A method and system for early warning of abnormal transformer load

Abstract

This invention relates to the field of anomaly early warning technology, specifically to a method and system for transformer load anomaly early warning. The method includes the following steps: acquiring electrical parameter sequences and correcting phase; constructing current and power change rate sequences; extracting load fluctuation trend factors; calculating disturbance growth rate and determining whether it exceeds limits; identifying power gradient abrupt change characteristics; determining whether periodic overlap exists; and outputting a joint anomaly early warning identifier sequence. In this invention, by introducing phase difference and time base offset calculations, a phase-aligned electrical parameter sequence group is constructed. Disturbance trend assessment is achieved through periodic rate mean and integral analysis. The power gradient sequence is used to identify segmental abrupt interference signals. A joint anomaly identification mechanism is constructed in both periodic and segmental dimensions to achieve a composite determination of load disturbance trends and abrupt interference, effectively improving the accuracy and response time of load anomaly early warning and reducing the risk of identification errors caused by hidden fluctuations or local abrupt changes.

Description

Technical Field

[0001] This invention relates to the field of anomaly early warning technology, and in particular to a method and system for early warning of transformer load anomalies. Background Technology

[0002] The field of anomaly early warning technology primarily involves utilizing data analysis, pattern recognition, and artificial intelligence to detect, assess, and predict abnormal states that occur during the operation of equipment, systems, or processes in real time. This field aims to identify abnormal signs that may lead to performance degradation, functional failure, or safety incidents by constructing dynamic threshold models, behavioral feature models, or multivariate analysis mechanisms, thereby issuing early warnings to facilitate timely intervention and handling. This technology is widely used in scenarios with high stability requirements, such as power systems, industrial manufacturing, and transportation. Common methods include sliding window monitoring based on time-series data, cluster identification, Bayesian network modeling, principal component analysis, and neural network learning.

[0003] The transformer load anomaly early warning method is mainly used to detect abnormal load conditions that occur in transformers during operation. The method aims to identify situations that deviate from normal operating conditions, such as overload, drastic load fluctuations, and abnormal peak values, through analysis and continuous monitoring of transformer load data. It then issues timely early warning signals to prevent equipment damage, power outages, or potential grid security risks, thereby ensuring the stability and reliability of the power system.

[0004] Traditional early warning methods rely solely on trend monitoring and anomaly identification of load data. However, they struggle to correct for drift in the sampling time axis or phase deviations between electrical parameter channels. In the presence of sampling offsets, the actual phase relationship between voltage and current becomes distorted, leading to errors in load status identification. The extraction of load fluctuation characteristics depends on single-index analysis and lacks trend assessment and multi-dimensional cross-validation mechanisms based on multiple periods of data. This makes it easy to miss warnings when load disturbances are evolving slowly or experiencing sudden changes. For example, if the frequency does not exceed the threshold but the power fluctuation is significant within a certain period, traditional methods may miss the early warning opportunity due to the simplistic judgment criteria. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a method and system for early warning of abnormal transformer loads.

[0006] To achieve the aforementioned objective, the present invention employs the following technical solution: a transformer load anomaly early warning method, comprising the following steps:

[0007] S1: Obtain the phasor amplitude curves of the transformer voltage channel in multiple cycles, detect the corresponding sampling time of the peak point of the voltage waveform and the peak point of the current waveform in each cycle, calculate the phase difference between voltage and current, correct the time axis of the sampling time point of each electrical parameter channel, and generate a phase-aligned electrical parameter sequence group.

[0008] S2: Call the phase-aligned electrical parameter sequence group to construct the current change rate sequence and the power change rate sequence, detect whether the number of numerical changes within the period exceeds the current sudden change threshold frequency and the power disturbance threshold frequency respectively, and call the two rate averages and frequency terms to form a factor pair to generate a load fluctuation trend factor set.

[0009] S3: Based on the load fluctuation trend factor set, if the load disturbance trend growth rate exceeds the disturbance increase boundary value and the cumulative sum exceeds the risk threshold, then the period is marked as an abnormal trend period, and an abnormal load trend early warning period segment is generated.

