Piano key weir discharge spatial distribution prediction method, system and storage medium

Abstract

The application discloses a kind of keyboard weir discharge space distribution prediction method, system and storage medium, obtain the geometric parameter and hydraulic parameter of target keyboard weir and are converted into dimensionless geometric parameter group;Based on the parameter group, respectively extract physical constraint structure, calculate the dimensionless local discharge intensity of each overflow front edge of target keyboard weir;Extract the geometric length of each overflow front edge, calculate the product of local discharge intensity and corresponding length, and all products are ratio normalized operation, obtain the distribution coefficient of each part;Get the total discharge flow model constructed in advance and calculate total flow, then multiplied by each distribution coefficient to output partial discharge flow.The application reduces the incoherent deviation of normalization caused by traditional independent fitting through physical mechanism parameterization and normalization division logic, and improves the robust prediction ability of the model under extreme head working condition.

Description

Technical Field

[0001] This invention relates to the field of hydraulic engineering design and numerical prediction, and in particular to a method, system and storage medium for predicting the spatial distribution of discharge flow of a key weir. Background Technology

[0002] Piano key weirs, due to their polygonal overflow front design, possess excellent total discharge capacity under pre-set weir width conditions and are widely used in dam flood control renovation and water diversion projects. During the engineering design phase, in addition to understanding the total weir discharge, the precise layout of downstream energy dissipation and scour prevention facilities, as well as the physical strength verification of the sidewall structure, depends on the physical distribution ratio of the discharge borne by the three different overflow fronts: the inlet, outlet, and sidewalls. Therefore, developing a fundamental calculation method that can bypass costly computational fluid dynamics (CFD) numerical simulations and directly predict the spatial distribution of discharge based on basic geometric and hydraulic parameters is of significant technical importance for promoting the standardization of piano key weir engineering.

[0003] In existing studies on the flow distribution prediction of key weirs, a single numerical regression method is typically used to process physical experimental data. This type of method extracts a large number of model test records and establishes polynomial fitting networks for the sidewall distribution coefficients, inlet distribution coefficients, and outlet distribution coefficients, respectively. In practice, the model takes parameters such as relative head and gate width ratio as input variables and uses independently evolving positive power functions or multivariate linear equations to calculate the percentage of flow handled by each part.

[0004] However, in terms of deep hydraulic logic and mathematical consistency, the mathematical prediction models in existing prediction schemes lack a global normalization constraint mechanism. Since the fitting functions for the distribution coefficients of the three different overflow fronts are constructed independently, under any preset hydraulic conditions, the sum of the three ratios in the calculated output is difficult to be locked at 1, which can easily lead to water overflow or loss when multiplying back to the total flow. Summary of the Invention

[0005] Purpose of the invention: In order to solve the technical problems existing in the prior art, the present invention proposes a method, system and storage medium for predicting the spatial allocation of discharge flow of a piano key weir.

[0006] Technical solution: Firstly, a method for predicting the spatial distribution of discharge flow at a key-weir, comprising:

[0007] Obtain the geometric and hydraulic parameters of the target key weir and convert them into a dimensionless set of geometric parameters;

[0008] Based on the dimensionless geometric parameter set, the dimensionless local discharge intensity of each overflow front edge of the target key weir is calculated. Each overflow front edge includes the sidewall front edge, the inlet front edge, and the outlet front edge.

[0009] Based on geometric parameters, the geometric length of each overflow leading edge is extracted, the dimensionless local discharge intensity of each overflow leading edge is calculated as a product of the corresponding geometric length, and the ratio normalization operation is performed on all products to obtain the distribution coefficient of each overflow leading edge.

[0010] Obtain a pre-built model for calculating the total discharge flow rate, and calculate the total discharge flow rate of the target key weir based on geometric and hydraulic parameters;

[0011] Calculate the product of the total discharge flow rate and the distribution coefficient of each overflow front, and output the partial discharge flow rate of each overflow front.

[0012] According to one aspect of this application, the dimensionless local discharge intensity at the leading edge of the sidewall is calculated, specifically including:

[0013] Obtain the relative head and relative outlet gate width from the dimensionless geometric parameter set;

[0014] Based on the relative head and the relative outlet gate chamber width, calculate the water tongue collision parameters that characterize the ratio of the horizontal throw distance to the gate chamber width when the overflow water tongues from the sidewalls move towards each other in the gate chamber.

[0015] The water tongue collision parameters are input into a pre-constructed attenuation function for calculation to obtain the dimensionless local discharge intensity at the leading edge of the sidewall.

[0016] According to one aspect of this application, the water tongue collision parameters are calculated in the following manner:

[0017] The initial collision parameters are obtained by calculating the ratio of the arithmetic square root of the relative head to the relative width of the outlet gate chamber.

[0018] Obtain the relative retaining wall height from the dimensionless set of geometric parameters;

[0019] By using a linear correction factor that includes the relative height of the retaining wall, the influence of the drop height in the initial collision parameters is coupled and corrected to obtain the water tongue collision parameters.

[0020] According to one aspect of this application, the dimensionless local discharge intensity at the inlet leading edge and the outlet leading edge is calculated separately, including:

[0021] Extract the relative head and the corresponding geometric ratio parameters from the dimensionless geometric parameter set;

[0022] Based on relative head, a competitive response mechanism term is constructed to characterize the head-driven enhancement effect and the gate chamber filling inhibition effect in the inlet and outlet flow regimes;

[0023] Obtain the asymptotic lower bound term for characterizing the residual discharge capacity of a broad-crested weir under extreme inundation conditions;

[0024] The dimensionless local discharge intensity at the inlet leading edge and outlet leading edge is obtained by combining the competitive response mechanism term with the asymptotic lower bound term and multiplying it by a correction coefficient based on the corresponding geometric ratio parameter.

[0025] According to one aspect of this application, the total discharge capacity of the target key weir is calculated, specifically including:

[0026] By inputting the geometric and hydraulic parameters into the pre-constructed empirical formula or polynomial regression model of the total discharge coefficient of the key weir, the total discharge coefficient of the key weir can be obtained.

[0027] By combining the general weir flow theory formula with the total discharge coefficient of the key weir, the total discharge flow of the target key weir is calculated.

[0028] According to one aspect of this application, the basic and modified parameters for calculating the dimensionless local discharge intensity are obtained through the following hierarchical calibration strategy:

[0029] Obtain the first sample set without retaining walls and excluding extreme high head conditions. Use the least squares deviation of the predicted and measured values ​​of the normalized distribution coefficients as the objective function to fit the conventional undetermined coefficients of the sidewall and inlet / outlet strength formulas. The asymptotic lower bound proportional coefficient is temporarily taken as a preset initial constant value.

[0030] Fix this set of coefficients, obtain a second sample set containing extreme high head conditions, and calibrate the asymptotic lower bound ratio coefficients of the inlet and outlet;

[0031] With the parameters of the first two layers fixed, a third sample set containing multiple retaining wall heights is obtained, and retaining wall correction coefficients for coupling retaining wall effects are fitted separately through regression.

