A method and system for inferring the operating state of a high-voltage DC circuit breaker

Abstract

The present invention provides a method and system for inferring the operating status of a high-voltage direct current (HVDC) circuit breaker, comprising: collecting eigenvalues ​​of each characteristic parameter in a feature subset of the HVDC circuit breaker; calculating, based on the eigenvalues, a correlation coefficient matrix and a self-increase / decrease trend matrix of the characteristic parameters of the HVDC circuit breaker; and inferring the operating status of the HVDC circuit breaker using the correlation coefficient matrix and the self-increase / decrease trend matrix. The feature subset of the HVDC circuit breaker is obtained by screening the characteristic parameters in an initial feature quantity set of the HVDC circuit breaker based on the maximum correlation principle, the minimum redundancy principle, a multi-criteria weighted sorting algorithm, and an association algorithm. The present invention provides data support for operators to promptly understand the operating status of the HVDC circuit breaker, detect early defects of the HVDC circuit breaker, and scientifically formulate condition-based maintenance plans. This provides strong technical support for improving the overall operating efficiency and reliability of the HVDC circuit breaker.

Description

Technical Field

[0001] The present invention relates to the field of online monitoring of DC equipment, and in particular to a method and system for deducing the operating status of a high-voltage DC circuit breaker. Background Art

[0002] A high-voltage DC circuit breaker is a comprehensive electrical device that effectively integrates multiple devices with very different electrical characteristics through reasonable connection methods and set operating logic. It has the current-carrying and insulation capabilities of a mechanical switch and the breaking capacity of a solid-state switch.

[0003] During the steady-state operation of a HVDC breaker, analyzing the overall operating state trends, predicting equipment failures, and analyzing the lifecycle status are key tasks in online monitoring. Because the characteristic parameters of power electronic devices degrade slowly, exhibit subtle deviations, and are easily masked by unknown disturbances and noise, existing conventional monitoring methods struggle to characterize the degradation of power electronic devices during steady-state operation. Over time, device degradation can lead to potential failures that threaten the safe operation of the equipment.

[0004] The complex topology design of HVDC circuit breakers, coupled with the use of numerous power electronic components and switchgear, increases the vulnerability of these devices. The complex interrelationships between these components mean that even minor local degradation can propagate and diffuse through the interconnected pathways between subsystems. This can cause the degradation process to mutate within a short period of time, potentially leading to system failure.

[0005] At present, most of the research on HVDC circuit breaker monitoring focuses on the failure mechanism of power electronic devices or modules, fault characteristic parameter analysis, and electronic component fault diagnosis and location. There is a lack of detailed analysis of the gradual change process of the operating status of the HVDC circuit breaker, and a lack of relevant research on identifying or assisting in identifying minor degradation of power electronic devices and switching devices. Summary of the Invention

[0006] To overcome the above-mentioned deficiencies of the prior art, the present invention proposes a method for inferring the operating state of a high-voltage DC circuit breaker, comprising:

[0007] Collecting characteristic values ​​of each characteristic parameter in a characteristic subset of the high-voltage DC circuit breaker;

[0008] Based on the characteristic values, a correlation coefficient matrix and a self-increasing and decreasing trend matrix of characteristic parameters of the high-voltage DC circuit breaker are calculated respectively;

[0009] Deducing the operating state of the high-voltage direct current circuit breaker using the correlation coefficient matrix and the self-increasing and decreasing trend matrix;

[0010] The feature subset of the high-voltage DC circuit breaker is obtained by screening the feature parameters in the initial feature quantity set of the high-voltage DC circuit breaker based on the maximum correlation principle, the minimum redundancy principle, the multi-criteria weighted sorting algorithm and the association algorithm.

[0011] Preferably, the initial feature quantity set consists of multi-dimensional features, and each dimensional feature consists of multiple feature parameters.

[0012] Furthermore, the process of screening the initial feature quantity set of the high-voltage DC circuit breaker based on the maximum correlation principle, the minimum redundancy principle, the multi-criteria weighted sorting algorithm and the association algorithm includes:

[0013] Based on the maximum correlation principle, the minimum redundancy principle and the multi-criteria weighted sorting algorithm, the feature parameters in the initial feature quantity set are screened to obtain a set Q;

[0014] By using an association algorithm, feature parameters whose support and confidence are less than a threshold value in the set Q are eliminated to obtain the feature subset.

[0015] Furthermore, based on the maximum correlation principle, the minimum redundancy principle, and the multi-criteria weighted sorting algorithm, the feature parameters in the initial feature quantity set are screened to obtain a set Q, including:

[0016] Step 1): Let set Q be an empty set;

[0017] Step 2): Based on the maximum correlation principle, a feature parameter with the largest mean weight of the class separability measure is selected from the initial feature set and placed in the set Q;

[0018] Step 3): When the feature parameters in the initial feature set are completely redundant with the feature parameters in the set Q, based on the minimum redundancy principle, the feature parameters in the initial feature set that are redundant with the feature parameters in the set Q are deleted;

[0019] Step 4): Calculate the redundancy between each feature parameter in the initial feature set and the set Q;

[0020] Step 5): Calculate the state representation degree of each feature parameter in the initial feature set using the redundancy between each feature parameter in the initial feature set and the set Q, and select the feature parameter with the highest state representation degree from the initial feature set based on the multi-criteria weighted sorting algorithm and put it into the set Q;

[0021] Step 6: If the feature parameter exists in the initial feature quantity set, return to step 3); otherwise, output the set Q.

[0022] Furthermore, the calculation formula for the mean weight of the category separability measure of the feature parameters in the initial feature quantity set is as follows:

[0023]

[0024] Where, is the mean weight of the category separability measure of the kth feature parameter under the i-th dimension feature in the initial feature quantity set, avg(f ik ) is the mean of the historical sampling values ​​of the kth feature parameter under the i-th dimension feature in the initial feature quantity set, avg i (F i ) is the mean of the historical sampling values ​​of all feature parameters under the i-th dimension feature in the initial feature quantity set, x f (f ik ) is the f-th historical sampling value of the k-th feature parameter under the i-th dimension feature in the initial feature quantity set, is the total number of historical sampling values ​​of the kth feature parameter under the i-th dimension feature in the initial feature quantity set, i∈(1~n), n is the total number of dimensions in the initial feature quantity set, k∈(1~N i ), N i is the total number of feature parameters contained in the i-th dimension feature in the initial feature set.

