A representative monitoring data-driven water transfer project regulating gate over-flow formula partition construction method

Abstract

The application discloses a representative monitoring data-driven water diversion project check gate over-flow formula partition construction method, which comprises representative monitoring data set construction, over-flow state division, partition over-flow formula description, flow state transition interval segmentation point optimization model, each flow state partition over-flow formula coefficient calibration, partition over-flow formula result model verification and evaluation, check gate over-flow dynamic prediction and the like. Advantages are that different flow states are partitioned and fitted, and corresponding function expressions are constructed to improve over-flow simulation precision. The method provides a reference for water delivery scheduling work condition pre-play and scientific scheduling plan making, and further provides technical support for open channel water diversion project water supply safety.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of accurate calibration of the hydraulic parameters of a regulating gate in a open channel water diversion system, and in particular to a representative monitoring data driven method for constructing a flow formula partition of a regulating gate in a water diversion project. BACKGROUND

[0002] In the water delivery scheduling system of a large open channel water diversion project, the regulating gate, as a key flow regulating facility, its core hydraulic parameter, the gate flow formula, has important theoretical value and practical guiding significance for the optimization design of the project structure, the decision support of operation scheduling, the development of channel hydraulic control strategy, and the study of dynamic characteristics of the water delivery system. At the same time, the accurate calculation of the flow through the gate is also the basis for realizing the scientific scheduling and smooth operation of the water diversion system. At present, scholars at home and abroad have formed two basic modes for the study of the flow formula of the regulating gate: one is to construct an empirical formula based on the energy conservation equation (such as the classic hydraulic model of Tsinghua University formula, Wushui formula, etc.), and the other is to use the dimensional analysis method and its derivative technology (including data assimilation driven dimensional analysis optimization model, multi-objective dimensional analysis framework embedded with intelligent algorithm).

[0003] Research shows that although the traditional empirical formula system is widely used in engineering, it is subject to the strong empirical dependence of the submergence coefficient and the flow coefficient, and the nonlinear response characteristics of the water level-flow caused by the complex flow pattern in front of the gate, resulting in significant multiple solutions and uncertainties in parameter identification. Compared with the traditional empirical formula system, the dimensional analysis method effectively improves the numerical accuracy of flow simulation by mathematically representing the coupling mechanism of hydraulic elements and combining data-driven parameter identification methods. However, it is worth noting that the global fitting strategy of this method is sensitive to data distribution, and the model bias propagation effect caused by abnormal monitoring data can cause flow prediction errors. More importantly, the existing researches mostly use the dimensional analysis method under the dominant orifice flow state, and there is no clear characteristic discrimination standard for the flow state adaptability of the weir flow area and the transition area of the orifice weir flow. The lack of flow state representation leads to the fact that the predicted value of the flow through the gate is easy to exceed the allowable error threshold of the project. SUMMARY

[0004] The purpose of the present application is to provide a representative monitoring data driven method for constructing a flow formula partition of a regulating gate in a water diversion project, thereby solving the aforementioned problems in the prior art.

[0005] In order to achieve the above-mentioned purpose, the technical scheme adopted by the present application is as follows:

[0006] A representative monitoring data driven method for constructing a flow formula partition of a regulating gate in a water diversion project, comprising the following steps,

[0007] S1, gate flow state division and zoned flow formula description: according to the different flow state partition of orifice flow and weir flow, the corresponding zoned flow formula under the dimensional analysis method is determined;

[0008] S2, flow state transition interval segmentation point optimization model, and zoned flow formula coefficient calibration: according to the weir flow gate flow formula under the dimensional analysis method, the actual gate flow coefficient is inversely calculated through the flow monitoring data, and then the function expression of the gate flow coefficient in the weir flow gate flow formula under the dimensional analysis method is obtained. The NSGA-II algorithm is used to obtain the transition zone segmentation points of orifice flow and weir flow with the minimum root mean square error as the objective function, and the important parameters in the dimensional analysis method under the optimal transition partition and the gate flow formula are determined.

[0009] S3, zoned flow formula result model verification and evaluation: the gate flow formula verified by the optimal transition partition calibration is used to evaluate the model prediction flow accuracy with evaluation indexes.

[0010] Preferably, step S1 specifically includes the following contents,

[0011] S11, orifice flow interval: through dimensional analysis method, the gate flow relationship formula under the orifice flow submerged outflow state is derived,

[0012] q=f(e,g,H E ,β) (1)

[0013] H E =H0-H2 (2)

[0014] Wherein, q is the single-width flow; e is the gate opening; g is the gravity acceleration; H E is the water level difference before and after the gate; β is the absolute viscosity coefficient; H0 is the upstream water depth of the gate; H2 is the downstream water depth of the gate; if the flow state is free outflow, H2=0;

[0015] Suppose the gate flow has the following form,

[0016]

[0017] Wherein, α, δ, η, ω and z are constant coefficients:

[0018] Then formula (3) can be converted to,

[0019] L 2 T -1 =L α ×(LT -2 ) δ ×L η ×(ML -1 T -1 ) ω (4)

[0020]

[0021]

[0022] Simplify,

[0023]

[0024] wherein, Taking logarithm on both sides of the equation, the linear relationship between a and b is obtained as

[0025]

[0026] Let a = j, b = lgi, and the linear equation is simplified as

[0027] y = ax + b (9)

[0028] The gate flow formula based on dimensional analysis method for orifice flow submergence outflow state is derived as

[0029]

[0030] wherein, Q 孔 is the interval flow through the gate; B is the water surface width.

