Hydroelectric power generation system water-machine-electricity coupling transition process refined modeling simulation method, device and medium
By improving the transfer function method and convolutional neural network to construct a refined model of the hydropower generation system, the simulation accuracy and complexity problems of the water-machine-electric coupling transition process in the existing technology are solved, and high-precision simulation of the dynamic response of water-machine-electric coupling and flexible modeling switching are realized.
Patent Information
- Application Number
- CN202410981200.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-22
- Publication Date
- 2024-11-15
AI Technical Summary
Existing mathematical modeling of hydropower systems is inadequate in terms of simulation accuracy, time scale, and model complexity, and cannot accurately simulate the water-mechanical-electric coupling transition process, resulting in inaccurate calculation results and model simplification.
An improved transfer function method is used to construct a flow channel system model. A convolutional neural network is used to interpolate the full characteristic curve of the turbine. A multi-physical quantity coupling model of the turbine, governor, synchronous generator, excitation system and power grid is established to achieve refined modeling of the water-machine-electric coupling transition process.
It improves the simulation accuracy of hydropower systems, realizes the simulation of dynamic response characteristics of water-mechanical-electric coupling, has high flexibility and high precision modeling capabilities, and can quickly switch between different operating modes.
Smart Images

Figure CN118965646A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of hydropower generation, and more specifically, relates to a new modeling method for a hydropower generation system, which is used to simulate the water-machine-electrical multi-physical quantity coupling transition process in the process of changing the operating conditions of the hydropower generation system. Background Art
[0002] Hydropower is very important for the renewability, sustainability, and environmental friendliness of energy systems, and it plays an important role in the current energy market. Hydropower units need to frequently adjust their operating conditions, and fluctuations will inevitably occur during the transition process from one operating condition to another, and in some cases, accidents may even occur. Numerical simulation can provide relevant support for the actual operation of hydropower units. By simulating the corresponding operating conditions, the control parameters of the hydropower system can be changed, the adjustment method can be optimized, etc., so as to avoid dangerous conditions, increase the operating life of the unit and improve economic benefits. However, in many previous scientific research or engineering practices, many complex links of the hydropower system have been simplified to a certain extent, such as ignoring the nonlinear links in the turbine and speed governor system, ignoring the high-order terms of the system transfer function, etc. These simplifications may lead to ignoring some dangerous factors when simulating some operating conditions of the hydropower system and lead to inaccurate calculation results. Therefore, numerical simulation of the water-machine-electric multi-physical quantity coupling transition process of the hydropower system is very important.
[0003] Summarizing the existing background technology, the research on mathematical modeling and simulation of hydropower generation systems has the following main deficiencies: 1) Simulation accuracy: Currently, the transfer function method is commonly used in the mathematical modeling of hydropower generation systems, but most models have made a lot of simplifications to the flow system, and therefore cannot meet the needs of accurate simulation of the flow system; 2) Time scale: The conventional characteristic line method used to calculate the water hammer pressure in the flow system has a large step size and requires a fixed step size calculation, and cannot be numerically simulated simultaneously with the model on the electrical side; 3) Model complexity: Many studies are only focused on the refined modeling of the hydraulic machinery side, or the refined modeling of the electrical side model alone. There is currently a lack of refined models that consider the water-machine-electrical coupling of the hydropower generation system. Summary of the invention
[0004] The present invention mainly solves and improves the defects of the prior art, and provides a method for fine-grained modeling and simulation of the water-machine-electric coupling transition process of a hydroelectric power generation system. The method constructs a fine-grained mathematical model of the hydroelectric power generation system, wherein the flow channel system model can simulate the different connection modes of multiple pipes and the surge chamber in the flow channel system is taken into account, and the turbine model can simulate the full operating conditions. The electrical side model adopts a three-phase circuit model. The method can simultaneously simulate the multi-physical quantity transition process of the hydraulic machinery side and the electrical side, and can fully reflect the water-machine-electric coupling dynamic characteristics of the system.
[0005] In order to achieve the above-mentioned purpose, a refined modeling and simulation method for the water-machine-electricity coupling transition process of a hydropower generation system is provided, comprising the following steps:
[0006] Step (1): Establish a flow channel system model based on the improved transfer function method
[0007] Step 1: Construct a single pipe model using the basic theory of transfer function method
[0008] Step 1.1: According to the basic water hammer equation, including the continuity equation and the momentum equation, the non-steady flow (transient flow) in the pipeline system should follow the three basic laws of fluid mechanics: conservation of mass, conservation of momentum and conservation of energy. Without considering heat exchange, the continuity equation and momentum equation can be used to describe the pipeline system. Therefore, under the premise of one-dimensional flow, the water hammer equation of non-prismatic pressurized pipeline can be listed as:
[0009]
[0010]
[0011] In the formula: x is the coordinate distance along the pipeline axis, which can start from any starting point; α is the angle between the line connecting the centroids of each section of the pipeline and the horizontal plane; H is the pressure head; Q is the flow rate of the pipeline section; A is the cross-sectional area of the pipeline; a is the wave velocity of the water flow; λ is the friction coefficient.
