River flow monitoring method and system fusing multi-source data

Abstract

The invention relates to the technical field of hydrological monitoring, and discloses a river flow monitoring method and system fusing multi-source data, and the method comprises the steps: collecting river flow data and river water level data through employing a high-precision sensor group, and collecting river image data through employing an unmanned plane; filtering the river flow data and the river water level data by adopting a Kalman filtering method to obtain filtered river flow data and filtered river water level data; performing edge detection on the riverway image data by adopting a Canny edge detection method to obtain riverway boundary information; on the basis of the riverway boundary information, a cubic spline interpolation method is adopted for interpolation, a smooth first boundary line and a smooth second boundary line are obtained, and the actual flow of the riverway is calculated on the basis of the filtered riverway water level data and the riverway flow velocity data; judging whether the actual flow of the riverway is greater than a riverway flow preset value, and if yes, sending alarm information; the actual flow of the river channel is calculated by fusing the multi-source data, and the accuracy of monitoring the flow of the river channel is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of hydrological monitoring, and in particular to a river flow monitoring method and system integrating multi-source data. Background Art

[0002] Traditional river flow monitoring methods mainly rely on single sensor data, such as current meters and water level meters, which are easily affected by environmental factors and have a limited monitoring range. At the same time, drone technology is increasingly used in river monitoring, but using drone image data alone for flow monitoring also has the problem of insufficient accuracy. Summary of the invention

[0003] In view of the above-mentioned deficiencies in the prior art, the present invention provides a river flow monitoring method and system that integrates multi-source data, which fuses the river flow data and river water level data collected by the sensor group with the river image data collected by the drone, thereby calculating the actual flow of the river and solving the problem of low accuracy in river flow monitoring.

[0004] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is:

[0005] A river flow monitoring method integrating multi-source data comprises the following steps:

[0006] S1. In the river area of ​​the study area, a high-precision sensor group is used to collect river flow data and river water level data, and a drone is used to collect river image data;

[0007] The river channel image data includes the river channel shape and the river channel centerline;

[0008] S2, using Kalman filtering method to filter the river flow data and river water level data to obtain filtered river flow data and river water level data;

[0009] S3, using the Canny edge detection method to perform edge detection on the river channel image data to obtain the river channel boundary information;

[0010] S4. Based on the river boundary information, a cubic spline interpolation method is used to perform interpolation processing to obtain a smooth first boundary line and a smooth second boundary line, and the filtered river water level data and the river flow velocity data are combined to perform data fusion to calculate the actual flow of the river;

[0011] S5. Determine whether the actual flow of the river is greater than the preset value of the river flow. If so, send an alarm message; otherwise, store the data and upload it to the display terminal.

[0012] Furthermore, step S1 specifically includes:

[0013] S11. In the river area of ​​the study area, a drone carrying a high-definition camera is used to fly along the river area to collect river image data including the river shape and river water level;

[0014] S12. In the river area of ​​the study area, water level sensors are deployed to collect water level data of the river, and flow velocity sensors are deployed to collect flow velocity data of the river;

[0015] Among them, the water level sensor is arranged near the center line of the river, where the water flow is gentle and not easily affected by local terrain or obstacles;

[0016] The flow velocity sensors are arranged at equal intervals in the longitudinal direction of the river channel and at positions that are not easily affected by local terrain or obstacles.

[0017] Furthermore, step S2 specifically includes:

[0018] S21. Define the system state equation of river flow data and river water level data, namely:

[0019] x k =Ax k-1 +ω k-1

[0020] Among them, x k represents the system state vector at the kth moment, A represents the state transfer matrix, ω k-1 represents the process noise at the k-1th moment;

[0021] S22. Based on the system state equation of river flow data and river water level data, the observation equation of river flow velocity data and river water level data is defined, namely:

[0022] z k =Hx k +μ k

[0023] Among them, z k represents the observation vector at the kth moment, H represents the observation matrix, μ k represents the observation noise at the kth moment;

[0024] S23. Initialize the covariance matrix P 0 , system state vector x 0 , process noise covariance matrix Q, observation noise covariance matrix R;

[0025] S24. If the current time is k, predict the system prior state estimate value at the current time k and the prior covariance matrix at the current time k, that is:

[0026]

[0027] in, Table 1 shows the system prior state estimate at the kth moment, represents the estimated value of the system state at the k-1th moment, represents the prior covariance matrix at the kth moment, P k-1 represents the posterior covariance matrix at the k-1th moment, and T represents the transpose;

[0028] S25, based on the prior covariance matrix of the current time k, calculate the Kalman gain matrix, that is:

[0029]

[0030] Among them, K k represents the Kalman gain matrix at the kth moment;

[0031] S26. Based on the Kalman gain matrix, update the estimated value of the system state at the current time k, that is:

[0032]

[0033] in, represents the estimated value of the system state at the kth moment;

[0034] S27, based on the Kalman gain matrix, update the posterior covariance matrix of the current time k, that is:

[0035]

[0036] Among them, P k represents the posterior covariance matrix at the kth moment, and I represents the identity matrix;

[0037] S28. Repeat steps S24-S27 until the river flow data and river water level data at all times are processed to obtain the observed values ​​of the filtered river flow velocity data and river water level data, i.e., the filtered river flow velocity data and the filtered river water level data.

