CN115356096B - System and method for researching surge characteristics of compressor pipe network system - Google Patents
System and method for researching surge characteristics of compressor pipe network system Download PDFInfo
- Publication number
- CN115356096B CN115356096B CN202210994751.4A CN202210994751A CN115356096B CN 115356096 B CN115356096 B CN 115356096B CN 202210994751 A CN202210994751 A CN 202210994751A CN 115356096 B CN115356096 B CN 115356096B
- Authority
- CN
- China
- Prior art keywords
- compressor
- valve
- pipe network
- simulation
- network system
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000004088 simulation Methods 0.000 claims abstract description 92
- 238000012360 testing method Methods 0.000 claims abstract description 74
- 238000012549 training Methods 0.000 claims abstract description 25
- 230000006870 function Effects 0.000 claims description 50
- 238000005094 computer simulation Methods 0.000 claims description 45
- 230000001105 regulatory effect Effects 0.000 claims description 41
- 238000013528 artificial neural network Methods 0.000 claims description 34
- 238000011160 research Methods 0.000 claims description 27
- 238000013135 deep learning Methods 0.000 claims description 24
- 230000007246 mechanism Effects 0.000 claims description 22
- 238000011144 upstream manufacturing Methods 0.000 claims description 8
- 238000004134 energy conservation Methods 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 3
- 238000012546 transfer Methods 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 2
- 230000008859 change Effects 0.000 description 7
- 210000002569 neuron Anatomy 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 4
- 238000004891 communication Methods 0.000 description 4
- 230000001066 destructive effect Effects 0.000 description 3
- 239000012530 fluid Substances 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000011056 performance test Methods 0.000 description 2
- 230000000737 periodic effect Effects 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000005272 metallurgy Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 238000011946 reduction process Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 230000008685 targeting Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B17/00—Systems involving the use of models or simulators of said systems
- G05B17/02—Systems involving the use of models or simulators of said systems electric
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Control Of Positive-Displacement Air Blowers (AREA)
Abstract
The invention discloses a system and a method for researching surge characteristics of a compressor pipe network system. The method has high simulation precision after training the test data, can accurately simulate the dynamic characteristics of the compressor pipe network system when switching between working conditions, and has reliable simulation results.
Description
Technical Field
The invention belongs to the field of surge test and simulation research of a compressor pipe network system, and relates to a system and a method for researching surge characteristics of the compressor pipe network system.
Background
Centrifugal compressors are important equipment in the industrial field and have wide application in the fields of chemical industry, metallurgy, aviation and the like. The centrifugal compressor sucks working medium gas from the air inlet through the high-speed rotating impeller, and the working medium gas is centrifugally thrown out through the acceleration of the impeller, pressurized in the diffuser and then input into downstream equipment. The centrifugal compressor is usually matched with devices of pipelines and valves to realize pressurization and transportation of gas working media. When the compressor pipe network system works, the surge working condition of repeated fluctuation of the pressure of the compressor is easy to occur at low flow. The periodic intense vibration typically associated with a surge condition can cause devastating damage to the equipment and need to be avoided during operation.
The research on the surge problem is always a hot spot subject of attention of students at home and abroad, and a plurality of models are provided for the surge problem of the compressor. The most classical of these is Greitzer compressor model, which combines the performance curves to model the mechanism of the surge problem. In the subsequent study, scholars perform optimization on the basis of Greitzer models to perform simulation modeling study. With the development of testing technology, experimental research on a compressor pipe network system has also been developed. The test research of surge is usually destructive, the cost is high, the risk is high, and especially the destructive power is huge when a large-sized compressor enters surge, so that the surge test research is hindered.
Disclosure of Invention
The invention aims to overcome the defects of the prior art, and provides a system and a method for researching the surge characteristics of a compressor pipe network system, so as to solve the defect that the problem of surge of a centrifugal compressor is difficult to research in the prior art.
In order to achieve the purpose, the invention is realized by adopting the following technical scheme:
A method for researching surge characteristics of a compressor pipe network system comprises the following steps:
Step 1, a mechanism model of a compressor pipe network system is established, wherein the mechanism model comprises a mass conservation equation, a momentum conservation equation and an energy conservation equation;
Step 2, establishing a compressor zero-dimensional control equation, a pipe network zero-dimensional control equation and a valve zero-dimensional control equation through a mechanism model; the valve comprises a compressor inlet regulating valve and a compressor outlet regulating valve; respectively establishing simulation modules of a compressor, a pipe network and a valve through a zero-dimensional control equation;
Step 3, connecting three simulation modules to form a simulation system of the compressor pipe network system; the input of the compressor simulation module and the valve simulation module is the output of the pipe network simulation module, and the input of the pipe network simulation module is the output of the compressor simulation module and the valve simulation module;
Step 4, taking the opening of the inlet regulating valve of the compressor and the opening of the outlet regulating valve of the compressor as variables, and performing multi-station test on a pipe network system of the compressor to obtain a test number set; the test data set comprises a valve test data set and a compressor test data set, and the neural network is trained through the valve test data set to obtain a valve flow resistance characteristic function; training the neural network through a compressor test number set to obtain a compressor characteristic function;
Step 5, the valve flow resistance characteristic function is brought into a valve simulation module, the compressor characteristic function is brought into a compressor simulation module, the valve simulation module and the compressor simulation module are simulated, if the simulation result precision of the two simulation modules meets the standard, the step 6 is executed, otherwise, the step 4 is repeated, and the valve flow resistance characteristic function and the compressor characteristic function are obtained again through neural network learning;
step 6, bringing a valve flow resistance characteristic function and a compressor characteristic function meeting the precision requirement into a simulation system of a compressor pipe network system, and completing parameter transfer and control equation decoupling among three simulation modules to obtain a dynamic simulation system of the compressor pipe network system;
And 7, simulating a compressor surge working condition through a dynamic simulation system of the compressor pipe network system to obtain the compressor surge characteristic, wherein the compressor surge characteristic comprises compressor temperature, pressure and flow.