[0032] According to one aspect of this application, the horizontal throw distance is derived by equating the sidewall overflow flow state to projectile motion, and the derivation is based on the following physical mapping conditions:

[0033] The total flow velocity over the top of the target key weir sidewall is defined to include a normal velocity component. The magnitude of the normal velocity component is driven by the local head and is characterized by a normal velocity scaling factor.

[0034] Extract the geometric tilt characteristics along the sidewall and calculate the average drop height coefficient representing the drop from the top of the sidewall to the bottom plate of the outlet gate chamber;

[0035] The calculation structure of the water tongue collision parameters is a dimensionless form of the product of the normal velocity component and the fall time.

[0036] According to one aspect of this application, after outputting the partial discharge flow rates at each overflow leading edge, the method further includes a step of jointly evaluating the prediction accuracy based on error propagation.

[0037] Obtain the relative error of the allocation coefficient caused by the pre-stored dimensionless local discharge intensity model, and the pre-stored total flow relative error of the pre-built model used to calculate the total discharge flow rate;

[0038] The square of the relative error of the allocation coefficient is summed with the square of the relative error of the total flow.

[0039] Calculate the arithmetic square root of the summation result to obtain and output the relative prediction error of the discharge flow rate of each overflow leading edge.

[0040] In a second aspect, a key-weir discharge spatial allocation prediction system is provided. The system includes a processor coupled to a memory. The memory stores instructions, which, when executed by the processor, implement any one of the methods in the first aspect.

[0041] Thirdly, a computer-readable storage medium storing computer instructions that, when executed on a computer, implement any one of the methods in the first aspect.

[0042] Beneficial effects: By using physical mechanism parameterization and normalization division logic, the normalization inconsistency bias caused by traditional independent fitting is reduced, and the robust predictive ability of the model under extreme head conditions is improved. Attached Figure Description

[0043] Figure 1 This is a flowchart of the spatial distribution prediction process for the discharge flow of the key weir in this invention.

[0044] Figure 2 This is a flowchart of the present invention for calculating the dimensionless local discharge intensity at the leading edge of the sidewall.

[0045] Figure 3 This is a flowchart of the calculation of water tongue collision parameters according to the present invention.

[0046] Figure 4 This is a flowchart of the present invention for calculating the dimensionless local discharge intensity at the leading edge of the inlet and outlet.

[0047] Figure 5 This is a flowchart of the joint evaluation process for prediction accuracy based on error propagation, as described in this invention. Detailed Implementation

[0048] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0049] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein are implemented in a sequence other than those illustrated or described herein. Furthermore, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to the process, method, product, or apparatus.

[0050] The background section mentions that existing mathematical prediction models lack a global normalization constraint mechanism, leading to issues such as water overflow or missing values ​​when multiplying back the total flow. Furthermore, the use of an infinitely extrapolated power function to construct the sidewall attenuation law fails to consider the geometric threshold of projectile fall during water tongue collisions; the inlet and outlet models also fail to incorporate parameters of the residual base of the broad-crested weir under high heads. In other words, existing fitting functions fail to accurately map the true physical boundaries of water flow intersection and collision phenomena. When prediction conditions reach extreme conditions such as lower or ultra-high submerged heads, this type of physically unfounded fitting can easily lead to problems such as the output of the allocation prediction model tending towards infinity or the proportions being reversed contrary to common sense.

[0051] To solve these problems, combined with Figures 1 to 5 The present invention will be specifically described through the following embodiments.

[0052] Example 1

[0053] According to one aspect of this application, this embodiment provides a method for predicting the spatial allocation of discharge flow at a key weir, specifically including:

[0054] Step 101: Obtain the geometric and hydraulic parameters of the target key weir and convert them into a set of dimensionless geometric parameters.

[0055] In this embodiment, the fundamental physical quantities used to describe the physical morphology and fluid state of the keyway weir are its geometric and hydraulic parameters. The geometric parameters include the weir height P, weir width W, and inlet gate chamber width W. _i , Exit gate chamber width W _o Side wall length B, inverted length of the inlet B _i Export overhang length B _o The height h and thickness T of the retaining wall _sThis reveals the data consumption path of Ts. The hydraulic parameter is the total head H at the weir, characterizing the current fluid energy. After obtaining these basic parameters, a division operation is performed to convert them into a set of dimensionless geometric parameters in pure numerical form. This reduces the error impact of physical scale scaling and improves the universality of the prediction model.

[0056] The dimensionless geometric parameter set includes the inlet and outlet gate chamber width ratio W. _i / W _o Relative exit gate chamber width W _o / P, relative wall height h / P, overhang length ratio B _o / B _i Relative sidewall length B / P, relative head H _rel H _rel It is the ratio of H to P.

[0057] In some alternative implementations, certain derived parameters are not used as independent variables in subsequent local discharge intensity calculations. For example, the width ratio L / W, which is a geometrically related quantity determined by the gate chamber width, side weir length, and wall thickness, can be separated and used in subsequent total discharge calculations to avoid parameter collinearity and redundancy problems in multiple regression modeling.

[0058] Step 102: Based on the dimensionless geometric parameter set, calculate the dimensionless local discharge intensity of each overflow front edge of the target key weir. Each overflow front edge includes the sidewall front edge, the inlet front edge, and the outlet front edge.

[0059] In this embodiment, the dimensionless local discharge intensity is used to characterize the relative discharge capacity per unit length of a preset overflow front under the current head conditions. Unlike the traditional black-box method of directly fitting the flow rate ratio, this embodiment performs decoupled modeling based on the flow physics mechanism of different parts. Specifically, the system will analyze the specific values ​​in the dimensionless geometric parameter set and call the corresponding prediction model to independently calculate the dimensionless local discharge intensity of the inlet front, outlet front, and sidewall front.

[0060] Furthermore, for different overflow leading edges, dimensionless local discharge intensity calculation functions with different response mechanisms can be used according to their physical boundary conditions. For example, the intensity of the sidewall leading edge can be calculated using a decay function with a physical upper limit, and the intensity of the inlet leading edge and outlet leading edge can be calculated using a peak response function with a lower limit.

[0061] Step 103: Extract the geometric length of each overflow leading edge, calculate the product of the dimensionless local discharge intensity of each overflow leading edge and the corresponding geometric length, and perform ratio normalization on all products to obtain the distribution coefficient of each overflow leading edge.

[0062] The geometric length of each overflow leading edge refers to the actual unfolded length of the structure participating in the overflow. Specifically, the lengths of each part are extracted based on the geometric topology of the key weir. Among them, the geometric length L of the inlet leading edge... _i Equal to the width W of the inlet gate chamber _i Geometric length L of the exit leading edge _o Equal to the width W of the exit gate chamber _o Geometric length L of the leading edge of the sidewall _s This is equal to the sum of the lengths of the two weirs, i.e., 2 × B. After obtaining the geometric length, the dimensionless local discharge intensity of each leading edge segment is multiplied by its geometric length to obtain a product term characterizing the relative magnitude of the absolute flow capacity of that leading edge segment. Correspondingly, the product term of the preset leading edge is divided by the sum of all three leading edge product terms, and a ratio normalization operation is performed. This ratio structure ensures that the sum of the three allocation coefficients is always equal to 1, which avoids the loopholes of traditional independent multiple regression fitting, where the sum of the proportions of each part deviates from 100% and the normalization is inconsistent.