[0010] S4: Call the load anomaly trend early warning period segment, extract the power value of the target monitoring segment and the adjacent units before and after, calculate the power gradient linkage difference, compare it with the segment power mutation identification threshold, and if it exceeds the threshold, mark the mutation as existing and generate segment mutation interference signal features.

[0011] As a further aspect of the present invention, the phase-aligned electrical parameter sequence group includes the voltage phasor peak point position, the current root mean square sampling time node, the frequency offset rate correction record, the voltage and current phase deviation mapping table, and the time axis correction synchronization factor. The load fluctuation trend factor set includes the average current change rate, the power change amplitude index, the mutation frequency count item, the total number of disturbance events within the period, and the trend mutation configuration type. The load abnormal trend early warning period segment specifically includes the trend integral sudden increase period, the change rate exceeding the limit period, the cumulative disturbance deviation interval, the abnormal trend period number, and the fluctuation index superposition level. The segment mutation interference signal characteristics include the three-segment power difference value, the gradient comparison factor between the preceding and following segments, the segment fluctuation jump amplitude, the segment response instability label, and the monitoring segment mutation level identifier.

[0012] As a further aspect of the present invention, the step of obtaining the phase-aligned electrical parameter sequence group specifically includes:

[0013] S101: Obtain the phasor amplitude curves of the transformer voltage channel in multiple cycles, obtain the root mean square current curves and frequency offset sequences of the current channel in multiple cycles, locate the peak point position in the voltage waveform for each cycle, extract the sampling time point of the peak value in the current waveform in the corresponding cycle, calculate the sampling time difference between the voltage peak point and the current peak point, and obtain the voltage and current phase difference trend sequence.

[0014] S102: Call the voltage and current phase difference trend sequence, and according to the phase difference data of each cycle, perform difference calculation on the phase difference of the current cycle and the phase difference of the previous cycle in the cycle order to form a change sequence during the cycle. Extract the absolute amplitude of the difference of each cycle and judge the difference with the set sampling offset reference value to obtain the sampling time offset degree of the current cycle and obtain the time reference offset difference group.

[0015] S103: Based on the time reference offset difference group, and based on the offset value of each period, perform time axis translation operation on the sampling time points of voltage channel, current channel and frequency offset in the period, and form a corrected data index sequence after uniform adjustment of the time reference points, obtain the complete period dataset of each sampling channel under the corrected time reference, and generate a phase-aligned electrical parameter sequence group.

[0016] As a further aspect of the present invention, the step of obtaining the load fluctuation trend factor set specifically includes:

[0017] S201: Call the phase-aligned electrical parameter sequence group, extract the values ​​of continuous sampling points within the period according to the load current effective value sequence and the active power time sequence, calculate the difference amplitude of each pair of adjacent sampling points in the sampling order, construct the sequence of current change amplitude with time and the sequence of power change amplitude with time, and obtain the electrical parameter change rate sequence group.

[0018] S202: Based on the electrical parameter change rate sequence group, according to the number of differences in the sequence, detect the frequency of current change and the frequency of power change in the current period, determine whether the number of changes exceeds the current sudden change threshold frequency and the power disturbance threshold frequency respectively, extract the number of frequencies that meet the conditions and establish a time period mark, and obtain the electrical parameter sudden change frequency statistics.

[0019] S203: Call the average value data of the current change rate sequence and the power change rate sequence in the electrical parameter change rate sequence group, and combine them with the change frequency value recorded in the electrical parameter change frequency statistics information to combine the two types of values ​​into a set of factor pairs under the same period to obtain the load fluctuation trend factor set.

[0020] As a further aspect of the present invention, the step of obtaining the load anomaly trend early warning period segment specifically includes:

[0021] S301: Based on the load fluctuation trend factor set, the mean current change rate and the power change frequency term are called, and the product of the two is calculated in each cycle to obtain the periodic product value, construct a continuous sequence representing the load disturbance intensity, and generate a periodic integral fluctuation sequence.

[0022] S302: Based on the periodic integral fluctuation sequence, calculate the difference between the current period value and the previous period value according to the integral value of each period, form an integral change rate sequence, calculate the load disturbance trend growth rate, and obtain the load disturbance trend growth sequence.

