Method for building simulation platform of axial-flow hydroelectric generating set with paddle signal driver

Abstract

The application belongs to the technical field of safe and stable operation of hydroelectric generating set, and particularly provides a method for building a simulation platform of an axial flow and rotating slurry type hydroelectric generating set containing a paddle signal driver, which comprises the following steps: designing a paddle signal driver; building a water turbine regulating system simulation platform containing a real governor, a hydroelectric generating set simulator and a monitoring system simulator; constructing a water turbine flow and torque characteristic neural network model based on a BP neural network; saving the weights and thresholds of the water turbine neural network model in a txt format; importing the weights and thresholds saved in the txt format of the water turbine neural network into the host computer software of the hydroelectric generating set simulator; and issuing a command by the monitoring system simulator to test the precision of the simulation platform. The method fully considers the coordination relationship between the paddle angle and the guide vane opening, and truly simulates the actual operation of the axial flow and rotating slurry type hydroelectric generating set.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of safe and stable operation of hydroelectric generating sets, in particular, relates to a method for building a simulation platform of an axial flow and rotating blade hydroelectric generating set containing a blade signal driver. BACKGROUND

[0002] As an important hydroelectric power device, the axial flow and rotating blade hydroelectric generating set has a wide application in the field of hydroelectric energy development and utilization. The device introduces water flow into the axial flow runner to generate rotary kinetic energy, and then rotates the generator to finally convert it into electric energy. However, the performance and efficiency of such a hydroelectric generating set are affected by many factors, one of which is the design and adjustment of the blades.

[0003] Traditional blade design usually relies on empirical rules and trial-and-error methods based on simplified models. This method greatly limits the performance optimization of the blades and the improvement of the overall efficiency of the unit. In addition, due to the complexity of the operating environment of the hydroelectric generating set, the design and adjustment of the blades need to consider many factors, including water flow speed, water pressure, unit speed, etc. Therefore, developing an effective blade design and adjustment method is crucial to improving the performance of the hydroelectric generating set.

[0004] In recent years, with the development of computer simulation technology, numerical simulation-based performance evaluation and optimization methods of hydroelectric generating sets have received widespread attention. By establishing an accurate numerical model, the flow of water in the axial flow turbine and rotating blade machine can be simulated to evaluate the performance of different blade designs and optimize the geometry and parameters of the blades. However, existing simulation platforms are mostly limited to specific software or models, lacking universality and flexibility.

[0005] Therefore, a new method for building a simulation platform of an axial flow and rotating blade hydroelectric generating set is needed, which can fully utilize advanced numerical simulation technology to achieve rapid, accurate and flexible blade design and adjustment. This patent aims to provide such a method to solve the problems existing in the prior art and provide reliable technical support for the performance optimization and engineering application of the axial flow and rotating blade hydroelectric generating set. SUMMARY

[0006] The technical problem to be solved by the present application is to provide a method for building a simulation platform of an axial flow and rotating blade hydroelectric generating set containing a blade signal driver, which fully considers the coordination relationship between the blade angle and the guide vane opening and truly simulates the actual operation of the axial flow and rotating blade hydroelectric generating set.

[0007] To solve the above technical problems, the technical solution adopted by the present application is: a method for building a simulation platform of an axial flow and rotating blade hydroelectric generating set containing a blade signal driver, comprising the following steps:

[0008] S1, design a blade signal driver;

[0009] S2, build a water turbine regulating system simulation platform including a real governor, a hydroelectric generating set simulator and a monitoring system simulator;

[0010] S3, constructing a water turbine flow and torque characteristic neural network model based on a BP neural network;

[0011] S4, saving the weights and thresholds of the water turbine neural network model in txt format;

[0012] S5, importing the weights and thresholds of the water turbine neural network saved in txt format into the host computer software of the hydroelectric generating set simulator;

[0013] S6, the monitoring system simulator sends a command to test the precision of the simulation platform.