[0031] S12, for the weir flow interval: single-width flow function relationship is

[0032]

[0033] wherein, k, r, s, t are constant coefficients;

[0034] Then formula (11) can be transformed into

[0035] L 2 T -1 = (LT -2 ) r ×L s × (ML -1 T -1 ) t (12)

[0036]

[0037]

[0038] Through equation transformation, the gate flow formula based on dimensional analysis method for arc gate is derived as​​

[0039]

[0040] wherein, Q 堰 is the discharge through the gate in the weir flow section; λ is the discharge coefficient through the gate in the weir flow state.

[0041] Preferably, step S2 specifically comprises,

[0042] S21, based on the discharge formula of weir flow under dimensional analysis method, combined with the representative monitoring data set, the discharge coefficient expression is fitted by introducing the gate parameter;

[0043] S22, taking the root mean square error of the discharge formula in each subsection as the objective function, a subsection point optimization model of the flow relationship of the check gate is constructed, the constraint conditions and decision variables are set for the subsection point optimization model, and the NSGA-II algorithm is used for autonomous optimization to obtain the optimal transition interval corresponding to each check gate;

[0044] S23, based on the flow sequence of the corresponding orifice flow section and the weir flow section, the flow of the optimal transition interval is calculated.

[0045] Preferably, S21 specifically is, according to the discharge formula of weir flow under dimensional analysis method, the actual discharge coefficient λ is inversely calculated through the flow monitoring data,

[0046]

[0047] Based on the representative monitoring data set, according to λ~e / (H0-H2), a quadratic function curve fitting is performed, the gate parameter is introduced, and the discharge coefficient expression is fitted;

[0048] λ=l[e / (H0-H2)] 2 +m[e / (H0-H2)]+n (18)

[0049] wherein, l, m, n are three gate parameters introduced.

[0050] Preferably, S22 specifically comprises the following contents,

[0051] S221, taking the root mean square error of the discharge formula in each subsection as the objective function, a subsection point optimization model of the flow relationship of the check gate is constructed;

[0052]

[0053] MIN f(Q) =RMSE(Q obs ,Q cal ) min (20)

[0054] wherein Q obs is the monitored data of the passage flow; Q cal is the calculated data of the passage flow; RMSE Q is the root mean square error of the calculated value and the monitored value of the passage flow; N0 is the total number of data;

[0055] S222, set constraints for the section point optimization model, including the constraint of the number of sections and the maximum / minimum constraint of the relative opening section interval,

[0056] section min = 3 (21) 0.00 < u < v < 1.00 (22)

[0057] wherein section is the number of sections; u, v are the relative opening section points;

[0058] S223, the section point optimization model takes the relative opening section points u, v as the decision variables, and iteratively calculates and autonomously optimizes the two section points in the interval of the relative opening of the check gate being 0 and the maximum opening being 1 by combining the objective function with the NSGA-II algorithm, and the section points are respectively set as the right closed interval and the left closed interval,

[0059] Y section = (0.00, u] + (u, v) + [v, 1.00) (23)

[0060] wherein Y section is the relative opening interval of each section;

[0061] S224, the flow equation of the arc gate is divided into three sections of orifice flow, weir flow, and transition from orifice flow to weir flow; when the relative opening Y ≤ the upper limit opening Ya of orifice flow, it is orifice flow; when the relative opening Y ≥ the lower limit opening Yb of weir flow, it is weir flow; when Ya ≤ Y ≤ Yb, it is the transition from orifice flow to weir flow; the average of the flow under orifice flow and weir flow is used as the transition section flow;

[0062] Q = (Q1 + Q2) × 0.5 (24)

[0063] wherein Q1 is the flow calculated by the orifice flow formula in the transition interval data sequence, Q2 is the flow calculated by the weir flow formula in the transition interval data sequence; Q is the transition section flow;

[0064] In the initial interval preset around the relative opening criterion, the corresponding transition interval is calculated for each candidate group, and the transition interval is smoothed to obtain the average transition interval flow simulation value. The transition interval flow simulation value is compared with the measured flow data to calculate the root mean square error between them, which measures the pros and cons of the simulation effect. Through global search and optimization of all possible combinations in the initial interval, the transition interval combination that minimizes the root mean square error is determined, which is the optimal transition interval of the current check dam; finally, the optimal transition interval corresponding to each check dam, the optimal transition interval flow sequence in the interval and the minimum root mean square error value between the measured data are output.

