Converter station fault identification method and system fusing SER event and fault recording data

Abstract

The invention relates to the technical field of power system monitoring, and particularly discloses a converter station fault identification method and system fusing SER events and fault recording data, and the method comprises the steps: obtaining the SER event data of a converter station, and identifying an SER alarm event and event data; acquiring corresponding fault recording data based on the SER alarm event information, and identifying fault recording data features; based on the SER event and the fault recording data, a Mann-Kendall method is adopted to detect whether the fault is a false alarm; and on the basis of the SER event and the fault recording data, the identification information of the fault is determined by adopting a Binary Segmentation method. Compared with the prior art, the converter station fault can be monitored and identified in real time, and false alarm can be avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of power system monitoring, and in particular to a converter station fault identification method integrating SER event and fault recording data. Background Art

[0002] With the continuous development of smart grids and the accelerated integration of physical information systems in power systems, power data is showing an explosive growth trend. Converter stations have infrared temperature measurement systems, integrated monitoring systems, robot inspection systems, etc., which are characterized by a large number of equipment, excessive dispersion of equipment, and a large amount of equipment data. The increasing complexity of converter stations and the degree of intelligence of equipment have put forward higher requirements for the intelligent operation and maintenance level of converter stations. The manual comparison and analysis method has the defects of poor timeliness and inability to accurately predict equipment failures. The operation and maintenance personnel of the converter station cannot promptly clarify the operating status of the equipment in the converter station.

[0003] The converter valve is a key device for UHV DC transmission, responsible for key functions such as rectification and inversion. If problems such as commutation failure occur, the voltage on the DC side may drop and the current may increase. In severe cases, it may even cause power transmission interruption, threatening the safe operation of the power grid. At present, the DC transmission converter valve in my country mainly adopts thyristor technology. The thyristor converter valve is mainly composed of thyristor components, valve cooling components and valve arrester components. Any failure or abnormality in any part may affect the normal operation of the converter valve. At present, the converter station usually adopts traditional planned maintenance, and performs annual maintenance by arranging a power outage for the entire station. However, planned maintenance requires a long power outage time, which may have an adverse impact on the economic efficiency of the DC transmission system. Summary of the invention

[0004] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and provide a converter station fault identification method integrating SER events and fault recording data.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] As a first aspect of the present invention, a converter station fault identification method integrating SER events and fault recording data is provided, the steps comprising:

[0007] Obtain SER event data of converter stations, identify SER alarm events and event data;

[0008] Based on the SER alarm event information, the corresponding fault recording data is obtained and the characteristics of the fault recording data are identified;

[0009] Based on SER events and fault recording data, the Mann-Kendall method is used to check whether the fault is a false alarm;

[0010] Based on SER events and fault recording data, the Binary Segmentation method is used to determine the fault identification information.

[0011] As a preferred technical solution, the identifying SER event data includes: collecting SER data generated after a fault occurs and the location where the fault occurs.

[0012] As a preferred technical solution, the identification of fault recording data features includes: waveforms and data before, during and after the fault occurs, and the identification includes text information, recording graph scale, channel annotations, time scale and recording waveform.

[0013] As a preferred technical solution, the Mann-Kendall method is used to check whether the fault is a false alarm, as follows:

[0014] Read the file corresponding to the fault recording according to the SER alarm event sequence;

[0015] For the fault recording time series X={x1,…,x n}, construct the statistic UF k :

[0016]

[0017] In the formula, s k is the order column; E(s k ), They are s k The expected value and variance of

[0018] Reverse the fault recording time series X to obtain a new time series {x n ,x (n-1) …,x1}, repeat the above steps for the new time series to obtain a new order column and the corresponding statistic UB k :

[0019]

[0020] Judgment curve UB k , UF k Whether the part beyond the horizontal line contains a mutation point, if the curve UB k , UF k If there is no mutation point, the fault is a false alarm. k , UF k If there is a mutation point, the fault is a normal alarm.