The invention further improves that:
Preferably, in step 1, the mass conservation equation is:
the momentum conservation equation is:
The energy conservation equation is:
Preferably, in step2, the zero-dimensional control equation of the compressor is:
the zero-dimensional control equation of the pipe network is as follows:
the zero-dimensional control equation of the valve is as follows:
wherein p, T, w, v are pressure, temperature and mass flow and flow rate, respectively; h is specific enthalpy; cp is the specific heat of the working medium; v is the volume of the pipeline; r is a gas constant; psi is the performance curve of the compressor; n is the rotation speed; l c is the equivalent through-flow length; a 1 is the flow area; d is the diameter of the pipeline; c f is a valve flow resistance characteristic function; k is the valve flow area coefficient.
Preferably, in step 2, in the compressor simulation module, the input parameter is an upstream pressure of the compressor and a downstream pressure of the compressor, and the output parameter is a mass flow of the compressor; in the pipe network simulation module, the input parameters are the mass flow entering the pipe network and the mass flow output by the pipe network, and the output parameters are the pressure of the pipe network; in the valve simulation module, input parameters are valve opening, valve front pressure and valve back pressure, and output parameters are valve mass flow.
Preferably, in step 4, in the process of training the neural network through the valve test data set and training the neural network through the compressor test data set, each neural network performs deep learning through a generalized least square method by using the BP neural network and the NARX neural network at the same time, and compared with simulation precision of the deep learning of the two neural networks, the neural network with small precision is selected as the final neural network.
Preferably, in step 4, the test data set is used after the filtering noise reduction process.
In step 6, in the simulation system of the compressor pipe network, the input parameters are functions of the opening of the inlet regulating valve of the compressor and the opening of the outlet regulating valve of the compressor; the output parameters are the temperature, pressure and flow of the compressor, the temperature, pressure and flow of the pipe network, and the temperature, pressure and flow of the inlet and outlet sections of the valve.
Preferably, in step 6, the simulation result of the dynamic simulation system of the compressor pipe network system is tested and corrected until the precision requirement of the dynamic simulation is met, and step 7 is executed.
Preferably, in step 7, when the surge condition of the compressor is simulated, the compressor is caused to enter the surge condition by adjusting the opening of the inlet regulating valve of the compressor and the opening of the outlet regulating valve of the compressor; when the compressor surge characteristic is obtained, the compressor surge characteristic in the signal fluctuation is extracted.
The system for researching the surge characteristics of the compressor pipe network system for the research method according to any one of the above, comprises a centrifugal compressor, wherein a compressor inlet regulating valve is arranged in front of the centrifugal compressor, and a compressor outlet regulating valve is arranged at the outlet of the centrifugal compressor;
A test point is arranged between the inlet regulating valve of the compressor and the centrifugal compressor, a test point is arranged between the centrifugal compressor and the outlet regulating valve of the compressor, the test point is connected with an A/D converter, and the A/D converter is connected with a computer and a software part;
And the test point is provided with a pressure sensor, a temperature sensor and a flow sensor.
Compared with the prior art, the invention has the following beneficial effects:
The invention discloses a method for researching surge characteristics of a compressor pipe network system, which is characterized in that key parameters of all parts are deeply learned on the basis of a mechanism model, and identification of the key parameters is realized by training test data of a plurality of working conditions, so that a semi-mechanism model with high simulation precision is obtained, and the research method is advanced and has innovation. The method has high simulation precision after training the test data, can accurately simulate the dynamic characteristics of the compressor pipe network system when switching between working conditions, and has reliable simulation results. The method is based on experimental research equipment, a semi-mechanism model of a compressor pipe network system is obtained by utilizing mechanism modeling and deep learning, and the surge characteristic of the compressor is obtained by dynamic simulation. On the premise of meeting the dynamic simulation precision, the high cost of destructive test research is avoided, and the method has advancement and practicability.
The invention also discloses a system for researching the surging characteristics of the compressor pipe network system, which is a device for researching the surging characteristics of the compressor pipe network system based on mechanism modeling and deep learning, and comprises 1) a test research device which is divided into a hardware part and a software part; 2) The hardware part comprises a compressor pipe network system consisting of a valve, a pipeline and a compressor, a pressure sensor, a temperature sensor, a flow sensor, signal collection and processing equipment and an acquisition and control system consisting of a computer; 3) The software part is a dynamic simulation model of the target compressor pipe network system; 4) The dynamic simulation model of the compressor pipe network system is a half-mechanism model, and the half-mechanism model with higher precision is obtained by deep learning of test data on the basis of the compressor pipe mechanism model; 5) And dynamically simulating the surge working condition of the compressor pipe network system by using the semi-mechanism model, and outputting a simulation result. Compared with the prior art, the invention is oriented to the field of research on surge characteristics of the compressor pipe network system, avoids the risk of surge test while guaranteeing the reliability of simulation results, has low cost and high safety, and has practicability and universality. Compared with the traditional surge research equipment, the surge research device of the compressor can greatly reduce the research cost on the premise of ensuring the simulation precision, and avoid the high cost and high risk of surge test research.