[0025] Furthermore, the process of identifying redundant feature parameters in the initial feature set and feature parameters in the set Q includes:

[0026] If the feature parameter f in the initial feature set j The information entropy of the set Q, the characteristic parameter g z The information entropy and the characteristic parameters f in the initial feature set j and the characteristic parameter g in the set Q z The joint entropy of the initial feature quantity set is equal, then the feature parameter f j and the characteristic parameter g in the set Q z Completely redundant, otherwise, the feature parameter f in the initial feature set j and the characteristic parameter g in the set Q z There is no redundancy;

[0027] Among them, the characteristic parameter f in the initial characteristic quantity set j The information entropy of the set Q, the characteristic parameter g z The information entropy and the characteristic parameters f in the initial feature set j and the characteristic parameter g in the set Q z The joint entropy is based on the first characteristic parameter f in the initial feature set j and / or characteristic parameters g in set Q z The historical sampling value of g z ∈Q,f j∈U, U is the initial feature set.

[0028] Furthermore, the redundancy between each feature parameter in the initial feature set and the set Q is calculated as follows:

[0029]

[0030] Where, is the feature parameter f in the initial feature set j The redundancy between the set Q, I(f j ,g z ) is the characteristic parameter in the initial feature set and the characteristic parameter g in the set Q z The mutual information between z ∈Q,f j ∈U, U is the initial feature set.

[0031] Among them, the features in the initial feature set participate in the feature parameters g in the set Q z The mutual information I(f j ,g z ) is calculated as follows:

[0032] I(f j ,g z )=H(f j )+H(g z )-H(f j ,g z )

[0033] Where, H(f j ) is the feature parameter f in the initial feature set j The information entropy, H(g z ) is the characteristic parameter g in the set Q z The information entropy, H(f j ,g z ) is the feature parameter f in the initial feature set j and the characteristic parameter g in the set Q z The joint entropy of .

[0034] Furthermore, the calculation formula for the state representation degree of each feature parameter in the initial feature quantity set is as follows:

[0035]

[0036] Where, J(f j ) is the feature parameter f in the initial feature set j The degree of state representation, is the mean weight of the category separability measure of the kth feature parameter under the i-th dimension feature in the initial feature set, is the feature parameter f in the initial feature set j The redundancy between and set Q,

[0037] Preferably, the collecting of characteristic values ​​of characteristic parameters in the characteristic subset of the high-voltage DC circuit breaker includes:

[0038] Collect the characteristic values ​​of each characteristic parameter in the feature subset at the current time and the r-1 time before;

[0039] Here, r is the total number of moments included in the sampling time window.

[0040] Furthermore, the calculation of the correlation coefficient matrix of the characteristic parameters of the high-voltage DC circuit breaker based on the characteristic values ​​includes:

[0041] Based on the characteristic values, a correlation coefficient matrix of characteristic parameters of the high-voltage direct current circuit breaker at the current moment is determined using a correlation coefficient calculation method.

[0042] Furthermore, the calculating of the self-increasing and decreasing trend matrix of the characteristic parameters of the high-voltage DC circuit breaker based on the characteristic values ​​includes:

[0043] Determine the eigenvalue matrices of the high-voltage direct current circuit breaker at the current moment and h-1 moments before that, respectively, based on the eigenvalues;

[0044] Determine the self-increase and decrease trend matrix of the characteristic parameters of the high-voltage direct current circuit breaker at h-1 moments before the current moment using the eigenvalue matrix;

[0045] Based on the self-increasing and decreasing trend matrix of the characteristic parameters of the high-voltage DC circuit breaker at h-1 moments before the current moment, the self-increasing and decreasing trend matrix of the characteristic parameters of the high-voltage DC circuit breaker at the current moment is estimated using the least squares method;

[0046] The self-increasing and decreasing trend matrix of the characteristic parameter of the high-voltage direct current circuit breaker at the w-1th moment before the current moment is obtained by subtracting the eigenvalue matrix of the high-voltage direct current circuit breaker at the w-1th moment before the current moment from the eigenvalue matrix of the high-voltage direct current circuit breaker at the w-1th moment before the current moment;

[0047] The eigenvalue matrices of the high-voltage direct current circuit breaker at the current moment and the r-1 moments before it are all x-row and 1-column matrices, the αth row elements of which are the eigenvalues ​​of the αth characteristic parameters in the characteristic subset at the current moment and the r-1 moments before it, x is the total number of characteristic parameters in the characteristic subset, h is a positive integer less than r, w is a positive integer less than h-1, and h is the total number of moments included in the derivation time window.

[0048] Furthermore, the use of the correlation coefficient matrix and the self-increasing and decreasing trend matrix to deduce the operating state of the high-voltage DC circuit breaker includes:

[0049] The product of the correlation coefficient matrix of the characteristic parameters of the high-voltage DC circuit breaker at the current moment and the self-increasing and decreasing trend matrix and the eigenvalue matrix of the high-voltage DC circuit breaker at the current moment are added as the eigenvalue matrix of the high-voltage DC circuit breaker at a moment after the current moment;

[0050] The eigenvalue matrix of the high-voltage DC circuit breaker at a moment after the current moment represents the operating state of the high-voltage DC circuit breaker at a moment after the current moment.

[0051] Furthermore, the feature dimensions of the initial feature quantity set include, but are not limited to: thermal fault, voltage withstand breakdown fault, electrical circuit fault, transient fault, secondary system fault, water cooling system fault, and energy supply system fault;

[0052] The characteristic parameters under the thermal fault include but are not limited to: IGBT junction temperature, mechanical switch contact temperature, submodule temperature, MOV overheat protection action, overheat alarm and protection action information;

[0053] The characteristic parameters under the electrical circuit fault include but are not limited to: submodule voltage, submodule current, main branch current, transfer branch current, lightning arrester leakage current, main branch abnormal state information, transfer branch abnormal state information and energy consumption branch abnormal state information;

[0054] The characteristic parameters under the transient fault include but are not limited to: system voltage, system current, breaking current, breaking time, dissipated energy, abnormal state of fast mechanical switch and commutation timeout;

[0055] The characteristic parameters under the secondary system fault include but are not limited to: communication abnormality information, protection action information and control abnormality information;

[0056] The characteristic parameters under the water cooling system failure include but are not limited to: water cooling system failure status and water leakage detection device alarm;

[0057] The characteristic parameters of the energy supply system failure include but are not limited to: energy supply system failure information.