[0012] Step 1.2: Simplify the continuity equation and momentum equation of the pipeline and ignore some of the smaller terms. The continuity equation and momentum equation of pressurized non-steady flow of water can be simplified to the following two equations: H x , H t , Q x , Q t They are the corresponding partial differential forms:
[0013]
[0014]
[0015] The instantaneous pressure head H in the above formula can be divided into the average pressure head and oscillating pressure head h′. Similarly, the instantaneous flow is also divided into average flow and oscillating flow, as shown in the following formula:
[0016]
[0017]
[0018] Similar to the above two equations, the expression in partial derivative form is written as:
[0019]
[0020]
[0021]
[0022]
[0023] The steady state conditions are:
[0024]
[0025] After sorting, we get the following two formulas:
[0026]
[0027]
[0028] In the above formula: R is the linear resistance per unit length, turbulent flow Laminar flow
[0029] Define C as the flow capacity and L as the flow velocity of the fluid in the pipe, respectively, where: The following two formulas are obtained:
[0030] h′ x +Lq′ t +Rq′=0
[0031] q′ x +Ch′ t =0
[0032] Use the method of separation of variables to solve the above two equations, and let: h′(x,t)=H(x)e st ,q′(x,t)=Q(x)e st , where H(x) and Q(x) are the complex water head and complex flow rate at the distance x in the pipeline, respectively. U and complex flow Q U, we can get the expressions of the oscillating head and oscillating flow at the end of the pipe under the boundary conditions at the beginning of the pipe, as follows:
[0033] H(x)=H U cosh(γx)-Z C Q U sinh(γx)
[0034]
[0035] In the above formula: γ is the propagation constant, It is a complex constant that does not depend on x and t. s is the complex frequency or Laplace variable, s = σ + iω, σ is the attenuation factor, ω is the circular frequency; Z C is the pipeline characteristic impedance, Z C =γ / Cs, which is also a complex constant independent of x and t.
[0036] Set H at the end of the pipe x = l D and Q D Substituting the above formula, we can get the expressions of the oscillating head and oscillating flow at the end of the pipe represented by the boundary conditions at the beginning of the pipe:
[0037] H D =H U cosh(γl)-Z c Q U sinh(γl)
[0038]
[0039] Similarly, the expressions for the oscillating head and oscillating flow at the beginning of the pipe represented by the boundary conditions at the end of the pipe can also be obtained:
[0040] H U =H D cosh(γl)+Q D Z c sinh(γl)
[0041]
[0042] Step 1.3: Define the hydraulic impedance in the pipe model:
[0043]
[0044] The hydraulic impedance Z(x) is a complex function that does not depend on time. It is a function of the cross-sectional position x. From the above formula, we can get the hydraulic impedance Z at the head of the pipeline: U and the hydraulic impedance Z at the end D The relationship between:
[0045]
[0046]
[0047] Step 1.4: Get the expression of the pipeline model, considering the head loss along the pipeline, then The last term is defined as the quantized loss coefficient F, Z C =a / gA, Taylor expansion of the hyperbolic sine and cosine functions in the above formula can yield the following four equations:
[0048]
[0049]
[0050]
[0051]
[0052] Step 2: Construct a surge chamber model. The transfer function between the flow deviation and the head deviation of the surge chamber is:
[0053]
[0054] f t is the friction coefficient of the diversion tunnel; T et (s) is the elastic water hammer coefficient of the diversion tunnel; T s (s) is the overflow time of the surge tank, also known as the regulating time constant of the surge chamber; Z t is the hydraulic impedance of the diversion tunnel; where tanh(T et s) has an infinite factor expansion:
[0055]
[0056] Step 2.2: Keep the appropriate number of terms in the infinite factor expansion, take n = 0, and use the lumped parameter equivalent to approximate the transfer function;
[0057]
[0058] Step 3: Compare accuracy with other methods;
[0059] Step 3.1: Comparison of the theoretical transfer matrices of each method:
[0060] The solution to the water hammer model in transient flow can be solved by using a hyperbolic partial differential equation, which can be solved by the characteristic line method. Ignoring the head loss term, the transfer matrix of the hyperbolic partial differential equation is shown as follows, where: x = lj / a j ·s,Z C =gA j / a j .
[0061]
[0062] The equivalent circuit method is also a commonly used method for calculating water hammer pressure in pipelines. It treats the transient flow in the pipeline system as an RLC circuit. The impedance R can be used to simulate the head loss term, the inductive reactance L can be used to simulate the inertia term, and the capacitive reactance C can be used to simulate the compression term.
[0063] The transfer matrix of the central equivalent circuit is:
[0064]
[0065] Transfer matrix of distributed equivalent circuit:
[0066]
[0067] In the above two formulas: x = l j / a j ·s,
[0068] The transfer matrix form corresponding to the transfer function method (T e = l j / a j ):
[0069]
[0070] The above transfer matrices can be written uniformly in the following form:
[0071]
[0072] M in the transfer matrix of different methods 11 ,M 12 ,M 21 ,M 22 Four items are compared, taking a pipe with a length of 10m, a diameter of 1m, and a wave velocity of 1000m / s as the comparison object, ignoring the head loss in the pipe. Divide the frequency by a critical frequency (f c =a j / 10l j ) is normalized, and the wavelength corresponding to this frequency is 10 times the tube length. The comparison results are shown in Figure 1. From the transfer matrix in the figure, it can be seen that the first term M of the central and distributed equivalent circuit methods and the transfer function method 11 and the fourth term M 22The expressions of these three methods are completely consistent, so the frequency domain diagrams corresponding to these three methods are also completely consistent. c The frequency band of M is highly consistent with the hyperbolic equation in the continuous model. 12 and the third term M 21 , the transfer function method is consistent with the distributed equivalent circuit and the central equivalent circuit respectively. c The frequency band can also have a good fit with the hyperbolic equation. In general, compared with the transfer matrix of the continuous partial differential equation of the water hammer model, the transfer matrices of the two discrete methods, the equivalent circuit method and the transfer function method, can be lower than the critical frequency f. c The frequency bands of the two discrete methods have a high degree of agreement, and the accuracy of the two discrete methods is similar.
