Cascade pump station group optimization operation scheme recommendation method and system integrating stable operation

Abstract

The invention discloses a cascade pump station group optimization operation scheme recommendation method and system fusing stable operation, and the method comprises the steps: constructing a mathematical model; analyzing the correlation degree between the working condition parameters and the stability parameters of the unit by using a maximum information coefficient method, and calibrating the working condition parameters according to the correlation degree; a deep neural network model is constructed for each stability parameter, the calibrated working condition parameters of the pump station unit serve as input data, the stability parameters serve as output data, and a mapping relation between the working condition parameters and the stability parameters of the unit is obtained through training so as to establish a working condition-stability relation model; solving the mathematical model, and constructing a scheme library; and respectively calculating stability parameters of the pump station unit under each scheme, and obtaining an optimal operation scheme considering economy and stability according to a parameter comparison result. According to the method, operation stability analysis can be carried out on the schemes in the cascade pump station group optimization operation scheme library, and the optimal operation scheme considering economical efficiency and stability is selected through comprehensive comparison.

Description

Technical Field

[0001] The present invention relates to the technical field of optimized operation schemes for cascade pump station groups, and in particular to a method and system for recommending optimized operation schemes for cascade pump station groups with integrated stable operation. Background Art

[0002] The optimized operation of cascade pump station groups is one of the important measures to ensure the safety of water supply systems and improve the benefits and service levels of water conservancy projects. The optimized operation of cascade pump station groups is conducive to improving the water supply efficiency of pump stations, reducing energy consumption and costs, improving system stability and reliability, and realizing intelligent management, improving the overall operation effect and service level of water supply systems, and promoting the sustainable use of water resources. Taking the reduction of energy consumption and costs as the optimization goal, while considering the operational stability of the pump station group and integrating it into the recommendation of the optimized operation plan for cascade pump station groups, it is conducive to improving the reliability of the water supply system, reducing unit losses and maintenance costs, saving energy and reducing operating costs.

[0003] At present, the formulation of the optimal operation plan for cascade pump station groups generally requires the collection of data related to the operation of cascade pump station groups, such as water flow, water level, power consumption, etc., and the preprocessing and cleaning of the collected data to ensure the accuracy and completeness of the data. Then, based on the engineering characteristics and operation constraints of the cascade pump station group, a mathematical model is established to describe the operation mechanism of the cascade pump station group, and a hydraulic model or a dynamic simulation model is used to describe the hydraulic characteristics and operation process. Finally, according to the optimization objectives, such as minimum energy consumption and maximum transmission efficiency, the corresponding objective function is defined. At the same time, considering the requirements of operational stability, constraints can be introduced, such as the range of water level changes and the number of pump starts and stops. The objective function is solved using the optimization algorithm to obtain the optimal operation plan for the cascade pump station group.

[0004] The existing optimization operation and scheduling model for cascade pump station groups is a single-objective system with the objective function of "minimum operating cost within the scheduling cycle", without considering the effects of unit operation stability (such as pressure pulsation, vibration, spindle swing, etc.) and maintenance costs (number of unit starts and stops). The existing optimization operation and scheduling model for cascade pump station groups is designed according to the theoretical state of the pump station units during the engineering design phase, without considering the effects of actual parameter changes such as the units and the river channel during the operation of the pump station, resulting in the scheduling model losing its effective target and the optimization results seriously deviating from the optimal working conditions. The existing optimization operation and scheduling model for cascade pump station groups is too idealistic in theory, with too detailed time divisions, and the optimization results require frequent angle changes and frequent starts and stops, without considering actual operability, the impact on the stability of equipment units, and the investment in manpower and material maintenance costs. Summary of the invention

[0005] The present invention aims to solve one of the technical problems in the related art at least to a certain extent.

[0006] To this end, the present invention proposes a method for recommending an optimized operation plan for a cascade pump station group that integrates stable operation. In view of the problems existing in the above-mentioned prior art, such as single optimization target, over-idealization, poor operability, failure to consider changes in actual operating parameters, failure to consider the impact of unit operation stability and maintenance costs, a mathematical model for the optimized operation of a cascade pump station group with the goal of minimizing daily water extraction electricity consumption is constructed, a plan library is constructed, and a working condition-stability relationship model is established to comprehensively select the optimal operation plan, which is beneficial to improving the reliability of the water supply system, reducing unit losses and maintenance costs, saving energy, and reducing operating costs.

[0007] Another object of the present invention is to propose a system for recommending optimal operation plans for cascade pump station groups that integrate stable operation.

[0008] To achieve the above-mentioned purpose, the present invention proposes, on one hand, a method for recommending an optimal operation plan for a cascade pump station group integrating stable operation, comprising:

[0009] Taking the minimum daily water-lifting power consumption as the objective function, a mathematical model for the optimal operation of a series cascade pumping station group considering economics is constructed.

[0010] Obtain stability parameters that characterize the operation of the pump station unit, and use the maximum information coefficient method to analyze the correlation between the unit operating parameters and the stability parameters, and rate the operating parameters according to the correlation;

[0011] A deep neural network model is constructed for each stability parameter, and the operating parameters of the pump station unit after calibration are used as input data and the stability parameters are used as output data. The mapping relationship between the unit operating parameters and the stability parameters is obtained through training to establish an operating condition-stability relationship model;

[0012] The large system decomposition-dynamic programming aggregation method is used to solve the mathematical model of the optimal operation of the series cascade pump station group, and combined with the non-steady flow simulation, iterative optimization is performed to obtain multiple groups of economically optimized operation plans for the cascade pump station group, so as to build a plan library;

[0013] The operating condition-stability relationship model is used to calculate the stability parameters of the pump station units under each scheme in the scheme library, and the optimal operation scheme considering economy and stability is obtained based on the comparison results of the stability parameters of the pump station units.

[0014] The method for recommending an optimal operation plan for a cascade pump station group with integrated stable operation according to the embodiment of the present invention may also have the following additional technical features:

[0015] In one embodiment of the present invention, taking the minimum daily water extraction power consumption as the objective function, a mathematical model for optimizing the operation of a series cascade pumping station group considering economic efficiency is constructed, including:

[0016] The objective function is to minimize the total power consumption of daily startup operation, the divided time periods are stage variables, the blade placement angles in each time period are decision variables, the total amount of water pumped in the specified time period and the rated power of the motor are constraints, and a mathematical model for the optimal operation of a single unit is constructed.

[0017] Taking the minimum daily pumping cost of the pumping station as the objective function, the blade angle of each unit at each time period as the decision-making quantity, and the specified water pumping volume and unit power requirements as constraints, a mathematical model for the optimal operation of multiple units is constructed.

[0018] The objective function is to minimize the daily water-lifting power consumption of the cascade pump station group system, the divided time periods are stage variables, the blade placement angle and operating head of each unit at each level of pumping station in each time period are decision variables, the total water lifting volume of each level of pumping station within the specified time period, the rated power of the motor of each unit and the total water lifting head of the cascade pump station group are constraints, and a mathematical model for the optimal operation of the series cascade pump station group is constructed.

