Pump station efficiency characteristic curved surface dynamic updating method and system and storage medium

Abstract

The invention discloses a pump station efficiency characteristic curved surface dynamic updating method and system and a storage medium. The method comprises the following steps: 1) carrying out data processing on a pump station data abnormal value; 2) performing data reorganization on the pump station data, and dividing a sliding window data set in a time dimension; 3) establishing a relationship between the original pump station efficiency characteristic curve and actually measured pump station efficiency characteristic monitoring data by using a linear equation, and performing linear translation to obtain a corrected pump station efficiency characteristic curved surface; 4) according to an error minimization principle, selecting an optimal learning rate of the original pump station efficiency characteristic curved surface and the corrected pump station efficiency characteristic curved surface, and obtaining a comprehensive corrected pump station efficiency characteristic curved surface under the optimal learning rate; 5, historical measured data are regularly utilized, and the pump station efficiency characteristic curved surface is dynamically renewed.According to the method, on the time dimension, through dynamic rolling renewal, the influence that the pump station efficiency characteristic is not obvious due to the fact that the time span of a data set is too long is reduced, and the simulation accuracy of the pump station efficiency is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of optimized operation of pump station units, and in particular relates to a method, system and storage medium for dynamically updating an efficiency characteristic surface of a pump station. Background Art

[0002] The optimal operation of pumping station units is an important part of water resources planning and management. Accurately carrying out pumping station efficiency simulation can better support tasks such as water resources management, water volume scheduling, reducing pumping station energy consumption and reducing water delivery costs. Existing pumping station efficiency simulation technology focuses on data-driven theory or conceptual mechanism models for simulation, which does not deeply explore the efficiency characteristics of pumping stations and ignores the hidden time effects in the efficiency characteristics of pumping stations. In addition, due to the long time span of the data set, the efficiency characteristics of pumping stations are not obvious, resulting in poor simulation effects of pumping station efficiency. Therefore, the time effect of the efficiency characteristics of pumping stations is analyzed, and a method and system for dynamically updating the surface of the efficiency characteristics of pumping stations is proposed based on this, which can improve the simulation accuracy. Summary of the invention

[0003] The purpose of the present invention is to provide a method and system for dynamically updating the efficiency characteristic surface of a pumping station in view of the deficiencies in the prior art, so as to improve the simulation accuracy.

[0004] In order to achieve the above technical objectives, the present invention adopts the following technical solutions.

[0005] The present invention first provides a method for dynamically updating a pump station efficiency characteristic surface, comprising the following steps:

[0006] Step S1, performing data processing on the abnormal values ​​of the collected pump station data;

[0007] Step S2, reorganizing the pump station data and dividing the sliding window data set in the time dimension;

[0008] Step S3, using a linear equation to establish a relationship between the original pumping station efficiency characteristic curve and the measured pumping station efficiency characteristic monitoring data, linearly shifting the original pumping station efficiency characteristic curve surface to obtain a corrected pumping station efficiency characteristic curve surface;

[0009] Step S4, according to the error minimization principle, selecting the optimal learning rate of the original pumping station efficiency characteristic surface and the corrected pumping station efficiency characteristic surface, and obtaining the comprehensive corrected pumping station efficiency characteristic surface under the optimal learning rate;

[0010] Step S5: regularly use historical measured data to dynamically update the pump station efficiency characteristic surface.