[0038] Furthermore, step S3 specifically includes:

[0039] S31, graying the river channel image data to obtain a gray image, and filtering the gray image using a Gaussian filter method to obtain a filtered gray image;

[0040] S32, using the Sobel operator to calculate the gradient values ​​of the filtered grayscale image in the horizontal and vertical directions to obtain the gradient intensity and direction of each pixel;

[0041] S33, based on the gradient strength and direction of each pixel, the edge of the filtered grayscale image is refined by using a non-maximum suppression method, and the gradient values ​​of the two neighboring pixels of the current pixel in its gradient direction are compared to determine whether the gradient value of the current pixel is the largest. If so, the current pixel is retained as a candidate edge point, otherwise, the current pixel is set to 0, that is, the current pixel is eliminated;

[0042] S34, using a dual threshold method to detect candidate edge points, by setting a high threshold and a low threshold, determine whether the candidate edge point is higher than the high threshold, if so, mark it as a strong edge point, otherwise, execute step S35;

[0043] S35, determining whether the candidate edge point is lower than the low threshold, if so, marking it as a non-edge point, otherwise, if the candidate edge point is between the high threshold and the low threshold, marking it as a weak edge point;

[0044] S36, checking the neighborhood pixels of the weak edge point, if the weak edge point is surrounded by at least one strong edge point, the weak edge point is taken as a strong edge point, and all the strong edge points are connected to obtain the river boundary information.

[0045] Furthermore, step S4 specifically includes:

[0046] S41, based on the river boundary information, obtaining a first boundary line and a second boundary line of the river;

[0047] S42, using a cubic spline interpolation method to interpolate the first boundary line and the second boundary line to obtain a smooth first boundary line and a smooth second boundary line;

[0048] S43, obtaining the position of the water level sensor near the center line of the river channel, and using the position of the water level sensor as the position of the point to be measured;

[0049] S44, draw a perpendicular line along the direction of the smooth first boundary line and the smooth second boundary line at each point to be measured, obtain a first intersection point and a second intersection point of the perpendicular line with the smooth first boundary line and the smooth second boundary line, and use the Euclidean distance between the first intersection point and the second intersection point as the actual river channel width of the point to be measured, that is:

[0050]

[0051] Among them, w i represents the actual river width of the i-th measured point, x 1 ,y 1 Represents the horizontal and vertical coordinates of the first intersection point, x 2 ,y 2 represents the abscissa and ordinate of the second intersection point;

[0052] S45. According to the filtered river water level data, the actual water level at the location of the water level sensor is obtained, and combined with the actual river width at the point to be measured, the actual cross-sectional area of ​​the river at the point to be measured is calculated, that is:

[0053] h i =d i × i

[0054] Among them, h i represents the actual cross-sectional area of ​​the river at the i-th measured point, d i Indicates the actual water level height of the water level sensor at the i-th test point;

[0055] S46, obtaining the actual river flow velocity at the location of the flow velocity sensor according to the filtered river flow velocity data, and taking the actual river flow velocity at the location of the flow velocity sensor closest to the test point as the actual river flow velocity at the test point;

[0056] S47, based on the actual river flow velocity at the measured point and the actual cross-sectional area of ​​the river at the measured point, calculate the river flow at the measured point, that is:

[0057] Q i =v i ×h i

[0058] Among them, Q i represents the river flow at the i-th test point, v i represents the actual river flow velocity at the i-th measured point;

[0059] S48. According to the river flow at the test point, the average value of the river flow at all the test points is calculated to obtain the actual flow of the river, that is:

[0060]

[0061] Among them, Q represents the river flow, and n represents the number of points to be measured.

[0062] A river flow monitoring system integrating multi-source data, comprising:

[0063] The multi-source data acquisition module is used to collect river flow data and river water level data in the river area of ​​the study area using a high-precision sensor group, and to collect river image data using drones;

[0064] A multi-source data fusion module is used to receive river flow data, river water level data and river image data, and perform data preprocessing on the river flow data and river water level data to obtain filtered river flow data and river water level data, perform edge detection on the river water level data, obtain river boundary information, perform data fusion based on the filtered river flow data, river water level data and river boundary information, and calculate the actual flow of the river;

[0065] The early warning module is used to receive the actual flow of the river. If the calibrated river flow is greater than the preset value of the river flow, an alarm message is sent.

[0066] Furthermore, the multi-source data acquisition module includes a drone subsystem and a sensor subsystem;

[0067] The UAV subsystem is used to fly along the river channel area in the study area using a UAV carrying a high-definition camera to collect river channel image data including the river channel shape and river channel water level line;

[0068] The sensor subsystem is used to collect water level data of the river channel using a water level sensor and to collect flow velocity data of the river channel using a flow velocity sensor in the river channel area of ​​the study area.

