Optimized construction method of wing-shaped flow guide cover for improving stall flow field in axial flow pump

Abstract

The invention discloses a wing-shaped fairing optimization construction method for improving a stall flow field in an axial flow pump. The method comprises the following steps that a three-dimensional water body model of a horizontal axial flow pump station is constructed, wing-shaped fairing optimization parameters are selected, sample data are collected, and a three-dimensional water body model of a front guide vane provided with a wing-shaped fairing is drawn for analysis; constructing an objective function, and calculating an objective function value of each group of parameter sample data; performing regression optimization on the optimization variable and the target function of the wing-shaped air deflector through a PSO-BP neural network algorithm to obtain an optimal optimization parameter combination; and based on the obtained optimal parameter combination scheme, constructing a wing-shaped flow guide cover, and mounting the wing-shaped flow guide cover on the front guide vane of the horizontal axial flow pump. The wing-shaped flow guide cover can effectively eliminate the poor flow state near an impeller inlet, the water inlet condition of a water pump is improved, and the working condition range of safe and stable operation of a horizontal axial flow pump station is widened.

Description

Technical Field

[0001] The invention relates to the technical fields of water conservancy engineering, pump station engineering and mechanical engineering, and in particular to a method for optimizing the construction of a wing-shaped flow guide cover for improving a stall flow field in an axial flow pump. Background Art

[0002] Horizontal axial flow pump stations have the characteristics of large flow and low head, so they are widely used in agricultural irrigation and drainage, urban water supply and drainage, inter-basin water transfer and regional water transfer. However, for axial flow pumps, under low flow conditions, rotational stall is prone to occur inside the axial flow pump. The stall vortex blocks the impeller channel, seriously affecting the impeller's ability to work, and then leads to poor flows such as flow separation, secondary flow and vortex, which reduces the water pumping efficiency and induces hydraulic excitation problems such as poor pressure pulsation, cavitation, and excessive noise. These adverse hydraulic instability phenomena affect the safe, stable and efficient operation of the pump station, and in severe cases, they can also cause hydraulic failures such as cavitation, excessive noise and increased vibration.

[0003] To solve the stall problem in axial flow pumps, conventional measures mainly include blade optimization design, variable speed regulation and guide fences. Although blade optimization is effective, it is costly and complex to implement, too targeted and not applicable for promotion; variable speed regulation increases system complexity and cost, and cannot completely avoid stall; guide fences are easy to install and maintain, but their effect is limited. Arranging guide fences at the impeller inlet will also affect the hydraulic performance of the pump station unit under the design conditions.

[0004] Therefore, the existing technology still has limitations in improving stall, and a more efficient, economical and convenient solution is urgently needed. Designing a new wing-shaped fairing device to improve the stall flow field in the axial flow pump in a more efficient, economical and convenient way has important practical and engineering significance. Summary of the invention

[0005] The problem to be solved by the present invention is to provide a wing-shaped guide cover optimization construction method for improving the stall flow field in an axial flow pump, which is installed on the front guide vane of a horizontal axial flow pump to efficiently, economically and conveniently improve the stall flow field problem in the axial flow pump.

[0006] The present invention adopts the following technical solution: a method for optimizing the construction of a wing-shaped guide cover for improving the stall flow field in an axial flow pump, comprising the following steps:

[0007] S1. Construct a three-dimensional water body model of a horizontal axial flow pump station, select the optimization parameters of the wing-shaped guide cover, take several groups of optimization parameter sample data, draw corresponding three-dimensional water body models of several leading guide vanes equipped with the wing-shaped guide cover, and perform numerical analysis;

[0008] S2. Construct an objective function and calculate the objective function value of each group of optimization parameter sample data;

[0009] S3, dividing each group of sample data into a training set, a validation set, and a test set, and sequentially performing parameter learning and validation of the PSO-BP neural network algorithm;

[0010] S4. Perform regression optimization on the optimization variables and objective functions of the wing-shaped fairing through the PSO-BP neural network algorithm to obtain the best optimization parameter combination, and perform numerical simulation verification on the best parameter combination scheme;

[0011] S5. Based on the best parameter combination solution, a wing-shaped guide cover is constructed and installed on the front guide vane of the horizontal axial flow pump to improve the stall flow field in the axial flow pump.