[0023] S303: Call the load disturbance trend growth sequence and the periodic integral fluctuation sequence, accumulate the integral change rate within multiple consecutive periods, and jointly judge it with the load disturbance increase boundary value and the load risk accumulation threshold, filter the period number that meets the dual threshold conditions, establish a period number list and trend marker, and obtain the load abnormal trend early warning period segment.

[0024] As a further aspect of the present invention, the formula for calculating the load disturbance trend growth rate is specifically as follows:

[0025] ;

[0026] in, This indicates the growth rate of the cyclical load disturbance trend. Indicates the first The periodic integral fluctuation of each period. Indicates the first The periodic integral fluctuation of each period. Indicates the first The normalized value of the frequency of current jumps within each cycle. Indicates the first Normalized value of the number of power disturbances within a period. Indicates the first Normalized value of the duration of each period, This represents the coefficient indicating the influence of the current disturbance frequency on the weight calculation. This represents the proportionality coefficient by which the duration of the period modifies the outcome. This represents the total number of consecutive periods.

[0027] As a further aspect of the present invention, the step of obtaining the segment abrupt interference signal characteristics specifically includes:

[0028] S401: Call the load abnormality trend early warning cycle segment, and according to the active power data of the transformer secondary side section in the corresponding cycle, divide the power sampling position into a continuously arranged monitoring unit, extract the average power value of the specified target section and the two adjacent forward and backward sections in each cycle, and generate a three-segment power comparison group.

[0029] S402: Based on the power comparison group of the three segments, perform difference calculations to construct forward gradient and backward gradient according to the power of the target segment and the power values ​​of the adjacent segments before and after, arrange them in order to form a power gradient sequence, perform absolute difference calculation on the difference between the first and last terms in the sequence, and calculate the power gradient linkage difference degree.

[0030] The formula for obtaining the power gradient linkage difference is as follows:

[0031] ;

[0032] in, Indicates the first Segment power gradient linkage difference and They represent the first With the Section average power value, Indicates the first Normalized value of section load amplitude Indicates the first Normalized value of power fluctuation amplitude in the section Indicates the first Normalized segment length This is the amplitude adjustment coefficient;

[0033] S403: Call the power gradient linkage difference degree to mark the periodic segments that exceed the mutation judgment threshold, store the corresponding segment number and judgment label as a signal set, and obtain the segment mutation interference signal characteristics.

[0034] As a further aspect of the present invention, the method further includes the following steps:

[0035] S5: Call the load anomaly trend warning period segment and the segment sudden change interference signal characteristics to determine whether there is an overlapping interval in the time period. If both judgments are true in the same period segment, the output period segment is the joint anomaly identification period, and the risk level is marked to generate a load joint anomaly warning identification sequence.

[0036] The load joint anomaly early warning identifier sequence specifically refers to the overlapping cycle identification number, load interference level index, abnormal trend change cycle group, linkage early warning output trigger flag, and cycle risk status instruction.

[0037] As a further aspect of the present invention, the step of obtaining the load joint anomaly warning identifier sequence specifically includes:

[0038] S501: Call the load anomaly trend warning period segment and section sudden interference signal features, extract the anomaly trend identifier number and power sudden identifier number corresponding to each period, compare the period numbers of the two sequences one by one according to the time index, determine whether there is a time overlap interval, and obtain the period anomaly overlap interval information.

[0039] S502: Based on the periodic abnormal overlap interval information, extract the periodic index value with dual abnormal identifiers, establish the periodic segment set corresponding to the index, and perform classification statistics by combining the number of abnormal overlaps and frequency density values ​​to obtain the joint abnormal periodic screening number group.

[0040] S503: Call the joint anomaly cycle screening number group, classify the risk of each cycle segment, and perform risk level labeling based on the number of anomaly duration cycles and the power change response amplitude as part of the risk level judgment rule. Store the cycle number and corresponding level index in cycle order to generate a load joint anomaly early warning identifier sequence.

[0041] A transformer load anomaly early warning system is provided, the system being used to implement the transformer load anomaly early warning method, the system comprising:

[0042] The phase alignment processing module acquires the phasor amplitude curves of the voltage channel within multiple cycles, detects the corresponding sampling time of the peak point of the voltage waveform and the peak point of the current waveform in each cycle, calculates the phase difference between voltage and current, corrects the time axis of the sampling time point of each electrical parameter channel, and generates a phase-aligned electrical parameter sequence group.