[0069] Furthermore, the multi-source data fusion module includes a data preprocessing module, an edge feature extraction module, and a data fusion module;

[0070] A data preprocessing module is used to receive the water level data and the flow velocity data of the river and perform filtering processing to obtain filtered river flow data and river water level data;

[0071] The edge feature extraction module is used to receive the river channel image data and perform edge detection using the Canny edge detection method to obtain the river channel boundary information;

[0072] The data fusion module is used to receive the filtered river flow data, river water level data, and river boundary information, and perform data fusion to obtain the actual flow of the river.

[0073] Furthermore, it also includes a flow correction module, which is used to calibrate the actual flow of the river based on the actual flow of the river, using the historically monitored river flow and adopting a random forest model to obtain a calibrated river flow.

[0074] Furthermore, based on the actual flow of the river, the random forest model is used to calibrate the actual flow of the river using the historically monitored river flow. The specific process of obtaining the calibrated river flow is as follows:

[0075] The random forest model is used to train the model with the historical monitored river flow as the training set, and the trained random forest model is used to correct the calculated actual river flow to obtain the corrected river flow;

[0076] Among them, the historically monitored river flow includes river flow measurement values ​​at several historical time points.

[0077] The present invention has the following beneficial effects:

[0078] 1. The method for monitoring river flow by fusing multi-source data proposed in the present invention fuses the river flow data and river water level data collected by the sensor group with the river image data collected by the drone to calculate the actual flow of the river, thereby improving the accuracy of river flow monitoring;

[0079] 2. The river flow monitoring system proposed in the present invention integrates multi-source data, which combines high-precision sensor groups and multi-source data collected by drones, namely, river flow data, water level data and image data, so that the system can obtain more comprehensive and accurate information; the sensor subsystem provides accurate flow rate and water level data, while the drone subsystem captures subtle changes in river shape and water level through high-definition images and obtains river boundary information. The combination of the two significantly improves data accuracy;

[0080] 3. The introduction of early warning module and flow correction module at the same time improves the timeliness and accuracy of early warning response. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 A schematic diagram of the flow chart of a river flow monitoring method integrating multi-source data proposed by the present invention;

[0082] Figure 2 This is a flow chart of a river flow monitoring system that integrates multi-source data proposed by the present invention. DETAILED DESCRIPTION

[0083] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0084] like Figure 1 As shown, a river flow monitoring method integrating multi-source data includes the following steps S1-S5:

[0085] S1. In the river area of ​​the study area, a high-precision sensor group is used to collect river flow data and river water level data, and a drone is used to collect river image data; the river image data includes the river shape and the river centerline.

[0086] In this embodiment, through a high-precision sensor group and drone technology, real-time monitoring of key parameters such as river flow, water level and river shape in the study area is achieved, providing accurate data support for water resource management, flood prevention and disaster reduction, river management, etc.; at the same time, combined with drone technology, it can cover complex terrain and dangerous areas that are difficult to reach by traditional monitoring methods, reduce manpower investment, improve monitoring efficiency and ensure personnel safety.

[0087] Specifically, step S1 includes S11-S12:

[0088] S11. In the river area of ​​the study area, a drone carrying a high-definition camera is used to fly along the river area to collect river image data including the river shape and river water level.

[0089] In this embodiment, the drone is equipped with a high-definition camera and flies along a preset route to take all-round photos of the river. Through image processing technology, it can extract information such as the river shape, river centerline or water level line, and provide basic data for river morphology analysis and flood simulation. In this embodiment, the river boundary is obtained by extracting the edge information of the river image data, so that the river boundary information can be integrated with the water level data and flow velocity data of the river in subsequent steps to calculate the river flow, so as to improve the calculation accuracy of the river flow.

[0090] S12. In the river area of ​​the study area, water level data of the river is collected by deploying water level sensors, and flow velocity data of the river is collected by deploying flow velocity sensors.

[0091] Among them, the water level sensor is arranged near the center line of the river, where the water flow is gentle and not easily affected by local terrain or obstacles.

[0092] In this embodiment, the water level sensor can select a pressure water level sensor or an ultrasonic water level sensor to accurately measure the water level height near the center line of the river channel, and at the same time select a location where the water flow is slow and not easily disturbed to ensure the accuracy and stability of the measurement data.

[0093] The flow velocity sensors are arranged at equal intervals in the longitudinal direction of the river channel and at positions that are not easily affected by local terrain or obstacles.

[0094] In this embodiment, the flow sensor can be a laser Doppler flow sensor or a differential pressure flow sensor, etc., which are evenly spaced in the river channel through the Doppler effect or electromagnetic induction principles to measure the water flow velocity. In addition, the evenly spaced arrangement helps to analyze the flow velocity distribution and evaluate the dynamic characteristics of the water flow.

[0095] S2. Use Kalman filtering method to filter the river flow data and river water level data to obtain filtered river flow data and river water level data.

[0096] In this embodiment, the Kalman filter method is introduced to filter the river flow data and the river water level data, which can effectively filter the river flow and water level data, thereby removing noise and interference and improving the accuracy and reliability of the data. Such a design is of great significance to river monitoring, flood warning, and water resources management.

