A river video water level-flow velocity-flow three-dimensional surface calculation method and device

By constructing a three-dimensional surface calculation model of river video water level-velocity-flow rate, and combining information entropy and Bayesian inference to optimize parameters, the problems of long time consumption and large uncertainty in traditional flow measurement methods are solved, and rapid and accurate flow monitoring and uncertainty analysis are achieved.

CN118762146BActive Publication Date: 2024-11-08WUHAN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411242465.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-05
Publication Date
2024-11-08
Estimated Expiration
2044-09-05

AI Technical Summary

Technical Problem

Traditional flow measurement methods, such as the velocity-area method, are time-consuming, rely on manual operation, and have many sources of error. They are difficult to achieve continuous flow measurement around the clock, and the relationship between a single water level and flow rate has significant uncertainty in uncertainty analysis.

Method used

A three-dimensional surface calculation method based on river video water level, velocity, and flow rate is adopted. This method integrates information entropy theory and variational Bayesian inference to construct a model of the relationship between water level, maximum surface velocity, and flow rate. The model parameters are optimized, and non-contact measurement methods are combined to simplify the flow rate calculation process and reduce sources of error.

Benefits of technology

It enables rapid, real-time monitoring of flow, simplifies the calculation process, accurately reflects flow patterns, adapts to different cross-sectional conditions, and can clearly analyze the sources of uncertainty, thus improving measurement accuracy and reliability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118762146B_ABST
    Figure CN118762146B_ABST
Patent Text Reader

Abstract

The present application relates to the field of hydrological surveys, and specifically discloses a three-dimensional surface calculation method and device for river video water level-velocity-flow, the method comprising the following steps: collecting water level, maximum surface velocity, and flow data sets of the research section; establishing a water level-maximum surface velocity-flow calculation model based on the flow calculation principle, velocity distribution equation, and cross-sectional geometric characteristics; comprehensively considering the measurement error of the observed data and the error distribution of the model deviation, obtaining the expression of the measured flow with respect to the measured maximum surface velocity and the measured water level; using variational inference to approximate the posterior distribution of the simulation parameters, and selecting the optimal parameter combination. Based on the river water level-flow relationship model, the present application introduces the surface velocity to reflect the changes in the complex water flow motion state, and uses variational Bayesian inference to select the distribution of model parameters. The mathematical mechanism and physical meaning of the model are clear, the flow calculation form is concise, the error source is clear, and it can adapt to different cross-sectional conditions.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of hydrological surveys, and in particular to a method and device for calculating a three-dimensional surface of water level, velocity and flow rate of a river video. Background Art

[0002] Flow is one of the most basic and important observation elements in hydrological testing. It can be used as data support for hydrological forecasting, hydrodynamic simulation, and water resources management, and provide decision-making basis for flood control and drought relief, reservoir scheduling, and water resources planning and allocation. The basic method currently used for river flow measurement is the velocity area method. Its principle is to evenly set a number of velocity measuring vertical lines along the cross-section direction to divide the river section into several sub-areas, use a velocity measuring instrument to measure the average velocity of each velocity measuring vertical line, and calculate the water depth and sub-area area at each vertical line according to the measured water level and riverbed bottom elevation. The product of the average velocity of the vertical line and the area of ​​the sub-area is the flow of the sub-area, and the accumulation of the flow of each sub-area is the cross-section flow. The traditional contact flow measurement method uses a flow meter to measure the velocity of the velocity measuring vertical line at different depths to calculate the average velocity of the vertical line. It takes a long time to complete the velocity measurement of the entire section, and it is difficult to capture the instantaneous changes in the flow. It also relies on manual operation and is difficult to measure under conditions such as rainstorms and flash floods, and it is impossible to achieve all-weather uninterrupted continuous flow measurement. The velocity-area method requires frequent measurement of the water depth and velocity of the cross-section, and the calculation and summation of the flow of multiple sub-sections. The calculation process is relatively complicated, and there are many sources of error in the measurement process, including: measurement errors of water level and point velocity, calculation of the average velocity of the vertical line, setting the number of vertical lines, etc. It is difficult to quantify the uncertainty of the test results.