[0058] Based on the same inventive concept, the present invention also provides a high-voltage DC circuit breaker operating state derivation system, comprising:

[0059] An acquisition module, configured to acquire characteristic values ​​of characteristic parameters in a characteristic subset of a high-voltage DC circuit breaker;

[0060] A calculation module, configured to calculate, based on the characteristic values, a correlation coefficient matrix and a self-increasing and decreasing trend matrix of characteristic parameters of the high-voltage direct current circuit breaker;

[0061] an inference module, configured to infer the operating state of the high-voltage DC circuit breaker using the correlation coefficient matrix and the self-increasing and decreasing trend matrix;

[0062] The feature subset of the high-voltage DC circuit breaker is obtained by screening the feature parameters in the initial feature quantity set of the high-voltage DC circuit breaker based on the maximum correlation principle, the minimum redundancy principle, the multi-criteria weighted sorting algorithm and the association algorithm.

[0063] Compared with the closest prior art, the present invention has the following beneficial effects:

[0064] The present invention provides a method and system for inferring the operating state of a high-voltage direct current (HVDC) circuit breaker, comprising: collecting characteristic values ​​of characteristic parameters in a characteristic subset of a HVDC circuit breaker; calculating, based on the characteristic values, a correlation coefficient matrix and a self-increasing and decreasing trend matrix of the characteristic parameters of the HVDC circuit breaker; and inferring the operating state of the HVDC circuit breaker using the correlation coefficient matrix and the self-increasing and decreasing trend matrix. The characteristic subset of the HVDC circuit breaker is obtained by screening characteristic parameters from an initial characteristic quantity set of the HVDC circuit breaker based on a maximum correlation principle, a minimum redundancy principle, a multi-criteria weighted sorting algorithm, and an association algorithm. Based on the characteristic parameter subset of the HVDC circuit breaker, the present invention performs online analysis of the correlation and variation trend of the characteristic parameters of the HVDC circuit breaker through statistical and linear regression algorithms, thereby achieving online inference of the operating state of the HVDC circuit breaker. The method provides data support for operators to promptly understand the operating state of the HVDC circuit breaker, detect early defects of the HVDC circuit breaker, and scientifically formulate condition-based maintenance plans, thereby providing strong technical support for improving the overall operating efficiency and reliability of the HVDC circuit breaker.

[0065] The present invention utilizes the similarity and correlation between characteristic parameters of various modules inside the high-voltage DC circuit breaker to screen the characteristic parameters of the existing high-voltage DC circuit breaker. The extracted characteristic parameter subset of the high-voltage DC circuit breaker can more comprehensively reflect various fault conditions of the high-voltage DC circuit breaker.

[0066] The present invention provides a data prediction method for the operation status prediction and fault risk assessment of high-voltage DC circuit breakers. It can also achieve a refined description of the operation characteristic parameters of the high-voltage DC circuit breaker by adjusting the sampling time window and the inference time window, providing data support for the identification of small faults of the high-voltage DC circuit breaker and the assessment of its associated status. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 A flow chart of a method for deducing the operating status of a high-voltage DC circuit breaker provided by the present invention;

[0068] Figure 2 This is a diagram showing the overall idea of ​​the state derivation method in an embodiment of the present invention;

[0069] Figure 3 This is a flow chart of feature subset extraction in an embodiment of the present invention;

[0070] Figure 4 This is a flowchart of state derivation implementation in an embodiment of the present invention;

[0071] Figure 5 This is a structural diagram of a high-voltage DC circuit breaker operating status deduction system provided by the present invention. DETAILED DESCRIPTION

[0072] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0073] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0074] The present invention provides a method for inferring the operating state of a high-voltage DC circuit breaker. Figure 1 As shown, including:

[0075] Step 1: collecting characteristic values ​​of characteristic parameters in a characteristic subset of a high-voltage DC circuit breaker;

[0076] Step 2: Calculate the correlation coefficient matrix and the self-increase and decrease trend matrix of the characteristic parameters of the high-voltage direct current circuit breaker based on the characteristic values;

[0077] Step 3, using the correlation coefficient matrix and the self-increasing and decreasing trend matrix to deduce the operating state of the high-voltage DC circuit breaker;

[0078] The feature subset of the high-voltage DC circuit breaker is obtained by screening the feature parameters in the initial feature quantity set of the high-voltage DC circuit breaker based on the maximum correlation principle, the minimum redundancy principle, the multi-criteria weighted sorting algorithm and the association algorithm.

[0079] In this embodiment, before executing the deduction of the operating state of the high-voltage DC circuit breaker, it is necessary to pre-extract the feature subset of the high-voltage DC circuit breaker. After the feature subset is extracted, Figure 2 The overall idea of ​​the state derivation method shown in the figure implements the operation state derivation of the high-voltage DC circuit breaker proposed by the present invention;

[0080] The pre-extraction of feature subsets of the HVDC circuit breaker can be called the data preparation stage or offline analysis stage, which uses the original accumulated data of the HVDC circuit breaker to screen its feature parameters and extract feature subsets;

[0081] The feature subset extraction flow chart is as follows Figure 3 As shown in the figure, considering that the number of HVDC circuit breakers in operation is small and there is a lack of historical data accumulation samples, the feature parameters in the HVDC circuit breaker feature subset can be selected by referring to the historical operation database records and the HVDC circuit breaker preventive test data.

