A 10kv circuit breaker opening and closing fault early warning method of a ring main unit system

Abstract

The application relates to the technical field of fault early warning, in particular to a 10KV circuit breaker opening and closing fault early warning method of a ring network cabinet system, which comprises the following steps: collecting voltage signals to generate sequence data, extracting a main frequency to identify an abnormal frequency band, constructing a feature vector and performing opening and closing fault early warning. In the application, voltage signals in a circuit breaker operation cycle are collected to generate current waveform time sequence data, the whole operation process is recorded, frequency domain conversion and main frequency distribution extraction are combined, a feature structure reflecting operation state changes can be constructed, frequency change and amplitude difference between cycles are utilized to form a high-dimensional feature vector, the expression ability of abnormal modes is enhanced, on the basis, the dense distribution of abnormal frequency bands is judged to realize dynamic identification and risk early warning of potential fault trends, the timeliness and accuracy of fault identification are significantly improved, and the dependence on artificial patrol and fixed devices is reduced.

Description

Technical Field

[0001] This invention relates to the field of fault early warning technology, and in particular to a method for early warning of opening and closing faults in a 10KV circuit breaker of a ring main unit system. Background Technology

[0002] The field of fault early warning technology mainly involves real-time monitoring, anomaly analysis, and risk prediction of the operating status of power equipment. This includes equipment operating parameter acquisition, abnormal state identification, fault trend modeling, and the establishment of early warning mechanisms. Typically, this is achieved by constructing a multi-source data acquisition system and combining it with algorithmic models to dynamically analyze the collected information, enabling early identification of potential faults and the formulation of response strategies. Traditional circuit breaker opening and closing fault early warning methods refer to the methods of pre-identifying and alerting users to abnormal states of circuit breakers during opening or closing operations in power systems. These methods primarily target technical issues that may occur during circuit breaker operation, such as mechanical jamming, electromagnetic drive abnormalities, control signal interruptions, or abnormal position feedback. Traditional methods typically rely on manual periodic inspections, installing limit switches to detect travel positions, using current transformers to detect operating currents, and using time relays to monitor operation duration for preliminary identification.

[0003] Current circuit breaker opening and closing fault early warning mainly relies on regular manual inspections and traditional electrical testing devices to judge the circuit breaker status. This has significant limitations in terms of identification accuracy and response speed. Relying on components such as limit switches, current transformers, and time relays for indirect monitoring makes it difficult to capture dynamic changes during operation in a timely manner. In particular, when latent faults such as mechanical jamming or electromagnetic drive abnormalities have not yet been fully exposed, it is difficult to achieve effective early warning. In addition, existing technologies lack multi-dimensional parameter correlation analysis of equipment operating status and cannot identify potential anomalies caused by changes in frequency domain characteristics. This results in single early warning signals, a high risk of missed detection, and a lagging response strategy, affecting the overall safety of circuit breaker operation and the stability of the power system. Summary of the Invention

[0004] To address the technical problems existing in the prior art, this invention provides a method for early warning of opening and closing faults in a 10kV circuit breaker of a ring main unit system, comprising the following steps:

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a method for early warning of opening and closing faults of a 10KV circuit breaker in a ring main unit system, comprising the following steps:

[0006] S1: Collect the voltage signals of the opening coil, closing coil and energy storage motor in the 10KV circuit breaker of the ring main unit during the operation cycle, combine the operation start timestamp and operation category label to construct a data structure and generate a current waveform time series dataset.

[0007] S2: Based on the current waveform time series dataset, perform frequency domain transformation to obtain frequency and amplitude mapping groups, define fixed frequency intervals and determine the main frequency points, arrange them into frequency position vectors according to the segment order, and construct the main frequency distribution mapping structure table.

[0008] S3: Based on the main frequency points recorded in the main frequency distribution mapping structure table, match the frequency band main frequency point vector in the previous cycle, calculate the frequency offset difference segment by segment to identify abnormal frequency bands and establish a frequency band index, and output the abnormal frequency band index set.

[0009] S4: Read the frequency band index position recorded in the abnormal frequency band index set, extract the difference of the main frequency position and the amplitude change rate of the corresponding frequency band in the current cycle and the previous cycle, combine the current cycle operation category code to construct a multi-field matrix, unify the field order and length to perform standardization and normalization processing, and form a circuit breaker operation feature vector structure.

[0010] S5: Based on the circuit breaker operation feature vector structure, perform abnormal frequency band density judgment operation, confirm the spectrum offset clustering phenomenon within the same operation cycle, issue early warning information, and output opening and closing fault early warning record.

[0011] As a further aspect of the present invention, the main frequency point specifically refers to the frequency position corresponding to the maximum amplitude point.

[0012] As a further aspect of the present invention, the abnormal frequency band specifically refers to the frequency band where the frequency offset difference exceeds a set spectral change discrimination threshold.

[0013] As a further aspect of the present invention, the spectrum offset clustering phenomenon specifically refers to counting the number of abnormal frequency bands in the circuit breaker operation feature vector structure. If the number of consecutive abnormal frequency bands is greater than a set clustering length threshold, then the existence of spectrum offset clustering phenomenon is confirmed.

[0014] As a further embodiment of the present invention, the current waveform time series dataset includes the current characteristics of the tripping coil, the current characteristics of the closing coil, the current characteristics of the energy storage motor, the operation start time label, and the operation category label. The main frequency distribution mapping structure table includes the frequency interval number, the main frequency point of each interval, and the frequency position vector. The abnormal frequency band index set includes the frequency offset abnormal segment identifier, the frequency offset difference record, and the over-threshold judgment mark. The circuit breaker operation feature vector structure includes the main frequency difference vector, the amplitude change rate vector, and the operation category code. The tripping and closing fault early warning record includes the circuit breaker code, the abnormal frequency band clustering situation, and the operation timestamp.