[0073] Step 3.2: Comparison of simulation results in time and frequency domain of each method:
[0074] For a pipe with a length of 118.05m and a diameter of 2.06m, the time domain simulation results and spectrum characteristics of the water head at the end of the pipe section are shown in Figure 2. From the time domain simulation diagram, it can be seen that the results of the transfer function method and the characteristic line method are very consistent. From the spectrum analysis diagram, it can be seen that for frequencies less than the critical frequency f c =1.04Hz water hammer wave, the frequency spectrum characteristics obtained by the two methods are basically the same, but for the water hammer wave with frequency greater than the critical frequency, there is a certain gap in the comparison of its frequency spectrum characteristics.
[0075] Step 4: Improve the transfer function to achieve higher simulation accuracy:
[0076] When Taylor expansion is performed on the hyperbolic sine function and the hyperbolic cosine function to retain higher order terms, as shown in the following formula, better fitting accuracy can be obtained, thereby achieving higher frequencies (the critical frequency is about 2f c )Simulation of water shock waves.
[0077]
[0078] Step (2): Use convolutional neural network to interpolate and extend the full characteristic curve of the turbine
[0079] The general process of using Convolutional Neural Network (CNN) to interpolate and fit part of the full characteristic curve of the turbine and obtain the full characteristic curve data of other areas is as follows:
[0080] Step 1: Data collection and preparation: Collect some full characteristic curve data of the turbine, including input variables (speed, flow, etc.) and output variables (power, efficiency, etc.). Divide the data into training set and validation set to ensure the adequacy and diversity of the data.
[0081] Step 2: Build a CNN model of the full characteristic curve of the turbine: Use a deep learning framework (TensorFlow or PyTorch) to build a convolutional neural network model of the full characteristic curve of the turbine. Design the network architecture, including convolutional layers, pooling layers, and fully connected layers, as well as components such as activation functions and regularization. Choose the appropriate network depth and width based on the complexity of the problem and the size of the data.
[0082] Step 3: Train the CNN model of the full characteristic curve of the turbine: Use the data of the training set to input the model, and perform forward propagation and back propagation to optimize the weights and biases of the network. Use its loss function to measure the difference between the predicted value and the true value, and use optimization algorithms such as gradient descent to minimize the loss. Through multiple rounds of iterative training, the performance of the model is gradually improved.
[0083] Step 4: Model evaluation and tuning: Use the validation set data to evaluate the trained CNN model of the turbine full characteristic curve to evaluate the performance and generalization ability of the model. According to the evaluation results, adjust the architecture and hyperparameters of the CNN model, i.e., the size of the filter, the number of convolution kernels, etc., to further improve the model's fitting ability and prediction accuracy.
[0084] Step 5: Interpolation and prediction: Use the trained CNN model to interpolate and predict the input variable data in other regions to obtain the corresponding output variable data. Enter the new input variable data (ω and Q), and through the forward propagation of the CNN model, the corresponding output variable data (P and η) can be predicted.
[0085] Step (3): Establish a turbine model and a governor model based on the turbine full characteristic curve
[0086] Step 1: Extract characteristic parameters based on the full characteristic curve of the turbine, namely the guide vane opening y, unit speed n 11 , Unit moment M 11 and unit flow Q 11 .
[0087]
[0088] Where D1 is the turbine inlet diameter;
[0089] Step 2: Interpolation to obtain turbine flow and torque
[0090] Step 2.1: Take the per unit value y of the guide vane opening output by the speed regulator as input, calculate the actual stroke of the relay according to the base value, and calculate the guide vane opening a by looking up the table;
[0091] Step 2.2: Take the per unit water head value h at the end of the flow channel system output pipe as input, calculate the actual water head H based on the base value, and calculate the unit speed n′1;
[0092] Step 2.3: Calculate the unit flow rate Q by interpolating the flow characteristic curve based on the guide vane opening a (Step 2.1) and the unit speed n′1 (Step 2.2) 11 ;
[0093] Step 2.4: Calculate the actual flow rate Q, obtain and output the relative flow deviation value q;
[0094] Step 2.5: Interpolate the pump-turbine torque on the turbine characteristic curve
[0095] Step 3: Build the governor model
[0096] Step 3.1: Using the parallel PID regulation rule, after the signal is output, it is first amplified by the auxiliary relay, and then the hard feedback link performs negative feedback regulation on it, and then it is input into the main relay to control the guide vane opening. Assume that the auxiliary relay time constant T, which is smaller in magnitude, is ignored. y1 ;
[0097] Step 3.2: If the input value is the frequency setting and the feedback value is the unit frequency, the governor model under the frequency control mode is obtained:
[0098]
[0099] where b p is the permanent slip coefficient; T y is the main relay time constant; K p is the proportional gain; K i is the integral gain; K d is the differential gain; ω is the frequency deviation;
[0100] Step 3.3: If the input value is the power given and the feedback value is the unit mechanical power, the governor model under the power control mode is obtained:
[0101]
[0102] Where P is the power deviation;
[0103] Step (4): Establish synchronous generator, excitation system, transformer and its power grid model
[0104] Step 1: Establish the sixth-order standard parameter model of synchronous generator motor. Based on the equal power Park transformation (where the q-axis leads the d-axis by 90°, and the d-axis and the phase reference axis are at an angle, i.e., the a-phase angle), and after per-unit value processing, the mathematical expression of the synchronous generator motor considering steady-state, transient and sub-transient characteristics is:
[0105]
[0106] Where T a is the unit inertia time constant, P m is the mechanical power, P e is the electromagnetic power, D is the damping coefficient, δ is the rotor angle (or power angle) relative to the infinite busbar, T′ do ,T′ qo are the transient open circuit time constants of the d-axis and q-axis, T″ do ,T″ qo are the d-axis and q-axis subtransient open-circuit time constants, E d ,E q are the d-axis and q-axis components of the steady-state internal electromotive force, E′ d ,E′ q are the d-axis and q-axis components of the transient internal electromotive force, E″ d ,E″ q are the d-axis and q-axis components of the subtransient internal electromotive force, E f is the excitation electromotive force, I d ,I q are the d-axis and q-axis components of the armature current, respectively, and X d ,X′ d ,X″ d are d-axis synchronous, transient and subtransient reactances, respectively, and X q ,X′ q ,X″ q They are q-axis synchronous, transient and sub-transient reactances respectively; in addition, for variable speed units, since the reactance is proportional to the speed, it is necessary to calculate X x =(ω s +Δω)L x Reactance, where ω s The synchronous speed.