[0003] Compared with flow monitoring, continuous water level monitoring is less difficult to implement. Therefore, when compiling hydrological data, a water level-flow relationship is usually established based on multiple sets of measured water levels and flow values, and then a continuous flow sequence is calculated based on the water level data and the water level-flow relationship. Due to the influence of factors such as backwater caused by changes in downstream boundary conditions and flood fluctuations, the water flow will change in gradient when the cross-section control conditions are relatively stable. Therefore, the water level-flow relationship is usually expressed as a loop curve, and there will be greater uncertainty in inferring the flow using a single water level-flow relationship.

[0004] With the development of non-contact flow measurement technologies such as video and radar, surface velocity monitoring can be realized synchronously with water level monitoring. Since both water level and surface velocity can be used as intuitive expressions of river flow status and flow rate changes, a functional relationship for calculating flow rate based on video water level and surface velocity is constructed on the premise of conforming to physical meaning and mathematical principles, and the uncertainty of the fitting results is quantified. This can effectively simplify the calculation process of flow rate measurement using the velocity-area method. Compared with a single water level-flow relationship, it can better reflect the state changes of water flow and has strong application value and practical significance. Summary of the invention

[0005] In response to the shortcomings of the prior art, the present application provides a three-dimensional surface calculation method and device for river video water level-velocity-flow, which integrates the flow calculation principle, information entropy theory and cross-sectional geometric characteristics to obtain a simplified flow calculation model. On the basis of the river water level-flow relationship model, the surface flow velocity is introduced to reflect the changes in the water flow movement state, and then the variational Bayesian inference is used to optimize the distribution of model parameters. The mathematical mechanism and physical meaning are clear, the calculation process is concise, the error source is clear, and it can adapt to different cross-sectional conditions.

[0006] The present application provides a three-dimensional surface calculation method for river water level-velocity-flow rate using the following technical solution:

[0007] A method for calculating a three-dimensional surface of water level, velocity and flow rate of a river video includes the following steps:

[0008] Collect hydrological test data: Collect data set S at the section of the river to be studied, S={ };in, is the water level measurement, is the maximum surface velocity measurement, is the flow measurement;

[0009] Construct a water level-maximum surface velocity-flow calculation model: express the flow of the river section as the product of the average flow velocity of the section and the cross-sectional area, establish the relationship between the average flow velocity of the section and the maximum surface flow velocity, and the relationship between the cross-sectional area and the water level; express the average flow velocity of the section by the maximum surface flow velocity, and express the cross-sectional area by the water level, and obtain the general expression of the water level-maximum surface flow velocity-flow relationship model:

[0010]

[0011] in, Q For flow, V is the maximum surface velocity, H is the water level, a , b , c are all model parameters;

[0012] The relationship between the measured flow rate and the measured maximum surface velocity and the measured water level is established by comprehensively considering the measurement error of the observed data and the error distribution of the model bias;

[0013] Variational Bayesian inference is used to approximate the posterior distribution of simulation parameters, and the optimal model parameter combination is selected to obtain the water level-maximum surface velocity-discharge calculation model after parameter optimization.

[0014] Calculate flow rate: Use the water level-maximum surface velocity-flow calculation model with optimized parameters to calculate the flow rate based on the measured water level and the measured maximum surface velocity.

[0015] Furthermore, in the step of collecting hydrological test data, the means of collecting water level and maximum surface flow velocity are non-contact measurement means, including image measurement means, and the measurement means of water level and maximum surface flow velocity remain consistent in the model establishment stage and the model application stage; the flow measurement method includes flow meter and ADCP instrument.

[0016] Furthermore, in the step of constructing the water level-maximum surface flow velocity-flow rate calculation model, the method of establishing the relationship between the cross-sectional average flow velocity and the maximum surface flow velocity includes:

[0017] The vertical line where the maximum flow velocity and the maximum surface flow velocity of the cross section are located is taken as the y-axis, and the transverse direction of the cross section is taken as the z-axis. The relationship between the flow velocity and the depth on the y-axis and the cumulative distribution function of the flow velocity are established.