[0011] Furthermore, in step S1, the collected abnormal values ​​of the pump station data are processed, and the data processing process is as follows:

[0012] The collected pump station data include: head, flow, efficiency and time;

[0013] The data limits of head, flow and efficiency of different pumping stations are different. The abnormal value data exceeding the threshold are deleted. The mathematical expression is:

[0014] H={H i |H min ≤H i ≤H max};

[0015] Q={Q i |Q min ≤Q i ≤Q max};

[0016] E={E i |E min ≤E i ≤E max};

[0017] In the above formula, H is the effective head data after deleting the threshold; H min H is the lower limit of pump station head data; max is the upper limit of the pump station head data; Q is the effective flow data after deleting the threshold; Q min is the lower limit of the pump station flow data; Q max is the upper limit of the pump station flow data; E is the effective efficiency data after deleting the threshold; E min E is the lower limit of the efficiency data of the pump station unit; max It is the upper limit of the efficiency data of the pump station unit;

[0018] After inspection, troubleshooting and correction, the measured data of the pump station is recorded as follows:

[0019] Head: H'={H1,H2,...,H t-1 ,H t ,...};

[0020] Flow rate: Q' = {Q1, Q2, ..., Q t-1 ,Q t ,...};

[0021] Efficiency: E'={E1,E2,...,E t-1 ,E t ,...}.

[0022] Furthermore, the step S2 of reorganizing the pump station data and dividing the sliding window data set in the time dimension includes the following steps:

[0023] Step S21: The lift, data and efficiency data are collected by different collection terminals. The time information of the three is not one-to-one. Arrange the data in chronological order, perform a left join on the data table, use time as the index, find the closest timestamp in each data table, associate all data into one table, and the maximum time difference tolerance is T. max , re-sort the associated data table with time as the index to obtain the required data table:

[0024]

[0025]

[0026] In the above formula, D1 is the efficiency data of the pump station unit; D2 is the head data of the pump station; D3 is the flow data of the pump station; D is the data after the head, flow and efficiency data are combined and re-arranged according to time; the time difference of each set of head, flow and efficiency data is less than the maximum time difference tolerance T max ;

[0027] Step S22: re-index the merged data set, that is, D = {(t1, x1), (t2, x2), ..., (t n ,x n )}, the sliding window data set is recorded as: W k ={(t k ,x k ),(t k+1 ,x k+1 ),...,(t k+W-1 ,x k+W-1 )};

[0028] The sliding window data set is represented as:

[0029] W i ={(t i+j ,x i+j )|j=0,1,2,...,m-1};

[0030] i=1,1+S,1+2S,...,t i+W-1 ≤t n ;

[0031] In the above formula, i is the sliding window data set number; m is the number of data in the sliding window data set; S is the sliding window step size.

[0032] Furthermore, the process of obtaining the corrected pump station efficiency characteristic surface in step S3 is as follows:

[0033] The actual operation data of the pumping station is interpolated on the original pumping station efficiency characteristic surface, and a linear relationship between the interpolated original pumping station efficiency characteristic surface data and the actual operation data of the pumping station is established by using a linear equation, and the optimal parameters for translating the original pumping station efficiency characteristic surface are found, and the parameter optimization function is selected as the least square method; based on the optimal parameters, the linear equation for converting the original pumping station efficiency characteristic surface into the corrected pumping station efficiency characteristic surface is deduced;

[0034] The interpolation function uses radial basis function interpolation, and the mathematical expression is:

[0035]

[0036] In the above formula, f(x) is the interpolation function, r = || xx i || is the interpolation point x and the data point x i The Euclidean distance between i Determined by solving a system of equations or by the method of least squares;

[0037] The mathematical expression of the least squares method is:

[0038]

[0039] In the above formula, J1 is the parameters A and B related to efficiency obtained by the least square method; is the actually observed pumping station efficiency; The pumping station efficiency is obtained by interpolating the corrected pumping station efficiency characteristic surface through radial basis function;

[0040] The linear equation for converting the original pumping station efficiency characteristic surface into the corrected pumping station efficiency characteristic surface is:

[0041] Corrected pump station efficiency characteristic surface = A × original pump station efficiency characteristic surface + B.