[0082] The corresponding steps are:

[0083] S1: Perform a preliminary screening of the characteristic parameters of the high-voltage DC circuit breaker and record the initial characteristic parameter set after screening as U;

[0084] The principles for selecting characteristic parameters of high-voltage DC circuit breakers are:

[0085] 1) Characteristic parameters that can be collected online and reflect the operating status and faults of high-voltage DC circuit breakers;

[0086] 2) Statistical data stored in the historical database that can reflect the historical operating rules of the high-voltage DC circuit breaker;

[0087] 3) Preventive test data that can reflect the failure of high-voltage DC circuit breakers recorded in the historical database;

[0088] 4) External data that can affect the development trend of the operating status of the HVDC circuit breaker;

[0089] After screening, the initial feature quantity set U of the high-voltage DC circuit breaker includes n-dimensional features, that is, U = (F1…F i …F n ), the i-th dimension features in the initial feature set U include N i characteristic parameters, namely

[0090] Among them, f ik is the kth feature parameter in the i-th dimension feature in the initial feature quantity set U, N i is the total number of feature parameters contained in the i-th dimension feature in the initial feature quantity set U;

[0091] The dimensions of the features in the initial feature set U include but are not limited to:

[0092] Thermal faults, voltage withstand breakdown faults, electrical circuit faults, transient faults, secondary system faults, water cooling system faults, and energy supply system faults;

[0093] The feature parameters contained in each dimension feature are shown in Table 1:

[0094] Table 1

[0095]

[0096] S2: Based on the maximum correlation principle, minimum redundancy principle and multi-criteria weighted sorting algorithm, the feature parameters in the initial feature quantity set U are further screened to obtain the initial feature subset Q of the high-voltage DC circuit breaker, including:

[0097] S2-1. Initialize the initial feature subset Q to an empty set;

[0098] S2-2. Based on the maximum correlation principle, a feature parameter with the largest mean weight of the class separability metric is selected from the initial feature set U and placed in the initial feature subset Q as the first feature parameter of the initial feature subset Q;

[0099] It is expressed in formula form as follows:

[0100]

[0101] In the formula, g1 is the feature parameter with the largest mean weight of the category separability measure in the initial feature quantity set U, and is also the first feature parameter in the initial feature subset Q. is the mean weight of the category separability measure of the k-th feature parameter in the i-th dimension feature in the initial feature quantity set U. The symbol arg max f(x) represents the value of x when f(x) takes the maximum value;

[0102] in, The calculation formula is as follows:

[0103]

[0104] Where, avg(f ik ) is the average of the historical sampling values ​​of the kth feature parameter in the i-th dimension feature in the initial feature quantity set U, avg i (F i ) is the mean of the historical sampling values ​​of all feature parameters in the i-th dimension feature in the initial feature quantity set U, x f (f ik ) is the f-th historical sampling value of the k-th feature parameter in the i-th dimension feature in the initial feature quantity set U, is the total number of historical sampling values ​​of the kth feature parameter in the i-th dimension feature in the initial feature quantity set U.

[0105] S2-3. Based on the principle of minimum redundancy, for each feature parameter in the initial feature set U, if there is a feature parameter in the initial feature subset Q that satisfies its own information entropy, the information entropy of the feature parameter, and the joint entropy of the two are equal, then the feature parameter in the initial feature set U is removed. Otherwise, the mutual information between the feature parameter and each feature parameter in the initial feature subset Q is calculated, and the maximum value is used as the redundancy between the feature parameter and the initial feature subset Q;

[0106] Assumption: For the feature parameter f in the initial feature set U j , if there is a feature parameter g in the initial feature subset Q z Satisfy H(f j )=H(g z )=H(f j ,g z ), then f j With g z Completely redundant, then the initial feature set U in f j ; Otherwise, calculate f j The mutual information between the feature parameters of the initial feature subset Q and the maximum value is taken as f j The redundancy between the initial feature subset Q is denoted as I max (f j ,g z ),g z ∈Q;

[0107] Among them, H(f j )=-p(f j )lgp(f j ), H(g z )=-p(g z )lgp(g z ), H(f j ,g z )=-p(f j ,g z )lgp(f j ,g z ), I(f j ,g z )=H(f j )+H(g z )-H(f j ,g z );

[0108] Where, H(f j ) is the characteristic parameter f j The information entropy, p(f j ) is the characteristic parameter f j The probability of crossing the limit / occurrence probability, H(gz ) is the characteristic parameter g z The information entropy, p(g z ) is the characteristic parameter g z The probability of crossing the limit / occurrence probability, H(f j ,g z ) is the characteristic parameter f j and characteristic parameter g z The joint entropy of p(f j ,g z ) is the characteristic parameter f j Limit crossing / occurrence and characteristic parameter g z The probability of exceeding the limit / occurrence at the same time; the probability of exceeding the limit is for the analog quantity, and the probability of occurrence is for the state quantity.

[0109] The p(f j ) is based on the collected characteristic parameters f j The characteristic value of p(g z ) is based on the collected characteristic parameters g z The characteristic value of p(f j ,g z ) Based on the collected characteristic parameters f j The eigenvalues ​​and the collected characteristic parameters g z The eigenvalues ​​are determined.

[0110] S2-4. Based on the multi-criteria weighted sorting algorithm, select a feature parameter with the highest degree of state representation from the initial feature set U and put it into the initial feature subset Q, and return to step S2-3 until the initial feature set U does not contain feature parameters;

[0111] It is expressed in formula form as follows:

[0112]

[0113] Where g l It is the feature parameter with the highest state representation degree in the initial feature quantity set U, and is also the feature parameter currently placed in the initial feature subset Q. j ) is the feature parameter f in the initial feature set U j the degree of state representation;

[0114] Wherein, the J(f j ) is:

[0115]

[0116] Where, I max (f j ,g z ) is f jThe redundancy between the initial feature subset Q, is the feature parameter f in the initial feature set U j The mean weight of the class separability measure .

[0117] S3: The correlation of the feature parameters in the initial feature subset Q is verified by the association algorithm, and the feature parameters whose support and confidence are less than the threshold requirements are eliminated to obtain the feature subset S of the high-voltage DC circuit breaker.

[0118] The state derivation process of the present invention can be called the online analysis part, which uses the initial state determined by the feature subset to achieve online state derivation of the high-voltage DC circuit breaker by calculating the correlation matrix and the self-increase and decrease trend.

[0119] Among them, the state derivation implementation flow chart is as follows Figure 4 As shown, the step 1 includes:

[0120] Data preparation, i.e. determining the inference time window h for HVDC breaker state inference and the sampling time window r for each characteristic parameter in the characteristic subset S, and performing online sampling of each characteristic parameter in the characteristic parameter subset S;

[0121] Here, the current time is set to t, and the characteristic values ​​of each characteristic parameter in the feature subset S from time t-r+1 to the current time t are collected online.