[0097] Specifically, step S2 specifically includes S21-S28:

[0098] S21. Define the system state equation of river flow data and river water level data, namely:

[0099] x k =Ax k-1 +ω k-1

[0100] Among them, x k represents the system state vector at the kth moment, A represents the state transfer matrix, ω k-1 represents the process noise at the k-1th moment.

[0101] In this embodiment, the system state vector includes river flow data and river water level data; wherein, the system state equation describes the dynamic change of the system state over time, the state transfer matrix reflects the relationship between the state variables within the system, and the process noise is used to simulate the uncertainty of the system model and external random interference; therefore, the equation can predict the system state at the next moment and provide a basis for filtering.

[0102] S22. Based on the system state equation of river flow data and river water level data, the observation equation of river flow velocity data and river water level data is defined, namely:

[0103] z k =Hx k +μ k

[0104] Among them, z k represents the observation vector at the kth moment, H represents the observation matrix, μ k represents the observation noise at the kth moment.

[0105] In this embodiment, the observation vector includes direct river flow and water level observation data; wherein, the observation equation establishes the relationship between the system state and the observation data; the observation matrix describes how to extract the observation data from the system state vector; the observation noise reflects the errors and uncertainties in the observation process; therefore, the observation equation allows the actual observation data to be compared with the predicted system state, thereby performing error correction.

[0106] S23. Initialize the covariance matrix P 0 , system state vector x 0 , process noise covariance matrix Q, observation noise covariance matrix R.

[0107] In this embodiment, the purpose of initialization is to provide a basis for predicting and updating states in subsequent steps.

[0108] S24. If the current time is k, predict the system prior state estimate value at the current time k and the prior covariance matrix at the current time k, that is:

[0109]

[0110] in, Table 1 shows the system prior state estimate at the kth moment, represents the estimated value of the system state at the k-1th moment, represents the prior covariance matrix at the kth moment, P k-1 represents the posterior covariance matrix at the k-1th moment, and T represents the transpose.

[0111] In this embodiment, the state estimate and state transfer matrix at the previous moment are used to predict the prior state estimate and prior covariance matrix at the current moment, providing a preliminary estimate of the state at the current moment and providing a basis for subsequent observation updates.

[0112] S25, based on the prior covariance matrix of the current time k, calculate the Kalman gain matrix, that is:

[0113]

[0114] Among them, K k represents the Kalman gain matrix at the kth moment.

[0115] S26. Based on the Kalman gain matrix, update the estimated value of the system state at the current time k, that is:

[0116]

[0117] in, Represents the estimated value of the system state at the kth moment.

[0118] S27, based on the Kalman gain matrix, update the posterior covariance matrix of the current time k, that is:

[0119]

[0120] Among them, P k represents the posterior covariance matrix at the kth moment, and I represents the identity matrix.

[0121] S28. Repeat steps S24-S27 until the river flow data and river water level data at all times are processed to obtain the observed values ​​of the filtered river flow velocity data and river water level data, i.e., the filtered river flow velocity data and the filtered river water level data.

[0122] In this embodiment, steps S25-S28 use observation data and Kalman gain to correct the predicted state estimate, and update the covariance matrix to reflect the uncertainty of the estimate, that is, by fusing the predicted and observed information, a more accurate and reliable system state estimate is obtained, thereby effectively filtering the river flow and water level data to remove noise and interference, thereby improving the accuracy and reliability of the data.

[0123] S3. Use the Canny edge detection method to perform edge detection on the river channel image data to obtain the river channel boundary information.

[0124] In this embodiment, the Canny edge detection method is used for image processing to effectively extract boundary information from the river channel image, thereby providing a basis for calculating the river channel flow in the subsequent step.

[0125] Specifically, step S3 specifically includes S31-S36:

[0126] S31, graying the river channel image data to obtain a gray image, and filtering the gray image using a Gaussian filter method to obtain a filtered gray image.

[0127] In this embodiment, the purpose of grayscale processing of the river image data is to remove the color information of the image and retain only the brightness information, thereby simplifying the image data and reducing the amount of calculation for subsequent processing, while retaining sufficient image information for edge detection; the purpose of filtering the grayscale image using the Gaussian filtering method is to remove noise in the image and reduce the false detection rate during edge detection. At the same time, the filtered image is smoother and the noise is reduced.

[0128] S32, using the Sobel operator to calculate the gradient values ​​of the filtered grayscale image in the horizontal and vertical directions to obtain the gradient intensity and direction of each pixel.

[0129] In this embodiment, the Sobel operator is used to calculate the gradients of the image in the horizontal and vertical directions respectively; the Sobel operator can be implemented by two 3×3 convolution kernels, one for the horizontal gradient and the other for the vertical gradient, so as to detect the edge strength and direction at each pixel in the filtered grayscale image, so as to provide a basis for subsequent non-maximum suppression.