[0019] In one embodiment of the present invention, a stability parameter characterizing the operation of a pump station unit is obtained, and the correlation between the unit operating parameters and the stability parameters is analyzed using the maximum information coefficient method, and the operating parameters are determined according to the correlation, including:

[0020] Collecting stability parameters that characterize the operation of the pump station unit; wherein the stability parameters include pressure pulsation, vibration, noise and swing data;

[0021] The pressure pulsation, vibration, noise and swing data are processed respectively to obtain the stability parameters after processing, including:

[0022] The correlation between the processed stability parameter and the operating condition parameter is analyzed based on the maximum information coefficient; wherein the maximum information coefficient is obtained by calculating the maximum value of the mutual information in different grid divisions and normalizing it.

[0023] In one embodiment of the present invention, a deep neural network model is constructed for each stability parameter, and the operating condition parameters of the pump station unit after calibration are used as input data and the stability parameters are used as output data. The mapping relationship between the unit operating condition parameters and the stability parameters is obtained through training to establish an operating condition-stability relationship model, including:

[0024] According to the maximum information coefficient, multiple operating condition parameters with the highest correlation among the stability parameters of the pump station unit are selected as input data;

[0025] Divide the input data into training set and test set;

[0026] Determine the number of neurons in the input layer according to the dimension of the input data and build a deep neural network model:

[0027] The training set is input into the deep neural network model, and the RMSprop optimization algorithm is used to train the deep neural network model to obtain a trained deep neural network model;

[0028] Input the test set into the trained deep neural network model to obtain the network output result, and compare the network output result with the actual output data of the test set to obtain the comparison result, and calculate the mean absolute error, mean absolute percentage error and root mean square error of the comparison result;

[0029] When all error values ​​meet the target requirements, the model training is completed, and a mapping relationship between each operating parameter of the pump station unit and multiple operating parameters with the highest correlation is established to establish an operating condition-stability relationship model.

[0030] In one embodiment of the present invention, a large system decomposition-dynamic programming aggregation method is used to solve the mathematical model of the optimal operation of the series cascade pump station group, and combined with the non-steady flow simulation, iterative optimization is performed to obtain multiple groups of economically optimized operation schemes of the cascade pump station group, so as to construct a scheme library, including:

[0031] The performance curve equations of pumping station devices at each level are obtained by fitting the performance characteristic data of pumping station devices at each level in the mathematical model for optimizing the operation of the series cascade pumping station group;

[0032] Based on the basic data of the water supply channel and the water use status of water users along the inter-level water supply channel, a generalized map of the water supply channel network between every two pumping stations in the cascade pumping station group is constructed;

[0033] Based on the generalized diagram of the water supply channel network, the load water volume when the pump station is running at a fixed angle and constant speed is taken as the optimized target water volume, and the large system decomposition-dynamic programming aggregation method is adopted to calculate the optimal operation of the pump station under the constraints of different water volumes at each daily average head based on the performance curve equation to obtain the optimal operation plan of each level of pump stations;

[0034] Based on the known initial lift and water lifting load of each level of the secondary water lifting pump station group, the water lifting load of the first-level pump station is determined:

[0035] Taking the delayed initial head of the first-stage pumping station as the daily average head, re-taking the optimized operation process corresponding to the matching load under the daily average head as the boundary condition, combining the flow process of the second-stage pumping station and the water use process of the inter-stage water users, one-dimensional unsteady flow simulations under various schemes were performed to obtain the flow process lines and water level process lines of each section of the inter-stage water transfer channel.

[0036] According to the flow process lines and water level process lines of each section of the inter-stage water transfer channel, it is judged whether the initial water level assumptions of the first and second pumping stations are reasonable, and based on the judgment results, an optimized operation plan that meets the requirements is obtained to construct an optimized operation plan for the cascade pumping station group.

[0037] In one embodiment of the present invention, the working condition-stability relationship model is used to calculate the stability parameters of the pump station unit under each scheme in the scheme library, and the optimal operation scheme considering economy and stability is obtained according to the comparison results of the stability parameters of the pump station unit, including:

[0038] Input the working condition parameters into the working condition-stability relationship model to calculate the stability data under each scheme;

[0039] Based on the flow process lines, water level process lines and stability data of each section of the inter-stage water channel, each scheme is calculated and analyzed from the perspectives of head matching, optimized operation economy, water level fluctuation of the inter-stage water channel, flood control and drainage, navigation, ecological water level requirements and unit operation stability to obtain the stability parameters of the pump station units under each scheme;

[0040] The stability parameters of the pump station units under the various schemes are compared to obtain the optimal operation scheme of the cascade pump station group considering economy and stability according to the comparison results.

[0041] To achieve the above-mentioned purpose, the present invention proposes, on the other hand, a system for recommending an optimal operation plan for a cascade pump station group integrating stable operation, comprising:

[0042] The mathematical model building module is used to build a mathematical model for optimizing the operation of a series of cascade pumping stations taking economic factors into consideration, taking the minimum daily water extraction power consumption as the objective function;

[0043] The stability parameter calibration module is used to obtain the stability parameters that characterize the operation of the pump station unit, and use the maximum information coefficient method to analyze the correlation between the unit operating parameters and the stability parameters, and calibrate the operating parameters according to the correlation;

[0044] A stability relationship model building module is used to build a deep neural network model for each stability parameter, and use the calibrated operating parameters of the pump station unit as input data and the stability parameters as output data, and train to obtain a mapping relationship between the unit operating parameters and the stability parameters to establish an operating condition-stability relationship model;

[0045] The optimization operation model solving module is used to solve the mathematical model of the optimal operation of the series cascade pump station group by using the large system decomposition-dynamic programming aggregation method, and to perform iterative optimization in conjunction with the non-steady flow simulation to obtain multiple groups of economically optimized operation plans for the cascade pump station group, so as to build a plan library;

[0046] The comprehensive comparison module for optimized operation is used to calculate the stability parameters of the pump station units under each scheme in the scheme library using the operating condition-stability relationship model, and obtain the optimal operation scheme considering economy and stability based on the comparison results of the stability parameters of the pump station units.

[0047] The embodiment of the present invention integrates the method and system for recommending the optimal operation scheme of the cascade pump station group with stable operation, constructs a mathematical model for the optimal operation of the cascade pump station group with the goal of minimizing the daily water extraction power consumption, adopts the "large system decomposition-dynamic programming aggregation" method to solve the model and construct a scheme library, and integrates a one-dimensional hydraulic model to ensure the head matching degree and water level requirements between pump stations at all levels. At the same time, the maximum information coefficient method is used to calibrate the operating parameters related to the stable operation of the unit, and a deep neural network model is established for the calibrated operating parameters to obtain the mapping relationship between the operating parameters and the unit stability parameters, so as to perform an operation stability analysis on the schemes in the optimal operation scheme library of the cascade pump station group, and comprehensively compare and select the optimal operation scheme considering economy and stability.

