Adaptive scaling method and system for design flood process based on local weighted regression
Adaptively scaling the flood process through local weighted regression method, solving the problem of sudden change and discontinuity of flood process line, realizing the consistency between flood peak flow and flood volume in different periods and design standards, and is suitable for flood processes under different types of control indicators. Zoom.
Patent Information
- Application Number
- CN202410218984.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-02-28
- Publication Date
- 2024-12-13
- Estimated Expiration
- 2044-02-28
AI Technical Summary
The existing design flood process scaling method results in sudden changes and discontinuities in the flood process line when the target flood peak volume is different from the typical flood process peak volume. The results rely on manual correction, which is subjective and has a large workload. It is difficult to keep the peak flow and flooding at different times consistent with the design standards after scaling.
Adaptive scaling method for designing flood process based on local weighted regression is adopted. By obtaining the flood process time series and target peak eigenvalues, identifying characteristic parameters, calculating the scaling ratio, and using local weighted regression method to smooth the flood process, iterate Cycling until preset conditions are met to ensure that the scaled flood peak flow and flooding at different times are consistent with the design standards.
It effectively solved the problem of sudden changes in the flood process, realized the flood process was scaled according to the design standards, maintained the original typical flood distribution model, reduced the subjectivity and workload of manual adjustments, and improved the accuracy and consistency of the scaling results.
Smart Images
Figure CN117993207B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of flood process simulation, and in particular to a method and system for adaptively scaling a designed flood process based on local weighted regression. Background Art
[0002] The results of the design flood process are an important basis for flood control planning, flood control project design, and flood control scheduling scheme formulation. At present, the scaling methods of the design flood process line mainly include the same-ratio scaling method and the same-frequency scaling method. Among them, the same-ratio scaling method uses the flood peak of the flood process or the flood volume in a specified period as the control, and scales the entire flood process with the same scaling factor. The scaled flood peak volume is difficult to be completely consistent with the design value. The same-frequency scaling method uses the flood peak of the flood process and the flood volume in one or more periods as the control to scale the typical flood process. Compared with the same-ratio scaling method, the advantage of the same-frequency scaling method is that the characteristic values of the scaled process line, such as the flood peak flow and the flood volume in each period, can be consistent with the design value, and it has been more widely used in planning, design and other fields.
[0003] However, when the target flood peak volume is significantly different from the peak volume distribution of the typical flood process, the same frequency method will be used for amplification, and the amplification ratio of each period will be different. There will be a sudden change on the flood process line at the junction of the two periods, making the process discontinuous and the process line zigzag, which is quite different from the measured flood. Usually, it is smoothed manually, and the final result needs to be gradually tested and repeatedly adjusted. Therefore, the workload is large, the subjectivity is strong, the results are arbitrary, and the original typical flood pattern is easily destroyed. Summary of the invention
[0004] In order to overcome the defects and shortcomings of the prior art, the present invention provides a method and system for adaptively scaling a designed flood process based on local weighted regression. The present invention solves the problem of sudden changes in the flood process caused by direct scaling of traditional peak volumes, realizes the scaling of typical flood processes according to design standards, and ensures that the scaled flood peak flow and the total flood volume in different time periods are consistent with the design standards, thereby better maintaining the original typical flood distribution pattern.
[0005] In order to achieve the above object, the present invention adopts the following technical solutions:
[0006] The present invention provides a method for adaptively scaling a design flood process based on local weighted regression, comprising the following steps:
[0007] Obtain input parameters, including flood process time series and target peak volume characteristic values of scaled flood;
[0008] Identify the characteristic parameters of the input flood process;
[0009] According to the characteristic parameters of the identified flood process and the target peak characteristic value of the scaled flood, the scaling ratio of the flood peak and the flood volume in different periods is calculated, and the flood process is scaled according to the peak-volume relationship;
[0010] The local weighted regression method is used to smooth the scaled flood process. The regression weight of the flow rate at each time point in the corresponding span is calculated, and the second-order polynomial is used for weighted linear least squares regression. The center points of the regression curve at each moment are connected to obtain the smoothed flood process.
[0011] Identify the characteristic parameters of the smoothed flood process, and calculate the scaling factor of the smoothed flood process and then scale it;
[0012] Iterate the smoothing of the flood process, identify the characteristic parameters of the smoothed flood process, and scale the smoothed flood process until the preset stop iteration condition is met, and output the final scaled flood process line result.
[0013] As a preferred technical solution, the characteristic parameters of the flood process inputted for identification include peak flow, flood volume, peak occurrence time, and flood volume statistical period;
[0014] While counting the flood peak flow value, obtain the flood peak occurrence time;
[0015] The calculation method of flood peak flow is expressed as:
[0016] Q m =max(Q(t))
[0017] Among them, Q m is the peak flow of the flood process, Q(t) represents the flow value at each moment in the flood process;
[0018] Identify the time t s The total amount of flood with the initial statistical length T:
[0019]
[0020] t e =t s +T
[0021] Among them, W T,ts Indicates the time t s is the total amount of flood with the initial statistical length T, t s Indicates the starting time point of the statistical period, t e Indicates the end time of the statistical period;
[0022] Identify the maximum flood volume of the flood process with a statistical length of T:
[0023] WT =max(W T,ts )
[0024] Among them, W T It represents the maximum total flood volume of the flood process with a statistical length of T;
[0025] At the same time, get W T The corresponding t s and t e , [t s , t e ] is W T statistical period.
[0026] As a preferred technical solution, the scaling ratio of the flood peak and the flood volume in different periods is calculated based on the identified characteristic parameters of the flood process and the target peak volume characteristic value of the scaled flood, specifically including:
[0027]
[0028]
[0029] ...
