Storm surge single-point conventional forecasting method, medium and system

Abstract

The invention provides a storm surge single-point conventional forecasting method, medium and system, and belongs to the technical field of ocean storm surge prediction.The method comprises the steps that historical water level data, historical flow speed data and historical meteorological data are collected at a forecasting target point to serve as training data; performing time sequence decomposition on the training data to obtain a fundamental component and a harmonic component, constructing a characteristic matrix, calculating a water level contribution coefficient, a flow velocity contribution coefficient and a meteorological contribution coefficient, and constructing a storm surge water level forecasting equation set comprising a storm surge dynamic equation and a storm surge continuous equation according to the contribution coefficients, inputting real-time water level data, real-time flow velocity data and real-time meteorological data into the storm surge water level forecasting equation set to obtain a storm surge forecasting result, and outputting an early warning signal when the storm surge forecasting result exceeds an early warning threshold value; the physical basis of forecasting is guaranteed, the calculation efficiency is improved, and the technical problem that in the prior art, forecasting precision and timeliness of the sudden strong storm surge are not enough is solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of ocean storm surge prediction, and in particular, relates to a single-point conventional storm surge prediction method, medium and system. Background Art

[0002] Storm surge is an abnormal rise and fall of the sea level caused by meteorological factors such as strong winds and abnormal air pressure. It is characterized by strong suddenness and great destructive power. Traditional storm surge forecasting methods mainly include statistical model method and numerical simulation method. The statistical model method establishes a forecasting model based on historical observation data, and obtains the relationship between influencing factors and storm surge water increase through statistical analysis. This method is simple to calculate and efficient, but the forecast accuracy for sudden storm surges is low. The numerical simulation method is based on the basic equations of fluid mechanics, and the evolution process of storm surge is obtained through numerical solution. This method takes into account the physical mechanism of storm surge formation and has relatively high forecast accuracy, but the calculation is large and time-consuming, which is difficult to meet the needs of real-time forecasting. In recent years, with the development of machine learning methods, storm surge forecasting methods based on deep learning have been widely used. This method realizes storm surge forecasting by constructing a neural network model, and has strong nonlinear fitting ability.

[0003] At present, there are the following major problems in storm surge forecasting technology: First, it is difficult for existing methods to accurately describe the multi-scale characteristics of storm surges. Storm surges are affected by multiple factors such as astronomical tides and meteorological tides. The time scales of each factor are different. Simple statistical analysis or numerical simulations cannot effectively distinguish the contribution of each scale component. Second, existing methods do not fully consider the lag effect of storm surge formation. The formation of storm surges has obvious time delay characteristics. Ignoring this feature will reduce the forecast accuracy. Third, existing methods perform poorly when dealing with sudden and strong storm surges. Due to the lack of effective characterization of extreme events, there are large errors in the forecast results.

[0004] These problems make it difficult to effectively improve the accuracy and timeliness of storm surge forecasts, especially for sudden strong storm surges, where the problems of large forecast errors and short lead times are more prominent. It is urgent to develop a new forecasting method that can take into account both forecast accuracy and timeliness and effectively improve the forecast level of sudden strong storm surges.

[0005] In summary, the existing technology has the technical problem of insufficient forecast accuracy and timeliness for sudden severe storm surges. Summary of the invention

[0006] In view of this, the present invention provides a single-point conventional storm surge forecasting method, medium and system, which can solve the technical problem of insufficient forecasting accuracy and timeliness of sudden strong storm surges in the prior art.

[0007] The present invention is implemented as follows: In a first aspect, the present invention provides a conventional single-point storm surge forecasting method comprising the following steps: collecting historical water level data, historical flow velocity data and historical meteorological data as training data at a forecast target point, performing time series decomposition on the training data to obtain fundamental wave components and harmonic components and constructing a feature matrix, calculating water level contribution coefficients, flow velocity contribution coefficients and meteorological contribution coefficients, constructing a storm surge water level forecasting equation group including a storm surge dynamic equation and a storm surge continuity equation according to the contribution coefficients, inputting real-time water level data, real-time flow velocity data and real-time meteorological data into the storm surge water level forecasting equation group to obtain a storm surge forecast result, and outputting a warning signal when the storm surge forecast result exceeds a warning threshold; the storm surge water level forecasting equation group includes a storm surge dynamic equation and a storm surge continuity equation, the storm surge dynamic equation describes the relationship between the total potential energy of the storm surge, the historical water level data and the historical water level change rate, and the storm surge continuity equation describes the relationship between the historical water level data and the historical flow velocity data.

[0008] Among them, the step of collecting training data is specifically to set up water level monitoring equipment, acoustic Doppler flowmeter and automatic weather station at the forecast target point. The water level monitoring equipment includes ultrasonic water level gauge and pressure water level gauge. The automatic weather station is used to monitor wind direction, wind speed, air pressure and temperature.

[0009] Among them, the step of time series decomposition specifically adopts the empirical mode decomposition method to decompose the historical water level data into water level fundamental component and water level harmonic component, decompose the historical flow velocity data into flow velocity fundamental component and flow velocity harmonic component, and decompose the historical meteorological data into meteorological fundamental component and meteorological harmonic component.

[0010] Among them, the step of constructing the characteristic matrix specifically adopts the sliding window technology to construct the water level fundamental component and the water level harmonic component into a water level characteristic matrix, the flow velocity fundamental component and the flow velocity harmonic component into a flow velocity characteristic matrix, and the meteorological fundamental component and the meteorological harmonic component into a meteorological characteristic matrix.

[0011] Among them, the step of calculating the contribution coefficient is specifically to use the principal component analysis method to calculate the water level contribution coefficient of the water level characteristic matrix, the flow velocity contribution coefficient of the flow velocity characteristic matrix and the meteorological contribution coefficient of the meteorological characteristic matrix through singular value decomposition.

[0012] Among them, the step of constructing a storm surge water level forecast equation group is specifically to calculate the total potential energy of the storm surge based on the water level contribution coefficient and the meteorological contribution coefficient, the total potential energy of the storm surge is composed of the sum of the storm surge hydrodynamic potential energy and the storm surge meteorological potential energy, the storm surge dynamic equation is established based on the total storm surge potential energy, and the storm surge continuity equation is established based on the historical water level data and the historical flow rate data.

[0013] The parameters of the storm surge water level forecast equation group are optimized by a genetic algorithm, the population size of the genetic algorithm is 100, the evolutionary generation is 1000, the crossover probability is 0.8, and the mutation probability is 0.1.

[0014] Among them, the warning thresholds include yellow warning threshold, orange warning threshold and red warning threshold. The yellow warning threshold is a water increase of 1.5 meters or a duration of more than 6 hours, the orange warning threshold is a water increase of 2.0 meters or a duration of more than 12 hours, and the red warning threshold is a water increase of 2.5 meters or a duration of more than 24 hours.

