A method for reversing incoming flow based on flow trend feature recognition

Abstract

A method for reversing incoming flow correction based on flow trend feature identification, the method constructs a flow trend feature library and obtains the reservoir water level-capacity curve and the period water level and outflow process; reversing the period incoming flow process based on the latest period water level process and ternary the incoming flow trend process; performing feature library identification and matching on the incoming flow trend process; identifying and extracting the corresponding characteristic water level process and dynamically setting the water level correction constraint value; performing error correction on the water level in a specific period based on the least squares method; and finally achieving the overall correction of the reverse inflow process by rolling extraction and correction of the characteristic water level process period by period. The method is based on flow trend feature identification, has strong versatility, does not introduce subjective weight coefficients, and can eliminate the sawtooth fluctuation of the reverse inflow by only correcting the water level error in a specific period, while retaining the original data as much as possible.

Description

Technical Field

[0001] The invention belongs to the technical field of reservoir inflow correction, and in particular relates to a reverse inflow correction method based on flow trend feature identification. Background Art

[0002] Reservoir inflow is an important basic data to support reservoir flood forecasting and scientific scheduling, but there is generally no measured data on reservoir inflow. The measured reservoir inflow is calculated based on the water balance principle, from the changes in reservoir capacity (calculating the reservoir capacity based on the measured water level and the water level-reservoir capacity relationship curve) and the total outflow.

[0003] Theoretically, the reservoir inflow process is a continuous and smooth curve. The reverse calculation method has a water level error amplification effect. Due to the influence of factors such as water level observation errors and wind and wave influences, the reverse calculated reservoir inflow will inevitably show sawtooth fluctuations. Error correction is required to ensure the accuracy and reliability of the measured reservoir inflow, providing a more solid foundation for reservoir flood forecasting and scientific scheduling.

[0004] The existing methods for reverse inflow correction mainly include two types: flow correction method and water level correction method. Flow correction methods mainly include least square method, sliding average method, etc. Such correction methods cannot directly meet the water balance, and require additional secondary distribution of flow. Moreover, the correction of peak flow often requires the introduction of weight coefficients, which is subjective. The method based on water level correction objectively meets the water balance (water balance does not need to be considered during the correction process), but generally corrects the overall water level process once by establishing a global objective function. The correction of the overall water level process does not conform to the relative independence of the time period of the water level error. Setting a global objective function may lead to excessive correction of the overall flow process (especially the peak flow), and it is also impossible to correct the single sawtooth small flow fluctuations caused by the critical mutation of the water level during the dry season.

[0005] Based on the above situation, this application has developed a reverse inflow flow correction method based on flow trend feature identification. Summary of the invention

[0006] In view of the above shortcomings in the prior art, the present invention provides a method for correcting the reverse inflow flow based on flow trend feature identification. The method is based on flow trend feature identification, has strong versatility, does not introduce subjective weight coefficients, and can eliminate the sawtooth fluctuations of the reverse inflow flow by only correcting the water level error in a specific period of time while retaining the original data as much as possible.

[0007] The present invention is solved by the following technical solutions.

[0008] A method for reversing incoming flow based on flow trend feature recognition includes the following steps:

[0009] S10: Obtain the reservoir water level-storage capacity curve and the period water level and outflow flow process;

[0010] S20: Building a traffic trend feature library :

[0011]

[0012]

[0013]

[0014]

[0015] in, It represents a feature library built based on the “‌↗↘↗” or “‌↘↗↘” sawtooth fluctuations. Indicates the feature library constructed based on the mutation in the “→‌↗↘→” or “‌→↘↗→” period;

[0016] S30: Reverse the period of inflow flow process :

[0017]

[0018]

[0019]

[0020] in and Respectively represent The average inbound and outbound flow in each period, in m 3 / s;

[0021] Indicates The reservoir capacity at the end of each period, in 10,000 m 3 ;

[0022] Indicates The reservoir water level at the end of each period, in m, can be calculated by combining the water level-reservoir capacity curve interpolation function Calculate reservoir capacity ;