[0021] As a preferred technical solution, the Binary Segmentation method is used to determine the fault identification information, as follows:

[0022] The fault recording sequence is divided into two subsequences X1, X2, ..., X m and X m+1 , X m+2 , …, X n Check whether there is a mutation point and establish a hypothesis test for the fault recording sequence:

[0023] H0: β1=β2=…=β n

[0024] H1: β1=…=β m ≠β m+1 =…=β n and / or

[0025] In the formula, H0 represents the null hypothesis, H1 represents the alternative hypothesis; β n is the fault recording sequence X n The mean of is the fault recording sequence X n The variance of

[0026] If the null hypothesis H0 is not rejected, stop; if the null hypothesis H0 is rejected, find a mutation point m;

[0027] Repeat the above steps until no more mutation points are detected in the subsequences;

[0028] Confirm the fault information in combination with the fault recording diagram, including the fault start time and fault end time;

[0029] Match the information in the SER alarm event and summarize it into the final fault information.

[0030] As a second aspect of the present invention, a converter station fault identification system integrating SER events and fault recording data is provided, the system comprising:

[0031] SER fault event identification module, which obtains SER event data of the converter station and identifies SER alarm events and event data;

[0032] The fault recording data identification module obtains the corresponding fault recording data based on the SER alarm event information and identifies the characteristics of the fault recording data;

[0033] The false alarm judgment module uses the Mann-Kendall method to check whether the fault is a false alarm based on the SER event and fault recording data;

[0034] The fault information identification module uses the Binary Segmentation method to determine the fault identification information based on the SER event and fault recording data.

[0035] As a preferred technical solution, the SER event data identified by the SER fault event identification module includes: collecting SER data generated after the fault occurs and the location where the fault occurs.

[0036] As an optimal technical solution, the fault recording data identification module collects and organizes waveforms and data before, during and after the fault occurs, and identifies text information, recording scale, channel annotations, time scale and recording waveform.

[0037] As a preferred technical solution, the false alarm judgment module uses the Mann-Kendall method to check whether the fault is a false alarm, as follows:

[0038] Read the file corresponding to the fault recording according to the SER alarm event sequence;

[0039] For the fault recording time series X={x1,…,x n}, construct the statistic UF k :

[0040]

[0041]

[0042] In the formula, s k is the order column; E(s k ), They are s k The expected value and variance of

[0043] Reverse the fault recording time series X to obtain a new time series {x n ,x (n-k) …,x1}, repeat the above steps for the new time series to obtain a new order column and the corresponding statistic UB k :

[0044]

[0045] Judgment curve UB k , UF k Whether the part beyond the horizontal line contains a mutation point, if the curve UB k , UF k If there is no mutation point, the fault is a false alarm. k , UF k If there is a mutation point, the fault is a normal alarm.

[0046] As a preferred technical solution, the fault information identification module uses a Binary Segmentation method to determine the fault identification information, as follows:

[0047] The fault recording sequence is divided into two subsequences X1, X2, ..., X m and X m+1 , X m+2 , …, X n Check whether there is a mutation point and establish a hypothesis test for the fault recording sequence:

[0048] H0: β1=β2=…=β n

[0049] H1: β1=…=β m ≠β m+1 =…=β n and / or

[0050] In the formula, H0 represents the null hypothesis, H1 represents the alternative hypothesis; β n is the fault recording sequence X n The mean of is the fault recording sequence C n The variance of

[0051] If the null hypothesis H0 is not rejected, stop; if the null hypothesis H0 is rejected, find a mutation point m;

[0052] Repeat the above steps until no more mutation points are detected in the subsequences;

[0053] Confirm the fault information in combination with the fault recording diagram, including the fault start time and fault end time;

[0054] Match the information in the SER alarm event and summarize it into the final fault information.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] The present invention proposes a converter station fault identification method that integrates SER event and fault recording data. First, based on the fusion event sequence recording system SER identification, fault event data and fault recording data are obtained; based on the Mann-Kendall method, the obtained data are checked to see whether the fault is a false alarm; then the fault identification information is determined by the Binary Segmentation method. Based on the above method, the present invention can monitor and identify converter station faults in real time and avoid false alarms. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 A flow chart of a converter station fault identification method integrating SER events and fault recording data according to the present invention;

[0058] Figure 2It is a fault recording diagram of pole 1 in one specific embodiment of the present invention;

[0059] Figure 3 It is a fault recording diagram of pole 2 in one specific embodiment of the present invention;