[0042] Further, the process of obtaining the comprehensive correction pump station efficiency characteristic surface under the optimal learning rate in step S4 is as follows:

[0043] The mean absolute error MAE expression is:

[0044]

[0045] The root mean square error RMSE expression is:

[0046]

[0047] The mean square error MSE expression is:

[0048]

[0049] The maximum error MI expression is:

[0050]

[0051] In the above formula, is the actual operating efficiency of the pumping station; is the corrected pump station efficiency;

[0052] The mathematical expression of learning rate is:

[0053]

[0054] The expression of multi-objective weighted error function is:

[0055]

[0056] In the above formula, w1, w2, w3 and w4 are the weights of the error term of the error function;

[0057] The optimal learning rate mathematical expression is:

[0058]

[0059] The optimization goal of the optimal learning rate is to minimize the multi-objective weighted error is the pumping station efficiency after comprehensive correction; J(a) is the optimal learning rate a when the multi-objective weighted error between the actual operating efficiency of the pumping station and the pumping station efficiency after comprehensive correction is minimized;

[0060] The comprehensive corrected pumping station efficiency characteristic surface is a weighted average of the original pumping station efficiency characteristic surface and the corrected pumping station efficiency characteristic surface, thereby obtaining the comprehensive corrected pumping station efficiency characteristic surface.

[0061] Furthermore, in step S5, historical measured data is regularly used to dynamically update the pump station efficiency characteristic surface, and the updating process is as follows:

[0062] The pump station efficiency characteristic curve is updated regularly. The time span of regular update is the sliding window time step S. For each sliding window data set W k , calculate the pump station efficiency characteristic surface E k , denoted as E k =f(W k ); Regularly update the pump station efficiency characteristic surface dynamically, with an update time span of S, and the mathematical expression is:

[0063]

[0064] In the above formula, E k is the kth sliding window W k The updated pump station efficiency characteristic surface calculated above; W kis the kth sliding window dataset; f is the sliding window dataset W k Calculates and updates the function of the pump station efficiency characteristic surface.

[0065] On the basis of the above technical solution, the present invention also provides a system for dynamically updating the efficiency characteristic surface of a pumping station, the system comprising a processor, and the processor is used to execute the steps of the method for dynamically updating the efficiency characteristic surface of a pumping station given in the above technical solution.

[0066] Furthermore, the present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the method for dynamically updating the pump station efficiency characteristic surface given in the above technical solution are implemented.

[0067] Compared with the prior art, the present invention has the following beneficial effects:

[0068] 1. The present invention proposes a method for dynamically updating the surface of pump station efficiency characteristics. In the time dimension, after dynamic rolling update, the time effect in the pump station efficiency characteristics is revealed, thereby improving the simulation accuracy of the pump station efficiency; reducing the impact of the unclear pump station efficiency characteristics due to the long time span of the data set, thereby improving the simulation accuracy of the pump station efficiency.

[0069] 2. The method for dynamically updating the pumping station efficiency characteristic surface proposed in the present invention, based on the dynamic update analysis conclusions of the pumping station efficiency characteristic surface, uses a short time span pumping station data set to establish a dynamic update model for the pumping station efficiency characteristic surface, and then outputs the updated pumping station efficiency characteristic surface each time; the pumping station efficiency characteristic surface dynamic update method has a higher pumping station efficiency simulation accuracy than the existing efficiency simulation accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present disclosure and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying creative work.

[0071] Figure 1 The present invention is a flow chart of a method for dynamically updating a pump station efficiency characteristic surface. DETAILED DESCRIPTION

[0072] In order to facilitate those of ordinary skill in the art to understand and implement the present invention, the following detailed description is given to each step of the method proposed by the present invention, and it should be understood that these embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms also fall within the scope defined by the appended claims of the present application.