[0048] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0050] Figure 1 is a flow chart of a method for recommending an optimal operation plan for a cascade pump station group with integrated stable operation according to an embodiment of the present invention;

[0051] Figure 2 It is an architecture diagram of a method for recommending an optimized operation plan for a cascade pump station group with integrated stable operation according to an embodiment of the present invention;

[0052] Figure 3 It is a structural diagram of a system for recommending an optimal operation plan for a group of cascade pump stations with integrated and stable operation according to an embodiment of the present invention. DETAILED DESCRIPTION

[0053] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0054] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0055] The following describes a method and system for recommending an optimal operation plan for a cascade pump station group with integrated stable operation according to an embodiment of the present invention with reference to the accompanying drawings.

[0056] Figure 2 The architecture diagram of the method for recommending the optimal operation scheme of the cascade pump station group with stable operation integrated with the embodiment of the present invention is as follows. The objective function is to minimize the daily water extraction power consumption cost, and to construct a mathematical model for the optimal operation of the series cascade pump station group considering economy. The pressure pulsation, vibration, noise, swing and other index data that characterize the operational stability of the pump station unit are collected, and the maximum information coefficient (MIC) method is used to analyze the relationship and correlation between the unit operating parameters and the stability parameters, and the operating parameters are calibrated according to the correlation. The operating parameters of the pump station unit after calibration are selected as input data, and the stability parameters such as vibration, swing, noise, and pressure pulsation are selected as output data. A deep neural network (DNN) model is constructed for each stability parameter. After training, the mapping relationship between the unit operating parameters and the stability parameters is obtained, that is, a "operating condition-stability relationship" model is established. The large system decomposition-dynamic programming aggregation method is used to solve the constructed mathematical model for the optimal operation of the series cascade pump station group, and the non-steady flow simulation is combined for iterative optimization to obtain multiple groups of optimal operation schemes for the cascade pump station group considering economy, and a scheme library is constructed. The established "operating condition-stability relationship" model is used to calculate the specific stability parameters of the pump station units under each scheme in the scheme library, and a comprehensive comparison is made based on the head matching, optimized operation economy, water level fluctuation of the inter-stage water supply channel, flood control and drainage, navigation, and ecological water level requirements, to give the optimal operation scheme that takes into account economy and stability. Figure 2 shown.

[0057] Figure 1 It is a flow chart of a method for recommending an optimized operation plan for a cascade pump station group with integrated stable operation according to an embodiment of the present invention.

[0058] like Figure 1 As shown, the method includes but is not limited to the following steps:

[0059] S1, taking the minimum daily water-lifting electricity cost as the objective function, constructs a mathematical model for the optimal operation of a series of cascade pumping stations considering economy.

[0060] Specifically, step S1 includes the following steps:

[0061] S11, considering the peak and valley electricity prices, water level fluctuations, etc., 1d (24h) is divided into SN segments to study the optimization operation rules of single units in the pump station. The objective function is to minimize the total power consumption cost of daily startup operation, the time periods divided by 1d are stage variables, the blade placement angles in each time period are decision variables, the total amount of water lifted within the specified time period and the rated power of the motor are constraints, and a mathematical model for the optimization operation of a single unit is constructed:

[0062] Objective function:

[0063]

[0064] Total water volume constraint:

[0065]

[0066] Power Constraints:

[0067] N i (θ i )≤N0

[0068] Among them, f is the minimum power consumption cost of a single unit for 1 day, yuan; s i is the electricity consumption cost of the unit in the i-th period, yuan; ρ is the water density, kg / m 3 ; g is the acceleration due to gravity, m / s 2 ;H i is the average head in the i-th period, m; Q i (θ i ) is the pump flow rate in the i-th period, m 3 / s, average head per day H i When the blade is placed at a certain angle, i The function of Q i (θ i )=0 means shutdown at this stage; ΔT i is the length of the i-th period, h; P i is the electricity price in the i-th period, yuan / (kW·h); W j is the total daily water extraction volume of a single unit, m 3 ; η z，i (θ i ) is the efficiency of the water pump device, which is related to the flow rate and head in the i-th period; η mot is the motor efficiency. When the load is greater than 60%, it can be considered to be basically unchanged. The efficiency of large motors is around 94%; η int is the transmission efficiency, the transmission efficiency of the direct-connected unit is 100%; η f The frequency conversion efficiency of high-power and high-voltage inverters is about 96%. i (θ i ) is the actual power corresponding to the blade placement angle in the i-th period and should be less than or equal to the rated power N0 of the motor.

[0069] S12, taking a single pump station including JZ units as the research object, the optimal operation of multiple units in the pump station was studied. The objective function was to minimize the daily water extraction cost of the pump station, the blade angle of each unit at each time period was the decision-making quantity, and the specified water extraction volume and unit power requirements were the constraints. A mathematical model for the optimal operation of multiple units was constructed:

[0070] Objective function:

[0071]

[0072] Total water volume constraint:

[0073]

[0074] Power Constraints:

[0075] N i，j (θ i，j )≤N 0，j

[0076] Among them, F is the minimum power consumption cost of the pump station per day, yuan; f i is the daily operating electricity cost of the j-th water pump unit, RMB; JZ is the number of water pump units, units; W e is the total daily target water lifting volume of the pumping station, m 3 .

[0077] If the units in the pump station are of the same model, the efficiency η of the frequency conversion device can be ignored. f，j , the above mathematical model can be transformed into:

[0078] Objective function:

[0079]

[0080] Total water volume constraint:

[0081]

[0082] Power Constraints:

[0083] N i (θ i )≤N0

[0084] Among them, NS i is the number of pumping stations in operation during the i-th period.

[0085] S13, for the series cascade pumping station group including the TJ pumping station, the objective function is to minimize the daily water-lifting power consumption of the cascade pumping station group system, the several time periods divided by 1d are the stage variables, the blade placement angle and operating head of each unit of each pumping station in each time period are the decision variables, and the total water lifting volume of each pumping station at each level within the specified time period, the rated power of the motor of each unit and the total water lifting head of the cascade pumping station group are the constraints to construct the mathematical model for the optimal operation of the series cascade pumping station group:

[0086] Objective function:

[0087]

[0088] Total water volume constraint of the first-level pumping station:

[0089]

[0090] Power Constraints:

[0091] N i，j，l (θ i，j，l )≤N 0，j，l

[0092] Constraints on total lift of cascade water lifting:

[0093]

[0094] Among them, R is the minimum operating power consumption cost of the cascade pump station group, yuan; TJ is the number of stages included in the cascade pump station group, stage; M l is the electricity consumption of the first-level pump station, yuan; W e，l is the total water pumping volume of the first-level pumping station, m 3 ; HE is the total water lifting head of the entire cascade water diversion system, m.

[0095] S2, obtain the stability parameters that characterize the operation of the pump station unit, and use the maximum information coefficient method to analyze the correlation between the unit operating parameters and the stability parameters, and determine the operating parameters according to the correlation.