[0043] The load fluctuation analysis module calls the phase-aligned electrical parameter sequence group to construct the current change rate sequence and the power change rate sequence. It detects whether the number of numerical changes within the period exceeds the current sudden change threshold frequency and the power disturbance threshold frequency, respectively. It calls the two rate averages and the frequency term to form a factor pair and generates a load fluctuation trend factor set.

[0044] The abnormal trend identification module, based on the load fluctuation trend factor set, marks the period as an abnormal trend period if the load disturbance trend growth rate exceeds the disturbance increase boundary value and the cumulative sum exceeds the risk threshold, and generates a load abnormal trend early warning period segment.

[0045] The mutation interference identification module calls the load abnormal trend early warning period segment, extracts the power value of the target monitoring segment and the adjacent units before and after, calculates the power gradient linkage difference, compares it with the segment power mutation identification threshold, and if it exceeds the threshold, it marks the existence of mutation and generates segment mutation interference signal features.

[0046] The abnormal early warning implementation module calls the load abnormal trend early warning period segment and the segment sudden change interference signal characteristics to determine whether there is an overlapping interval in the time period. If both judgments are true in the same period segment, the output period segment is the joint abnormal identification period, and the risk level is marked to generate a load joint abnormal early warning identification sequence.

[0047] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0048] In this invention, by introducing phase difference and time base offset calculations, the sampling time axis is corrected, and a phase-aligned electrical parameter sequence group is constructed to enhance the time consistency and measurement accuracy between sampled data. The load fluctuation frequency term is extracted by combining the difference sequence of current and power. The disturbance trend is evaluated by periodic rate mean and integral analysis. The abnormal trend period is marked by comparing the cumulative sum of integral change rate with a set threshold. The power gradient sequence is used to identify segmental sudden interference signals. A joint anomaly identification mechanism is constructed in the two dimensions of period and segment to achieve composite judgment of load disturbance trend and sudden interference. This effectively improves the accuracy and response time of load anomaly early warning and reduces the risk of identification error caused by hidden fluctuations or local sudden changes. Attached Figure Description

[0049] 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.

[0050] Figure 1 This is a schematic diagram of the workflow of the present invention;

[0051] Figure 2 This is a detailed flowchart of S1 of the present invention;

[0052] Figure 3 This is a detailed flowchart of the S2 process of the present invention;

[0053] Figure 4 This is a detailed flowchart of the S3 process of the present invention;

[0054] Figure 5 This is a detailed flowchart of the S4 process of the present invention;

[0055] Figure 6 This is a detailed flowchart of S5 of the present invention;

[0056] Figure 7 This is a system flowchart of the present invention. Detailed Implementation

[0057] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0058] 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.

[0059] 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.

[0060] 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.

[0061] 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.

[0062] Please see Figure 1 This invention provides a technical solution: a method for early warning of abnormal transformer load, comprising the following steps:

[0063] S1: Obtain the phasor amplitude curves of the transformer voltage channel over multiple cycles, obtain the root mean square current curves and frequency offset sequences of the current channel over multiple cycles, detect the corresponding sampling times of the peak points of the voltage waveform and the peak points of the current waveform in each cycle, calculate the time difference between the two and convert it into voltage and current phase difference, call the phase difference change value between the current cycle and the previous cycle for difference comparison, obtain the sampling time base offset, and perform time axis correction on the sampling time points of each electrical parameter channel to generate a phase-aligned electrical parameter sequence group;

[0064] S2: Call the phase-aligned electrical parameter sequence group, calculate the numerical difference between adjacent sampling points based on the load current effective value sequence and the active power time sequence, construct the current change rate sequence and the power change rate sequence, detect whether the number of numerical changes in the period exceeds the current sudden change threshold frequency and the power disturbance threshold frequency respectively, form the load fluctuation frequency term of the current period, call the two rate averages and the frequency term to form a factor pair, and generate the load fluctuation trend factor set.