[0018] Establish the relationship between the maximum flow velocity of the cross section and the surface flow velocity;

[0019] Based on the velocity distribution equation based on information entropy theory, the maximum flow velocity of the cross section is converted into the average flow velocity of the cross section;

[0020] Based on the relationship between the maximum flow velocity of the cross section and the surface flow velocity, and the relationship between the maximum flow velocity of the cross section and the average flow velocity of the cross section, the relationship between the average flow velocity of the cross section and the maximum surface flow velocity is obtained.

[0021] Furthermore, in the step of constructing the water level-maximum surface velocity-flow calculation model, the relationship between the cross-sectional area and the water level is constructed according to the cross-sectional geometric characteristics. The cases where the cross-sectional shapes of the river are rectangular, trapezoidal, and parabolic are analyzed respectively. Combined with the relationship between the average flow velocity of the cross section and the maximum surface velocity, the relationship between the flow, the water level, and the maximum surface velocity is obtained as follows: .

[0022] Furthermore, considering the measurement error of the observation data and the error distribution of the model bias, the method of obtaining the expression of the measured flow rate with respect to the measured maximum surface velocity and the measured water level includes:

[0023] Observed values ​​of water level, maximum surface velocity, and flow rate is the true value The estimated value of satisfies the following relationship with the true value:

[0024]

[0025] in, are the observed values ​​of water level, maximum surface velocity and flow rate, respectively. are the true values ​​of water level, maximum surface velocity, and flow rate, respectively. i is the number of the observation data; is the error distribution of the maximum surface velocity observation, which satisfies the mean of 0 and the standard deviation of u V Normal distribution of is the error distribution of the flow observation value, which satisfies the mean of 0 and the standard deviation of u Q Normal distribution; N() represents normal distribution;

[0026] The water level-maximum surface velocity-discharge relationship model writing In the form of, considering the model bias:

[0027]

[0028] in, i are model parameters, e i f is the error distribution of the model simulation, which satisfies the mean of 0 and the standard deviation of s f Normal distribution of

[0029] Assuming that the error components are independent of each other, the expression of the measured flow rate with respect to the measured maximum surface velocity and the measured water level is obtained:

[0030] .

[0031] Further, the method of using variational Bayes to infer the posterior distribution of the approximate simulation parameters includes:

[0032] With the goal of minimizing the KL divergence, we search for a probability distribution in the probability space that is close to the posterior distribution of the parameter. :

[0033]

[0034] arg min KL() represents the value of the independent variable when minimizing the KL divergence;

[0035] Minimizing the KL divergence is equivalent to maximizing The ELBO function:

[0036]

[0037] in, represents the likelihood function, represents the prior distribution of the parameter, and E[ ] is the mathematical expectation;

[0038] The likelihood function is expressed in the following form:

[0039]

[0040] in, N is the total number of observations, i is the number of the observation data; Expressing obedience expectations m , the standard deviation is s The probability density function of the normal distribution is x The value of Expressing obedience expectations , standard deviation is The probability density function of the normal distribution is The value of ;

[0041] Assuming that the parameters to be determined are independent of each other, the probability density function of the prior distribution is expressed as:

[0042]

[0043] in, n is the total number of parameters, i is the number of the parameter, i n is the probability density function of the nth parameter, p ( s f ), p ( a ), p ( b ), p ( c ) are parameters s f , a , b , c The probability density function of .

[0044] Furthermore, the step of using variational inference to approximate the posterior distribution of the simulation parameters and selecting the optimal model parameter combination also includes: optimizing the probability distribution by using a stochastic gradient ascent algorithm Approximate the posterior distribution of the parameters.

[0045] Furthermore, the stochastic gradient ascent algorithm comprises the following steps:

[0046] Step 1: The initial distribution of is a multivariate normal distribution;

[0047] Step 2: Assumptions The parameters to be requested are f , find the gradient of the ELBO function in this round:

[0048]

[0049] in, is the gradient operator, and the expectation E in the formula is approximated by Monte Carlo sampling to obtain the gradient of this iteration round;

[0050] Step 3: After obtaining the gradient, update the parameters in the direction that makes the ELBO function value increase f ;

[0051] Step 4: Repeat steps 2 and 3 until the gradient converges and the optimized parameters are obtained. f .