[0073] Example

[0074] like Figure 1 As shown, the present invention provides a method for dynamically updating a pump station efficiency characteristic surface, comprising the following steps:

[0075] Step S1, performing data processing on the abnormal values ​​of the collected pump station data;

[0076] Step S2, reorganizing the pump station data and dividing the sliding window data set in the time dimension;

[0077] Step S3, using a linear equation to establish a relationship between the original pumping station efficiency characteristic curve and the measured pumping station efficiency characteristic monitoring data, linearly shifting the original pumping station efficiency characteristic curve surface to obtain a corrected pumping station efficiency characteristic curve surface;

[0078] Step S4, according to the error minimization principle, selecting the optimal learning rate of the original pumping station efficiency characteristic surface and the corrected pumping station efficiency characteristic surface, and obtaining the comprehensive corrected pumping station efficiency characteristic surface under the optimal learning rate;

[0079] Step S5: regularly use historical measured data to dynamically update the pump station efficiency characteristic surface.

[0080] Example 1

[0081] In this embodiment, four units in a certain river section are used to carry out dynamic update simulation of pump station efficiency characteristic surface. The specific implementation process steps are as follows:

[0082] Step S1, performing data processing on the abnormal values ​​of the collected pump station data;

[0083] The pump station data used include: head, flow rate, efficiency and time, etc.; the pump station data used in this embodiment include head, flow rate, efficiency and time.

[0084] The data limits of head, flow and efficiency of different pumping stations are different. The data limits of pumping stations should be clearly defined to avoid the impact of erroneous data on the model. Due to the aging of the equipment at the collection end or the influence of factors such as nature, the data of head, flow and unit efficiency are prone to fluctuations and exceed the data limit range. It is necessary to delete the abnormal value data exceeding the threshold. The mathematical expression is:

[0085] H'={H i |H min ≤H i ≤H max};

[0086] Q'={Q i |Q min ≤Q i ≤Q max};

[0087] E'={E i |E min ≤E i ≤E max};

[0088] In the above formula, H' is the effective head data after deleting the threshold; H min H is the lower limit of pump station head data; max is the upper limit of the pump station head data; Q' is the effective flow data after deleting the threshold; Q min is the lower limit of the pump station flow data; Q max is the upper limit of the pump station flow data; E' is the effective efficiency data after deleting the threshold; E min E is the lower limit of the efficiency data of the pump station unit; max It is the upper limit of the efficiency data of the pump station unit;

[0089] After inspection, troubleshooting and correction, the measured data of the pump station is recorded as follows:

[0090] Head: H'={H1,H2,...,H t-1 ,H t ,...};

[0091] Flow rate: Q' = {Q1, Q2, ..., Q t-1 ,Q t ,...};

[0092] Efficiency: E'={E1,E2,...,E t-1 ,E t ,...}.

[0093] The pump station data used in this embodiment are:

[0094] The data set for Unit 1 is the actual operating data from November 16, 2022 to March 9, 2024;

[0095] The data set for Unit 2 is the actual operating data from November 23, 2022 to March 26, 2024;

[0096] The data set for Unit 3 contains actual operating data from November 16, 2022 to March 26, 2024;

[0097] The data set for Unit 4 contains actual operating data from December 3, 2022 to March 26, 2024;

[0098] The actual working condition data table of the pump station is shown in Table 1 below; the limit range of the pump station data is: pump station head range: 0≤H≤4.1m; pump station flow range: 18.5≤Q≤51.4m 3 / s; Pump station unit efficiency range: 0≤E≤100%; The available data of the pump station after eliminating abnormal values ​​are head H', flow rate Q' and efficiency E'.

[0099] Table 1. Actual working condition data of Pizhou Station

[0100]

[0101] Step S2, reorganizing the pump station data and dividing the sliding window data set in the time dimension;

[0102] The lift, flow and efficiency data are collected by different collection terminals, and their time information does not correspond one to one. If the lengths of the three data are different, the data will be arranged in chronological order based on the data type with less data volume, and the data table will be left-joined. Using time as the index, the closest timestamp in each data table is found, and all data are associated into one table. The maximum time difference tolerance is T max , re-sort the associated data table with time as the index to obtain a usable data table. Its mathematical expression is:

[0103]