[0065]

[0066] Wherein, N is the number of time in the transition interval time step; is the i1th time point; RMSE(tra) is the root mean square error of the transition interval flow; Q sim and Q obs are the simulation value and the monitoring value of the transition interval flow, respectively.

[0067] Preferably, S23 specifically comprises taking the average of the orifice flow interval and the weir flow interval flow sequence as the flow value of the optimal transition interval.

[0068]

[0069] Wherein, Q 孔 is the orifice flow interval flow; Q 堰 is the weir flow interval flow; Q 过 is the transition interval flow.

[0070] Preferably, step S3 specifically comprises using the optimal transition interval flow formula verified by the rating to predict the flow of the non-rating year check dam monitoring data, and using the root mean square error, the mean absolute error and the Nash coefficient as evaluation indexes to evaluate the prediction accuracy of the model.

[0071] Preferably, step S1 further comprises,

[0072] S0, representative monitoring data set acquisition: based on the data requirements of model simulation, the long sequence historical water regime and working condition monitoring data are preprocessed to obtain the representative monitoring data set.

[0073] Preferably, step S0 specifically comprises,

[0074] S01, eliminate obvious error data such as null, abnormal value and missing value, and preliminarily exclude interference;

[0075] S02, based on the following criteria, the relative steady-state data are screened: on the one hand, the dynamic change of the flow under the joint action of the two gates and the canal pool and the water outlet of the research section is determined, and on the other hand, whether the adjacent regulation gate of the research object is determined; the steady-state screening of whether the monitoring data of the regulation gate is stable is considered by the above two factors.

[0076] S03, the screened data are taken as representative monitoring data set, which provides data basis for subsequent partition relationship description.

[0077] Preferably, step S3 further comprises,

[0078] S4, regulation gate flow dynamic prediction: based on the evaluation verified flow formula of each partition, the flow through the gate is dynamically simulated according to the input basic water regime and working condition data of the project, and the flow through the gate of the regulation gate is dynamically predicted.

[0079] The beneficial effects of the present application are: 1, the method of the present application is based on long sequence of historical water regime data, the flow pattern is divided into three partitions, the fitting flow formula of different partitions is calibrated and verified, and the purpose of improving the flow simulation accuracy is achieved, so as to realize the accurate perception of water regime state of water transfer system. 2, the method of the present application determines whether the dynamic change of flow under the joint action of the two gates and the canal pool and the water outlet of the research section is changed, and on the other hand, whether the adjacent regulation gate of the research object is determined, and the steady-state screening of whether the monitoring data of the regulation gate is stable is considered. This method avoids the interference of abnormal fluctuation of monitoring data on the accurate construction of flow formula. BRIEF DESCRIPTION OF DRAWINGS

[0080] Figure 1 is the logical sequence diagram of the partition construction method in the embodiment of the present application;

[0081] Figure 2 is the comparison diagram of the prediction effect of the turbulent river regulation gate partition in the embodiment of the present application;

[0082] Figure 3 is the comparison diagram of the overall prediction effect of the turbulent river regulation gate in the embodiment of the present application. DETAILED DESCRIPTION

[0083] In order to make the purpose, technical scheme and advantages of the present application more clear, the present application is further described in detail below with reference to the drawings. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

[0084] Embodiment one

[0085] In this embodiment, for the problem of regulating gate passing relationship description of water diversion project, a representative monitoring data driven over-flow formula partition construction method is provided to better solve the deficiencies and defects existing in the prior art and improve the prediction accuracy of passing gate flow. The method filters steady-state data by considering the dynamic flow of two-gate-one-channel and water outlet, and whether the adjacent gate is in action, as the basis for constructing the representative monitoring data set. At the same time, different flow states are partitioned and fitted to construct the corresponding function expression to improve the simulation accuracy of passing gate flow. For water delivery scheduling simulation, it provides reference for scientific scheduling plan, and provides technical support for the safety of open channel water diversion project. Figure 1 As shown in the following,

[0086] I. Representative monitoring data set construction

[0087] Based on the data requirements of model simulation, the historical long sequence of water regime measured data is preprocessed, mainly including: eliminating obvious error data such as null, abnormal value and missing value, and preliminarily excluding interference. Then, the relatively steady-state data is screened based on the following standards: on the one hand, the dynamic change of the flow under the joint action of the two-gate-one-channel and the water outlet is determined, and on the other hand, whether the adjacent regulating gate of the research object is in action is identified, and the steady-state screening of whether the monitoring data of the regulating gate is stable is considered under the above double factors. The screened data is used as the representative monitoring data set, which provides data basis for subsequent partition relationship description.