[0031]
[0032] in, represents the peak flow target value, Characteristic parameters representing peak flow, It represents the target value of the total flood volume in the first statistical period T1. It indicates the target value of the total flood volume in the next statistical period T2. Indicates the nth statistical duration T n The target value of the total flood volume, It represents the characteristic parameter of the total flood volume in the first statistical time period T1. It represents the characteristic parameter of the total flood volume in the next statistical time period T2. Indicates the nth statistical duration T n Characteristic parameter of total flood volume, K m Indicates the peak flow scaling ratio, Indicates the scaling ratio of traffic in each time period.
[0033] As a preferred technical solution, the scaling ratio of the flood peak and the flood volume in different periods is calculated based on the identified characteristic parameters of the flood process and the target peak volume characteristic value of the scaled flood, specifically including:
[0034] Calculate the flood peak and the scaling ratio of flood volume in different periods, expressed as:
[0035]
[0036]
[0037]
[0038]
[0039]
[0040]
[0041] in, They represent the characteristic parameters corresponding to the peak flow, the maximum 3-day flood volume, the maximum 5-day flood volume, the maximum 7-day flood volume, the maximum 15-day flood volume, and the maximum 30-day flood volume, respectively. They represent the target values corresponding to the peak flow, the maximum 3-day flood volume, the maximum 5-day flood volume, the maximum 7-day flood volume, the maximum 15-day flood volume, and the maximum 30-day flood volume, respectively. K m represents the peak flow scaling factor, K 3d , K 5d , K 7d , K 16d , K 30d They respectively represent the scaling ratio of the traffic in each time period.
[0042] As a preferred technical solution, the typical flood process is scaled according to the peak-volume relationship. The specific scaling method is expressed as follows:
[0043]
[0044]
[0045]
[0046]
[0047]
[0048]
[0049] Among them, Q′ m is the scaled peak flow; Q′ t3-tm , Q′ t5-t3 , Q′ t7-t5 , Q′ t15-t7 , Q′ t30-t15They are the flow rates at each time in the flood statistical period of three days after the flood peak period is removed after scaling, the flow rates at each time in the flood statistical period of five days after the three-day flow statistical period is removed, the flow rates at each time in the flood statistical period of seven days after the five-day flow statistical period is removed, the flow rates at each time in the flood statistical period of fifteen days after the seven-day flow statistical period is removed, and the flow rates at each time in the flood statistical period of thirty days after the fifteen-day flow statistical period is removed; is the peak flow of the flood process before scaling; Respectively represent the traffic in each statistical period before scaling.
[0050] As a preferred technical solution, the scaling ratio of the flood peak and the flood volume in different periods is calculated based on the characteristic parameters of the identified flood process and the target peak volume characteristic value of the scaled flood, and the typical flood process is scaled according to the peak volume relationship, specifically including:
[0051] Calculate the flood peak and the scaling ratio of flood volume in different periods, expressed as:
[0052]
[0053]
[0054]
[0055]
[0056]
[0057] in, They represent the characteristic parameters corresponding to the peak flow, the maximum 3-day flood volume, the maximum 7-day flood volume, the maximum 15-day flood volume, and the maximum 30-day flood volume, respectively. They represent the target values corresponding to the peak flow, the maximum 3-day flood volume, the maximum 7-day flood volume, the maximum 15-day flood volume, and the maximum 30-day flood volume, respectively. m represents the peak flow scaling factor, K 3d , K 7d , K 15d , K 30d They respectively represent the scaling ratio of the traffic in each time period.
[0058] As a preferred technical solution, the typical flood process is scaled according to the peak-volume relationship. The specific scaling method is expressed as follows:
[0059]
[0060]
[0061]
[0062]
[0063]
[0064] Among them, Q′ m is the scaled peak flow; Q′ t3-tm , Q′ t7-t3 , Q′ t15-t7 , Q′ t30-t15 They are the flow rates at each moment in the flood volume statistics period of three days after the flood peak period is removed after scaling, the flow rates at each moment in the flood volume statistics period of seven days after the three-day flow statistics period is removed, the flow rates at each moment in the flood volume statistics period of fifteen days after the seven-day flow statistics period is removed, and the flow rates at each moment in the flood volume statistics period of thirty days after the fifteen-day flow statistics period is removed; is the peak flow of the flood process before scaling; Respectively represent the traffic in each statistical period before scaling.
[0065] As a preferred technical solution, the local weighted regression method is used to smooth the scaled flood process, and the specific steps include:
[0066] Calculate the regression weight of the flow rate at each time point in the corresponding span at each time point, expressed as:
[0067]
[0068] Among them, w i Yes i The smoothing weight of the flow at the moment, t represents the flow value to be smoothed at that moment, t i represents the time point of the flood process, dt represents the interval along the horizontal axis from time t to the farthest time point in the span;
[0069] Using a second-order polynomial for weighted linear least squares regression, the quadratic regression at time t is expressed as:
[0070]
[0071] Among them, a, b, c represent the parameters to be determined;
[0072] The loss function of local weighted regression is expressed as:
[0073]
[0074] Among them, q i Indicates the flow rate at each time point of the flood process;
[0075] Based on the least squares method, the unknown parameters a, b, and c are solved to minimize the loss function J(θ) and obtain the fitting curve at each moment;
[0076] Connect the center points of the fitting curves at each moment to obtain the smoothed flood process.