[0107] In addition, the armature voltage equation is:
[0108]
[0109] Where V d ,V q are the direct-axis and quadrature-axis voltage components at the generator end, R s is the resistance of the generator armature winding;
[0110] The electromagnetic active power and reactive power expressions of the generator are:
[0111]
[0112] In the electrical subsystem, the excitation system, transformer and its power grid model are modeled using modules in Simscape Electrical, which will not be described in detail in this description.
[0113] Step (5): Connect the above model modules to build the overall model and set the model parameters
[0114] Step 1: The input of the speed regulator module is P m and P ref , output y is connected to the turbine module;
[0115] Step 2: The governor output y is used as the input of the turbine module, and the turbine outputs mechanical power P m Connected to synchronous generator module;
[0116] Step 3: The flow rate of the turbine is used as the input of the flow channel system, and the output head of the flow channel system is used as the input of the turbine module;
[0117] Step 4: The input of the excitation system module is V ref 、V d and V q , the output is V f connected to a synchronous motor generator module;
[0118] Step 5: Excitation system output V f , turbine output mechanical power P m As the input of the synchronous generator, the synchronous generator outputs ω as the input of the turbine, and the synchronous generator outputs the stator voltage V d and V q As the input of the excitation system, the synchronous generator outputs electromagnetic active power as the input of the speed regulator. In addition, the three-phase circuit of the synchronous generator is connected to the transformer, and multiple units are connected through the transformer before being connected to the grid model;
[0119] Step 6: Set all step model parameters, including hydraulic machinery side system parameters and electrical side system parameters. Note that except for the time constant which is a nominal value (s), all other parameters are taken as per unit values.
[0120] Step (6): Change the module connection logic according to the control model and set the control command for simulation Step 1: When the unit is in the grid-connected mode, set the load parameter of the grid model to 1e12;
[0121] Step 2: When the unit is in isolated grid mode, set the load parameters of the grid model to be consistent with the unit power;
[0122] Step 3: When the unit is in load shedding state, disconnect the circuit breaker of the circuit model;
[0123] Step 4: Check the time domain simulation results of each physical quantity;
[0124] Compared with the prior art, the present invention has the following advantages and effects:
[0125] (1) The present invention designs a refined modeling and simulation method for the water-machine-electric coupling transition process of a hydropower generation system. The improved transfer function method model is used. The simulation has high accuracy in the hydraulic side and can realize variable step-size simulation. This method is connected with the electrical side model to achieve high-precision simulation accuracy of the hydraulic side and the electrical side, and can effectively simulate the dynamic response characteristics of the water-machine-electric coupling of the hydropower generation system.
[0126] (2) The modeling method proposed in the present invention can achieve rapid switching between different modes of the hydropower generation system by changing the connection logic between subsystem modules, and has high modeling flexibility. BRIEF DESCRIPTION OF THE DRAWINGS
[0127] Figure 1a to Figure 1d Comparison of various frequency domain characteristics in the transfer matrix between the transfer function method adopted by the present invention and other methods;
[0128] Figure 2a to Figure 2b The time-frequency domain simulation comparison of the transfer function method used in the present invention and other methods for calculating the water hammer phenomenon of a single pipeline;
[0129] Figure 3 is a layout diagram of the flow channel system of the present invention;
[0130] Figure 4 A model block diagram of the flow channel system of the present invention;
[0131] Figure 5a to Figure 5c The characteristic curve of the water turbine of the present invention and the corresponding relationship between the servomotor stroke and the guide vane opening;
[0132] Figure 6 It is a connection diagram of an example model of the present invention;
[0133] Figures 7a to 7f It is the simulation result diagram of the working condition of the example of the present invention.
[0134] Figure 8a to Figure 8b It is a structural diagram of the simulation system of the present invention. DETAILED DESCRIPTION
[0135] In order to make the purpose, technical scheme and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in each step of the present invention described below can be applied to simulate any other simulation object with a reasonable parameter combination.
[0136] The present invention relates to a refined modeling and simulation method for the water-machine-electricity coupling transition process of a hydroelectric power generation system, wherein the system comprises seven parts: a station flow channel system, a turbine, a speed regulator, an excitation system, a synchronous generator, a transformer and a power grid. The speed regulator has a power control mode and a frequency control mode. The purpose of the present invention is to propose a new simulation modeling method for the system, so as to dynamically simulate the water-machine-electricity coupling transition process of the hydroelectric power generation system and simulate the dynamic response law of multiple physical quantities of the system.
[0137] First, the principle of the method of the present invention is introduced.
[0138] The steps include:
[0139] Step (1): Establish a flow channel system model based on the improved transfer function method
[0140] Step 1: Construct a single pipe model using the basic theory of transfer function method
[0141] Step 1.1: According to the basic water hammer equation, including the continuity equation and the momentum equation, the non-steady flow (transient flow) in the pipeline system should follow the three basic laws of fluid mechanics: conservation of mass, conservation of momentum and conservation of energy. Without considering heat exchange, the continuity equation and momentum equation can be used to describe the pipeline system. Therefore, under the premise of one-dimensional flow, the water hammer equation of non-prismatic pressurized pipeline can be listed as:
[0142]
[0143]
[0144] In the formula: x is the coordinate distance along the pipeline axis, which can start from any starting point; α is the angle between the line connecting the centroids of each section of the pipeline and the horizontal plane; H is the pressure head; Q is the flow rate of the pipeline section; A is the cross-sectional area of the pipeline; a is the wave velocity of the water flow; λ is the friction coefficient.