[0088] II. Flow state division and partition over-flow formula description

[0089] The representative monitoring data set is divided according to the flow state, i.e. orifice flow interval, transition interval and weir flow interval. According to the dimensional analysis method, the dimensionless form of the passing gate flow formula (i.e. dimensionless function expression) in each interval is derived. Specifically, the dimensionless function expression of the passing gate flow formula in the orifice flow and weir flow intervals is derived.

[0090] 2.1. Orifice flow interval

[0091] The orifice flow relationship formula under the submerged outflow state is derived by the dimensional analysis method:

[0092] q=f(e,g,H E ,β) (1)

[0093] H E =H0-H2 (2)

[0094] Where q is the unit discharge, unit m 2 / s; e is the gate opening, unit m; g is the acceleration of gravity, unit m / s 2 ; H Eis the water level difference in front of and behind the gate, with unit of m; β is the absolute viscosity coefficient; H0 is the water depth upstream of the gate, with unit of m; H2 is the water depth downstream of the gate, with unit of m. If the flow state is free outflow, then H2 = 0.

[0095] Assume that the flow through the gate has the following form, where α, δ, η, ω and z are constant coefficients:

[0096]

[0097] Through dimensional analysis, it can be transformed as:

[0098] L 2 T -1 = L α × (LT -2 ) δ × L η × (ML -1 T -1 ) ω (4)

[0099]

[0100]

[0101] Simplifying, we get:

[0102]

[0103] wherein,

[0104] Taking the logarithm of both sides of the equation, it can be transformed as a linear relationship between and and is:

[0105]

[0106] Let a = j and b = lgi, and simplify it into a linear equation:

[0107] y = ax + b (9)

[0108] The formula for the flow through the gate in the submerged outflow state of the orifice flow is derived based on the dimensional analysis method as:

[0109]

[0110] wherein, Q 孔 is the flow through the gate, with unit of m 3 / s; B is the water section width, with unit of m.

[0111] 2.2, the weir flow range

[0112] Single-width flow function relationship:

[0113]

[0114] wherein k is a constant coefficient, r, s, t are constant coefficients, which can be converted through dimension analysis:

[0115] L 2 T -1 =(LT -2 ) r ×L s ×(ML -1 T -1 ) t (12)

[0116]

[0117]

[0118] Through equation transformation, according to the form of the conventional gate flow formula, the arc gate flow formula based on the dimension analysis method is derived as:

[0119]

[0120] wherein Q 堰 is the flow through the gate, the unit is m 3 / s; λ is the flow coefficient.

[0121] Three, the flow state transition interval segmentation point optimization model, each flow state partition flow formula coefficient rating

[0122] Based on the screened weir flow zone steady-state data and the flow formula coefficient λ, the combination of λ~e / H, λ~e / (H0-H2) and the quadratic function is fitted, and the flow formula coefficient λ function expression is obtained. After determining the weir flow zone coefficient λ, the basic engineering data is collected and arranged. Through the NSGA-II optimization algorithm, the minimum root mean square error is taken as the optimization objective, and two transition points of the transition zone, i.e. the orifice weir flow partition, are obtained, and each flow state partition is determined. The transition points divided by the relative opening e / H (wherein e is the gate opening, and H is the upstream water depth) are u, v (wherein 0

[0123] Orifice flow interval: 0

[0124] Transition interval: u

[0125] Weir flow interval: v

[0126] Through the dimension analysis method of orifice and weir flow zone, the average value of the calculation results of orifice and weir flow in the transition zone is taken, and the parameters of each partition flow formula through the gate are rated.

[0127] 3.1, Determine the weir flow formula function expression

[0128] According to the weir flow formula under the dimensional analysis method, the actual discharge coefficient λ can be inverted from the flow monitoring data, and the formula is as follows:

[0129] Weir flow-submerged outflow condition:

[0130]

[0131] Based on the screened relative steady-state data, λ~e / H, λ~e / (H0-H2) are sequentially fitted with quadratic function curve, three gate parameters l, m, n are introduced, and the discharge relationship expression is fitted, here taking the upstream and downstream water head difference H0-H2 of the check gate as an example, taking e / (H0-H2) as the variable, and the specific formula is as follows:

[0132] λ=l[e / (H0-H2)] 2 +m[e / (H0-H2)]+n (18)

[0133] Where λ is the discharge coefficient under the weir flow state; e is the gate opening, unit is m; l, m, n are gate parameters; H0, H2 are the water levels before and after the check gate, respectively, unit is m.

[0134] 3.2, Coupling NSGA-Ⅱ algorithm for autonomous partitioning

[0135] (1) Objective function

[0136] Taking the root mean square error (RMSE) of the gate flow relationship fitting expression in each segment as the objective function, a segmented point optimization model of the check gate flow relationship is constructed, and the objective function form is as follows:

[0137]

[0138] MIN f(Q) =RMSE(Q obs ,Q cal ) min (20)

[0139] Where Q obs is the discharge monitoring data, unit is m 3 / s, Q cal is the discharge calculation data, unit is m 3 / s, RMSE Q is the root mean square error of the discharge calculation value and the monitoring value, MIN f(Q) is the fitting result under the condition of minimum root mean square error of each segment.