[0096] Specifically, step S2 includes the following steps:

[0097] S21, collects four types of index data, namely pressure pulsation, vibration, noise and swing, which represent the operating stability of the pump station unit. The collection methods are as follows:

[0098] 1) Vibration data collection: According to the "Pump Vibration Measurement and Evaluation Method" (GB / T29531-2013) and the "Technical Management Regulations for Pump Stations" (GB / T 30948-2021), the main measuring points can be installed at the bearing seat, and the auxiliary measuring points can be installed at the motor frame, the water pump outlet elbow, etc. The measurement conditions are the vibration conditions of the full operating conditions of different unit combinations.

[0099] 2) Noise data collection: According to the "Measurement and Evaluation Method of Pump Noise" (GB / T29529-2013), the noise measurement point can be selected at a noise-sensitive location 1m outside the unit boundary and above 1.2m in height. The motor layer is measured in the walkway, and 4 measurement points are arranged around the water pump layer. Each measurement point measures about 20 data, and the measurement conditions are the noise conditions of the full operating conditions of different unit combinations.

[0100] 3) Pressure pulsation data collection: According to the "Field Test Procedure for Vibration and Pulsation of Hydraulic Machinery (Turbines, Storage Pumps and Pump-Turbines)" (GB / T17189-2017), main measuring points can be arranged at the impeller inlet, between the impeller and the guide vane, and at the guide vane outlet. Auxiliary measuring points can be arranged at the bend of the outlet flow channel (left and right sides of the water inlet direction). The measurement conditions are the pressure pulsation conditions of the full operating conditions of different unit combinations.

[0101] 4) Swing data collection: According to "Measurement and Evaluation of Radial Vibration of Rotating Machinery Shafts" (GB / T11348.5-2008), measuring points can be arranged at the coupling to obtain swing data in the XY direction. The measurement conditions are the swing conditions of the full operating conditions of different unit combinations.

[0102] S22, respectively process the pressure pulsation, vibration, noise and swing data, the methods are as follows:

[0103] 1)1 Vibration data processing: a) Perform displacement time-history analysis on the vibration signal, perform vector summation on the measured vibration displacement values ​​X, Y, and Z of each measuring point, and then derive the velocity value with respect to time. Use the velocity value as the ordinate and time as the abscissa to make a time domain diagram of the vibration velocity signal. b) Perform frequency domain analysis on the vibration signal, perform fast Fourier transform (FFT) on the vibration velocity value of the measuring point, and obtain the vibration frequency domain distribution of each measuring point. Assuming a non-periodic time domain function signal f(t) is given, the transformation formula is:

[0104]

[0105] Where F(ω) is the complex function of f(t) in the frequency domain, ω is the frequency, e -iωt and e iωt is a complex exponential function.

[0106] 2) Noise data processing: The measured noise values ​​are averaged to obtain the average noise values ​​of the pump layer and the motor layer under different unit combinations and different operating conditions.

[0107] 3) Pressure pulsation data processing: a) Perform displacement time history analysis on the pressure pulsation signal, and the pressure pulsation amplitude is calculated by the pressure coefficient C p It means that the calculation formula is:

[0108]

[0109] Where p is the instantaneous pressure value of the flow field, is the mean pressure, and u2 is the impeller circumferential speed. b) Perform Fourier transform on the pressure pulsation signal to obtain the frequency domain characteristics of the pressure pulsation. The processing method is the same as the frequency domain analysis of the vibration signal.

[0110] 4) Swing data processing: Perform displacement time history analysis and frequency domain analysis on the swing signal respectively. The processing method is the same as that of vibration data.

[0111] S23, since the stability parameters of the pump station unit are affected by a variety of operating parameters, the maximum information coefficient (MIC) is introduced to analyze the correlation between the stability parameters and operating parameters such as head, blade angle, speed, and power.

[0112] The maximum information coefficient is a statistical concept used to calculate the linear or nonlinear correlation between two variables. It measures the correlation between the two variables by calculating their mutual information through the approximate probability density distribution of the two variables in the grid a×b. The calculation formula for mutual information is:

[0113]

[0114] Among them, p(x, y) is the joint probability density distribution function, p(x) and p(y) are the marginal probability density distributions of the two variables. Divide the variable X into a grids and the variable Y into b grids, calculate the maximum mutual information in different grid divisions and normalize them, and get the maximum information coefficient MIC:

[0115]

[0116] Where B(n) is a variable, which is set to n by default. 0.6 .

[0117] S3, construct a deep neural network model for each stability parameter, and use the operating parameters of the pump station unit after calibration as input data and the stability parameters as output data, and train to obtain the mapping relationship between the unit operating parameters and the stability parameters to establish an operating condition-stability relationship model.

[0118] Specifically, step S3 includes the following steps:

[0119] S31, selection of relevant operating parameters: for a pump station unit stability parameter V(t), select the N operating parameters X1(t), X2(t), ..., X3(t) with the highest correlation according to the MIC value obtained in step S2. n (t) as input data.

[0120] S32, data preprocessing: first divide the data into a training set and a test set, and then normalize the data to the interval [0, 1]. The normalization formula is:

[0121]

[0122] Among them, x max and x minare the maximum and minimum values ​​in the data respectively.

[0123] S33, construction of a deep fully connected network: determine the number of neurons in the input layer according to the input data dimension, build several hidden layers and an output layer containing one neuron. To prevent overfitting of the model training, a dropout layer can be added to the hidden layer. The hidden layer uses the ReLU activation function, and the output layer uses the linear activation function. The ReLU activation function is:

[0124]

[0125] The linear activation function is:

[0126] f(x)=x

[0127] S34, network training: matrix the input and output data of the training set and input them into the established network in sequence, set the training batch and number of iterations, and use the RMSprop optimization algorithm to train the network.

[0128] S35, training result testing and evaluation: After the test set input data is matrixed, it is input into the trained network in sequence, the output of the network is compared with the output data of the test set, and the mean absolute error (MAE), mean absolute percentage error (MAPE) and root mean square error (RMSE) of the results are calculated. The calculation formula is:

[0129]

[0130] Among them, x i is the true value, is the model output value, and n is the length of the sequence.

[0131] S36, when all error values ​​meet the target requirements, the model training is completed, and the mapping relationship between each operating parameter of the pump station unit and the N operating parameters with the highest correlation therewith can be established:

[0132] V(t)=f(X1(t),X2(t),...,X n (t))

[0133] S4, adopts the large system decomposition-dynamic programming aggregation method to solve the mathematical model of the optimal operation of the series cascade pump station group, and combines iterative optimization with non-steady flow simulation to obtain multiple groups of optimal operation plans of the cascade pump station group considering economy, so as to build a plan library.

