Intelligent extension method for water level flow relation based on similar flood

Abstract

The invention relates to the technical field of hydrological data analysis, in particular to a similar flood-based water level flow relationship intelligent extension method. The method comprises the following steps: acquiring actually measured water level and flow data and performing curve fitting; carrying out three-property test on the fitting curve, and if the fitting curve does not meet the requirement, adopting multi-point sectional type fitting to obtain an output curve; screening a similar curve based on a historical flood graphic library, and carrying out extension processing to form an extended water level flow relation curve; and the real-time water level is subjected to flow pushing by utilizing the extension curve, so that accurate prediction and dynamic extension of the water level and flow relationship are realized. According to the method, the reliability of the curve is guaranteed through actually measured water level flow data fitting and segmented optimization in combination with three-property test, and intelligent complementation and real-time correction under the condition of insufficient data are realized by utilizing a historical flood similar curve extension and dynamic matching mechanism, so that the precision, continuity and adaptability of the water level flow relation curve are remarkably improved; and the accuracy and practicability of the hydrological plug flow process are ensured.

Description

Technical Field

[0001] This invention relates to the field of hydrological data analysis technology, and in particular to a method for intelligently extending the relationship between water level and flow rate based on similar floods. Background Technology

[0002] In recent years, the timeliness requirements for hydrological data have become increasingly stringent in areas such as flood and drought disaster prevention, water resource assessment, and water ecological protection. Real-time extension of flow rate data has become a new requirement for current data compilation. However, existing real-time flow rate extension methods largely rely on manual experience for curve setting, resulting in an over-reliance on manual experience and technical management. This makes it impossible to extend the water level-flow rate relationship curve in a timely and accurate manner, leading to significant errors in the real-time flow rate estimates. Therefore, how to extend the water level-flow rate relationship curve in a timely and accurate manner is a pressing issue that needs to be addressed in the field of hydrology. Summary of the Invention

[0003] Therefore, it is necessary to provide a method for intelligent extension of the water level-discharge relationship based on similar floods to solve at least one of the above-mentioned technical problems.

[0004] To achieve the above objectives, a method for intelligent extension of water level-discharge relationship based on similar floods is proposed, the method comprising the following steps: Step S1: Obtain measured flow rate and water level process data; fit the measured flow rate and water level process data to a water level-flow rate relationship line based on the measured water level point and flow rate point to obtain the fitted water level-flow rate relationship curve; Step S2: Determine whether the fitted water level-discharge relationship curve passes the three-property test. If yes, mark the fitted water level-discharge relationship curve as the output curve; otherwise, perform multi-point segmented fitting on the fitted water level-discharge relationship curve to obtain the output curve after secondary fitting. Step S3: Use a preset historical flood graph library to filter similar curves that match the output curve, and extend the water level-discharge process of the output curve through similar curves to obtain the extended water level-discharge relationship curve; Step S4: Perform water level propagation based on the collected real-time water level using the extended water level-flow relationship curve.

[0005] Preferably, step S2, which involves performing multi-point piecewise fitting on the fitted water level-discharge relationship curve, includes: The fitted water level-discharge relationship curve is divided into several continuous water level segments according to the slope change characteristics of the water level process, and the measured water level point and discharge point in each segment are used as the feature point set of that segment. For any water level section, the first difference feature of the water level section is obtained based on the difference between the flow value of each characteristic point in the section and the mean flow value of all characteristic points in the section. The second difference feature of the water level section is obtained based on the difference between the water level slope of each feature point in the section and the mean of the water level slope of all feature points in the section. Based on the first and second difference features, determine the piecewise fitting curve for this water level section; Connect the piecewise fitted curves of all water level sections in order of water level to obtain the output curve after secondary fitting.

[0006] Preferably, the piecewise fitting curve for the water level segment is determined based on the first difference feature and the second difference feature, including: Based on the first and second difference features, the largest first difference feature point, the largest second difference feature point, and the remaining feature points within the water level range are selected. Based on the largest first difference feature point and the largest second difference feature point, a piecewise curve profile is established; By fitting the remaining feature points with the piecewise curve profile, a piecewise fitted curve for the water level section is obtained.

[0007] Preferably, fitting curve points to the remaining feature points using a piecewise curve profile includes: Determine the number of feature points of the remaining feature points. If the number of remaining feature points is greater than the preset number of fitting standards, then merge the adjacent remaining feature points so that the number of feature points is equal to the preset number of fitting standards. When the number of remaining feature points is less than the preset number of fitting standards, a preset compensation point is inserted based on the interval between adjacent remaining feature points, so that the number of feature points is equal to the preset number of fitting standards.

[0008] Preferably, step S2, determining whether the fitted water level-discharge relationship curve passes the three tests, specifically includes: sign test, fitting test, and deviation from numerical value test.

[0009] Preferably, step S3 includes the following steps: Step S31: Use the water level range of the output curve as the initial search boundary, and filter similar curves that match the output curve from the preset historical flood graphic library; Step S32: Determine the extension segment of the output curve based on the curve shape of similar curves and perform smoothing to obtain the smoothed curve extension segment; Step S33: Extend the water level-flow rate process of the output curve by extending the curve extension segment to obtain the extended water level-flow rate relationship curve.

[0010] Preferably, step S31 includes the following steps: Step S311: Use the water level range of the output curve as the initial search boundary; Step S312: Based on the initial search boundary, filter similar curves that match the output curve from the preset historical flood graphic library. If the match is successful, output the similar curve directly. Step S313: If the matching fails, the matching range is expanded cyclically with a preset step size until the matching is successful and a similarity curve is output. The number of expansions and the corresponding similarity change data are recorded.