[0015] A second aspect of the present invention provides a computer-readable storage medium, in which program instructions are stored. When the program instructions are run in a computer, they are used to execute the above-mentioned single-point conventional storm surge forecasting method.

[0016] The third aspect of the present invention provides a single-point conventional storm surge forecasting system, comprising the above-mentioned computer-readable storage medium, wherein the system is any one of a computer, a server, and a single-chip microcomputer, and the computer-readable storage medium is arranged in the system, and the system is provided with a microprocessor for executing program instructions stored in the computer-readable storage medium.

[0017] Compared with the prior art, the present invention provides a single-point conventional storm surge forecasting method, medium and system. The single-point conventional storm surge forecasting method proposed by the present invention realizes multi-scale analysis of signals through empirical mode decomposition, uses characteristic matrix to characterize the contribution of each scale component, and establishes a forecasting equation group based on the principle of energy conservation, thereby realizing an accurate description of the storm surge formation mechanism. While ensuring the forecast accuracy, the method has high computational efficiency and can meet the needs of real-time forecasting. The present invention uses the empirical mode decomposition method to realize the effective identification of the multi-scale characteristics of the storm surge, obtains components of different scales through adaptive decomposition, and avoids the limitations of artificially setting the decomposition basis function in the traditional method; introduces a time delay term in the forecast equation, effectively describes the lag effect of the storm surge formation, and improves the accuracy of the forecast; constructs the forecast equation based on the principle of energy conservation, unifies the hydrodynamic and meteorological factors into the energy framework, and enhances the model's forecasting ability for sudden strong storm surges. The present invention organically combines signal decomposition, feature extraction and physical modeling, which not only ensures the physical basis of the forecast, but also improves the calculation efficiency, realizes the unity of forecast accuracy and timeliness, and solves the technical problem of insufficient forecast accuracy and timeliness for sudden strong storm surges in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0020] like Figure 1 FIG. 1 is a flow chart of a conventional storm surge single-point forecasting method provided by the first aspect of the present invention. The method comprises the following steps:

[0021] S01, setting water level monitoring equipment and meteorological monitoring equipment at the forecast target point, collecting historical water level data, historical flow rate data and historical meteorological data of the forecast target point as training data;

[0022] S02, performing time series decomposition on the training data to obtain the water level fundamental component and water level harmonic component of the historical water level data, the flow velocity fundamental component and flow velocity harmonic component of the historical flow velocity data, and the meteorological fundamental component and meteorological harmonic component of the historical meteorological data;

[0023] S03, constructing a water level characteristic matrix according to the water level fundamental wave component and the water level harmonic component, constructing a flow velocity characteristic matrix according to the flow velocity fundamental wave component and the flow velocity harmonic component, and constructing a meteorological characteristic matrix according to the meteorological fundamental wave component and the meteorological harmonic component;

[0024] S04, calculating the water level contribution coefficient according to the water level characteristic matrix, calculating the flow velocity contribution coefficient according to the flow velocity characteristic matrix, and calculating the meteorological contribution coefficient according to the meteorological characteristic matrix;

[0025] S05. Calculating the total potential energy of the storm surge according to the water level contribution coefficient and the meteorological contribution coefficient, wherein the total potential energy of the storm surge is composed of the sum of the hydrodynamic potential energy of the storm surge and the meteorological potential energy of the storm surge;

[0026] S06. Calculating the historical water level change rate according to the historical water level data;

[0027] S07, establishing a storm surge water level forecast equation group, wherein the storm surge water level forecast equation group includes a storm surge dynamic equation and a storm surge continuity equation, wherein the storm surge dynamic equation describes the relationship between the total potential energy of the storm surge, the historical water level data and the historical water level change rate, and the storm surge continuity equation describes the relationship between the historical water level data and the historical flow velocity data;

[0028] S08, optimizing parameters of the storm surge water level forecast equation group using the training data;

[0029] S09, obtaining real-time water level data, real-time flow velocity data and real-time meteorological data of the forecast target point;

[0030] S10, performing time series decomposition on the real-time water level data, the real-time flow velocity data and the real-time meteorological data to obtain a real-time water level fundamental wave component and a real-time water level harmonic component, a real-time flow velocity fundamental wave component and a real-time flow velocity harmonic component, and a real-time meteorological fundamental wave component and a real-time meteorological harmonic component;

[0031] S11, constructing a real-time water level characteristic matrix according to the real-time water level fundamental wave component and the real-time water level harmonic component, constructing a real-time flow velocity characteristic matrix according to the real-time flow velocity fundamental wave component and the real-time flow velocity harmonic component, and constructing a real-time meteorological characteristic matrix according to the real-time meteorological fundamental wave component and the real-time meteorological harmonic component;

[0032] S12, inputting the real-time water level characteristic matrix, the real-time flow velocity characteristic matrix and the real-time meteorological characteristic matrix into the storm surge water level forecast equation group to obtain the storm surge forecast water level;

[0033] S13. Calculating a storm surge forecast result according to the storm surge forecast water level, and outputting a warning signal when the storm surge forecast result exceeds a warning threshold.

[0034] The specific implementation methods of the above steps are described in detail below. The specific implementation method of step S01 is to realize comprehensive monitoring of the hydrological and meteorological elements of the forecast target point by rationally deploying a monitoring equipment network. First, a high-precision water level monitoring device is set up in the area where the tidal effect of the forecast target point is significant. The water level monitoring device includes an ultrasonic water level gauge and a pressure water level gauge, wherein the ultrasonic water level gauge has a measurement accuracy of ±1 cm and a sampling frequency of 1 time per minute, and the pressure water level gauge has a measurement accuracy of ±0.5 cm and a sampling frequency of 2 times per minute. Then, an acoustic Doppler current meter is deployed at the forecast target point to monitor the three-dimensional flow velocity of the water body. The measurement accuracy of the acoustic Doppler current meter is ±0.1 meters per second and the sampling frequency is 1 time per minute. Finally, an automatic weather station is deployed at the forecast target point. The automatic weather station is used to monitor meteorological elements such as wind direction, wind speed, air pressure, and temperature. The measurement accuracy of the wind speed sensor is ±0.3 meters per second, the measurement accuracy of the wind direction sensor is ±3 degrees, the measurement accuracy of the air pressure sensor is ±0.1 hPa, and the measurement accuracy of the temperature sensor is ±0.1 degrees Celsius. The sampling frequency of all meteorological elements is once per minute. The above-mentioned monitoring equipment is used to obtain no less than 1 year of historical observation data as training data, and the training data includes historical water level data, historical flow rate data, and historical meteorological data. The purpose of this step is to establish a complete monitoring data system to provide data support for subsequent data analysis and model construction.