[0065] S3: Based on the load fluctuation trend factor set, the load disturbance integral is constructed by performing periodic multiplication operations on the mean current change rate and the power change frequency term. The load disturbance trend growth rate is calculated based on the difference between the current period integral and the previous period integral. The integral change rate is accumulated over multiple consecutive periods, and the accumulated sum is compared with the load risk accumulation threshold. If the load disturbance trend growth rate exceeds the disturbance increase boundary value and the accumulated sum exceeds the risk threshold, the period is marked as an abnormal trend period, and an abnormal load trend warning period segment is generated.

[0066] S4: Call the load anomaly trend early warning period segment, divide the monitoring segment into multiple independent units according to the active power data of the corresponding transformer secondary side section, extract the power value of the target monitoring segment and the adjacent units before and after, calculate the power difference between the forward and backward sides respectively, construct a three-point power gradient sequence, calculate the power gradient linkage difference, compare it with the section power change identification threshold, if it exceeds the threshold, mark the change as existing, and generate section change interference signal characteristics;

[0067] S5: Call the load anomaly trend warning period segment and the segment sudden change interference signal characteristics. Based on the judgment result, determine whether there is an overlapping interval in the time period. If both judgments are true in the same period segment, the output period segment is the joint anomaly identification period, and the risk level is marked to generate the load joint anomaly warning identification sequence.

[0068] Phasor amplitude is a complex number representing the change of voltage or current over time in a power system, commonly used to describe instantaneous values ​​and phase; root mean square current is a standard current index commonly used in calculating the effective power of alternating current; frequency offset is used to assess the degree of fluctuation of the system frequency from the rated value and is a key parameter for monitoring power grid stability; voltage and current phase difference describes the relative time difference between voltage and current waveforms and determines the power factor; current change rate, power change rate, and abrupt change frequency are key quantities used in load dynamic analysis to reveal the abrupt change trend of electrical parameters during operation; the three-point power gradient sequence is composed of power values ​​at three points (front, middle, and back) and is used to assess the degree of abrupt change in spatial distribution.

[0069] The phase-aligned electrical parameter sequence includes the voltage phasor peak point position, current root mean square sampling time node, frequency offset rate correction record, voltage and current phase deviation mapping table, and time axis correction synchronization factor. The load fluctuation trend factor set includes the average current change rate, power change amplitude index, sudden change frequency count item, total number of disturbance events within the cycle, and trend sudden change configuration type. The load abnormal trend warning cycle segment specifically includes the trend integral sudden increase cycle, change rate exceeding limit period, cumulative disturbance deviation interval, abnormal trend cycle number, and fluctuation index superposition level. The segment sudden change interference signal characteristics include the three-segment power difference value, gradient comparison factor between the preceding and following segments, segment fluctuation jump amplitude, segment response instability label, and monitoring segment sudden change level identifier. The load joint abnormal warning identifier sequence specifically refers to the overlapping cycle identification number, load interference level index, abnormal trend sudden change cycle group, linkage warning output trigger flag, and cycle risk status instruction.

[0070] Please see Figure 2 The specific steps for obtaining the phase-aligned electrical parameter sequence set are as follows:

[0071] S101: Obtain the phasor amplitude curves of the transformer voltage channel in multiple cycles, obtain the root mean square current curves and frequency offset sequences of the current channel in multiple cycles, locate the peak point position in the voltage waveform for each cycle, extract the sampling time point of the peak value in the current waveform in the corresponding cycle, calculate the sampling time difference between the voltage peak point and the current peak point, and obtain the voltage and current phase difference trend sequence.

[0072] To acquire phasor amplitude curves over multiple cycles from the voltage channel and root-mean-square current curves and frequency offset sequences over multiple cycles from the current channel, taking a three-phase transformer system as an example, 128 sets of voltage and current data points are collected per cycle. The position with the largest amplitude in each set of voltage data is extracted and recorded as the peak point of the voltage waveform for that cycle, with the corresponding sampling time recorded as... ,in For periodic indexing, the effective value sequence of the current channel is extracted, and a curve is constructed based on the quadratic average of the current waveform. The peak point in the curve is retrieved, and the corresponding sampling time is recorded as . To ensure sample consistency, a unified time base with the same sampling frequency was used for synchronous extraction. After the first round of time point pairing was completed, the data for each period were calculated separately. The time difference value is converted into an angle value and recorded as the phase difference in degrees. This process is achieved through the following calculation method: ,in The sampling period is set to 20 milliseconds, i.e., for a 50Hz system. If in a certain period... , Then the phase difference of this period is The operation iterates through each cycle sequentially to form a time series data set of voltage and current phase differences. The cycle length can be set to 60 seconds, so a total of 3000 cycles of data points are processed to form a complete cycle phase difference sequence. In the calculation, the harmonic interference in the voltage and current waveforms should be considered. High-frequency interference filtering and fundamental wave reconstruction should be completed in advance before peak extraction to reduce phase error. The frequency offset data can be used as a subsequent correction step to dynamically adjust the reliability judgment index of the phase difference results. After processing, the voltage and current phase difference trend sequence under each cycle is obtained.

