Converter valve assembly state evaluation method based on Simulink simulation

Abstract

The invention discloses a converter valve assembly state evaluation method based on Simulink simulation. The method comprises the following steps: S1, acquiring monitoring data of a converter valve in actual operation through a sensor; s2, through a converter valve which is actually operated by a Simulink-based flexible DC power transmission simulation module, obtaining corresponding simulation generation data; s3, taking the monitoring data of the converter valve as comparison data, and comparing the comparison data with simulation generation data through a data consistency analysis algorithm; s3.1, constructing a similarity evaluation method based on time-frequency feature fusion; s3.2, simulating a component state by adopting an improved PSO method, and obtaining simulation data; and S4, verifying analysis results of the detection data and the simulation generation data. According to the converter valve assembly state evaluation method based on Simulink simulation, the specific state of the sub-assemblies of the converter valve can be obtained based on background software simulation without independently monitoring the sub-assemblies and manually checking the sub-assemblies on site one by one, so that the state judgment of the converter assembly can be accurately determined.

Description

Technical Field

[0001] The invention relates to the technical field of circulating valves, and in particular to a method for evaluating the state of a converter valve component based on Simulink simulation. Background Art

[0002] As the core equipment of the UHVDC transmission system, the flexible direct current converter valve plays a key role in rectification and inversion. However, the converter valve is usually composed of a large number of power electronic modules and bridge arm resistors and inductors. The current fault diagnosis method requires the converter valve to monitor each component and transmit the data through communication to determine the status of each component, which is time-consuming and labor-intensive. In addition, a large number of monitoring devices will increase the complexity of the system and reduce reliability. If the operating status of the converter valve can be simulated, the status of the converter valve component can be determined only by comparing the input and output characteristics, which will help guide the inspection and maintenance of the converter valve and improve the safety level of power grid operation. Summary of the invention

[0003] The purpose of the present invention is to provide a method for evaluating the status of a converter valve component based on Simulink simulation. It does not require individual monitoring of each sub-component and manual on-site inspection one by one. The specific sub-component status of the converter valve can be obtained based on background software simulation, thereby accurately determining the status of the converter component.

[0004] The present invention provides a method for evaluating the state of a converter valve assembly based on Simulink simulation, comprising the following steps:

[0005] S1. Obtaining monitoring data of the actually operated converter valve through a sensor;

[0006] S2, the converter valve is actually operated through the flexible direct current transmission simulation module based on Simulink and the corresponding simulation generated data is obtained;

[0007] S3, using the monitoring data of the converter valve as comparison data and the simulation generated data for comparison through a data consistency analysis algorithm;

[0008] S3.1. Construct a similarity evaluation method based on time-frequency feature fusion;

[0009] S3.2, using the improved PSO method to simulate component states and obtain simulation data;

[0010] S4. Verify the analysis results of the test data and the simulation generated data.

[0011] Preferably, in step S1, the actually operating converter valve is an object model for actual fault diagnosis.

[0012] Preferably, in step S2, the flexible direct current transmission simulation of Simulink requires an object model for actual fault diagnosis and an environment model for actual operation of the converter valve.

[0013] Preferably, in step S2, when simulating and modeling the converter valve, the submodules, resistor and inductor modules of the converter valve are replaced by three switching modules.

[0014] Preferably, in step S3, S3.1, constructing a similarity evaluation method based on time-frequency feature fusion;

[0015] First, construct the time domain characteristic matrix; transform the DC voltage U dc 、Upper bridge arm voltage curve U up , lower bridge arm voltage curve U down , the three-phase AC current i of the converter valve a 、i b 、i c , three-phase AC voltage u a 、u b 、u c , the environmental temperature and humidity time series data monitoring phase time series sequence is evenly divided;

[0016] Each monitored quantity is expressed as:

[0017] X i ={x i,t |t=1,2,...,N};

[0018] Among them, X i Represents different monitoring quantities, x i,t is the value of a certain monitoring quantity at the tth sampling moment, and N represents the Nth sampling moment; The segmented aggregation approximation method PAA is expressed as