[0140] (2) Constraint conditions

[0141] The constraint conditions of the piecewise fitting model include the constraint of the number of sections and the constraint of the maximum / minimum value of the relative opening section interval, which are expressed as follows:

[0142] section min = 3 (21) 0.00 < u < v < 1.00 (22)

[0143] wherein, section is the number of sections; u and v are the relative opening section points.

[0144] (3) Decision variables

[0145] The decision variables of the model are the relative opening section points u and v. The NSGA-II algorithm is combined with the objective function to iteratively calculate and autonomously optimize the two section points in the interval of the relative opening of the regulation gate being 0 and the maximum opening being 1. The right closed interval and the left closed interval are respectively set at the section points, which are expressed as follows:

[0146] Y section = (0.00, u] + (u, v) + [v, 1.00) (23)

[0147] wherein, Y section is the relative opening section interval of each section, and u and v have the same meaning as above.

[0148] (4) Optimization of flow state transition interval

[0149] The flow equation of the radial gate can be divided into three sections of orifice flow, weir flow and transition from orifice flow to weir flow. Let Ya and Yb represent the upper limit opening of orifice flow and the lower limit opening of weir flow, respectively, that is, the relative opening Y ≤ Ya is orifice flow, Y ≥ Yb is weir flow, and Ya ≤ Y ≤ Yb is the transition from orifice flow to weir flow. After the upper limit opening of orifice flow and the lower limit opening of weir flow are calculated, the flow calculation formula of the transition section uses the average value of the flow calculated by the orifice flow and weir flow methods:

[0150] Q = (Q1 + Q2) × 0.5 (24)

[0151] wherein, Q1 is the flow calculated by the orifice flow formula in the data sequence of the transition section, Q2 is the flow calculated by the weir flow formula in the data sequence of the transition section, and Q is the flow of the transition section.

[0152] The artificial direct setting of the flow regime transition interval can cause a large error with the actual interval, and therefore, the data sequence of the gate flow coefficient λ is calculated by using the orifice flow formula and the weir flow formula respectively. The relationship between λ and e / h is fitted by taking λ as the dependent variable and e / h as the independent variable, and the approximate range of the transition of the orifice flow and the weir flow is qualitatively analyzed through the distribution of each data point. The appropriate initial value of the preset flow regime transition interval is set, so that the process of searching for the optimal solution by the optimization algorithm is avoided due to the divergence phenomenon, and the rationality of the result is avoided. In the application, the model aims to determine the actual transition interval range of the regulating gate and improve the simulation precision of the flow through the gate, and the method of minimizing the root mean square error is used to optimize the transition interval.

[0153] Specifically, first, the possible transition zone flow sequence is extracted in the initial interval preset around the previous relative opening degree judgment standard, such as the floating range [0.55, 0.75]. For each candidate transition interval, the corresponding flow through the gate sequence is calculated, and the sequence is smoothed to obtain the average transition interval flow simulation value. Then, the simulation value is compared with the measured flow data, and the RMSE value is calculated to measure the pros and cons of the simulation effect. Through global search and optimization of all possible combinations of the entire initial interval, the transition interval combination that minimizes the RMSE is determined, and the combination is the optimal transition interval range of the current regulating gate. Finally, the optimal transition interval corresponding to each regulating gate is output, the optimal flow through the gate sequence in the interval, and the minimum RMSE value between the measured data.

[0154]

[0155] Wherein, n represents the number of time points in the transition interval time step, t i is the i-th time point, RMSE(tra) is the root mean square error of the transition interval sequence, Q sim and Q obs are the simulation value and the monitoring value of the transition interval flow, respectively. The final result of the transition interval is output in the form of a table, including the optimal transition interval range of each regulating gate and the error evaluation index of the model fitting.

[0156] 3.3, transition interval division and flow calculation

[0157] The average of the flow sequence in the orifice flow interval and the weir flow interval is taken as the value of the transition zone flow.

[0158]

[0159] Wherein, Q 孔 is the flow through the gate in the orifice flow interval, the unit is m 3 / s; Q 堰 is the flow through the gate in the weir flow interval, the unit is m 3 / s; Q过 The flow rate through the gate in the transition section is expressed in cubic meters per second (m³). 3 / s.

[0160] IV. Verification and Evaluation of the Regional Overcurrent Formula Model

[0161] The flow rate formulas for each zone under the optimal transition interval are derived through calibration. The Nash efficiency coefficient, mean absolute error, and root mean square error are used as evaluation indicators to evaluate the accuracy of the model's flow rate prediction.

[0162] (1) Nash efficiency coefficient (NSE)

[0163]

[0164] Among them, Q obs This is the flow rate monitoring data for the gate, in meters (m³). 3 / s;Q cal The data for calculating the flow rate through the gate is in cubic meters (m³). 3 / s; NSE is the Nash coefficient between the calculated and monitored values ​​of the gate flow rate.