[0023] S40: Three-valued inbound flow trend process :

[0024]

[0025]

[0026]

[0027] in express The threshold of the three-valued inbound flow trend at each moment is generally set to Flow error caused by +0.001m water level error at the moment, unit is m 3 / s;

[0028] S50: Identification Whether match:

[0029]

[0030]

[0031] in Indicates For the start period, For the end period Select trend feature process, ;

[0032] Indicates the length of the feature library series for identification matching;

[0033] Indicates the traffic trend feature recognition result. express and Match, where express and match, express and match; express and No match;

[0034] like , then extract The characteristic water level process corresponding to the time period, dynamically setting the water level correction constraint value and performing water level error correction for a specific time period, and then performing steps S30, S40, and S50;

[0035] like , then determine whether the traffic trend feature recognition is completed. If the recognition is completed, end the process; if the recognition is not completed, proceed to steps S30, S40, and S50.

[0036] Specifically, in step S50 of the present application:

[0037] like , then extract Characteristic water level process corresponding to the time period :

[0038]

[0039] Specifically, in step S50 of the present application:

[0040] Dynamically set the water level correction constraint value:

[0041]

[0042] in express The water level correction constraint value corresponding to the time period, in meters.

[0043] Specifically, in step S50 of the present application:

[0044] Water level error correction for a specific period of time:

[0045]

[0046]

[0047] in It represents the characteristic water level process after the error correction of the water level process in a specific period;

[0048] Represents the result calculated based on the least squares method The water level error value that needs to be corrected during the time period is in meters.

[0049] Compared with the prior art, the present invention has the following beneficial effects: it provides a method for correcting the reverse inflow flow based on flow trend feature identification. The method is based on flow trend feature identification, has strong versatility, does not introduce subjective weight coefficients, and can eliminate the sawtooth fluctuations of the reverse inflow flow by only correcting the water level error in a specific period of time while retaining the original data as much as possible. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a schematic diagram of the flow chart of the reverse inflow flow correction method in the present invention.

[0051] Figure 2 It is a diagram of the measured inflow flow (solid line) and the corrected inflow flow (dashed line) in Example 1.

[0052] Figure 3 for Figure 2 A zoomed-in view of part of the time period. DETAILED DESCRIPTION

[0053] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0054] In the following embodiments, the embodiments described with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0055] See also Figure 3 The present invention relates to a method for reversing incoming flow based on flow trend feature recognition, comprising the following steps:

[0056] S10: Obtain the reservoir water level-storage capacity curve and the period water level and outflow flow process;

[0057] S20: Building a traffic trend feature library :

[0058]

[0059]

[0060]

[0061]

[0062] in, It represents a feature library built based on the “‌↗↘↗” or “‌↘↗↘” sawtooth fluctuations. Indicates the feature library constructed based on the mutation in the “→‌↗↘→” or “‌→↘↗→” period;

[0063] S30: Reverse the period of inflow flow process :

[0064]

[0065]

[0066]

[0067] in and Respectively represent The average inbound and outbound flow in each period, in m 3 / s;

[0068] Indicates The reservoir capacity at the end of each period, in 10,000 m 3 ;

[0069] Indicates The reservoir water level at the end of each period, in m, can be calculated by combining the water level-reservoir capacity curve interpolation function Calculate reservoir capacity ;

[0070] S40: Three-valued inbound flow trend process :

[0071]

[0072]

[0073]

[0074] in express The threshold of the three-valued inbound flow trend at each moment is generally set to The absolute value of the flow error caused by the +0.001m water level error at the moment, in m 3 / s;

[0075] S50: Identification Whether match:

[0076]

[0077]

[0078] in Indicates For the start period, For the end period Select trend feature process, ;

[0079] Indicates the length of the feature library series for identification matching;

[0080] Indicates the traffic trend feature recognition result. express and Match, where express and match, express and match; express and No match;