[0019]

[0020] in, for The mth sampling value, k is the length ratio before and after PAA; According to x i,t calculate:

[0021]

[0022] At the same time, the maximum and minimum values ​​are normalized The value ranges from 0 to 1, as shown below:

[0023]

[0024] in, They are The maximum and minimum values ​​of

[0025] Then, the standardized According to the following formula, it is transformed into the polar coordinate system and forms the time domain characteristic matrix:

[0026]

[0027] Among them, r i is the polar diameter of the polar coordinates, is the polar angle; t i is the sampling time step;

[0028] Then, construct the frequency domain feature matrix;

[0029] The following maximum overlap discrete wavelet transform (MODWT) method is used to decompose the above monitoring signal to obtain its frequency domain characteristics;

[0030]

[0031] Among them, j is the scale parameter, k is the translation parameter, i represents different monitoring quantities; h is the high-frequency decomposition filter, g is the low-frequency decomposition filter, and n is the number of decomposition layers; the frequency domain matrix is ​​obtained As shown below:

[0032] F F ={w i,j,k ,v i,j,k};

[0033] Concatenate the time domain / frequency domain feature matrices to obtain the time-frequency domain matrix TF ;

[0034] TF=[T F ; F F ];

[0035] Finally, the time-frequency domain matrix is ​​input into the convolutional neural network for feature extraction and evaluation. A ;

[0036] A = sigmod[tanh(W TF ·TF)].

[0037] Preferably, in step S3,

[0038] The converter valve has N components, and the module states of the converter valve components input for simulation are set to be N tuples;

[0039] S=(s1,s2,…,s i ,…,s N )i=1,2,3,…N;

[0040] Among them, si The converter valve assembly module is in short circuit, normal or open circuit state; as shown in the following formula:

[0041]

[0042] The particle swarm algorithm is used to use the L2-norm as an adaptive function to approximate the similarity of the measurement data, as shown in the following formula:

[0043]

[0044] Among them, i is the monitoring quantity;

[0045] The state of the converter valve assembly is used as the particle coordinates, and the particle position and velocity of the particle swarm algorithm are updated as follows:

[0046] v i k+1 =wv i k +c1rand(p bi k -q i k )+c2rand(g b k -q i k );

[0047] q i k+1 =q i k +v i k+1 ;

[0048] In the formula, v i k ,q i k 、p i k is the speed, position and historical optimal point of the i-th particle at the k-th iteration; c1 and c2 are acceleration factors; rand is a random number; g b k is the historical optimal point of the group at the kth iteration, and w is the inertia factor.

[0049] Preferably, in step S4, it is verified whether the simulation state is correct and the state is output correctly, otherwise the cycle continues.

[0050] Therefore, the present invention adopts the above-mentioned method for evaluating the status of the converter valve component based on Simulink simulation. There is no need to monitor each sub-component separately and there is no need to manually check them one by one on site. The specific sub-component status of the converter valve can be obtained based on the background software simulation, thereby realizing accurate determination of the converter component status judgment.

[0051] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A schematic diagram of a fault diagnosis model of a method for evaluating a state of a converter valve assembly based on Simulink simulation according to the present invention;

[0053] Figure 2 A schematic diagram of a converter valve assembly module with status in a converter valve assembly status evaluation method based on Simulink simulation according to the present invention;

[0054] Figure 3 A schematic diagram of a modular multi-level converter valve simulation model of a converter valve assembly state assessment method based on Simulink simulation of the present invention. DETAILED DESCRIPTION

[0055] The technical solution of the present invention is further described below through the accompanying drawings and embodiments.

[0056] Unless otherwise defined, technical or scientific terms used in the present invention shall have the common meanings understood by one having ordinary skills in the field to which the present invention belongs.

[0057] The words "first", "second" and similar terms used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. Words such as "include" or "comprises" and similar terms mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Words such as "connect" or "connected" and similar terms are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0058] Embodiment 1

[0059] like Figure 1-Figure 3 As shown, the present invention provides a method for evaluating the state of a converter valve assembly based on Simulink simulation, comprising the following steps:

[0060] The fault diagnosis model of converter valve components based on Simulink simulation consists of three parts: the converter valve in actual operation, the flexible direct current transmission simulation based on Simulink, and the converter valve component status determination module based on the optimization algorithm.

[0061] S1. Obtaining monitoring data of the actually operated converter valve through a sensor;

[0062] The actual operating converter valve is the object model of actual fault diagnosis;

[0063] S2, simulating the actual operation of the converter valve through the flexible direct current transmission simulation based on Simulink and obtaining the corresponding simulation generated data;

[0064] By setting different states of the converter valve components in the Simulink simulation program, the actual operating state of the converter valve can be simulated and the simulation operation data can be obtained; Simulink's flexible direct current transmission simulation requires an object model for actual fault diagnosis and an environmental model for the actual operation of the converter valve; the PoweSystem device library is used in Simulink to build a simulation program corresponding to the actual situation, and the parameters are set according to the actual operating conditions of the converter valve.

[0065] These include: the operating environment of the converter valve that is consistent with the actual situation. Including but not limited to the sending end, receiving end, converter transformer, closing resistor, AC line, DC line, etc.

[0066] A converter valve that is consistent with actual operating characteristics consisting of power electronic modules such as IGBT\IGCT and their control strategies.

[0067] When the converter valve simulation program is used to model the converter valve, the submodules, resistor and inductor modules of the converter valve are replaced by a three-switch module, wherein the first switch is short-circuited, the second switch is normal, and the third switch is open-circuited.

[0068] Therefore, the modular multilevel converter valve simulation model consisting of the newly designed state-controlled valve assembly module is as follows: Figure 2 As shown;

[0069] S3, using the monitoring data of the converter valve as comparison data and the simulation generated data for comparison through a data consistency analysis algorithm;

[0070] (1) Constructing a similarity evaluation method based on time-frequency feature fusion

[0071] First, construct the time domain feature matrix;

[0072] The DC voltage U dc 、Upper bridge arm voltage curve U up , lower bridge arm voltage curve U down , the three-phase AC current i of the converter valve a、i b 、i c , three-phase AC voltage u a 、u b 、u c , the environmental temperature and humidity time series data monitoring phase time series sequence is evenly divided;

[0073] Each monitored quantity is expressed as:

[0074] X i ={x t |t=1,2,…,N t};

[0075] Among them, X i Represents different monitoring quantities; represents the maximum and minimum standardization; X i The segmented aggregation approximation method PAA is expressed as:

[0076]

[0077] in, for The mth sampling value, k is the length ratio before and after PAA; According to x i,t calculate:

[0078] As shown below:

[0079]

[0080] At the same time, the maximum and minimum standardization is adopted The value ranges from 0 to 1, as shown below:

[0081]

[0082] in, They are The maximum and minimum values ​​of

[0083] Then, the standardized According to the following formula, it is transformed into the polar coordinate system and forms the time domain characteristic matrix:

[0084]

[0085] Among them, r i is the polar diameter of the polar coordinates, is the polar angle; t i is the sampling time step;

[0086] Then, construct the frequency domain feature matrix;

[0087] The following maximum overlap discrete wavelet transform (MODWT) method is used to decompose the above monitoring signal to obtain its frequency domain characteristics;

[0088]

[0089] Among them, j is the scale parameter, k is the translation parameter, i represents different monitoring quantities; h is the high-frequency decomposition filter, g is the low-frequency decomposition filter, and n is the number of decomposition layers; the frequency domain matrix is ​​obtained As shown below:

[0090] F F ={w i,j,k ,v i,j,k};

[0091] Concatenate the time domain / frequency domain feature matrices to obtain the time-frequency domain matrix TF ;

[0092] TF=[T F ; F F ];

[0093] Finally, the time-frequency domain matrix is ​​input into the convolutional neural network for feature extraction and evaluation. A ;

[0094] A = sigmod[tanh(W TF TF)];

[0095] (2) Use the improved PSO method to simulate component status and obtain simulation data

[0096] According to the newly designed converter valve assembly module with state, assuming that the converter valve has N components, the state of the converter valve assembly module input for simulation is set to be an N-tuple;

[0097] S=(s1,s2,…,s i ,…,s N )i=1,2,3,…N;

[0098] Among them, s i The converter valve assembly module is in short circuit, normal or open circuit state; as shown in the following formula:

[0099]

[0100] The particle swarm algorithm is used to use the L2-norm as an adaptive function to approximate the similarity of the measurement data, as shown in the following formula:

[0101]

[0102] Among them, i is the monitoring quantity;

[0103] The state of the converter valve assembly is used as the particle coordinates, and the particle position and velocity of the particle swarm algorithm are updated as follows:

[0104] v i k+1 =wv i k +c1rand(p bi k -q i k )+c2rand(g b k -q i k );

[0105] q i k+1 =q i k +v i k+1 ;

[0106] In the formula, v i k ,q i k 、p i k is the speed, position and historical optimal point of the i-th particle at the k-th iteration; c1 and c2 are acceleration factors; rand is a random number; g b k is the historical optimal point of the group at the kth iteration, and w is the inertia factor.

[0107] Due to the different requirements for step size refinement in the early and late stages of optimization, the ordinary particle swarm algorithm with fixed inertia factor and acceleration factor often has the problem of "premature maturity" of particle swarm. Adaptive inertia weight and acceleration factor are used to improve this problem. The adaptive weight calculation formula is as follows:

[0108]

[0109] In the formula, w n 、c 1,n 、c 2,n is the inertia weight and acceleration factor at the nth iteration; w max is the maximum value of the inertia weight; c 1,max 、c 2,max is the maximum value of the acceleration factor; N max is the maximum number of iterations.

[0110] If |X|2 approaches 0, it means that the current simulated converter valve component state is similar to the actual converter valve component state. Therefore, when the fitness function converges, the converter valve state is output and input into the flexible direct current transmission simulation program based on Simulink for simulation and operation, and the simulation data is generated to record its output voltage curve U dc 、Upper bridge arm voltage curve U up , the lower bridge arm voltage curve U down . And the three-phase AC current i of the converter valve a 、i b and i c And three-phase AC voltage u a 、u b and u c and ambient temperature and humidity;

[0111] S4. Verify the analysis results of the test data and the simulation generated data.

[0112] In step S4, verify whether the simulation state is correct and output the state correctly, otherwise continue the loop.

[0113] According to the simulation generated data and the actual operation data, the similarity evaluation method of time-frequency feature fusion established in step S3 (1) is used to determine whether the states are consistent. If they are consistent, the states of the components of the converter valve are output to realize the operation state judgment of the components of the converter valve.

[0114] If not consistent, repeat step (2) in step S3.

[0115] Therefore, the present invention adopts the above-mentioned method for evaluating the status of the converter valve component based on Simulink simulation. There is no need to monitor each sub-component separately and there is no need to manually check them one by one on site. The specific sub-component status of the converter valve can be obtained based on the background software simulation, thereby realizing accurate determination of the converter component status judgment.

[0116] The above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for evaluating the state of a converter valve assembly based on Simulink simulation, characterized in that: The following steps are involved: S1. Obtaining monitoring data of the actually operated converter valve through a sensor; S2, the converter valve is actually operated through the flexible direct current transmission simulation module based on Simulink and the corresponding simulation generated data is obtained; S3, using the monitoring data of the converter valve as comparison data and the simulation generated data for comparison through a data consistency analysis algorithm; S3.

1. Construct a similarity evaluation method based on time-frequency feature fusion; S3.2, using the improved PSO method to simulate component states and obtain simulation data; S4. Verify the analysis results of the test data and the simulation generated data.

2. A method for evaluating the state of a converter valve assembly based on Simulink simulation according to claim 1, characterized in that: In step S1, the actually operating converter valve is used as the object model for actual fault diagnosis.

3. The method for evaluating the state of a converter valve assembly based on Simulink simulation according to claim 1, characterized in that: In step S2, the flexible direct current transmission simulation of Simulink requires an object model for actual fault diagnosis and an environment model for actual operation of the converter valve.

4. The method for evaluating the state of a converter valve assembly based on Simulink simulation according to claim 1, characterized in that: In step S2, when simulating and modeling the converter valve, the submodules, resistor and inductor modules of the converter valve are replaced by three switching modules.

5. The method for evaluating the state of a converter valve assembly based on Simulink simulation according to claim 1, characterized in that: In step S3, S3.1, construct a similarity evaluation method based on time-frequency feature fusion; First, construct the time domain characteristic matrix; transform the DC voltage U dc 、Upper bridge arm voltage curve U up , lower bridge arm voltage curve U down , the three-phase AC current i of the converter valve a 、i b 、i c , three-phase AC voltage u a 、u b 、u c , the environmental temperature and humidity time series data monitoring phase time series sequence is evenly divided; Each monitored quantity is expressed as: X i ={x i,t |t=1,2,…,N}; Among them, X i Represents different monitoring quantities, x i,t is the value of a certain monitoring quantity at the tth sampling time, N represents the Nth sampling time; X i The segmented aggregation approximation method PAA is expressed as in, for The mth sampling value, k is the length ratio before and after PAA; According to x i,t calculate: At the same time, the maximum and minimum values ​​are normalized The value ranges from [0,1] as shown below: in, They are The maximum and minimum values ​​of Then, the standardized According to the following formula, it is transformed into the polar coordinate system and forms the time domain characteristic matrix: Among them, r i is the polar diameter of the polar coordinates, is the polar angle; t i is the sampling time step; Then, construct the frequency domain feature matrix; The following maximum overlap discrete wavelet transform (MODWT) method is used to decompose the above monitoring signal to obtain its frequency domain characteristics; Among them, j is the scale parameter, k is the translation parameter, i represents different monitoring quantities; h is the high-frequency decomposition filter, g is the low-frequency decomposition filter, and n is the number of decomposition layers; the frequency domain matrix is ​​obtained As shown below: F F ={w i,j,k ,v i,j,k }; Concatenate the time domain / frequency domain feature matrices to obtain the time-frequency domain matrix TF ; TF=[T F ;F F ]; Finally, the time-frequency domain matrix is ​​input into the convolutional neural network for feature extraction and evaluation. A ; A=sigmod[tanh(W TF ·TF)]。 6. The method for evaluating the state of a converter valve assembly based on Simulink simulation according to claim 1, characterized in that: In step S3, The converter valve has N components, and the module states of the converter valve components input for simulation are set to be N tuples; S=(s1,s2,…,s i ,…,s N )i=1,2,3,…N; Among them, s i The converter valve assembly module is in short circuit, normal or open circuit state; as shown in the following formula: The particle swarm algorithm is used to use the L2-norm as an adaptive function to approximate the similarity of the measurement data, as shown in the following formula: Among them, i is the monitoring quantity; The state of the converter valve assembly is used as the particle coordinates, and the particle position and velocity of the particle swarm algorithm are updated as follows: v i k+1 =wv i k +c1rand(p bi k -q i k )+c2rand(g b k -q i k ); q i k+1 =q i k +v i k+1 ; In the formula, v i k ,q i k 、p i k is the speed, position and historical optimal point of the i-th particle at the k-th iteration; c1 and c2 are acceleration factors; rand is a random number; g b k is the historical optimal point of the group at the kth iteration, and w is the inertia factor.

7. The method for evaluating the state of a converter valve assembly based on Simulink simulation according to claim 1, characterized in that: In step S4, verify whether the simulation state is correct and output the state correctly, otherwise continue the loop.

Images