Drawings
FIG. 1 is a technical block diagram of the present invention;
FIG. 2 is a block diagram of a compressor grid system test apparatus;
FIG. 3 is a connection diagram of a compressor pipe network system half-mechanism model module;
FIG. 4 is a graph of dynamic simulated pressure changes for a compressor piping system;
FIG. 5 is a diagram of a dynamic simulated flow variation of a compressor piping system;
Wherein, 1-compressor inlet regulating valve; 2-centrifugal compressor; 3-a compressor downstream conduit; 4-compressor outlet regulator valve; 5-a sensor; a 6-A/D converter; 7-compressor test signal; 8-valve control signal; 9-data transmission; 10-computer and software part.
Detailed Description
The invention is described in further detail below with reference to the attached drawing figures and to specific examples:
In the description of the present invention, it should be noted that, directions or positional relationships indicated by terms such as "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., are based on directions or positional relationships shown in the drawings, are merely for convenience of description and simplification of description, and do not indicate or imply that the apparatus or element to be referred to must have a specific direction, be constructed and operated in the specific direction, and thus should not be construed as limiting the present invention; the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance; furthermore, unless explicitly specified and limited otherwise, the terms "mounted," "connected," and "connected" are to be construed broadly, and may be either fixed or removable, for example; can be directly connected or indirectly connected through an intermediate medium, and can be communication between two elements. The specific meaning of the above terms in the present invention will be understood in specific cases by those of ordinary skill in the art.
The invention discloses a system and a method for researching surge characteristics of a compressor pipe network system, and referring to fig. 1, the device comprises a hardware part and a software part, wherein the hardware part comprises the compressor pipe network system and a data acquisition and control system, and the software part comprises a mechanism modeling module and a deep learning module.
Fig. 2 shows a block diagram of the apparatus of the present invention, divided into a compressor pipe network device and a signal collection and processing device. The compressor pipe network device consists of a regulating valve, a pipeline, a compressor, a pipeline and a regulating valve, and all the components are connected by flanges. The signal collecting and processing device consists of a temperature sensor, a pressure sensor, a flow sensor, signal collecting and converting equipment and a computer, wherein the sensors are mainly distributed at an upstream inlet and a downstream outlet of the compressor.
The compressor pipe network system consists of a compressor inlet regulating valve 1, a centrifugal compressor 2, a compressor downstream pipeline 3 and a compressor outlet regulating valve 4. Specifically, a compressor inlet regulating valve 1 is arranged on an inlet pipeline of the centrifugal compressor 2, an outlet of the centrifugal compressor 2 is connected with a compressor downstream pipeline 3, and a compressor outlet regulating valve 4 is arranged on the compressor downstream pipeline 3. A test point is arranged between the compressor inlet regulating valve 1 and the centrifugal compressor 2, and a test point is arranged on the inlet of the downstream pipeline 3 of the compressor. One sensor 5 is arranged for each test point, and the sensor 5 for each test point comprises a pressure sensor, a temperature sensor and a flow sensor. Specifically, the sensor mainly comprises a conical restrictor, a pitot tube and a thermometer. The specific parts are connected through flanges.
The signal acquisition and processing means comprise registers, relays and a/D converters 6. In the compressor performance test, the compressor operating point is shifted by changing the opening degrees of the compressor inlet regulating valve 1 and the compressor outlet regulating valve 4. In the compressor pipe network system, a sensor 5 is arranged at an inlet and an outlet of a compressor, test data of the system under different working conditions are obtained, a sensor on each test point transmits a compressor test signal 7 to an A/D converter 6, and finally the test data are transmitted to a software part 10 in a computer. The various sensors are connected with data acquisition software in the computer through a signal conversion device based on a ModBus communication protocol. When the compressor pipe network system is tested, the pressure, the temperature and the flow are collected by a sensor, input into a computer through an A/D converter and are collected by computer software. And finally, compressor valve system test data of pressure, temperature and flow rate which change with time are formed.
The research device of the invention realizes valve control through software, specifically inputs a valve control instruction into a computer and a software part 10, inputs the valve control instruction into an A/D converter 6 through a data transmission 9, outputs a communication instruction and a relay through a ModBus communication protocol and a register, realizes control of a regulating valve through computer software, inputs a valve control signal 8 to a compressor outlet regulating valve 4, realizes change of action of the compressor outlet regulating valve 4, and changes working conditions of the compressor.
Through the process, the performance test of the compressor pipe network system under various working conditions can be performed, and test data for processing the pressure, the temperature and the flow of the compressor are collected.
Based on the data collected by the test equipment, a specific technical route in the computer and software part 10 is shown in fig. 3, and fig. 2 shows a technical route of mechanism modeling and deep learning, wherein the technical route is obtained by firstly obtaining a control equation of a compressor pipe network system through a conservation equation, then taking out key parameters in the equation to carry out deep learning, and further obtaining a semi-mechanism model of the compressor pipe network system. When the key parameters are deeply learned, firstly, test data of the compressor pipe network system under each working condition are used as data, and the neural network is used for training the key parameters so as to realize model identification. And finally, a semi-mechanism model conforming to the dynamic simulation precision is obtained. Therefore, the compressor pipe network system provided by the invention has higher accuracy and practicability, and can greatly reduce the cost of surge test research.