[0081] like , then extract Characteristic water level process corresponding to the time period , Dynamically set water level correction constraint value And perform water level error correction for a specific period of time, and then proceed to steps S30, S40, and S50;

[0082] in, ;

[0083] Among them, the water level correction constraint value is dynamically set as follows:

[0084]

[0085] Among them, the water level error correction for a specific period is as follows:

[0086]

[0087]

[0088] in It represents the characteristic water level process after the error correction of the water level process in a specific period; Represents the result calculated based on the least squares method The water level error value that needs to be corrected during the period, in m;

[0089] like , then determine whether the traffic trend feature recognition is completed. If the recognition is completed, end the process; if the recognition is not completed, proceed to steps S30, S40, and S50.

[0090] The reverse inflow process affected by factors such as water level observation errors and wind and wave influences often shows a "‌↗↘↗" or "‌↘↗↘" sawtooth fluctuation characteristic that violates the natural flow law; the reverse inflow process affected by the critical change of water level in the dry season often shows a "→‌↗↘→" or "‌→↘↗→" period mutation characteristic that violates the natural flow law.

[0091] The method in this application only makes effective and targeted corrections to the water level errors in specific periods that do not conform to the natural inflow process, which satisfies the relative independence of the water level errors in the time periods, objectively satisfies the water balance, and avoids excessive correction of the water level series. At the same time, this method can also correct the single sawtooth small flow fluctuations caused by the critical change of the water level in the dry season, and has certain guiding significance for estimating the inflow flow and evaporation in the dry season.

[0092] Specifically, the method in this application considers the regular characteristics of water level errors and the relative independence of time periods, and proposes a method for reverse inflow correction based on flow trend feature identification by implementing a water level correction method based on a limited time period and range. The method constructs a flow trend feature library and obtains the reservoir water level-storage capacity curve and the time period water level and outflow process; based on the latest time period water level process, the time period inflow process is reversed and the inflow flow trend process is ternary; the inflow flow trend process is identified and matched by the feature library; the corresponding characteristic water level process is identified and extracted, and the water level correction constraint value is dynamically set; based on the least squares method, only the water level in a specific time period is corrected for errors; by rolling extraction and correction of the characteristic water level process time by time period, the overall correction of the reverse inflow flow process is finally achieved.

[0093] The following is a specific example 1 of this application.

[0094] The measured inflow process of Zhougongzhai Reservoir, the "highest arch dam in East China", during the 2022 Super Typhoon No. 11 Xuanlannuo was selected. Figure 1 As shown by the solid line, we can see that the flow process is obviously jagged and has negative values, which does not conform to the law of natural flow process.

[0095] The specific steps of the reverse inflow flow correction method based on flow trend feature identification provided by the present invention are as follows:

[0096] Step 1: Obtain the water level-storage capacity curve of Zhougongzhai Reservoir and the water level and outflow process for a total of 155 periods from 08:00 on August 29, 2022 to 18:00 on September 6, 2022;

[0097] Step 2: Build a traffic trend feature library , including feature libraries based on “‌↗↘↗” or “‌↘↗↘” sawtooth fluctuations And the feature library constructed based on the mutation of the “→‌↗↘→” or “‌→↘↗→” period ;

[0098] Step 3: Reverse calculation of the inflow process for a total of 154 periods from 09:00 on August 29, 2022 to 18:00 on September 6, 2022 The measured flood peak before correction was 178m 3 / s, the measured flood volume entering the reservoir is 29.7 million m 3 ;

[0099] Step 4: Ternaryize the inbound flow trend process for a total of 153 periods from 09:00 on August 29, 2022 to 17:00 on September 6, 2022;

[0100] Step 5: Identify and select trend characteristics Whether match;

[0101] like and If it matches, extract The characteristic water level process corresponding to the time period, dynamically set the water level correction constraint value and perform water level error correction for a specific time period, and then proceed to steps three, four, and five;

[0102] Among them, extraction Characteristic water level process corresponding to the time period as follows:

[0103]

[0104] Among them, the water level correction constraint value is dynamically set as follows:

[0105]

[0106] Among them, the water level error correction for a specific period is as follows:

[0107]

[0108]

[0109] like and If there is no match, it is determined whether the traffic trend feature recognition is complete. If the recognition is complete, the process ends; if the recognition is not complete, proceed to steps 3, 4, and 5;

[0110] The overall correction effect of the reverse inflow flow correction method based on flow trend feature recognition provided by the present invention is as follows: Figure 2 As shown by the dotted line in Figure 3 The flow process after correction is smooth, without unreasonable negative values, which conforms to the law of natural flow process.