[0077] As a preferred technical solution, the iterative cyclic flood process smoothing, the identification of characteristic parameters of the smoothed flood process, and the scaling of the smoothed flood process are stopped until it is determined that a preset stop iteration condition is met, specifically including:
[0078] When the number of iterations reaches the preset maximum number, the iteration stops;
[0079] Or by calculating the Euclidean distance between the two flood processes before and after scaling, specifically expressed as:
[0080]
[0081] Among them, d q,j represents the Euclidean distance between the j-th iteration and the j-1-th iteration, q i,j represents the flow at time i in the jth iteration result, q i,j-1 represents the flow at time i in the j-1th iteration result;
[0082] The iteration stops when it is determined that the set Euclidean distance threshold is met.
[0083] The present invention also provides a design flood process adaptive scaling system based on local weighted regression, comprising: an input parameter acquisition module, a characteristic parameter identification module, a scaling ratio calculation module, a flood process scaling module, a smoothing processing module, a post-smoothing adjustment module, a loop iteration module, a stop iteration determination module, and an output module;
[0084] The input parameter acquisition module is used to acquire input parameters, including the flood process time series and the target peak volume characteristic value of the scaled flood;
[0085] The characteristic parameter identification module is used to identify the characteristic parameters of the input flood process;
[0086] The scaling ratio calculation module is used to calculate the scaling ratios of the flood peak and the flood volume in different periods according to the characteristic parameters of the identified flood process and the target peak volume characteristic value of the scaled flood;
[0087] The flood process scaling module is used to scale the flood process according to the peak volume relationship;
[0088] The smoothing processing module is used to smooth the scaled flood process using a local weighted regression method, by calculating the regression weight of the flow rate corresponding to each time point in the corresponding span at each time point, using a second-order polynomial to perform weighted linear least squares regression, and connecting the center points of the regression curve at each moment to obtain the smoothed flood process;
[0089] The post-smoothing adjustment module is used to identify characteristic parameters of the smoothed flood process, and to scale the smoothed flood process after calculating the scaling ratio;
[0090] The loop iteration module is used for loop iteration of the smoothing process of the flood process, identification of characteristic parameters of the flood process after smoothing, and scaling of the flood process after smoothing, until the iteration is stopped when it is determined that a preset stop iteration condition is met;
[0091] The stop iteration determination module is used to determine whether the stop iteration condition is met;
[0092] The output module is used to output the final scaled flood process line result.
[0093] Compared with the prior art, the present invention has the following advantages and beneficial effects:
[0094] The present invention realizes adaptive scaling of flood processes based on the design of local weighted regression, solves the problem of sudden changes in flood processes caused by traditional direct scaling of peak volumes, realizes scaling of typical flood processes according to design standards, and ensures that the scaled flood peak flow and the total flood volume in different time periods are consistent with the target control indicators, and better maintains the distribution pattern of the original typical flood. The present invention is better suitable for scaling of flood processes of different magnitudes under different types of control indicator combinations, and can meet the needs of designing flood processes in planning and design work. BRIEF DESCRIPTION OF THE DRAWINGS
[0095] Figure 1 It is a flow chart of the adaptive scaling method for designing flood process based on local weighted regression of the present invention;
[0096] Figure 2 To directly scale the flood process diagram based on the flood peak volume relationship;
[0097] Figure 3 This is a schematic diagram of the local weighted regression smoothing flood process;
[0098] Figure 4 This is a schematic diagram of the flood scaling and smoothing process of the present invention;
[0099] Figure 5 It is a schematic diagram of the Euclidean distance change curve of the flood scaling smoothing process of the present invention;
[0100] Figure 6 It is a schematic diagram of the scaling results of the flood process designed with different frequencies according to the present invention;
[0101] Figure 7 This is a schematic diagram of the flood process scaling results under different peak volume control conditions of the present invention. DETAILED DESCRIPTION
[0102] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.
[0103] Example 1
[0104] like Figure 1 As shown, this embodiment provides a method for adaptively scaling a design flood process based on local weighted regression. The process can be smoothed by local weighted regression, and the peak quantity characteristics of the process will be changed while smoothing. This embodiment achieves smoothing and corresponding peak quantity relationship by iterating two processes in a loop, including the following steps:
[0105] S1: Determine input parameters. The input of this embodiment is a typical flood process time series and a target peak volume characteristic value of the scaled flood. The target peak volume characteristic value mainly includes a flood peak flow target value. and the target value of total flood volume for a specified duration The superscript o represents the target value, and the subscript T represents the statistical duration. For example, the maximum three-day flood volume target value is recorded as Maximum 7-day flood volume target value Maximum 15-day flood volume target value wait;
[0106] S2: Identify the characteristic parameters of the input typical flood process, including peak flow, flood volume, peak occurrence time, flood volume statistical period and other characteristic parameters;
[0107] In this embodiment, the peak flow is calculated as follows:
[0108] Q m =max(Q(t))
[0109] Among them, Q m is the peak flow of the flood process, Q(t) represents the flow value at each moment in the flood process, and while counting the peak flow value, the peak time t is recorded m ;
[0110] In this embodiment, the flood volume of the subsequent specified duration is counted moment by moment from the flood start time point, the maximum value is taken as the specified duration characteristic flood volume of the input flood process, and the start time point and end time point of the maximum flood volume are recorded at the same time, which is specifically expressed as:
[0111]
[0112] Among them, W T,ts Indicates the time t s is the total amount of flood with a starting statistical length of T, such as W 3d,1hIt means that the total flood volume for the next three days is counted starting from the first hour of the flood process. s Indicates the starting time point of the statistical period, t e Indicates the end time of the statistical period, t e =t s +T, get W T The corresponding t s and t e , [t s , t e ] is W T Statistical period;
[0113] In this embodiment, the maximum total flood volume of the flood process with a statistical length of T is expressed as:
[0114] W T =max(W T,ts )
[0115] Among them, W T It represents the maximum total flood volume of the flood process with a statistical length of T, such as: W3 represents the maximum total flood volume of 3 days, and the occurrence interval of the maximum flood volume is [ts T ,te T ], such as [ts 3d ,te 3d ] indicates the period of occurrence of the maximum three-day flood total.