[0052] The present application also provides a river video water level-flow velocity-flow three-dimensional surface calculation device, which is used to implement a river video water level-flow velocity-flow three-dimensional surface calculation method, and the device includes:

[0053] Data acquisition module: used to collect the water level, maximum surface velocity and flow data corresponding to the same time required to build the water level-velocity-flow distribution relationship;

[0054] Model building module: used to construct a flow calculation formula based on the collected water level, maximum surface velocity and flow data using variational Bayesian inference to optimize parameters;

[0055] Flow calculation module: used to calculate cross-sectional flow and confidence interval based on water level and maximum surface velocity.

[0056] The present application also provides a computer program product, including a computer program, which, when executed by a processor, implements a method for calculating a three-dimensional surface of water level-flow velocity-flow in a river video.

[0057] In summary, the present application includes at least one of the following beneficial technical effects:

[0058] 1. This application can quickly calculate the flow rate based on the water level and the maximum surface flow velocity, and realize remote real-time monitoring of the flow rate. Compared with the traditional velocity-area method, the function form and calculation process are simplified;

[0059] 2. The traditional water level-flow relationship only considers a single curve relationship. This application introduces the maximum surface velocity based on the water level-flow relationship, which can accurately reflect the flow law and adapt to different cross-sectional conditions;

[0060] 3. In terms of uncertainty analysis, the traditional velocity-area method has many sources of error when measuring flow, the uncertainty transfer process is complex, and the uncertainty analysis is difficult; the traditional water level-flow relationship curve only considers the model fitting deviation, and the parameters are often fixed values, and the source of uncertainty is inaccurate; this application comprehensively considers the measurement error and model fitting deviation, the source of uncertainty is clear, and the probability distribution of the model parameters is obtained according to the Bayesian principle, which is convenient for the uncertainty calculation of the results. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 is a flow chart of the method involved in the embodiment of the present application;

[0062] Figure 2 Schematic diagram of vertical velocity distribution where the maximum surface velocity is located in the embodiment of the present application;

[0063] Figure 3 This is a graph showing the relationship between water level, maximum surface velocity and flow rate in the embodiment of the present application. DETAILED DESCRIPTION

[0064] The following is combined with Figure 1-3 This application is described in further detail.

[0065] The present application embodiment discloses a method and device for calculating a three-dimensional surface of water level, velocity and flow rate in a river video. Figure 1 The river video water level-velocity-flow three-dimensional surface calculation method includes the following steps:

[0066] S100, collecting hydrological test data: collect a set of data sets S at the section of the river to be studied, S={ };in, is the water level measurement, is the maximum surface velocity measurement, is the flow measurement value.

[0067] When calibrating parameters, the water level and maximum surface flow velocity are collected by non-contact measurement methods, including images, radar and other measurement methods; in order to ensure the applicability of the model, the measurement methods of water level and maximum surface flow velocity remain consistent during the model establishment stage and the model application stage; the flow measurement methods include instruments such as flow meters and ADCP.

[0068] S200, constructing a water level-maximum surface velocity-flow calculation model: expressing the flow of the cross section as the product of the cross section average velocity and the cross section area, establishing the relationship between the cross section average velocity and the maximum surface velocity, and the relationship between the cross section area and the water level; expressing the cross section average velocity by the maximum surface velocity, and expressing the cross section area by the water level, and obtaining a general expression form of the water level-maximum surface velocity-flow relationship model in the form of a power function:

[0069] Formula (2-1)

[0070] in, Q For flow, V is the maximum surface velocity, H is the water level, a , b , c These are all model parameters. Specifically, the following steps are included:

[0071] S210, establishing a relationship between the average flow velocity of the cross section and the maximum surface flow velocity, comprising the following steps:

[0072] S211, such as Figure 2 As shown in the figure, the vertical line where the maximum flow velocity and the maximum surface flow velocity of the cross section are located is used as the y-axis, and the cross section in the direction perpendicular to the flow direction (horizontal) is used as the z-axis to draw the equal flow velocity distribution line. x and its normal or , Figure 2 D is the vertical distance from the bottom of the riverbed to the water surface on the y-axis, h is the vertical distance from the water surface to the point of maximum flow velocity on the y-axis, and B1 is the distance from the vertical line of maximum flow velocity to the shore;