[0122] The step 2 comprises:

[0123] Step 2-1: Calculate the correlation coefficient matrix P(t) of each feature parameter in the feature subset S at the current time t, that is, use the eigenvalues ​​of each feature parameter in the feature subset S from the time t-r+1 to the current time t collected online to calculate the autocorrelation coefficient of each feature parameter and the mutual correlation coefficient between the feature parameters, and construct the correlation coefficient matrix P(t) of each feature parameter in the feature subset S at the current time t;

[0124] Among them, the expression of P(t) is as follows:

[0125]

[0126] Where χ is the total number of feature parameters in the feature subset S, and the element p in the αth row and βth column of P(t) is αβ (t) is the correlation coefficient between the αth feature parameter and the βth feature parameter in the feature subset S at the current time t;

[0127] The p αβ The calculation formula of (t) is as follows:

[0128]

[0129] Where Cov(xα (t),x β (t)) is the covariance of the αth feature parameter and the βth feature parameter in the feature subset S at the current time t, Var[X(t)] is the variance of the feature parameter matrix X(t) of the high-voltage DC circuit breaker at the current time t, and X(t) = [x1(t)…x α (t)…x χ (t)] T , x α (t) is the eigenvalue of the αth feature parameter in the feature subset S at the current time t;

[0130] The Cov(x α (t),x β (t)) and Var[X(t)] are calculated based on the eigenvalues ​​of each feature parameter in the feature subset S from time t-r+1 to the current time t collected online, using the covariance calculation formula and variance calculation formula respectively.

[0131] Step 2-2: Based on the characteristic parameter matrix of the HVDC circuit breaker from time t-r+1 to the current time t, use the least squares regression algorithm to estimate the self-increase and decrease trend matrix Δ(t) of the characteristic parameters in the characteristic subset S at the current time t;

[0132] Among them, the estimated value of Δ(t) is Represented by univariate linear regression analysis. Expand Δ(t) forward to Δ(t-1), Δ(t-2), ..., Δ(th), where Δ(t-σ) = X(t-σ+1) - X(t-σ), σ = 1, 2, ..., h;

[0133] The least squares regression algorithm is used to The formula for making the estimate is:

[0134]

[0135] Solve the above formula to get the minimum self-increasing and decreasing trend of the characteristic parameters in the characteristic subset S. It is equal to the sum of the minimum self-increasing and decreasing trend and Δ(t-1).

[0136] The step 3 comprises:

[0137] State deduction, that is, using the estimated value of the self-increasing and decreasing trend matrix Δ(t) of the characteristic parameters in the feature subset S at the current time t The increase / decrease matrix of the characteristic parameters of the high-voltage DC circuit breaker at the current time t is calculated using the correlation coefficient matrix P(t) of each characteristic parameter in the characteristic subset S at the current time t, and the characteristic value of each characteristic parameter in the characteristic subset S at the time t+1 is deduced using the increase / decrease matrix.

[0138] That is: ε(t)=P(t)Δ(t), X(t+1)=X(t)+ε(t), where ε(t) is the increase / decrease matrix of the characteristic parameters of the HVDC circuit breaker at the current time t.

[0139] It should be noted that in the process of HVDC circuit breaker state derivation, the derivation time window h and sampling time window r in S need to be considered. If the derivation time window h and sampling time window r are too large, the amount of calculation will increase, resulting in unnecessary computational cost; if the derivation time window h and sampling time window r are too small, the information granularity will be reduced, resulting in inaccurate self-increase and decrease trend estimation.

[0140] In a specific embodiment of the present invention, the method provided by the present invention utilizes abundant electrical information of the HVDC circuit breaker to predict and extrapolate the operating characteristic parameters of the HVDC circuit breaker within a short time scale, providing auxiliary analysis information for the evaluation of the degradation process of power electronic devices, thereby assisting operation and maintenance personnel in quickly discovering and capturing early defects of the HVDC circuit breaker, and providing a data basis for the online identification, location and diagnosis of minor faults of the HVDC circuit breaker, which is of great significance for ensuring the safe and reliable operation of the HVDC circuit breaker equipment.

[0141] In a specific embodiment of the present invention, the premise for deducing the operating state of the high-voltage DC circuit breaker is:

[0142] 1. The HVDC circuit breaker operates smoothly and is in a non-fault state. All electrical parameters are not subject to sudden changes caused by external impacts, and data collection is continuous.

[0143] 2. The current characteristic parameter value of the high-voltage DC circuit breaker is affected by its past characteristic parameter value and the changes in the surrounding characteristic parameter values, and there is a correlation between the characteristic parameter values.

[0144] The present invention uses data inference to depict the detailed state change process of the high-voltage DC circuit breaker within a short time scale, providing data support for operators to timely understand the operating status of the high-voltage DC circuit breaker, capture early defects of the high-voltage DC circuit breaker, and scientifically formulate condition-based maintenance plans, thereby providing strong technical support for improving the overall operating efficiency and reliability of the high-voltage DC circuit breaker.

[0145] The present invention collects information based on existing high-voltage DC circuit breakers and does not require additional measurement devices. The method itself not only considers the changing trends of various characteristic parameters during the operation of the high-voltage DC circuit breaker on the basis of characterizing the accumulated operating losses of components, but also considers the mutual influence relationship between various characteristic parameters from the perspective of system structure. Therefore, it can more comprehensively realize the deduction of the operating status of the DC circuit breaker.

[0146] Example 2:

[0147] The present invention also provides a high voltage DC circuit breaker operating state deduction system, such as Figure 5 Shown, including:

[0148] An acquisition module, configured to acquire characteristic values ​​of characteristic parameters in a characteristic subset of a high-voltage DC circuit breaker;

[0149] A calculation module, configured to calculate, based on the characteristic values, a correlation coefficient matrix and a self-increasing and decreasing trend matrix of characteristic parameters of the high-voltage direct current circuit breaker;

[0150] an inference module, configured to infer the operating state of the high-voltage DC circuit breaker using the correlation coefficient matrix and the self-increasing and decreasing trend matrix;

[0151] The feature subset of the high-voltage DC circuit breaker is obtained by screening the feature parameters in the initial feature quantity set of the high-voltage DC circuit breaker based on the maximum correlation principle, the minimum redundancy principle, the multi-criteria weighted sorting algorithm and the association algorithm.