[0060] Figure 4 FIG. 4 is a Mann-Kendall curve diagram of polar neutral current in one specific embodiment of the present invention. DETAILED DESCRIPTION

[0061] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0062] Example 1

[0063] The object of the present invention is to provide a converter station fault identification method integrating Sequence Events Recorder (SER) events and fault recording data, such as Figure 1 As shown, the specific steps of the program are as follows:

[0064] Step S1, obtaining SER event data and identifying SER alarm event data;

[0065] Step S2: acquiring corresponding fault recording data based on the SER alarm event information, and identifying the characteristics of the fault recording data;

[0066] Step S3, integrating SER events, fault recording data and Mann-Kendall method to check whether the fault is a false alarm;

[0067] Step S4: Determine fault identification information by integrating SER events, fault recording data and Binary Segmentation method.

[0068] In step S1, SER event data is collected and sorted, and SER event data is identified, as follows:

[0069] In terms of time, SER data generated after a fault occurs is collected;

[0070] The host locates the fault location;

[0071] In step S2, the fault recording data is collected and sorted, and the characteristics of the fault recording data are identified, as follows;

[0072] Collect and organize the waveforms and data before, during and after the fault occurs, and identify the fault recording data features including text information, recording scale, channel annotations, time scale, recording waveform, etc.

[0073] In step S3, the SER event, fault recording data and Mann-Kendall method are combined to check whether the fault is a false alarm, as follows:

[0074] According to the SER alarm event sequence, use Python to read the corresponding files of the fault recording (*.CFG, *.DAT);

[0075] Check whether the fault is a false alarm based on the Mann-Kendall method, including:

[0076] For the time series of fault recording X={x1,…,x n}, construct the statistic UF k :

[0077]

[0078] In the formula, s k is the order column; E(s k ), Sequential columns s k The expected value and variance of .

[0079] Reverse the fault recording time series X to obtain a new time series {x n ,x (n-1) …,x1}, repeat the above steps for the new time series to obtain a new order column and the corresponding statistic UB k :

[0080]

[0081] Curve UB k , UF k If the curve UB exceeds the horizontal line, it means that the time series X has a significant change trend. It is necessary to determine whether the exceeding part contains a mutation point. k , UF k If there is no mutation point, it means that the fault is a false alarm. k , UF k If there is a mutation point, it means that the fault is a normal alarm.

[0082] In step S4, the SER event, the fault recording data and the binary segmentation method are integrated to determine the fault identification information, which is as follows:

[0083] The fault mutation point is checked based on the Binary Segmentation test method, including:

[0084] Fault recording sequence X1, X2, ..., X nIndependent, subject to means β1, β2, …, β n , the variance is Normal distribution, establish hypothesis test:

[0085] H0: β1=β2=…=β n

[0086] H1: β1=…=β m ≠β m+1 =…=β n and / or

[0087] In the formula, H0 represents the null hypothesis and H1 represents the alternative hypothesis.

[0088] If the null hypothesis H0 is not rejected, stop; if the null hypothesis H0 is rejected, find a mutation point m;

[0089] The present invention divides the sequence into two subsequences X1, X2, ..., X m and X m+1 , X m+2 , …, X n Check whether there are mutation points;

[0090] Repeat the above steps until no more mutation points are detected in the subsequences;

[0091] Based on the Binary Segmentation test method, the fault mutation point is obtained and combined with the fault recording diagram to confirm the fault information, including the fault start time and fault end time;

[0092] Match the information in the SER alarm event and summarize it into the final fault information.

[0093] Example 2

[0094] As one of the specific implementation examples of the present invention, this embodiment performs fault identification on the SER event data and the fault recording data collected from a converter station in Shanghai. The specific process is as follows:

[0095] Collect and organize SER event data, identify SER alarm events and event data;

[0096] Select the corresponding fault recording file based on the SER alarm event information, such as Figure 2 , Figure 3 The fault recording diagrams of pole 1 and pole 2 of the converter station are shown respectively, and the fault recording data features are identified for the fault recording file;

[0097] The SER event, fault recording data and Mann-Kendall method are used to test whether the fault is a false alarm. The test results are as follows: Figure 4 As shown, curve UB k , UF k The part beyond the horizontal line contains a mutation point, the test result is that there is a mutation point, and the alarm is a valid alarm.