[0145] Step 1.2: Simplify the continuity equation and momentum equation of the pipeline and ignore some of the smaller terms. The continuity equation and momentum equation of pressurized non-steady flow of water can be simplified to the following two equations: H x , H t , Q x , Qt They are the corresponding partial differential forms:
[0146]
[0147]
[0148] The instantaneous pressure head H in the above formula can be divided into the average pressure head and oscillating pressure head h′. Similarly, the instantaneous flow is also divided into average flow and oscillating flow, as shown in the following formula:
[0149]
[0150]
[0151] Similar to the above two equations, the expression in partial derivative form is written as:
[0152]
[0153]
[0154]
[0155]
[0156] The steady state conditions are:
[0157]
[0158] After sorting, we get the following two formulas:
[0159]
[0160]
[0161] In the above formula: R is the linear resistance per unit length, turbulent flow Laminar flow
[0162] Define C as the flow capacity and L as the flow velocity of the fluid in the pipe, respectively, where: The following two formulas are obtained:
[0163] h′ x +Lq′ t +Rq′=0
[0164] q′ x +Ch′ t =0
[0165] Use the method of separation of variables to solve the above two equations, and let: h′(x,t)=H(x)est ,q′(x,t)=Q(x)e st , where H(x) and Q(x) are the complex water head and complex flow rate at the distance x in the pipeline, respectively. U and complex flow Q U , we can get the expressions of the oscillating head and oscillating flow at the end of the pipe under the boundary conditions at the beginning of the pipe, as follows:
[0166] H(x)=H U cosh(γx)-Z C Q U sinh(γx)
[0167]
[0168] In the above formula: γ is the propagation constant, It is a complex constant that does not depend on x and t. s is the complex frequency or Laplace variable, s = σ + iω, σ is the attenuation factor, ω is the circular frequency; Z C is the pipeline characteristic impedance, Z C =γ / Cs, which is also a complex constant independent of x and t.
[0169] Set H at the end of the pipe x = l D and Q D Substituting the above formula, we can get the expressions of the oscillating head and oscillating flow at the end of the pipe represented by the boundary conditions at the beginning of the pipe:
[0170] H D =H U cosh(γl)-Z c Q U sinh(γl)
[0171]
[0172] Similarly, the expressions for the oscillating head and oscillating flow at the beginning of the pipe represented by the boundary conditions at the end of the pipe can also be obtained:
[0173] H U =H D cosh(γl)+Q D Z c sinh(γl)
[0174]
[0175] Step 1.3: Define the hydraulic impedance in the pipe model:
[0176]
[0177] The hydraulic impedance Z(x) is a complex function that does not depend on time. It is a function of the cross-sectional position x. From the above formula, we can get the hydraulic impedance Z at the head of the pipeline: U and the hydraulic impedance Z at the end D The relationship between:
[0178]
[0179]
[0180] Step 1.4: Get the expression of the pipeline model, considering the head loss along the pipeline, then The last term is defined as the quantized loss coefficient F, Z C =a / gA, Taylor expansion of the hyperbolic sine and cosine functions in the above formula can yield the following four equations:
[0181]
[0182]
[0183]
[0184]
[0185] Step 2: Construct a surge chamber model. The transfer function between the flow deviation and the head deviation of the surge chamber is:
[0186]
[0187] f t is the friction coefficient of the diversion tunnel; T et (s) is the elastic water hammer coefficient of the diversion tunnel; T s (s) is the overflow time of the surge tank, also known as the regulating time constant of the surge chamber; Z t is the hydraulic impedance of the diversion tunnel; where tanh(T et s) has an infinite factor expansion:
[0188]
[0189] Step 2.2: Keep the appropriate number of terms in the infinite factor expansion, take n = 0, and use the lumped parameter equivalent to approximate the transfer function;
[0190]
[0191] Step 3: Compare accuracy with other methods;
[0192] Step 3.1: Comparison of the theoretical transfer matrices of each method:
[0193] The solution to the water hammer model in transient flow can be solved by using a hyperbolic partial differential equation, which can be solved by the characteristic line method. Ignoring the head loss term, the transfer matrix of the hyperbolic partial differential equation is shown as follows, where: x = l j / a j ·s,Z C =gA j / a j .
[0194]
[0195] The equivalent circuit method is also a commonly used method for calculating water hammer pressure in pipelines. It treats the transient flow in the pipeline system as an RLC circuit. The impedance R can be used to simulate the head loss term, the inductive reactance L can be used to simulate the inertia term, and the capacitive reactance C can be used to simulate the compression term.
[0196] The transfer matrix of the central equivalent circuit is:
[0197]
[0198] Transfer matrix of distributed equivalent circuit:
[0199]
[0200] In the above two formulas: x = l j / a j ·s,
[0201] The transfer matrix form corresponding to the transfer function method (T e = l j / a j ):
[0202]
[0203] The above transfer matrices can be written uniformly in the following form:
[0204]
[0205] M in the transfer matrix of different methods 11 ,M 12 ,M 21 ,M 22 Four items are compared, taking a pipe with a length of 10m, a diameter of 1m, and a wave velocity of 1000m / s as the comparison object, ignoring the head loss in the pipe. Divide the frequency by a critical frequency (f c =a j / 10lj ) is normalized, and the wavelength corresponding to this frequency is 10 times the tube length. The comparison results are shown in Figure 1. From the transfer matrix in the figure, it can be seen that the first term M of the central and distributed equivalent circuit methods and the transfer function method 11 and the fourth term M 22 The expressions of these three methods are completely consistent, so the frequency domain diagrams corresponding to these three methods are also completely consistent. c The frequency band of M is highly consistent with the hyperbolic equation in the continuous model. 12 and the third term M 21 , the transfer function method is consistent with the distributed equivalent circuit and the central equivalent circuit respectively. c The frequency band can also have a good fit with the hyperbolic equation. In general, compared with the transfer matrix of the continuous partial differential equation of the water hammer model, the transfer matrices of the two discrete methods, the equivalent circuit method and the transfer function method, can be lower than the critical frequency f. c The frequency bands of the two discrete methods have a high degree of agreement, and the accuracy of the two discrete methods is similar.