[0035] The specific implementation method of step S02 is to use the empirical mode decomposition method to perform time series decomposition on the training data. First, the historical water level data is subjected to empirical mode decomposition, and the nonlinear and non-stationary water level signal is decomposed into several intrinsic mode functions through repeated screening processes, wherein the sum of the energy of the first two intrinsic mode functions accounts for more than 85% of the total energy and is defined as the water level harmonic component, and the sum of the energy of the remaining intrinsic mode functions is defined as the water level fundamental component. Then the same empirical mode decomposition method is used to decompose the historical flow velocity data to obtain the flow velocity fundamental component and the flow velocity harmonic component. Finally, the wind speed, air pressure and other elements in the historical meteorological data are subjected to empirical mode decomposition respectively to obtain the meteorological fundamental component and the meteorological harmonic component. The advantage of using the empirical mode decomposition method is that it has strong adaptability, does not require preset basis functions, and can effectively process nonlinear and non-stationary signals. The purpose of this step is to decompose the complex hydrological and meteorological elements into components of different scales, so as to facilitate the subsequent analysis of their respective contributions.

[0036] The specific implementation method of step S03 is to construct a feature matrix using a matrix transformation method. First, the water level fundamental wave component and the water level harmonic component are arranged in chronological order to construct a water level feature matrix, where the number of rows of the matrix is ​​equal to the number of sampling points and the number of columns is 2. Then, the flow velocity fundamental wave component and the flow velocity harmonic component are constructed into a flow velocity feature matrix using the same method. Finally, the meteorological fundamental wave component and the meteorological harmonic component are constructed into a meteorological feature matrix. During the matrix construction process, a sliding window technique is used for data enhancement, with a window length of 24 hours and a sliding step length of 1 hour. In this way, the time characteristics of the data can be fully mined. The purpose of this step is to convert the various components obtained by time series decomposition into a matrix form that is easy to calculate.

[0037] The specific implementation method of step S04 is to use the principal component analysis method to calculate the contribution coefficients of each characteristic matrix. First, the water level characteristic matrix is ​​subjected to singular value decomposition, and it is decomposed into a left singular matrix, a singular value matrix and a right singular matrix, wherein the singular value reflects the importance of each component, and the water level contribution coefficient is obtained by normalization. Then the velocity contribution coefficient of the velocity characteristic matrix is ​​calculated by the same method. Finally, the meteorological contribution coefficient of the meteorological characteristic matrix is ​​calculated. When calculating the contribution coefficient, the cross-validation method is used to determine the optimal number of principal components, and generally the number of principal components corresponding to the cumulative contribution rate of 95% is selected. The purpose of this step is to quantify the contribution of each component to the formation of storm surges.

[0038] The specific implementation method of step S05 is to calculate the total potential energy of the storm surge based on the principle of conservation of energy. First, the hydrodynamic potential energy of the storm surge is calculated according to the water level contribution coefficient, and the conversion relationship between gravitational potential energy and kinetic energy is considered during the calculation. Then, the meteorological potential energy of the storm surge is calculated according to the meteorological contribution coefficient, and the effects of wind stress and air pressure difference on the water body are considered during the calculation. Finally, the hydrodynamic potential energy of the storm surge and the meteorological potential energy of the storm surge are added to obtain the total potential energy of the storm surge. In the potential energy calculation process, the energy integration method is adopted, and the integration time step is 1 hour. The purpose of this step is to describe the formation mechanism of the storm surge from an energy perspective.

[0039] The specific implementation method of step S06 is to calculate the historical water level change rate using the finite difference method. First, the historical water level data is subjected to data quality control to remove outliers, and the criterion for judging outliers is to exceed the mean by plus or minus 3 times the standard deviation. Then the central difference format is used to calculate the water level change rate at each moment, and the calculation time step is 1 hour. Finally, the calculated water level change rate is smoothed, and the 5-point sliding average method is used to eliminate high-frequency noise. The purpose of this step is to obtain the dynamic characteristics of water level changes.

[0040] The specific implementation method of step S07 is to establish a storm surge water level forecast equation group based on fluid mechanics theory. First, a storm surge dynamic equation is established, which describes the water level change process driven by the total potential energy of the storm surge. The equation contains inertia terms, Coriolis force terms, pressure gradient terms and friction terms. Then a storm surge continuity equation is established, which is based on the principle of mass conservation and describes the relationship between water level and flow velocity. In the process of equation construction, the Bernoulli equation is used to derive the relationship between potential energy and water level change rate, and the Navier-Stokes equation is used to derive the relationship between water level and flow velocity. The purpose of this step is to establish a theoretical basis for storm surge forecasting.

[0041] The specific implementation method of step S08 is to use a genetic algorithm to optimize the parameters of the storm surge water level forecast equation group. First, determine the parameters to be optimized, including the friction coefficient, energy conversion coefficient, etc. in the equation. Then construct a fitness function and use the root mean square error as an evaluation index. Finally, a genetic algorithm is used to optimize the parameters, with the population size set to 100, the evolutionary algebra set to 1000, the crossover probability set to 0.8, and the mutation probability set to 0.1. During the optimization process, the cross-validation method is used to evaluate the generalization performance of the parameters. The purpose of this step is to improve the accuracy of the forecast equation.

[0042] The specific implementation method of step S09 is to collect the monitoring data of the forecast target point in real time. First, the real-time water level data is obtained through the water level monitoring equipment, and the sampling frequency is once per minute. Then the real-time flow velocity data is obtained through the acoustic Doppler current meter, and the sampling frequency is once per minute. Finally, the real-time meteorological data is obtained through the automatic weather station, and the sampling frequency is once per minute. During the data collection process, the real-time quality control method is used to eliminate abnormal values. The purpose of this step is to obtain the input data required for real-time forecasting.

[0043] The specific implementation method of step S10 is to perform time series decomposition on the real-time monitoring data. First, the real-time water level data is decomposed by the online empirical mode decomposition method to obtain the real-time water level fundamental component and the real-time water level harmonic component. Then the real-time flow velocity data is decomposed by the same method to obtain the real-time flow velocity fundamental component and the real-time flow velocity harmonic component. Finally, the real-time meteorological data is decomposed to obtain the real-time meteorological fundamental component and the real-time meteorological harmonic component. In the decomposition process, the sliding window technology is used to realize real-time processing, and the window length is 6 hours. The purpose of this step is to obtain each component of the real-time data.