[0165] (2) Mean Absolute Error (MAE)

[0166]

[0167] Where N0 is the total number of samples; Q obs Q cal The meaning is the same as above. MAE is the average absolute error between the calculated value and the monitored value of the gate flow rate.

[0168] (3) Root Mean Square Error (RMSE)

[0169]

[0170] Where N0 is the total number of samples; Q obs Q cal The meaning is the same as above. RMSE is the root mean square error between the calculated value and the monitored value of the gate flow rate.

[0171] V. Dynamic Prediction of Throughflow at Control Gates

[0172] Finally, the formula for the flow rate through the gate is obtained. The flow rate through the gate is dynamically simulated based on the basic hydrological and engineering data of the input project. At the same time, the flow rate formula calibrated in segments further improves the accuracy of the flow rate prediction and provides a reference for scheduling and operation.

[0173] Example 2

[0174] In this embodiment, taking the Tonghe regulation gate of a certain middle line project as an example, the relative steady-state data automatic identification is carried out according to the method described in the application, the flow state transition interval segmentation point optimization model is constructed and calculated, the over-flow formula coefficient is calibrated, the partition function expression is described, and the improvement of the flow prediction accuracy is realized.

[0175] The middle line project has a total length of 1432 km, and there are various types of water passing structures along the line, wherein there are 61 regulation gates as key regulation structures. In actual engineering operation, the sensing equipment for water level and flow usually has certain monitoring errors, and even abnormal or wrong data such as missing data and outlier data may occur. In addition, due to objective environment, seasonal changes, regulation structure operation and other disturbance factors, the hydraulic parameters of various structures have dynamic characteristics in space-time changes, which subtly affect the water conveying state of the project. These disturbance factors increase the difficulty of gate flow relationship description to some extent, and put forward higher requirements for high-precision flow prediction. Therefore, the application uses representative monitoring data, takes the upstream Tonghe regulation gate of the middle line project as an example, and realizes dynamic prediction of gate flow and improves calculation accuracy by segmenting the gate flow formula.

[0176] I. Construction of representative monitoring data set

[0177] The long sequence monitoring data is preprocessed: the obviously wrong data such as null value, invalid value and missing value is eliminated, and the measured data is screened for relative steady state. On the one hand, whether the flow dynamic change under the joint action of two gates and a channel pool and a water distribution port changes is judged; on the other hand, whether the adjacent regulation gate of the research object is in action is identified, and the data after screening is screened for stability under the double factors, which is used as the representative monitoring data set.

[0178] II. Flow state division and partition flow formula description

[0179] According to the different flow state partition of orifice flow and weir flow, the dimensionless function relationship formula derived by the dimension analysis method of gate flow is determined.

[0180] III. Flow state transition interval segmentation point optimization model and calibration of flow formula coefficient of each flow state partition

[0181] According to the weir flow formula under the dimension analysis method, the actual gate flow coefficient λ can be inversely calculated through the flow monitoring data, and then the function expression of the gate flow coefficient in the weir flow dimensionless formula is calibrated. The NSGA-II multi-objective genetic algorithm takes the minimum root mean square error as the optimization target, and independently optimizes the segmentation points of the orifice flow and weir flow transition partition, and finally determines the data sequence length of each partition and the important parameters in the flow formula.

[0182] IV. Verification and evaluation of partition flow formula result model

[0183] Based on the steady-state data and the partition results, the orifice and weir flow interval is based on the dimensional analysis method, the over-flow formula is segmented and fitted, and the transition interval takes the average of the results of the two.

[0184] V. Dynamic prediction of the flow through the regulation gate

[0185] After verification, the final determination of the description of the segmented over-flow formula is compared with the overall fitting results of the dimensional analysis method, and the calculation results are as shown in Figure 1 、 Figure 2 as shown, and the index calculation results are shown in Table 1.

[0186] After introducing the orifice and weir flow transition interval, the flow prediction accuracy is obviously improved. Compared with the root mean square error (RMSE), the mean absolute error (MAE) and the Nash efficiency coefficient (NSE) of the fitting flow simulation value and the measured value of the overall dimensional analysis method of the turbulent river regulation gate, the three indicators of the segmented fitting flow simulation value and the measured value of the regulation gate, the results are shown in Table 1.

[0187] Table 1: Fitting index results of the turbulent river regulation gate

[0188]

[0189] Therefore, it can be seen from the results that the method of the present application can realize the simulation accuracy of the representative monitoring data driven regulation gate flow formula of the water transfer project. The method introduces the concept of transition interval, and segments the long sequence of historical monitoring data to fit the flow relationship. The method passes the NSGA-II genetic algorithm to independently optimize the transition segment point position, and more accurately reflects the gate flow relationship. Mainly realizes the high-precision prediction of the gate flow process under the support of the regulation gate special model and the driving of the representative data, and provides necessary support for the engineering operation. The method has the characteristics of simplicity, convenience and universality, and basically meets the simulation accuracy demand of the flow prediction of the flow relationship.