[0014] In the preferred scheme, in step S1, the paddle signal driver includes an electrical part and a mechanical part, the electrical part includes a brushless motor driver and a brushless DC speed reducer, the brushless motor driver is used to drive the brushless DC speed reducer, the mechanical part includes a paddle displacement signal transmission wheel arranged on the output shaft of the brushless DC speed reducer, the paddle displacement signal transmission wheel is connected with a paddle displacement sensor outer wheel through a steel wire rope, and the paddle displacement sensor outer wheel is arranged on the paddle displacement sensor.

[0015] In the preferred scheme, the operation process of step S2 is as follows:

[0016] The real governor, the hydroelectric generating set simulator and the monitoring system simulator are combined to build a simulation platform of the axial flow and rotating paddle type hydroelectric generating set regulating system according to the matching relationship between the built-in paddle angle and the guide vane opening, the monitoring system simulator outputs a switch pulse to the real governor, the hydroelectric generating set simulator is used to simulate the hydroelectric generating set, realize paddle control, the paddle signal driver is used to adjust the paddle angle, the hydroelectric generating set simulator is used to transmit the simulation parameters of the hydroelectric generating set to the real governor, and the real governor is used to control the guide vane and the paddle of the hydroelectric generating set simulator.

[0017] In the preferred scheme, in step S3, according to the comprehensive characteristic data of the water turbine model, a water turbine neural network model is constructed based on the BP neural network and the gravitational search algorithm by using the Matlab software, the water turbine neural network model takes the unit speed and the guide vane opening as the input, and the water turbine torque and the flow as the output.

[0018] In the preferred scheme, the gravitational search algorithm is used to optimize the weights and thresholds of the BP neural network; for the adopted gravitational search algorithm, it is assumed that there are particles, and the = represents the dimensionali position of a particle, and = denotes the corresponding velocity parameter, where ; the fitness function value and inertia of the t th iteration are denoted as and , respectively, where The root mean square error RMSE is expressed as:

[0019] ;

[0020] In the formula, n is the number of sample data, d i is the predicted data of the i th hydraulic turbine torque or flow rate, y i is the sample data of the i th hydraulic turbine torque or flow rate;

[0021] The optimal and worst fitness values of the entire particle swarm are denoted as best(t) and worst(t) , The expressions are:

[0022] (1) ;

[0023] (2) ;

[0024] According to the Newton's law of universal gravitation, d the particle i in the j dimension acts on the particle t at time

[0025] (3) ;

[0026] where denotes a small constant, denotes the Euclidean distance between the particle t and i at the j th iteration, i.e.,

[0027] (4) ;

[0028] The expression of the gravitational coefficient is:

[0029] (5) ;

[0030] In the formula, is the initial gravitational constant,α is the attenuation exponent, i t and N Gen are the current iteration number and the maximum iteration number respectively; during the calculation of the gravitational coefficient, a random number between 0 and 1 will be generated rand assigned to i the gravitational resultant force of the particle:

[0031] (6);

[0032] At this time, in d the dimensional space, the i acceleration of the particle is:

[0033] / (7);

[0034] In d the dimension, the i velocity and position of the particle are respectively:

[0035] (8);

[0036] (9);

[0037] Among them: rand is a random number of 0 or 1.

[0038] In a preferred solution, in step S4, the weight and threshold information of the trained torque characteristic neural network and flow characteristic neural network models are saved in txt format in Matlab software.

[0039] In a preferred solution, step S5 is specifically: importing the weights and thresholds saved in txt format of the turbine neural network model into the upper computer supporting the water turbine generator set simulator.

[0040] In a preferred solution, step S6 is specifically: sending commands by a monitoring system simulator that can issue start-up, shutdown, load increase / decrease commands, and comparing the simulation values with the reference values to test the accuracy of the simulation platform.