With reference to fig. 2, the method is specifically based on a dynamic simulation model of a target compressor pipe network system, wherein the dynamic simulation model of the compressor pipe network system is a semi-mechanism model, and a semi-mechanism model with higher precision is obtained by performing deep learning on test data on the basis of a compressor pipe mechanism model; and dynamically simulating the surge working condition of the compressor pipe network system by using the semi-mechanism model, and outputting a simulation result. The method for researching the surge characteristics of the compressor pipe network system by combining mechanism modeling and deep learning comprises the following steps.
The software part is a dynamic simulation model of the target compressor pipe network system. The established dynamic simulation model of the compressor pipe network system is a semi-mechanical ash box model, and is characterized by mainly comprising the following components: ① Deducing a mechanism control equation of the compressor, the pipe network and the valve based on mass conservation, a momentum equation and an energy equation, and connecting the compressor, the pipe network, the valve and other parts into a system to realize data transmission and equation decoupling; ② The key parameters in the mechanism control equation (such as a flow-pressure ratio function of a compressor, an opening degree of a valve, a pressure ratio-flow function and the like) are replaced by a fitting function obtained by deep learning by a general experience function.
And for the obtained dynamic simulation system, the user dynamically simulates typical working conditions of the target compressor pipe network system by setting boundary conditions and initial conditions and outputs simulation results. The simulation results include dynamic changes in pressure, temperature, and flow rate upstream and downstream of the compressor over time. The dynamic simulation model of the compressor pipe network system can simulate the test working conditions of the compressor pipe network system of the hardware part, obtain the dynamic changes of the pressure, the temperature and the flow of the corresponding working conditions and carry out comparison analysis; and the surge working condition which is not suitable for testing on hardware equipment can be simulated, a simulation result is output, and the characteristic analysis is performed.
More specifically, the research method comprises the following steps:
Step 1, a mechanism model of a compressor pipe network system is established, a unit comprises a centrifugal compressor, a pipe network, a valve and other components, a characteristic function of pressure, temperature and flow of the unit is obtained, and the mechanism model of each component is obtained by combining mass, momentum and energy conservation equations:
The mass conservation equation is:
the conservation of momentum equation is:
the energy conservation equation is:
Wherein the mass conservation equation and the momentum conservation equation are in tensor form, and i and j represent tensor orders; u, p, T represent speed, pressure and temperature, respectively; ρ is the density. The energy conservation equation is in an unfolding form, u, v and w respectively represent the speed in the x, y and z directions, and h is the specific enthalpy; alpha is the coefficient of thermal conductivity of the fluid; s h is a source item of the fluid; Φ is the thermal energy converted from mechanical energy due to viscosity.
And deducing the conservation equation according to the characteristics of the compressor, the pipe network and the valve to obtain a zero-dimensional control equation of the compressor, the pipe network and the valve as follows.
Control equation for compressor:
Control equation of pipe network:
Valve control equation:
Wherein p, T, w, v are pressure, temperature and mass flow and flow rate; subscripts in and out represent the ingress and egress; h is specific enthalpy; cp is the specific heat of the working medium; v is the volume of the pipeline; r is a gas constant; psi is the performance curve of the compressor; ; n is the rotation speed; l c is the equivalent through-flow length; a 1 is the flow area; d is the diameter of the pipeline; c f is a valve flow resistance characteristic function; k is the valve flow area coefficient.
And 2, obtaining control equations (4) - (6) of the compressor, the pipe network and the valve component according to the equation, and establishing simulation modules of the compressor, the pipe network and the valve module according to the control equations.
For the compressor simulation module, the initial conditions include a compressor dynamic characteristic function ψ and a compressor parameter rotational speed N, an equivalent flow length L c, a flow area a 1. During simulation, the input parameters are pressures at the upper part and the lower part of the compressor, and the mass flow is calculated by the formula (4) when the parameters are output.
For the pipe network simulation module, the initial condition is the equivalent volume V of the pipe network. During simulation, the input parameter is the mass flow entering and exiting the pipe network system, the output parameter is the pressure of the pipe network, and the pipe network pressure is calculated by the formula (5).
For the valve simulation module, the initial conditions include a valve flow resistance function C f, a valve flow area coefficient k, and a valve diameter D. During simulation, the input parameters are the valve opening (the opening is a percentage, the corresponding angle is theta) and the pressures before and after the valve, the output parameters are the valve mass flow, and the valve mass flow is calculated by the formula (6).
And step 3, building a dynamic simulation model of the compressor pipe network system based on the building of the dynamic simulation modules of the compressor, the pipe network and the valve. At this time, the simulation modules of the three components are connected in order. Referring to FIG. 3, the simulation modules can be divided into two types, the first type is a compressor and valve simulation module, the input parameter is upstream and downstream pressure, and the output parameter is mass flow; the second type is a pipe network module, wherein the input parameter is inlet and outlet mass flow, and the output parameter is upstream and downstream pressure. Therefore, when the system is modeled, the two types of modules are needed to be alternately arranged and connected in sequence, so that conservation of input and output parameters is realized, and transmission of pressure, temperature and flow is realized. The boundary conditions of the system are ambient temperature, ambient pressure and the initial temperature, pressure and flow of the components. During simulation, the input condition of the system is the opening degree of each valve in the system, and the output condition is the temperature, pressure and flow of the compressor, the pipe network and the inlet and outlet cross sections of the valves.