[0063] Step 104: Obtain the pre-built model for calculating the total discharge flow rate, and calculate the total discharge flow rate of the target key weir based on the key weir's geometric parameters and hydraulic parameters.

[0064] To achieve the final physical quantity output, the system needs to incorporate the total discharge flow of the entire weir. Specifically, the system calls an externally pre-built empirical model, inputting the acquired parameters to obtain the global flow rate. In this embodiment, the general weir flow theory formula from subsequent step 402 can be used as the total discharge flow rate calculation model.

[0065] Step 105: Calculate the product of the total discharge flow rate and the distribution coefficient of each overflow front edge, and output the partial discharge flow rate of each overflow front edge.

[0066] In this embodiment, the system performs a linear scalar multiplication operation, multiplying the total discharge flow, representing the overall discharge capacity of the key weir, by the distribution coefficients of the three overflow fronts mentioned above. This restores the relative proportions to the absolute physical quantities actually required for engineering design and structural verification. Based on this, the output discharge flow rates of each overflow front can be used for the scour-resistant arrangement of downstream energy dissipation facilities and the structural stress analysis of the weir sidewalls.

[0067] In addition to a single flow rate value, the output can also incorporate the theoretical confidence range of the prediction result by combining the model error propagation mechanism, thereby improving the reliability of the engineering prediction system.

[0068] Example 2

[0069] Without considering complex flow mechanisms, existing technologies typically employ a purely empirical negative power function to describe the decrease in sidewall discharge efficiency with increasing head. The corresponding calculation formula is:

[0070] γ _s =c _s0 ×(H _rel ) -βs ×M _geo,s ;

[0071] Where βs is a positive number to reflect the decay trend, M _geo,s This is the geometric correction factor for the sidewall.

[0072] When the relative head H _rel When the power function approaches zero, the local discharge intensity calculated by the negative power function will approach infinity, which contradicts the common sense that the maximum discharge capacity of the sidewall is limited under low head. To solve this technical problem, based on Example 1, this embodiment provides a quantitative prediction and physical deduction scheme for the sidewall discharge intensity.

[0073] One possible implementation includes the following steps:

[0074] Step 201: Obtain the relative head and relative outlet gate width from the dimensionless geometric parameter set.

[0075] Specifically, the corresponding dimensionless scalar is extracted from the geometric parameter dataset. The relative head is H. _rel The relative width of the exit gate chamber is represented by W. _o / P, two parameters used to characterize the boundary conditions of water collision within the sluice gate.

[0076] In some alternative implementations, if the underlying input set contains only absolute geometric values, a division operation between H and P, and W can be performed. _o The division operation with P is used to obtain the desired dimensionless extraction amount.

[0077] Step 202: Based on the relative head and the relative outlet gate chamber width, calculate the water tongue collision parameters, which characterize the ratio of the horizontal throw distance to the gate chamber width when the overflow water tongues from the sidewalls move towards each other in the gate chamber.

[0078] In this embodiment, the total flow velocity over the top of the target key weir sidewall is defined to include a normal velocity component, which is the velocity perpendicular to the sidewall as it enters the outlet gate chamber. This normal velocity component v _n Driven by the total head H above the weir, the corresponding calculation formula is:

[0079] v _n =φ×sqrt(2×g×H).

[0080] Where φ is the normal velocity scaling factor. Simultaneously, the geometric tilt characteristics along the sidewall are extracted, and the average drop height coefficient ξ, representing the fall from the top of the sidewall to the bottom of the exit gate chamber, is calculated, with sqrt representing the square root operation.

[0081] Furthermore, the horizontal throw distance is derived by equating the overflow flow pattern of the sidewall to projectile motion. This derivation is based on the following physical mapping conditions: the total flow velocity crossing the top of the target key weir sidewall includes a normal velocity component, the magnitude of which is driven by the local head and characterized by a normal velocity scaling factor; the geometric tilt characteristics along the sidewall are extracted, and the average drop height coefficient representing the fall from the top of the sidewall to the bottom of the outlet gate chamber is calculated; the calculation structure of the water tongue collision parameters is a dimensionless form of the product of the normal velocity component and the drop time. In other words, the mathematical structure of the water tongue collision parameters is analytically derived from the projectile motion distance calculation formula, which combines the normal velocity component and the average drop height coefficient, where the normal velocity scaling factor and the average drop height coefficient are merged and absorbed into a pre-calibrated collision comprehensive coefficient.

[0082] Accordingly, based on the equations of free fall and uniform linear motion, the horizontal throw distance of the water jet reaching the bottom of the outlet gate chamber is calculated, and the corresponding calculation formula is as follows:

[0083] x=v _n ×sqrt(2z _f / g);

[0084] z _f =ξ×P;

[0085] Among them, z _f This refers to the height of the fall.

[0086] In this embodiment, after substituting the expressions for normal velocity and drop height and merging constant terms, the projection distance expression is transformed into one that is proportional to sqrt(H×P). The water tongue collision parameter λ is defined as the sum of the total projection distance 2x of the water tongues on both sides and the width W of the outlet gate chamber. _o The ratio of , and thus its mathematical structure is proportional to sqrt(H) _rel ) / (W _o / P). To simplify the calculation process, the system incorporates physical constants that are difficult to measure directly, such as the normal velocity scaling factor φ and the average drop height coefficient ξ, into the calibrable collision composite coefficient c. _λ middle.

[0087] In practical engineering, the value of the average drop height coefficient ξ depends on geometric characteristics such as the difference in inverted overhang lengths at the inlet and outlet of the keyway and the slope of the bottom plate. Since ξ has been incorporated into the collision comprehensive coefficient c... _λ In this process, its value does not need to be determined independently and can be determined during offline calibration. _λ It was determined uniformly.

[0088] In this embodiment, in order to quantify the qualitative phenomenon of water tongue collision in hydraulics into a calculable dimensionless geometric ratio, the system incorporates underlying computational logic based on projectile dynamics mapping derivation.

[0089] Furthermore, in one possible implementation, the water tongue collision parameters are calculated as follows:

[0090] Step 2021: Calculate the ratio of the arithmetic square root of the relative head to the relative width of the outlet gate chamber to obtain the initial collision parameters.

[0091] Specifically, the square root operation is performed on the extracted relative head to obtain sqrt(H). _rel The square root result is used as the dividend, and the relative outlet gate width is used as the divisor to perform a division operation, thereby obtaining the initial collision parameters that reflect the conflict trend of the foundation hydraulic structure.

[0092] Step 2022: Obtain the relative retaining wall height from the dimensionless geometric parameter set.

[0093] To enhance the model's topology adaptability, the system introduces a retaining wall existence determination. Specifically, the relative retaining wall height is the pure number h / P obtained by dividing the absolute retaining wall height h by the weir height P. When the key weir does not contain a retaining wall structure, the value extracted by this operation is 0.

[0094] Furthermore, in the engineering drawing digitization input system, the system determines that the data of the retaining wall geometry layer is not empty, thereby triggering the extraction command of the relative retaining wall height data, so that the model parameter input is complete.