[0015] As a further aspect of the present invention, the step of obtaining the current waveform time series dataset is as follows:

[0016] S111: Collect the voltage signals of the opening coil, closing coil and energy storage motor in the 10KV circuit breaker of the ring main unit during the operation cycle, and uniformly align the collected voltage signals according to the sampling time axis. Then, perform truncation processing based on the operation start timestamp, retain the signal sequence within the operation cycle range, and generate a set of voltage signals within the operation cycle.

[0017] S112: Based on the voltage signal set within the operation cycle, the voltage signals of each channel are synchronously sampled in terms of voltage value and time dimension by the analog-to-digital converter, discrete voltage sample points are extracted, a sequence of sample points with a uniform time interval is constructed, and they are combined according to the sensor channel order to obtain a discrete multi-channel voltage sequence.

[0018] S113: Based on the discretized multi-channel voltage sequence, extract the operation start timestamp and operation category label, perform structural calibration and classification coding on each sequence, inject the identifier according to the operation type and complete the unified timing organization to establish a current waveform time series dataset.

[0019] As a further aspect of the present invention, the step of obtaining the main frequency distribution mapping structure table is as follows:

[0020] S211: Based on the current waveform time series dataset, perform fast Fourier transform on the voltage sequence of each channel, perform complex domain transformation operation at a fixed sampling frequency, extract the real and imaginary parts in the corresponding frequency domain, calculate the amplitude of each frequency point, and generate a frequency amplitude corresponding array.

[0021] S212: Based on the frequency amplitude corresponding array, the frequency axis interval is defined and divided into equally wide segments. The frequency value corresponding to the largest amplitude is retrieved in each segment. If the frequency value is greater than the set amplitude judgment benchmark, it is marked as a candidate point of the main frequency. The amplitude of all candidate points is compared, and the frequency point with the largest amplitude in each segment is located to obtain the interval main frequency set.

[0022] S213: Based on the set of dominant frequencies in the interval, number and arrange each frequency point according to the frequency interval order, construct a dominant frequency vector structure, bind the value of each frequency point with the index of the interval to generate an ordered mapping pair, and encapsulate it into a two-dimensional table structure to establish a dominant frequency distribution mapping structure table.

[0023] As a further aspect of the present invention, the step of obtaining the abnormal frequency band index set is as follows:

[0024] S311: Based on the main frequency points recorded in the main frequency distribution mapping structure table, extract the main frequency values ​​of each frequency band in the current operation cycle, read the frequency position vector in the previous cycle, perform one-to-one index comparison of each corresponding frequency band, construct a two-cycle main frequency comparison matrix under the same frequency band number, perform vector level difference calculation operation, calculate the frequency offset value sequence of the current cycle relative to the previous cycle, and generate the frequency band main frequency offset vector.

[0025] S312: Based on the frequency band main frequency offset vector, if the frequency band frequency offset value is greater than the set spectrum change discrimination threshold, the corresponding frequency band is marked as a spectrum fluctuation abnormal segment, and the corresponding segment number is recorded to identify the abnormal and normal frequency band distribution status, thereby obtaining a spectrum abnormality identification array.

[0026] S313: Based on the spectrum anomaly identifier array, retrieve the index number of the frequency band with an abnormal status, aggregate all the segment numbers that meet the condition that the spectrum offset is greater than the set spectrum change discrimination threshold into a set structure, and perform deduplication and ascending sorting on the set structure to output the abnormal frequency band index set.

[0027] As a further aspect of the present invention, the step of obtaining the circuit breaker operation feature vector structure is as follows:

[0028] S411: Read all the frequency band numbers recorded in the abnormal frequency band index set, index the main frequency position value and amplitude value corresponding to the frequency band in the current period and the previous period item by item, construct a two-field vector structure, calculate the difference of the main frequency position value and obtain the frequency change value, and at the same time perform a proportional calculation operation on the amplitude field to obtain the rate of change vector, and generate the main frequency band change parameter set;

[0029] S412: Based on the main frequency band change parameter set, aggregate the operation category code of the current cycle, and construct a three-field parameter matrix by combining the main frequency position difference, amplitude change rate and operation category number in the order of fields. Perform interpolation operation on the missing fields of each frequency band and mark the missing sections to obtain the circuit breaker operation parameter matrix.

[0030] S413: Based on the circuit breaker operation parameter matrix, unify the field length and arrangement order, standardize the main frequency position difference and amplitude change rate respectively, and keep the operation category field as a discrete value unchanged. After completing all field transformations, store and output in the form of a structure to establish the circuit breaker operation feature vector structure.

[0031] As a further aspect of the present invention, the step of obtaining the opening and closing fault early warning record is as follows:

[0032] S511: Based on the circuit breaker operation feature vector structure, extract feature records of all frequency bands within the operation cycle, arrange them in order according to the frequency band number, and use the normalized main frequency position difference field as the basis to determine whether each frequency band constitutes an abnormal state, count the number of consecutively appearing abnormal markers in continuous frequency bands, and generate a spectrum abnormal continuous segment count value.

[0033] S512: Based on the spectrum abnormal continuous segment count value, if the number of abnormal segments in any group of continuous frequency bands is greater than the set aggregation length threshold, it is confirmed that there is a spectrum offset aggregation phenomenon within the operation cycle. At the same time, the circuit breaker number and the current operation cycle timestamp information are extracted to generate a spectrum aggregation status record item.

[0034] S513: Based on the spectrum aggregation status record item, summarize the records marked as having aggregation status, aggregate the circuit breaker number, operation timestamp and aggregation identifier fields to construct the record structure, add warning tags, and establish a circuit breaker opening and closing fault warning record.