[0206] Step 3.2: Comparison of simulation results in time and frequency domain of each method:
[0207] For a pipe with a length of 118.05m and a diameter of 2.06m, the time domain simulation results and spectrum characteristics of the water head at the end of the pipe section are shown in Figure 2. From the time domain simulation diagram, it can be seen that the results of the transfer function method and the characteristic line method are very consistent. From the spectrum analysis diagram, it can be seen that for frequencies less than the critical frequency f c =1.04Hz water hammer wave, the frequency spectrum characteristics obtained by the two methods are basically the same, but for the water hammer wave with frequency greater than the critical frequency, there is a certain gap in the comparison of its frequency spectrum characteristics.
[0208] Step 4: Improve the transfer function to achieve higher simulation accuracy:
[0209] When Taylor expansion is performed on the hyperbolic sine function and the hyperbolic cosine function to retain higher order terms, as shown in the following formula, better fitting accuracy can be obtained, thereby achieving higher frequencies (the critical frequency is about 2f c )Simulation of water shock waves.
[0210]
[0211] Step (2): Use convolutional neural network to interpolate and extend the full characteristic curve of the turbine
[0212] The general process of using Convolutional Neural Network (CNN) to interpolate and fit part of the full characteristic curve of the turbine and obtain the full characteristic curve data of other areas is as follows:
[0213] Step 1: Data collection and preparation: Collect some full characteristic curve data of the turbine, including input variables (speed, flow, etc.) and output variables (power, efficiency, etc.). Divide the data into training set and validation set to ensure the adequacy and diversity of the data.
[0214] Step 2: Build a CNN model of the full characteristic curve of the turbine: Use a deep learning framework (TensorFlow or PyTorch) to build a convolutional neural network model of the full characteristic curve of the turbine. Design the network architecture, including convolutional layers, pooling layers, and fully connected layers, as well as components such as activation functions and regularization. Choose the appropriate network depth and width based on the complexity of the problem and the size of the data.
[0215] Step 3: Train the CNN model of the full characteristic curve of the turbine: Use the data of the training set to input the model, and perform forward propagation and back propagation to optimize the weights and biases of the network. Use its loss function to measure the difference between the predicted value and the true value, and use optimization algorithms such as gradient descent to minimize the loss. Through multiple rounds of iterative training, the performance of the model is gradually improved.
[0216] Step 4: Model evaluation and tuning: Use the validation set data to evaluate the trained CNN model of the turbine full characteristic curve to evaluate the performance and generalization ability of the model. According to the evaluation results, adjust the architecture and hyperparameters of the CNN model, i.e., the size of the filter, the number of convolution kernels, etc., to further improve the model's fitting ability and prediction accuracy.
[0217] Step 5: Interpolation and prediction: Use the trained CNN model to interpolate and predict the input variable data in other regions to obtain the corresponding output variable data. Enter the new input variable data (ω and Q), and through the forward propagation of the CNN model, the corresponding output variable data (P and η) can be predicted.
[0218] Step (3): Establish a turbine model and a governor model based on the turbine full characteristic curve
[0219] Step 1: Extract characteristic parameters based on the full characteristic curve of the turbine, namely the guide vane opening y, unit speed n 11 , Unit moment M 11 and unit flow Q 11 .
[0220]
[0221] Where D1 is the turbine inlet diameter;
[0222] Step 2: Interpolation to obtain turbine flow and torque
[0223] Step 2.1: Take the per unit value y of the guide vane opening output by the speed regulator as input, calculate the actual stroke of the relay according to the base value, and calculate the guide vane opening a by looking up the table;
[0224] Step 2.2: Take the per unit water head value h at the end of the flow channel system output pipe as input, calculate the actual water head H based on the base value, and calculate the unit speed n′1;
[0225] Step 2.3: Calculate the unit flow rate Q by interpolating the flow characteristic curve based on the guide vane opening a (Step 2.1) and the unit speed n′1 (Step 2.2) 11 ;
[0226] Step 2.4: Calculate the actual flow rate Q, obtain and output the relative flow deviation value q;
[0227] Step 2.5: Interpolate the pump-turbine torque on the turbine characteristic curve
[0228] Step 3: Build the governor model
[0229] Step 3.1: Using the parallel PID regulation rule, after the signal is output, it is first amplified by the auxiliary relay, and then the hard feedback link performs negative feedback regulation on it, and then it is input into the main relay to control the guide vane opening. Assume that the auxiliary relay time constant T, which is smaller in magnitude, is ignored. y1 ;
[0230] Step 3.2: If the input value is the frequency setting and the feedback value is the unit frequency, the governor model under the frequency control mode is obtained:
[0231]
[0232] where b p is the permanent slip coefficient; T y is the main relay time constant; K p is the proportional gain; K i is the integral gain; K d is the differential gain; ω is the frequency deviation;
[0233] Step 3.3: If the input value is the power given and the feedback value is the unit mechanical power, the governor model under the power control mode is obtained:
[0234]
[0235] Where P is the power deviation;
[0236] Step (4): Establish synchronous generator, excitation system, transformer and its power grid model
[0237] Step 1: Establish the sixth-order standard parameter model of synchronous generator motor. Based on the equal power Park transformation (where the q-axis leads the d-axis by 90°, and the d-axis and the phase reference axis are at an angle, i.e., the a-phase angle), and after per-unit value processing, the mathematical expression of the synchronous generator motor considering steady-state, transient and sub-transient characteristics is:
[0238]
[0239] Where T a is the unit inertia time constant, P m is the mechanical power, P e is the electromagnetic power, D is the damping coefficient, δ is the rotor angle (or power angle) relative to the infinite busbar, T′ do ,T′ qo are the transient open circuit time constants of the d-axis and q-axis, T″ do ,T″ qo are the d-axis and q-axis subtransient open-circuit time constants, E d ,E q are the d-axis and q-axis components of the steady-state internal electromotive force, E′ d ,E′ q are the d-axis and q-axis components of the transient internal electromotive force, E″ d ,E″ q are the d-axis and q-axis components of the subtransient internal electromotive force, E f is the excitation electromotive force, I d ,I q are the d-axis and q-axis components of the armature current, respectively, and X d ,X′ d ,X″ d are d-axis synchronous, transient and subtransient reactances, respectively, and X q ,X′ q ,X″ q They are q-axis synchronous, transient and sub-transient reactances respectively; in addition, for variable speed units, since the reactance is proportional to the speed, it is necessary to calculate X x =(ω s +Δω)L x Reactance, where ω s The synchronous speed.
[0240] In addition, the armature voltage equation is:
[0241]
[0242] Where V d ,V q are the direct-axis and quadrature-axis voltage components at the generator end, R s is the resistance of the generator armature winding;
[0243] The electromagnetic active power and reactive power expressions of the generator are:
[0244]
[0245] In the electrical subsystem, the excitation system, transformer and its power grid model are modeled using modules in Simscape Electrical, which will not be described in detail in this description.
[0246] Step (5): Connect the above model modules to build the overall model and set the model parameters
[0247] Step 1: The input of the speed regulator module is P m and P ref , output y is connected to the turbine module;
[0248] Step 2: The governor output y is used as the input of the turbine module, and the turbine outputs mechanical power P m Connected to synchronous generator module;
[0249] Step 3: The flow rate of the turbine is used as the input of the flow channel system, and the output head of the flow channel system is used as the input of the turbine module;
[0250] Step 4: The input of the excitation system module is V ref 、V d and V q , the output is V f connected to a synchronous motor generator module;
[0251] Step 5: Excitation system output V f , turbine output mechanical power P m As the input of the synchronous generator, the synchronous generator outputs ω as the input of the turbine, and the synchronous generator outputs the stator voltage V d and V q As the input of the excitation system, the synchronous generator outputs electromagnetic active power as the input of the speed regulator. In addition, the three-phase circuit of the synchronous generator is connected to the transformer, and multiple units are connected through the transformer before being connected to the grid model;
[0252] Step 6: Set all step model parameters, including hydraulic machinery side system parameters and electrical side system parameters. Note that except for the time constant which is a nominal value (s), all other parameters are taken as per unit values.
[0253] Step (6): Change the module connection logic according to the control model and set the control command for simulation Step 1: When the unit is in the grid-connected mode, set the load parameter of the grid model to 1e12;
[0254] Step 2: When the unit is in isolated grid mode, set the load parameters of the grid model to be consistent with the unit power;
[0255] Step 3: When the unit is in load shedding state, disconnect the circuit breaker of the circuit model;
[0256] Step 4: Check the time domain simulation results of each physical quantity;
[0257] The method of the present invention is described in detail below using a conventional hydropower station as an implementation object of the present invention:
[0258] Step (1): Establish a flow channel system model based on the improved transfer function method
[0259] Firstly, the accuracy of the transfer function method is theoretically analyzed and compared with the accuracy of the theoretical analytical solution and the equivalent circuit method. Figure 1a -d, Figure 1a is the first component of the water hammer transfer function matrix, Figure 1b is the second component of the water hammer transfer function matrix, Figure 1c is the third component of the water hammer transfer function matrix, Figure 1d is the fourth component of the water hammer transfer function matrix. At the same time, the simulation results of a single pipe are also compared. Figure 2a For the comparison of time domain simulation results, Figure 2b is the corresponding frequency result. It can be seen from the results that the simulation results are basically consistent with the theoretical analysis.
[0260] Assuming that the flow channel system of the hydropower station is designed as a one-hole two-machine water diversion pipeline with an upstream surge chamber, the change in water head of the pump turbine caused by the flow is the sum of the transfer functions of the water diversion pipeline and the surge chamber. Figure 3 As shown in Figure 2, the improved transfer function model of the flow channel system is as follows: Figure 4 shown.
[0261] Step (2): Establish a turbine model and a governor model based on the turbine full characteristic curve
[0262] The pump turbine model is obtained by using the full characteristic curve of the hydropower unit of the hydropower station. When the unit is in the grid-connected mode, the speed governor adopts the power control mode; when the unit is in the isolated grid mode, the speed governor adopts the frequency control mode; when the unit is in the accident load shedding state, the speed governor directly gives the guide vane closing instruction. The corresponding relationship between the speed governor's relay formation and the guide vane opening is shown in the attached figure. Figure 5a The turbine torque characteristic curve is shown in the attached Figure 5b The turbine flow characteristic curve is shown in the attached Figure 5c shown.
[0263] Step (3): Establish synchronous generator, excitation system, transformer and power grid model
[0264] The simulation models of synchronous generator, excitation system and transformer are established according to the corresponding mathematical models.