[0095] Step 2023: Using a linear correction factor that includes the relative wall height, the drop height effect in the initial collision parameters is coupled and corrected to obtain the water tongue collision parameters.

[0096] Adding a retaining wall raises the starting position of the effective overflow from the sidewall, resulting in a longer vertical drop distance for the water jet, which in turn changes the collision intensity of the water jet within the gate chamber. To incorporate the geometric variation introduced by the retaining wall into the physical collision mechanism, this embodiment introduces a linear correction factor [1+κ×(h / P)], where κ is a pre-calibrated correction coefficient for the drop height of the retaining wall.

[0097] Multiply the initial collision parameters by the linear correction factor and the collision comprehensive coefficient c. _λ The water tongue collision parameter formula with wall correction is constructed as follows:

[0098] λ * =c _λ ×[(H _rel ) 0.5 / (W _o / P)]×[1+κ×(h / P)];

[0099] Where, λ * To account for the influence of the retaining wall height in the corrected water jet collision parameters, the collision comprehensive coefficient, after the retaining wall is introduced, absorbs physical effects such as the increase in water jet drop height during offline calibration. When h is 0 in the parameter dataset, the formula automatically degenerates and eliminates the correction term to ensure compatibility with the basic body shape.

[0100] Step 203: Input the water tongue collision parameters into the pre-constructed attenuation function for calculation to obtain the dimensionless local discharge intensity at the leading edge of the sidewall.

[0101] To overcome the drawback of infinitely large local discharge intensity, this embodiment uses a pre-defined S-shaped structure instead of the negative power structure. The physically verified S-shaped function can simulate the three-stage flow characteristics: gradual attenuation under low head, accelerated attenuation after collision, and convergence under high head. In this embodiment, the water tongue collision parameters are input as independent variables into the mathematical logic for evaluation.

[0102] In some alternative embodiments, if complex floating-point exponentiation is difficult to perform due to system computing power limitations, piecewise linear lookup table curves corresponding to different collision parameters can be generated offline in advance, and the output result of the decay function can be simulated by linear interpolation method.

[0103] In some embodiments, the water tongue collision parameters are input into a pre-constructed attenuation function for calculation, specifically including:

[0104] Step 2031: Obtain the inlet / outlet gate chamber width ratio, inverted length ratio, and relative sidewall length from the dimensionless geometric parameter set.

[0105] Besides being affected by collisions, the discharge intensity of the sidewalls is also influenced by fine-tuning of the geometric scale of the planar and cross-sectional dimensions. Therefore, the system obtains the inlet and outlet gate chamber width ratio W in parallel. _i / W _o Inverted length ratio B _o / B _i And the B / P parameter, representing the relative sidewall length, is used for subsequent geometric multi-dimensional correction and compensation. This acquisition operation is achieved by reading the data frame corresponding to the address of the parsed data array in system memory.

[0106] Step 2032: Calculations are performed using the S-shaped attenuation structure formula with physical upper bounds. The specific S-shaped attenuation structure formula is as follows:

[0107] γ _s =c _s0 ×[1 / (1+λ ns )]×(W _i / W _o ) p1×(B _o / B _i ) p2 ×(B / P) p3 ;

[0108] Where, γ _s Let λ be the dimensionless local discharge intensity at the leading edge of the sidewall, and λ be the collision parameter of the water tongue. When the target key weir does not contain a retaining wall structure, λ is calculated by multiplying the initial collision parameter by the collision comprehensive coefficient c. _λ When the target key weir includes a retaining wall structure, λ is taken as the λ calculated above. * Value, W _i / W _o B is the ratio of the width of the inlet and outlet gate chambers. _o / B _i The ratio of the inverted lengths is B / P, where B / P is the relative sidewall length; c _s0 ns, p1, p2 and p3 are all pre-constructed constant coefficients.

[0109] Accordingly, the above multiplication and exponentiation compound operation is performed. In this structure, the term [1 / (1+λ]... ns The calculation provides a physical bound: when the relative head approaches 0, causing λ to approach 0, the denominator approaches 1, and the overall attenuation term approaches the maximum theoretical limit of 1, making the calculated dimensionless local discharge intensity γ... _s Locked in c _s0 Nearby, that is, supplemented by a geometric ratio multiplication term. The parameter ns is used to control the steepness of the S-curve near the water flow collision threshold. The larger the value of ns, the more violent the sudden drop in capacity caused by the flow regime change.

[0110] Example 3

[0111] When calculating the inlet and outlet discharge intensity, a simple empirical power function can be used for fitting, for example, setting the inlet discharge intensity to be proportional to H. _rel βi βi is greater than zero. However, physical flow observations revealed that the actual discharge efficiency at the inlet and outlet does not follow a monotonically increasing law with changes in head, but rather exhibits a non-monotonic peak phenomenon of first rising and then falling. That is, the conventional positive power function cannot mathematically express this unimodal phenomenon.

[0112] This embodiment provides a computational scheme based on the coupling of two opposing physical flow mechanisms to calculate the dimensionless local discharge intensity at the inlet and outlet leading edges. One possible implementation includes the following steps:

[0113] Step 301: Extract the relative head and corresponding geometric ratio parameters from the dimensionless geometric parameter set. The corresponding geometric ratio parameters include the inlet and outlet gate chamber width ratio and the inverted length ratio.

[0114] Read the subset required to perform the operations of this module from the pre-generated data structure. For the inlet or outlet leading edge, extract the relative head value calculated from the current total head over the weir, and extract the inlet / outlet gate chamber width ratio W in parallel. _i / W _o The ratio of the overhang length to B _o / B _i As the corresponding geometric ratio parameter.

[0115] In processor environments involving parallel computing architectures, the relative computation of the input and the relative computation of the output can be distributed to different computing cores for execution. Each core executes the extraction instruction independently, thereby reducing the overall model derivation time.

[0116] Step 302: Based on the relative head, construct a competitive response mechanism term that characterizes the head-driven enhancement effect and the gate chamber filling inhibition effect in the inlet and outlet flow regimes.

[0117] The competitive response mechanism term is the core mathematical carrier mapping the non-monotonic flow characteristics at the inlet and outlet. In this embodiment, the system constructs this term using two mutually constraining operators. The first operator characterizes the head-driven enhancement effect, meaning that as the head initially increases, the overflow front area increases and the flow velocity rises, resulting in a positive enhancement of the discharge capacity. The second operator characterizes the gate chamber filling inhibition effect, meaning that when the head continues to increase to a certain limit, the water level in the gate chamber rises, causing backflow and flooding, and this flooding effect begins to suppress the discharge efficiency.

[0118] Specifically, a positive power function based on head is used to map the enhancement operator, and a negative exponential function based on head is used to map the suppression operator, and the two are coupled multiplicatively. This composite structure is mathematically equivalent to the density function shape of the Γ distribution. Taking the derivative of this composite function, the calculated response value will have a peak at a certain relative head, which can well fit the laws observed in physical hydraulics.

[0119] Furthermore, in order to improve the accuracy of the model within the prediction range, the model constants need to be subject to preset physical extreme value constraints.

[0120] Step 303: Obtain a pre-defined asymptotic lower bound term to characterize the residual discharge capacity of a broad-crested weir under extreme inundation conditions.