[0035] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0036] In this invention, voltage signals are collected during the circuit breaker's operating cycle to generate current waveform time series data, enabling full recording of the operation process. By combining frequency domain conversion and main frequency distribution extraction, a feature structure reflecting changes in operating status can be constructed, thereby accurately locating spectral offset anomalies. High-dimensional feature vectors are formed using frequency changes and amplitude differences during the cycle, enhancing the ability to express abnormal patterns. Based on this, the dynamic identification and risk warning of potential fault trends are achieved by judging the dense distribution of abnormal frequency bands, significantly improving the timeliness and accuracy of fault identification, while reducing reliance on manual inspections and fixed devices. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is a schematic diagram of the steps of the present invention;

[0039] Figure 2 This is a flowchart illustrating the process of acquiring the current waveform time series dataset for this invention.

[0040] Figure 3 This is a flowchart of the process for obtaining the main frequency distribution mapping structure table in this invention;

[0041] Figure 4 This is a flowchart illustrating the process of obtaining the abnormal frequency band index set in this invention.

[0042] Figure 5 This is a flowchart of the process for obtaining the circuit breaker operation feature vector structure of the present invention;

[0043] Figure 6 This is a flowchart of the process for obtaining early warning records of circuit breaker opening and closing faults in this invention. Detailed Implementation

[0044] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0045] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0046] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0047] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0048] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0049] Please see Figure 1 This invention provides a method for early warning of opening and closing faults in a 10kV circuit breaker of a ring main unit system, comprising the following steps:

[0050] S1: The voltage signals of the opening coil, closing coil and energy storage motor in the 10KV circuit breaker of the ring main unit are collected by sensors (opening coil current sensor, closing coil current sensor and energy storage motor current sensor) during the operation cycle. The signal is discretized by analog-to-digital conversion. The data structure is constructed by combining the operation start timestamp and operation category label to generate a current waveform time series dataset.

[0051] S2: Based on the current waveform time series dataset, combine FFT calculation to perform frequency domain transformation to obtain frequency and amplitude mapping groups, define fixed frequency intervals and determine the main frequency point (locate the frequency position corresponding to the maximum amplitude point), arrange the main frequency points of each frequency interval into frequency position vectors according to the segment order, and construct the main frequency distribution mapping structure table.

[0052] S3: Based on the main frequency points recorded in the main frequency distribution mapping structure table, match the main frequency point vector of the frequency band in the previous period, calculate the frequency offset difference segment by segment and compare it with the set spectrum change discrimination threshold, identify abnormal frequency bands whose frequency offset difference exceeds the set spectrum change discrimination threshold, establish a frequency band index, and output the abnormal frequency band index set.

[0053] S4: Read the frequency band index position recorded in the abnormal frequency band index set, extract the difference of the main frequency position and the amplitude change rate of the corresponding frequency band in the current cycle and the previous cycle, combine it with the operation category code of the current cycle to construct a multi-field matrix, unify the field order and length to perform standardization and normalization processing, and form a circuit breaker operation feature vector structure.

[0054] S5: Based on the circuit breaker operation feature vector structure, perform abnormal frequency band density judgment operation, confirm the spectrum offset clustering phenomenon within the same operation cycle (count the number of abnormal frequency bands in the circuit breaker operation feature vector structure; if the number of consecutive abnormal frequency bands is greater than the set clustering length threshold, then the spectrum offset clustering phenomenon is confirmed), register the corresponding circuit breaker code and timestamp, issue early warning information, and output the opening and closing fault early warning record.

[0055] The current waveform time series dataset includes the current characteristics of the tripping coil, the current characteristics of the closing coil, the current characteristics of the energy storage motor, the operation start time label, and the operation category label. The main frequency distribution mapping structure table includes the frequency interval number, the main frequency point of each interval, and the frequency position vector. The abnormal frequency band index set includes the frequency offset abnormal segment identifier, the frequency offset difference record, and the over-threshold judgment mark. The circuit breaker operation feature vector structure includes the main frequency difference vector, the amplitude change rate vector, and the operation category code. The tripping and closing fault early warning record includes the circuit breaker code, the abnormal frequency band clustering situation, and the operation timestamp.

[0056] Please see Figure 2 The specific steps of S1 are as follows:

[0057] S111: Acquire the voltage signals output by the trip coil current sensor, the closing coil current sensor, and the energy storage motor current sensor during the operation cycle, and uniformly align the acquired voltage signals according to the sampling time axis, perform truncation processing based on the operation start timestamp, retain the signal sequence within the operation cycle range, and generate a voltage signal set within the operation cycle.

[0058] When acquiring the voltage signals output by the trip coil current sensor, closing coil current sensor, and energy storage motor current sensor during the operating cycle, it is necessary to clarify the corresponding access point locations and monitoring sequence of the three types of sensors. For example, the trip coil current sensor is connected to the trip output terminal of the control circuit to monitor its voltage response during the circuit breaker operation. The closing coil current sensor and the energy storage motor current sensor are connected to the closed drive output and the energy storage circuit output terminal, respectively. The sampling frequency is set to 5kHz, and the sampling duration covers the entire operation process. The operation control command trigger time is recorded by the external control system to obtain the operation start timestamp. This timestamp is recorded in Unix timestamp format, such as 1702280425.123, which represents December 11, 2023, 18:20:25.123. Subsequently, this timestamp is used as a reference. The key point is to extract a 1.5-second voltage signal segment from each channel in the self-sampled data, uniformly setting the signal length to 7500 sampling points. Simultaneously, it detects whether there are signal interference segments within this time range for each channel. If a sudden voltage drop exceeding 20V or a rise exceeding 30V occurs, it is considered an interference segment and is truncated to remove discontinuous signal parts. Then, linear interpolation is used to fill in the missing segments, ensuring a consistent time-domain structure for each voltage sequence. After processing, the voltage signal sets within the effective operating segments corresponding to the three types of sensors are obtained. For example, in a single circuit breaker tripping operation, the signals from all three channels can construct a data sequence of length 7500, which is then numbered and stored as CH1_seq001, CH2_seq001, and CH3_seq001, ultimately generating the voltage signal set within the operating cycle.

[0059] S112: Based on the voltage signal set within the operation cycle, the voltage signals of each channel are synchronously sampled in terms of voltage value and time dimension by an analog-to-digital converter, discrete voltage sample points are extracted, a sequence of sample points with a uniform time interval is constructed, and they are combined according to the sensor channel order to obtain a discrete multi-channel voltage sequence.

[0060] Based on the voltage signal set within the aforementioned operating cycle, an analog-to-digital converter (ADC) is invoked to quantize the analog signal sequence. The conversion resolution is set to a 12-bit ADC, dividing the voltage range from -100V to +100V into... There are discrete levels, each with a voltage resolution of [value missing]. V. All channel signals are processed using fixed-point quantization. Then, the sampling time interval is set to 0.2ms, and time-stamping is applied to the 7500-point sequence, i.e., a time index is added to each point. ms, where Here, i represents the starting timestamp of the operation, and i is the sample point index number. Then, channels CH1, CH2, and CH3 are each constructed as column vectors, and their columns are concatenated sequentially to form a unified channel matrix structure. The matrix form is as follows: The data is then saved as OpSeqMatrix_001. During this process, the sampling stability needs to be checked. The mean square error is used to determine whether the squared difference of 10 consecutive points exceeds the preset fluctuation benchmark value, which is set to 2.0V. If 5 out of 10 segments exceed this value, the current channel data is excluded as an abnormal signal. After filtering, the stable sequence is retained as the input source data, and finally the discretized multi-channel voltage sequence is obtained.

[0061] S113: Based on the discretized multi-channel voltage sequence, extract the operation start timestamp and operation category label, perform structural calibration and classification coding on each sequence, inject the label according to the operation type and complete the unified time sequence organization to establish a current waveform time series dataset;

[0062] Based on the discretized multi-channel voltage sequence, the operation start timestamp and operation category label are constructed as independent parameter fields. The operation start timestamp adopts Unix format, and the operation category label adopts binary identification, such as 0 for opening, 1 for closing, and 2 for energy storage operation. These two types of information are called and structurally bound to the voltage sequence to form a triplet format. ,in Represents a multi-channel sequence matrix. Represents a timestamp. This represents the category label. Taking the current operation example, it is set as follows: , , Then the data frame is constructed as follows Next, the data frame set is structured uniformly, arranged in ascending order of timestamps, and all data is categorized by time series to form a continuous structure of operation cycles. A unified numbering system is created using index numbers, such as SeqID_001, SeqID_002, etc. Finally, the structured data set is saved in a unified format with fields: Sequence ID, Multi-channel Sequence, Timestamp, and Operation Type, as shown in the table below.

[0063] Table 1. Operation Sequence Data Frame Structure

[0064]

[0065] As shown in Table 1, once the data frame structure is established, the current waveform time series dataset can be obtained.

[0066] Please see Figure 3 The specific steps of S2 are as follows:

[0067] S211: Based on the current waveform time series dataset, perform fast Fourier transform on the voltage sequence of each channel, perform complex domain transformation operation at a fixed sampling frequency, extract the real and imaginary parts in the corresponding frequency domain, calculate the amplitude of each frequency point, and generate a frequency amplitude corresponding array.

[0068] Based on a current waveform time series dataset, data from each channel of a three-channel current signal is extracted as an input vector. The sample length per channel is set to 7500 sampling points, with a sampling interval of 0.2 milliseconds. A discrete signal of equal duration with a sampling frequency of 5000 Hz is obtained. In frequency domain processing, the channel signal is acquired and a complex sequence is constructed. A Fast Fourier Transform (FFT) is performed on the sequence, transforming each time-domain signal vector into a complex frequency-domain vector. The real and imaginary parts of this complex vector are squared, summed, and then the square root is taken to obtain the amplitude at each frequency point. This process is called amplitude extraction. Specifically, for the point at index k, assuming its real part is 3 and its imaginary part is 4, the amplitude is... This process is repeated for all frequency points to obtain a complete spectral amplitude sequence. Then, based on the symmetry of the Fourier transform, the resulting sequence is pruned into its first half, retaining 3750 valid frequency points. The frequency value of each frequency point is determined according to... Calculation, where hertz, In the example, the frequency of point 100 is Hertz, by associating all frequency indices with amplitudes, creates a one-to-one mapping between frequency and amplitude, recorded as a frequency-amplitude array, as shown in the table below:

[0069] Table 2 Frequency Amplitude Mapping Table

[0070]

[0071] As shown in Table 2, a correspondence has been established between frequency points and amplitude points, and a frequency-amplitude correspondence array is finally generated.

[0072] S212: Based on the frequency amplitude corresponding array, the frequency axis interval is defined and divided into equally wide segments. The frequency value corresponding to the largest amplitude value is retrieved in each segment. If the frequency value is greater than the set amplitude judgment benchmark, it is marked as the main frequency candidate point. The amplitude of all candidate points is compared to locate the frequency point with the largest amplitude in each segment, and the interval main frequency set is obtained.

[0073] Based on the frequency amplitude corresponding array, the frequency analysis range is set to 0 to 2500 Hz. This range is divided into 10 equal-length segments, each 250 Hz wide. The first segment is 0 to 250 Hz, the second is 250 to 500 Hz, and so on up to the tenth segment. For each segment, the amplitude of all frequency points is extracted and averaged to obtain the amplitude baseline value for the current segment. This value is then used to set the amplitude judgment threshold, which is set as 1.5 times the average amplitude of the segment. If the amplitude of a frequency point exceeds this threshold, that frequency point is marked as a candidate for the main frequency. Taking the first segment as an example, if the average amplitude within 250 Hz is 2.8, then the judgment threshold is... If the amplitude of a frequency point is greater than 4.2, it is considered a candidate dominant frequency point. Then, the amplitudes of the candidate dominant frequency points in each segment are compared, and the one with the largest amplitude is extracted as the dominant frequency point of that segment. Its frequency position is the dominant frequency of that segment. After processing 10 segments in sequence, the following results can be obtained:

[0074] Table 3. Set of dominant frequencies in intervals

[0075]

[0076] Table 3 lists the locations of the dominant frequency points in each frequency range, ultimately yielding the set of dominant frequencies for each range.