[0104] In the above formula, D1 is the efficiency data of the pump station unit; D2 is the head data of the pump station; D3 is the flow data of the pump station; D 123 The data of head, flow and efficiency are combined and re-arranged according to time. The time difference of each set of head, flow and efficiency data is less than the maximum time difference tolerance T. max ;

[0105] Reindex the merged data set, that is, D = {(t1,x1),(t2,x2),...,(t n ,x n )}, the sliding window data set is recorded as: W k ={(t k ,x k ),(t k+1 ,x k+1 ),...,(t k+W-1 ,x k+W-1 )};

[0106] The sliding window data set expression is:

[0107] W i ={(t i+j ,x i+j )|j=0,1,2,…,m-1};

[0108] i=1,1+S,1+2S,…,t i+W-1 ≤t n ;

[0109] In the above formula, i is the number of sliding window data sets; m is the number of data in the sliding window data set; S is the sliding window step size;

[0110] When the pump station data used in this embodiment are merged, the maximum time difference tolerance is T max = 10min, the merged data set is D = {(t1, e1, h1, q1), (t2, e2, h2, q2), … (t n ,e n ,h n ,q n )}; The pump station division sliding window data set is divided into six modes, which are six time spans. The specific modes are shown in Table 2 below.

[0111] Table 2. Dynamic update mode of pump station efficiency characteristic surface

[0112]

[0113] Step S3, using a linear equation to establish a relationship between the original pumping station efficiency characteristic surface and the measured pumping station efficiency characteristic monitoring data, linearly translating the original pumping station efficiency characteristic surface to obtain a corrected pumping station efficiency characteristic surface;

[0114] The actual operation data of the pumping station is interpolated on the original pumping station efficiency characteristic surface, and the linear relationship between the interpolated original pumping station efficiency characteristic surface data and the actual operation data of the pumping station is established using a linear equation to find the optimal parameters for translating the original pumping station efficiency characteristic surface. The parameter optimization function is selected as the least squares method. Based on the optimal parameters, the linear equation for converting the original pumping station efficiency characteristic surface into the corrected pumping station efficiency characteristic surface is deduced.

[0115] The interpolation function uses radial basis function interpolation, and the kernel function uses the inverse quadratic kernel function. This kernel function is suitable for data interpolation problems that require smoothing and have good boundary conditions. Its mathematical expression is:

[0116]

[0117] In the above formula, f(x) is the interpolation function, r = || xx i || is the interpolation point x and the data point x iThe Euclidean distance between ,ò is determined by solving a system of equations or by the least squares method;

[0118] The objective function of parameter optimization uses the least squares method, and its mathematical expression is:

[0119]

[0120] In the above formula, J1 is the parameters A and B related to efficiency obtained by the least square method; is the actual observed pumping station efficiency, The pumping station efficiency is obtained by interpolating the corrected pumping station efficiency characteristic surface through radial basis function;

[0121] The linear equation for converting the original pumping station efficiency characteristic surface into the corrected pumping station efficiency characteristic surface is:

[0122] Corrected pump station efficiency characteristic surface = A × original pump station efficiency characteristic surface + B.

[0123] Step S4, according to the error minimization principle, selecting the optimal learning rate of the original pumping station efficiency characteristic surface and the corrected pumping station efficiency characteristic surface, and obtaining the comprehensive corrected pumping station efficiency characteristic surface under the optimal learning rate;

[0124] The mean absolute error MAE expression is:

[0125]

[0126] The root mean square error RMSE expression is:

[0127]

[0128] The mean square error MSE expression is:

[0129]

[0130] The maximum error MI expression is:

[0131]

[0132] is the actual operating efficiency of the pumping station; is the corrected pump station efficiency;

[0133] The mathematical expression of learning rate is:

[0134] a n+1 =a n +Δa;

[0135]

[0136] The expression of multi-objective weighted error function is:

[0137]

[0138] The normalization method is used to determine the weights of each error term in the multi-objective weighted error function. The weight expression of each error term is:

[0139]

[0140] In the above formula, MAE0 is the actual value of the mean absolute error MAE; MSE0 is the actual value of the mean square error MSE; RMSE0 is the actual value of the root mean square error RMSE; MI0 is the actual value of the maximum error MI.

[0141] The optimal learning rate mathematical expression is:

[0142]

[0143] The learning rate a is iteratively increased until the optimal learning rate is found; the comprehensive correction pumping station efficiency characteristic surface is the weighted average of the original pumping station efficiency characteristic surface and the corrected pumping station efficiency characteristic surface; w1, w2, w3 and w4 are the weights of the error term of the error function; the optimization goal of the optimal learning rate is to minimize the multi-objective weighted error is the efficiency of the pumping station after comprehensive correction; J(a) is the optimal learning rate a when the multi-objective weighted error between the actual operating efficiency of the pumping station and the efficiency of the pumping station after comprehensive correction is minimized, thus obtaining the characteristic surface of the efficiency of the pumping station after comprehensive correction.

[0144] The best learning rate in this embodiment has been verified by experiments to be a=0, that is, the multi-objective weighted error of the fully-corrected pump station efficiency characteristic surface is minimized.

[0145] Step S5: regularly using historical measured data to dynamically update the pump station efficiency characteristic surface;

[0146] The pump station efficiency characteristic surface is updated regularly, and the time span of the regular update is the sliding window time step S; for each sliding window data set W k , calculate the pump station efficiency characteristic surface E k , denoted as E k =f(W k ); Regularly update the pump station efficiency characteristic surface dynamically, with an update time span of S, and the expression is:

[0147]

[0148] In the above formula, E k is the kth sliding window W k The updated pump station efficiency characteristic surface calculated above; W kis the kth sliding window dataset; f is the sliding window dataset W k Calculates and updates the function of the pump station efficiency characteristic surface.

[0149] After each round of update, the sliding window data set is updated once in the time dimension to achieve dynamic rolling update, and the efficiency characteristic surface of the pumping station after each update is output.

[0150] In this embodiment, the pump station efficiency simulation method adopts the dynamic update method of the pump station efficiency characteristic surface, polynomial regression, multivariate linear regression, Gaussian process regression, regression tree, support vector regression and artificial neural network. The pump station station is divided into six modes, which are six time spans. The time range of the data set used for the dynamic rolling update of the six modes of the pump station is shown in Table 3 below.

[0151] Table 3. Time range of dynamic rolling update data sets of six modes at Pizhou Station

[0152]

[0153]

[0154]

[0155]

[0156]

[0157] The following evaluation indicators are used to evaluate the simulation results of the dynamic update method of the pump station efficiency characteristic surface and the results of the traditional efficiency simulation method, so as to obtain an objective analysis.

[0158] The mean absolute error (MAE) reflects the average absolute error between the simulated efficiency and the actual efficiency. The formula is:

[0159] The root mean square error RMSE reflects the discrete degree of simulation efficiency, and its formula is:

[0160]

[0161] The mean square error (MSE) reflects the accuracy level of the model simulation, and its formula is:

[0162]

[0163] The maximum error A reflects the most extreme situation of the model simulation results, and its formula is: A = max(|S i -0 i |)

[0164] In the above formula, S iis the simulation efficiency; i For actual efficiency; is the mean of actual efficiency; N is the number of simulated efficiency data.

[0165] The simulation results of the six modes of the dynamic update method of the pumping station efficiency characteristic surface are compared and analyzed with the simulation results of polynomial regression, multivariate linear regression, Gaussian process regression, regression tree, support vector regression and artificial neural network models, as shown in Table 4 below.