[0011] Preferably, after step S33, the method further includes: Real-time water level data is collected. Confirm the water level range based on real-time water level; If the maximum or minimum value of the water level range does not belong to the extended water level-discharge relationship curve, then repeat steps S31-S33.

[0012] Preferred methods for obtaining measured water level and flow rate points include: Acquire station location data; deploy water level observation devices based on station location data, and collect water level data at preset time intervals to obtain a continuous measured water level sequence as the measured water level point; Obtain the location data of the water level observation section; deploy the flow measurement device based on the location data of the water level observation section, and calculate the flow value by collecting the measured flow velocity and cross-sectional area to form the corresponding measured flow sequence as flow point.

[0013] Preferred methods for constructing a pre-defined historical flood graphic database include: Obtain historical water level and flow rate curve data; Extract curve features from historical water level and flow rate curve data to construct a curve feature vector library; The curve feature vector library is updated in real time to obtain a preset historical flood graphic library.

[0014] The beneficial effects of this invention are as follows: First, by fitting curves to measured water level and flow rate points and combining this with a three-property test for validity assessment, the problem of excessive deviation or insufficient stability inherent in traditional single-fitting methods can be avoided. When the fitting result fails to meet the three-property test, this invention employs multi-point segmented fitting, segmenting the curve according to the slope change characteristics of the water level process, and optimizing the fitting by combining the difference characteristics. This effectively improves the accuracy and rationality of the fitted curve, ensures the continuity and stability of the output curve, and thus enhances the reliability of the water level-flow rate relationship.

[0015] Second, by establishing a historical flood graph database and extracting curve feature vectors, similar curves can be quickly retrieved from the historical database and extended when the current fitted curve data is limited. This overcomes the limitation of existing methods that cannot deduce complete relationship curves when flood data is insufficient or incomplete. Simultaneously, through similar curve extension and smoothing, the natural transition of the extended curve segments is effectively ensured, avoiding abrupt changes or discontinuities during curve splicing, thus improving the overall fitting effect of the water level-discharge relationship.

[0016] Third, a dynamic matching and iterative expansion retrieval strategy is introduced during the curve extension process. If a match is not successfully found at the initial retrieval boundary, similar curves are obtained by gradually expanding the retrieval range, and similarity change data is recorded to provide a reference for subsequent accuracy optimization of curve extension. This dynamic retrieval method enhances the adaptability and robustness of the method, enabling it to adapt to the differences in curve characteristics under different hydrological conditions.

[0017] Fourth, by combining real-time water level data, the extended water level-discharge relationship curve is corrected in real time. When the real-time hydrological range exceeds the current curve range, the historical database can be called again for iterative extension, ensuring that the derived results always remain consistent with the actual hydrological process, thereby improving the real-time performance and adaptability of the method. Through this mechanism, this invention can support the dynamic flow projection needs during flood processes, providing more accurate support for hydrological forecasting and flood control scheduling. Attached Figure Description

[0018] Figure 1 A schematic diagram illustrating the steps of an intelligent extension method based on the water level-discharge relationship of similar floods; Figure 2 for Figure 1 A detailed flowchart illustrating the implementation steps of step S3. Figure 3 This is a water level-discharge propagation curve diagram based on the intelligent extension method of water level-discharge relationship of similar floods in this application; The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] The technical method of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0020] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0021] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0022] To achieve the above objectives, please refer to Figures 1 to 3 A method for intelligent extension of water level-discharge relationship based on similar floods, the method comprising the following steps: Step S1: Obtain measured flow rate and water level process data; fit the measured flow rate and water level process data to a water level-flow rate relationship line based on the measured water level point and flow rate point to obtain the fitted water level-flow rate relationship curve; In one embodiment, the measured water level data is collected in real time using automatic water level gauges (such as ultrasonic or pressure gauges) deployed at the cross-section of the river or channel; the measured flow rate data is calculated using a flow velocity meter (such as an electromagnetic flow velocity meter or an acoustic Doppler current profiler) combined with the cross-sectional area. To ensure the reliability of the fitted curve, the sampling time interval is preferably set to 5 to 30 minutes, and the sampling interval is appropriately shortened during flood events or periods of sudden increase in water flow to improve monitoring accuracy.

[0023] In another embodiment, the scope of measured data collection should cover low water level, normal water level, and high water level stages to ensure the applicability and accuracy of the fitted relationship curve. For example, no fewer than 20 sets of water level-discharge correspondence points can be collected during the dry season, wet season, and flood season respectively, ultimately forming a dataset covering the entire hydrological process. For the fitting method, empirical formula methods (such as power function forms) can be selected. The least squares regression method or piecewise fitting method is used to obtain the water level-discharge relationship curve. Indicates flow rate. Indicates water level. Indicates the starting water level. These are the fitting coefficients. A continuous and smooth water level-flow curve can be obtained through fitting.

[0024] For example, measured data collected from a certain river section shows that when the water level is between 1.2m and 3.5m, the measured flow rate ranges from 10m³ / s to 320m³ / s; the curve obtained after power function regression fitting is... Coefficient of determination The value of 0.985 indicates a good fit.