[0130] S33. Based on the gradient strength and direction of each pixel, the edge of the filtered grayscale image is refined using the non-maximum suppression method, and the gradient values ​​of the two neighboring pixels of the current pixel in its gradient direction are compared to determine whether the gradient value of the current pixel is the largest. If so, the current pixel is retained as a candidate edge point, otherwise, the current pixel is set to 0, that is, the current pixel is eliminated.

[0131] In this embodiment, the purpose of using the non-maximum suppression method to refine the edge of the filtered grayscale image is to refine the edge, remove non-edge points, and retain the most likely edge points, thereby reducing edge redundancy and making the edge clearer and more accurate.

[0132] S34, using the dual threshold method to detect candidate edge points, by setting a high threshold and a low threshold, determine whether the candidate edge point is higher than the high threshold, if so, mark it as a strong edge point, otherwise, execute step S35.

[0133] S35. Determine whether the candidate edge point is below the low threshold. If so, mark it as a non-edge point. Otherwise, if the candidate edge point is between the high threshold and the low threshold, mark it as a weak edge point.

[0134] S36, checking the neighborhood pixels of the weak edge point, if the weak edge point is surrounded by at least one strong edge point, the weak edge point is taken as a strong edge point, and all the strong edge points are connected to obtain the river boundary information.

[0135] In this embodiment, steps S34-S36 further screen edge points through the double threshold method to improve the accuracy and robustness of edge detection, thereby obtaining the final edge detection result, which includes strong edge points and weak edge points confirmed by neighborhood relationships. These edge points together constitute the boundary information of the river channel.

[0136] S4. Based on the river boundary information, cubic spline interpolation method is used to perform interpolation processing to obtain a smooth first boundary line and a smooth second boundary line, and the filtered river water level data and river flow velocity data are combined for data fusion to calculate the actual flow of the river.

[0137] In this embodiment, accurate calculation of river flow is achieved by combining river boundary information, water level data, and flow rate data.

[0138] Specifically, step S4 specifically includes S41-S48:

[0139] S41. Based on the river boundary information, obtain a first boundary line and a second boundary line of the river.

[0140] S42: Use a cubic spline interpolation method to interpolate the first boundary line and the second boundary line to obtain a smooth first boundary line and a smooth second boundary line.

[0141] In this embodiment, steps S41-S42 use cubic spline interpolation to interpolate the first boundary line and the second boundary line of the river channel to generate a smooth boundary line, thereby reducing the discreteness of the boundary line through interpolation and improving the accuracy of subsequent calculations.

[0142] S43, obtaining the position of the water level sensor near the center line of the river channel, and using the position of the water level sensor as the position of the point to be measured.

[0143] S44, draw a perpendicular line along the direction of the smooth first boundary line and the smooth second boundary line at each point to be measured, obtain a first intersection point and a second intersection point of the perpendicular line with the smooth first boundary line and the smooth second boundary line, and use the Euclidean distance between the first intersection point and the second intersection point as the actual river channel width of the point to be measured, that is:

[0144]

[0145] Among them, w i represents the actual river width of the i-th measured point, x 1 ,y 1 Represents the horizontal and vertical coordinates of the first intersection point, x 2 ,y 2 Indicates the horizontal and vertical coordinates of the second intersection point.

[0146] In this embodiment, steps S43-S44 first determine the position of the water level sensor near the center line of the river channel as the test point, then draw a perpendicular line along the direction of the smooth boundary line at each test point, find the first and second intersection points of the perpendicular line with the smooth first boundary line and the smooth second boundary line, and calculate the Euclidean distance between the first intersection point and the second intersection point as the river channel width, thereby providing a basis for the subsequent steps of calculating the cross-sectional area of ​​the river channel at each test point.

[0147] S45. According to the filtered river water level data, the actual water level at the location of the water level sensor is obtained, and combined with the actual river width at the point to be measured, the actual cross-sectional area of ​​the river at the point to be measured is calculated, that is:

[0148] h i =d i × i

[0149] Among them, h i represents the actual cross-sectional area of ​​the river at the i-th measured point, d i Represents the actual water level height of the water level sensor at the i-th test point.

[0150] S46. According to the filtered river flow velocity data, the actual river flow velocity at the location of the flow velocity sensor is obtained, and the actual river flow velocity at the location of the flow velocity sensor closest to the test point is used as the actual river flow velocity of the test point.

[0151] S47, based on the actual river flow velocity at the measured point and the actual cross-sectional area of ​​the river at the measured point, calculate the river flow at the measured point, that is:

[0152] Q i =v i ×h i

[0153] Among them, Q i represents the river flow at the i-th test point, v i represents the actual river flow velocity at the i-th measured point;

[0154] S48. According to the river flow at the test point, the average value of the river flow at all the test points is calculated to obtain the actual flow of the river, that is:

[0155]

[0156] Among them, Q represents the river flow, and n represents the number of points to be measured.

[0157] In this embodiment, the average value is calculated as the actual flow of the river to reflect the overall water flow of the river, thereby providing an important basis for water resource management, flood prevention and disaster reduction, etc.