[0166] Table 4. Simulation results of each model method and evaluation results of the test period

[0167]

[0168]

[0169] The simulation results based on the dynamic update method of the pump station efficiency characteristic surface above show that: ① The simulation results based on the dynamic update method of the measured data of the pump station, the two-month data simulation, the weekly update mode, the average absolute error, mean square error, root mean square error, maximum error, a total of four indicators, are mostly better than the previous traditional simulation methods. ② The use of the dynamic update method of the pump station efficiency characteristic surface can significantly reduce the maximum error in the most unfavorable situation, and obtain more ideal simulation results in extreme cases, and it is more feasible and reliable to use it for pump station efficiency simulation.

[0170] The above is only a preferred embodiment of the present invention, and does not limit the present invention in other forms. Any technician familiar with the profession may use the above disclosed technical content to change or modify it into an equivalent embodiment with equivalent changes. However, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the technical solution of the present invention still belongs to the protection scope of the technical solution of the present invention.

Claims

1. A method for dynamically updating a pump station efficiency characteristic surface, characterized in that: The following steps are involved: Step S1, performing data processing on the abnormal values ​​of the collected pump station data; Step S2, reorganizing the pump station data and dividing the sliding window data set in the time dimension; Step S3, using a linear equation to establish a relationship between the original pumping station efficiency characteristic curve and the measured pumping station efficiency characteristic monitoring data, linearly shifting the original pumping station efficiency characteristic curve surface to obtain a corrected pumping station efficiency characteristic curve surface; Step S4, according to the error minimization principle, selecting the optimal learning rate of the original pumping station efficiency characteristic surface and the corrected pumping station efficiency characteristic surface, and obtaining the comprehensive corrected pumping station efficiency characteristic surface under the optimal learning rate; Step S5: regularly use historical measured data to dynamically update the pump station efficiency characteristic surface.

2. A method for dynamically updating a pump station efficiency characteristic surface according to claim 1, characterized in that: In step S1, the collected pump station data abnormal values ​​are processed, and the data processing process is as follows: the collected pump station data includes: head, flow, efficiency and time; The data limits of head, flow and efficiency of different pumping stations are different. The abnormal value data exceeding the threshold are deleted. The mathematical expression is: H={H i |H min ≤H i ≤H max }; Q={Q i |Q min ≤Q i ≤Q max }; And={And i |And min ≤E i ≤E max }; In the above formula, H is the effective head data after deleting the threshold; H min H is the lower limit of pump station head data; max is the upper limit of the pump station head data; Q is the effective flow data after deleting the threshold; Q min is the lower limit of the pump station flow data; Q max is the upper limit of the pump station flow data; E is the effective efficiency data after deleting the threshold; E min E is the lower limit of the efficiency data of the pump station unit; max It is the upper limit of the efficiency data of the pump station unit; After inspection, troubleshooting and correction, the measured data of the pump station is recorded as follows: Head: H'={H1,H2,...,H t-1 ,H t ,...}; Flow rate: Q' = {Q1, Q2, ..., Q t-1 ,Q t ,...}; Efficiency: E'={E1,E2,...,E t-1 ,E t ,...}.

3. A method for dynamically updating a pump station efficiency characteristic surface according to claim 1, characterized in that: The step S2 is to reorganize the pump station data and divide the sliding window data set in the time dimension, including the following steps: Step S21: The lift, data and efficiency data are collected by different collection terminals. The time information of the three is not one-to-one. Arrange the data in chronological order, perform a left join on the data table, use time as the index, find the closest timestamp in each data table, associate all data into one table, and the maximum time difference tolerance is T. max , re-sort the associated data table with time as the index to obtain the required data table: In the above formula, D1 is the efficiency data of the pump station unit; D2 is the head data of the pump station; D3 is the flow data of the pump station; D is the data after the head, flow and efficiency data are combined and re-arranged according to time; the time difference of each set of head, flow and efficiency data is less than the maximum time difference tolerance T max ; Step S22: re-index the merged data set, that is, D = {(t1, x1), (t2, x2), ..., (t n ,x n )}, the sliding window data set is recorded as: W k ={(t k ,x k ),(t k+1 ,x k+1 ),...,(t k+W-1 ,x k+W-1 )}; The sliding window data set is represented as: W i ={(t i+j ,x i+j )∣j=0,1,2,...,m-1}; i=1,1+S,1+2S,...,t i+W-1 ≤t n ; In the above formula, i is the sliding window data set number; m is the number of data in the sliding window data set; S is the sliding window step size.