[0041] The application provides a method for building a simulation platform of an axial flow water turbine generator set containing a paddle signal driver. BRIEF DESCRIPTION OF DRAWINGS

[0042] The application will be further described below in combination with the drawings and examples:

[0043] Figure 1 The method design steps of the application are as follows:

[0044] Figure 2 The paddle signal driver design principle of the application is as follows:

[0045] Figure 3 The internal structure diagram of the paddle signal driver of the application is as follows:

[0046] Figure 4 The simulation platform schematic diagram of the axial flow water turbine generator set containing the paddle signal driver of the application is as follows:

[0047] Figure 5 The water turbine model construction flowchart based on the BP neural network and the gravitational search algorithm of the application is as follows:

[0048] Figure 6 The start-up process waveform diagram in the example of the application is as follows:

[0049] In the figure: brushless motor driver 1, brushless DC speed reduction motor 2, paddle displacement signal transmission wheel 3, paddle displacement sensor outer wheel 4, steel wire rope 5. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical scheme and advantages of the application clearer and more understandable, the application will be further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the application and do not limit the application.

[0051] A method for building a simulation platform of an axial flow water turbine generator set containing a paddle signal driver, as shown in Figure 1 The method comprises the following steps:

[0052] S1, design a paddle signal driver.

[0053] As shown in Figure 2As shown, the paddle signal driver includes an electrical part and a mechanical part, the electrical part includes a brushless motor driver and a brushless DC speed reducer motor, the brushless motor driver is used to drive the brushless DC speed reducer motor, the mechanical part includes a paddle displacement signal transmission wheel arranged on the output shaft of the brushless DC speed reducer motor, the paddle displacement signal transmission wheel is connected with a paddle displacement sensor outer wheel through a steel wire rope, and the paddle displacement sensor outer wheel is arranged on the paddle displacement sensor. Figure 3 As shown.

[0054] The brushless motor driver is selected as model: AISIQMD6010BLS. The internal specific wiring of the paddle signal driver is as follows.

[0055] 1) Incoming line: the positive electrode of the unit simulator output voltage signal is connected to the input signal In1 of the driver, and the negative electrode of the unit simulator output voltage signal is connected to the input signal com of the driver.

[0056] 2) Outgoing line: the 485 communication is connected to the driver host computer, and the paddle displacement signal is connected to the governor.

[0057] 3) Internal: input 220V alternating current to connect the special power supply 1-2 terminals of the driver, and select one of 4-6 as 24V negative electrode output and 7-9 as 24V negative electrode output.

[0058] The internal motor driver initialization setting method of the paddle signal driver is as follows:

[0059] a. Install the driver program and open the motor learning program.

[0060] b. Connect the driver, motor, power supply and host computer.

[0061] c. Turn the code switch 8 to ON.

[0062] d. Set the learning parameters on the host computer: change the serial number to COM4, click "open serial port", click "read", the rated current of the motor is 8.4A, the steady speed control algorithm is selected as "time-position closed loop control", the position control algorithm is selected as "horizontal positioning control", the position control allowed error is 2 pulses, the motor pole pair number is 4, the reducer reduction ratio is 216:1, click "start learning", the driver will "ding" a short sound for each learning phase, and if "ding" a long sound is heard after learning, it means that the learning is successful, if three "ding-ding-ding" sounds are heard in succession, it means that the learning fails, then click "configuration" (if you want to check the configuration effect, click "motor control").

[0063] After setting the parameters, click "reciprocating motion", set the reset mode as "no reset", and calculate the total stroke pulse number. Assuming that the effective stroke corresponds to α ° range, the actual stroke is mm.

[0064] If the diameter of the paddle displacement signal transmission wheel is 15.5 mm, the stroke is in pulse units, the speed of the brushless DC motor is 3000 r / min, and the selected reduction ratio is 216, then 1 pulse corresponds to , that is, .

[0065] The host computer on the driver is set: the required number of pulses is . If = 270°, then the required number of pulses is .

[0066] e. Turn the code switch 8 to OFF to turn off the motor power supply, and the learning is complete.

[0067] f. Turn the code switches 4, 6, and 7 to ON, and the parameter initialization is complete.

[0068] g. Turn on the power supply, and the driver control mode is changed from 485 communication control to external analog signal control.