[0158] S5. Determine whether the actual flow of the river is greater than the preset value of the river flow. If so, send an alarm message; otherwise, store the data and upload it to the display terminal.

[0159] In this embodiment, when the actual flow of the river is greater than the preset value of the river flow, the system can promptly and quickly identify and send an alarm message, which is crucial for flood prevention and disaster reduction, because excessive water flow may cause natural disasters such as floods and riverbank erosion. Timely warning can help relevant departments and the public take countermeasures in advance to reduce losses; in addition, by storing actual flow data and uploading it to the display terminal, a long-term and continuous flow monitoring record can be formed; these data not only provide training data for subsequent flow prediction models to improve prediction accuracy, but also provide a scientific basis for water resources management, ecological environment protection, etc., and support more accurate and scientific decision-making; at the same time, understanding the real-time flow of the river will help to more reasonably allocate and utilize water resources; for example, in the dry season, irrigation plans can be adjusted according to flow data; in the flood season, reservoirs, gates and other facilities can be dispatched in advance to reduce downstream pressure; and the establishment of the alarm system is not limited to instant notification when the flow exceeds the standard, but also can quickly launch emergency plans in emergency situations and coordinate the forces of all parties to carry out rescue and disaster relief; at the same time, through data analysis, potential disaster risk points can be identified and reinforced or protected in advance.

[0160] like Figure 2 As shown, a river flow monitoring system integrating multi-source data includes:

[0161] The multi-source data acquisition module is used to collect river flow data and river water level data in the river area of ​​the study area using a high-precision sensor group, and to collect river image data using drones.

[0162] Specifically, the multi-source data acquisition module includes a drone subsystem and a sensor subsystem; the drone subsystem is used to fly along the river area in the study area using a drone carrying a high-definition camera to collect river image data including the river shape and river water level line; the sensor subsystem is used to collect river water level data using a water level sensor and collect river flow velocity data using a flow velocity sensor in the river area of ​​the study area.

[0163] The multi-source data fusion module is used to receive river flow data, river water level data and river image data, and perform data preprocessing on the river flow data and river water level data to obtain filtered river flow data and river water level data, perform edge detection on the river water level data, obtain river boundary information, perform data fusion based on the filtered river flow data, river water level data and river boundary information, and calculate the actual flow of the river.

[0164] Specifically, the multi-source data fusion module includes a data preprocessing module, an edge feature extraction module, and a data fusion module; the data preprocessing module is used to receive the water level data and the flow velocity data of the river and perform filtering processing to obtain filtered river flow data and river water level data; the edge feature extraction module is used to receive the river image data and use the Canny edge detection method to perform edge detection to obtain river boundary information; the data fusion module is used to receive the filtered river flow data, river water level data, and river boundary information, and perform data fusion to obtain the actual flow of the river.

[0165] The early warning module is used to receive the actual flow of the river. If the calibrated river flow is greater than the preset value of the river flow, an alarm message is sent.

[0166] Specifically, it also includes a flow correction module, which is used to calibrate the actual flow of the river based on the actual flow of the river, using the historically monitored river flow and adopting a random forest model to obtain a calibrated river flow.

[0167] Specifically, based on the actual flow of the river, using the historically monitored river flow, the random forest model is used to calibrate the actual flow of the river. The specific process of obtaining the calibrated river flow is as follows:

[0168] The random forest model is used to train the model with the historically monitored river flow as the training set, and the trained random forest model is used to correct the calculated actual river flow to obtain the corrected river flow; wherein the historically monitored river flow includes the river flow measurement values ​​at several historical time points.

[0169] In this embodiment, by introducing a flow correction module and using a random forest model to calibrate the calculated actual flow of the river, the flow data deviation caused by measurement errors, environmental interference and other factors can be effectively reduced, thereby improving the accuracy and reliability of the data; at the same time, the corrected river flow data is used as the input of the subsequent prediction model, which can improve the adaptability of the prediction model to real-time flow changes, making the prediction results closer to the actual situation, thereby enhancing the robustness of the prediction model; in addition, based on more accurate flow data, water resource managers can make more scientific and reasonable decisions, such as adjusting the reservoir water storage capacity, formulating irrigation plans, etc., to achieve optimal allocation and efficient utilization of water resources.

[0170] In summary, the river flow monitoring system that integrates multi-source data proposed in the present invention has the following advantages: (1) Improved data accuracy and reliability; by combining a high-precision sensor group and multi-source data collected by drones (i.e., river flow data, water level data, and image data), the system can obtain more comprehensive and accurate information; the sensor subsystem provides accurate flow rate and water level data, while the drone subsystem captures subtle changes in river shape and water level lines through high-definition images and obtains river boundary information. The combination of the two significantly improves data accuracy; (2) Enhanced data fusion and processing capabilities; the multi-source data fusion module effectively integrates data from different sources through steps such as data preprocessing, edge feature extraction, and data fusion; the data preprocessing module reduces noise and interference and improves data quality; the edge feature extraction module uses image processing technology to accurately identify river boundaries. , which provides an important basis for flow calculation; the data fusion module calculates a more accurate actual river flow based on these processed data; (3) improves the timeliness and accuracy of the early warning response; the early warning module can monitor the actual flow of the river in real time and compare it with the preset flow threshold; once the flow exceeds the preset value, the system immediately sends an alarm message, providing valuable early warning time for flood prevention and disaster reduction; in addition, through the introduction of the flow correction module, the random forest model is used to calibrate the initial flow, which further improves the accuracy and reliability of the early warning; (4) enhances the robustness and adaptability of the system; the system enhances the robustness and adaptability of the system by fusing multi-source data and adopting advanced data processing and fusion technology; even when some sensor data is missing or abnormal, the system can still supplement and correct it through other data sources to ensure the continuity and accuracy of flow monitoring.