4. A method for dynamically updating a pump station efficiency characteristic surface according to claim 1, characterized in that: The process of obtaining the corrected pump station efficiency characteristic surface in step S3 is as follows: The actual operation data of the pumping station is interpolated on the original pumping station efficiency characteristic surface, and a linear relationship between the interpolated original pumping station efficiency characteristic surface data and the actual operation data of the pumping station is established using a linear equation, and the optimal parameters for translating the original pumping station efficiency characteristic surface are found. The parameter optimization function is selected as the least square method; based on the optimal parameters, a linear equation for converting the original pumping station efficiency characteristic surface into the corrected pumping station efficiency characteristic surface is deduced; The interpolation function uses radial basis function interpolation, and the mathematical expression is: In the above formula, f(x) is the interpolation function, r = || xx i || is the interpolation point x and the data point x i The Euclidean distance between i Determined by solving a system of equations or by the method of least squares; The mathematical expression of the least squares method is: In the above formula, J1 is the parameters A and B related to efficiency obtained by the least square method; is the actually observed pumping station efficiency; The pumping station efficiency is obtained by interpolating the corrected pumping station efficiency characteristic surface through radial basis function; The linear equation for converting the original pumping station efficiency characteristic surface into the corrected pumping station efficiency characteristic surface is: Corrected pump station efficiency characteristic surface = A × original pump station efficiency characteristic surface + B.

5. A method for dynamically updating a pump station efficiency characteristic surface according to claim 1, characterized in that: The process of obtaining the comprehensive correction pump station efficiency characteristic surface under the optimal learning rate in step S4 is as follows: The mean absolute error MAE expression is: The root mean square error RMSE expression is: The mean square error MSE expression is: The maximum error MI expression is: In the above formula, is the actual operating efficiency of the pumping station; is the corrected pump station efficiency; The mathematical expression of learning rate is: The expression of multi-objective weighted error function is: In the above formula, w1, w2, w3 and w4 are the weights of the error term of the error function; The optimal learning rate mathematical expression is: The optimization goal of the optimal learning rate is to minimize the multi-objective weighted error is the efficiency of the pumping station after comprehensive correction; J(a) is the optimal learning rate a when the multi-objective weighted error between the actual operating efficiency of the pumping station and the efficiency of the pumping station after comprehensive correction is minimized; The comprehensive corrected pumping station efficiency characteristic surface is a weighted average of the original pumping station efficiency characteristic surface and the corrected pumping station efficiency characteristic surface, thereby obtaining the comprehensive corrected pumping station efficiency characteristic surface.

6. A method for dynamically updating a pump station efficiency characteristic surface according to claim 1, characterized in that: In step S5, historical measured data is regularly used to dynamically update the pump station efficiency characteristic surface, and the updating process is as follows: The pump station efficiency characteristic curve is updated regularly. The time span of regular update is the sliding window time step S. For each sliding window data set W k , calculate the pump station efficiency characteristic surface E k , denoted as E k =f(W k ); Regularly update the pump station efficiency characteristic surface dynamically, with an update time span of S, and the mathematical expression is: In the above formula, E k is the kth sliding window W k The updated pump station efficiency characteristic surface calculated above; W k is the kth sliding window dataset; f is the sliding window dataset W k Calculates and updates the function of the pump station efficiency characteristic surface.

7. A system for dynamically updating the efficiency characteristic surface of a pumping station, characterized in that: It includes a processor, which is used to execute the steps of the method for dynamically updating the pump station efficiency characteristic surface as described in claims 1-6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a method for dynamically updating a pump station efficiency characteristic surface as described in any one of claims 1 to 6 are implemented.

Images