[0134] Specifically, step S4 includes the following steps:

[0135] S41, according to the performance characteristic data of each pump station in the series cascade pump station group, the performance curve equation of each pump station device is fitted. The specific method is to regress the performance under different blade placement angles and construct a mathematical expression based on the device model test results. For the flow-head (QH) curve and flow-efficiency (Q-η) curve, the accuracy of the quadratic and cubic polynomials is fully able to meet the engineering requirements, that is:

[0136]

[0137] Among them, the performance at different blade placement angles is converted to the prototype according to the model test results, and the prototype data is fitted with polynomials to obtain the corresponding coefficients a, b, c, d, e, f and g.

[0138] S42, respectively, generalize the water channels between every two pumping stations in the cascade pumping station group. Collect basic information on the water channels (channel length, bottom width, river bottom elevation, slope, cross-section shape, roughness, etc.), clarify the water use and water consumption of water users along the inter-stage water channels (agricultural water users, industrial water users, lock water users, etc.), and construct a generalized map of the water channel network.

[0139] S43, taking each level of pumping station as the object, considering the daily average head unchanged, within the feasible domain of pumping station head, discretizes into multiple daily average heads. Under each daily average head, consider the 100%, 80%, and 60% load water volume of the pumping station when it is running at a fixed angle and constant speed as the optimization target water volume, adopt the large system decomposition-dynamic programming aggregation method, respectively perform the optimization operation calculation of the pumping station under different water volume constraints of each daily average head, and obtain a complete set of optimization operation schemes for pumping stations at all levels, that is, corresponding to the water lifting load W j The minimum daily operating cost f of a single pump unit is calculated based on the water lifting lift at each time period, the number of units in operation at each pump station, the blade angle of the operated pumps and the unit speed.

[0140] Among them, the large system decomposition-dynamic programming aggregation method is specifically as follows:

[0141] ① Decompose the optimization operation model of the cascade series pumping station group into a multi-unit optimization operation model of TJ pumping stations.

[0142] ② Decompose the multi-unit optimization operation model of pump stations at all levels into JZ single-unit optimization operation models.

[0143] ③ The single unit optimization operation model is a typical one-dimensional dynamic programming model, which is solved by dynamic programming method. The stage variable is i (i = 1, 2, ..., SN), and the decision variable is the blade placement angle θ i , the state variable λ is the water extraction amount at different stages, and the corresponding recursive equation is:

[0144] Phase i=1:

[0145]

[0146] Among them, the state variable λ1 can be discretized in the corresponding feasible domain: λ1=0,W1,W2,...,W j The blade placement angle θ1 can be discretized in the corresponding feasible domain and should satisfy: Q1(θ1)ΔT1≥λ1. At this time, the corresponding average flow rate and device efficiency η are obtained from the performance relationship of the water pump device by the blade placement angle θ1 and the average head H1. z，1 .

[0147] Phase I:

[0148]

[0149] Among them, the state variable λ i and blade placement angle θ i The discretization is the same as above and should satisfy Q i (θ i )ΔT i ≥λ i The state transfer equation is:

[0150] λ i-1 =λ i -Q i (θ i )ΔT i

[0151] Among them, i=1, 2, ..., SN-1.

[0152] Stage SN:

[0153]

[0154] Among them, the state variable λ SN =W j , blade placement angle θ SN The discretization of is the same as above, and should satisfy λ SN-1 =λ SN -Q SN (θ SN )ΔT SN ,λ SN ≥W j .

[0155] ④ The total amount of water lifted when running at the maximum blade angle in each discrete period with a certain step length W j，max , the single unit optimization operation model is used to calculate the water lifting capacity W of each pump unit j，m The minimum water extraction cost f under the requirement (m=1, 2, ..., max) j，m , thus obtaining a series of W j，m ~fj，m (W j，m )relation.

[0156] ⑤ A series of W obtained j，m ~f j，m (W j，m ) relationship (j = 1, 2, ..., JZ; m = 1, 2, ..., max) transforms the single pump station multi-unit optimization operation model into the following aggregation model:

[0157] Objective function:

[0158]

[0159] Total water volume constraint:

[0160]

[0161] Power Constraints:

[0162] N i，j (θ i，j )≤N 0，j

[0163] The aggregation model can be used as a typical one-dimensional dynamic programming model, with the stage variable j (j = 1, 2, ..., JZ) and the decision variable being the water extraction capacity W of each unit. j , and its discrete range is the discrete range W of the target water extraction capacity when optimizing a single unit j，m (m=1,2,...,max), the state variable λ is the discrete value of the total amount of water pumped by each unit in the pumping station. The dynamic programming method is used to solve the model and obtain the total amount of water pumped by the pumping station W e The M value, and the optimal water lifting capacity combination W of each unit in each pumping station corresponding to M j * (j=1, 2, ..., JZ). Then, the optimal startup mode of each pump unit can be obtained by checking the optimization results of the single unit, that is, the optimal blade placement angle θ of each unit in each period i，j * .

[0164] S44, taking the secondary water pumping station group as a research case (the results can be extended to multi-stage pumping stations), the initial head and water lifting load of each stage of the pumping station group are given as known conditions. The water pumping capacity W of the second stage pumping station is known. e，2 , the head at the start-up time is the average daily head of the second-level pump station, then the initial optimized operation plan and corresponding operation cost of the second-level pump station, as well as the 24-hour water lifting flow process can be obtained by step S43. At the same time, the water consumption of various types of water users in the pump station group and the losses such as evaporation and leakage in the river channel W1′ are taken into account, and the water lifting load W of the first-level pump station can be determined e，1 :

[0165] We,1 =W e，2 +W1′

[0166] S45, considering the second-stage pumping station to be started up after being delayed by -2h, -1h, 0h, 1h, 2h and 3h respectively compared with the first-stage pumping station, the delayed initial head of the first-stage pumping station is taken as the average daily head, and the optimized operation process corresponding to the matching load under the average daily head is re-taken as the boundary condition. Combined with the flow process of the second-stage pumping station and the water use process of the inter-stage water users, one-dimensional non-steady flow simulations are carried out under 6 schemes to obtain the flow process lines and water level process lines of each section of the inter-stage water transfer channel.

[0167] The basic principle of one-dimensional unsteady flow simulation is the Saint-Venant equations:

[0168]

[0169] Among them, Z is the water level, Q is the flow rate, B is the water surface width, A is the cross-sectional area of ​​the water flow, R is the hydraulic radius, C is the Xie Cai coefficient, q is the lateral inflow per unit length of the river, and t and s are the time and space coordinates respectively.

[0170] Preissmann's four-point implicit difference format is used to discretize the Saint-Venant equations. The discretization format is as follows:

[0171]

[0172] Among them, n is the divided calculation period, j is the divided calculation section, C j , D j 、E j 、F j , G j ,φ j It is a function of water level Z and flow rate Q (including initial value of iteration):

[0173]

[0174] Where Δx j is the interval between the divided river sections, Δt is the calculation time step, θ is the weight of the Preissmann implicit format, α is the kinetic energy correction coefficient, u is the cross-sectional flow velocity, and g is the gravitational acceleration.