On the basis of the dynamic simulation system of the compressor pipe network based on the physical principle, in order to further improve the simulation precision, a semi-mechanical dynamic simulation model needs to be obtained through deep learning, namely, a key function compressor dynamic characteristic function ψ and a valve flow resistance characteristic function C f are optimized from a general experience value to a deep learning value with higher precision and stronger targeting.
And 4, combining a mechanism model of the compressor pipe network system, and performing deep learning by taking a dynamic characteristic function psi of the compressor and a flow resistance characteristic function C f of the valve as key functions. And training key parameters in the model by taking test data of the compressor pipe network system obtained by the hardware part under a plurality of working conditions as a data set to finish model identification.
Specifically, the deep learning technical route with the dynamic characteristic function psi and the valve flow resistance function C f of the compressor is that firstly, multi-station testing is carried out on hardware equipment of a compressor pipe network system, and a plurality of groups of data are collected by changing working conditions of an opening adjusting system of a compressor inlet adjusting valve (1) and a compressor outlet adjusting valve (4) to form a training set of testing data. Secondly, training and verifying key parameters of a compressor pipe network model by using test data acquired by a hardware part through a neural network deep learning method, so as to realize model identification, wherein the method comprises the following steps of:
(1) The collected experimental data are arranged into a two-dimensional array according to a time-variable mode, the obtained data are screened, and the inlet and outlet pressure of the compressor and the flow of the compressor are selected to be used as a data set for training ψ. The dynamic characteristic function psi of the compressor is a key function reflecting the performance characteristic of the compressor, and represents the characteristic that the pressure ratio of the compressor changes along with the flow, and is related to the design parameters of the compressor. For a compressor with a definite structure and a certain rotating speed, the dynamic characteristic function is certain. And training by taking the inlet pressure, the outlet pressure and the flow of the compressor acquired in the test as data sets when the neural network deep learning is performed, wherein the flow of the compressor is taken as an input parameter, and the pressure ratio (the ratio of the outlet pressure to the inlet pressure) of the compressor is taken as an output parameter.
(2) The front and back valve pressure, the valve opening and the valve flow are used as a data set of training C f, and the valve flow resistance function C f reflects a key function of valve flow characteristics and reflects the flow capacity of the valve under the conditions of different opening degrees and the front and back valve pressure ratios. When the neural network deep learning is performed, training is performed by taking the valve front pressure, the valve back pressure, the valve opening and the valve flow acquired in the test as data sets, and taking the valve front-back pressure ratio (the ratio of the valve back pressure to the valve front pressure) and the valve opening as input parameters and the valve flow as output parameters.
(3) Filtering target data, and carrying out noise reduction treatment on the data for training;
(4) The reduced-noise dataset of ψ and the dataset of C f are used for training and recognition and to determine the best training method. In the step 3, two simulation modules, namely a compressor, a valve simulation module and a pipe network module, are indicated; by the generalized least square method, the BP neural network and the NARX neural network are used for deep learning at the same time, compared with the two neural network methods, the simulation precision is taken as a judgment condition, and the neural network with small precision is selected as a final deep learning neural network. And (3) obtaining a training algorithm of the compressor module and the valve module through iteration, determining the proper number of neurons of the hidden layer by using a grid traversal method during each iteration, determining a proper time delay order and a proper neuron training period through debugging parameters, and obtaining the neural network learning method of the current iteration step. And then comparing simulation precision of the current iteration step, if the simulation precision does not meet the precision requirement, repeating calculation to obtain the number of hidden layer neurons in the next iteration step, carrying out Shi Yanjie times and neuron training periods, and carrying out iterative correction for a plurality of times to obtain a neural network learning method with the highest precision, the number of corresponding hidden layer neurons, the optimal time delay order and the neuron training period. After the optimal neural network learning method is obtained, parameter training and identification are carried out on the compressor module and the valve module, and finally a compressor characteristic function ψ and a valve flow resistance characteristic function C f are obtained.
(5) And (3) carrying the compressor dynamic characteristic function psi and the valve flow resistance function C f obtained in the steps into the compressor and valve dynamic simulation module established in the step (3), and carrying out simulation accuracy verification. When the simulation system is verified, the input parameter of the simulation module of the compressor is upstream and downstream pressure, the output parameter is compressor mass flow, and the simulation value is compared with the test value of the compressor under the same condition; the input parameters of the valve simulation module are valve opening and front and back pressures of the valve, the output parameters are valve mass flow, and the simulation is compared with valve mass flow test values under the same condition. If the precision meets the standard, the deep learning and the model identification are considered to be successful, the obtained dynamic characteristic function psi of the compressor and the simulation module of the compressor, the valve flow resistance function C f and the valve simulation module are reliable, and the system precision verification of the next link is input; and if the precision does not meet the standard, repeating the step of deep learning.
(6) And (3) obtaining a compressor dynamic simulation module and a compressor dynamic characteristic function psi, a valve dynamic simulation module and a flow resistance function C f which meet simulation precision through multiple iterations and precision tests, and connecting the compressor dynamic simulation module, the pipe network dynamic simulation module and the valve dynamic simulation module according to the compressor pipe network system dynamic modeling method in the step (3), so as to realize parameter transfer and decoupling of equations (4) - (6). The input parameters of the dynamic simulation system of the compressor pipe network system are functions of the changes of the opening of each valve in the system along with time, and the output parameters are the temperatures, the pressures and the flow of the compressor, the pipe network and the inlet and outlet cross sections of the valves. The working point of the compressor pipe network system migrates along with the action of the valve, and the temperature, the pressure and the flow are dynamically changed.