[0077] S213: Based on the interval dominant frequency set, number and arrange each frequency point according to the frequency interval order, construct the dominant frequency vector structure, bind the value of each frequency point with the interval index to generate an ordered mapping pair, and encapsulate it into a two-dimensional table structure to establish a dominant frequency distribution mapping structure table.

[0078] Based on the dominant frequency set of the intervals, each frequency point is extracted and combined with its corresponding interval number to generate a frequency position vector. This vector is then arranged in ascending order according to the interval number to form a frequency vector structure. Subsequently, combined with the amplitude of the dominant frequency point in each segment and the frequency judgment threshold, a two-dimensional structure table is constructed. The fields are set as interval number, dominant frequency value, amplitude threshold, and dominant frequency amplitude value. This structure is used to display the intervality and dominant frequency prominence of the frequency distribution. The table is indexed and numbered, and a unified number prefix FMap_ is set to facilitate subsequent identification and management. For example, the dominant frequency of the first interval is 66.67 Hz, and the number is set as FMap_01. Its dominant frequency amplitude is 5.31, and the amplitude judgment threshold is 4.2. The structure fields are as follows: [FMap_01, 66.67, 4.2, 5.31]. The data of all ten frequency intervals are filled into the structure table accordingly to establish a dominant frequency mapping information set with a sequence structure. Finally, the dominant frequency distribution mapping structure table is established.

[0079] Please see Figure 4 The specific steps of S3 are as follows:

[0080] S311: Based on the main frequency points recorded in the main frequency distribution mapping structure table, extract the main frequency values ​​of each frequency band in the current operation cycle, read the frequency position vector in the previous cycle, perform one-to-one indexing and comparison of each corresponding frequency band, construct a two-cycle main frequency comparison matrix under the same frequency band number, perform vector level difference calculation operation, calculate the frequency offset value sequence of the current cycle relative to the previous cycle, and generate the frequency band main frequency offset vector.

[0081] Based on the main frequency points recorded in the main frequency distribution mapping structure table, firstly, extract all segment numbers and corresponding main frequency values ​​from the frequency distribution results in the current operating cycle to construct the frequency vector for the current cycle. Then, access the frequency distribution results stored in the previous period and extract the frequency vectors in the same format. The requirement is that the two vectors must be completely identical in frequency band index number, and sorted in ascending order to unify the structure. Then, the numerical difference calculation for the one-to-one correspondence of frequency bands is performed, using... The frequency offset value for each segment is obtained in a specific way. Taking the sample data as an example, if the main frequency of the third segment in the current cycle is 133.33Hz and the previous cycle was 138.25Hz, then the frequency offset value is... Hz, organize all frequency band offset values ​​into a vector structure, and retain the corresponding frequency band number to form a frequency band offset mapping, as shown in Table 4:

[0082] Table 4: Frequency Band Dominant Frequency Offset Values

[0083]

[0084] As shown in Table 4, the offset values ​​have been calculated and recorded segment by segment, and finally the frequency band main frequency offset vector is generated.

[0085] S312: Based on the frequency band main frequency offset vector, if the frequency band frequency offset value is greater than the set spectrum change discrimination threshold, the corresponding frequency band is marked as a spectrum fluctuation abnormal segment, and the corresponding segment number is recorded to identify the abnormal and normal frequency band distribution status, thus obtaining a spectrum abnormality identification array.

[0086] Based on the frequency band main frequency offset vector, the offset value of each frequency band is read. A spectral variation discrimination threshold of 30Hz is set. That is, when the main frequency offset value of a frequency band is greater than 30Hz, it is considered that the frequency band has abnormal fluctuation behavior. Each offset value is compared with the threshold. If the offset value is greater than the threshold, the corresponding frequency band number is added to the tag list. Simultaneously, a status flag array is constructed, assigning each segment a value of 0 or 1. 0 indicates that the spectral offset is within the normal range, and 1 indicates that the offset exceeds the discrimination threshold and constitutes an anomaly. Taking Table 5 as an example, the offset value of segment 3 is 51.87Hz, significantly exceeding the 30Hz threshold. This frequency band is judged as abnormal, and its status flag is set to 1. The remaining segments remain at 0. The generated flag array is as follows:

[0087]

[0088] By creating index key-value pairs between each frequency band number and its status value, it can be represented as follows:

[0089]

[0090] This array is used to represent the offset status of each frequency band and to assist in subsequent index filtering processing, ultimately resulting in a spectrum anomaly identifier array.

[0091] S313: Based on the spectrum anomaly identifier array, retrieve the index number of the frequency band with an abnormal status, aggregate all the segment numbers that meet the condition that the spectrum offset is greater than the set spectrum change discrimination threshold into a set structure, and perform deduplication and ascending sorting on the set structure to output the abnormal frequency band index set.

[0092] Based on the spectrum anomaly identifier array, frequency band index numbers with a status value of 1 are retrieved and aggregated into an independent index set structure to identify all segments identified as having spectrum shift anomalies. The extracted index numbers undergo deduplication and are arranged in ascending order to ensure structural consistency. If multiple anomalous frequency bands exist, a set structure is formed as follows: If only the third segment of the current sample has an abnormal offset, then the output set structure will be displayed. The data is then output as an array, with a set name and anomaly band index set added, as shown in the table below:

[0093] Table 5. Collection of Abnormal Frequency Band Indexes

[0094]

[0095] As shown in Table 5, the index extraction of the spectrum shift anomaly segments has been completed, and the set of anomaly frequency band indexes has been output.

[0096] Please see Figure 5 The specific steps of S4 are as follows:

[0097] S411: Read all frequency band numbers recorded in the abnormal frequency band index set, index the main frequency position value and amplitude value corresponding to the frequency band in the current period and the previous period item by item, construct a two-field vector structure, calculate the difference of the main frequency position value and obtain the frequency change value, and at the same time perform a proportional calculation operation on the amplitude field to obtain the rate of change vector, and generate the main frequency band change parameter set;

[0098] After reading the frequency band index number information recorded in the abnormal frequency band index set, the main frequency value and corresponding amplitude value corresponding to the corresponding frequency band number are first extracted from the frequency distribution structure table of the current operation cycle and the previous operation cycle. Two frequency comparison vectors and amplitude comparison vectors are established respectively, and a difference matrix and a ratio matrix are constructed based on these. In the specific execution process, taking the main frequency of the 4th frequency band in the current cycle as 1250 Hz and the previous cycle as 1235 Hz as an example, the difference in main frequency can be obtained as follows: Hertz, with amplitudes of 4.2 and 3.5 respectively, the rate of change of amplitude is calculated as follows: That is, the rate of change is 20%. These differences and ratios are uniformly recorded in a two-dimensional structure matrix using the frequency band number as the index. Each record includes the field name, frequency band number, frequency difference, and amplitude ratio value, forming the following structure:

[0099] Table 6. Parameters for Main Frequency Variation

[0100]

[0101] As shown in Table 6, the frequency difference and amplitude change rate have been accurately extracted and classified into the structural data, and finally the set of parameters for the main frequency band change is generated.

[0102] S412: Based on the set of parameters for changes in the main frequency band, aggregate the operation category code for the current cycle, and construct a three-field parameter matrix by combining the main frequency position difference, amplitude change rate and operation category number in the order of the fields. Perform interpolation operation on the missing fields of each frequency band and mark the missing sections to obtain the circuit breaker operation parameter matrix.

[0103] After obtaining the set of parameters for the main frequency band changes, the frequency difference field and amplitude change rate field are extracted in order. An additional operation category code field for the current cycle is added. This operation category code is taken from the cycle label in the operation control record. Opening, closing, and energy storage operations are marked as 0, 1, and 2, respectively. If the current cycle is a closing operation, the operation category field is assigned a value of 1. A three-column matrix is ​​constructed based on the above three field values, with the frequency band number as the row index, forming a two-dimensional matrix structure. If the difference field of a certain frequency band is empty due to a missing signal from the previous cycle, the field is filled with the default value -1, and this frequency band is marked as a missing record. In the actual example, the ratio of the 7th frequency band cannot be solved because the amplitude is 0, so its amplitude change rate field is assigned a value of -1. The matrix structure after filling is as follows:

[0104] Table 7 Operation Parameter Matrix

[0105]

[0106] As shown in Table 7, the three-field matrix has completed field alignment and missing data filling, resulting in the circuit breaker operation parameter matrix.

[0107] S413: Based on the circuit breaker operation parameter matrix, unify the field length and arrangement order, standardize the main frequency position difference and amplitude change rate respectively, and keep the operation category field as discrete value unchanged. After completing all field transformations, store the output in the form of a structure to establish the circuit breaker operation feature vector structure.

[0108] Based on the circuit breaker operating parameter matrix, each field undergoes standardized processing to unify its structure. First, the main frequency position difference field is extracted, the range of its maximum and minimum values ​​is determined, and a linear normalization function is set as follows: If the maximum difference value is 50 and the minimum is 15, then after normalization, the normalized value for the 4th frequency band is... The normalized value of the 7th frequency band is Similarly, after taking the maximum value of 0.20 and the minimum value of -0.24 for the amplitude change rate field and performing the same normalization process, the 4th frequency band is... The 7th frequency band is The operation category field remains unchanged as a discrete variable. All fields are combined in a uniform order and reconstructed into a structure array, with the following format:

[0109] Table 8 Standardized Feature Vector Structure

[0110]

[0111] As shown in Table 8, each field has been standardized and restructured to establish a circuit breaker operation feature vector structure.

[0112] Please see Figure 6 The specific steps of S5 are as follows:

[0113] S511: Based on the circuit breaker operation feature vector structure, extract the feature records of all frequency bands within the operation cycle, arrange them in order according to the frequency band number, and use the normalized main frequency position difference field as the basis to determine whether each frequency band constitutes an abnormal state, count the number of consecutively occurring abnormal markers in continuous frequency bands, and generate a spectrum abnormal continuous segment count value.

[0114] Based on the circuit breaker operation feature vector structure, the normalized value field of the main frequency position difference for all frequency bands within each operation cycle is extracted and sorted by frequency band number in ascending order. Anomaly judgment is performed on each frequency band in the sorting results. The anomaly judgment threshold for the normalized value of the main frequency position difference is set to 0.7. When the normalized value of a frequency band is greater than this threshold, the frequency band is marked as abnormal, and its index position is recorded. Subsequently, a sliding window is used to navigate through all frequency bands from front to back. The window size is a set value for the number of consecutive frequency bands within the operation cycle. For example, if the number of consecutive segments is set to 5, then a group is constructed for every 5 consecutive segments, and the anomalies in this group are counted. Taking frequency bands 4 to 8 as an example, if the normalized values ​​of the main frequency positions of bands 5, 6, and 7 are 0.72, 0.84, and 0.75 respectively, and the values ​​of the other two frequency bands are less than 0.7, then the number of consecutive abnormal frequency bands in this window is 3. If the number of abnormalities in the number of consecutive frequency bands is greater than the threshold, it is considered that there is a cluster of spectral offsets in the current operation period. This statistical operation will be performed once for all possible consecutive segment windows and the number of abnormalities in each window will be recorded. Finally, multiple consecutive frequency band groups and their corresponding number of abnormalities will be obtained. The maximum value is the number of consecutive abnormal spectral segments in this period. The results are as follows:

[0115] Table 9 Continuous Statistics of Abnormal Spectrum Segments

[0116]

[0117] As shown in Table 9, the maximum number of consecutive anomalies is 3, and the final count value of consecutive spectral anomalies is generated.

[0118] S512: Based on the count value of abnormal continuous segments of the spectrum, if the number of abnormal segments in any group of continuous frequency bands is greater than the set aggregation length threshold, it is confirmed that there is a spectrum offset aggregation phenomenon within the operation cycle. At the same time, the circuit breaker number and the current operation cycle timestamp information are extracted to generate a spectrum aggregation status record item.

[0119] Based on the count of consecutive abnormal segments in the spectrum, a threshold of 3 is set for spectrum offset aggregation. This threshold is used to determine whether aggregation occurs. If the number of abnormal segments in any window segment group is greater than the aggregation length threshold, the current period is marked as having spectrum offset aggregation. In this embodiment, the aggregation length threshold is set to 3. The data in Table 10 is judged. The number of abnormal segments in segment groups 3-7 and 4-8 is equal to 3, which meets the judgment condition and satisfies the judgment criterion of being greater than or equal to the aggregation threshold. Then, the circuit breaker's encoding information and the timestamp information corresponding to the current operation period are called. For example, the circuit breaker number is "BRK_0923" and the timestamp is 1702280425.123. A triplet record structure is constructed: [BRK_0923, 1702280425.123, aggregation exists], which is used for subsequent summary processing. If no abnormal segment count meets the condition, the field is marked as "aggregation does not exist". This process is executed periodically and stored in the record table, forming the following structure:

[0120] Table 10 Spectrum Aggregation Status Record Table

[0121]

[0122] As shown in Table 10, each operation cycle corresponds to a status record, and finally a spectrum aggregation status record item is generated.

[0123] S513: Based on the spectrum aggregation status record items, summarize the records marked as having aggregation status, and aggregate the circuit breaker number, operation timestamp and aggregation identifier fields to construct the record structure, and add early warning tags to establish a circuit breaker opening and closing fault early warning record;

[0124] After obtaining the spectrum aggregation status records, the records with an aggregation status of "exist" are filtered out. The corresponding circuit breaker number, operation timestamp, and status value are aggregated to construct the early warning output field structure. The fields are defined as "Circuit Breaker Number", "Operation Timestamp", and "Fault Early Warning Identifier". The fault early warning identifier is uniformly set to "Fault Early Warning". Then, the structure records are arranged in ascending order of timestamp. Invalid data entries with "aggregation not present" are filtered out, and only entries with the aggregation status of "exist" are retained to generate the early warning information structure. Assuming that three sets of entries with the aggregation status of "exist" are recorded in the same monitoring period, corresponding to the numbers BRK_0923, BRK_0960 and BRK_1021 respectively, the final output structure is as follows:

[0125] Table 11 Fault Warning Record for Circuit Breaker Opening and Closing

[0126]

[0127] As shown in Table 11, spectrum aggregation anomalies have been screened and identified, and a fault warning record for opening and closing circuits has been established.

[0128] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for early warning of opening and closing faults in a 10kV circuit breaker of a ring main unit system, characterized in that, Includes the following steps: S1: Collect the voltage signals of the opening coil, closing coil and energy storage motor in the 10KV circuit breaker of the ring main unit during the operation cycle, combine the operation start timestamp and operation category label to construct a data structure and generate a current waveform time series dataset. S2: Based on the current waveform time series dataset, perform frequency domain transformation to obtain frequency and amplitude mapping groups, define fixed frequency intervals and determine the main frequency points, arrange them into frequency position vectors according to the segment order, and construct the main frequency distribution mapping structure table. S3: Based on the main frequency points recorded in the main frequency distribution mapping structure table, match the frequency band main frequency point vector in the previous cycle, calculate the frequency offset difference segment by segment to identify abnormal frequency bands and establish a frequency band index, and output the abnormal frequency band index set. S4: Read the frequency band index position recorded in the abnormal frequency band index set, extract the difference of the main frequency position and the amplitude change rate of the corresponding frequency band in the current cycle and the previous cycle, combine the current cycle operation category code to construct a multi-field matrix, unify the field order and length to perform standardization and normalization processing, and form a circuit breaker operation feature vector structure. S5: Based on the circuit breaker operation feature vector structure, perform abnormal frequency band density judgment operation, confirm the spectrum offset clustering phenomenon within the same operation cycle, issue early warning information, and output opening and closing fault early warning record. The main frequency point specifically refers to the frequency position corresponding to the point of maximum amplitude.

2. The method for early warning of 10kV circuit breaker opening and closing faults in a ring main unit system according to claim 1, characterized in that: The abnormal frequency bands specifically refer to frequency bands where the frequency offset difference exceeds the set spectral change discrimination threshold.

3. The method for early warning of opening and closing faults of a 10kV circuit breaker in a ring main unit system according to claim 1, characterized in that: The term "spectral offset clustering phenomenon" specifically refers to counting the number of abnormal frequency bands in the circuit breaker operation feature vector structure. If the number of consecutive abnormal frequency bands is greater than a set clustering length threshold, then the existence of a spectral offset clustering phenomenon is confirmed.

4. The method for early warning of opening and closing faults of a 10kV circuit breaker in a ring main unit system according to claim 1, characterized in that: The current waveform time series dataset includes the current characteristics of the tripping coil, the current characteristics of the closing coil, the current characteristics of the energy storage motor, the operation start time label, and the operation category label. The main frequency distribution mapping structure table includes the frequency interval number, the main frequency point of each interval, and the frequency position vector. The abnormal frequency band index set includes the frequency offset abnormal segment identifier, the frequency offset difference record, and the over-threshold judgment mark. The circuit breaker operation feature vector structure includes the main frequency difference vector, the amplitude change rate vector, and the operation category code. The tripping and closing fault early warning record includes the circuit breaker code, the abnormal frequency band clustering situation, and the operation timestamp.