[0073] The relationship between flow velocity and depth on the y-axis is expressed as:

[0074] Formula (2-2)

[0075] in, u max is the maximum flow velocity of the cross section, M are the parameters describing the distribution, F ( u ) is the cumulative distribution function of flow velocity, that is, the flow velocity is less than or equal to u The probability of , e is a natural constant;

[0076] The cumulative distribution function of flow velocity is:

[0077] Formula (2-3)

[0078] in, D is the water depth on the y-axis, h is the distance from the water surface to the point of maximum flow velocity, y is the distance from the bottom of the riverbed to the water surface, and exp() is an exponential function with the natural constant e as the base.

[0079] S212. Establish the relationship between the maximum flow velocity of the cross section and the surface flow velocity, expressed as:

[0080] Formula (2-4)

[0081] in, u D is the y-axis surface velocity, i.e., the maximum surface velocity.

[0082] S213. Based on the velocity distribution equation based on information entropy theory, the maximum flow velocity of the cross section is converted into the average flow velocity of the cross section, which is expressed as:

[0083] Formula (2-5)

[0084] in, u mean is the average flow velocity in the cross section.

[0085] S214. Based on the relationship between the maximum flow velocity of the cross section and the surface flow velocity, and the relationship between the maximum flow velocity of the cross section and the average flow velocity of the cross section, formula (2-4) and formula (2-5) are combined to obtain the relationship between the average flow velocity of the cross section and the maximum surface flow velocity, which is expressed as:

[0086] Formula (2-6)

[0087] in, f It is the coefficient of maximum surface velocity and average cross-sectional velocity.

[0088] S220, establishing a relationship between the cross-sectional area and the water level, comprising the following steps:

[0089] S221. Construct the relationship between the cross-sectional area and the water level based on the cross-sectional geometric characteristics, and analyze the cases where the shapes of the river channel control cross-section are rectangular, trapezoidal, and parabolic:

[0090] Formula (2-7)

[0091] in, is the area of ​​the rectangle, B is the width of the rectangular section, b is the water level when the flow rate is 0, is the area of ​​the trapezoid, is the width of the trapezoidal section, is the riverbed width, is the area of ​​the parabola, C is the parabolic shape coefficient;

[0092] For irregular cross-sectional shapes, n evenly spaced vertical sounding lines can be evenly set along the riverbed to divide the cross-section into n+1 sub-sections. The two sub-sections near the shore are generalized into triangles, and the remaining sub-sections are generalized into trapezoids. The area of ​​each sub-section is:

[0093] Formula (2-8)

[0094] Among them, S1 and S n+1 are the areas of the two triangular sub-sections near the shore, S i is the area of ​​the ith trapezoidal sub-section, oh is the interval of the sounding vertical lines, d is the water depth at the sounding vertical line, d The subscript indicates the number of the sounding vertical line;

[0095] The cross-sectional area is:

[0096] Formula (2-9)

[0097] in, B For the width of the river, is the average water depth;

[0098] According to Chebyshev's inequality:

[0099] Formula (2-10)

[0100] in, d max is the maximum water depth, k is the coefficient, s 2 is the variance of river depth distribution;

[0101] The relationship between the irregular cross-sectional area and the water level can be expressed as:

[0102] Formula (2-11)

[0103] The above steps have deduced that the relationship between cross-sectional area and water level under different geometric conditions is consistent with of the form, in which l It is a coefficient that reflects the geometric shape of the river channel.

[0104] S222. Combining the relationship between the average flow velocity of the cross section and the maximum surface velocity, the relationship between the flow rate and the water level and the maximum surface velocity is obtained as follows: .