[0044] The specific implementation method of step S11 is to construct a real-time characteristic matrix. First, the real-time water level fundamental wave component and the real-time water level harmonic component are constructed as a real-time water level characteristic matrix. Then the real-time flow velocity fundamental wave component and the real-time flow velocity harmonic component are constructed as a real-time flow velocity characteristic matrix. Finally, the real-time meteorological fundamental wave component and the real-time meteorological harmonic component are constructed as a real-time meteorological characteristic matrix. In the matrix construction process, a real-time update strategy is adopted to update the matrix every 1 hour. The purpose of this step is to convert real-time data into a form suitable for model input.

[0045] The specific implementation method of step S12 is to perform storm surge forecast calculation. First, the real-time characteristic matrix is ​​input into the optimized storm surge water level forecast equation group. Then the Runge-Kutta method is used to solve the equation group, with a calculation time step of 1 hour and a forecast duration of 24 hours. Finally, the storm surge forecast water level is output. During the calculation process, parallel computing technology is used to improve the calculation efficiency. The purpose of this step is to obtain the forecast value of the future water level.

[0046] The specific implementation method of step S13 is to make a storm surge warning judgment. First, the storm surge forecast result is calculated based on the storm surge forecast water level, including the maximum water increase value and duration. Then the storm surge forecast result is compared with the warning threshold, where the yellow warning threshold is a water increase of 1.5 meters or a duration of more than 6 hours, the orange warning threshold is a water increase of 2.0 meters or a duration of more than 12 hours, and the red warning threshold is a water increase of 2.5 meters or a duration of more than 24 hours. Finally, when the forecast result exceeds the warning threshold, a warning signal is issued through the early warning system. The purpose of this step is to timely detect storm surge disaster risks and issue warnings.

[0047] In addition, the warning threshold is set according to the warning tide level, wherein the blue warning tide level, the yellow warning tide level, the orange warning tide level and the red warning tide level are the warning tide levels approved in accordance with GB / T 17839-2011. The blue warning tide level is the tide level value at which the marine disaster warning department issues a blue warning for storm surges. When the tide level reaches this set value, the coast must enter a state of alert to prevent the occurrence of tidal disasters. The yellow warning tide level is the tide level value at which the marine disaster warning department issues a yellow warning for storm surges. When the tide level reaches this set value, minor marine disasters may occur along the coast. The orange warning tide level is the tide level value at which the marine disaster warning department issues an orange warning for storm surges. When the tide level reaches this set value, major marine disasters may occur along the coast. The red warning tide level is the tide level value at which the marine disaster warning department issues a red warning for storm surges. When the tide level reaches this set value, major marine disasters may occur along the coast.

[0048] The equations or mathematical models involved in the present invention are described in detail below.

[0049] The mathematical expression of the empirical mode decomposition method is as follows:

[0050]

[0051] Where x(t) is the original signal; c i (t) is the i-th eigenmode function; r n (t) is the residual term; n is the number of eigenmode functions.

[0052] The derivation and establishment process of the empirical mode decomposition method is described as follows: first determine all extreme points, use cubic spline interpolation to connect all maximum points to obtain the upper envelope, and connect all minimum points to obtain the lower envelope. Then calculate the mean of the upper and lower envelopes, and subtract the mean from the original signal to obtain the first component. Repeat the above process until the residual becomes a monotonic function. The optimization of the empirical mode decomposition method is mainly reflected in the use of an improved endpoint extension method to deal with boundary effects, and the decomposition results at the boundary are made more accurate through mirror extension. The effect of this method is reflected in the ability to adaptively decompose complex signals into intrinsic mode functions of different frequencies. The parameter n is determined by the standard deviation criterion, and the decomposition is stopped when the standard deviation of two adjacent decompositions is less than 0.0001.

[0053] The construction of the water level feature matrix is ​​expressed as follows:

[0054]

[0055] Where H is the water level characteristic matrix; h i1 is the fundamental wave component of water level at the ith moment; h i2is the water level harmonic component at the ith moment; n is the number of sampling points.

[0056] Construction of water level characteristic matrix: First, the sliding window technology is used to segment the fundamental component and the harmonic component. The window length is determined to be 24 hours according to the tidal cycle, and the sliding step is 1 hour. Then the data in each window is standardized to remove the dimension effect. Finally, the processed data is arranged in chronological order to form a matrix. The effect of this construction method is reflected in both retaining the continuity of the time series and achieving data dimensionality reduction. The elements in the matrix are derived from the results of empirical mode decomposition and have clear physical meanings.

[0057] The construction of the flow rate characteristic matrix is ​​expressed as follows:

[0058]

[0059] Where V is the velocity characteristic matrix; v i1 is the fundamental component of the velocity at the ith moment; v i2 is the harmonic component of flow velocity at the ith moment; n is the number of sampling points.

[0060] The construction of the meteorological characteristic matrix is ​​expressed as follows:

[0061]

[0062] Where M is the meteorological characteristic matrix; m i1 is the meteorological fundamental wave component at the i-th moment; m i2 is the meteorological harmonic component at the ith moment; n is the number of sampling points.

[0063] The construction methods of velocity characteristic matrix and meteorological characteristic matrix are similar to those of water level characteristic matrix, except that the physical quantities of the data are different. The velocity data are derived from the measurement of acoustic Doppler current meter, and the meteorological data are derived from the observation of automatic weather station.

[0064] The mathematical expression of the singular value decomposition is as follows:

[0065] X=USV T ;

[0066] Where X is the matrix to be decomposed; U is the left singular matrix; S is the singular value matrix; V is the right singular matrix; T represents the matrix transpose.

[0067] Calculation of singular value decomposition: First, the feature matrix is ​​centered and the mean of each column is subtracted. Then the covariance matrix is ​​calculated to solve its eigenvalues ​​and eigenvectors. Finally, the eigenvectors are sorted by eigenvalue to obtain the singular value matrix. The optimization of this method is mainly reflected in the use of power iteration to accelerate the calculation of eigenvalues. The effect of singular value decomposition is reflected in the ability to extract the main features of the data and reduce the impact of noise.

[0068] The calculation expression of the contribution coefficient is as follows:

[0069]

[0070] In the formula, w i is the contribution coefficient of the i-th component; s i is the i-th singular value; n is the number of singular values.

[0071] Calculation of contribution coefficient: Calculate the relative contribution of each component according to the size of the singular value. The optimization process is mainly reflected in the introduction of a threshold mechanism, which only retains components with a contribution rate of more than 5%. The effect of this method is reflected in the ability to quantitatively evaluate the impact of various factors on the formation of storm surges. The cross-validation method is used to determine the parameters and select the optimal threshold.

[0072] The calculation expression of the storm surge hydrodynamic potential energy is as follows:

[0073]

[0074] In the formula, E h is the hydrodynamic potential energy of the storm surge; ρ is the seawater density; g is the acceleration of gravity; h is the water depth; A is the area of ​​the calculation area; and z is the vertical coordinate.

[0075] Calculation of storm surge hydrodynamic potential energy: Based on the definition of potential energy, the position energy of the water body in the gravity field is taken into account. The optimization process is reflected in the use of numerical integration methods to improve the calculation accuracy, and the integration step is determined according to the change in water depth. The effect of this method is that it can accurately describe the energy changes of water body movement. The parameter ρ is the seawater density, which is 1025 kilograms per cubic meter, and g is the acceleration of gravity, which is 9.81 meters per square second.

[0076] The calculation expression of the storm surge meteorological potential energy is as follows:

[0077] E m =∫ A (τ w x+τ w y+Δp)dA;

[0078] In the formula, E m is the meteorological potential energy of storm surge; τ w is wind stress; x, y are horizontal coordinates; Δp is the air pressure difference; A is the area of ​​the calculation region.

[0079] Calculation of storm surge meteorological potential energy: It is derived based on the effect of wind field and pressure field on water body. The optimization process takes into account the nonlinear effect of wind stress and introduces the relationship between wind speed and water surface roughness. The effect of this method is that it can accurately describe the influence of atmospheric forcing. The wind stress coefficient is calibrated according to the measured data, and the pressure gradient is calculated by interpolation of the pressure field.

[0080] The storm surge dynamic equation is expressed as follows:

[0081]

[0082] Where η is the water level; t is the time; x, y are the spatial coordinates; E t is the total potential energy of storm surge; τ is the time delay; α, β, γ, δ are unknown coefficients; ε is the error term.

[0083] The derivation and establishment process of the storm surge dynamic equation is described as follows: it is simplified based on the Navier-Stokes equation, taking into account the effects of inertia, Coriolis force, pressure gradient force and friction. The optimization process is reflected in the introduction of a time delay term to describe the hysteresis effect of the system, and the addition of an error term to characterize the uncertainty of the model. The effect of this equation is reflected in the ability to describe the generation and propagation process of storm surges. The coefficients are determined by genetic algorithm optimization, and the error term obeys a normal distribution.

[0084] The storm surge continuity equation is expressed as follows:

[0085]

[0086] Where η is the water level; h is the water depth; u and v are the horizontal flow velocity components.

[0087] The derivation and establishment process of the storm surge continuity equation is described as follows: It is based on the law of conservation of mass and describes the continuous movement of water bodies. The optimization process is reflected in the consideration of the influence of water depth changes. The effect of this equation is reflected in the guarantee of the stability of numerical calculations. The parameters are derived from measured terrain data.

[0088] The fitness function of the genetic algorithm is expressed as follows:

[0089]

[0090] Where F is the fitness value; η i To forecast water levels; is the measured water level; n is the number of samples.

[0091] Design of the fitness function of the genetic algorithm: Based on the prediction error definition, the inverse form of the root mean square error is adopted. The optimization process is reflected in the introduction of penalty terms to prevent overfitting. The effect of this function is that it can effectively guide parameter optimization. The number of samples is determined according to the length of historical data.

[0092] A second aspect of the present invention provides a computer-readable storage medium, in which program instructions are stored. When the program instructions are run in a computer, they are used to execute the above-mentioned single-point conventional storm surge forecasting method.

[0093] The third aspect of the present invention provides a single-point conventional storm surge forecasting system, comprising the above-mentioned computer-readable storage medium, wherein the system is any one of a computer, a server, and a single-chip microcomputer, and the computer-readable storage medium is arranged in the system, and the system is provided with a microprocessor for executing program instructions stored in the computer-readable storage medium.

[0094] Specifically, the principle of the present invention is: based on the principle of multi-scale analysis, the storm surge signal can be decomposed into eigenmode functions of different frequencies, which reflect the changing characteristics of the storm surge on different time scales. The empirical mode decomposition method decomposes the complex signal into a finite number of eigenmode functions through an iterative screening process, and each eigenmode function represents an oscillation of a characteristic time scale. This adaptive decomposition method does not require pre-setting of basis functions, and can be decomposed according to the characteristics of the signal itself, ensuring the physical meaning of the decomposition result.

[0095] From the energy perspective, the formation of storm surge is essentially a process of energy transfer and transformation. The hydrodynamic potential energy of storm surge reflects the position energy change of water body in the gravity field, and the meteorological potential energy of storm surge reflects the effect of atmospheric forcing on water body. By establishing the relationship between potential energy and water level change, the evolution process of storm surge can be described from the physical mechanism. The introduction of time delay term takes into account the lag effect of energy transfer, making the model more consistent with the actual physical process.

[0096] The construction of the forecast equation group is based on the basic laws of fluid mechanics, including the dynamic equations describing the movement of water bodies and the mass conservation equation. By introducing energy-based driving terms and time-lag terms, the equation's ability to describe sudden strong storm surges is enhanced. The optimization of model parameters uses genetic algorithms, which achieve global optimization of parameters by simulating the evolutionary process, thereby improving the adaptability of the model.

[0097] A specific embodiment 1 of the present invention is provided below, and the specific implementation method of each step in this embodiment 1 is described in detail as follows.

[0098] The specific implementation method of step S01 is: first, build a comprehensive monitoring system at the forecast target point, set up an ultrasonic water level meter and a pressure water level meter, the ultrasonic water level meter uses a high-precision sensor with a resolution of 1 mm, the beam divergence angle is controlled within 3 degrees, and the sampling frequency is once per minute. The pressure water level meter uses a crystal resonant sensor with an accuracy of 0.01% of the full scale, a range of 0 to 20 meters, and a sampling frequency of 2 times per minute. The data of the two water level meters are transmitted to the data processing center in real time through the data acquisition device, and the median filtering method is used to eliminate abnormal values; secondly, an acoustic Doppler flow meter is installed, using a 600 k The acoustic signal is measured with an accuracy of ±0.1 meters per second and a sampling frequency of once per minute. The velocity meter is installed on a fixed bracket 1 meter away from the seabed to ensure the stability of the measurement data. Finally, an automatic weather station is deployed, including wind direction sensor, wind speed sensor, air pressure sensor and temperature sensor. The wind direction sensor adopts the magnetoelectric principle with an accuracy of ±3 degrees. The wind speed sensor adopts a three-cup structure with an accuracy of ±0.3 meters per second. The air pressure sensor adopts the capacitive principle with an accuracy of ±0.1 hPa. The temperature sensor adopts a platinum resistance with an accuracy of ±0.1 degrees Celsius. The sampling frequency of all meteorological elements is once per minute. This step provides high-quality basic data for subsequent data analysis and model calculation by building a complete monitoring system.

[0099] The specific implementation of step S02 is: perform empirical mode decomposition on the acquired training data. This method decomposes the complex signal into a finite number of intrinsic mode functions based on the time scale characteristics of the signal itself. The mathematical expression of empirical mode decomposition is as follows: Where x(t) is the original signal, c i (t) is the i-th eigenmode function, r n (t) is the residual term, and n is the number of intrinsic mode functions. In the specific implementation, the signal is firstly preprocessed by endpoint extension, and the data of one tidal cycle is extended at both ends by the mirror extension method. Then, all the extreme points of the signal are determined by using the cubic spline difference, and the maximum points are connected to form the upper envelope, and the minimum points are connected to form the lower envelope. The average value of the upper and lower envelopes is calculated to obtain the average envelope, and the original signal is subtracted from the average envelope to obtain the first component. The above process is repeated until the decomposition termination condition is met. The termination condition is that the standard deviation of two adjacent decompositions is less than 0.0001 or the residual term becomes a monotonic function. For water level data, the sum of the energies of the first two intrinsic mode functions is defined as the water level harmonic component, and the sum of the energies of the remaining intrinsic mode functions is defined as the water level fundamental component; the same decomposition method is used for flow rate data and meteorological data to obtain the corresponding fundamental component and harmonic component. The role of this step is to decompose complex hydrological and meteorological elements into components of different scales, so as to facilitate the analysis of the contribution of each component.

[0100] The specific implementation of step S03 is: using sliding window technology to construct a feature matrix, the mathematical expression of the water level feature matrix is ​​as follows: Where H is the water level characteristic matrix, h i1 is the fundamental wave component of the water level at the ith moment, h i2 is the water level harmonic component at the i-th moment, and n is the number of sampling points. The mathematical expression of the velocity characteristic matrix is ​​as follows: Where v is the velocity characteristic matrix, v i1 is the fundamental component of velocity at the ith moment, v i2 is the harmonic component of flow velocity at the ith moment. The mathematical expression of the meteorological characteristic matrix is ​​as follows: Where M is the meteorological characteristic matrix, m i1 is the meteorological fundamental component at the i-th moment, m i2 is the meteorological harmonic component at the i-th moment. When constructing the feature matrix, a 24-hour sliding window is used with a sliding step of 1 hour. The data in each window is standardized to eliminate the dimension effect. The standardization adopts the minimum and maximum normalization method. The role of this step is to convert the time series decomposition results into a matrix form that is easy to calculate.

[0101] The specific implementation of step S04 is: using the singular value decomposition method to calculate the contribution coefficient of each characteristic matrix. The mathematical expression of singular value decomposition is as follows: X = USV T , where X is the matrix to be decomposed, U is the left singular matrix, S is the singular value matrix, V is the right singular matrix, and T represents the matrix transpose. The calculation expression of the contribution coefficient is as follows: Where w i is the contribution coefficient of the i-th component, s i is the i-th singular value. In the specific implementation, the characteristic matrix is ​​first centered, the mean of each column is subtracted, and then the covariance matrix is ​​calculated. The power iteration method is used to solve the eigenvalues ​​and eigenvectors of the covariance matrix, and the eigenvectors are sorted by the eigenvalue size to obtain the singular value matrix. Finally, the contribution coefficient of each component is calculated according to the singular value. This method introduces a threshold mechanism, which only retains components with a contribution rate of more than 5%. The threshold is selected using a cross-validation method. The purpose of this step is to quantitatively evaluate the impact of various factors on the formation of storm surges.

[0102] The specific implementation of step S05 is: calculate the total potential energy of the storm surge based on the principle of energy conservation. The calculation expression of the storm surge hydrodynamic potential energy is as follows: Where E h E is the storm surge hydrodynamic potential energy, ρ is the seawater density, which is 1025 kg / m3, g is the gravitational acceleration, which is 9.81 m / s2, h is the water depth, A is the calculation area, and z is the vertical coordinate. The calculation expression of storm surge meteorological potential energy is as follows: Em =∫ A (τ w x+τ w y+Δp)dA, where E m is the meteorological potential energy of storm surge, τ w is wind stress, x, y are horizontal coordinates, and Δp is the pressure difference. The numerical integration method is used in the calculation. The integration step is dynamically adjusted according to the change of water depth, and the integration area is determined according to the impact range of the storm surge. The wind stress coefficient is calibrated by measured data, and the pressure gradient is calculated by interpolation of the pressure field. The role of this step is to describe the formation mechanism of storm surge from an energy perspective.

[0103] The specific implementation of step S06 is: using the finite difference method to calculate the historical water level change rate, firstly perform quality control on the historical water level data, remove outliers that exceed the mean plus or minus 3 times the standard deviation, and then use the central difference format to calculate the water level change rate, the expression is: Where η is the water level, t is the time, Δt is the time step, which is 1 hour, and i is the time series index. In order to eliminate high-frequency noise, the calculated water level change rate is processed by a 5-point sliding average, and the expression is: In the formula is the water level change rate after smoothing. The purpose of this step is to obtain the dynamic characteristics of water level changes.

[0104] The specific implementation of step S07 is: establish a storm surge water level forecast equation group based on fluid mechanics theory. The storm surge dynamic equation is expressed as follows: Where η is the water level, t is the time, x and y are the spatial coordinates, and E t is the total potential energy of the storm surge, τ is the time delay, α, β, γ, δ are unknown coefficients, and ε is the error term. This equation contains the inertia term Diffusion term Energy driving term γE t , time lag term and the error term ε. The storm surge continuity equation is expressed as follows: Where h is the water depth, u and v are the horizontal flow velocity components. The relationship between potential energy and water level change rate is derived based on the Bernoulli equation, and the relationship between water level and flow velocity is derived based on the Navier-Stokes equation. The purpose of this step is to establish the theoretical basis for storm surge forecasting.

[0105] The specific implementation of step S08 is: using a genetic algorithm to optimize the parameters of the storm surge water level forecast equation group. The mathematical expression of the fitness function is as follows: Where F is the fitness value, η i To predict water levels, is the measured water level, and n is the number of samples. The parameters of the genetic algorithm are set as follows: population size 100, chromosome length equal to the number of parameters to be optimized, gene encoding using real number encoding, evolutionary generations 1000, crossover probability 0.8, and mutation probability 0.1. The selection operation uses the roulette method, the crossover operation uses arithmetic crossover, and the mutation operation uses uniform mutation. In order to prevent overfitting, a penalty term is introduced Where λ is the penalty coefficient, which takes the value of 0.01, θ j is the jth parameter to be optimized, and m is the number of parameters. The purpose of this step is to improve the accuracy of the prediction equation.

[0106] The specific implementation method of step S09 is: collect the hydrological and meteorological elements of the forecast target point in real time through the established monitoring system. The water level monitoring equipment collects water level data at a frequency of once per minute, the acoustic Doppler current meter collects flow velocity data at a frequency of once per minute, and the automatic weather station collects meteorological data at a frequency of once per minute. The collected data is transmitted to the data processing center in real time through the wireless transmission module, and the real-time quality control method is used to eliminate abnormal values. The judgment of abnormal values ​​adopts the 3 times standard deviation criterion. The role of this step is to obtain the input data required for real-time forecasting.