[0012] Among them, the wing-shaped fairing is an annular structure made of stainless steel. It is installed on the leading guide vane by welding. There are multiple circular holes evenly arranged on the surface of the wing-shaped fairing, and the cross-section is a standard NACA0009 airfoil.

[0013] A horizontal axial flow pump comprises a water inlet channel, a front guide vane, an impeller, a rear guide vane and a water outlet channel which are connected in sequence, wherein the front guide vane is a straight blade and the rear guide vane is a twisted blade.

[0014] Preferably, the optimization parameters in step S1 are geometric variables of the airfoil fairing, including: the length L1 of the airfoil fairing, the radius R1 of the airfoil, the radius R2 of the circular hole, the axial spacing L2 of the circular hole and the circumferential spacing L3 of the circular hole, and the initial value range of the geometric parameters is set according to engineering requirements and experience.

[0015] Preferably, in step S1, constructing a three-dimensional water body model of a horizontal axial flow pump station includes the following sub-steps:

[0016] S1.1. Extract several groups of optimization parameter sample data through Latin hypercube algorithm;

[0017] S1.2, using CREO software to automatically draw a three-dimensional water model of several leading guide vanes equipped with wing-shaped guide covers corresponding to the sample data;

[0018] S1.3. Automatically mesh the water model using ANSYS Mesh and perform numerical analysis using CFX software.

[0019] Preferably, in step S2, calculating the objective function value of each group of optimization parameter sample data includes the following sub-steps:

[0020] S2.1. Use ANSYS Mesh and ANSYA CFX software to complete the automatic meshing and numerical calculation of each water body model;

[0021] S2.2, through ANSYS Post processing and entropy production theory, obtain each group of sample data and the corresponding objective function value;

[0022] The objective function value is a performance index of the horizontal axial flow pump station under a small flow condition, including: head, efficiency, time-averaged loss, pulsation loss and wall loss.

[0023] The relationship between sample data and corresponding objective function values ​​is as follows:

[0024] By adjusting the wing-shaped shroud length L1, the control range of the shroud on the fluid flow is controlled. By adjusting the shroud radius R1, the radial installation position of the shroud is determined, the flow distribution of the inner and outer channels of the pump is controlled, and the head and efficiency of the axial flow pump are improved.

[0025] By adjusting the circular hole radius R2, the circular hole axial spacing L2 and the circular hole circumferential spacing L3, the passing area and flow smoothness of the fluid in the inner and outer channels are controlled, the pressure gradient change when the fluid passes through the guide cover is reduced, and the pulsation loss and time-averaged loss of the flow are reduced.

[0026] Preferably, in step S3, there is no intersection among the training set, the validation set, and the test set, and the ratio is 7:1:2;

[0027] The training set is used to determine learning parameters such as network weights; the validation set is used to determine hyperparameters, and the test set does not participate in the learning process and is used to judge the accuracy of the prediction model;

[0028] The hyperparameters include: the number of network layers, the number of network nodes, and the number of iterations.

[0029] Preferably, in step S4, the optimization variables and the objective function of the wing-shaped fairing are regressively optimized by a PSO-BP neural network algorithm, including the following sub-steps:

[0030] S4.1. Determine the PSO algorithm structure: set the particle swarm size and initialize the position x of each particle i and speed v i , where position x i Represents the optimization variable of the wing-shaped fairing, and sets the particle position search space range as the upper and lower limits of the optimization variable;

[0031] Set the inertia weight w, acceleration factors c1 and c2, and maximum number of iterations T of the PSO algorithm, calculate the initial fitness value of the particle, and evaluate it based on the prediction error of the neural network model;

[0032] Initialize the global state and find the historical optimal position p of each particle i and the global optimal position g of the particle swarm;

[0033] S4.2. Use neural network error to construct fitness:

[0034] Construct the objective function and use the neural network prediction error as the fitness function F, which is expressed as:

[0035]

[0036] Among them, y i is the actual objective function value, is the objective function value predicted by the neural network, and N is the number of samples;

[0037] According to the current position of the particle, the neural network model is used to calculate the objective function value and update the fitness F(x i ); if the current fitness is better than the historical optimal fitness, then update the corresponding optimal position p i and the global optimal position g;

[0038] S4.3, performing neural network training, including the following sub-steps:

[0039] S4.3.1. Training BP neural network: Use the particle position obtained by PSO optimization as the input of the neural network and the objective function value as the output to train the BP neural network; take the mean square error MSE as the error function of the network, adjust the weight and bias of the neural network to minimize the error;

[0040] S4.3.2. Iteratively update particle position and velocity: After each training and validation, update the particle position and velocity according to the following formula:

[0041] v i (t+1)=wv i (t)+c1r1(p i -x i (t))+c2r2(gx i (t))

[0042] r1x i (t+1)=x i (t)+v i (t+1)

[0043] Among them, r1 and r2 are random numbers in [0, 1], x i (t), v i (t) is the position and velocity of the particle at time t;

[0044] S4.3.3. Verify the neural network model: Use the validation set data to verify the performance of the neural network. If the validation set error is lower than the preset threshold ε, stop training and output the current best parameter combination; otherwise, continue training;

[0045] S4.3.4. Output the best parameter combination: After multiple iterations, the optimal particle position g is obtained as the best parameter combination of the wing-shaped fairing.

[0046] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:

[0047] 1. The wing-shaped guide cover optimization construction method of the present invention uses the CFD-EP coupling PSO-BP neural network method to quickly design a wing-shaped guide cover with excellent performance through a workstation. After the wing-shaped guide cover with optimized design is installed on the front guide vane, the unfavorable flow state near the impeller inlet can be effectively eliminated, thereby improving the water inlet conditions of the water pump and broadening the working condition range of the horizontal axial flow pump station for safe and stable operation.

[0048] 2. The wing-shaped fairing of the present invention is low-cost and easy to install. It can not only improve the stall flow field of the water pump unit under low flow conditions, but also improve the hydraulic performance of the unit under other conditions.

[0049] 3. The wing-shaped air guide cover optimization construction method of the present invention can also be used for the optimization design of impeller blades, guide vane blades and flow channel profiles. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a schematic diagram of a horizontal axial flow pump station of the present invention;

[0051] Figure 2 This is a schematic diagram of installing the wing-shaped fairing of the present invention;

[0052] Figure 3 It is a schematic diagram of the arrangement of circular holes on the surface of the wing-shaped fairing of the present invention;

[0053] Figure 4 The top view and left side view of the wing-shaped fairing of the present invention;

[0054] Figure 5 The cross section of the wing-shaped fairing of the present invention;

[0055] Figure 6 This is a flow chart of the method for optimizing the construction of a wing-shaped fairing according to the present invention;

[0056] Figure 7 This is the flow state of the embodiment of the present invention without the wing-shaped fairing device;

[0057] Figure 8 The flow state after the wing-shaped fairing device is added to the embodiment of the present invention;

[0058] Explanation of the numbers in the figure: 1-water inlet channel, 2-front guide vane, 3-impeller, 4-rear guide vane, 5-water outlet channel, 6-rim; 7-straight blade; 8-hub; 9-wing-shaped guide cover. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the application is further elaborated in detail below in conjunction with the accompanying drawings. The described embodiments are only a part of the embodiments involved in the present invention. All non-innovative embodiments of other researchers in the field on this embodiment belong to the protection scope of the present invention. At the same time, for the step numbering in the embodiment of the present invention, it is only set for the convenience of explanation, and the order between the steps is not limited in any way. The execution order of each step in the embodiment can be adaptively adjusted according to the understanding of those skilled in the art.

[0060] In one embodiment of the present invention, a horizontal axial flow pump station such as Figure 1 As shown, it comprises: an inlet flow channel 1, a front guide vane 2, an impeller 3, a rear guide vane 4 and an outlet flow channel 5, wherein the front guide vane is a straight blade 7 and the rear guide vane is a twisted blade.

[0061] The present embodiment relates to a wing-shaped flow guide cover for improving the stall flow field inside a horizontal axial flow pump station. Specifically, the wing-shaped flow guide cover 9 is made of stainless steel.

[0062] like Figure 2 As shown, inside the horizontal axial flow pump station, there are several straight blades 7, which are evenly distributed on the side of the hub 8. The wing-shaped guide cover 9 is welded to the straight blades 7 of the horizontal axial flow pump station by welding, and the straight blades and the hub 8 are distributed inside the rim 6.

[0063] In particular, the wing-shaped air guide cover 9 of this embodiment is an annular structure as a whole, and a plurality of circular holes are evenly arranged on the wing-shaped air guide cover, such as Figure 3 shown.

[0064] The overall top view and left view of the wing-shaped fairing, such as Figure 4 As shown, when the water flowing out of the water inlet channel flows through the front guide vane, the incoming flow is divided into two streams of water. One stream flows into the impeller through the inner channel (located between the wing-shaped guide cover and the hub), and the other stream flows through the outer channel (located between the rim and the wing-shaped guide cover).

[0065] The cross section of the wing-shaped guide cover is a standard NACA0009 airfoil, as shown in the following figure: Figure 5 shown.

[0066] The present embodiment also relates to the above-mentioned optimized construction method of the wing-shaped fairing for improving the stall flow field in the axial flow pump, which is proposed based on the CFD-EP method and the PSO-BP neural network algorithm, and is implemented by CREO, ANSYS Mesh, ANSYS CFX and MATLAB software.

[0067] Specifically, the optimized construction method of the wing-shaped fairing is as follows: Figure 6As shown, the following steps are included:

[0068] Step 1: Construct a three-dimensional water model of the axial flow pump and select appropriate parameters for the optimal design of the wing-shaped guide cover. By selecting several sets of optimization parameter sample data, a three-dimensional water model of the front guide vane equipped with the wing-shaped guide cover can be drawn. The influence of different optimization parameters on the pump performance is analyzed through CFD (computational fluid dynamics) simulation, focusing on head, efficiency, time-averaged loss, pulsation loss and wall loss.

[0069] Step 2: Define the objective function to evaluate the pump performance of different optimization parameter combinations, including head, efficiency, time-averaged loss, pulsation loss, and wall loss. The larger the head and efficiency values, the smaller the time-averaged loss, pulsation loss, and wall loss values, indicating a better optimization effect, that is, the better the pump performance.

[0070] Step 3: Divide the optimized parameter sample data into training set, validation set and test set to provide training data for the neural network algorithm. The training set is used to train the neural network so that it can learn the relationship between different optimization parameters and the objective function value. The validation set is used to adjust the network's hyperparameters to avoid overfitting. The test set is used to verify the generalization ability of the neural network model and ensure the accuracy of the optimization results.

[0071] Step 4: The combination of PSO (particle swarm optimization) and BP (back propagation) neural network (PSO-BP algorithm) is used for regression optimization, that is, to find the optimal combination of wing-shaped fairing geometric parameters. The PSO algorithm searches for the global optimal solution by simulating the collective behavior of a swarm of particles. PSO can quickly find potential optimal solutions and effectively avoid local optimal problems. The BP neural network trains the neural network through the back propagation algorithm, enabling it to fit the complex nonlinear relationship between input and output. During the optimization process, the neural network is used to model the mapping relationship between the optimization parameters and the objective function.

[0072] Step 5: Based on the optimal parameter combination obtained from the PSO-BP algorithm, design and manufacture the actual wing-shaped guide cover, and install it on the front guide vane of the horizontal axial flow pump to improve the flow field characteristics of the axial flow pump. Verify the effect of the optimized guide cover through CFD simulation to ensure that the new wing-shaped guide cover can effectively reduce stall, improve efficiency, reduce losses, etc. If conditions permit, physical experiments can also be carried out to further verify the actual effect of the theoretical optimization results.

[0073] Specifically, the first step is to construct a three-dimensional water body model of a horizontal axial flow pump station. The optimization parameters of the airfoil guide fence are selected, and the optimization parameters are the geometric parameters of the airfoil guide cover, including: the airfoil guide cover length L1, the airfoil guide cover radius R1, the circular hole radius R2, the circular hole axial spacing L2 and the circular hole circumferential spacing L3.

[0074] In this embodiment, based on engineering experience and design requirements, the initial value ranges of the above-mentioned geometric parameters are set, the range of the shroud length L1 is [0.05D, 0.25D] (D is the impeller diameter), the range of the shroud radius R1 is [0.1D, 0.4D], the range of the circular hole radius R2 is [0.01D, 0.02D], the range of the circular hole axial spacing L2 is [0.01D, 0.05D], and the range of the circular hole circumferential spacing L3 is [5°, 25°] (expressed in angle).

[0075] 200 groups of sample data (i.e. parameter combinations) are extracted from the decision space by the Latin hypercube algorithm, and CREO software is used to automatically draw 200 different three-dimensional water models of leading guide vanes equipped with wing-shaped guide covers.

[0076] Specifically, the second step is to calculate the objective function values ​​of 200 sets of optimization solutions.

[0077] The automatic meshing of the water model was completed using ANSYS Mesh, and the numerical calculation was completed using CFX software. Further, through ANSYS Post post-processing and entropy production theory, 200 sets of sample data and corresponding objective function values ​​were obtained.

[0078] The objective function is the performance index of the horizontal axial flow pump station under small flow conditions, including head, efficiency, average loss, pulsation loss and wall loss.

[0079] Among them, the length L1 of the wing-shaped shroud directly affects the range of control of the shroud on the fluid flow. A longer shroud helps to guide the fluid more effectively, reduce flow turbulence, and thus improve the head and efficiency of the axial flow pump. However, an overly long shroud may also cause additional flow losses and increase wall losses.

[0080] The shroud radius R1 directly determines the radial installation position of the shroud, which in turn affects the flow distribution of the inner and outer channels of the pump. The outer channel is located between the wheel rim and the wing-shaped shroud, and the inner channel is located between the wing-shaped shroud and the wheel hub. A suitable shroud radius R1 will enable the fluid to transition more smoothly, reduce turbulence and flow separation, and reduce flow losses, thereby improving head and efficiency.

[0081] The hole radius R2, the hole axial spacing L2 and the hole circumferential spacing L3 affect the fluid passing area and flow smoothness in the inner and outer channels. Reasonable hole size helps to reduce the pressure gradient change when the fluid passes through the guide cover, thereby reducing the pulsation and time-averaged loss of the flow. Too small a hole radius may cause excessive flow velocity, generate eddy currents and additional losses; while too large a hole radius may reduce the guiding effect of the guide cover and affect the head.

[0082] Preferably, the objective function of this embodiment is calculated as follows:

[0083] H=(P out -P in ) / ρg

[0084] η=(ρgQH) / P

[0085]

[0086] Where H is the head, P out and P in are the total pressure at the pump station inlet and outlet, ρ is the density of water, g is the acceleration of gravity, η is the efficiency, Q is the flow rate, P is the power, is the average loss over time, is the pulsation loss, S W is the wall loss, v is the effective viscosity, t is the time, is x i The velocity in the direction, V is the volume, A is the area, ω is the turbulent frequency, k is the turbulent kinetic energy, τ w is the wall shear stress, u w is the velocity at the center of the first layer of grid on the wall; It is the integer symbol for partial derivatives, which means the partial derivative of the numerator with respect to the denominator after the symbol.

[0087] Specifically, the third step: train, verify and predict 200 groups of samples in turn through the PSO-BP neural network, and perform regression optimization on the optimization variables to obtain the most reasonable regression model.

[0088] The 200 sets of sample data are divided into training set, validation set and test set for parameter learning and verification of PSO-BP algorithm. The training set is used to determine learning parameters such as network weights; the validation set is only involved in determining hyperparameters such as the number of network layers, the number of network nodes, and the number of iterations; the test set does not participate in the learning process and is used to judge the accuracy of the prediction model. In this process, there is no intersection between the training set, validation set and test set, and the selection ratio is 7:1:2.

[0089] Specifically, the fourth step: perform numerical simulation verification on the obtained regression model (i.e., the optimal parameter combination).

[0090] The PSO-BP neural network algorithm is used to perform regression optimization on the optimization variables and objective functions of the wing-shaped fairing to obtain the best parameter combination scheme. Finally, the best parameter combination scheme is verified by numerical simulation.

[0091] Specifically, the fifth step is to construct a wing-shaped fairing based on the best parameter combination solution obtained. The optimized wing-shaped fairing is as follows: Figure 2 shown.

[0092] The wing-shaped guide cover is installed on the front guide vane of the horizontal axial flow pump to improve the stall flow field in the axial flow pump. The flow pattern without the wing-shaped guide cover, such as Figure 7 As shown, the flow pattern of the wing-shaped fairing after the embodiment is optimized is added, as shown in FIG. Figure 8 shown.

[0093] In summary, the present invention adopts computational fluid dynamics (CFD-EP) based on entropy production theory and a particle swarm optimization algorithm combined with a neural network (PSO-BP) of a back propagation algorithm to propose a low-cost, easy-to-install wing-shaped flow guide cover device and its optimization design method. Its core advantage is that by combining modern optimization algorithms (PSO and BP neural networks) and numerical simulation technology, it can find the best balance between multi-dimensional pump performance indicators and optimize the working efficiency and stability of the axial flow pump. The optimized construction method of the wing-shaped flow guide cover of the present invention not only accurately optimizes parameters through digital means, but also combines an efficient artificial intelligence optimization algorithm to find the optimal solution in a complex fluid mechanics environment, adapt to different working conditions, and improve the overall performance of the pump. This optimization design method is particularly suitable for the optimization of axial flow pumps under low flow conditions, and is an important means to improve the performance of pump stations.

[0094] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for optimizing the construction of a wing-shaped guide cover for improving the stall flow field in an axial flow pump, characterized in that: The steps include: S1. Construct a three-dimensional water body model of a horizontal axial flow pump station, select the optimization parameters of the wing-shaped guide cover, take several groups of optimization parameter sample data, draw corresponding three-dimensional water body models of several leading guide vanes equipped with the wing-shaped guide cover, and perform numerical analysis; S2. Construct an objective function and calculate the objective function value of each group of optimization parameter sample data; S3, dividing each group of sample data into a training set, a validation set, and a test set, and sequentially performing parameter learning and validation of the PSO-BP neural network algorithm; S4. Perform regression optimization on the optimization variables and objective functions of the wing-shaped fairing through the PSO-BP neural network algorithm to obtain the best optimization parameter combination, and perform numerical simulation verification on the best parameter combination scheme; S5. Based on the best parameter combination solution, a wing-shaped guide cover is constructed and installed on the front guide vane of the horizontal axial flow pump to improve the stall flow field in the axial flow pump.

2. The method for optimizing the construction of a wing-shaped guide cover for improving the stall flow field in an axial flow pump according to claim 1, characterized in that: The wing-shaped air guide cover is an annular structure, made of stainless steel, and is installed on the front guide vane by welding. A plurality of circular holes are evenly arranged on the surface of the wing-shaped air guide cover, and the cross section is a standard NACA0009 airfoil.

3. The method for optimizing the construction of a wing-shaped guide cover for improving the stall flow field in an axial flow pump according to claim 1, characterized in that: The horizontal axial flow pump comprises a water inlet channel, a front guide vane, an impeller, a rear guide vane and a water outlet channel which are connected in sequence. The front guide vane is a straight blade, and the rear guide vane is a twisted blade.

4. The method for optimizing the construction of a wing-shaped guide cover for improving the stall flow field in an axial flow pump according to claim 1, characterized in that: The optimization parameters described in step S1 are geometric variables of the airfoil fairing, including: airfoil fairing length L1, airfoil fairing radius R1, circular hole radius R2, circular hole axial spacing L2 and circular hole circumferential spacing L3, and the initial value range of the geometric parameters is set according to engineering requirements and experience.

5. The method for optimizing the construction of a wing-shaped guide cover for improving the stall flow field in an axial flow pump according to claim 3, characterized in that: In step S1, a three-dimensional water body model of a horizontal axial flow pump station is constructed, including the following sub-steps: S1.

1. Extract several groups of optimization parameter sample data through Latin hypercube algorithm; S1.2, using CREO software to automatically draw a three-dimensional water model of several leading guide vanes equipped with wing-shaped guide covers corresponding to the sample data; S1.

3. Automatically mesh the water model using ANSYS Mesh and perform numerical analysis using CFX software.

6. The method for optimizing the construction of a wing-shaped guide cover for improving the stall flow field in an axial flow pump according to claim 5, characterized in that: In step S2, the objective function value of each group of optimization parameter sample data is calculated, including the following sub-steps: S2.

1. Use ANSYS Mesh and ANSYA CFX software to complete the automatic meshing and numerical calculation of each water body model; S2.2, through ANSYS Post processing and entropy production theory, obtain each group of sample data and the corresponding objective function value; The objective function value is a performance index of the horizontal axial flow pump station under a small flow condition, including: head, efficiency, time-averaged loss, pulsation loss and wall loss.

7. The method for optimizing the construction of a wing-shaped guide cover for improving the stall flow field in an axial flow pump according to claim 6, characterized in that: In step S2.2, the relationship between sample data and corresponding objective function values ​​is as follows: Adjust the length L1 of the wing-shaped shroud to control the range of the shroud on the fluid flow, adjust the shroud radius R1 to determine the radial installation position of the shroud, control the flow distribution of the inner and outer channels of the pump, and improve the head and efficiency of the axial flow pump; The circular hole radius R2, the circular hole axial spacing L2 and the circular hole circumferential spacing L3 are adjusted to control the passage area and flow smoothness of the fluid in the inner and outer channels, reduce the pressure gradient change when the fluid passes through the guide cover, and reduce the pulsation loss and time-averaged loss of the flow.

8. The method for optimizing the construction of a wing-shaped guide cover for improving the stall flow field in an axial flow pump according to claim 6, characterized in that: In step S3, there is no intersection among the training set, validation set, and test set, and the ratio is 7:1:2; The training set is used to determine learning parameters such as network weights; the validation set is used to determine hyperparameters, and the test set does not participate in the learning process and is used to judge the accuracy of the prediction model; The hyperparameters include: the number of network layers, the number of network nodes, and the number of iterations.

9. The method for optimizing the construction of a wing-shaped guide cover for improving the stall flow field in an axial flow pump according to claim 8, characterized in that: In step S4, the optimization variables and the objective function of the wing-shaped fairing are regressively optimized by the PSO-BP neural network algorithm, which includes the following sub-steps: S4.

1. Determine the PSO algorithm structure: set the particle swarm size and initialize the position x of each particle i and speed v i , where position x i Represents the optimization variable of the wing-shaped fairing, and sets the particle position search space range as the upper and lower limits of the optimization variable; Set the inertia weight w, acceleration factors c1 and c2, and maximum number of iterations T of the PSO algorithm, calculate the initial fitness value of the particle, and evaluate it based on the prediction error of the neural network model; Initialize the global state and find the historical optimal position p of each particle i and the global optimal position g of the particle swarm; S4.

2. Use neural network error to construct fitness: Construct the objective function and use the neural network prediction error as the fitness function F, which is expressed as: Among them, y i is the actual objective function value, is the objective function value predicted by the neural network, and N is the number of samples; According to the current position of the particle, the neural network model is used to calculate the objective function value and update the fitness F(x i ); if the current fitness is better than the historical optimal fitness, then update the corresponding optimal position p i and the global optimal position g; S4.

3. Perform neural network training: output the best parameter combination of the trained belt, and obtain the optimal particle position g as the best parameter combination of the wing-shaped fairing.

10. The method for optimizing the construction of a wing-shaped guide cover for improving the stall flow field in an axial flow pump according to claim 9, characterized in that: In step S4.3, the neural network training includes the following sub-steps: S4.3.

1. Training BP neural network: Use the particle position obtained by PSO optimization as the input of the neural network and the objective function value as the output to train the BP neural network; take the mean square error MSE as the error function of the network, adjust the weight and bias of the neural network to minimize the error; S4.3.

2. Iteratively update particle position and velocity: After each training and validation, update the particle position and velocity according to the following formula: v i (t+1)=wv i (t)+c1r1(p i -x i (t))+c2r2(g-x i (t)) r1x i (t+1)=x i (t)+v i (t+1) Among them, r1 and r2 are random numbers in [0, 1], x i (t), v i (t) is the position and velocity of the particle at time t; S4.3.

3. Verify the neural network model: Use the validation set data to verify the performance of the neural network. If the validation set error is lower than the preset threshold ε, stop training and output the current best parameter combination; otherwise, continue training; S4.3.

4. Output the best parameter combination: After multiple iterations, the optimal particle position g is obtained as the best parameter combination of the wing-shaped fairing.

Images