[0152] Specifically, the initial feature quantity set is composed of multi-dimensional features, and each dimensional feature is composed of multiple feature parameters.

[0153] Specifically, the system further includes a screening module for screening the initial feature quantity set of the high-voltage DC circuit breaker based on the maximum correlation principle, the minimum redundancy principle, the multi-criteria weighted sorting algorithm, and the association algorithm. The screening module includes:

[0154] A screening unit, configured to screen the feature parameters in the initial feature quantity set to obtain a set Q based on a maximum correlation principle, a minimum redundancy principle, and a multi-criteria weighted sorting algorithm;

[0155] The elimination unit is used to eliminate the feature parameters whose support and confidence are less than the threshold requirements in the set Q by using the association algorithm to obtain the feature subset.

[0156] Furthermore, the screening unit includes:

[0157] Setting submodule, used to set the set Q to be an empty set;

[0158] The first selection subunit is used to select a feature parameter with the largest mean weight of the class separability measure from the initial feature quantity set based on the maximum correlation principle and put it into the set Q;

[0159] A deletion subunit is used to delete the redundant feature parameters in the initial feature set and the feature parameters in the set Q based on the principle of minimum redundancy when the feature parameters in the initial feature set are completely redundant with the feature parameters in the set Q;

[0160] The calculation subunit is used to calculate the redundancy between each feature parameter in the initial feature quantity set and the set Q;

[0161] The second selection subunit is used to calculate the state representation degree of each feature parameter in the initial feature set by using the redundancy between each feature parameter in the initial feature set and the set Q, and select a feature parameter with the highest state representation degree from the initial feature set based on a multi-criteria weighted sorting algorithm and put it into the set Q;

[0162] The output sub-unit is used to return to the deletion sub-unit if the feature parameter exists in the initial feature quantity set; otherwise, the output set Q is output.

[0163] Specifically, the calculation formula for the mean weight of the category separability measure of the feature parameters in the initial feature quantity set is as follows:

[0164]

[0165] Where, is the mean weight of the category separability measure of the kth feature parameter under the i-th dimension feature in the initial feature quantity set, avg(f ik ) is the mean of the historical sampling values ​​of the kth feature parameter under the i-th dimension feature in the initial feature quantity set, avg i (F i ) is the mean of the historical sampling values ​​of all feature parameters under the i-th dimension feature in the initial feature quantity set, x f (f ik ) is the f-th historical sampling value of the k-th feature parameter under the i-th dimension feature in the initial feature quantity set, is the total number of historical sampling values ​​of the kth feature parameter under the i-th dimension feature in the initial feature quantity set, i∈(1~n), n is the total number of dimensions in the initial feature quantity set, k∈(1~N i ), N i is the total number of feature parameters contained in the i-th dimension feature in the initial feature set.

[0166] Specifically, the process of identifying redundancy between the feature parameters in the initial feature set and the feature parameters in the set Q includes:

[0167] If the feature parameter f in the initial feature set j The information entropy of the set Q, the characteristic parameter g z The information entropy and the characteristic parameters f in the initial feature set j and the characteristic parameter g in the set Q z The joint entropy of the initial feature quantity set is equal, then the feature parameter f j and the characteristic parameter g in the set Q z Completely redundant, otherwise, the feature parameter f in the initial feature setj and the characteristic parameter g in the set Q z There is no redundancy;

[0168] Among them, the characteristic parameter f in the initial characteristic quantity set j The information entropy of the set Q, the characteristic parameter g z The information entropy and the characteristic parameters f in the initial feature set j and the characteristic parameter g in the set Q z The joint entropy is based on the first characteristic parameter f in the initial feature set j and / or characteristic parameters g in set Q z The historical sampling value of g z ∈Q,f j ∈U, U is the initial feature set.

[0169] Specifically, the redundancy between each feature parameter in the initial feature set and the set Q is calculated as follows:

[0170]

[0171] Where, is the feature parameter f in the initial feature set j The redundancy between the set Q, I(f j ,g z ) is the characteristic parameter in the initial feature set and the characteristic parameter g in the set Q z The mutual information between z ∈Q,f j ∈U, U is the initial feature set.

[0172] Among them, the features in the initial feature set participate in the feature parameters g in the set Q z The mutual information I(f j ,g z ) is calculated as follows:

[0173] I(f j ,g z )=H(f j )+H(g z )-H(f j ,g z )

[0174] Where, H(f j ) is the feature parameter f in the initial feature set j The information entropy, H(g z ) is the characteristic parameter g in the set Q z The information entropy, H(f j ,g z ) is the feature parameter f in the initial feature setj and the characteristic parameter g in the set Q z The joint entropy of .

[0175] Specifically, the calculation formula for the state representation degree of each feature parameter in the initial feature quantity set is as follows:

[0176]

[0177] Where, J(f j ) is the feature parameter f in the initial feature set j The degree of state representation, is the mean weight of the category separability measure of the kth feature parameter under the i-th dimension feature in the initial feature set, is the feature parameter f in the initial feature set j The redundancy between and set Q,

[0178] Specifically, the acquisition module is used to:

[0179] Collect the characteristic values ​​of each characteristic parameter in the feature subset at the current time and the r-1 time before;

[0180] Here, r is the total number of moments included in the sampling time window.

[0181] Specifically, the computing unit includes:

[0182] The first determining submodule is configured to determine a correlation coefficient matrix of characteristic parameters of the high-voltage direct current circuit breaker at a current moment based on the characteristic value and using a correlation coefficient calculation method.

[0183] Specifically, the computing unit further includes:

[0184] A second determining submodule is configured to determine, based on the eigenvalues, the eigenvalue matrices of the high-voltage DC circuit breaker at the current moment and at h-1 moments before that;

[0185] A third determining submodule is configured to determine, using the eigenvalue matrix, a self-increasing and decreasing trend matrix of characteristic parameters of the high-voltage DC circuit breaker at h-1 moments before the current moment;

[0186] An estimation submodule, for estimating the self-increasing and decreasing trend matrix of the characteristic parameters of the high-voltage DC circuit breaker at the current moment using the least squares method based on the self-increasing and decreasing trend matrix of the characteristic parameters of the high-voltage DC circuit breaker at h-1 moments before the current moment;

[0187] The self-increasing and decreasing trend matrix of the characteristic parameter of the high-voltage direct current circuit breaker at the w-1th moment before the current moment is obtained by subtracting the eigenvalue matrix of the high-voltage direct current circuit breaker at the w-1th moment before the current moment from the eigenvalue matrix of the high-voltage direct current circuit breaker at the w-1th moment before the current moment;

[0188] The eigenvalue matrices of the high-voltage direct current circuit breaker at the current moment and the r-1 moments before it are all x-row and 1-column matrices, the αth row elements of which are the eigenvalues ​​of the αth characteristic parameters in the characteristic subset at the current moment and the r-1 moments before it, x is the total number of characteristic parameters in the characteristic subset, h is a positive integer less than r, w is a positive integer less than h-1, and h is the total number of moments included in the derivation time window.