[0105] S300, comprehensively considering the measurement error of the observation data and the error distribution of the model deviation, establishing the relationship between the measured flow rate and the measured maximum surface flow velocity and the measured water level, specifically including the following steps:

[0106] Observed values ​​of water level, maximum surface velocity, and flow rate is the true value The estimated value of satisfies the following relationship with the true value:

[0107] Formula (3-1)

[0108] in, are the observed values ​​of water level, maximum surface velocity and flow rate, respectively. are the true values ​​of water level, maximum surface velocity, and flow rate, respectively. i is the number of the observation data; is the error distribution of the maximum surface velocity observation, which satisfies the mean of 0 and the standard deviation of u V Normal distribution of is the error distribution of the flow observation value, which satisfies the mean of 0 and the standard deviation of u Q Normal distribution; N() represents normal distribution;

[0109] Formula (3-1) assumes that the error in water level observation can be ignored. There are uncertainties in the measurement of maximum surface velocity and flow rate, which can be obtained by synthesizing the components of each standard uncertainty after multiple independent measurements.

[0110] Since the derivation of the water level-maximum surface velocity-flow relationship model proposed in formula (2-1) is based on a variety of assumptions, such as: the cross-sectional flow velocity should conform to a uniform distribution, the maximum surface velocity and the cross-sectional maximum flow velocity are on the same vertical line, and the cross-sectional geometry is generalized; even if the actual water level and maximum surface velocity are known, the model cannot perfectly represent the actual flow relationship.

[0111] The water level-maximum surface velocity-discharge relationship model writing In the form of, considering the model bias:

[0112] Formula (3-2)

[0113] in, i are model parameters, e i f is the error distribution of the model simulation, which satisfies the mean of 0 and the standard deviation of s f The normal distribution of .

[0114] Combining formula (3-1) and formula (3-2), assuming that the error components are independent of each other, we can get the expression of the measured flow rate with respect to the measured maximum surface velocity and the measured water level:

[0115] Formula (3-3)

[0116] S400, using variational inference to approximate the posterior distribution of simulation parameters, selecting the optimal model parameter combination, and obtaining a water level-maximum surface velocity-flow calculation model after parameter optimization, the specific method includes:

[0117] S410, with the goal of minimizing KL divergence, find a probability distribution in the probability space that is close to the posterior distribution of the parameter :

[0118] Formula (4-1)

[0119] arg min KL() represents the value of the independent variable when minimizing the KL divergence;

[0120] Minimizing the KL divergence is equivalent to maximizing The ELBO function:

[0121] Formula (4-2)

[0122] in, represents the likelihood function, represents the prior distribution of the parameter, and E[ ] is the mathematical expectation;

[0123] The likelihood function is expressed in the following form:

[0124] Formula (4-3)

[0125] in, N is the total number of observations, i is the number of the observation data; Expressing obedience expectations m , standard deviation is s The probability density function of the normal distribution is x The value of Expressing obedience expectations , standard deviation is The probability density function of the normal distribution is The value of ;

[0126] Assuming that the parameters to be determined are independent of each other, the probability density function of the prior distribution is expressed as:

[0127] Formula (4-4)

[0128] in, n is the total number of parameters, i is the number of the parameter, i n is the probability density function of the nth parameter, p ( s f ), p ( a ), p ( b ),p ( c ) are parameters s f , a , b , c The probability density function of .

[0129] S420, Optimize the probability distribution by stochastic gradient ascent algorithm To approximate the posterior distribution of the parameters, the stochastic gradient ascent algorithm consists of the following steps:

[0130] S421, Order The initial distribution of is a multivariate normal distribution;

[0131] S422, Assumption The parameters to be requested are f , find the gradient of the ELBO function in this round:

[0132] Formula (4-5)

[0133] in, is the gradient operator, and the expectation E in the formula is approximated by Monte Carlo sampling to obtain the gradient of this iteration round;

[0134] S423, after obtaining the gradient, update the parameters in the direction that makes the ELBO function value increase f ;

[0135] S424, repeat steps 2 and 3 until the gradient converges and the optimized parameters are obtained f .

[0136] Multivariate Normal Distribution Parameters f Including three means and three standard deviations, the three optimized means are used as parameters a , b , c Three standard deviations are used to determine the parameters a , b , c The confidence interval of . After that, this application considers that the posterior distribution of the parameters approximately follows , taking the expectation of the posterior distribution as the value of the parameter and twice the standard deviation as the uncertainty of the parameter.

