending a monitoring request to a satellite, receiving remote sensing images fed back by the satellite, and counting the remote sensing images and water level matrices corresponding to the same time as the first sample include: Send monitoring requests to satellites and receive remote sensing images fed back by satellites; Extract standard building parameters from remote sensing images based on ENVI software; Color matching of all remote sensing images according to standard architectural parameters; The remote sensing image and the water level matrix after color value matching at the same time are counted as the first sample.

5. The hierarchical reservoir capacity intelligent measurement method according to claim 1 is characterized in that: The step of calculating the difference images of the remote sensing images at different times and the difference matrix of the water level matrix, and counting the corresponding difference images and difference matrices as the second sample includes: receiving the number of difference images input by a staff member, and randomly combining remote sensing images at different times based on the number; Read the water level matrix corresponding to the combined remote sensing image and calculate the difference matrix; The difference images and difference matrices of different combinations are counted as the second samples; wherein the difference images and the difference matrices are in a many-to-one relationship.

6. The hierarchical reservoir capacity intelligent measurement method according to claim 1 is characterized in that: The step of identifying the acquired remote sensing image based on the trained first recognition model and the second recognition model to determine the current water level matrix includes: After each preset period, the acquired remote sensing image is input into the trained first recognition model, a water level matrix is ​​output, and the remote sensing image is marked as a reference image; For each acquired remote sensing image, calculate the difference image between it and the nearest reference image, input the trained second recognition model, and output the water level matrix; The output of the same remote sensing image in the first recognition model and the output in the second recognition model are compared to determine and display the accuracy of the second recognition model.

7. A hierarchical reservoir capacity intelligent measurement system, characterized in that: The system comprises: A water level statistics module is used to obtain the absolute position of the water level measuring instrument, create a measuring instrument matrix according to the absolute position, and count the water level at each moment based on the measuring instrument matrix to obtain a water level matrix; The first sample acquisition module is used to send a monitoring request to the satellite, receive the remote sensing image fed back by the satellite, and count the remote sensing image and water level matrix corresponding to the same time as the first sample; The second sample acquisition module is used to calculate the difference images of the remote sensing images at different times and the difference matrix of the water level matrix, and count the corresponding difference images and difference matrices as the second samples; A model training module, used for training a first recognition model from a remote sensing image to a water level matrix according to a first sample, and training a second recognition model from a difference image to a difference matrix according to a second sample; A model application module is used to identify the acquired remote sensing image based on the trained first recognition model and the second recognition model to determine the current water level matrix; wherein the use frequency of the second recognition model is greater than the use frequency of the first recognition model; Regarding the sample construction process, the main process of this application is: First select a water level and record the time of that water level. When the water level rises or sinks, record the time of another water level. By analogy, you can get the corresponding time of different water levels. At this time, query the remote sensing image by time, measure the area in the remote sensing image with the help of calculus, and multiply it by the current water level to get the reservoir capacity.

8. The hierarchical reservoir capacity intelligent measurement system according to claim 7 is characterized in that: The water level statistics module includes: A distance query unit is used to obtain the absolute position of the water level measuring instrument, and calculate the maximum distance of the water level measuring instrument in the east-west direction and the maximum distance of the water level measuring instrument in the north-south direction according to the absolute position; A matrix creation unit, used for selecting a larger value as the span of the measuring instrument, determining the matrix size corresponding to the measuring instrument according to a preset scale, and creating a matrix of the measuring instrument; A position marking unit, used for marking the row and column positions corresponding to the water level measuring instruments in the measuring instrument matrix; A matrix filling unit is used to receive the water level obtained by the water level measuring instrument in real time at the predicted frequency and fill the water level to the marked row and column positions; The data fitting unit is used to perform data fitting on the unmarked row and column positions according to the water levels at the row and column positions to obtain a water level matrix; wherein the water level matrix uses the acquisition time of the water level as a label.

9. The hierarchical reservoir capacity intelligent measurement system according to claim 7 is characterized in that: The first sample acquisition module includes: An image receiving unit is used to send monitoring requests to the satellite and receive remote sensing images fed back by the satellite; Standard extraction unit, used to extract standard building parameters from remote sensing images based on ENVI software; A color value matching unit is used to match the color values ​​of all remote sensing images according to standard building parameters; The sample construction unit is used to count the remote sensing image and the water level matrix after color value matching at the same time as the first sample.

10. The hierarchical reservoir capacity intelligent measurement system according to claim 7 is characterized in that: The second sample acquisition module includes: An image combining unit, used for receiving the number of difference images input by a staff member, and randomly combining remote sensing images at different times based on the number; A difference calculation unit, used for reading the water level matrix corresponding to the combined remote sensing image and calculating the difference matrix; The statistical unit is used to count the difference images and difference matrices of different combinations as the second sample; wherein the difference images and the difference matrices are in a many-to-one relationship.

Images