[0175] The boundary conditions of the model are: the flow process of the first-stage pumping station, the flow process of the second-stage pumping station, and the water intake process of water users along the line. The initial conditions of the model are: the initial flow and initial operating water level of the inter-stage water channel. For a river with L sections, 2L nonlinear equations about Z and Q can be formed.

[0176] The pursuit method is used to solve the nonlinear equations. For the case of known flow boundary conditions, As parameter variables, the difference equations can be written as the following pursuit equations:

[0177]

[0178] Among them, the catch-up coefficient is:

[0179]

[0180] The specific solution is: Known, will As a parameter variable, proceed from the upstream boundary to the downstream boundary and find and Downstream Boundary Known, by and The recursive relationship can be obtained Then substitute back from the downstream boundary to the upstream boundary to calculate the water level Z and flow Q of each section in each time period.

[0181] S46, based on the river flow process line and water level process line obtained in step S45, check whether the initial water level assumptions of the first and second level pumping stations are reasonable (i.e., the water level above the first level pumping station and below the second level pumping station is basically the same as the 24-hour average water level). If not, re-assume the initial water level of the first level pumping station and repeat step S45 to make the simulation results basically consistent. Collect the operation plans that meet the requirements and build a library of optimized operation plans for cascade pumping station groups.

[0182] S5, using the operating condition-stability relationship model to calculate the stability parameters of the pump station unit under each scheme in the scheme library, and obtain the optimal operation scheme considering economy and stability based on the comparison results of the stability parameters of the pump station unit.

[0183] Specifically, step S5 includes the following steps:

[0184] S51, using the "operating condition-stability relationship" model established in step S3, taking the operating condition parameters obtained in step S2 as input, respectively calculate the pressure pulsation, vibration, noise, and swing data under each scheme.

[0185] S52, based on the flow process lines and water level process lines of each section of the inter-stage water transfer channel under each scheme obtained in step S4, and the stability data obtained in step S51, each scheme is calculated and analyzed from five aspects: head matching, optimized operation economy, water level fluctuation of the inter-stage water transfer channel, flood control and drainage, navigation, ecological water level requirements, and unit operation stability:

[0186] 1) Analysis of head matching of cascade pumping stations: Calculate the water level fluctuation above and below the pumping stations under each operation scheme respectively, and compare the scheme with smaller water level fluctuation above and below the pumping station, which will have a higher matching degree with the established head during the optimized operation process.

[0187] 2) Analysis of the economic efficiency of optimized operation of cascade pumping station system: Calculate the minimum unit water lifting cost of each pumping station under each operation scheme, using fixed-angle constant speed operation and the minimum unit water lifting cost under the optimized operation scheme, and obtain the total savings of the cascade pumping station group. Compare which scheme has the smaller total savings, and the optimized operation economy of the pumping station system under this scheme is higher.

[0188] 3) Analysis of water level fluctuations in inter-stage water transfer channels: Calculate the water level fluctuations in the inter-stage water transfer channels for each pump station under each operation scheme, and compare which scheme has the smaller water level fluctuation. The water level changes in the water transfer channels under this scheme will be relatively more stable.

[0189] 4) Analysis of flood control, drainage, navigation and ecological water level requirements of water conveyance channels: Based on the non-steady flow simulation results of water conveyance channels, analyze whether the water level and water level fluctuation of the channel under each operation scheme can meet the requirements of flood control, drainage, navigation and ecological water level in each period. For the optimization scheme that meets the requirements, the scheme with the smallest water level fluctuation can be selected.

[0190] 5) Analysis of pump station unit operation stability: Based on the stability parameters of each unit under each operation scheme, a comprehensive analysis is conducted to compare which scheme has the smaller pressure pulsation, vibration, noise, and swing data of the unit. The pump station unit has higher operation stability under this scheme.

[0191] S53, comprehensively compare the five aspects in step S52 with the stability parameters of the pump station units under each scheme obtained by calculation. For each indicator involved, determine the importance and weight of each indicator according to actual needs and relevant expert opinions, use the analytic hierarchy process (AHP) to establish a multi-indicator comprehensive evaluation model, take various indicators into consideration, and finally give the optimal operation scheme for the cascade pump station group considering economy and stability.

[0192] According to the method for recommending an optimal operation plan for a cascade pump station group integrating stable operation according to an embodiment of the present invention, a mathematical model for the optimal operation of a cascade pump station group with the goal of minimizing the daily water extraction power consumption is constructed, and the "large system decomposition-dynamic programming aggregation" method is used to solve the model and construct a solution library, and a one-dimensional hydraulic model is integrated to ensure the head matching degree and water level requirements between pump stations at all levels. At the same time, the maximum information coefficient method is used to calibrate the operating parameters related to the stable operation of the unit, and a deep neural network model is established for the calibrated operating parameters to obtain the mapping relationship between the operating parameters and the unit stability parameters, so as to perform an operation stability analysis on the plans in the optimal operation plan library of the cascade pump station group, and comprehensively compare and select the optimal operation plan considering economy and stability. The present invention uses deep learning technology to integrate the operation stability factor into the optimal operation plan of the cascade pump station group. The obtained operation plan takes into account both economy and stability. Compared with the traditional pump station group optimization operation model, it has the advantages of being more in line with engineering practice and high feasibility.

[0193] In order to implement the above embodiment, Figure 3 As shown, the present embodiment also provides a system 10 for recommending optimized operation plans for cascade pump stations integrating stable operation, the system 10 comprising a mathematical model building module 100, a stability parameter calibration module 200, a stability relationship model building module 300, an optimized operation model solving module 400 and an optimized operation comprehensive comparison module 500;

[0194] The mathematical model building module 100 is used to build a mathematical model for optimizing the operation of a series cascade pumping station group taking economic efficiency into consideration, taking the minimum daily water extraction power consumption cost as the objective function;

[0195] The stability parameter calibration module 200 is used to obtain the stability parameters characterizing the operation of the pump station unit, and analyze the correlation between the unit operating parameters and the stability parameters using the maximum information coefficient method, and calibrate the operating parameters according to the correlation;

[0196] The stability relationship model building module 300 is used to build a deep neural network model for each stability parameter, and use the calibrated operating parameters of the pump station unit as input data and the stability parameters as output data, and train to obtain a mapping relationship between the unit operating parameters and the stability parameters to establish an operating condition-stability relationship model;

[0197] The optimization operation model solving module 400 is used to solve the optimization operation mathematical model of the series cascade pump station group by adopting the large system decomposition-dynamic programming aggregation method, and to perform iterative optimization in conjunction with the non-steady flow simulation to obtain multiple groups of economic optimization operation schemes of the cascade pump station group, so as to build a scheme library;

[0198] The optimized operation comprehensive comparison module 500 is used to calculate the stability parameters of the pump station units under each scheme in the scheme library using the operating condition-stability relationship model, and obtain the optimal operation scheme considering economy and stability based on the comparison results of the pump station unit stability parameters.