[0025] Step S2: Determine whether the fitted water level-discharge relationship curve passes the three-property test. If yes, mark the fitted water level-discharge relationship curve as the output curve; otherwise, perform multi-point segmented fitting on the fitted water level-discharge relationship curve to obtain the output curve after secondary fitting. In one embodiment, the three-property test specifically includes a sign test, a fit test, and a deviation value test: when performing the sign test, the number of positive and negative signs of the deviation of the measuring point from the curve is counted respectively (those with a deviation value of zero are equally distributed as positive and negative measuring points), and then... Calculate the statistic value, where For statistical purposes, The total number of measurement points. The number of positive or negative signs. The probability of a positive or negative sign is 0.5 each. The results are converted to continuous corrections (discrete to continuous), and the distribution of the statistical values ​​is checked for randomness and regularity. If the residuals are alternately positive and negative, and there is no sustained single sign or systematic deviation within a large interval, the result is considered passed; otherwise, it is considered failed.

[0026] Fit test: Arrange the measuring points in ascending order of water level, starting from the second point, and count the sign changes of deviations. A sign change is recorded as 1, otherwise as 0, and compared with the theoretical expected number of deviations. If the number of sign changes falls within a reasonable range below the significance level, it is considered a pass; if it is significantly less or more than expected, it is considered a fail.

[0027] Deviation from numerical value test: by formula and formula Calculate the statistic and the standard deviation of the mean relative deviation, respectively, where the formula is as follows: For statistical purposes, This represents the average relative deviation. The standard deviation of the average relative deviation. The standard deviation of the relative deviation value. The total number of measurement points. The relative deviation between the measuring point and the relationship curve, and... The statistic is compared with the theoretical expected frequency. If the statistic falls within a reasonable range below the significance level, it is considered reasonable and the test is accepted; otherwise, the null hypothesis is rejected.

[0028] In another embodiment, when the fitted curve fails any of the three tests mentioned above, the system will automatically perform multi-point segmented fitting: that is, the water level range is divided into several segments (such as the low water level range of 1.0m to 2.5m, the medium water level range of 2.5m to 3.8m, and the high water level range of 3.8m to 5.0m), and local function fitting (power function, quadratic polynomial, or logarithmic function, etc.) is performed on each segment. Smooth transition processing is applied at the junction of the segments to generate a quadratic fitting output curve that conforms to the sign rule, has good adaptability, and has a deviation within the allowable range.

[0029] For example, in monitoring a certain river section, the sign test of the overall fitted curve failed, and the deviation in the high water level section exceeded 20%. After adopting piecewise fitting: the low water level section adopted a power function form. , =0.981; the middle water level section uses a quadratic polynomial. , =0.989; Logarithmic function is used for high water level sections. , =0.985. After segmented splicing and smoothing transition, all three properties tests were passed, and the final output curve can be accurately used for subsequent water level and flow rate conversion.

[0030] Step S3: Use a preset historical flood graph library to filter similar curves that match the output curve, and extend the water level-discharge process of the output curve through similar curves to obtain the extended water level-discharge relationship curve; In one embodiment, the historical flood graph database pre-stores water level-discharge relationship curve data for multiple typical flood processes, covering different watersheds and different flood types (such as slow-rising and slow-falling, rapid-rising and rapid-falling, bimodal, etc.). After obtaining the output curve in step S2, similar curve filtering and extension are performed through the following steps: The output curve is normalized to each curve in the historical flood graph database. A similarity index is calculated using methods such as Dynamic Time Warping (DTW), correlation coefficient matching, or Euclidean distance minimization. Curves with a similarity greater than 0.9 are considered candidate similar curves. Further error analysis is performed on multiple candidate curves, and the curve with the smallest sum of squared residuals and the closest trend is selected as the final similar curve. The endpoint of the output curve is used as the junction point. The similar curves are extended and spliced ​​using the water level and flow rate variation patterns after this junction point. A smooth transition algorithm (such as weighted average or spline interpolation) is applied at the splicing point to generate the extended water level and flow rate relationship curve.

[0031] In another embodiment, to ensure the rationality of the extended section, constraints are set on the extended curve: the rate of change of flow in the extended section does not exceed twice that of the original curve's final segment to prevent non-physical abrupt changes; the peak flow and peak water level of the extended section should remain within the 95% confidence interval of historical similar flood curves; and the overall fitting determination coefficient of the extended curve... Keep it above 0.95.

[0032] For example, during the 2022 flood season monitoring of a certain river basin, the output curve lacked data in the high-water level range, resulting in an incomplete flood peak interval. By accessing the "rapid rise and slow fall" flood process curve from the historical flood graph database, a similarity score of 0.936 was obtained, and this curve was selected as the optimal candidate curve. Finally, by extending and splicing the curves, a complete water level-discharge relationship curve was obtained, extending the peak discharge from the originally fitted 2100 m³ / s to 2650 m³ / s, which showed good agreement with similar historical flood processes.

[0033] Step S4: Perform water level propagation based on the collected real-time water level using the extended water level-flow relationship curve.

[0034] In one embodiment, reference Figure 3 First, real-time water level data of the target section is acquired through a real-time water level monitoring device (such as a water level gauge or radar water level sensor), and this real-time water level is used as input. The extended water level-flow rate relationship curve obtained in step S3 is then called, and the real-time water level is matched with the curve's correspondence to obtain the corresponding real-time flow rate value. The specific processing steps include: positioning the real-time water level value on the coordinate axis of the extended water level-flow rate relationship curve to confirm the corresponding interval; when the real-time water level falls between known sampling points, piecewise linear interpolation or cubic spline interpolation methods are used to calculate the corresponding flow rate value to avoid insufficient accuracy due to discrete points; the real-time water level is paired with the calculated flow rate value to form a real-time water level-flow rate pair, which is output as the water level-driven flow result.