[0073] S102: Call the voltage and current phase difference trend sequence. Based on the phase difference data of each cycle, perform difference calculation between the phase difference of the current cycle and the phase difference of the previous cycle in the cycle order to form a change sequence during the cycle. Extract the absolute amplitude of the difference of each cycle and judge the difference with the set sampling offset reference value to obtain the sampling time offset degree of the current cycle and obtain the time reference offset difference group.

[0074] The phase difference data for each period in the voltage-current phase difference trend sequence is retrieved, and the processing period is set to two consecutive periods. A sliding window approach is used for comparison and processing, and each period is numbered. phase difference Calculate its relationship with the previous period. The difference is used to perform a calculation operation: Construct a periodic difference sequence, and extract the absolute value of each difference as a measure of the offset. If the offset magnitude is greater than the sampling offset benchmark value, the offset is considered to be offset. If the deviation is abnormal, it is marked as an abnormal offset period. The sampling offset reference value is set based on the maximum allowable voltage and current phase difference variation range of the transformer system under stable operating conditions. According to GB / T14549 standard, the allowable phase deviation of a conventional 10kV system does not exceed 5°. Therefore, the setting is... This value is fixed and does not change with the cycle, but can be adjusted to 3° or 7° depending on the equipment level. In this embodiment, 5° is used as the judgment benchmark. During the processing, if the difference between two consecutive cycles exceeds this value, such as... , For continuous anomalies, trend accumulation recording should be performed. Furthermore, to enhance the identification of abrupt change segments, an amplitude weighting factor is added to amplify changes exceeding 10°. The amplitude weighting factor is set based on the gain ratio of the phase abrupt change response, and through experimentation, a reasonable range of 1.1 to 1.5 has been obtained. The current setting is [insert value here]. That is, when the difference of a certain period is 10°, it is adjusted to 12° after weighting. This value is determined by the tolerance characteristics of transformer load imbalance to short-term impact. It is suitable for power supply scenarios where the load dynamic adjustment is more frequent. It is evaluated together with the original difference to obtain a more stable offset trend change sequence. After the process is completed, the set of offset values ​​corresponding to all periods that meet the conditions is extracted to obtain the time base offset difference group.

[0075] S103: Based on the time base offset difference group, and based on the offset value of each cycle, perform time axis translation operation on the sampling time points of voltage channel, current channel and frequency offset in the cycle, and form a corrected data index sequence after unified adjustment of the time base points. Obtain the complete cycle dataset of each sampling channel under the corrected time base, and generate a phase-aligned electrical parameter sequence group.

[0076] Based on the offset value of each cycle in the time base offset difference group, the data point index of the voltage channel, current channel, and frequency offset in each cycle is located according to the original sampling index table. The sampling points are synchronously shifted on the time axis. If the offset is positive, the weighted offset steps are added to the sampling point index. For example, if the offset is 1.5ms, corresponding to 128 points per cycle of 20ms at a sampling frequency of 50Hz, then 1.5ms corresponds to an index movement of approximately 10 points. Conversely, if it is negative, the index number is shifted to the left. The same applies to the voltage and current channels, as well as the sequence corresponding to the frequency offset. In the time axis correction, the problem of boundary data loss caused by the movement of the sampling index needs to be addressed. This can be achieved by using mirror filling at the beginning and end of the cycle or by delaying the data shift by one cycle. After the correction is completed, all channels are renumbered, a unified index mapping table is constructed, and the data under the new index is repackaged to form a complete data packet under the same time reference point. Finally, all data are output according to the cycle sequence to obtain the phase-aligned electrical parameter sequence group.