[0199] Furthermore, the mathematical model building module 100 is also used for:

[0200] The objective function is to minimize the total power consumption of daily startup operation, the divided time periods are stage variables, the blade placement angles in each time period are decision variables, the total amount of water pumped in the specified time period and the rated power of the motor are constraints, and a mathematical model for the optimal operation of a single unit is constructed.

[0201] Taking the minimum daily pumping cost of the pumping station as the objective function, the blade angle of each unit at each time period as the decision-making quantity, and the specified water pumping volume and unit power requirements as constraints, a mathematical model for the optimal operation of multiple units is constructed.

[0202] The objective function is to minimize the daily water-lifting power consumption of the cascade pump station group system, the divided time periods are stage variables, the blade placement angle and operating head of each unit at each level of pumping station in each time period are decision variables, the total water lifting volume of each level of pumping station within the specified time period, the rated power of the motor of each unit and the total water lifting head of the cascade pump station group are constraints, and a mathematical model for the optimal operation of the series cascade pump station group is constructed.

[0203] Furthermore, the stability parameter calibration module 200 is also used for:

[0204] Collect stability parameters that characterize the operation of pump station units; stability parameters include pressure pulsation, vibration, noise and swing data;

[0205] The pressure pulsation, vibration, noise and swing data are processed respectively to obtain the stability parameters after processing, including:

[0206] The correlation between the processed stability parameter and the operating condition parameter is analyzed based on the maximum information coefficient; wherein the maximum information coefficient is obtained by calculating the maximum value of the mutual information in different grid divisions and normalizing it.

[0207] Furthermore, the stability relationship model building module 300 is also used for:

[0208] According to the maximum information coefficient, multiple operating condition parameters with the highest correlation among the stability parameters of the pump station unit are selected as input data;

[0209] Divide the input data into training set and test set;

[0210] Determine the number of neurons in the input layer according to the dimension of the input data and build a deep neural network model:

[0211] The training set is input into the deep neural network model, and the RMSprop optimization algorithm is used to train the deep neural network model to obtain a trained deep neural network model;

[0212] Input the test set into the trained deep neural network model to obtain the network output result, and compare the network output result with the actual output data of the test set to obtain the comparison result, and calculate the mean absolute error, mean absolute percentage error and root mean square error of the comparison result;

[0213] When all error values ​​meet the target requirements, the model training is completed, and a mapping relationship between each operating parameter of the pump station unit and multiple operating parameters with the highest correlation is established to establish an operating condition-stability relationship model.

[0214] According to the embodiment of the present invention, a cascade pump station group optimization operation plan recommendation system integrating stable operation is constructed to construct a mathematical model for the optimization operation of the cascade pump station group with the goal of minimizing the daily water extraction power consumption cost. The "large system decomposition-dynamic programming aggregation" method is used to solve the model and build a plan library, and a one-dimensional hydraulic model is integrated to ensure the head matching degree and water level requirements between pump stations at all levels. At the same time, the maximum information coefficient method is used to calibrate the operating parameters related to the stable operation of the unit, and a deep neural network model is established for the calibrated operating parameters to obtain the mapping relationship between the operating parameters and the unit stability parameters. In this way, the plans in the cascade pump station group optimization operation plan library are analyzed for operation stability, and the optimal operation plan considering economy and stability is selected by comprehensive comparison. The present invention uses deep learning technology to integrate the operation stability factor into the optimization operation plan of the cascade pump station group. The obtained operation plan takes into account both economy and stability. Compared with the traditional pump station group optimization operation model, it has the advantages of being more in line with engineering practice and high feasibility.

[0215] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0216] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

Claims

1. A method for recommending an optimal operation plan for a cascade pump station group integrating stable operation, characterized in that: The following steps are involved: Taking the minimum daily water-lifting power consumption as the objective function, a mathematical model for the optimal operation of a series cascade pumping station group considering economics is constructed. Obtain stability parameters that characterize the operation of the pump station unit, and use the maximum information coefficient method to analyze the correlation between the unit operating parameters and the stability parameters, and rate the operating parameters according to the correlation; A deep neural network model is constructed for each stability parameter, and the operating parameters of the pump station unit after calibration are used as input data and the stability parameters are used as output data. The mapping relationship between the unit operating parameters and the stability parameters is obtained through training to establish an operating condition-stability relationship model; The large system decomposition-dynamic programming aggregation method is used to solve the mathematical model of the optimal operation of the series cascade pump station group, and combined with the non-steady flow simulation, iterative optimization is performed to obtain multiple groups of economically optimized operation plans for the cascade pump station group, so as to build a plan library; The operating condition-stability relationship model is used to calculate the stability parameters of the pump station units under each scheme in the scheme library, and the optimal operation scheme considering economy and stability is obtained based on the comparison results of the stability parameters of the pump station units.

2. The method according to claim 1, characterized in that Taking the minimum daily water extraction power consumption as the objective function, a mathematical model for the optimal operation of a series cascade pumping station group considering economic efficiency is constructed, including: The objective function is to minimize the total power consumption of daily startup operation, the divided time periods are stage variables, the blade placement angles in each time period are decision variables, the total amount of water pumped in the specified time period and the rated power of the motor are constraints, and a mathematical model for the optimal operation of a single unit is constructed. Taking the minimum daily pumping cost of the pumping station as the objective function, the blade angle of each unit at each time period as the decision-making quantity, and the specified water pumping volume and unit power requirements as constraints, a mathematical model for the optimal operation of multiple units is constructed. The objective function is to minimize the daily water-lifting power consumption of the cascade pump station group system, the divided time periods are stage variables, the blade placement angle and operating head of each unit at each level of pumping station in each time period are decision variables, the total water lifting volume of each level of pumping station within the specified time period, the rated power of the motor of each unit and the total water lifting head of the cascade pump station group are constraints, and a mathematical model for the optimal operation of the series cascade pump station group is constructed.

3. The method according to claim 1, characterized in that Obtain the stability parameters that characterize the operation of the pump station unit, and use the maximum information coefficient method to analyze the correlation between the unit operating parameters and the stability parameters, and rate the operating parameters according to the correlation, including: Collecting stability parameters that characterize the operation of the pump station unit; wherein the stability parameters include pressure pulsation, vibration, noise and swing data; The pressure pulsation, vibration, noise and swing data are processed respectively to obtain the stability parameters after processing, including: The correlation between the processed stability parameter and the operating condition parameter is analyzed based on the maximum information coefficient; wherein the maximum information coefficient is obtained by calculating the maximum value of the mutual information in different grid divisions and normalizing it.