[0035] In another embodiment, to ensure the stability of the flow projection results, smoothing processing and constraints are set for the real-time estimated flow values: when the flow rate change rate exceeds 15% in two consecutive sampling periods, anomaly correction is triggered, and the result is smoothed using a moving average method; when the real-time water level exceeds the defined range of the extended water level-flow relationship curve, the system triggers an extrapolation algorithm to extend the calculation using the end trend, but the extrapolation range does not exceed 10% of the end water level of the curve to prevent distortion.

[0036] For example, in real-time monitoring at a hydrological station, when the measured water level is 12.8m, the corresponding flow rate is found to be 1840 m³ / s using the extended water level-flow curve; as the water level gradually rises to 14.2m, the corresponding flow rate is calculated to be 2460 m³ / s. By using the water level-driven flow method, real-time monitoring and prediction of flood events can be achieved even in the absence of actual measured flow rates.

[0037] Preferably, step S2, which involves performing multi-point piecewise fitting on the fitted water level-discharge relationship curve, includes: The fitted water level-discharge relationship curve is divided into several continuous water level segments according to the slope change characteristics of the water level process, and the measured water level point and discharge point in each segment are used as the feature point set of that segment. For any water level section, the first difference feature of the water level section is obtained based on the difference between the flow value of each characteristic point in the section and the mean flow value of all characteristic points in the section. The second difference feature of the water level section is obtained based on the difference between the water level slope of each feature point in the section and the mean of the water level slope of all feature points in the section. Based on the first and second difference features, determine the piecewise fitting curve for this water level section; Connect the piecewise fitted curves of all water level sections in order of water level to obtain the output curve after secondary fitting.

[0038] In one embodiment, when the fitted water level-flow rate relationship curve fails the three-property test, a multi-point segmented fitting is performed to improve the matching accuracy between the fitted curve and the measured data.

[0039] First, the fitted water level-discharge relationship curve is segmented according to the slope change characteristics of the water level process. Specifically, the slope change rate at each point on the curve is calculated. When the slope change rate of adjacent points exceeds a set threshold (e.g., 15%–25%), a boundary point is automatically generated, thus dividing the overall curve into several continuous water level segments. Within each water level segment, the measured water level point and discharge point of that segment are extracted as a feature point set. For any given water level segment, the difference between the discharge value of each feature point and the mean discharge value of all feature points in the segment is calculated and recorded as the first difference feature of that segment. Simultaneously, the difference between the water level slope of each feature point in the segment and the mean water level slope of all feature points in the segment is calculated and recorded as the second difference feature of that segment.

[0040] In another implementation, the system determines the piecewise fitting curve for the section based on a first difference feature and a second difference feature. Specifically: when the first difference feature is small and the second difference feature is large, quadratic polynomial fitting is preferred to reflect rapid changes in the slope; when the first difference feature is large and the second difference feature is small, piecewise linear fitting is used to ensure the stability of the flow rate within the section; when both are large, piecewise cubic spline fitting is used to balance the smoothness and accuracy of the fitting. Finally, the piecewise fitting curves of all water level sections are spliced ​​and connected according to the water level order to obtain a continuous and smooth quadratic fitting output curve.

[0041] For example, in a certain measured dataset, the original fitted curve deviated significantly in the water level range of 8.5m to 10.2m. By using multi-point piecewise fitting, this range was divided into three sub-segments: the first sub-segment used linear fitting, the second sub-segment used quadratic fitting, and the third sub-segment used cubic spline fitting. The resulting quadratic fitted curve, obtained after splicing, significantly improved the overall fitting accuracy and successfully passed the three-property test.

[0042] Preferably, the piecewise fitting curve for the water level segment is determined based on the first difference feature and the second difference feature, including: Based on the first and second difference features, the largest first difference feature point, the largest second difference feature point, and the remaining feature points within the water level range are selected. Based on the largest first difference feature point and the largest second difference feature point, a piecewise curve profile is established; By fitting the remaining feature points with the piecewise curve profile, a piecewise fitted curve for the water level section is obtained.

[0043] In one embodiment, a first difference feature (i.e., the degree to which the flow rate deviates from the mean) and a second difference feature (i.e., the degree to which the water level slope deviates from the mean) are calculated for all feature points within the water level section. After the calculation is completed, the feature point with the largest first difference feature and the feature point with the largest second difference feature are selected, and the remaining feature points are regarded as ordinary points.

[0044] Secondly, based on the spatial location and numerical relationship of the largest first difference feature point and the largest second difference feature point, a piecewise curve profile for this water level segment is established. The largest first difference feature point primarily reflects the extreme variation trend of the flow rate, while the largest second difference feature point primarily reflects the extreme variation trend of the water level slope. Together, they determine the key inflection points and curve trends of this water level segment.

[0045] Then, using the piecewise curve profile as a constraint, the remaining feature points within the segment are fitted. Specifically, using weighted least squares or spline interpolation, ordinary points are fitted to the curve points while ensuring consistency with the piecewise curve profile, thus obtaining the piecewise fitted curve for the water level segment.

[0046] In another implementation, if the water level distance between the largest first difference feature point and the largest second difference feature point exceeds a preset threshold (e.g., 0.3m to 0.5m), additional auxiliary points are introduced in the piecewise curve profile to avoid excessive bending or fitting distortion of the fitted curve.