(7) And testing and correcting the dynamic simulation system of the compressor pipe network system until the precision requirement of dynamic simulation is met. At the moment, the dynamic simulation precision of the system is verified by taking the compressor pipe network system as an object, the boundary condition of the compressor pipe network system is consistent with the test condition during verification, the opening degree of an inlet valve, the environmental parameter and the power of the compressor are taken as input quantities, the pressure, the temperature and the flow of each section before and after the compressor and the valve before and after the valve are taken as output flows, and the simulation and the test results are compared. If the precision is met, the obtained dynamic simulation system of the compressor pipe network system can be used for researching the dynamic simulation characteristics of surge; if the precision does not meet the requirement, error analysis is carried out, and dynamic modeling of the system is carried out again.
(8) And simulating the surge working condition of the compressor based on the semi-mechanism dynamic simulation model of the compressor pipe network system, and enabling the compressor pipe network system to enter the surge working condition by adjusting the opening of the upstream and downstream valves of the compressor. And outputting a dynamic simulation result of the temperature, the pressure and the flow of the compressor in the process of entering the surge working condition by the compressor pipe network system, extracting the surge characteristic of the compressor in the signal fluctuation, and taking the surge characteristic as a vibration part in the simulation result, wherein the vibration part comprises amplitude frequency and the like.
(9) And (3) based on the semi-mechanism model obtained in the step (8), the precision requirement of dynamic simulation is met, and the dynamic change and response characteristics of the pressure, the temperature and the flow under the surge working condition of the compressor pipe network system are obtained. By simulating the surge working condition, the cost of the surge test research can be reduced, and the risk existing in the surge test research is avoided.
And carrying out dynamic simulation and analysis on the surge working condition by obtaining a dynamic simulation system of the compressor pipe network. According to compressor theory, surge is caused by excessive resistance downstream of the compressor and failure of the compressor's power capacity to impart sufficient energy to the working fluid out of the system. At this time, the pressure of the downstream pipe network is transiently higher than the outlet pressure of the compressor, the flow is reduced, even the backflow condition occurs, and then the pressure oscillation is caused, and the system enters the surge working condition. In actual working conditions, when the opening of a valve at the downstream of the compressor is reduced, the resistance of a pipe network at the downstream of the compressor is increased, so that the system can enter surge. In the simulation system, the system enters a surge working condition by adjusting the opening of inlet and outlet valves of the compressor. In a certain dynamic simulation study, the opening of the valve is gradually adjusted, and the method concretely comprises the following steps: during 3-8s, the opening of the outlet valve is closed from 0.5 to 0.1 in stages, the working condition of the compressor system is changed, and as the opening of the outlet valve is reduced, the outlet pressure of the compressor is increased, the flow is reduced, and the change of the outlet pressure is stepped, and the outlet pressure is consistent with the opening of the valve. The pressure change lag and the flow change accord with the working mechanism of the compressor. At 11s, the outlet valve is closed from 0.1 to 0.03, the system enters a surge state, the pressure and the flow show periodic fluctuation, the pressure fluctuation is consistent with the flow fluctuation period, and the pressure fluctuation lags behind the flow fluctuation. The opening of the inlet valve is gradually reduced at 15s and 18s, the surge characteristic is changed, the vibration amplitude of the flow and the pressure is reduced, and the frequency is increased.
Fig. 3 and 4 show the results of dynamic simulation of pressure and flow during surge in the semi-mechanical model according to the invention. In a semi-mechanism simulation model of a compressor pipe network system, the opening degrees of a compressor inlet regulating valve and a compressor outlet regulating valve are changed, the compressor outlet regulating valve is gradually closed from a full opening to a small opening at a specific position, a compressor working condition point is changed, a working condition point moves leftwards along a performance curve, and outlet pressure is gradually increased until a surge state is entered. Meanwhile, the opening degree of the inlet regulating valve of the compressor is gradually reduced after the surge, and the surge characteristic is changed. The calculation example fully proves that the dynamic simulation system of the compressor system can carry out simulation research on the surge characteristics of the compressor.
The surge characteristic research system of the compressor pipe network system with the probability mechanism modeling and the deep learning fully plays the advantages of test research and simulation research, adopts an advanced modeling method to obtain dynamic simulation of pressure, temperature and flow change under typical working conditions of the compressor pipe network system, simulates and analyzes working conditions with test risks such as surge and the like, and has reliable simulation results and wide application range.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (10)
1. A method for researching surge characteristics of a compressor pipe network system is characterized by comprising the following steps:
Step 1, a mechanism model of a compressor pipe network system is established, wherein the mechanism model comprises a mass conservation equation, a momentum conservation equation and an energy conservation equation;
Step 2, establishing a compressor zero-dimensional control equation, a pipe network zero-dimensional control equation and a valve zero-dimensional control equation through a mechanism model; the valve comprises a compressor inlet regulating valve (1) and a compressor outlet regulating valve (4); respectively establishing simulation modules of a compressor, a pipe network and a valve through a zero-dimensional control equation;
Step 3, connecting three simulation modules to form a simulation system of the compressor pipe network system; the input of the compressor simulation module and the valve simulation module is the output of the pipe network simulation module, and the input of the pipe network simulation module is the output of the compressor simulation module and the valve simulation module;
Step 4, taking the opening of the compressor inlet regulating valve (1) and the opening of the compressor outlet regulating valve (4) as variables, and performing multi-station test on a compressor pipe network system to obtain a test number set; the test data set comprises a valve test data set and a compressor test data set, and the neural network is trained through the valve test data set to obtain a valve flow resistance characteristic function; training the neural network through a compressor test number set to obtain a compressor characteristic function;
Step 5, the valve flow resistance characteristic function is brought into a valve simulation module, the compressor characteristic function is brought into a compressor simulation module, the valve simulation module and the compressor simulation module are simulated, if the simulation result precision of the two simulation modules meets the standard, the step 6 is executed, otherwise, the step 4 is repeated, and the valve flow resistance characteristic function and the compressor characteristic function are obtained again through neural network learning;
step 6, bringing a valve flow resistance characteristic function and a compressor characteristic function meeting the precision requirement into a simulation system of a compressor pipe network system, and completing parameter transfer and control equation decoupling among three simulation modules to obtain a dynamic simulation system of the compressor pipe network system;
And 7, simulating a compressor surge working condition through a dynamic simulation system of the compressor pipe network system to obtain the compressor surge characteristic, wherein the compressor surge characteristic comprises compressor temperature, pressure and flow.