[0137] S500, calculate flow: use the water level-maximum surface velocity-flow calculation model after parameter optimization, calculate the flow after measuring the water level and surface velocity, and analyze the confidence interval of the model fitting flow. The three-dimensional surface of water level H-maximum surface velocity V-flow Q constructed at a certain site is as follows: Figure 3shown.

[0138] This embodiment also provides a river video water level-flow velocity-flow three-dimensional surface calculation device, which is used to implement a river video water level-flow velocity-flow three-dimensional surface calculation method, and the device includes:

[0139] Data acquisition module: used to collect the water level, maximum surface velocity and flow data corresponding to the same time required to build the water level-velocity-flow distribution relationship;

[0140] Model building module: used to construct a flow calculation formula based on the collected water level, maximum surface velocity and flow data using variational Bayesian inference to optimize parameters;

[0141] Flow calculation module: used to calculate cross-sectional flow and confidence interval based on water level and maximum surface velocity.

[0142] This embodiment also provides a computer program product, including a computer program, which, when executed by a processor, implements a method for calculating a three-dimensional surface of water level-flow velocity-flow in a river video.

[0143] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A three-dimensional surface calculation method for river video water level-flow velocity-flow, characterized by: The following steps are involved: Collect hydrological test data: Collect data set S at the section of the river to be studied, S={ };in, is the water level measurement, is the maximum surface velocity measurement, is the flow measurement; Construct a water level-maximum surface velocity-flow calculation model: express the flow of the river section as the product of the average flow velocity of the section and the cross-sectional area, establish the relationship between the average flow velocity of the section and the maximum surface flow velocity, and the relationship between the cross-sectional area and the water level; express the average flow velocity of the section by the maximum surface flow velocity, and express the cross-sectional area by the water level, and obtain the general expression of the water level-maximum surface flow velocity-flow relationship model: in, Q For flow, V is the maximum surface velocity, H is the water level, a , b , c are all model parameters; Taking into account the measurement error of the observation data and the error distribution of the model bias, the relationship between the measured flow rate and the measured maximum surface velocity and the measured water level is established. The specific methods include: Observed values ​​of water level, maximum surface velocity, and flow rate is the true value The estimated value of satisfies the following relationship with the true value: in, are the observed values ​​of water level, maximum surface velocity and flow rate, respectively. are the true values ​​of water level, maximum surface velocity, and flow rate, respectively. i is the number of the observation data; is the error distribution of the maximum surface velocity observation, which satisfies the mean of 0 and the standard deviation of u V Normal distribution of is the error distribution of the flow observation value, which satisfies the mean of 0 and the standard deviation of u Q Normal distribution; N() represents normal distribution; The water level-maximum surface velocity-discharge relationship model writing In the form of, considering the model bias: in, θ are model parameters, ε i f is the error distribution of the model simulation, which satisfies the mean of 0 and the standard deviation of σ f Normal distribution of Assuming that the error components are independent of each other, the expression of the measured flow rate with respect to the measured maximum surface velocity and the measured water level is obtained: ; Variational Bayesian inference is used to approximate the posterior distribution of simulation parameters, and the optimal model parameter combination is selected to obtain the water level-maximum surface velocity-discharge calculation model after parameter optimization. Calculate flow rate: Use the water level-maximum surface velocity-flow calculation model with optimized parameters to calculate the flow rate based on the measured water level and the measured maximum surface velocity.

2. The method for calculating the three-dimensional surface of water level, velocity and flow rate of a river video according to claim 1, characterized in that: In the step of collecting hydrological test data, the water level and maximum surface flow velocity are collected by non-contact measurement methods, including image measurement methods. The measurement methods of water level and maximum surface flow velocity remain consistent in the model establishment stage and the model application stage; the flow measurement methods include flow meters and ADCP instruments.