[0069] S2, build a water turbine regulating system simulation platform including a real governor, a hydroelectric generator set simulator, and a monitoring system simulator, as shown in Figure 4 .

[0070] A simulation platform for regulating the axial flow and rotating water turbine generator set is built by combining the real governor, the virtual hydroelectric generator set simulator, and the virtual monitoring system simulator based on the blade angle and the guide vane opening relationship.

[0071] The monitoring system simulator outputs the switch pulse quantity to the real governor, the hydroelectric generator set simulator is used to simulate the hydroelectric generator set, the paddle signal driver is used to adjust the paddle angle, and the hydroelectric generator set simulator is used to transmit the simulation parameters of the hydroelectric generator set to the real governor, which is used for guide vane control and paddle control.

[0072] S3, based on BP neural network and gravitational search algorithm, a neural network model of water turbine flow and torque characteristics is constructed.

[0073] According to the comprehensive characteristic data of the water turbine model, using Matlab software, based on BP neural network and gravitational search algorithm, a torque characteristic neural network and a flow characteristic neural network model are constructed with unit speed and guide vane opening as input and water turbine torque and flow as output, wherein the gravitational search algorithm is used to optimize the weight and threshold of the BP neural network, and the specific process is as shown in Figure 5The adopted BP neural network structure includes input layer, hidden layer and output layer, the number of hidden layer neurons can be adjusted, the output layer is linear, and the hidden layer neuron transfer function is log-signoid function. The corresponding weight and threshold can be obtained through neural network learning, so as to establish the neural network-based water turbine model. In practical application, only the unit speed and guide vane opening need to be input, and the established water turbine neural network model can be used to calculate the water turbine output flow and torque.

[0074] To improve the model accuracy, the weight and threshold of the BP neural network are optimized based on the gravitational search algorithm (GSA). The GSA is a heuristic optimization algorithm that simulates the law of universal gravitation in astrophysics and is used to solve optimization problems in continuous space. The basic idea of GSA is to regard each solution in the search space as a celestial body with mass, and the interaction force (i.e. gravity) between these celestial bodies determines their movement direction and speed, thereby guiding the solution to move to a better position. In GSA, each candidate solution (or "agent") has a fitness value associated with it, which can be regarded as the mass of the celestial body. The higher the fitness value, the greater the mass of the corresponding "celestial body". The force of action (gravity) between celestial bodies is proportional to their mass and inversely proportional to the distance between them. In each iteration, each celestial body adjusts its position according to the gravity of all other celestial bodies on it, simulating the process of gravitational interaction between celestial bodies.

[0075] For the adopted GSA algorithm, it is assumed that there are particles, represented by = , where is the position of the i th particle in -dimensional space, and is its velocity parameter, where ; the fitness function value and inertia obtained in the t th iteration are represented by and , respectively, where The root mean square error RMSE is used to represent, and the expression is:

[0076] ;

[0077] In the formula, n is the number of sample data, d i is the predicted data of the i th water turbine torque or flow, y i is thei sample data of the water turbine torque or flow;

[0078] The optimal and worst fitness values of the entire particle swarm are respectively represented as best(t) and worst(t) , The expression is:

[0079] (1);

[0080] (2);

[0081] According to the Newton's law of universal gravitation, d particles in the dimension i and the particle j at time t The force of action is:

[0082] (3);

[0083] wherein, represents a minimum constant, represents the iteration t times of the particle i and j Euclidean distance, that is:

[0084] (4);

[0085] The expression of the gravitational coefficient is:

[0086] (5);

[0087] In the formula: is the initial gravitational constant, α is the decay index, i t and N Gen are the current iteration number and the maximum iteration number respectively; in the process of calculating the gravitational coefficient, a random number of 0-1 rand is given to the gravitational force of the particle i :

[0088] (6);

[0089] At this time, the acceleration of the particle d in the i dimensional space is:

[0090] / (7);

[0091] In the process of calculating the gravitational coefficient, a random number of 0-1 dDimensional particle i The speed and position of the particle in the dimension are respectively:

[0092] (8);

[0093] (9);

[0094] wherein: rand is a random number of 0 or 1.