5. The method for early warning of opening and closing faults of a 10kV circuit breaker in a ring main unit system according to claim 1, characterized in that, The steps for obtaining the current waveform time series dataset are as follows: S111: Collect the voltage signals of the opening coil, closing coil and energy storage motor in the 10KV circuit breaker of the ring main unit during the operation cycle, and uniformly align the collected voltage signals according to the sampling time axis. Then, perform truncation processing based on the operation start timestamp, retain the signal sequence within the operation cycle range, and generate a set of voltage signals within the operation cycle. S112: Based on the voltage signal set within the operation cycle, the voltage signals of each channel are synchronously sampled in terms of voltage value and time dimension by the analog-to-digital converter, discrete voltage sample points are extracted, a sequence of sample points with a uniform time interval is constructed, and they are combined according to the sensor channel order to obtain a discrete multi-channel voltage sequence. S113: Based on the discretized multi-channel voltage sequence, extract the operation start timestamp and operation category label, perform structural calibration and classification coding on each sequence, inject the identifier according to the operation type and complete the unified timing organization to establish a current waveform time series dataset.

6. The method for early warning of 10kV circuit breaker opening and closing faults in a ring main unit system according to claim 1, characterized in that, The steps for obtaining the main frequency distribution mapping structure table are as follows: S211: Based on the current waveform time series dataset, perform fast Fourier transform on the voltage sequence of each channel, perform complex domain transformation operation at a fixed sampling frequency, extract the real and imaginary parts in the corresponding frequency domain, calculate the amplitude of each frequency point, and generate a frequency amplitude corresponding array. S212: Based on the frequency amplitude corresponding array, the frequency axis interval is defined and divided into equally wide segments. The frequency value corresponding to the largest amplitude is retrieved in each segment. If the frequency value is greater than the set amplitude judgment benchmark, it is marked as a candidate point of the main frequency. The amplitude of all candidate points is compared, and the frequency point with the largest amplitude in each segment is located to obtain the interval main frequency set. S213: Based on the set of dominant frequencies in the interval, number and arrange each frequency point according to the frequency interval order, construct a dominant frequency vector structure, bind the value of each frequency point with the index of the interval to generate an ordered mapping pair, and encapsulate it into a two-dimensional table structure to establish a dominant frequency distribution mapping structure table.

7. The method for early warning of 10kV circuit breaker opening and closing faults in a ring main unit system according to claim 1, characterized in that, The steps for obtaining the abnormal frequency band index set are as follows: S311: Based on the main frequency points recorded in the main frequency distribution mapping structure table, extract the main frequency values ​​of each frequency band in the current operation cycle, read the frequency position vector in the previous cycle, perform one-to-one index comparison of each corresponding frequency band, construct a two-cycle main frequency comparison matrix under the same frequency band number, perform vector level difference calculation operation, calculate the frequency offset value sequence of the current cycle relative to the previous cycle, and generate the frequency band main frequency offset vector. S312: Based on the frequency band main frequency offset vector, if the frequency band frequency offset value is greater than the set spectrum change discrimination threshold, the corresponding frequency band is marked as a spectrum fluctuation abnormal segment, and the corresponding segment number is recorded to identify the abnormal and normal frequency band distribution status, thereby obtaining a spectrum abnormality identification array. S313: Based on the spectrum anomaly identifier array, retrieve the index number of the frequency band with an abnormal status, aggregate all the segment numbers that meet the condition that the spectrum offset is greater than the set spectrum change discrimination threshold into a set structure, and perform deduplication and ascending sorting on the set structure to output the abnormal frequency band index set.

8. The method for early warning of opening and closing faults of a 10kV circuit breaker in a ring main unit system according to claim 1, characterized in that, The steps for obtaining the circuit breaker operation feature vector structure are as follows: S411: Read all the frequency band numbers recorded in the abnormal frequency band index set, index the main frequency position value and amplitude value corresponding to the frequency band in the current period and the previous period item by item, construct a two-field vector structure, calculate the difference of the main frequency position value and obtain the frequency change value, and at the same time perform a proportional calculation operation on the amplitude field to obtain the rate of change vector, and generate the main frequency band change parameter set; S412: Based on the main frequency band change parameter set, aggregate the operation category code of the current cycle, and construct a three-field parameter matrix by combining the main frequency position difference, amplitude change rate and operation category number in the order of fields. Perform interpolation operation on the missing fields of each frequency band and mark the missing sections to obtain the circuit breaker operation parameter matrix. S413: Based on the circuit breaker operation parameter matrix, unify the field length and arrangement order, standardize the main frequency position difference and amplitude change rate respectively, and keep the operation category field as a discrete value unchanged. After completing all field transformations, store and output in the form of a structure to establish the circuit breaker operation feature vector structure.

9. The method for early warning of opening and closing faults of a 10kV circuit breaker in a ring main unit system according to claim 1, characterized in that, The steps for obtaining the opening and closing fault early warning record are as follows: S511: Based on the circuit breaker operation feature vector structure, extract the normalized value field of the main frequency position difference of all frequency bands in each operation cycle, and sort them in ascending order of frequency band number. In the sorting result, perform abnormal state judgment on each frequency band, count the number of consecutive segments in the continuous frequency band, and generate the spectrum abnormal continuous segment count value. S512: Based on the spectrum abnormal continuous segment count value, if the number of abnormal segments in any group of continuous frequency bands is greater than the set aggregation length threshold, it is confirmed that there is a spectrum offset aggregation phenomenon within the operation cycle. At the same time, the circuit breaker number and the current operation cycle timestamp information are extracted to generate a spectrum aggregation status record item. S513: Based on the spectrum aggregation status record item, summarize the records marked as having aggregation status, aggregate the circuit breaker number, operation timestamp and aggregation identifier fields to construct the record structure, add warning tags, and establish a circuit breaker opening and closing fault warning record.

Images