4. The method according to claim 3, characterized in that A deep neural network model is constructed for each stability parameter, and the operating parameters of the pump station unit after calibration are used as input data and the stability parameters are used as output data. The mapping relationship between the unit operating parameters and the stability parameters is obtained through training to establish an operating condition-stability relationship model, including: According to the maximum information coefficient, multiple operating condition parameters with the highest correlation among the stability parameters of the pump station unit are selected as input data; Divide the input data into training set and test set; Determine the number of neurons in the input layer according to the dimension of the input data and build a deep neural network model: The training set is input into the deep neural network model, and the RMSprop optimization algorithm is used to train the deep neural network model to obtain a trained deep neural network model; Input the test set into the trained deep neural network model to obtain the network output result, and compare the network output result with the actual output data of the test set to obtain the comparison result, and calculate the mean absolute error, mean absolute percentage error and root mean square error of the comparison result; When all error values ​​meet the target requirements, the model training is completed, and a mapping relationship between each operating parameter of the pump station unit and multiple operating parameters with the highest correlation is established to establish an operating condition-stability relationship model.

5. The method according to claim 4, characterized in that The large system decomposition-dynamic programming aggregation method is used to solve the mathematical model of the optimal operation of the series cascade pump station group, and combined with the unsteady flow simulation, iterative optimization is performed to obtain multiple groups of economically optimized operation schemes for the cascade pump station group, so as to build a scheme library, including: The performance curve equations of pumping station devices at each level are obtained by fitting the performance characteristic data of pumping station devices at each level in the mathematical model for optimizing the operation of the series cascade pumping station group; Based on the basic data of the water supply channel and the water use status of water users along the inter-level water supply channel, a generalized map of the water supply channel network between every two pumping stations in the cascade pumping station group is constructed; Based on the generalized diagram of the water supply channel network, the load water volume when the pump station is running at a fixed angle and constant speed is taken as the optimized target water volume, and the large system decomposition-dynamic programming aggregation method is adopted to calculate the optimal operation of the pump station under the constraints of different water volumes at each daily average head based on the performance curve equation to obtain the optimal operation plan of each level of pump stations; Based on the known initial lift and water lifting load of each level of the secondary water lifting pump station group, the water lifting load of the first-level pump station is determined: Taking the delayed initial head of the first-stage pumping station as the daily average head, re-taking the optimized operation process corresponding to the matching load under the daily average head as the boundary condition, combining the flow process of the second-stage pumping station and the water use process of the inter-stage water users, one-dimensional unsteady flow simulations under various schemes were performed to obtain the flow process lines and water level process lines of each section of the inter-stage water transfer channel. According to the flow process lines and water level process lines of each section of the inter-stage water transfer channel, it is judged whether the initial water level assumptions of the first and second pumping stations are reasonable, and based on the judgment results, an optimized operation plan that meets the requirements is obtained to construct an optimized operation plan for the cascade pumping station group.

6. The method according to claim 5, characterized in that The working condition-stability relationship model is used to calculate the stability parameters of the pump station units under each scheme in the scheme library, and the optimal operation scheme considering economy and stability is obtained based on the comparison results of the stability parameters of the pump station units, including: Input the working condition parameters into the working condition-stability relationship model to calculate the stability data under each scheme; Based on the flow process lines, water level process lines and stability data of each section of the inter-stage water channel, each scheme is calculated and analyzed from the perspectives of head matching, optimized operation economy, water level fluctuation of the inter-stage water channel, flood control and drainage, navigation, ecological water level requirements and unit operation stability to obtain the stability parameters of the pump station units under each scheme; The stability parameters of the pump station units under the various schemes are compared to obtain the optimal operation scheme of the cascade pump station group considering economy and stability according to the comparison results.

7. A system for recommending optimal operation plans for cascade pump station groups with integrated stable operation, characterized in that: include: The mathematical model building module is used to build a mathematical model for optimizing the operation of a series of cascade pumping stations taking economic factors into consideration, taking the minimum daily water extraction power consumption as the objective function; The stability parameter calibration module is used to obtain the stability parameters that characterize the operation of the pump station unit, and use the maximum information coefficient method to analyze the correlation between the unit operating parameters and the stability parameters, and calibrate the operating parameters according to the correlation; A stability relationship model building module is used to build a deep neural network model for each stability parameter, and use the calibrated operating parameters of the pump station unit as input data and the stability parameters as output data, and train to obtain a mapping relationship between the unit operating parameters and the stability parameters to establish an operating condition-stability relationship model; The optimization operation model solving module is used to solve the mathematical model of the optimal operation of the series cascade pump station group by using the large system decomposition-dynamic programming aggregation method, and to perform iterative optimization in conjunction with the non-steady flow simulation to obtain multiple groups of economically optimized operation plans for the cascade pump station group, so as to build a plan library; The comprehensive comparison module for optimized operation is used to calculate the stability parameters of the pump station units under each scheme in the scheme library using the operating condition-stability relationship model, and obtain the optimal operation scheme considering economy and stability based on the comparison results of the stability parameters of the pump station units.

8. The system according to claim 7, characterized in that Mathematical model building blocks, also used for: The objective function is to minimize the total power consumption of daily startup operation, the divided time periods are stage variables, the blade placement angles in each time period are decision variables, the total amount of water pumped in the specified time period and the rated power of the motor are constraints, and a mathematical model for the optimal operation of a single unit is constructed. Taking the minimum daily pumping cost of the pumping station as the objective function, the blade angle of each unit at each time period as the decision-making quantity, and the specified water pumping volume and unit power requirements as constraints, a mathematical model for the optimal operation of multiple units is constructed. The objective function is to minimize the daily water-lifting power consumption of the cascade pump station group system, the divided time periods are stage variables, the blade placement angle and operating head of each unit at each level of pumping station in each time period are decision variables, the total water lifting volume of each level of pumping station within the specified time period, the rated power of the motor of each unit and the total water lifting head of the cascade pump station group are constraints, and a mathematical model for the optimal operation of the series cascade pump station group is constructed.

9. The system according to claim 8, characterized in that The stability parameter calibration module is also used for: Collecting stability parameters that characterize the operation of the pump station unit; wherein the stability parameters include pressure pulsation, vibration, noise and swing data; The pressure pulsation, vibration, noise and swing data are processed respectively to obtain the stability parameters after processing, including: The correlation between the processed stability parameter and the operating condition parameter is analyzed based on the maximum information coefficient; wherein the maximum information coefficient is obtained by calculating the maximum value of the mutual information in different grid divisions and normalizing it.

10. The system according to claim 9, characterized in that Stability relation model building blocks are also used to: According to the maximum information coefficient, multiple operating condition parameters with the highest correlation among the stability parameters of the pump station unit are selected as input data; Divide the input data into training set and test set; Determine the number of neurons in the input layer according to the dimension of the input data and build a deep neural network model: The training set is input into the deep neural network model, and the RMSprop optimization algorithm is used to train the deep neural network model to obtain a trained deep neural network model; Input the test set into the trained deep neural network model to obtain the network output result, and compare the network output result with the actual output data of the test set to obtain the comparison result, and calculate the mean absolute error, mean absolute percentage error and root mean square error of the comparison result; When all error values ​​meet the target requirements, the model training is completed, and a mapping relationship between each operating parameter of the pump station unit and multiple operating parameters with the highest correlation is established to establish an operating condition-stability relationship model.

Images