[0190] By adopting the above technical scheme of the present application, the following beneficial effects are obtained:

[0191] The application provides a representative monitoring data-driven regulation gate over-flow formula partition construction method. The application method is based on long sequence historical water regime data, discriminates flow state division into three partitions, calibrates and verifies fitting over-flow formulas of different partitions, and aims to improve flow simulation precision, so as to realize accurate perception of water regime state of the water transfer system. The application method on one hand studies whether the flow dynamic change under the joint action of the two gates and the canal pool and the water outlet of the research section changes, and on the other hand identifies whether the adjacent regulation gate of the research object is in action, and considers whether the monitoring data of the regulation gate is stable under the double factors. This method avoids the interference of abnormal fluctuation of the monitoring data on the accurate construction of the over-flow formula.

[0192] The above only describes the preferred embodiments of the application, and it should be noted that those skilled in the art can make several improvements and refinements without departing from the principles of the application, and these improvements and refinements should be considered as the protection scope of the application.

Claims

1. A representative monitoring data-driven water transfer project regulation gate over-flow formula partitioning construction method, characterized in that: Comprising the following steps, S1, over gate flow state division and partition over flow formula description: according to the different flow state partition of orifice flow and weir flow, the over gate flow formula under the dimensional analysis method of the corresponding partition is determined; S2, flow state transition interval segmentation point optimization model, each flow state partition over flow formula coefficient calibration: according to the over gate flow formula under the dimensional analysis method, the actual over gate flow coefficient is inverted through the flow monitoring data, and then the function expression of the over gate flow coefficient in the over gate flow formula under the dimensional analysis method is obtained by calibration; The NSGA-II algorithm is used to independently optimize the transition interval segmentation points of orifice flow and weir flow with the minimum root mean square error as the objective function, so as to determine the important parameters in the over gate flow formula under the dimensional analysis method; Step S2 specifically comprises, S21, based on the over gate flow formula under the dimensional analysis method, the over gate flow coefficient expression is fitted by introducing the gate parameter in combination with the representative monitoring data set; S22, the root mean square error of the over gate flow formula in each segmented calculation flow is taken as the objective function, a segmentation point optimization model of the over flow relationship of the check dam is constructed, the constraint conditions and decision variables are set for the segmentation point optimization model, and the NSGA-II algorithm is used for independent optimization to obtain the optimal transition interval corresponding to each check dam; S23, based on the flow sequence of the corresponding orifice flow interval and weir flow interval, the flow of the optimal transition interval is calculated; The constraint conditions for the segmentation point optimization model include the segmentation number constraint and the maximum / minimum value constraint of the relative opening segmentation interval; Segmented point optimization model with relative opening degree segmented point 、 As a decision variable, the NSGA-II algorithm combined with the objective function is used to iteratively calculate and autonomously optimize two segmented points in the interval of the relative opening degree of the regulating gate being 0 and the maximum opening degree 1. The segmented points are set as the right closed interval and the left closed interval, respectively. S3, verification and evaluation of the result model of the partition over flow formula: the over gate flow formula after the calibration verification of the optimal transition partition is used to evaluate the prediction flow accuracy of the model by the evaluation index; Step S3 specifically is that the over gate flow formula after the calibration verification of the optimal transition partition is used to predict the flow of the non-calibration year check dam monitoring data, and the root mean square error, the average absolute error and the Nash coefficient are taken as the evaluation indexes to evaluate the prediction flow accuracy of the model; Before step S1, it further comprises, S0, acquisition of representative monitoring data set: based on the data demand of model simulation, the long sequence historical water regime and working condition monitoring data are preprocessed to obtain the representative monitoring data set.

2. The representative monitoring data-driven regulation sluice over-flow formula partitioning construction method according to claim 1, characterized in that: Step S1 specifically comprises the following contents, S11, orifice flow interval: the gate over flow relationship formula under the submerged outflow state of orifice flow is derived through dimensional analysis method, （1） （2） wherein, is the single-width flow rate; is the gate opening; is the gravitational acceleration; is the water level difference before and after the gate; is the absolute viscosity coefficient; is the upstream water depth of the gate; is the downstream water depth of the gate; if the flow regime is free outflow, then ; It is assumed that the over gate flow has the following form, （3） wherein , , , and are constant coefficients: Then formula (3) can be transformed into, （6） After simplification, （7） wherein , , ; Taking logarithm on both sides of the equation, the conversion is obtained A linear relationship between is （8） Let , , , , simplifying to linear equations is, （9） The gate over flow formula under the submerged outflow state of orifice flow based on dimensional analysis method is derived as, （10） wherein, is the lockage flow for the orifice flow section; is the width of the flow section. S12, for weir flow interval: the single width flow function relationship formula is, （11） wherein , , , are constant coefficients; Then formula (11) can be transformed into, （14） Through equation transformation, according to the form of conventional gate over flow formula, the over flow formula of arc gate based on dimensional analysis method is derived as, （15） （16） wherein, is the discharge through the lock for the weir flow regime; is the discharge coefficient for the weir flow regime.