[0189] Specifically, the derivation unit is used to:

[0190] The product of the correlation coefficient matrix of the characteristic parameters of the high-voltage DC circuit breaker at the current moment and the self-increasing and decreasing trend matrix and the eigenvalue matrix of the high-voltage DC circuit breaker at the current moment are added as the eigenvalue matrix of the high-voltage DC circuit breaker at a moment after the current moment;

[0191] The eigenvalue matrix of the high-voltage DC circuit breaker at a moment after the current moment represents the operating state of the high-voltage DC circuit breaker at a moment after the current moment.

[0192] Specifically, the feature dimensions of the initial feature quantity set include but are not limited to: thermal fault, voltage withstand breakdown fault, electrical circuit fault, transient fault, secondary system fault, water cooling system fault and energy supply system fault;

[0193] The characteristic parameters under the thermal fault include but are not limited to: IGBT junction temperature, mechanical switch contact temperature, submodule temperature, MOV overheat protection action, overheat alarm and protection action information;

[0194] The characteristic parameters under the electrical circuit fault include but are not limited to: submodule voltage, submodule current, main branch current, transfer branch current, lightning arrester leakage current, main branch abnormal state information, transfer branch abnormal state information and energy consumption branch abnormal state information;

[0195] The characteristic parameters under the transient fault include but are not limited to: system voltage, system current, breaking current, breaking time, dissipated energy, abnormal state of fast mechanical switch and commutation timeout;

[0196] The characteristic parameters under the secondary system fault include but are not limited to: communication abnormality information, protection action information and control abnormality information;

[0197] The characteristic parameters under the water cooling system failure include but are not limited to: water cooling system failure status and water leakage detection device alarm;

[0198] The characteristic parameters of the energy supply system failure include but are not limited to: energy supply system failure information.

[0199] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0200] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0201] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0202] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0203] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A method for inferring the operating state of a high-voltage DC circuit breaker, characterized in that: The method comprises: Collecting characteristic values ​​of each characteristic parameter in a characteristic subset of the high-voltage DC circuit breaker; Based on the characteristic values, a correlation coefficient matrix and a self-increasing and decreasing trend matrix of characteristic parameters of the high-voltage DC circuit breaker are calculated respectively; Deducing the operating state of the high-voltage direct current circuit breaker using the correlation coefficient matrix and the self-increasing and decreasing trend matrix; The feature subset of the high-voltage DC circuit breaker is obtained by screening the feature parameters in the initial feature quantity set of the high-voltage DC circuit breaker based on the maximum correlation principle, the minimum redundancy principle, the multi-criteria weighted sorting algorithm and the association algorithm; The collecting of characteristic values ​​of each characteristic parameter in the characteristic subset of the high-voltage DC circuit breaker includes: Collect the characteristic values ​​of each characteristic parameter in the feature subset at the current time and the r-1 time before; Wherein, r is the total number of moments included in the sampling time window; The step of calculating a correlation coefficient matrix of characteristic parameters of the high-voltage DC circuit breaker based on the characteristic values ​​includes: Based on the characteristic values, a correlation coefficient matrix of characteristic parameters of the high-voltage DC circuit breaker at the current moment is determined using a correlation coefficient calculation method; The step of calculating the self-increasing and decreasing trend matrix of the characteristic parameters of the high-voltage DC circuit breaker based on the characteristic values ​​includes: Determine the eigenvalue matrices of the high-voltage direct current circuit breaker at the current moment and h-1 moments before that, respectively, based on the eigenvalues; Determine the self-increase and decrease trend matrix of the characteristic parameters of the high-voltage direct current circuit breaker at h-1 moments before the current moment using the eigenvalue matrix; Based on the self-increase and decrease trend matrix of the characteristic parameters of the high-voltage DC circuit breaker at h-1 moments before the current moment, the least squares method is used to estimate the self-increase and decrease trend matrix of the characteristic parameters of the high-voltage DC circuit breaker at the current moment; where h is a positive integer less than r, and h is the total number of moments included in the inference time window.

2. The method according to claim 1, wherein The initial feature quantity set is composed of multi-dimensional features, and each dimensional feature is composed of multiple feature parameters.

3. The method according to claim 2, wherein The process of screening the initial feature quantity set of the high-voltage DC circuit breaker based on the maximum correlation principle, the minimum redundancy principle, the multi-criteria weighted sorting algorithm and the association algorithm includes: Based on the maximum correlation principle, the minimum redundancy principle and the multi-criteria weighted sorting algorithm, the feature parameters in the initial feature quantity set are screened to obtain a set Q; By using an association algorithm, feature parameters whose support and confidence are less than a threshold value in the set Q are eliminated to obtain the feature subset.

4. The method according to claim 3, wherein The characteristic parameters in the initial characteristic quantity set are screened based on the maximum correlation principle, the minimum redundancy principle, and the multi-criteria weighted sorting algorithm to obtain a set Q, including: Step 1): Let set Q be an empty set; Step 2): Based on the maximum correlation principle, a feature parameter with the largest mean weight of the class separability measure is selected from the initial feature set and put into the set Q; Step 3): When the feature parameters in the initial feature set are completely redundant with the feature parameters in set Q, based on the principle of minimum redundancy, the feature parameters in the initial feature set that are redundant with the feature parameters in set Q are deleted; Step 4): Calculate the redundancy between each feature parameter in the initial feature set and the set Q; Step 5): Calculate the state representation degree of each feature parameter in the initial feature set using the redundancy between each feature parameter in the initial feature set and the set Q, and select the feature parameter with the highest state representation degree from the initial feature set based on the multi-criteria weighted sorting algorithm and put it into the set Q; Step 6: If the feature parameter exists in the initial feature set, return to step 3); otherwise, output the set Q.