[0171] The present invention uses specific embodiments to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

[0172] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A river flow monitoring method integrating multi-source data, characterized in that: The following steps are involved: S1. In the river area of ​​the study area, a high-precision sensor group is used to collect river flow data and river water level data, and a drone is used to collect river image data; The river channel image data includes the river channel shape and the river channel centerline; S2, using Kalman filtering method to filter the river flow data and river water level data to obtain filtered river flow data and river water level data; S3, using the Canny edge detection method to perform edge detection on the river channel image data to obtain the river channel boundary information; S4. Based on the river boundary information, a cubic spline interpolation method is used to perform interpolation processing to obtain a smooth first boundary line and a smooth second boundary line, and the filtered river water level data and the river flow velocity data are combined to perform data fusion to calculate the actual flow of the river; S5. Determine whether the actual flow of the river is greater than the preset value of the river flow. If so, send an alarm message; otherwise, store the data and upload it to the display terminal.

2. The method for monitoring river flow by integrating multi-source data according to claim 1, characterized in that: Step S1 specifically includes: S11. In the river area of ​​the study area, a drone carrying a high-definition camera is used to fly along the river area to collect river image data including the river shape and river water level; S12. In the river area of ​​the study area, water level sensors are deployed to collect water level data of the river, and flow velocity sensors are deployed to collect flow velocity data of the river; Among them, the water level sensor is arranged near the center line of the river, where the water flow is gentle and not easily affected by local terrain or obstacles; The flow velocity sensors are arranged at equal intervals in the longitudinal direction of the river channel and at positions that are not easily affected by local terrain or obstacles.

3. The method for monitoring river flow by integrating multi-source data according to claim 2 is characterized in that: Step S2 specifically includes: S21. Define the system state equation of river flow data and river water level data, namely: x k =Ax k-1 +ω k-1 Among them, x k represents the system state vector at the kth moment, A represents the state transfer matrix, ω k-1 represents the process noise at the k-1th moment; S22. Based on the system state equation of river flow data and river water level data, the observation equation of river flow velocity data and river water level data is defined, namely: z k =Hx k +μ k Among them, z k represents the observation vector at the kth moment, H represents the observation matrix, μ k represents the observation noise at the kth moment; S23, initialize the covariance matrix P0, the system state vector x0, the process noise covariance matrix Q, and the observation noise covariance matrix R; S24. If the current time is k, predict the system prior state estimate value at the current time k and the prior covariance matrix at the current time k, that is: in, Table 1 shows the system prior state estimate at the kth moment, represents the estimated value of the system state at the k-1th moment, represents the prior covariance matrix at the kth moment, P k-1 represents the posterior covariance matrix at the k-1th moment, and T represents the transpose; S25, based on the prior covariance matrix of the current time k, calculate the Kalman gain matrix, that is: Among them, K k represents the Kalman gain matrix at the kth moment; S26. Based on the Kalman gain matrix, update the estimated value of the system state at the current time k, that is: in, represents the estimated value of the system state at the kth moment; S27, based on the Kalman gain matrix, update the posterior covariance matrix of the current time k, that is: Among them, p k represents the posterior covariance matrix at the kth moment, and I represents the identity matrix; S28. Repeat steps S24-S27 until the river flow data and river water level data at all times are processed to obtain the observed values ​​of the filtered river flow velocity data and river water level data, i.e., the filtered river flow velocity data and the filtered river water level data.

4. The method for monitoring river flow by integrating multi-source data according to claim 3 is characterized in that: Step S3 specifically includes: S31, graying the river channel image data to obtain a gray image, and filtering the gray image using a Gaussian filter method to obtain a filtered gray image; S32, using the Sobel operator to calculate the gradient values ​​of the filtered grayscale image in the horizontal and vertical directions to obtain the gradient intensity and direction of each pixel; S33, based on the gradient strength and direction of each pixel, the edge of the filtered grayscale image is refined by using a non-maximum suppression method, and the gradient values ​​of the two neighboring pixels of the current pixel in its gradient direction are compared to determine whether the gradient value of the current pixel is the largest. If so, the current pixel is retained as a candidate edge point, otherwise, the current pixel is set to 0, that is, the current pixel is eliminated; S34, using a dual threshold method to detect candidate edge points, by setting a high threshold and a low threshold, determine whether the candidate edge point is higher than the high threshold, if so, mark it as a strong edge point, otherwise, execute step S35; S35, determining whether the candidate edge point is lower than the low threshold, if so, marking it as a non-edge point, otherwise, if the candidate edge point is between the high threshold and the low threshold, marking it as a weak edge point; S36, checking the neighborhood pixels of the weak edge point, if the weak edge point is surrounded by at least one strong edge point, the weak edge point is taken as a strong edge point, and all the strong edge points are connected to obtain the river boundary information.