[0047] For example, in a certain water level range (9.0m to 10.0m), the calculation results show that the flow rate at a certain characteristic point deviates from the average of the range by 15%, which is identified as the first feature point with the largest difference; another feature point has a slope that deviates from the average of the range by 20%, which is identified as the second feature point with the largest difference. After establishing the piecewise curve profile based on these two extreme points, a weighted fitting is then performed on other ordinary points. The final piecewise curve can simultaneously reflect the abrupt change characteristics of the flow rate and the changing trend of the water level slope, significantly improving the fitting accuracy.

[0048] Preferably, fitting curve points to the remaining feature points using a piecewise curve profile includes: Determine the number of feature points of the remaining feature points. If the number of remaining feature points is greater than the preset number of fitting standards, then merge the adjacent remaining feature points so that the number of feature points is equal to the preset number of fitting standards. When the number of remaining feature points is less than the preset number of fitting standards, a preset compensation point is inserted based on the interval between adjacent remaining feature points, so that the number of feature points is equal to the preset number of fitting standards.

[0049] In one embodiment, a set of remaining feature points, excluding the largest first and second largest difference feature points, is obtained, and their number is counted. If the number of remaining feature points is determined to be greater than a preset number of fitting criteria (e.g., 6 points), a feature point merging operation is performed. Specifically, based on the order of the feature points on the water level coordinate axis, two adjacent feature points are merged into one representative point. The merging method uses an arithmetic mean or a weighted average, ensuring that the final number of retained feature points is exactly equal to the preset number of fitting criteria. This method avoids the problem of curve overfitting caused by too many feature points.

[0050] When the number of remaining feature points is determined to be less than the preset number of fitting standards, a feature point compensation operation is performed. Specifically, one or more compensation points are inserted in the water level interval between adjacent feature points using linear interpolation or spline interpolation until the number of remaining feature points reaches the number of fitting standards. This method avoids the problem of an overly coarse curve due to insufficient feature points. Finally, the processed set of feature points (equal to the number of fitting standards) is combined with the piecewise curve profile, and a smooth piecewise fitted curve for that water level segment is generated through least-squares fitting or spline curve fitting.

[0051] In one specific embodiment, the number of fitting standards can preferably be set to 5 to 8 points. For example, in a certain water level section, if the number of other feature points is 12 and the number of fitting standards is set to 6, then by merging adjacent points, 6 representative points are finally obtained; if the number of other feature points is only 3, then 3 compensation points are inserted between adjacent points to make the total number reach 6, ensuring the stability and reliability of the fitting curve.

[0052] Preferably, step S2, determining whether the fitted water level-discharge relationship curve passes the three tests, specifically includes: sign test, fitting test, and deviation from numerical value test.

[0053] In one embodiment, determining whether the fitted water level-discharge relationship curve passes the three-property test specifically includes the following testing method: For the fitted water level-discharge relationship curve, calculate the difference between the measured flow rate and the fitted flow rate point by point, and count the number of times the sign of the difference changes. If the number of sign changes is greater than a set reasonable range (e.g., not exceeding 30% of the total number of points), it indicates that the curve is consistent with the measured data in the overall trend, and the sign test is determined to be passed; otherwise, it is determined to be failed.

[0054] Calculate the correlation coefficient between the fitted water level-discharge relationship curve and the measured points. When the correlation coefficient reaches a preset threshold (e.g., above 0.95), it indicates that the fitted curve can reflect the linear or nonlinear relationship of the measured data well, thus determining that the fitting test has passed; otherwise, it is determined that it has not passed.

[0055] Calculate the absolute value of the deviation between the measured flow rate and the fitted flow rate, and obtain the root mean square error (RMSE) of all deviations. If the RMSE is less than a set threshold (e.g., not exceeding 5% to 10% of the average measured flow rate), it indicates that the fitting accuracy of the fitted curve to the measured data meets the requirements, and the deviation value test is passed; otherwise, it is considered to have failed.

[0056] When all three of the above three tests are passed, the fitted water level-flow rate relationship curve is considered to be an output curve that meets the accuracy and stability requirements; if any one test fails, subsequent multi-point segmented fitting is triggered to improve the fitting effect and stability of the curve.

[0057] In one specific embodiment, the threshold for the sign test can be selected to be no more than 25% to 30% of the total number of points; the correlation coefficient threshold for the fitting test can preferably be 0.95 to 0.98; and the root mean square error threshold for the deviation from the numerical value test can be set to no more than 10% of the average measured flow rate. Through the comprehensive judgment of the above indicators, the overall reliability of the fitted curve in terms of trend consistency, correlation, and accuracy can be effectively guaranteed.

[0058] As an example of the present invention, reference is made to... Figure 2 As shown, step S3 in this example includes: Step S31: Use the water level range of the output curve as the initial search boundary, and filter similar curves that match the output curve from the preset historical flood graphic library; Step S32: Determine the extension segment of the output curve based on the curve shape of similar curves and perform smoothing to obtain the smoothed curve extension segment; Step S33: Extend the water level-flow rate process of the output curve by extending the curve extension segment to obtain the extended water level-flow rate relationship curve.