[0116] S3: According to the characteristic parameters of the identified flood process and the characteristic values of the scaled target peak volume, the scaling ratios of the flood peak and the flood volume in different periods are calculated, which can be specifically expressed as:
[0117]
[0118]
[0119] ...
[0121]
[0122] in, represents the peak flow target value, Characteristic parameters representing peak flow, It represents the target value of the total flood volume in the first statistical period T1. It indicates the target value of the total flood volume in the next statistical period T2. Indicates the nth statistical duration T n The target value of the total flood volume, It represents the characteristic parameter of the total flood volume in the first statistical time period T1. It represents the characteristic parameter of the total flood volume in the next statistical time period T2. Indicates the nth statistical duration T n Characteristic parameter of total flood volume, K m Indicates the peak flow scaling ratio, Indicates the scaling ratio of the traffic in each time period;
[0123] In this embodiment, the peak flow rate (Q m ), the maximum 3-day total flood volume (W3), the maximum 5-day total flood volume (W5), the maximum 7-day total flood volume (W7), the maximum 15-day total flood volume (W 15 ), the maximum 30-day flood volume (W 30 )6 parameters as an example, according to the method described in S2, the characteristic parameters of the input typical flood process are identified and recorded as And according to the characteristic parameters of the identified flood process and the characteristic values of the scaled target peak volume, the scaling ratios of the flood peak and the flood volume in different periods are calculated. When calculating the scaling ratio, the influence of the scaled period needs to be deducted. The specific method for calculating the scaling ratio is as follows:
[0124]
[0125]
[0126]
[0127]
[0128]
[0129] In the formula, K m is the peak flow scaling factor, K 3d , K 5d , K 7d , K 15d , K 30d Respectively represent the scaling ratio of the flow in each time period;
[0130] Then, the typical flood process is scaled according to the peak-volume relationship. The specific scaling method is as follows:
[0131]
[0132]
[0133]
[0134]
[0135]
[0136]
[0137] Among them, Q′ m is the scaled peak flow; Q′ t3-tm , Q′ t5-t3 , Q′ t7-t5 , Q′ t15-t7 , Q′ t30-t15 They are the flow rates at each time in the flood statistical period of three days after the flood peak period is removed after scaling, the flow rates at each time in the flood statistical period of five days after the three-day flow statistical period is removed, the flow rates at each time in the flood statistical period of seven days after the five-day flow statistical period is removed, the flow rates at each time in the flood statistical period of fifteen days after the seven-day flow statistical period is removed, and the flow rates at each time in the flood statistical period of thirty days after the fifteen-day flow statistical period is removed; is the peak flow of the flood process before scaling; Respectively represent the traffic in each statistical period before scaling.
[0138] S4: Local weighted regression method is used to smooth the scaled flood process;
[0139] In this embodiment, the peak volume characteristics of the flood process after the method described in step S3 directly scales the flood process are consistent with the target value, but due to the different scaling ratios of the flow in each time period, a sudden change in flow will occur at the junction of the statistical time periods. This embodiment uses a local weighted regression method to smooth the scaled flood process, specifically by calculating the regression weight of the flow corresponding to each time point in the span corresponding to each time point, and then using a second-order polynomial to perform weighted linear least squares regression, and finally connecting the center points of the regression curves at each moment to obtain the smoothed flood process;
[0140] In this embodiment, a local weighted regression method is used to smooth the scaled flood process, which specifically includes:
[0141] First, the flood samples are weighted and then fitted, as follows: i and q i (i=1, 2, 3, ... n) are the time points of the flood process and the corresponding flow at each time point. With a time point t as the center, a segment of data with a length of frac is intercepted forward and backward. In this embodiment, the value is 5. A weighted linear regression is performed on the data using the weight function w. Note is the center value of the regression line, where is the flow rate corresponding to the curve after fitting. For all data points of the entire time series, n weighted regression lines can be drawn, and the line connecting the center points of each regression line is the local weighted regression curve of the entire data segment.
[0142] For each time point, the local regression smoothing process follows these steps:
[0143] S41: Calculate the regression weight of the flow rate corresponding to each time point in the span corresponding to each time point. The weight is calculated by the following formula:
[0144]
[0145] In the formula, w i Yes i The smoothing weight of the flow at time t is the flow value to be smoothed, t i are several time values near time t defined by the span, and dt is the interval along the horizontal axis from time t to the farthest time point in the span. The weight has the following two characteristics: on the one hand, the data point to be smoothed has the largest weight and has the greatest impact on the fit; on the other hand, the weight of the data point outside the span is zero and has no impact on the fit.
[0146] S42: Use a second-order polynomial to perform weighted linear least squares regression, and record the quadratic regression at time t as:
[0147]
[0148] In the formula, a, b, and c are unknown parameters.