[0121] If only the constructed Γ-like function is used, when faced with extremely high water head inputs, i.e., when the relative water head approaches infinity, the decay rate of the negative exponential function will exceed the growth rate of the power function. This will cause the local discharge intensity at the inlet and outlet of the model output to unreasonably approach zero, which does not conform to basic physical laws. Even if the key weir is completely submerged, the physical weir body at the top of the inlet and outlet still exists, and it should at least have a residual drainage capacity equivalent to a broad-crested weir at the same elevation.

[0122] Based on this, this embodiment retrieves a pre-calibrated non-zero constant parameter and uses it as an asymptotic lower bound term. This parameter serves as a physical lower bound fallback, preventing the prediction calculation model from experiencing numerical reversal or collapse when faced with inputs that exceed the limit.

[0123] Step 304: Combine the competitive response mechanism term and the asymptotic lower bound term, and multiply them by the correction coefficient based on the corresponding geometric ratio parameters to obtain the dimensionless local discharge intensity at the inlet leading edge and the dimensionless local discharge intensity at the outlet leading edge, respectively.

[0124] Specifically, this embodiment uses the following formula for a Γ-type hybrid structure with an asymptotic lower bound for calculation:

[0125] γ _x =c _x0 ×[(1-η _x )×(H _rel ) ax ×exp(-b _x ×H _rel )+η _x ]×M _geo,x ;

[0126] Where x represents the inlet leading edge or the outlet leading edge; γ _x H represents the dimensionless local discharge intensity at the inlet leading edge or the dimensionless local discharge intensity at the outlet leading edge; _rel Relative head; (H) _rel ) ax ×exp(-b _x ×H _rel ) represents the competition response mechanism item; η _x The asymptotic lower bound scaling factor is set in advance, and η is set. _x The value range of M is (0,1), representing the asymptotic lower bound term; _geo,x It is constructed based on the power product of the ratio of the width of the inlet / outlet gate chamber and the ratio of the overhang length; c _x0 ax and b _x All are pre-constructed constant coefficients, and exp(•) is the natural exponential function.

[0127] Accordingly, the asymptotic lower bound term η _xA weighted summation is performed with the competition mechanism term, which reflects the changing characteristics, to construct a hybrid distributed logic. When the input head is relatively small, the power function term outputs zero, and the hybrid formula outputs the basis constant. When the input head is high, the exponential term causes the competition mechanism term to approach zero, and the hybrid formula also outputs the basis constant. Within the normal operating head range, the intermediate variable term occupies an absolute computational weight, realistically presenting the discharge efficiency curve that first rises and then falls. Next, the calculated main trunk characteristic value is multiplied by the geometric correction coefficient composed of the power of the width ratio and the power of the overhang ratio to complete the anisotropy correction and output the characteristic value of the specified leading edge.

[0128] Specifically, the mathematical structure of the geometric correction coefficient is as follows:

[0129] M _geo,x =(W _i / W _o ) r1 ×(B _o / B _i ) r2 ;

[0130] r1 and r2 are pre-constructed constant coefficients that reflect the second-order correction effects of the inlet / outlet gate width ratio and the overhang length ratio on the leading-edge discharge efficiency, respectively. The constant coefficients are calibrated independently for the inlet and outlet leading edges.

[0131] Step 305: When it is determined that the target key weir includes a retaining wall, the formula for the Γ-type hybrid structure with asymptotic lower bound also includes a retaining wall reinforcement correction factor, and the corresponding calculation formula is as follows:

[0132] γ _x * =c _x0 ×[(1-η _x )×(H _rel * ) ax ×exp(-b _x ×H _rel * )+η _x ]×[1+δ _x ×(h / P)]×M _geo,x ;

[0133] Where, γ _x * is the dimensionless local discharge intensity at the inlet leading edge or the dimensionless local discharge intensity at the outlet leading edge when a retaining wall is present; h / P is the relative retaining wall height extracted from the dimensionless geometric parameter set; [1+δ _x [×(h / P)] is the wall reinforcement correction factor, δ _x H is the pre-built retaining wall correction factor. _rel *The equivalent relative head, taking into account the height of the retaining wall, is equal to the ratio of the total head H on the weir to the sum of the weir height P and the retaining wall height h.

[0134] Considering that the retaining walls located on both sides of the overflow front edge will restrict the lateral diffusion of water flow, thereby increasing the effective longitudinal discharge unit width flow rate of the inlet and outlet gate chambers, this embodiment adds branch decision execution logic. When a retaining wall height value greater than zero is detected in the input data structure, the basic calculation logic of step 304 is terminated, and the execution switches to an alternative calculation path containing enhanced parameters.

[0135] Specifically, a positive linear amplification operator [1+δ] is introduced. _x ×(h / P)] is used as the product factor. Since the retaining wall structure raises the overall overflow reference level, this embodiment updates the equivalent relative head variable H through algebraic substitution. _rel * The value of is set to equal H / (P+h). Under this structure, as the input relative to the wall height increases, the product factor proportionally amplifies the local discharge intensity at the inlet and outlet. The change in the definition of the equivalent head shifts the peak efficiency to an axial position with a higher absolute head. Through this parameter closed-loop coupling, the model in this embodiment can predict the flow distribution of derived engineering shapes without rebuilding the architecture, demonstrating good generalization ability.

[0136] Example 4

[0137] Based on Example 1, this example further elaborates on the calculation process of ratio normalization and the allocation of flow through closed-loop mathematical equations.

[0138] In one possible implementation, the ratio of all products is normalized to obtain the distribution coefficients for each overflow leading edge, specifically calculated using the following normalization formula:

[0139] α _k =(L _k ×γ _k ) / Σ(L _j ×γ _j );

[0140] Where k represents any overflow leading edge among the sidewall leading edge, inlet leading edge, or outlet leading edge, and j represents the summation subscript for traversing the sidewall leading edge, inlet leading edge, and outlet leading edge; α _k L is the distribution coefficient for the leading edge k; _k and L _j Let γ be the geometric lengths of leading edge k and leading edge j, respectively; _k and γ _j denoted as dimensionless local discharge intensities at leading edges k and j, respectively.

[0141] Accordingly, the three pure number features containing the relative capabilities of each overflow component, which have been calculated and stored, are read, and their corresponding geometric lateral unfolding lengths are extracted. In this mathematical structure, L _k ×γ _k This represents the absolute magnitude of the weighted efficiency that the overflow front can contribute, Σ(L _j ×γ _j This constitutes the benchmark for the overall cohesion capacity of the key weir under the current physical structural state.

[0142] In calculating the allocation coefficient α _k At that time, the weighted efficiency of the current part is divided by the aggregation capability benchmark. Therefore, for any input parameter conditions, the above three output variables mathematically satisfy the boundary condition that the summation is always equal to 1.

[0143] Through the closed-loop design of this algorithm, this embodiment effectively integrates the independent sidewall attenuation models and inlet / outlet Γ models. It can also eliminate errors such as predicted overflow and missing flow caused by residual deviations in existing independent space allocation ratio curve fitting methods.