3. The representative monitoring data-driven regulation sluice over-flow formula partitioning construction method according to claim 2, characterized in that: S21 is specifically, according to the dimensional analysis under the weir flow through the gate flow formula, through the flow monitoring data to retrieve the actual through the gate flow coefficient , （17） Based on the representative monitoring dataset, according to A quadratic function curve fitting is performed, a gate parameter is introduced, and a fitted over-gate flow coefficient expression is obtained. （18） wherein , , are three gate parameters introduced.

4. The representative monitoring data-driven water transfer project regulation gate over-flow formula partitioning construction method according to claim 3, characterized in that: S22 specifically comprises the following contents, S221, taking the root mean square error of the over gate flow formula in each segmented calculation flow as the objective function, a segmentation point optimization model of the over flow relationship of the check dam is constructed; （19） （20） wherein, is the monitored data of the passage flow rate; is the calculated data of the passage flow rate; is the root mean square error of the calculated value and the monitored value of the passage flow rate; is the total number of data; S222, the constraint conditions for the segmentation point optimization model include the segmentation number constraint and the maximum / minimum value constraint of the relative opening segmentation interval, （21） （22） wherein is the number of segments; , is the relative opening segment point; S223, segment point optimization model with relative opening degree segment point 、 As a decision variable, the NSGA-II algorithm is combined with the objective function to iteratively calculate and autonomously optimize two segment points in the interval of the relative opening degree of the regulation gate being 0 and the maximum opening degree 1. The segment points are set to the right closed interval and the left closed interval, respectively. （23） wherein is the relative opening interval for each segment; S224, the overfall equation of the arc gate is divided into three sections of orifice flow, weir flow, and transition from orifice flow to weir flow; when the relative opening Y is less than or equal to the upper limit opening Ya of orifice flow, the flow is orifice flow; when the relative opening Y is greater than or equal to the lower limit opening Yb of weir flow, the flow is weir flow; when Ya < Y < Yb, the flow is the transition from orifice flow to weir flow; the average of the flow under orifice flow and weir flow is used as the flow in the transition section; （24） wherein, Qtrans is the flow rate calculated from the orifice flow equation in the transition interval data sequence, Qweir is the flow rate calculated from the weir flow equation in the transition interval data sequence; Qtrans is the flow rate calculated from the orifice flow equation in the transition interval data sequence, In the initial interval preset around the relative opening judgment standard, the corresponding overfall flow sequence of each candidate transition interval is calculated, and the overfall flow sequence is smoothed to obtain the average transition interval flow simulation value; the transition interval flow simulation value is compared with the measured flow data, the root mean square error between them is calculated, and the pros and cons of the simulation effect are measured; through global search and optimization of all possible combinations of the entire initial interval, the transition interval combination that minimizes the root mean square error is determined, that is, the optimal transition interval range of the current check gate; finally, the optimal transition interval corresponding to each check gate, the optimal overfall flow sequence in the interval, and the minimum root mean square error value between the measured data are output; （25） wherein, is the number of time steps in the transition interval; is the time step in the transition interval; is the time step in the transition interval; is the root mean square error of the flow rate in the transition interval; and are the simulated and monitored values of the flow rate in the transition interval, respectively.

5. The representative monitoring data-driven regulation sluice over-flow formula partitioning construction method according to claim 4, characterized in that: S23 is specifically to take the average of the flow sequence of the corresponding orifice flow interval and weir flow interval as the flow value of the optimal transition interval; （26） wherein, is the orifice flow interval flow rate; is the weir flow interval flow rate; is the transition interval flow rate.

6. The representative monitoring data-driven regulation sluice over-flow formula partitioning construction method according to claim 5, characterized in that: Step S0 specifically includes, S01, eliminate obviously wrong data such as null, abnormal value and missing value, and preliminarily exclude interference; S02, based on the following standards, the relative stable data is screened: on the one hand, the flow dynamic change under the joint action of the two gates and the canal pool of the research section and the water distribution port is judged, and on the other hand, whether the adjacent check gate of the research object is actuated is identified; the stability of the check gate monitoring data is screened by considering the above two factors; S03, the screened data is taken as the representative monitoring data set, which provides a data basis for subsequent partition relationship description.

7. The representative monitoring data-driven regulation sluice over-flow formula partitioning construction method according to claim 1, characterized in that: Step S3 further includes, S4, check gate overfall flow dynamic prediction: based on the evaluation verified overfall flow formula of each partition, the overfall flow condition is dynamically simulated according to the input basic water regime and working condition data of the project, and the overfall flow of the check gate is dynamically predicted.

Images