5. The method for monitoring river flow by integrating multi-source data according to claim 4 is characterized in that: Step S4 specifically includes: S41, based on the river boundary information, obtaining a first boundary line and a second boundary line of the river; S42, using a cubic spline interpolation method to interpolate the first boundary line and the second boundary line to obtain a smooth first boundary line and a smooth second boundary line; S43, obtaining the position of the water level sensor near the center line of the river channel, and using the position of the water level sensor as the position of the point to be measured; S44, draw a perpendicular line along the direction of the smooth first boundary line and the smooth second boundary line at each point to be measured, obtain a first intersection point and a second intersection point of the perpendicular line with the smooth first boundary line and the smooth second boundary line, and use the Euclidean distance between the first intersection point and the second intersection point as the actual river channel width of the point to be measured, that is: Among them, w i represents the actual river width of the i-th measured point, x1 and y1 represent the abscissa and ordinate of the first intersection, and x2 and y2 represent the abscissa and ordinate of the second intersection; S45. According to the filtered river water level data, the actual water level at the location of the water level sensor is obtained, and combined with the actual river width at the point to be measured, the actual cross-sectional area of ​​the river at the point to be measured is calculated, that is: h i =d i ×w i Among them, h i represents the actual cross-sectional area of ​​the river at the i-th measured point, d i Indicates the actual water level height of the water level sensor at the i-th test point; S46, obtaining the actual river flow velocity at the location of the flow velocity sensor according to the filtered river flow velocity data, and taking the actual river flow velocity at the location of the flow velocity sensor closest to the test point as the actual river flow velocity at the test point; S47, based on the actual river flow velocity at the measured point and the actual cross-sectional area of ​​the river at the measured point, calculate the river flow at the measured point, that is: Q i =v i ×h i Among them, Q i represents the river flow at the i-th test point, v i represents the actual river flow velocity at the i-th measured point; S48. According to the river flow at the test point, the average value of the river flow at all the test points is calculated to obtain the actual flow of the river, that is: Among them, Q represents the river flow, and n represents the number of points to be measured.

6. A river flow monitoring system integrating multi-source data, characterized in that: include: The multi-source data acquisition module is used to collect river flow data and river water level data in the river area of ​​the study area using a high-precision sensor group, and to collect river image data using drones; A multi-source data fusion module is used to receive river flow data, river water level data and river image data, and perform data preprocessing on the river flow data and river water level data to obtain filtered river flow data and river water level data, perform edge detection on the river water level data, obtain river boundary information, perform data fusion based on the filtered river flow data, river water level data and river boundary information, and calculate the actual flow of the river; The early warning module is used to receive the actual flow of the river. If the calibrated river flow is greater than the preset value of the river flow, an alarm message is sent.

7. The method for monitoring river flow by integrating multi-source data according to claim 6 is characterized in that: The multi-source data acquisition module includes the drone subsystem and the sensor subsystem; The UAV subsystem is used to fly along the river channel area in the study area using a UAV carrying a high-definition camera to collect river channel image data including the river channel shape and river channel water level line; The sensor subsystem is used to collect water level data of the river channel using a water level sensor and to collect flow velocity data of the river channel using a flow velocity sensor in the river channel area of ​​the study area.

8. The method for monitoring river flow by integrating multi-source data according to claim 7 is characterized in that: The multi-source data fusion module includes a data preprocessing module, an edge feature extraction module, and a data fusion module; A data preprocessing module is used to receive the water level data and the flow velocity data of the river and perform filtering processing to obtain filtered river flow data and river water level data; The edge feature extraction module is used to receive the river channel image data and perform edge detection using the Canny edge detection method to obtain the river channel boundary information; The data fusion module is used to receive the filtered river flow data, river water level data, and river boundary information, and perform data fusion to obtain the actual flow of the river.

9. The method for monitoring river flow by integrating multi-source data according to claim 8, characterized in that: It also includes a flow correction module for calibrating the actual flow of the river based on the actual flow of the river and using the historically monitored river flow and adopting a random forest model to obtain a calibrated river flow.

10. The method for monitoring river flow by integrating multi-source data according to claim 9, characterized in that: Based on the actual flow of the river, the random forest model is used to calibrate the actual flow of the river using the historical monitored river flow. The specific process of obtaining the calibrated river flow is as follows: The random forest model is used to train the model with the historical monitored river flow as the training set, and the trained random forest model is used to correct the calculated actual river flow to obtain the corrected river flow; Among them, the historically monitored river flow includes river flow measurement values ​​at several historical time points.

Images