[0111] The revised flood peak is 178m 3 / s, and the flood volume entering the reservoir is 29.7 million m 3 , which is consistent with that before correction, indicating that this correction method satisfies the water balance and will not over-correct the flood process (especially the peak flow) without obvious errors.

[0112] The protection scope of the present invention includes but is not limited to the above embodiments. The protection scope of the present invention shall be based on the claims. Any replacement, deformation, and improvement of the technology that can be easily thought of by technicians in this field shall fall within the protection scope of the present invention.

Claims

1. A method for reversing incoming flow based on flow trend feature recognition, characterized in that: The following steps are involved: S10: Obtain the reservoir water level-storage capacity curve and the period water level and outflow flow process; S20: Build a traffic trend feature library; S30: reverse the inflow flow process during the period; S40: three-valued inflow flow trend process; S50: Identify whether the selected trend feature process matches the traffic trend feature library: If the recognition is complete, the process ends; If the recognition is not completed, proceed to steps S30, S40, and S50; The specific steps in step S20 are as follows: S20: Building a traffic trend feature library : ； ； ； ； in, It indicates the feature library built based on the sawtooth fluctuation of "‌↗↘↗" or "‌↘↗↘". Indicates the feature library constructed based on the mutation in the "→‌↗↘→" or "‌→↘↗→" period; The specific steps in step S40 are as follows: S40: Three-valued inbound flow trend process : ； ； ； in express The threshold of the three-valued inbound flow trend at each moment is generally set to The absolute value of the flow error caused by the +0.001m water level error at the moment, in m 3 / s; The specific steps in step S50 are as follows: S50: Identification Whether match: ； ； in Indicates For the start period, For the end period Select trend feature process, ; Indicates the length of the feature library series for identification matching; Indicates the traffic trend feature recognition result. express and Match, where express and match, express and match; express and Mismatch; like , then extract The characteristic water level process corresponding to the time period, dynamically setting the water level correction constraint value and performing water level error correction for a specific time period, and then performing steps S30, S40, and S50; like , then determine whether the traffic trend feature recognition is completed. If the recognition is completed, the process ends; If the recognition is not completed, the steps S30, S40 and S50 are performed again.

2. A method for reversing incoming flow based on flow trend feature recognition according to claim 1, characterized in that: The specific steps in step S30 are as follows: S30: Reverse the period of inflow flow process : ； ； ； in and Respectively represent The average inbound and outbound flow in each period, in m 3 / s; Indicates The reservoir capacity at the end of each period, in 10,000 m 3 ; Indicates The reservoir water level at the end of each period, in m, can be calculated by combining the water level-reservoir capacity curve interpolation function Calculate reservoir capacity .

3. The method for reversing incoming flow based on flow trend feature recognition according to claim 2 is characterized in that: In step S50: like ,Right now and If it matches, extract Characteristic water level process corresponding to the time period : 。 4. The method for reversing incoming flow based on flow trend feature recognition according to claim 3 is characterized in that: In step S50: Dynamically set the water level correction constraint value: ； in express The water level correction constraint value corresponding to the time period, in meters.

5. The method for reversing incoming flow based on flow trend feature recognition according to claim 4 is characterized in that: In step S50: Water level error correction for a specific period of time: ； ； in It represents the characteristic water level process after the error correction of the water level process in a specific period; Represents the result calculated based on the least squares method The water level error value that needs to be corrected during the time period is in meters.

Images