[0098] The fault mutation point is tested by combining SER events, fault recording data and Binary Segmentation test method. The detection result of pole 1 is [4947, 5086, 6635, 7132, 8402]; the detection result of pole 2 is [4764, 5763, 8270, 9639]. Therefore, the fault identification results are: pole 1 fault start time: 0.4947s, fault end time: 0.8402s; pole 2 fault start time: 0.4764s, fault end time: 0.9639s.

[0099] Example 3

[0100] As another implementation manner of the present invention, this embodiment further provides a converter station fault identification system integrating SER events and fault recording data. The system adopts the converter station fault identification method described in the above embodiment, including:

[0101] SER fault event identification module, which obtains SER event data of the converter station, identifies SER alarm events and event data, and the SER event data includes: collecting SER data generated after the fault occurs and the location of the fault;

[0102] The fault recording data recognition module obtains the corresponding fault recording data based on the SER alarm event information and identifies the characteristics of the fault recording data; including collecting and collating the waveforms and data before, during and after the fault occurs, and identifying text information, recording scale, channel annotations, time scale and recording waveform;

[0103] The false alarm judgment module uses the Mann-Kendall method to check whether the fault is a false alarm based on the SER event and fault recording data, as follows:

[0104] Read the file corresponding to the fault recording according to the SER alarm event sequence;

[0105] For the fault recording time series X={x1,…,x n}, construct the statistic UF k :

[0106]

[0107]

[0108] In the formula, s k is the order column; E(s k), They are s k The expected value and variance of

[0109] Reverse the fault recording time series X to obtain a new time series {x n ,x (n-k) …,x1}, repeat the above steps for the new time series to obtain a new order column and the corresponding statistic UB k :

[0110]

[0111] Judgment curve UB k , UF k Whether the part beyond the horizontal line contains a mutation point, if the curve UB k , UF k If there is no mutation point, the fault is a false alarm. k , UF k If there is a mutation point, the fault is a normal alarm.

[0112] The fault information identification module uses the Binary Segmentation method to determine the fault identification information based on the SER event and fault recording data, as follows:

[0113] The fault recording sequence is divided into two subsequences X1, X2, ..., X m and X m+1 , X m+2 , …, X n Check whether there is a mutation point and establish a hypothesis test for the fault recording sequence:

[0114] H0: β1=β2=…=β n

[0115] H1: β1=…=β m ≠β m+1 =…=β n and / or

[0116] In the formula, H0 represents the null hypothesis, H1 represents the alternative hypothesis; β n is the fault recording sequence X n The mean of is the fault recording sequence C n The variance of

[0117] If the null hypothesis H0 is not rejected, stop; if the null hypothesis H0 is rejected, find a mutation point m;

[0118] Repeat the above steps until no more mutation points are detected in the subsequences;

[0119] Confirm the fault information in combination with the fault recording diagram, including the fault start time and fault end time;

[0120] Match the information in the SER alarm event and summarize it into the final fault information.

[0121] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A converter station fault identification method integrating SER events and fault recording data, characterized in that the steps include: Obtain SER event data of converter stations, identify SER alarm events and event data; Based on the SER alarm event information, the corresponding fault recording data is obtained and the characteristics of the fault recording data are identified; Based on SER events and fault recording data, the Mann-Kendall method is used to check whether the fault is a false alarm; Based on SER events and fault recording data, the Binary Segmentation method is used to determine the fault identification information.

2. A converter station fault identification method integrating SER event and fault recording data according to claim 1, characterized in that: The identifying SER event data includes: collecting SER data generated after a fault occurs and a location where the fault occurs.

3. A converter station fault identification method integrating SER event and fault recording data according to claim 1, characterized in that: The identification of fault recording data features includes: waveforms and data before, during and after the fault occurs, and the identification includes text information, recording diagram scale, channel annotations, time scale and recording waveform.