[0265] Step 1: Synchronous generator model: First, determine the type of synchronous generator model to be used (for example, composite generator model or refined generator model). Add a new subsystem to the simulation model to represent the synchronous generator model. In this subsystem, use electrical components such as inductors, capacitors, and resistors to build an equivalent circuit model of the synchronous generator. Use a suitable controller to simulate the speed and voltage regulation of the generator.
[0266] Step 2: Excitation system model: In the subsystem of the synchronous generator model, add a subsystem to represent the excitation system model. Use electrical components such as inductors, capacitors, and resistors to build an equivalent circuit model of the excitation system, such as the exciter, excitation transformer, and voltage regulator. Add appropriate controller modules according to the required control strategy.
[0267] Step 3: Transformer model: Add a subsystem to represent the transformer model in the subsystem of the synchronous generator model. Use electrical components such as inductors, capacitors, and resistors to build an equivalent circuit model of the transformer, such as the transformer's inductance, resistance, and mutual inductance. Consider the transformer's short-circuit and open-circuit characteristics as needed.
[0268] Step 4: Grid model: In the subsystem of the synchronous generator model, add a subsystem to represent the grid model. Use electrical components such as inductors, capacitors, and resistors to build an equivalent circuit model of the grid, such as inductors, capacitors, and resistors. Consider the characteristics of the grid such as frequency and voltage regulation.
[0269] Step 5: Connect the models: Use transfer signal lines to connect the synchronous generator, excitation system, transformer, and grid models together to simulate the hydroelectric system's water-machine-electric coupling transition process. Make sure to set the signal connections and model parameters correctly.
[0270] Step 6: Simulation and verification: Configure simulation parameters, such as simulation time, step size, etc. Run the simulation to verify the correctness and accuracy of the established model. Analyze the simulation results to evaluate the performance and response of the hydropower system during the transition process.
[0271] Step (4): Connect the above model modules to build the overall model and set the model parameters
[0272] The above modules are connected according to the data flow transmission logic to form the overall model. The connection sequence between modules is as follows: Figure 6 shown.
[0273] Step (5): Change the module connection logic according to the control mode and set the control command for simulation
[0274] Since this embodiment adopts the method of load shedding in case of an accident of one unit and the normal operation of the other unit in parallel with the main grid, the module connection logic has been described in the above steps. The specific module connection sequence is as follows: Figure 6 shown.
[0275] Finally, a hydraulic interference condition is set for simulation. The specific simulation results are shown in Figure 7. Figure 7a are the simulation results of the stator currents of the two units. Figure 7b are the simulation results of the stator voltage of the two units. Figure 7c is the simulation result of the power of the two units, Figure 7d The simulation results of the flow of the two units are as follows: Figure 7e The simulation results of the volute pressure of the two units are shown in Figure 2. Figure 7f The simulation results of the tailwater pipe pressure of the two units. The accident load shedding of Unit 1 will cause the head and flow in the flow system shared by the two units to change, thereby affecting the working state of Unit 2 and causing overload of Unit 2.
[0276] It is easy for those skilled in the art to understand that the above-described working conditions are only preferred embodiments of the present invention and are not intended to limit the present invention. Any example analysis, equivalent substitution, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A refined modeling and simulation method for the water-machine-electricity coupling transition process of a hydropower generation system, characterized in that: The method comprises the following steps: Establish a flow channel system model based on the improved transfer function method; Establish a turbine model and a governor model based on the turbine's full characteristic curve; Establish synchronous generator, excitation system, transformer and power grid models; Connect the above model modules to build the overall model and set the model parameters; Change the module connection logic according to the control mode and set the control command for simulation.
2. According to claim 1, a method for fine-tuning modeling and simulation of water-machine-electricity coupling transition process of a hydroelectric power generation system is characterized in that: The flow channel system model includes a pressure pipeline and a pressure regulating chamber, and adopts an improved transfer function method to perform one-dimensional numerical simulation on the transient flow in the flow channel system.
3. The method for fine-tuning modeling and simulation of water-machine-electricity coupling transition process of a hydropower generation system according to claim 1 is characterized in that: The nonlinear turbine model is interpolated based on the full characteristic curve of the turbine, and the governor model adopts PID control.
4. The method for fine-tuning modeling and simulation of water-machine-electricity coupling transition process of a hydroelectric power generation system according to claim 1 is characterized in that: The synchronous generator model adopts a 6th-order synchronous generator model based on a rotor rotating coordinate system, and the transformer, power grid and other models are all three-phase circuit models.
5. The method for fine-tuning modeling and simulation of water-machine-electricity coupling transition process of a hydroelectric power generation system according to claim 1 is characterized in that: The overall model includes seven parts: power station flow system, turbine, speed governor, excitation system, synchronous generator, transformer and power grid. The model parameters include hydraulic side parameters, mechanical side parameters and electrical side parameters.
6. The method for fine-tuning modeling and simulation of water-machine-electricity coupling transition process of a hydroelectric power generation system according to claim 1, characterized in that: The control modes include a speed control mode of the speed regulator and a power control mode of the speed regulator, and the control command will change the internal control structure of the speed regulator according to the control mode.
7. An electronic device, characterized in that: A computer-readable storage medium storing computer-executable instructions; and one or more processors, the one or more processors being coupled to the computer-readable storage medium and configured to execute the computer-executable instructions so that the device performs the method according to any one of claims 1-6.
8. A readable storage medium, characterized in that: Computer executable instructions are stored, which, when executed by a processor, configure the processor to perform the method according to any one of claims 1-6.
Citation Information
Cited By
Guide vane and power angle double-loop cooperative feedback control method of one-pipe multi-machine hydroelectric system
CN120848352A
Numerical simulation method for hydraulic transition process of pressure regulating equipment
CN122616428A
A numerical simulation method for hydraulic transient process of a pressure regulating device
CN122616428B