[0059] In one embodiment, the water level range of the output curve is determined, and this water level range is used as the initial search boundary. Then, curves with similar line shapes to the output curve within the water level range are selected from a preset historical flood graph database. The selection process may include the following operations: normalizing the output curve and the historical database curves to unify the dimensions of water level and flow; calculating the similarity index between the output curve and the historical database curve within the initial search boundary, which can be done using methods such as Euclidean distance, dynamic time warping (DTW), or correlation coefficient; and selecting several similar curves based on the similarity index as a candidate extended curve set.

[0060] Based on the linear characteristics of candidate similar curves, the extended segment of the output curve is determined. Specific methods include: analyzing the linear trend of the similar curve outside the water level range of the output curve, such as slope changes and convexity / concavity; smoothing the extension segment of the similar curve and the endpoint of the output curve to ensure no significant jumps or abrupt changes at the junction, using spline smoothing or moving average filtering methods; obtaining the smoothed extension segment curve, ensuring that the extension segment's water level change trend is continuous and consistent with the output curve.

[0061] The output curve is extended by smoothing the extension segment to extend the water level-flow process. Specifically, this involves: connecting the smoothed extension segment to the end of the output curve so that the output curve can extend beyond the original water level range; making local fine adjustments at the connection point to ensure continuous and abrupt flow changes; and finally obtaining the extended water level-flow relationship curve, which can be used for subsequent real-time water level propagation or flood forecasting.

[0062] In one specific embodiment, the junction between the end of the output curve and the extended section of the similar curve can be set to a flow rate variation tolerance of ±0.5% to ensure that the extended curve is smooth and practically applicable.

[0063] Preferably, step S31 includes the following steps: Step S311: Use the water level range of the output curve as the initial search boundary; Step S312: Based on the initial search boundary, filter similar curves that match the output curve from the preset historical flood graphic library. If the match is successful, output the similar curve directly. Step S313: If the matching fails, the matching range is expanded cyclically with a preset step size until the matching is successful and a similarity curve is output. The number of expansions and the corresponding similarity change data are recorded.

[0064] In one embodiment, an initial search boundary is determined based on the water level range of the output curve. The initial search boundary is the water level interval between the lowest and highest water levels in the output curve, used to limit the screening range of the historical flood image database. Within the initial search boundary, similar curves matching the output curve are screened from the preset historical flood image database. Specific methods may include: normalizing the output curve and the historical database curves to unify their dimensions; using similarity calculation methods, such as Euclidean distance, dynamic time warping (DTW), or correlation coefficient, to compare the degree of matching between the output curve and the historical curve within the search boundary; if the similarity of the screened historical curve is higher than a preset threshold, a successful match is determined, and the historical curve is directly used as the output similar curve.

[0065] If no curve meeting the similarity threshold is found within the initial search boundary, the matching range is expanded cyclically: the water level range of the output curve is gradually expanded by a preset step size, for example, by ±0.01 meters each time; after each expansion, the similar curve screening is re-executed, and the matching similarity is calculated; the above operation is repeated until a match is successful; after a successful match, the number of expansion cycles and the similarity change data corresponding to each expansion are recorded for subsequent analysis of the stability and reliability of the output curve matching.

[0066] In one specific embodiment, the preset matching similarity threshold is 0.85 to 0.95, and the matching range step size is set to 0.1 to 0.2 meters, which can ensure that even when the historical database lacks accurate matching curves, curves with similar trends can still be found and reasonably extended.

[0067] Preferably, after step S33, the method further includes: Real-time water level data is collected. Confirm the water level range based on real-time water level; If the maximum or minimum value of the real-time hydrological range does not belong to the extended water level-discharge relationship curve, then repeat steps S31-S33.

[0068] In one embodiment, real-time water levels at the target hydrological station or monitoring point are collected to obtain the latest water level data sequence. Based on the collected real-time water level data, the real-time hydrological range is determined, that is, the minimum and maximum values ​​in the current water level data sequence are identified to reflect the fluctuation range of the current hydrological conditions.

[0069] Compare the real-time hydrological range with the extended water level-discharge relationship curve: If the maximum or minimum value of the real-time hydrological range is not within the water level range covered by the extended water level-discharge relationship curve, it is determined that the extended curve cannot completely cover the current hydrological situation; in this case, repeat steps S31 to S33, that is, use the water level range of the current output curve as the initial search boundary again, filter similar curves in the historical flood graphic library, and generate a new extended curve until the extended water level-discharge relationship curve covers the real-time hydrological range.

[0070] In a preferred embodiment, a maximum number of loop executions can be set, for example, no more than 5 times, to avoid infinite loops; at the same time, the water level range changes and historical curve matching similarity can be recorded each time the curve is extended, in order to analyze the stability and reliability of the curve extension.

[0071] Preferred methods for obtaining measured water level and flow rate points include: Acquire station location data; deploy water level observation devices based on station location data, and collect water level data at preset time intervals to obtain a continuous measured water level sequence as the measured water level point; Obtain the location data of the water level observation section; deploy the flow measurement device based on the location data of the water level observation section, and calculate the flow value by collecting the measured flow velocity and cross-sectional area to form the corresponding measured flow sequence as flow point.

[0072] In one embodiment, the location data of the target hydrological station is first acquired, including the station's latitude and longitude coordinates and elevation information. Based on the station location data, water level observation devices, such as water level gauges, water gauges, or ultrasonic water level sensors, are deployed on-site and automatically collect water level data at preset time intervals (e.g., every 5 to 30 minutes), forming a continuous water level time series. Each sampling point in this water level time series is the measured water level point.