[0144] To objectively present the response characteristics and theoretical consistency of the algorithm under various stringent hydraulic boundary conditions, it is necessary to verify the rationality of three extreme conditions, thereby ensuring that the predicted values ​​under any input conditions satisfy the fundamental laws of physics. Based on this, this embodiment analyzes the asymptotic behavior of the leading-edge operators of each submodule to describe the robustness of the algorithm.

[0145] As an example, let's take the first extreme condition corresponding to a lower head state. When the input relative head H... _rel As the value approaches zero, the collision parameters in the sidewall prediction module approach zero, causing the denominator of its S-shaped attenuation function operator to approach 1. This results in the calculated local strength value of the sidewall approaching the upper limit preset constant of the function. Because the positive power factor of the head in the Γ-shaped function of the inlet and outlet modules causes the product to become zero, the dimensionless local discharge intensity at the inlet and outlet degrades and is equivalent to a preset weak basis constant. After summing and ratio operations on this parameter in the denominator, the sidewall effectiveness value in the denominator becomes dominant, causing the output sidewall allocation coefficient α to... _s This approach approaches its maximum possible proportion while suppressing import and export proportions to a low level. This reflects the objective fact that in real hydraulic environments under low head conditions, physical discharge is dominated by side-flow paths.

[0146] As an example, let's take the second intermediate normal operating condition corresponding to the active design flow area. When the head variable is in the intermediate scale range, the sidewall input collision variable is in the nonlinear steep drop section of the function, and the discharge shows an accelerated decay trend. The input data in the inlet and outlet models are in the extreme value range where the positive and negative exponential terms tug at each other, and the output inlet and outlet operators reach their peak values. The three model parameters, after calculation, tug at each other and jointly constrain each other, so that each overflow front edge divides the remaining flow according to the dynamic competition ratio, maintaining the dynamic balance of flow distribution.

[0147] As an example, let's take the third extreme condition, corresponding to a completely submerged flood control state, as an example. When faced with a large flow rate leading to a dimensionless head H... _rel As the value approaches infinity, existing single-function regression algorithms are prone to producing anomalous calculation results due to the combined constraints of the negative exponent and the collision amplification factor. For example, the sidewall proportion may reverse and tend towards 100%. To avoid these anomalous calculation results, in this embodiment, the higher-order terms in the denominator of the sidewall prediction module tend to infinity, causing the system output of the local intensity term for that location to return to 0. At the same time, the exponential terms of the inlet and outlet modules completely decay. However, because the system has a built-in constant proportionality parameter η used to characterize the asymptotic lower bound in the preceding steps, the system... _i η _o The process of further normalization to zero was truncated, retaining the final static numerical term. Next, the normalization division operation was performed. Since the sidewall product term was 0, the numerator was 0, and its final output distribution ratio α... _s The result is 0. Therefore, the summation in the denominator only retains the sum of the products of the inlet and outlet due to the asymptotic lower bound terms. The inlet and outlet allocation coefficients are calculated by amortizing the final residual total flow under this operating condition based solely on their own asymptotic efficiency product ratio. This mechanism ensures that the algorithm, even when faced with illegal boundary data, still possesses a safe, reliable, and physically responsive truncation output function.

[0148] In this embodiment, a multi-sub-model joint prediction system is established through ratio normalization, ensuring that the sum of the three leading-edge allocation coefficients is always equal to 1 under any input conditions; and that the variation trend of the discharge intensity at each leading edge satisfies the qualitative laws of existing publicly available physical model experiments under extreme head conditions. Those skilled in the art will understand that the specific prediction accuracy of the model may vary depending on the geometry of the target key weir, the range of hydraulic conditions, and the size and quality of the calibration dataset. Before practical engineering application, it is recommended to use physical model experimental data corresponding to the target project or verified CFD numerical simulation results to specifically calibrate the model parameters and verify the prediction accuracy.

[0149] Example 5

[0150] This embodiment illustrates how to achieve a closed-loop prediction of the total discharge flow at the physical level by calling an externally pre-built empirical model of the key weir. It also describes a mathematical synthesis evaluation and output in the form of root mean square (RMS) to address the system prediction errors caused by the coupling of different sub-models. In one possible implementation, the total discharge flow of the target key weir is calculated based on its geometric and hydraulic parameters, specifically including the following steps:

[0151] Step 401: Input the geometric and hydraulic parameters of the key weir into the pre-constructed empirical formula or polynomial regression model of the key weir total discharge coefficient to obtain the key weir total discharge coefficient.

[0152] Determining the total discharge coefficient is a prerequisite for predicting the overall flow capacity. In this embodiment, global data containing features such as the broad crest widening state of the target key weir, stored in memory, is extracted and substituted into a total flow prediction algorithm interface compatible with this weir type. The method framework of this embodiment is designed with compatibility and openness, and does not force a pre-defined empirical calculation method for the total discharge coefficient. As long as the external interface can output a pure numerical total discharge coefficient reflecting the current geometric and head conditions, the calculation requirements are met.

[0153] In some specific implementations, this embodiment sets up two parallel mathematical structure forms for switching between engineering tasks of different sizes.

[0154] As an example, for a key weir with a low to medium head and meeting the preset size ratio conditions, the system can select the first structural form. This form is expressed as a dimensionless multiplication formula based on the sharp-flange weir, specifically:

[0155] C _d =C _d_sharp ×[a1×(L / W) a2 ×(H _rel ) a3 ×(W _i / W _o ) a4 ];

[0156] Among them, C _d C is the total discharge coefficient of the key weir. _d_sharp The discharge coefficient of the sharp-edge weir under the corresponding head is calculated in advance. This coefficient is obtained by looking up the discharge coefficient of the sharp-edge weir in advance based on the total head above the weir at the current time. L / W is the width ratio, and W is the height ratio. _i / W _o The width ratio of the inlet and outlet gate chambers is given, and a1 to a4 are the first set of preset empirical constants. By calling the constant set parameter block pre-extracted from relevant water conservancy literature, the total flow characteristic coefficient can be quickly solved on low-computing-power hardware using this multiplication polynomial.

[0157] In some alternative implementations, when the target key weir involves complex overhangs, varying bottom slopes, or even retaining structures, the second structural form is used. This structural form typically involves a function formula obtained through multivariate nonlinear regression training based on extensive physical model test results, specifically:

[0158] C _d =f(H _rel L / W, W _i / W _o B _o / B _i ,h / P);

[0159] Among them, B _o / B _i , where h / P is the relative height of the retaining wall, and f() is a multivariate mapping function prefitted by multivariate nonlinear regression based on physical model test data. This function can be implemented using a nested polynomial algorithm network constructed based on complex multivariate response surfaces, and can be received as a comprehensive dimension input to output a high-precision total discharge coefficient of the key weir.

[0160] Step 402: Combine the general weir flow theory formula with the total discharge coefficient of the key weir to calculate the total discharge flow of the target key weir.