[0095] S4, save the weights and thresholds of the hydraulic turbine neural network model as txt format.

[0096] In the Matlab software, save the weight and threshold information of the trained moment characteristic neural network and flow characteristic neural network model as txt format.

[0097] S5, import the weights and thresholds of the water turbine neural network saved in txt format into the host computer software of the hydroelectric generator set simulator.

[0098] Import the weights and thresholds of the water turbine neural network model saved in txt format into the host computer software GTS4_TG of the water turbine simulator of the GTS4 type.

[0099] S6, the simulation instrument of the monitoring system sends a command to test the precision of the simulation platform.

[0100] The monitoring system simulation instrument can send commands to start, stop, increase / decrease load, compare the simulation value with the reference value to test the simulation platform precision, and ensure the reliability of the simulation platform.

[0101] Simulation example:

[0102] In order to verify the reliability of the present application, a certain axial flow propeller unit is taken as an example, an axial flow propeller water turbine simulation platform containing a blade signal driver is built based on real data, and an automatic start test is carried out, and the test steps are:

[0103] 1) First of all, it should be ensured that the governor can automatically run, the servomotor is freely operated, and the governor is in shutdown state;

[0104] 2) According to the set simulation unit parameters, select appropriate governor adjustment parameters;

[0105] 3) Set the frequency disturbance output to be the rated frequency (if this signal is used as the frequency given input of the governor);

[0106] 4) The unit operation mode is grid-connected mode;

[0107] 5) Press the "start / stop simulation" button, and confirm that the simulation unit frequency is below 10%;

[0108] 6) Click the "start sampling" button to start the waveform recording;

[0109] 7) The governor executes the automatic start command issued by the simulation instrument;

[0110] 8) If the governor adjustment parameters are reasonable, the output of the speed regulation system (simulated frequency output) will eventually stabilize at around 50 Hz. The recorded waveform is the start-up process, as shown in Figure 6 .

[0111] Note:

[0112] Once the "start simulation" button is pressed, the speed regulation system is in a closed-loop state, and at this time the system can be judged to be working normally through the measured value data box of each waveform; due to the convergence of the simulation algorithm, the lower limit of the simulated unit frequency output is 2 Hz in all cases, so the simulation frequency during the start-up process only contains the part above 2 Hz; all operations in the data acquisition and analysis control area, including routine index analysis, are the same as dynamic characteristic testing and will not be repeated here.

[0113] As can be seen from Figure 6 , the frequency change curve is a monotonous rising process. The start-up time is about 28s, and after the frequency reaches about 45Hz, it enters the no-load opening degree and stabilizes at the reference value of 50Hz, which is consistent with the actual operation, that is, the simulation platform of the axial-flow and mixed-flow hydroelectric generating set containing the blade signal driver is reliable.

[0114] In view of the difficulty in establishing a simulation platform of the axial-flow and mixed-flow hydroelectric generating set considering the blades, a method for building a simulation platform of the axial-flow and mixed-flow hydroelectric generating set containing the blade signal driver is proposed. This method will allow a variety of high-precision hydroelectric generating set simulation tests to be carried out, thereby facilitating the improvement and upgrading of hydroelectric equipment and improving the power generation efficiency.