[0073] Acquire location data of the water level observation section, including the start and end points of the section and its geometric shape. Deploy flow measurement devices, such as multi-point velocity meters, electromagnetic flowmeters, or radar velocity sensors, at the water level observation section location. Calculate the flow rate by combining the measured flow velocity data collected at the section with the geometric area of ​​the section. Integrate the flow rate values ​​from each sampling point at the section to obtain a continuous flow rate time series, which corresponds to the measured water level series, forming the measured flow rate points.

[0074] To ensure the time correspondence between water level and flow rate points, a unified time synchronization device, such as a GPS time synchronization module, can be used to ensure that the sampling time error of water level and flow rate measurements is within ±1 second.

[0075] Preferred methods for constructing a pre-defined historical flood graphic database include: Obtain historical water level and flow rate curve data; Extract curve features from historical water level and flow rate curve data to construct a curve feature vector library; The curve feature vector library is updated in real time to obtain a preset historical flood graphic library.

[0076] In one embodiment, historical hydrological station water level and flow curve data are collected, including water level time series and corresponding flow time series, covering the longest possible historical time range (such as records of consecutive flood events over the past decade or many years) to ensure the integrity and representativeness of the data in the database.

[0077] Feature extraction is performed on the collected historical water level-discharge curve data, including but not limited to: the start and end water levels of the curve, peak flow rate, rate of rise / fall, number and location of inflection points, and curve morphology index. The features of each curve are organized into curve feature vectors to form a curve feature vector library.

[0078] To ensure the real-time performance and scalability of the historical flood graphic database, a real-time data update mechanism is implemented for the curve feature vector database. Once new water level and flow measurement data or newly occurring flood event data are collected, the corresponding curve feature vectors are calculated and added to the feature vector database. Simultaneously, if existing feature vectors are duplicated or outdated, they can be overwritten or marked to maintain the validity and usability of the data in the database. After the above processing, the resulting curve feature vector database constitutes the pre-defined historical flood graphic database, which can be used for similar curve matching of subsequent output curves and for water level and flow process extension operations, providing reliable reference data for hydrological forecasting and water level projection.

[0079] Most importantly, the flow rate value calculated from the collected measured flow velocity and cross-sectional area also includes: The collected measured flow velocities are sampled and processed to obtain flow velocity sequence data; The cross-sectional area is measured and calibrated to obtain cross-sectional area sequence data; Dynamic deviation correction product calculation and smoothing are performed based on velocity sequence data and cross-sectional area sequence data to generate smoothed corrected flow data. The smoothed and corrected traffic data is arranged in a time sequence to generate measured traffic sequence data.

[0080] In one embodiment, a flow velocity measuring device deployed at the water level observation cross-section collects real-time flow velocity data at each cross-section location. Data is collected at preset sampling time intervals (e.g., once every 5 seconds or once per minute) to obtain continuous flow velocity sequence data. The cross-section is then measured and calibrated on-site, including measuring the cross-section width, water depth, and cross-section profile. Based on the measurement results, the effective cross-sectional area is calculated to form a cross-sectional area sequence data. For irregular cross-sections, the cross-section can be divided into several small unit areas to measure the area and flow velocity separately, and then a comprehensive calculation is performed.

[0081] Based on the collected velocity and cross-sectional area sequence data, a dynamic deviation correction method is used to calculate instantaneous flow. Specifically, this includes: outlier removal and mean smoothing of the velocity sequence; seasonal or water level variation correction for the cross-sectional area sequence; and multiplying the corrected velocity and cross-sectional area hourly or point-by-point to obtain instantaneous flow data. The instantaneous flow data is then smoothed to obtain smoothed corrected flow data. The smoothed corrected flow data calculated at each time point are arranged in time series to form a complete measured flow sequence, which is used for subsequent water level-flow relationship curve fitting and flood thrust analysis.

[0082] Of particular importance, the calculation of the dynamic deviation correction product based on velocity sequence data and cross-sectional area sequence data also includes: Perform local trend analysis on velocity sequence data to generate velocity trend correction factor data; Perform differential calibration on the cross-sectional area sequence data to generate area correction factor data; Based on the velocity trend correction factor data and the area correction factor data, the dynamic deviation correction product of velocity and cross-sectional area is calculated to generate instantaneous flow rate data. Short-term smoothing filtering is applied to instantaneous flow data to obtain smoothed and corrected flow data.

[0083] In one embodiment, local trend analysis is performed on the collected flow velocity sequence data. Specifically, in the continuous flow velocity sequence, the average rate of change and standard deviation within each window are calculated using a sliding window (e.g., 3 to 5 sampling points) to generate flow velocity trend correction factor data, which is used to correct for local abrupt changes or short-term abnormal fluctuations.

[0084] Micro-differential calibration is performed on the cross-sectional area sequence data, including correcting for minor area deviations caused by water level changes or measurement errors, and generating area correction factor data. Micro-differential calibration can be performed by fitting a standard function of cross-sectional area changes with water level to historical measurement data, and then correcting for deviations between real-time area data and the standard function.

[0085] By utilizing velocity trend correction factor data and area correction factor data, point-by-point product calculations are performed on the velocity sequence and cross-sectional area sequence to achieve dynamic deviation correction and generate instantaneous flow data. Specifically, the instantaneous flow at each sampling point is equal to the velocity at that moment multiplied by the cross-sectional area at that moment, and then multiplied by the corresponding correction factor.