[0161] After obtaining the comprehensive dimensionless total discharge coefficient with corrective significance, a conversion process with physical dimension correlation is performed. Specifically, the gravitational acceleration constant g is called and the fluid characteristic environment is extracted. According to the physical laws of free discharge volumetric flow rate of classical broad-crested weirs or overflow weirs, formula expansion and calculation are performed. The general weir flow theory formula is:

[0162] Q=C _d ×W×(2g) 0.5 ×H 3 / 2 ;

[0163] Where Q is the total discharge flow rate, and C _d Let C be the total discharge coefficient of the keyway weir, W be the width of the keyway weir in a single cycle, and H be the total head above the weir. In this formula, the power coefficients 0.5 and 3 / 2 are inherent exponential relationships derived from weir flow theory, which are well-known theoretical concepts in hydraulics. _d This already includes factors such as weir type correction and flow coefficient. Through this multiplication operation, the absolute flow rate passing through the target section per unit time can be calculated, that is, the total discharge flow rate of the target key weir under the current set head control boundary, thereby completing the locking of the global hydraulic parameters.

[0164] Furthermore, after outputting the partial discharge flow rates at each overflow leading edge, the method also includes a joint evaluation of the prediction accuracy based on error propagation, specifically including the following steps:

[0165] Step 403: Obtain the relative error of the distribution coefficient caused by the constructed dimensionless local discharge intensity model, and the total flow relative error of the pre-constructed model used to calculate the total discharge flow.

[0166] In this embodiment, to provide a confidence benchmark for the structural mechanics safety analysis and computational fluid dynamics-assisted verification of hydraulic structures, the system triggers a built-in post-evaluation and error assessment computational flow. This technical architecture includes two model intervals with independent sources of mathematical error propagation. The first source of error is the set of root mean square errors of model feature maps left over from the spatial local relative intensity curve fitting regression calculation and ratio normalization process; the second source of error is the set of calibration tolerance domains between the empirical structural network and the actual physical prototype measurements. By calling the deviation feature dictionary stored in each fitting training record file pre-stored in read-only memory, the system retrieves the first statistical relative error parameter caused by structural differences and the second statistical relative error parameter caused by the external network constraints of the total flow interface.

[0167] Step 404: Summate the square of the relative error of the allocation coefficient with the square of the relative error of the total flow.

[0168] In this embodiment, data computation is performed using the law of superposition of variances of independent random error variables. Considering the algorithmic independence of the two calculation processes, the actual positive and negative sign probabilities of the fluctuations are not synchronously correlated.

[0169] Accordingly, the extracted relative error values ​​of the allocation coefficients and the relative error values ​​of the total flow are multiplied by themselves. The corresponding real square values ​​of the two generated items are added together to generate a single variance sink quantitative characteristic value.

[0170] Step 405: Calculate the arithmetic square root of the summation result, and obtain and output the relative prediction error of the partial discharge flow rate at each overflow leading edge.

[0171] The system uses the floating-point arithmetic unit within the microprocessor to extract the arithmetic square root of the variance sink's quantitative characteristic value. The data generated by this error orthogonal synthesis formula is directly defined, retained, and output by the system as the relative prediction error limit for the partial discharge flow rate at the specified overflow leading edge.

[0172] Example 6

[0173] This embodiment describes an offline method for obtaining all undetermined basic parameters and correction parameters used in the network driving the local discharge intensity calculation formula. Since the sub-model combination of this invention contains multiple undetermined coefficients, using a conventional globally synchronous multivariate fitting network for optimization can easily trigger matrix inversion collapse due to parameter collinearity, as well as the risk of overfitting the preset samples.

[0174] In one possible implementation, the undetermined basic parameters and correction parameters used to calculate the dimensionless local overflow intensity are obtained through the following pre-executed offline hierarchical calibration strategy. The obtained parameters are used in the step of calculating the dimensionless local overflow intensity of each overflow front.

[0175] Step 501, First layer of basic calibration: Obtain the first sample set without retaining walls and excluding extreme high head conditions. Use the least squares deviation between the predicted and measured values ​​of the normalized allocation coefficients as the objective function to fit and determine the conventional undetermined coefficients in the strength formulas of the sidewall, inlet, and outlet. The asymptotic lower bound proportional coefficient is temporarily taken as a preset initial constant value.

[0176] In the first-level basic calibration stage, physical laws within the normal operating area are extracted. Specifically, a first sample set is extracted from offline hydraulic physics model test databases or high-fidelity fluid mechanics numerical simulation test result databases. The first sample set needs to meet the following data cleaning conditions, including removing all data frames containing geometric input records of non-zero retaining wall height; removing high-frequency overflow records where the relative head parameter exceeds the preset boundary value, and reducing the interference of edge mutation factors on the optimization direction of the backbone constant network. The preset boundary value can be determined by those skilled in the art based on the distribution characteristics of the relative head and the transition characteristics of the discharge flow state in the calibration dataset. Usually, the critical value before the relative head enters the fully submerged flow state is taken as the dividing point.

[0177] Next, a nonlinear least squares optimization solver is invoked, such as a Levenberg-Marquardt optimization algorithm network or a genetic algorithm search architecture with an adaptive mutation factor. The residual objective function is set by comparing the sequence of calculated allocation coefficients generated by the current iteration parameters with the sequence of actual spatial proportions captured in the corresponding test bench measurement records, using the minimization of the sum of the variances between the two as the optimization objective.

[0178] The solver performs iterations within the search solution space to determine and fix the series of control coefficients that control the S-shaped attenuation characteristics of the sidewall collision, as well as the series of exponents that control the distribution position of the dual competition mechanism at the inlet and outlet, and other conventional undetermined coefficients.

[0179] Furthermore, when the parameters of extreme overflow are in an inactive state, a constant small initial value parameter is assigned to the asymptotic lower bound scaling factor to prevent them from participating in the competition for computing resources in this frequency band.

[0180] Step 502, Second-layer boundary calibration: Fix the calibrated conventional undetermined coefficients, obtain a second sample set including the high head limit condition, and use the same objective function to separately calibrate the asymptotic lower bound proportional coefficients of the inlet and outlet.

[0181] After completing the basic feature extraction, the process proceeds to the second layer of boundary protection calibration. To avoid compromising the accuracy of the converged backbone fitting, the numerical constants locked in the first stage are frozen, transforming their state from adjustable variables to fixed constant parameter configurations. Accordingly, a second sample set is obtained from the original offline recording array. This involves selecting multiple test arrays whose relative head exceeds the upper limit of normal operations and whose data covers the ultimate flood control inundation boundary, thereby constructing the second sample set.

[0182] During the optimization process involving a single variable or a small number of variables, the system iteratively calls the objective function analysis module based on least squares bias. Since the remaining parameters are frozen for protection, the algorithm engine's optimization partial derivative gradient and weight update resources are used to update the asymptotic lower bound scaling coefficient η of the corresponding inlet and outlet leading edges. _i With η _o This embodiment quickly identifies the optimal convergence parameters by comparing the measured residual drainage characteristics of a broad-crested weir under extreme conditions, thereby capturing the truncation tail point characteristics of the mathematical model.

[0183] Step 503, Third layer of retaining wall calibration: Fix the parameters of the first two layers that have been calibrated, obtain a third sample set containing different retaining wall heights, and perform separate regression fitting of the retaining wall correction coefficients used to couple the retaining wall effect.

[0184] In the third and final retaining wall extension control calibration task, the system inherits and solidifies all numerical variables generated in the first two stages. It calls upon the extended physical prototype test feature library corresponding to the multi-tiered non-zero retaining wall geometric layout, loads the data, and generates the third sample set. This quantifies the influence rate of a single structural variation variable on the flow mechanism output.

[0185] In this embodiment, the system maintains the overall computational architecture and the form of the loss assessment equation, only opening the numerical space of preset parameters such as the retaining wall correction coefficient κ for the equivalent drop height and the retaining wall correction coefficient δ for controlling the enhanced dimension of the retaining wall flow regime. Based on this variable stripping regression method, after superimposing physical constraints, the slope change of the corresponding single parameter is solved, completing the optimization process for all parameters. This isolated execution pipeline mechanism can improve the optimization iteration efficiency of the computing system, allowing each set of empirical parameters to be bound to the corresponding fluid dynamics mechanism segment, thus improving the generalization robustness when loading and using online modules.

[0186] Based on the above embodiments, this application also provides a key weir discharge spatial allocation prediction system. The system includes a processor coupled to a memory. The memory stores instructions, which, when executed by the processor, implement the methods performed in the above method embodiments.

[0187] This application also provides an electronic device, including: one or more processors and a memory, wherein the one or more processors are coupled to the memory, the memory is used to store computer program code, the computer program code including computer instructions, and when the one or more processors execute the computer instructions, the electronic device executes the methods executed in the above-described method embodiments.

[0188] This application also provides a chip, including: a processor and a communication interface, wherein the processor calls a computer program or instruction through the communication interface to execute the methods executed in the above-described method embodiments.

[0189] This application also provides a computer program product, which, when run on a computer, executes the methods described in the above-described method embodiments.

[0190] This application also provides a computer-readable storage medium storing a computer program. When the computer program is run on a computer, the computer executes the methods described in the above-described method embodiments.

[0191] The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific details of the above embodiments. Within the scope of the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all fall within the protection scope of the present invention.

Claims

1. A method for predicting the spatial distribution of discharge flow at a key-weir, characterized in that, include: Obtain the geometric and hydraulic parameters of the target key weir and convert them into a dimensionless set of geometric parameters; Based on the dimensionless geometric parameter set, the dimensionless local discharge intensity of each overflow front edge of the target key weir is calculated. Each overflow front edge includes the sidewall front edge, the inlet front edge, and the outlet front edge. Based on geometric parameters, the geometric length of each overflow leading edge is extracted, the dimensionless local discharge intensity of each overflow leading edge is calculated as a product of the corresponding geometric length, and the ratio normalization operation is performed on all products to obtain the distribution coefficient of each overflow leading edge. Obtain a pre-built model for calculating the total discharge flow rate, and calculate the total discharge flow rate of the target key weir based on geometric and hydraulic parameters; Calculate the product of the total discharge flow rate and the distribution coefficient of each overflow front, and output the partial discharge flow rate of each overflow front.

2. The method according to claim 1, characterized in that, Calculate the dimensionless local discharge intensity at the leading edge of the sidewall, specifically including: Obtain the relative head and relative outlet gate width from the dimensionless geometric parameter set; Based on the relative head and the relative outlet gate chamber width, calculate the water tongue collision parameters that characterize the ratio of the horizontal throw distance to the gate chamber width when the overflow water tongues from the sidewalls move towards each other in the gate chamber. The water tongue collision parameters are input into a pre-constructed attenuation function for calculation to obtain the dimensionless local discharge intensity at the leading edge of the sidewall.

3. The method according to claim 2, characterized in that, The water tongue collision parameters are calculated as follows: The initial collision parameters are obtained by calculating the ratio of the arithmetic square root of the relative head to the relative width of the outlet gate chamber. Obtain the relative retaining wall height from the dimensionless set of geometric parameters; By using a linear correction factor that includes the relative height of the retaining wall, the influence of the drop height in the initial collision parameters is coupled and corrected to obtain the water tongue collision parameters.

4. The method according to claim 1, characterized in that, Calculate the dimensionless local discharge intensity at the inlet leading edge and outlet leading edge, respectively, including: Extract the relative head and the corresponding geometric ratio parameters from the dimensionless geometric parameter set; Based on relative head, a competitive response mechanism term is constructed to characterize the head-driven enhancement effect and the gate chamber filling inhibition effect in the inlet and outlet flow regimes; Obtain the asymptotic lower bound term for characterizing the residual discharge capacity of a broad-crested weir under extreme inundation conditions; The dimensionless local discharge intensity at the inlet leading edge and outlet leading edge is obtained by combining the competitive response mechanism term with the asymptotic lower bound term and multiplying it by a correction coefficient based on the corresponding geometric ratio parameter.

5. The method according to claim 1, characterized in that, The total discharge capacity of the target key weir is calculated, specifically including: By inputting the geometric and hydraulic parameters into the pre-constructed empirical formula or polynomial regression model of the total discharge coefficient of the key weir, the total discharge coefficient of the key weir can be obtained. By combining the general weir flow theory formula with the total discharge coefficient of the key weir, the total discharge flow of the target key weir is calculated.

6. The method according to claim 1, characterized in that, The fundamental and corrected parameters for calculating the dimensionless local discharge intensity are obtained through the following hierarchical calibration strategy: Obtain the first sample set without retaining walls and excluding extreme high head conditions. Use the least squares deviation of the predicted and measured values ​​of the normalized distribution coefficients as the objective function to fit the conventional undetermined coefficients of the sidewall and inlet / outlet strength formulas. The asymptotic lower bound proportional coefficient is temporarily taken as a preset initial constant value. Fix this set of coefficients, obtain a second sample set containing extreme high head conditions, and calibrate the asymptotic lower bound ratio coefficients of the inlet and outlet; With the parameters of the first two layers fixed, a third sample set containing multiple retaining wall heights is obtained, and retaining wall correction coefficients for coupling retaining wall effects are fitted separately through regression.

7. The method according to claim 2, characterized in that, The horizontal throw distance is derived by equating the sidewall overflow flow pattern to projectile motion, and its derivation is based on the following physical mapping conditions: The total flow velocity over the top of the target key weir sidewall is defined to include a normal velocity component. The magnitude of the normal velocity component is driven by the local head and is characterized by a normal velocity scaling factor. Extract the geometric tilt characteristics along the sidewall and calculate the average drop height coefficient representing the drop from the top of the sidewall to the bottom plate of the outlet gate chamber; The calculation structure of the water tongue collision parameters is a dimensionless form of the product of the normal velocity component and the fall time.

8. The method according to claim 1, characterized in that, After outputting the partial discharge flow rates at each overflow leading edge, the process also includes a step of jointly evaluating the prediction accuracy through error propagation. Obtain the relative error of the allocation coefficient caused by the pre-stored dimensionless local discharge intensity model, and the pre-stored total flow relative error of the pre-built model used to calculate the total discharge flow rate; The square of the relative error of the allocation coefficient is summed with the square of the relative error of the total flow. Calculate the arithmetic square root of the summation result to obtain and output the relative prediction error of the discharge flow rate of each overflow leading edge.

9. A spatial distribution prediction system for the discharge of a key-weir, characterized in that, The system includes a processor coupled to a memory storing instructions which, when executed by the processor, implement the method as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed on a computer, implement the method as described in any one of claims 1 to 8.

Images