[0115] Those skilled in the art will readily understand that the above description is only a preferred embodiment of the present application and is not intended to limit the present application, and any modifications, equivalent replacements and improvements made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for constructing a simulation platform for an axial-flow propeller-type hydroelectric generator including a blade signal driver, characterized in that, The steps include the following: S1. Design a blade signal driver: The blade signal driver includes an electrical part and a mechanical part. The electrical part includes a brushless motor driver and a brushless DC reduction motor. The brushless motor driver is used to drive the brushless DC reduction motor. The mechanical part includes a blade displacement signal transmission wheel arranged on the output shaft of the brushless DC reduction motor. The blade displacement signal transmission wheel is connected to the outer wheel of the blade displacement sensor through a steel wire rope. The outer wheel of the blade displacement sensor is arranged on the blade displacement sensor. S2. Build a simulation platform for a water turbine regulating system including a real governor, a water turbine generator set simulator, and a monitoring system simulator. The operation process is as follows: Build a simulation platform for an axial flow Kaplan turbine regulating system by combining a real governor with built-in blade angle and wicket gate opening association relationship, a water turbine generator set simulator, and a monitoring system simulator. The monitoring system simulator outputs a switch pulse quantity to the real governor. The water turbine generator set simulator is used to simulate a water turbine generator set to achieve blade control. The blade signal driver is used to adjust the blade angle. The water turbine generator set simulator is used to transmit water turbine generator set simulation parameters to the real governor. The real governor is used to conduct wicket gate control and blade control on the water turbine generator set simulator. S4. Build a neural network model for water turbine flow and torque characteristics based on a BP neural network. S5. Save the weights and thresholds of the water turbine neural network model in txt format. S6. Import the weights and thresholds of the water turbine neural network saved in txt format into the upper computer software of the water turbine generator set simulator. S6. The monitoring system simulator issues a command to test the accuracy of the simulation platform.

2. The method for constructing a simulation platform for an axial-flow propeller-type hydroelectric generator including a blade signal driver according to claim 1, characterized in that, In step S3, specifically: According to the comprehensive characteristic data of the water turbine model, use Matlab software to build a water turbine neural network model based on a BP neural network and a gravitational search algorithm. The water turbine neural network model takes unit speed and wicket gate opening as inputs and water turbine torque and flow as outputs.

3. The method for constructing a simulation platform for an axial-flow propeller-type hydroelectric generator including a blade signal driver according to claim 2, characterized in that, The gravity search algorithm is used to optimize the weights and thresholds of a BP neural network; for the gravity search algorithm used, it is assumed that... A particle, using = express Dimension i The position of each particle, and using = This represents the corresponding speed parameter, where, ;No. t The fitness function value and inertia obtained from the second iteration are expressed as follows: and ,in Using root mean square error RMSE The expression is: ； In the formula, n For the number of sample data, d i For the first i Predicted data for turbine torque or flow rate, y i For the first i Sample data of turbine torque or flow rate; The optimal and worst fitness values ​​of the entire particle swarm are expressed as follows: best(t) and worst(t) , The expression is: (1)； (2)； According to Newton's law of universal gravitation, d Particles in Dimensions i With particles j exist t The force at any given time is: (3)； in, Represents a minimal constant. Iteration t Next-time particles i and j The Euclidean distance, i.e.: (4 )； Gravitational coefficient The expression is: (5)； In the formula: Let be the initial gravitational constant. α The decay exponent, i t and N Gen These represent the current iteration number and the maximum iteration number, respectively; a random number between 0 and 1 will be generated during the calculation of the gravity coefficient. rand Assigned i The resultant gravitational force of particles: (6)； At this time, d Particles in 1-dimensional space i The acceleration is: / (7)； exist d Particles in Dimension i The speed and position are as follows: (8)； (9)； in: rand A random number that is either 0 or 1.

4. The method for constructing a simulation platform for an axial-flow propeller-type hydroelectric generator including a blade signal driver according to claim 1, characterized in that, In step S4, in Matlab software, save the weight and threshold information of the trained torque characteristic neural network and flow characteristic neural network models in txt format.

5. The method for constructing a simulation platform for an axial-flow propeller-type hydroelectric generator including a blade signal driver according to claim 1, characterized in that, Step S5 specifically: Import the weights and thresholds of the water turbine neural network model saved in txt format into the upper computer supporting the water turbine generator set simulator.

6. The method for constructing a simulation platform for an axial-flow propeller-type hydroelectric generator including a blade signal driver according to claim 1, characterized in that, Step S6 specifically: The monitoring system simulator that can issue commands for starting, stopping, increasing / decreasing load issues a command to compare the simulation value with the reference value to test the accuracy of the simulation platform.

Images