4. A converter station fault identification method integrating SER event and fault recording data according to claim 1, characterized in that: The Mann-Kendall method is used to test whether the fault is a false alarm, as follows: Read the file corresponding to the fault recording according to the SER alarm event sequence; For the fault recording time series X={x1,…,x n }, construct the statistic UF k : In the formula, s k is the order column; E(s k ), They are s k The expected value and variance of Reverse the fault recording time series X to obtain a new time series {x n ,x (n-1) …,x1}, repeat the above steps for the new time series to obtain a new order column and the corresponding statistic UB k : Judgment curve UB k , UF k Whether the part beyond the horizontal line contains a mutation point, if the curve UB k , UF k If there is no mutation point, the fault is a false alarm. k , UF k If there is a mutation point, the fault is a normal alarm.

5. The converter station fault identification method integrating SER event and fault recording data according to claim 1, characterized in that: The Binary Segmentation method is used to determine the fault identification information, as follows: The fault recording sequence is divided into two subsequences X1, X2, ..., X m and X m+1 , X m+2 , …, X n Check whether there is a mutation point and establish a hypothesis test for the fault recording sequence: H0：β1＝β2＝…＝β n H1: β1 = … = β m ≠ β m+1 = … = β n and / or In the formula, H0 represents the null hypothesis, H1 represents the alternative hypothesis; β n is the fault recording sequence X n The mean of is the fault recording sequence X n The variance of If the null hypothesis H0 is not rejected, stop; if the null hypothesis H0 is rejected, find a mutation point m; Repeat the above steps until no more mutation points are detected in the subsequences; Confirm the fault information in combination with the fault recording diagram, including the fault start time and fault end time; Match the information in the SER alarm event and summarize it into the final fault information.

6. A converter station fault identification system integrating SER events and fault recording data, characterized in that: The system comprises: SER fault event identification module, which obtains SER event data of the converter station and identifies SER alarm events and event data; The fault recording data identification module obtains the corresponding fault recording data based on the SER alarm event information and identifies the characteristics of the fault recording data; The false alarm judgment module uses the Mann-Kendall method to check whether the fault is a false alarm based on the SER event and fault recording data; The fault information identification module uses the Binary Segmentation method to determine the fault identification information based on the SER event and fault recording data.

7. A converter station fault identification system integrating SER events and fault recording data according to claim 6, characterized in that: The SER event data identified by the SER fault event identification module includes: SER data generated after the fault occurs and the location where the fault occurs.

8. A converter station fault identification system integrating SER events and fault recording data according to claim 6, characterized in that: The fault recording data identification module collects and organizes waveforms and data before, during and after a fault occurs, and identifies text information, recording diagram scale, channel annotations, time scales and recording waveforms.

9. A converter station fault identification system integrating SER events and fault recording data according to claim 6, characterized in that: The false alarm judgment module uses the Mann-Kendall method to check whether the fault is a false alarm, as follows: Read the file corresponding to the fault recording according to the SER alarm event sequence; For the fault recording time series X={x1,…,x n }, construct the statistic UF k : In the formula, s k is the order column; E(s k ), They are s k The expected value and variance of Reverse the fault recording time series X to obtain a new time series {x n ,x (n-1) …,x1}, repeat the above steps for the new time series to obtain a new order column and the corresponding statistic UB k : Judgment curve UB k , UF k Whether the part beyond the horizontal line contains a mutation point, if the curve UB k , UF k If there is no mutation point, the fault is a false alarm. k , UF k If there is a mutation point, the fault is a normal alarm.

10. A converter station fault identification system integrating SER events and fault recording data according to claim 6, characterized in that: The fault information identification module uses the Binary Segmentation method to determine the fault identification information, as follows: The fault recording sequence is divided into two subsequences X1, X2, ..., X m and X m+1 , X m+2 , …, X n Check whether there is a mutation point and establish a hypothesis test for the fault recording sequence: H0：β1＝β2＝…＝β n H1: β1 = … = β m ≠ β m+1 = … = β n and / or In the formula, H0 represents the null hypothesis, H1 represents the alternative hypothesis; β n is the fault recording sequence X n The mean of is the fault recording sequence X n The variance of If the null hypothesis H0 is not rejected, stop; if the null hypothesis H0 is rejected, find a mutation point m; Repeat the above steps until no more mutation points are detected in the subsequences; Confirm the fault information in combination with the fault recording diagram, including the fault start time and fault end time; Match the information in the SER alarm event and summarize it into the final fault information.

Images