[0149] According to the weights determined in step S41, the loss function of the local weighted regression is defined as:
[0150]
[0151] Then, the least square method is used to solve the optimization problem, that is, to find the unknown parameters a, b, c in θ to minimize the loss function J(θ) and obtain the fitting curve at each time;
[0152] S43: Connect the center points of the fitting curves at each moment to obtain a smoothed flood process;
[0153] S5: After the flood process is smoothed by the method of step S4, the peak characteristic statistical value of the flood process may change and cannot be consistent with the target control value. In this embodiment, the method in step S2 is used again to identify the characteristic parameters of the smoothed flood process, and the method of step S3 is used to scale the calculation scaling ratio of the smoothed flood process;
[0154] S6: loop iterate steps S4 and S5 until the maximum number of iterations is reached or the error between the characteristic value of the smoothed flood process and the characteristic value of the target flood meets the requirement (i.e., the results before and after the iteration are basically the same), then exit the loop and output the final scaled flood process line result;
[0155] In this embodiment, the steps S4 and S5 are iterated in a loop. When the flood volume is directly scaled, a sudden change will occur in the flood process. When the flood process is smoothed by local weighted regression, the flood statistical characteristic values such as the peak flow and the maximum 3-day flood volume will be changed. By repeating these two steps, the scaling result is continuously approached to the ideal. The error between the characteristic value of the smoothed flood process and the characteristic value of the target flood meets the requirements. The Euclidean distance (Euclidean distance) between the two flood processes before and after scaling is calculated to evaluate the closeness of the processes before and after scaling. When the flood processes before and after scaling meet the set Euclidean distance threshold, they are determined to be close enough, and the loop is exited at this time.
[0156] The specific calculation formula is as follows:
[0157]
[0158] Among them, d q,j represents the Euclidean distance between the j-th iteration and the j-1-th iteration, q i,j represents the flow at time i in the jth iteration result, q i,j-1 Represents the flow at time i in the j-1th iteration result.
[0159] Example 2
[0160] In this embodiment, a typical flood process measured at a certain site is selected as a research example, and the specific process of scaling a single flood process by the adaptive scaling method for the designed flood process based on local weighted regression is specifically described.
[0161] In this example, a typical flood process is as follows: Figure 2 As shown in Figure 1, six parameters including flood peak flow, maximum 3-day flood volume, maximum 5-day flood volume, maximum 7-day flood volume, maximum 15-day flood volume and maximum 30-day flood volume are selected as control characteristic parameters, and the typical flood process is scaled to a once-in-a-century event (P=1%). The specific control characteristic values are shown in Table 1.
[0162] Table 1 Flood scaling control characteristic parameters
[0163]
[0164] In this embodiment, the specific implementation process is as follows:
[0165] S1: Arrange the above parameters as input parameters;
[0166] S2: Count the characteristic parameters of typical flood processes and scale the flood peak volume;
[0167] S21: According to the input typical flood process, the characteristic parameters of the typical flood process are calculated. The results are shown in Table 2:
[0168] Table 2 Characteristic parameters of typical flood processes
[0169]
[0170] S22: Calculate the traffic scaling period and scaling ratio for each period. The calculation results in this iteration are shown in Table 3:
[0171] Table 3. Scaling coefficients and time periods for the first iteration
[0172]
[0173] S23: Scale the flow rate of each period of a typical flood process according to Table 3. The process lines before and after scaling are as follows: Figure 2 As shown. After scaling the flood process to the target control characteristic value according to the scaling ratio, due to the different scaling factors of each time period, a sudden change will occur at the junction of the statistical time periods, especially when the difference in scaling factors is large. For example, the peak flow occurs at the 514th hour, the peak flow scaling factor is 1.493, and the scaling factor of the three-day flood total (W3) statistical period (482-553 hours) is 1.367, resulting in the flow at the 513th hour of the flow process from 8702.2m 3 / s suddenly changes to 9807.85m at 514h 3 / s. At the junction of the W3 statistical period and the W5 statistical period, the flow value increased from 2331.52m / s at 553h. 3 / s suddenly changes to 3006.15m at 554h 3 / s.
[0174] S3: Perform local weighted regression on the scaled flood process, and compare the before and after smoothing. Figure 3 The flood characteristics after smoothing are shown in Table 4. It can be seen that the smoothing method solves the problem of discontinuity of the flood process, but the flood characteristic parameters are changed after smoothing and cannot be consistent with the target flood characteristics. For example, the flood peak flow value is consistent with the target flood peak flow after scaling, which is 9807.85m 3 / s, after smoothing, the peak flow becomes 8947.63m 3 / s.
[0175] Table 4 Flood characteristic parameters after the first smoothing
[0176]
[0177] S4: Scale the smoothed flood process, then iterate and repeat steps S2 and S3 until one of the following two conditions is met and then exit the loop:
[0178] Condition 1: The maximum number of iterations is reached. After testing, the maximum number of iterations is finally set to 500 in this embodiment.
[0179] Condition 2: The two flood processes before and after the iteration are close enough, and the Euclidean distance is used as the judgment basis. In this embodiment, the Euclidean distance d between the two flood processes before and after the iteration is Q <=100, then exit the loop.
[0180] According to the above method, the flood process is corrected and smoothed iteratively, and the final result is as follows: Figure 4 As shown, the number of iterations and the Euclidean distance change process are as follows Figure 5 As shown in Figure 2, it can be seen that by repeating the two processes of iterative scaling and smoothing, the Euclidean distance between the two flood processes before and after the iteration gradually decreases, and the flood process gradually tends to be ideal. When the number of iterations is 107, d q =98.983, the condition is met and the loop is exited. The final output result not only ensures that the characteristic value is consistent with the target, but also overcomes the problem of discontinuity of the flood process and better retains the distribution pattern of the typical flood process.
[0181] S5: Output the final flood process data to the storage device in txt / excel format.