[0086] To eliminate short-term noise fluctuations, short-term smoothing filtering is performed on instantaneous flow data. Methods such as moving average, weighted average, or exponential smoothing can be used to obtain smoothed and corrected flow data, ensuring that the generated measured flow sequence reflects the real flow changes and has high stability and reliability.

[0087] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0088] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A method for intelligently extending the water level-discharge relationship based on similar floods, characterized in that, Includes the following steps: Step S1: Obtain measured flow rate and water level process data; fit the measured flow rate and water level process data to a water level-flow rate relationship line based on the measured water level point and flow rate point to obtain the fitted water level-flow rate relationship curve; Step S2: Determine whether the fitted water level-discharge relationship curve passes the three-property test. If yes, mark the fitted water level-discharge relationship curve as the output curve; otherwise, perform multi-point segmented fitting on the fitted water level-discharge relationship curve to obtain the output curve after secondary fitting. Step S3: Use a preset historical flood graph library to filter similar curves that match the output curve, and extend the water level-discharge process of the output curve through similar curves to obtain the extended water level-discharge relationship curve; Step S4: Perform water level propagation based on the collected real-time water level using the extended water level-flow relationship curve.

2. The intelligent extension method for water level-discharge relationship based on similar floods according to claim 1, characterized in that, Step S2 involves performing a multi-point piecewise fitting of the fitted water level-discharge relationship curve, including: The fitted water level-discharge relationship curve is divided into several continuous water level segments according to the slope change characteristics of the water level process, and the measured water level point and discharge point in each segment are used as the feature point set of that segment. For any water level section, the first difference feature of the water level section is obtained based on the difference between the flow value of each characteristic point in the section and the mean flow value of all characteristic points in the section. The second difference feature of the water level section is obtained based on the difference between the water level slope of each feature point in the section and the mean of the water level slope of all feature points in the section. Based on the first and second difference features, determine the piecewise fitting curve for this water level section; Connect the piecewise fitted curves of all water level sections in order of water level to obtain the output curve after secondary fitting.

3. The intelligent extension method for water level-discharge relationship based on similar floods according to claim 2, characterized in that, Based on the first and second difference characteristics, the piecewise fitting curves for this water level segment are determined as follows: Based on the first and second difference features, the largest first difference feature point, the largest second difference feature point, and the remaining feature points within the water level range are selected. Based on the largest first difference feature point and the largest second difference feature point, a piecewise curve profile is established; By fitting the remaining feature points with the piecewise curve profile, a piecewise fitted curve for the water level section is obtained.

4. The intelligent extension method for water level-discharge relationship based on similar floods according to claim 3, characterized in that, The process of fitting curve points to the remaining feature points using a piecewise curve profile includes: Determine the number of feature points of the remaining feature points. If the number of remaining feature points is greater than the preset number of fitting standards, then merge the adjacent remaining feature points so that the number of feature points is equal to the preset number of fitting standards. When the number of remaining feature points is less than the preset number of fitting standards, a preset compensation point is inserted based on the interval between adjacent remaining feature points, so that the number of feature points is equal to the preset number of fitting standards.

5. The intelligent extension method for water level-discharge relationship based on similar floods according to claim 2, characterized in that, Step S2 determines whether the fitted water level-discharge relationship curve passes the three tests, specifically including: sign test, line fit test, and deviation from numerical value test.

6. The intelligent extension method for water level-discharge relationship based on similar floods according to claim 1, characterized in that, Step S3 includes the following steps: Step S31: Use the water level range of the output curve as the initial search boundary, and filter similar curves that match the output curve from the preset historical flood graphic library; Step S32: Determine the extension segment of the output curve based on the curve shape of similar curves and perform smoothing to obtain the smoothed curve extension segment; Step S33: Extend the water level-flow rate process of the output curve by extending the curve extension segment to obtain the extended water level-flow rate relationship curve.

7. The intelligent extension method for water level-discharge relationship based on similar floods according to claim 6, characterized in that, Step S31 Includes the following steps: Step S311: Use the water level range of the output curve as the initial search boundary; Step S312: Based on the initial search boundary, filter similar curves that match the output curve from the preset historical flood graphic library. If the match is successful, output the similar curve directly. Step S313: If the matching fails, the matching range is expanded cyclically with a preset step size until the matching is successful and a similarity curve is output. The number of expansions and the corresponding similarity change data are recorded.

8. The intelligent extension method for water level-discharge relationship based on similar floods according to claim 6, characterized in that, Step S33 is followed by: Real-time water level data is collected. Confirm the water level range based on real-time water level; If the maximum or minimum value of the water level range does not belong to the extended water level-discharge relationship curve, then repeat steps S31-S33.

9. The intelligent extension method for water level-discharge relationship based on similar floods according to claim 1, characterized in that, Methods for obtaining measured water level and flow rate points include: Acquire station location data; deploy water level observation devices based on station location data, and collect water level data at preset time intervals to obtain a continuous measured water level sequence as the measured water level point; Obtain the location data of the water level observation section; deploy the flow measurement device based on the location data of the water level observation section, and calculate the flow value by collecting the measured flow velocity and cross-sectional area to form the corresponding measured flow sequence as flow point.

10. The intelligent extension method for water level-discharge relationship based on similar floods according to claim 1, characterized in that, The pre-defined methods for constructing a historical flood graphic library include: Obtain historical water level and flow rate curve data; Extract curve features from historical water level and flow rate curve data to construct a curve feature vector library; The curve feature vector library is updated in real time to obtain a preset historical flood graphic library.

Images