3. The method for calculating the three-dimensional surface of water level, velocity and flow rate of a river video according to claim 1, characterized in that: In the step of constructing the water level-maximum surface velocity-flow calculation model, the method of establishing the relationship between the average flow velocity of the cross section and the maximum surface velocity includes: The vertical line where the maximum flow velocity and the maximum surface flow velocity of the cross section are located is taken as the y-axis, and the transverse direction of the cross section is taken as the z-axis. The relationship between the flow velocity and the depth on the y-axis and the cumulative distribution function of the flow velocity are established. Establish the relationship between the maximum flow velocity of the cross section and the surface flow velocity; Based on the velocity distribution equation based on information entropy theory, the maximum flow velocity of the cross section is converted into the average flow velocity of the cross section; Based on the relationship between the maximum flow velocity of the cross section and the surface flow velocity, and the relationship between the maximum flow velocity of the cross section and the average flow velocity of the cross section, the relationship between the average flow velocity of the cross section and the maximum surface flow velocity is obtained.

4. The method for calculating the three-dimensional surface of water level, velocity and flow rate of a river video according to claim 1, characterized in that: In the step of constructing the water level-maximum surface velocity-flow calculation model, the relationship between the cross-sectional area and the water level is constructed according to the cross-sectional geometric characteristics. The cases where the river cross-sectional shapes are rectangular, trapezoidal, and parabolic are analyzed respectively. Combined with the relationship between the average flow velocity of the cross section and the maximum surface velocity, the relationship between the flow, the water level, and the maximum surface velocity is obtained as follows: .

5. The method for calculating the three-dimensional surface of water level, velocity and flow rate of a river video according to claim 1, characterized in that: Methods for using variational Bayes to infer the posterior distribution of approximate simulation parameters include: With the goal of minimizing the KL divergence, we search for a probability distribution in the probability space that is close to the posterior distribution of the parameter. : arg min KL() represents the value of the independent variable when minimizing the KL divergence; Minimizing the KL divergence is equivalent to maximizing The ELBO function: in, represents the likelihood function, represents the prior distribution of the parameter, and E[ ] is the mathematical expectation; The likelihood function is expressed in the following form: in, N is the total number of observations, i is the number of the observation data; Expressing obedience expectations m , standard deviation is s The probability density function of the normal distribution is x The value of Expressing obedience expectations , standard deviation is The probability density function of the normal distribution is The value of ; Assuming that the parameters to be determined are independent of each other, the probability density function of the prior distribution is expressed as: in, n is the total number of parameters, i is the number of the parameter, θ n is the probability density function of the nth parameter, p ( σ f ), p ( a ), p ( b ), p ( c ) are parameters σ f , a , b , c The probability density function of .

6. A river video water level-flow velocity-flow three-dimensional surface calculation method according to claim 5, characterized in that: Using variational Bayesian inference to approximate the posterior distribution of simulation parameters, the step of selecting the optimal model parameter combination also includes: optimizing the probability distribution by stochastic gradient ascent algorithm Approximate the posterior distribution of the parameters.

7. A river video water level-flow velocity-flow three-dimensional surface calculation method according to claim 6, characterized in that: The stochastic gradient ascent algorithm comprises the following steps: Step 1: The initial distribution of is a multivariate normal distribution; Step 2: Assumptions The parameters to be requested are φ , find the gradient of the ELBO function in this round: in, is the gradient operator, and the expectation E in the formula is approximated by Monte Carlo sampling to obtain the gradient of this iteration round; Step 3: After obtaining the gradient, update the parameters in the direction that makes the ELBO function value increase φ ; Step 4: Repeat steps 2 and 3 until the gradient converges and the optimized parameters are obtained. φ .

8. A three-dimensional surface calculation device for river video water level-flow velocity-flow, characterized in that: The device is used to implement the method for calculating the three-dimensional surface of water level, velocity and flow rate of a river video according to any one of claims 1 to 7, and comprises: Data acquisition module: used to collect the water level, maximum surface velocity and flow data corresponding to the same time required to build the water level-velocity-flow distribution relationship; Model building module: used to construct a flow calculation formula based on the collected water level, maximum surface velocity and flow data using variational Bayesian inference to optimize parameters; Flow calculation module: used to calculate cross-sectional flow and confidence interval based on water level and maximum surface velocity.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, a river video water level-velocity-flow three-dimensional surface calculation method as described in any one of claims 1-7 is implemented.