[0182] Example 3
[0183] In flood control planning, flood control engineering design, flood control scheduling scheme formulation and other work, it is often necessary to scale a typical flood process to different frequencies. This embodiment is used to illustrate the adaptability of the method of the present invention to the scaling of design floods under the control of flood parameters of different design frequencies. The typical flood process measured at a certain station in Example 2 is still taken as the typical process. The peak flow of the flood, the maximum 3-day flood total, the maximum 5-day flood total, the maximum 7-day flood total, the maximum 15-day flood total and the maximum 30-day flood total are selected as control indicators. The typical flood process is scaled to obtain design floods of once in 5 years (P = 20%), once in 10 years (P = 10%), once in 20 years (P = 5%), once in 30 years (P = 3.3%), once in 50 years (P = 2%), once in 100 years (P = 1%), once in 200 years (P = 0.5%), and once in 300 years (P = 0.33%). The flood characteristic control indicators of each frequency are shown in Table 5:
[0184] Table 5 Flood characteristic control indicators at different flood frequencies
[0185]
[0186] This embodiment provides a batch processing tool for batch scaling flood processes, scaling typical flood processes according to characteristic parameters of different frequencies, and specifically includes the following steps:
[0187] S1: Arrange typical flood process data and different frequency control parameters and input them into the system;
[0188] S2: Repeatedly call the flood scaling process in Example 1 to scale the flood process respectively, and the specific process will not be repeated;
[0189] S3: The processed result data is output to a storage device.
[0190] In this embodiment, the typical flood process is scaled according to the control parameters in Table 5, and the results are as follows: Figure 6 As shown, it can be seen that the present invention shows good applicability under the control of different frequency characteristic parameters. The scaling result better solves the problem of discontinuity in the flood process caused by traditional direct peak volume scaling. At the same time, it ensures that the characteristic parameters of the results are consistent with the target values, and better retains the distribution characteristics of typical floods.
[0191] Example 4
[0192] In flood control planning, flood control project design, flood control scheduling plan formulation and other work, different types of flood characteristic control parameters may be selected when scaling typical flood processes. For example, for typical flood processes with shorter duration, only the peak flow, the maximum 3-day flood volume, the maximum 5-day flood volume, the maximum 7-day flood volume, and the maximum 15-day flood volume can be selected for control, while for flood processes with longer duration, the maximum 30-day flood volume needs to be increased. This embodiment is used to illustrate the adaptability of the method of the present invention to the scaling of design floods under the control of different types of flood parameters. The typical flood process measured at a certain site in Example 2 is still used as a typical process. The control indicators of different types of flood characteristic parameters are shown in Table 6:
[0193] Table 6 Flood control parameters of different methods
[0194]
[0195] This embodiment provides a processing tool, which can select different types of flood characteristic control parameter combinations as needed, and scale the typical flood process according to different types of characteristic parameters, specifically including the following steps:
[0196] S1: Arrange typical flood process data and parameters of different control methods and input them into the system;
[0197] S2: Identify the control parameters and call the corresponding processing method to scale the flood process. In this embodiment, the flood scaling process in Example 1 needs to be modified, mainly the calculation of the scaling coefficient and the determination of the scaling time period when scaling the flood process. For example, when the control parameter is C2(Q m , W3, W7, W15 , W 30 When the control parameter is set, the scaling formula after removing the 3-day flood total period from the 7-day flood total statistical period is modified as shown below. The scaling method under other control parameters is changed accordingly and will not be repeated here.
[0198]
[0199]
[0200] The other processes are similar to those in Example 1, including scaling and smoothing iteration.
[0201] S3: The processed result data is output to a storage device.
[0202] In this embodiment, the typical flood process is scaled according to the control parameter form in Table 5, and the characteristic statistical parameters of the scaling result are shown in Table 7. It can be seen that the present invention shows good applicability under the control of different characteristic parameters. The specified parameters can be controlled according to the set target values, and the remaining parts more completely retain the peak distribution characteristics of the typical flood process. For example, comparing the scaling results of the two types of control parameters C1 and C3, when the control parameter type is C1, the output Q of the flood process m ,W3,W5,W7,W 15 ,W 30 A total of 6 characteristic parameters are consistent with the target values. When the control parameter type is C3, since there is no restriction on W3 and W5, these two characteristic parameters will be scaled to 1.541 billion m according to the scaling ratio of W7. 3 and 1.891 billion m 3 ,from Figure 7 It can be seen that the distribution characteristics of the typical flood process are more completely preserved during the statistical period of W7.
[0203] Table 7 Flood characteristic parameters after scaling in different ways
[0204]
[0205] Example 5
[0206] This embodiment provides a design flood process adaptive scaling system based on local weighted regression, including: an input parameter acquisition module, a characteristic parameter identification module, a scaling ratio calculation module, a flood process scaling module, a smoothing processing module, a post-smoothing adjustment module, a loop iteration module, a stop iteration determination module, and an output module;
[0207] In this embodiment, the input parameter acquisition module is used to acquire input parameters, including the flood process time series and the target peak volume characteristic value of the scaled flood;
[0208] In this embodiment, the characteristic parameter identification module is used to identify the characteristic parameters of the input flood process;
[0209] In this embodiment, the scaling ratio calculation module is used to calculate the scaling ratios of the flood peak and the flood volume in different periods according to the identified characteristic parameters of the flood process and the target peak volume characteristic value of the scaled flood;
[0210] In this embodiment, the flood process scaling module is used to scale the flood process according to the peak volume relationship;
[0211] In this embodiment, the smoothing processing module is used to smooth the scaled flood process using a local weighted regression method, by calculating the regression weight of the flow rate corresponding to each time point in the corresponding span at each time point, using a second-order polynomial to perform weighted linear least squares regression, and connecting the center points of the regression curve at each moment to obtain the smoothed flood process;
[0212] In this embodiment, the post-smoothing adjustment module is used to identify characteristic parameters of the smoothed flood process, and to calculate the scaling ratio of the smoothed flood process and then scale it;
[0213] In this embodiment, the cyclic iteration module is used for cyclically iterating the smoothing process of the flood process, identifying the characteristic parameters of the flood process after smoothing, and scaling the flood process after smoothing, until the iteration is stopped when it is determined that the preset stop iteration condition is met;
[0214] In this embodiment, the stop iteration determination module is used to determine whether the stop iteration condition is met;
[0215] In this embodiment, the output module is used to output the final scaled flood process line result.
[0216] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be equivalent replacement methods and are included in the protection scope of the present invention.
Claims
1. A method for adaptive scaling of design flood processes based on local weighted regression, characterized in that: The steps include: Obtain input parameters, including flood process time series and target peak volume characteristic values of scaled flood; Identify the characteristic parameters of the input flood process; According to the characteristic parameters of the identified flood process and the target peak characteristic value of the scaled flood, the scaling ratio of the flood peak and the flood volume in different periods is calculated, and the flood process is scaled according to the peak-volume relationship; The local weighted regression method is used to smooth the scaled flood process. The regression weight of the flow rate at each time point in the corresponding span is calculated, and the second-order polynomial is used for weighted linear least squares regression. The center points of the regression curve at each moment are connected to obtain the smoothed flood process. The local weighted regression method is used to smooth the scaled flood process, and the specific steps include: Calculate the regression weight of the flow rate at each time point in the corresponding span at each time point, expressed as: Among them, w i Yes i The smoothing weight of the flow at the moment, t represents the flow value to be smoothed at that moment, t i represents the time point of the flood process, dt represents the interval along the horizontal axis from time t to the farthest time point in the span; Using a second-order polynomial for weighted linear least squares regression, the quadratic regression at time t is expressed as: Among them, a, b, c represent the parameters to be determined; The loss function of local weighted regression is expressed as: Among them, q i Indicates the flow rate at each time point of the flood process; Based on the least squares method, the unknown parameters a, b, and c are solved to minimize the loss function J(θ) and obtain the fitting curve at each moment; Connect the center points of the fitting curves at each moment to get the smoothed flood process; Identify the characteristic parameters of the smoothed flood process, and calculate the scaling factor of the smoothed flood process and then scale it; Iterate the smoothing of the flood process, identify the characteristic parameters of the smoothed flood process, and scale the smoothed flood process until it is determined that the preset stop iteration condition is met, and output the final scaled flood process line result.
2. The method for adaptive scaling of design flood processes based on local weighted regression according to claim 1, characterized in that: The characteristic parameters of the flood process input by the identification include peak flow, flood volume, peak occurrence time, and flood volume statistical period; While counting the flood peak flow value, obtain the flood peak occurrence time; The calculation method of flood peak flow is expressed as: Q m =max(Q(t)) Among them, Q m is the peak flow of the flood process, Q(t) represents the flow value at each moment in the flood process; Identify the time t s The total amount of flood with the initial statistical length T: t e =t S +T Among them, W T,ts Indicates the time t s is the total amount of flood with the initial statistical length T, t s Indicates the starting time point of the statistical period, t e Indicates the end time of the statistical period; Identify the maximum flood volume of the flood process with a statistical length of T: W T =max(W T,ts ) Among them, W t It represents the maximum total flood volume of the flood process with a statistical length of T; At the same time, get W T The corresponding t s and t e , [t s , t e ] is W T statistical period.
3. The method for adaptively scaling a design flood process based on local weighted regression according to claim 1, characterized in that: The method of calculating the flood peak and the scaling ratio of the flood volume in different periods according to the identified characteristic parameters of the flood process and the target peak volume characteristic value of the scaled flood specifically includes: ... in, represents the peak flow target value, Characteristic parameters representing peak flow, It represents the target value of the total flood volume in the first statistical period T1. It indicates the target value of the total flood volume in the next statistical period T2. Indicates the nth statistical duration T n The target value of the total flood volume, It represents the characteristic parameter of the total flood volume in the first statistical time period T1. It represents the characteristic parameter of the total flood volume in the next statistical time period T2. Indicates the nth statistical duration T n Characteristic parameter of total flood volume, K m Indicates the peak flow scaling ratio, Indicates the scaling ratio of traffic in each time period.
4. The method for adaptive scaling of design flood process based on local weighted regression according to claim 3, characterized in that: According to the characteristic parameters of the identified flood process and the characteristic value of the target peak volume of the scaled flood, the scaling ratio of the flood peak and the flood volume in different periods is calculated, which is expressed as: in, They represent the characteristic parameters corresponding to the peak flow, the maximum 3-day flood volume, the maximum 5-day flood volume, the maximum 7-day flood volume, the maximum 15-day flood volume, and the maximum 30-day flood volume, respectively. They represent the target values corresponding to the peak flow, the maximum 3-day flood volume, the maximum 5-day flood volume, the maximum 7-day flood volume, the maximum 15-day flood volume, and the maximum 30-day flood volume, respectively. K m represents the peak flow scaling factor, K 3d , K 5d , K 7d , K 15d , K 30d They respectively represent the scaling ratio of the traffic in each time period.
5. The method for adaptively scaling a design flood process based on local weighted regression according to claim 4, characterized in that: The flood process is scaled according to the peak-volume relationship, and the specific scaling method is expressed as: Among them, Q′ m is the scaled peak flow; Q t ' 3-tm , Q t ' 5-t3 , Q t ' 7-t5 , Q t ' 15-t7 , Q t ' 30-t15 They are the flow rates at each time in the flood statistical period of three days after the flood peak period is removed after scaling, the flow rates at each time in the flood statistical period of five days after the three-day flow statistical period is removed, the flow rates at each time in the flood statistical period of seven days after the five-day flow statistical period is removed, the flow rates at each time in the flood statistical period of fifteen days after the seven-day flow statistical period is removed, and the flow rates at each time in the flood statistical period of thirty days after the fifteen-day flow statistical period is removed; is the peak flow of the flood process before scaling; Respectively represent the traffic in each statistical period before scaling.
6. The method for adaptive scaling of design flood process based on local weighted regression according to claim 3, characterized in that: According to the characteristic parameters of the identified flood process and the characteristic value of the target peak volume of the scaled flood, the scaling ratio of the flood peak and the flood volume in different periods is calculated, which is expressed as: in, They represent the characteristic parameters corresponding to the peak flow, the maximum 3-day flood volume, the maximum 7-day flood volume, the maximum 15-day flood volume, and the maximum 30-day flood volume, respectively. They represent the target values corresponding to the peak flow, the maximum 3-day flood volume, the maximum 7-day flood volume, the maximum 15-day flood volume, and the maximum 30-day flood volume, respectively. m represents the peak flow scaling factor, K 3d , K 7d , K 15d , K 30d They respectively represent the scaling ratio of the traffic in each time period.
7. The method for adaptive scaling of design flood processes based on local weighted regression according to claim 4, characterized in that: The flood process is scaled according to the peak-volume relationship, and the specific scaling method is expressed as: Among them, Q′ m is the scaled peak flow; Q t ' 3-tm , Q t ' 7-t3 , Q t ' 15-t7 , Q t ' 30-t15 They are the flow rates at each moment in the flood volume statistics period of three days after the flood peak period is removed after scaling, the flow rates at each moment in the flood volume statistics period of seven days after the three-day flow statistics period is removed, the flow rates at each moment in the flood volume statistics period of fifteen days after the seven-day flow statistics period is removed, and the flow rates at each moment in the flood volume statistics period of thirty days after the fifteen-day flow statistics period is removed; is the peak flow of the flood process before scaling; Respectively represent the traffic in each statistical period before scaling.
8. The method for adaptive scaling of design flood processes based on local weighted regression according to claim 1, characterized in that: The iterative cyclic flood process smoothing, identification of characteristic parameters of the smoothed flood process, and scaling of the smoothed flood process, until it is determined that a preset stop iteration condition is met, specifically includes: When the number of iterations reaches the preset maximum number, the iteration stops; Or by calculating the Euclidean distance between the two flood processes before and after scaling, specifically expressed as: Among them, d q,j represents the Euclidean distance between the j-th iteration and the j-1-th iteration, q i,j represents the flow at time i in the jth iteration result, q i,j-1 represents the flow at time i in the j-1th iteration result; The iteration stops when it is determined that the set Euclidean distance threshold is met.
9. A design flood process adaptive scaling system based on local weighted regression, characterized in that: include: Input parameter acquisition module, characteristic parameter identification module, scaling ratio calculation module, flood process scaling module, smoothing processing module, post-smoothing adjustment module, loop iteration module, stop iteration determination module, output module; The input parameter acquisition module is used to acquire input parameters, including the flood process time series and the target peak volume characteristic value of the scaled flood; The characteristic parameter identification module is used to identify the characteristic parameters of the input flood process; The scaling ratio calculation module is used to calculate the scaling ratios of the flood peak and the flood volume in different periods according to the characteristic parameters of the identified flood process and the target peak volume characteristic value of the scaled flood; The flood process scaling module is used to scale the flood process according to the peak volume relationship; The smoothing processing module is used to smooth the scaled flood process using a local weighted regression method, by calculating the regression weight of the flow rate corresponding to each time point in the corresponding span at each time point, using a second-order polynomial to perform weighted linear least squares regression, and connecting the center points of the regression curve at each moment to obtain the smoothed flood process; The local weighted regression method is used to smooth the scaled flood process, specifically including: Calculate the regression weight of the flow rate at each time point in the corresponding span at each time point, expressed as: Among them, w i Yes i The smoothing weight of the flow at the moment, t represents the flow value to be smoothed at that moment, t i represents the time point of the flood process, dt represents the interval along the horizontal axis from time t to the farthest time point in the span; Using a second-order polynomial for weighted linear least squares regression, the quadratic regression at time t is expressed as: Among them, a, b, c represent the parameters to be determined; The loss function of local weighted regression is expressed as: Among them, q i Indicates the flow rate at each time point of the flood process; Based on the least squares method, the unknown parameters a, b, and c are solved to minimize the loss function J(θ) and obtain the fitting curve at each moment; Connect the center points of the fitting curves at each moment to get the smoothed flood process; The post-smoothing adjustment module is used to identify characteristic parameters of the smoothed flood process, and to scale the smoothed flood process after calculating the scaling ratio; The loop iteration module is used for loop iteration of the smoothing process of the flood process, identification of characteristic parameters of the flood process after smoothing, and scaling of the flood process after smoothing, until the iteration is stopped when it is determined that a preset stop iteration condition is met; The stop iteration determination module is used to determine whether the stop iteration condition is met; The output module is used to output the final scaled flood process line result.