2. The method for studying surge characteristics of a compressor pipe network system according to claim 1, wherein in the step 1, the mass conservation equation is:
the momentum conservation equation is:
The energy conservation equation is:
3. the method for studying surge characteristics of a pipe network system of a compressor according to claim 1, wherein in the step 2, a zero-dimensional control equation of the compressor is:
the zero-dimensional control equation of the pipe network is as follows:
the zero-dimensional control equation of the valve is as follows:
wherein p, T, w, v are pressure, temperature and mass flow and flow rate, respectively; h is specific enthalpy; cp is the specific heat of the working medium; v is the volume of the pipeline; r is a gas constant; psi is the performance curve of the compressor; n is the rotation speed; l c is the equivalent through-flow length; a 1 is the flow area; d is the diameter of the pipeline; c f is a valve flow resistance characteristic function; k is the valve flow area coefficient.
4. The method for studying surge characteristics of a pipe network system of a compressor according to claim 1, wherein in the step 2, in the compressor simulation module, input parameters are upstream pressure of the compressor and downstream pressure of the compressor, and output parameters are mass flow of the compressor; in the pipe network simulation module, the input parameters are the mass flow entering the pipe network and the mass flow output by the pipe network, and the output parameters are the pressure of the pipe network; in the valve simulation module, input parameters are valve opening, valve front pressure and valve back pressure, and output parameters are valve mass flow.
5. The method for researching surge characteristics of a compressor pipe network system according to claim 1, wherein in step 4, in the process of training the neural network through a valve test data set and training the neural network through a compressor test data set, each neural network performs deep learning simultaneously through a generalized least square method by using a BP neural network and a NARX neural network, simulation precision of the two neural networks is compared, and a neural network with small precision is selected as a final neural network.
6. The method for researching surge characteristics of a pipe network system of a compressor according to claim 1, wherein in the step 4, the test data set is used after filtering and noise reduction treatment.
7. The method for researching surge characteristics of a compressor pipe network system according to claim 1, wherein in step 6, in the compressor pipe network simulation system, the input parameters are functions of the opening of a compressor inlet regulating valve (1) and the opening of a compressor outlet regulating valve (4) with time; the output parameters are the temperature, pressure and flow of the compressor, the temperature, pressure and flow of the pipe network, and the temperature, pressure and flow of the inlet and outlet sections of the valve.
8. The method for researching surge characteristics of a compressor pipe network system according to claim 1, wherein in step 6, simulation results of a dynamic simulation system of the compressor pipe network system are tested and corrected until the precision requirement of the dynamic simulation is met, and step 7 is executed.
9. The method for researching the surging characteristics of the compressor pipe network system according to claim 1, wherein in the step 7, when the surging working condition of the compressor is simulated, the compressor is caused to enter the surging working condition by adjusting the opening of a compressor inlet regulating valve (1) and the opening of a compressor outlet regulating valve (4); when the compressor surge characteristic is obtained, the compressor surge characteristic in the signal fluctuation is extracted.