5. The method according to claim 4, wherein The calculation formula of the mean weight of the category separability measure of the feature parameters in the initial feature quantity set is as follows: Where, is the first feature in the initial feature set The mean weight of the category separability measure of the k-th feature parameter under the dimensional feature, is the first feature in the initial feature set The mean of the historical sampling values ​​of the k-th feature parameter under the dimensional feature, is the first feature in the initial feature set The mean of the historical sampling values ​​of all feature parameters under the dimensional feature, is the first feature in the initial feature set The kth feature parameter under the k-dimensional feature Historical sampling values, , is the first feature in the initial feature set The total number of historical sampling values ​​of the k-th feature parameter under the dimensional feature, , is the total number of dimensions in the initial feature set, , is the first feature in the initial feature set The total number of feature parameters contained in the dimension feature.

6. The method according to claim 4, wherein The process of identifying redundancy between the feature parameters in the initial feature set and the feature parameters in the set Q includes: If the feature parameters in the initial feature set The information entropy and characteristic parameters in set Q The information entropy and the characteristic parameters in the initial feature set and the characteristic parameters in set Q The joint entropy of are equal, then the characteristic parameters in the initial feature set and the characteristic parameters in set Q Completely redundant, otherwise, the feature parameters in the initial feature set and the characteristic parameters in set Q There is no redundancy; Among them, the characteristic parameters in the initial characteristic quantity set The information entropy of the set Q and the characteristic parameters The information entropy and the characteristic parameters in the initial feature set and the characteristic parameters in set Q The joint entropy is based on the characteristic parameters in the initial feature set and / or characteristic parameters in set Q The historical sampling value is determined by , , is the initial feature set.

7. The method according to claim 4, wherein The redundancy between each feature parameter in the initial feature set and the set Q is calculated as follows: Where, is the feature parameter in the initial feature set The redundancy between and set Q, The characteristic parameters in the initial feature set and the characteristic parameters in the set Q The information between , , is the initial feature set; Among them, the features in the initial feature set participate in the feature parameters in the set Q Mutual information between The calculation formula is as follows: Where, is the feature parameter in the initial feature set The information entropy of is the characteristic parameter in set Q The information entropy of is the feature parameter in the initial feature set and the characteristic parameters in set Q The joint entropy of .

8. The method according to claim 4, wherein The calculation formula for the state representation degree of each feature parameter in the initial feature quantity set is as follows: Where, is the feature parameter in the initial feature set The degree of state representation, is the first feature in the initial feature set The mean weight of the category separability measure of the k-th feature parameter under the dimensional feature, is the feature parameter in the initial feature set The redundancy between Q and the set Q.

9. The method according to claim 1, wherein The self-increasing and decreasing trend matrix of the characteristic parameter of the high-voltage direct current circuit breaker at the w-1th moment before the current moment is obtained by subtracting the eigenvalue matrix of the high-voltage direct current circuit breaker at the w-th moment before the current moment from the eigenvalue matrix of the high-voltage direct current circuit breaker at the w-1th moment before the current moment; The eigenvalue matrices of the high-voltage DC circuit breaker at the current moment and the r-1 moments before are A matrix with 1 row and 1 column, The row element is the first The characteristic values ​​of the characteristic parameters at the current moment and the r-1 moments before, is the total number of feature parameters in the feature subset, h is a positive integer less than r, w is a positive integer less than h-1, and h is the total number of moments included in the inference time window.

10. The method according to claim 9, characterized in that The method of deducing the operating state of the high-voltage DC circuit breaker by using the correlation coefficient matrix and the self-increasing and decreasing trend matrix includes: The product of the correlation coefficient matrix of the characteristic parameters of the high-voltage DC circuit breaker at the current moment and the self-increasing and decreasing trend matrix and the eigenvalue matrix of the high-voltage DC circuit breaker at the current moment are added as the eigenvalue matrix of the high-voltage DC circuit breaker at a moment after the current moment; The eigenvalue matrix of the high-voltage DC circuit breaker at a moment after the current moment represents the operating state of the high-voltage DC circuit breaker at a moment after the current moment.

11. The method according to claim 2, characterized in that The feature dimensions of the initial feature quantity set include but are not limited to: thermal fault, voltage withstand breakdown fault, electrical circuit fault, transient fault, secondary system fault, water cooling system fault and energy supply system fault; The characteristic parameters under thermal fault include but are not limited to: IGBT junction temperature, mechanical switch contact temperature, submodule temperature, MOV overheat protection action, overheat alarm and protection action information; The characteristic parameters under the electrical circuit fault include but are not limited to: submodule voltage, submodule current, main branch current, transfer branch current, lightning arrester leakage current, main branch abnormal state information, transfer branch abnormal state information and energy consumption branch abnormal state information; The characteristic parameters under the transient fault include but are not limited to: system voltage, system current, breaking current, breaking time, dissipated energy, abnormal state of fast mechanical switch and commutation timeout; The characteristic parameters under the secondary system fault include but are not limited to: communication abnormality information, protection action information and control abnormality information; The characteristic parameters under the water cooling system failure include but are not limited to: water cooling system failure status and water leakage detection device alarm; The characteristic parameters of the energy supply system failure include but are not limited to: energy supply system failure information.

12. A high-voltage DC circuit breaker operating state deduction system, used to implement the method according to claim 1, characterized in that: The system comprises: An acquisition module, configured to acquire characteristic values ​​of characteristic parameters in a characteristic subset of a high-voltage DC circuit breaker; A calculation module, configured to calculate, based on the characteristic values, a correlation coefficient matrix and a self-increasing and decreasing trend matrix of characteristic parameters of the high-voltage direct current circuit breaker; an inference module, configured to infer the operating state of the high-voltage DC circuit breaker using the correlation coefficient matrix and the self-increasing and decreasing trend matrix; The feature subset of the high-voltage DC circuit breaker is obtained by screening the feature parameters in the initial feature quantity set of the high-voltage DC circuit breaker based on the maximum correlation principle, the minimum redundancy principle, the multi-criteria weighted sorting algorithm and the association algorithm.

Images