[0107] The specific implementation of step S10 is: decomposing the real-time monitoring data by online empirical mode decomposition method. The mathematical expression of this method is the same as step S02, except that the sliding window technology is used to achieve real-time processing, the window length is 6 hours, and the decomposition results in the window are updated every time new data arrives. In order to improve the computational efficiency, an incremental update strategy is adopted, and only the newly added data is decomposed and merged with the original results. The function of this step is to obtain the various components of the real-time data.

[0108] The specific implementation method of step S11 is: construct a real-time feature matrix. The mathematical expressions of the real-time water level feature matrix, the real-time flow velocity feature matrix and the real-time meteorological feature matrix are the same as those of step S03, except that a real-time update strategy is adopted to update the matrix every hour. The recursive least squares method is used during the update to avoid repeated calculations and improve the operation efficiency. The function of this step is to convert the real-time data into a form suitable for model input.

[0109] The specific implementation of step S12 is: perform storm surge forecast calculation. Input the real-time characteristic matrix into the optimized storm surge water level forecast equation group, and use the fourth-order Runge Kutta method to solve the equation group. The expression of the fourth-order Runge Kutta method is: k1=f(t n ,y n ), k4=f(t n +Δt,y n +Δtk3), Where t nis the current moment, y n is the current value, Δt is the time step, which is 1 hour, f is the right-hand side of the equation, and k1, k2, k3, and k4 are the slopes of the four stages. The forecast time is 24 hours. To improve the calculation efficiency, parallel computing technology is used to divide the forecast area into several sub-areas for simultaneous calculation. The purpose of this step is to obtain the forecast value of the future water level.

[0110] The specific implementation of step S13 is: perform storm surge warning judgment. First, calculate the storm surge water value, which is defined as the difference between the actual water level and the astronomical tide level, expressed as: Δη=η-η a , where Δη is the water increase value, η is the actual water level, and η a is the astronomical tide level. Then the duration of the water increase is calculated, and the duration is defined as the continuous length of time that the water increase value exceeds the specified threshold. The warning level is determined according to the water increase value and duration, and the warning threshold is set as follows: the yellow warning threshold is a water increase of 1.5 meters or a duration of more than 6 hours; the orange warning threshold is a water increase of 2.0 meters or a duration of more than 12 hours; the red warning threshold is a water increase of 2.5 meters or a duration of more than 24 hours. When the forecast result exceeds the warning threshold, the warning signal is automatically issued through the warning system. The issuance of the warning signal adopts a multi-channel method, including SMS, telephone, network, etc., to ensure that the warning information is delivered to relevant personnel in a timely manner. The purpose of this step is to timely detect the risk of storm surge disasters and issue warnings.

[0111] The core innovation of the entire method in this embodiment is: using the empirical mode decomposition method to achieve multi-scale analysis of hydrological and meteorological elements; introducing a time delay term to describe the lag effect of storm surge; constructing a forecast equation based on the principle of conservation of energy; and using a genetic algorithm to optimize model parameters.

[0112] In order to better understand and implement the present invention, the following provides an embodiment 2 of a specific application scenario of the present invention: In January 2024, when a coastal city marine forecast center carried out storm surge forecasting business, it implemented the storm surge single-point conventional forecasting method of the present invention in response to the storm surge disasters caused by frequent strong typhoons in the sea area. The forecast center selected an observation station located at the main port of the city as the forecast target point, which is located at 22.5 degrees north latitude and 114.5 degrees east longitude, with an average water depth of 25 meters.

[0113] First, a comprehensive monitoring system was installed at the forecast target point, including 1 set of ultrasonic water level gauges and 1 set of pressure water level gauges, which were installed at different locations on the observation platform to achieve data complementarity. The ultrasonic water level gauge uses the RLS model of OTT, with a measurement accuracy of ±1 cm and a sampling frequency of 1 time per minute; the pressure water level gauge uses the U20 model of HOBO, with a measurement accuracy of ±0.5 cm and a sampling frequency of 2 times per minute. An acoustic Doppler current meter was installed 10 meters south of the water level gauge, using the Workhorse model of RDI, with a measurement accuracy of ±0.1 meters per second and a sampling frequency of 1 time per minute. An automatic weather station was installed on the top of the observation platform, including a wind direction sensor, a wind speed sensor, an air pressure sensor, and an air temperature sensor. The sampling frequency of all sensors is 1 time per minute.

[0114] The forecast center collected historical observation data from January 1, 2023 to December 31, 2023 as training data, with a data volume of 525,600. These data were subjected to empirical mode decomposition. Taking a set of data at 12:00 on July 15, 2023 as an example, the water level data was decomposed into 8 intrinsic mode functions, of which the sum of the energy of the first and second eigenmode functions accounted for 87.3% of the total energy and was defined as the water level harmonic component. The sum of the energy of the remaining eigenmode functions was defined as the water level fundamental component. Similar decomposition was performed on flow velocity data and meteorological data, as shown in Table 1.

[0115] Table 1 Decomposition results of each factor at a certain moment

[0116] Feature Type fundamental component Harmonic Components Water level (m) 2.15 0.85 Flow rate (m / s) 0.45 0.25 Wind speed (m / s) 12.5 3.5

[0117] According to the decomposition results, the characteristic matrix is ​​constructed, and a 24-hour sliding window is used to obtain the dimension of the water level characteristic matrix of 1440×2, the dimension of the flow velocity characteristic matrix of 1440×2, and the dimension of the meteorological characteristic matrix of 1440×2. The contribution coefficients of each matrix are calculated by singular value decomposition, and the results show that the contribution coefficient of the fundamental wave component of the water level is 0.72, and the contribution coefficient of the harmonic component is 0.28; the contribution coefficient of the fundamental wave component of the flow velocity is 0.65, and the contribution coefficient of the harmonic component is 0.35; the contribution coefficient of the fundamental wave component of the meteorological component is 0.81, and the contribution coefficient of the harmonic component is 0.19.

[0118] Based on the calculated contribution coefficient, the total potential energy of the storm surge is calculated. The calculation area of ​​the forecast target point is set to 5 km × 5 km, and the area is divided into 50 × 50 grids for numerical integration. The calculation results show that the hydrodynamic potential energy of the storm surge is 3.2 × 106 joules, the meteorological potential energy of the storm surge is 2.8 × 106 joules, and the total potential energy of the storm surge is 6.0 × 106 joules.