10. A compressor pipe network system surge characteristic research system for implementing the research method according to any one of claims 1-9, characterized by comprising a centrifugal compressor (2), wherein a compressor inlet regulating valve (1) is arranged in front of the centrifugal compressor (2), and a compressor outlet regulating valve (4) is arranged at the outlet of the centrifugal compressor (2);
A test point is arranged between the compressor inlet regulating valve (1) and the centrifugal compressor (2), a test point is arranged between the centrifugal compressor (2) and the compressor outlet regulating valve (4), the test point is connected with an A/D converter (6), and the A/D converter is connected with a computer and a software part (10);
And the test point is provided with a pressure sensor, a temperature sensor and a flow sensor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210994751.4A CN115356096B (en) | 2022-08-18 | 2022-08-18 | System and method for researching surge characteristics of compressor pipe network system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210994751.4A CN115356096B (en) | 2022-08-18 | 2022-08-18 | System and method for researching surge characteristics of compressor pipe network system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115356096A CN115356096A (en) | 2022-11-18 |
| CN115356096B true CN115356096B (en) | 2024-05-14 |
Family
ID=84002477
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210994751.4A Active CN115356096B (en) | 2022-08-18 | 2022-08-18 | System and method for researching surge characteristics of compressor pipe network system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115356096B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116517867B (en) * | 2023-06-28 | 2023-10-03 | 国网江苏省电力有限公司常州供电分公司 | Method and device for diagnosing and suppressing surge of compressor |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20140124082A (en) * | 2013-04-15 | 2014-10-24 | 두산중공업 주식회사 | Compressor performance test apparatus and method for turbine |
| CN109779894A (en) * | 2018-12-29 | 2019-05-21 | 西安交通大学 | A kind of reciprocating compressor fault diagnosis system and method based on neural network algorithm |
| CN110307138A (en) * | 2018-03-20 | 2019-10-08 | 恩尔赛思有限公司 | A kind of design, measurement and the optimization method of the multi-compressor system about energy efficiency |
| CN112594062A (en) * | 2020-11-13 | 2021-04-02 | 南京航空航天大学 | Simulation method for surge detection and surge elimination control verification |
| CN113688510A (en) * | 2021-08-09 | 2021-11-23 | 同济大学 | Surging prediction method of centrifugal compression system for fuel cell vehicle |
| CN114348295A (en) * | 2022-01-05 | 2022-04-15 | 北京临近空间飞艇技术开发有限公司 | Test system and method for measuring aerodynamic performance of helium gas compressor |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5514047B2 (en) * | 2010-09-02 | 2014-06-04 | 株式会社日立製作所 | Apparatus and method for simulating motor-driven compressor system |
-
2022
- 2022-08-18 CN CN202210994751.4A patent/CN115356096B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20140124082A (en) * | 2013-04-15 | 2014-10-24 | 두산중공업 주식회사 | Compressor performance test apparatus and method for turbine |
| CN110307138A (en) * | 2018-03-20 | 2019-10-08 | 恩尔赛思有限公司 | A kind of design, measurement and the optimization method of the multi-compressor system about energy efficiency |
| CN109779894A (en) * | 2018-12-29 | 2019-05-21 | 西安交通大学 | A kind of reciprocating compressor fault diagnosis system and method based on neural network algorithm |
| CN112594062A (en) * | 2020-11-13 | 2021-04-02 | 南京航空航天大学 | Simulation method for surge detection and surge elimination control verification |
| CN113688510A (en) * | 2021-08-09 | 2021-11-23 | 同济大学 | Surging prediction method of centrifugal compression system for fuel cell vehicle |
| CN114348295A (en) * | 2022-01-05 | 2022-04-15 | 北京临近空间飞艇技术开发有限公司 | Test system and method for measuring aerodynamic performance of helium gas compressor |
Non-Patent Citations (2)
| Title |
|---|
| 基于双通道伪键合图的压气机建模与仿真研究;黄林;程刚;许伟;范仕龙;;海军工程大学学报;20180215(第01期);全文 * |
| 离心式压缩机组喘振控制系统特性的模拟研究;侯大立;孙志广;张芳;;石油机械;20180710(第07期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115356096A (en) | 2022-11-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110197049B (en) | Non-metal pipeline leakage positioning method based on transient inverse problem | |
| CN105042339B (en) | One kind is based on nondimensional leakage of finished oil pipeline amount estimating system and method | |
| CN105550405B (en) | A kind of city planting ductwork hydraulic model construction method | |
| CN115356096B (en) | System and method for researching surge characteristics of compressor pipe network system | |
| CN111859746B (en) | Method for predicting variable working condition performance of turbomachinery based on flow field reconstruction | |
| CN113551744A (en) | Ultrasonic flowmeter performance online monitoring method and system | |
| CN102419827B (en) | Radial basis function (RBF) neural network-based boiling heat exchanging prediction method | |
| CN102135021A (en) | Method for predicting shaft power of industrial extraction condensing steam turbine | |
| CN111272436B (en) | Gas turbine NOx prediction emission monitoring system and method | |
| CN110146249A (en) | A kind of pressure-measuring pipe road optimum design method for wind tunnel test | |
| CN117189713A (en) | Hydraulic system fault diagnosis method based on digital twin driving | |
| Huang et al. | Nox emission estimation in gas turbines via interpretable neural network observer with adjustable intermediate layer considering ambient and boundary conditions | |
| CN115344019A (en) | Natural gas metering flow adjusting process based on composite intelligent algorithm | |
| Fernelius et al. | Mapping efficiency of a pulsing flow-driven turbine | |
| CN105447256B (en) | A kind of enhancing excitation emulation genetic optimization method | |
| CN117090831A (en) | Hydraulic system fault diagnosis framework with twinning application layer | |
| CN110736526A (en) | High-temperature gas flowmeter calibration device and method for liquid oxygen kerosene engine | |
| CN114462336B (en) | Method for calculating average temperature of coolant of main pipeline of nuclear reactor | |
| Schillaci et al. | Air ejector analysis in normal and abnormal modes, oriented to control purposes in aircraft systems | |
| Hu et al. | Predictions of on-engine efficiency for the radial turbine of a pulse turbocharged engine | |
| CN109977598A (en) | For the loading analysis model building method and analysis method of valve downstream drain pipe | |
| Bailey | The aerodynamic performance of an annular S-shaped duct | |
| CN113341694B (en) | Equivalent verification system and method for flow control valve control strategy | |
| CN115470726A (en) | Hypersonic inlet channel flow field rapid prediction method based on deep learning | |
| Gu et al. | A modified clustering procedure for energy consumption monitoring in the steam turbine considering volume effect |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |