SLCC converter fault detection method based on multi-module cooperation

Abstract

The invention provides an SLCC converter fault detection method based on multi-module cooperation, and belongs to the technical field of power systems, and the method comprises the steps: collecting the electrical parameter data of each module of an SLCC, calculating a coupling parameter, and constructing an incidence matrix; dynamically updating the incidence matrix to obtain an updated incidence matrix; an SLCC topological graph is constructed; on the basis of the incidence relation of the topological graph, node features are updated through an iterative formula, and final target features are obtained; inputting the final target features into a trained linear classifier, and judging the fault type at the node i and at the (t + 1) th moment by the linear classifier; and continuing to obtain the updated incidence matrix along with the time, and judging the fault type at the node i at the next time through the linear classifier. According to the invention, accurate tracing of fault causes of multi-module cooperation is realized, and the problem that only a single fault module can be positioned and cross-module cascading fault causes cannot be identified in the prior art is solved.

Description

Technical Field

[0001] This invention belongs to the field of power system technology, specifically relating to a fault detection method for SLCC converters based on multi-module collaboration. Background Technology

[0002] DC technology based on SLCC (Multi-Source Adaptive Commutation Converter) has become a key solution for adapting to ultra-high voltage DC transmission in new power systems due to its integration of the large capacity and low loss of LCC and the flexible control characteristics of VSC. It has significant advantages in reducing the risk of commutation failure and improving reactive power management efficiency, and is currently a research hotspot in the field of power electronics.

[0003] The SLCC (Single-Module Converter) is composed of multiple coupled modules, including the LCC valve group, VSC submodule, and SVG branch, with close electrical connections between them. The patent CN202411857249, entitled "An Overvoltage Suppression Method for SLCC Converters," suppresses fault overvoltages by optimizing control strategies, improving system transient stability. However, this technology only focuses on suppressing the single fault of overvoltage and lacks a fault detection and tracing mechanism. It cannot determine whether the overvoltage is caused by LCC commutation failure, SVG reactive power surge, or VSC submodule breakdown. Maintenance personnel still need to manually locate the fault source, which can take several hours. The aforementioned existing technologies do not overcome the limitations of "single-module local detection" and lack adaptability to the multi-module topology of SLCCs, failing to meet the engineering requirements for system-level collaborative detection and accurate tracing. Summary of the Invention

[0004] This invention proposes a fault detection method for SLCC converters based on multi-module collaboration. It analyzes the fault causes of SLCC based on the topology correlation of SLCC multi-modules, thus achieving accurate fault tracing.

[0005] To achieve the above objectives, the present invention proposes the following technical content: The fault detection method for SLCC converters based on multi-module collaboration includes the following steps: S1: Collect electrical parameter data for each module of the SLCC and clean up outliers; electrical parameters include: commutation angle of the LCC valve group. Trigger angle DC current I dc ;VSC submodule capacitor voltage U c Bridge arm current I arm Modulation ratio M Reactive power output of SVG branch Q SVG Branch current I SVGEquivalent output voltage U SVG ; S2: Based on the electrical parameter data obtained in S1, construct the correlation matrix; specifically including the following steps: S2.1: Calculate the coupling parameters based on the electrical parameter data obtained in S1; Calculate the coupling parameters between the same bridge arm of the VSC submodule. K 1. The formula is:

[0006] In the formula, U c ( t +1) indicates the first t The capacitor voltage at time +1; U c ( t ) indicates the first t The capacitor voltage at a given time, set t The time is the initialization time; I arm ( t +1) indicates the first t Bridge arm current at time +1; I arm ( t ) indicates the first t Bridge arm current at time +1; Calculate the unidirectional coupling parameters of the SVG branch to the LCC valve manifold. K 2; The formula is:

[0007] In the formula, Q SVG ( t +1) indicates the first t The reactive power output at +1 moment; Q SVG ( t ) indicates the first t Effortlessly exerting oneself in moments of ineffectiveness; Indicates the first t Commutation angle at time +1; Indicates the first t Phase angle at any given moment; Calculate the coupling parameters of the LCC valve assembly to the VSC submodule. K 3; The formula is:

[0008] In the formula, I dc ( t +1) indicates the firstt DC current at time +1; I dc ( t ) indicates the first t DC current at any given moment; M ( t + 1) indicates the first t Modulation ratio at time +1; M ( t ) indicates the first t Modulation ratio at time; S2.2: Based on the topology, the VSC submodule is subdivided into 3 bridge arm groups; the coupling parameters of the LCC valve group, the 3 bridge arm groups and the SVG branch are used as the elements in the row and column of the correlation matrix to construct the correlation matrix; The formula for the correlation matrix is:

[0009] S3: Dynamically update the elements in the correlation matrix that have coupling relationships to obtain the updated correlation matrix; S4: Construct the SLCC topology graph using the updated correlation matrix as the adjacency matrix; S5: Based on the association relationship of the topology graph, update the node features through iterative formulas to obtain the final target features; S6: Input the final target features into the trained linear classifier, and the linear classifier determines the fault type at node i at time t+1. S7: Repeat S3-S6, and as time progresses, continue to obtain the updated correlation matrix, and continue to obtain the next time step and node. i The final target features are used to determine the next time step and node using a linear classifier. i The type of fault at the location.

[0010] Furthermore, step S3 specifically includes the following steps: S3.1: Calculate the parameter changes of elements in the correlation matrix that have coupling relationships; Set this element as a ij ( t +1), then in this element i and j The formula for the change in parameters is:

[0011] S3.2: Calculate the differential coupling parameters; The formula is:

[0012] In the formula, Indicates the first i The parameter for the first j Differential coupling parameters; S3.3: The real-time parameters are fitted using the least squares method to obtain the... i The parameter, the first t The fitted data at time +1, and the data at time +1. j The parameter, the first t The fitted data at time +1; and the calculation of the... t The fitting coupling parameters at time +1 are denoted as... The formula is:

[0013] In the formula, Indicates the first i The parameter, the first t Fitted data at time +1; x i ( t ) indicates the first i The parameter, the first t Data at any given time; Indicates the first j The parameter, the first t Fitted data at time +1; x j ( t ) indicates the first j The parameter, the first t Data at time step; S3.4: Calculate the updated parameters and obtain the updated correlation matrix; The updated parameter formula is:

[0014] In the formula, This represents the updated elements in the correlation matrix; λ Indicates the set parameter weights; After obtaining the updated parameters, the updated correlation matrix is ​​obtained, denoted as...

[0015] Further, step S4 includes the following steps: S4.1: Construct the topology graph framework; A topological graph is defined as:

[0016] in, V Represents a set of nodes. V ={ v 1. v 2, ... v5}, each node represents a parameter;

[0017] The adjacency matrix is: ; E Indicates the first i The parameter and the first j The relationship between the edges of the parameters; S4.2: Define the feature point matrix, i.e. the real-time state of each node;

[0018] The real-time status of each node includes: commutation angle. Trigger angle DC current I dc capacitor voltage U c Bridge arm current I arm Modulation ratio M Unproductive efforts Q SVG Branch current I SVG Equivalent output voltage U SVG .

[0019] Furthermore, the formula for updating node features in step S5 is:

[0020] In the formula, Indicates the first k Layer weights; Indicates the activation function; Represents a node j No. k The feature vector of the layer; Indicates the first k Neighbor feature weights of the layer; Represents a node i With nodes j The coupling parameters between them; This represents the final target characteristic.

[0021] The beneficial effects that can be achieved by adopting the above technologies are: This invention enables precise tracing of fault causes through multi-module collaboration, solving the problem that existing technologies can only locate a single faulty module and cannot identify cross-module cascading fault causes (e.g., existing technologies cannot determine whether LCC commutation failure is due to its own defect or SVG abnormality). By constructing a multi-module electrical coupling correlation matrix and using the GNN algorithm, this invention can accurately trace the root cause in scenarios such as "SVG branch impedance abnormality causing LCC commutation failure". Attached Figure Description

[0022] Figure 1 This is the flowchart of this solution; Figure 2 This is a graph of the SVG branch impedance over time in the example. Figure 3 This is a graph of the LCC commutation angle over time in the example. Figure 4 It is a graph of direct current over time; Figure 5 It is a graph of VCC bridge arm current and capacitor voltage over time; Figure 6 It is a graph of the GNN response score over time; Figure 7 This is a graph showing the changes in the correlation matrix before and after the fault. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] like Figure 1 As shown, the fault detection method for SLCC converters based on multi-module collaboration includes the following steps: S1: Collect electrical parameter data of each module of SLCC and clean up outliers in the data; Specifically, the SLCC module includes the LCC valve assembly, the VSC submodule, and the SVG branch; The following electrical parameters are collected in real time: The electrical parameters of the LCC valve assembly include: commutation angle. Trigger angle DC current I dc ; The electrical parameters of the VSC submodule include: capacitor voltage. U c Bridge arm currentI arm Modulation ratio M ; The electrical parameters of the SVG branch include: reactive power output. Q SVG Branch current I SVG Equivalent output voltage U SVG .

[0025] The "3σ criterion + wavelet transform" is used to clean outliers in the real-time acquired data.

[0026] S2: Based on the electrical parameter data obtained in S1, construct the correlation matrix. This includes the following steps: S2.1: Calculate the coupling parameters based on the electrical parameter data obtained in S1.

[0027] Coupling parameters include: coupling parameters between the same bridge arm of the VSC submodule. K 1. Coupling parameters of the SVG branch to the LCC valve assembly in one direction K 2. Unidirectional coupling parameters of the LCC valve group to the VSC submodule K 3; In this context, considering that the capacitor element plays a dominant role in the VSC submodule during transient electrical coupling, the capacitor voltage is used. U c With bridge arm current I arm Calculate the coupling parameters between the same bridge arm of the VSC submodule. K 1. Used to reflect the instantaneous charge-discharge coupling within the VSC submodule. The formula is:

[0028] In equation (1), U c ( t +1) indicates the first t The capacitor voltage at time +1; U c ( t ) indicates the first t The capacitor voltage at a given time, set t The time is the initialization time; I arm ( t +1) indicates the first t Bridge arm current at time +1; I arm ( t ) indicates the first t Bridge arm current at time +1.

[0029] Similarly, the unidirectional coupling of the SVG branch to the LCC valve group is mainly based on the coupling between reactive power output and commutation angle. Therefore, reactive power output is used. Q SVG With phase angle Calculate the unidirectional coupling parameters of the SVG branch to the LCC valve manifold. K 2; The formula is:

[0030] In equation (2), Q SVG ( t +1) indicates the first t The reactive power output at +1 moment; Q SVG ( t ) indicates the first t Effortlessly exerting oneself in moments of ineffectiveness; Indicates the first t Commutation angle at time +1; Indicates the first t The phase angle at any given moment.

[0031] Similarly, in the VSC submodule, the change in modulation ratio M directly affects the AC current, which in turn affects the DC current. I dc Therefore, by modulating the ratio M and the DC current I dc Calculate the coupling parameters of the LCC valve assembly to the VSC submodule. K 3; The formula is:

[0032] In equation (3), I dc ( t +1) indicates the first t DC current at time +1; I dc ( t ) indicates the first t DC current at any given moment; M ( t+ 1) indicates the first t Modulation ratio at time +1; M ( t ) indicates the first t Modulation ratio at time.

[0033] S2.2: Based on the topology, the VSC submodule is subdivided into 3 bridge arm groups; the coupling parameters of the LCC valve group, the 3 bridge arm groups and the SVG branch are used as elements in the rows and columns of the correlation matrix to construct the correlation matrix.

[0034] The formula for the correlation matrix is:

[0035] In the matrix, a ij Indicates the first i The parameter for the first j The coupling parameters of each parameter, to a 31 For example, it represents the coupling parameter of the second bridge arm group of the VSC submodule to the LCC valve group; parameters that are not directly coupled to the physical path or have a very weak coupling are represented by a zero value.

[0036] S3: Dynamically update the elements in the correlation matrix that have coupling relationships (i.e., non-zero elements) to obtain an updated correlation matrix, which is then adapted to the real-time SLCC converter data. This includes the following steps: S3.1: Calculate the parameter changes of elements in the correlation matrix that have coupling relationships.

[0037] Set this element as a ij ( t +1), then in this element i The formula for the change in parameters is:

[0038] In the formula, x i ( t ) indicates the first i The parameter, the first t Data at any given time; x i ( t +1) indicates the first i The parameter, the first t Data at time +1; Similarly, in this element j The formula for the change in parameters is:

[0039] In the formula, x j ( t ) indicates the first j The parameter, the first t Data at any given time; x j ( t +1) indicates the first j The parameter, the first t Parameter data at time +1; S3.2: Calculate the differential coupling parameters of the elements in S3.1, denoted as... ; The formula is: ; In the formula, Indicates the first i The parameter for the first j Differential coupling parameters; S3.3: The real-time parameters are fitted using the least squares method to obtain the... i The parameter, the first t The fitted data at time +1, and the data at time +1. j The parameter, the first t The fitted data at time +1; and the calculation of the... t The fitting coupling parameters at time +1 are denoted as... .

[0040] The formula is:

[0041] In equation (4), Indicates the first i The parameter, the first t Fitted data at time +1; x i ( t ) indicates the first i The parameter, the first t Data at any given time; Indicates the first j The parameter, the first t Fitted data at time +1; x j ( t ) indicates the first j The parameter, the first t Data at any given time; S3.4: Calculate the updated parameters and obtain the updated correlation matrix, denoted as... .

[0042] The formula is:

[0043] In the formula, This represents the updated elements in the correlation matrix; λ This indicates the set parameter weights.

[0044] For example: In the correlation matrix a 13 For example, calculate this a 13 Parameter change:

[0045] a13 The differential coupling parameters are:

[0046] The updated parameters are:

[0047] S4: Construct the SLCC topology graph using the updated incidence matrix as the adjacency matrix. This includes the following steps: S4.1: Construct the topology graph framework.

[0048] A topological graph is defined as: G = ( V , E );in, V ={ v 1. v 2, ... v 5}, V This represents a set of nodes, where each node represents a parameter (such as an LCC valve manifold, three VSC arms, or an SVG branch). v 1 indicates the LCC valve assembly; v 2~ v 4 represents bridge arm group 1, bridge arm group 2 and bridge arm group 3 of the VSC submodule respectively; v 5 indicates an SVG branch;

[0049] The adjacency matrix is: ; E Indicates the first i The parameter and the first j The relationship between the edges of the parameters.

[0050] S4.2: Define the feature point matrix, which represents the real-time state of each node.

[0051]

[0052] In the formula, Represents a node N Real-time status; the real-time status of each node includes: commutation angle. Trigger angle DC current I dc capacitor voltage U c Bridge arm current I arm Modulation ratio M Unproductive efforts Q SVG Branch current I SVG Equivalent output voltage USVG .

[0053] S5: Based on the association relationship of the topology graph, update the node features through iterative formulas to obtain the final target features.

[0054] The formula is:

[0055] In the formula, Indicates the first k Layer weights; Indicates the activation function; Represents a node j No. k The feature vector of the layer; Indicates the first k Neighbor feature weights of the layer; Represents a node i With nodes j The coupling parameters between them; This represents the final target characteristic.

[0056] S6: Final target features The input is fed into a trained linear classifier, which then identifies the node. i Place, No. t Fault type at time +1; S7: Repeat S3-S6, and as time progresses, continue to obtain the updated correlation matrix, and continue to obtain the next time step and node. i The final target features are used to determine the next time step and node using a linear classifier. i The type of fault at the location.

[0057] Calculation example: The correlation matrix constructed in a certain instance is as follows:

[0058] above K 1. K 2 and K All three are within the normal range; Using this correlation matrix as the adjacency matrix, construct the SLCC topology graph; like Figure 6 Under normal operating conditions, the scores of each module are evenly distributed, that is... Figure 6 In the process, the scores of each module remained stable; however, when the SVG branch impedance jumped from 5Ω to 10Ω (as shown in the image), the scores remained stable. Figure 2The SVG module score jumped from 0.1 to 0.85, the LCC module score rose from 0.15 to 0.7, the commutation angle changed from 18° to 28°, and the capacitor voltage, DC current, and VCC arm current all decreased (e.g., Figure 4 and Figure 5 ),pass Figure 7 The correlation matrix shows that the algorithm determines that "the abnormal impedance of the SVG branch is the root cause" and the failure of LCC commutation is a secondary fault, with a tracing time of 10ms.

[0059] Based on the above-described preferred embodiments of the present invention, and through the foregoing description, those skilled in the art can make various changes and modifications without departing from the inventive concept. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A fault detection method for SLCC converters based on multi-module collaboration, characterized in that, Includes the following steps: S1: Collect electrical parameter data of each module of SLCC and clean up outliers in the data; Electrical parameters include: commutation angle of the LCC valve assembly. Trigger angle DC current I dc ;VSC submodule capacitor voltage U c Bridge arm current I arm Modulation ratio M Reactive power output of SVG branch Q SVG Branch current I SVG Equivalent output voltage U SVG ; S2: Based on the electrical parameter data obtained in S1, construct the correlation matrix; specifically including the following steps: S2.1: Calculate the coupling parameters based on the electrical parameter data obtained in S1; Calculate the coupling parameters between the same bridge arm of the VSC submodule. K 1. The formula is: ； In the formula, U c ( t +1) indicates the first t The capacitor voltage at time +1; U c ( t ) indicates the first t The capacitor voltage at a given time, set t The time is the initialization time; I arm ( t +1) indicates the first t Bridge arm current at time +1; I arm ( t ) indicates the first t Bridge arm current at time +1; Calculate the unidirectional coupling parameters of the SVG branch to the LCC valve manifold. K 2; The formula is: ； In the formula, Q SVG ( t +1) indicates the first t The reactive power output at +1 moment; Q SVG ( t ) indicates the first t Effortless exertion at all times; Indicates the first t Commutation angle at time +1; Indicates the first t Phase angle at any given moment; Calculate the coupling parameters of the LCC valve assembly to the VSC submodule. K 3; The formula is: ； In the formula, I dc ( t +1) indicates the first t DC current at time +1; I dc ( t ) indicates the first t DC current at any given moment; M ( t+ 1) indicates the first t Modulation ratio at time +1; M ( t ) indicates the first t Modulation ratio at time; S2.2: Based on the topology, the VSC submodule is subdivided into 3 bridge arm groups; the coupling parameters of the LCC valve group, the 3 bridge arm groups and the SVG branch are used as the elements in the row and column of the correlation matrix to construct the correlation matrix; The formula for the correlation matrix is: ； S3: Dynamically update the elements in the correlation matrix that have coupling relationships to obtain the updated correlation matrix; S4: Construct the SLCC topology graph using the updated correlation matrix as the adjacency matrix; S5: Based on the association relationship of the topology graph, update the node features through iterative formulas to obtain the final target features; S6: Input the final target features into the trained linear classifier, which then determines the node. i Place, No. t Fault type at time +1; S7: Repeat S3-S6, and as time progresses, continue to obtain the updated correlation matrix, and continue to obtain the next time step and node. i The final target features are used to determine the next time step and node using a linear classifier. i The type of fault at the location.

2. The SLCC converter fault detection method based on multi-module collaboration according to claim 1, characterized in that, Step S3 specifically includes the following steps: S3.1: Calculate the parameter changes of elements in the correlation matrix that have coupling relationships; Set this element as a ij ( t +1), then in this element i and j The formula for the change in parameters is: ； S3.2: Calculate the differential coupling parameters; The formula is: ； In the formula, Indicates the first i The parameter for the first j Differential coupling parameters; S3.3: The real-time parameters are fitted using the least squares method to obtain the... i The parameter, the first t The fitted data at time +1, and the data at time +1. j The parameter, the first t The fitted data at time +1; and the calculation of the... t The fitting coupling parameters at time +1 are denoted as... The formula is: ； In the formula, Indicates the first i The parameter, the first t Fitted data at time +1; x i ( t ) indicates the first i The parameter, the first t Data at any given time; Indicates the first j The parameter, the first t Fitted data at time +1; x j ( t ) indicates the first j The parameter, the first t Data at any given time; S3.4: Calculate the updated parameters and obtain the updated correlation matrix; The updated parameter formula is: ； In the formula, This represents the updated elements in the correlation matrix; λ Indicates the set parameter weights; After obtaining the updated parameters, the updated correlation matrix is ​​obtained, denoted as... .

3. The SLCC converter fault detection method based on multi-module collaboration according to claim 2, characterized in that, Step S4 includes the following steps: S4.1: Construct the topology graph framework; A topological graph is defined as: ； in, V Represents a set of nodes. V ={ v 1. v 2, ... v 5}, each node represents a parameter; ； The adjacency matrix is: ; E Indicates the first i The parameter and the first j The relationship between the edges of the parameters; S4.2: Define the feature point matrix, i.e. the real-time state of each node; ； The real-time status of each node includes: commutation angle. Trigger angle DC current I dc capacitor voltage U c Bridge arm current I arm Modulation ratio M Unproductive efforts Q SVG Branch current I SVG Equivalent output voltage U SVG .

4. The SLCC converter fault detection method based on multi-module collaboration according to claim 3, characterized in that, The formula for updating node features in step S5 is: ； In the formula, Indicates the first k Layer weights; Indicates the activation function; Represents a node j No. k The feature vector of the layer; Indicates the first k Neighbor feature weights of the layer; Represents a node i With nodes j The coupling parameters between them; This represents the final target characteristic.

Images