[0119] When establishing the storm surge water level forecast equation group, the genetic algorithm is used to optimize the model parameters. The population size is set to 100, the evolutionary generation is set to 1000, the crossover probability is set to 0.8, and the mutation probability is set to 0.1. After optimization, the parameter values ​​obtained are: α = 0.015, β = 0.012, γ = 0.008, δ = 0.6, and the time delay τ = 3 hours.

[0120] On January 10, 2024, the sea area was affected by a strong typhoon, and the forecast center launched a storm surge forecast program. The system obtains real-time monitoring data every hour, processes the data in real time and calculates forecasts. Taking the forecast at 14:00 on January 10 as an example, the system predicts that the maximum water increase in the next 24 hours will be 2.3 meters, which will occur at 2:00 on January 11 and last for 8 hours. According to the warning threshold, the forecast result meets the orange warning standard. The forecast center immediately issued an orange storm surge warning signal.

[0121] The actual observation results show that the actual water increase at 2:00 on January 11 was 2.1 meters, which was 0.2 meters different from the forecast value, and the forecast time error was 30 minutes. The traditional statistical model method predicted a water increase of 1.7 meters with a time error of 2 hours; the numerical simulation method predicted a water increase of 2.4 meters with a time error of 1.5 hours. Comparative analysis shows that the forecast method of the present invention is superior to the traditional method in terms of accuracy and timeliness.

[0122] Before adopting the method of the present invention, the forecast center mainly used statistical modeling and numerical simulation methods to forecast storm surges. The statistical modeling method establishes a regression equation based on historical data. The calculation speed is fast but the accuracy is not high. It often under-predicts sudden storm surges. Although the numerical simulation method has high forecast accuracy, it takes a long time to calculate. It usually takes more than 3 hours to complete a forecast calculation, which is difficult to meet the needs of real-time forecasting. The method of the present invention not only ensures the forecast accuracy but also improves the calculation efficiency by introducing multi-scale analysis and the principle of energy conservation. The single forecast calculation time is shortened to 15 minutes, which provides strong support for the timely implementation of early warning work.

[0123] It can be seen from this embodiment that the method of the present invention has shown significant technical advantages in practical applications: high forecast accuracy, with the average error controlled within 0.2 meters; long forecast lead time, which can be predicted 24 hours in advance; high calculation efficiency, a complete forecast calculation can be completed in 15 minutes; and a reasonable warning level meets actual disaster prevention needs.

[0124] It should be noted that the variables involved in the present invention are explained in detail as shown in Table 2 below.

[0125] Table 2 Variable explanation table

[0126]

[0127]

[0128] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A conventional single-point storm surge forecasting method, characterized in that: The following steps are involved: At the forecast target point, historical water level data, historical flow velocity data and historical meteorological data are collected as training data, the training data are time-series decomposed to obtain fundamental wave components and harmonic components and a feature matrix is ​​constructed, the water level contribution coefficient, the flow velocity contribution coefficient and the meteorological contribution coefficient are calculated, and a storm surge water level forecast equation group including a storm surge dynamic equation and a storm surge continuity equation is constructed according to the contribution coefficients, and real-time water level data, real-time flow velocity data and real-time meteorological data are input into the storm surge water level forecast equation group to obtain a storm surge forecast result, and a warning signal is output when the storm surge forecast result exceeds a warning threshold; the storm surge water level forecast equation group includes a storm surge dynamic equation and a storm surge continuity equation, the storm surge dynamic equation describes the relationship between the total potential energy of the storm surge, the historical water level data and the historical water level change rate, and the storm surge continuity equation describes the relationship between the historical water level data and the historical flow velocity data.

2. The storm surge single point conventional forecasting method according to claim 1, characterized in that: The step of collecting training data is to set up water level monitoring equipment, acoustic Doppler flowmeter and automatic weather station at the forecast target point. The water level monitoring equipment includes ultrasonic water level gauge and pressure water level gauge. The automatic weather station is used to monitor wind direction, wind speed, air pressure and temperature.

3. The single-point conventional storm surge forecasting method according to claim 1, characterized in that: The step of time series decomposition is to use the empirical mode decomposition method to decompose the historical water level data into water level fundamental component and water level harmonic component, decompose the historical flow velocity data into flow velocity fundamental component and flow velocity harmonic component, and decompose the historical meteorological data into meteorological fundamental component and meteorological harmonic component.

4. The single-point conventional storm surge forecasting method according to claim 1, characterized in that: The step of constructing a characteristic matrix is ​​specifically to use a sliding window technology to construct the water level fundamental component and the water level harmonic component into a water level characteristic matrix, the flow velocity fundamental component and the flow velocity harmonic component into a flow velocity characteristic matrix, and the meteorological fundamental component and the meteorological harmonic component into a meteorological characteristic matrix.

5. The single-point conventional storm surge forecasting method according to claim 1, characterized in that: The step of calculating the contribution coefficient is specifically to use the principal component analysis method to calculate the water level contribution coefficient of the water level characteristic matrix, the flow velocity contribution coefficient of the flow velocity characteristic matrix and the meteorological contribution coefficient of the meteorological characteristic matrix through singular value decomposition.

6. The single-point conventional storm surge forecasting method according to claim 1, characterized in that: The step of constructing a storm surge water level forecast equation group is specifically to calculate the total potential energy of the storm surge based on the water level contribution coefficient and the meteorological contribution coefficient, the total potential energy of the storm surge is composed of the sum of the storm surge hydrodynamic potential energy and the storm surge meteorological potential energy, the storm surge dynamic equation is established based on the total potential energy of the storm surge, and the storm surge continuity equation is established based on the historical water level data and the historical flow rate data.

7. The single-point conventional storm surge forecasting method according to claim 6, characterized in that: The parameters of the storm surge water level forecast equation group are optimized using a genetic algorithm, the population size of the genetic algorithm is 100, the evolutionary generation is 1000, the crossover probability is 0.8, and the mutation probability is 0.

1.

8. The single-point conventional storm surge forecasting method according to claim 1, characterized in that: The warning thresholds include yellow warning threshold, orange warning threshold and red warning threshold. The yellow warning threshold is a water increase of 1.5 meters or a duration of more than 6 hours, the orange warning threshold is a water increase of 2.0 meters or a duration of more than 12 hours, and the red warning threshold is a water increase of 2.5 meters or a duration of more than 24 hours.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program instructions, and when the program instructions are executed in a computer, they are used to execute the conventional single-point storm surge forecasting method according to any one of claims 1 to 8.

10. A single-point conventional storm surge forecasting system, characterized in that: The system comprises the computer-readable storage medium as claimed in claim 9, wherein the system is any one of a computer, a server, and a single-chip microcomputer, the computer-readable storage medium is arranged in the system, and the system is provided with a microprocessor for executing program instructions stored in the computer-readable storage medium.

Images