[0077] Please see Figure 3 The specific steps for obtaining the load fluctuation trend factor set are as follows:

[0078] S201: Call the phase-aligned electrical parameter sequence group, extract the values ​​of continuous sampling points within the period according to the load current effective value sequence and the active power time sequence, calculate the difference amplitude of each pair of adjacent sampling points in the sampling order, construct the sequence of current change amplitude with time and the sequence of power change amplitude with time, and obtain the electrical parameter change rate sequence group.

[0079] The load current RMS sequence and active power time series from the phase-aligned electrical parameter sequence group are called, and the values ​​of continuous sampling points within the period are extracted respectively. Taking 128 sampling points per period as an example, the current RMS sequence is extracted point by point and sorted by time index. arrive Numbering, similarly, the power time series is numbered as arrive The difference calculation is performed sequentially on adjacent sampling points of current and power, i.e., the calculation is performed. and The data were recorded into the current rate of change sequence and the power rate of change sequence, respectively, to construct a difference sequence group of length 127. To avoid spurious fluctuations caused by numerical jitter, a 3-point moving average was performed on the original sampled data before constructing the difference sequence. Adjusted to This yields a smoothed base sequence, which is then used for subsequent difference calculations. Assuming three current sampling points within one period are 14.2A, 14.9A, and 14.5A, the moving average for this segment is (14.2 + 14.9 + 14.5) / 3 = 14.53A, and the difference is |14.9 - 14.2| = 0.7A. Each interval is processed sequentially, and the difference is added to the current change rate sequence. The power sequence is processed in the same way. , If the change rate is 0.5kW, then after completing the processing of all cycle data, a set of difference sequences reflecting the amplitude change characteristics between sampling points is constructed, and finally the electrical parameter change rate sequence group is obtained.

[0080] S202: Based on the electrical parameter change rate sequence group, the frequency of current change and power change in the current period are detected according to the number of differences in the sequence. It is determined whether the number of changes exceeds the current sudden change threshold frequency and the power disturbance threshold frequency, respectively. The number of frequencies that meet the conditions is extracted and a time period is established to obtain the statistical information of electrical parameter sudden change frequency.

[0081] Based on the number of differences between the sequences in the electrical parameter change rate sequence group, the judgment criterion is set as follows: within one period, if the current change exceeds 0.5A, it is considered a sudden change; if the power change exceeds 0.8kW, it is considered a disturbance. The number of times the difference exceeds this threshold is counted, and the frequency of current sudden changes is calculated. With power disturbance frequency For example, if there are 17 current changes greater than 0.5A within a certain period, then If the power disturbance count is 9, then The frequency values ​​of these two items are compared with the judgment thresholds. The current mutation threshold frequency is set based on the load current jitter limit under the stable operation condition of the transformer load, and the selected value is 15 times / cycle. The power disturbance threshold frequency is set to 10 times / cycle. This value is determined based on the sampling statistics when the difference between the upper and lower limits of power fluctuation per unit time within a certain cycle is within 5%. If any frequency value in the current cycle exceeds the threshold, it is recorded as the corresponding abnormal event cycle. In subsequent analysis, the abnormal cycle is tagged, numbered and archived. The frequency values ​​and cycle positions of all those that meet the abnormal conditions are extracted, and finally the statistical information of electrical parameter mutation frequency is obtained.

[0082] S203: Call the average data of the current change rate sequence and the power change rate sequence in the electrical parameter change rate sequence group, and combine them with the change frequency value recorded in the electrical parameter change frequency statistics information to combine the two types of values ​​into a set of factor pairs under the same period to obtain the load fluctuation trend factor set.

[0083] The current rate of change sequence and the power rate of change sequence are retrieved from the electrical parameter rate of change sequence group. Their respective average differences over the entire cycle are extracted as the fluctuation amplitude benchmark, set as follows: and For example, if the total rate of change of current within a certain period is 63.5A, with a total of 127 variation intervals, then The total rate of change of power is 101.6kW, then Simultaneously, the mutation frequency was extracted from the statistical information of electrical parameter mutation frequency. , For each group corresponding to a period , Structured index units are formed to create a set of load fluctuation factor pairs, which are recorded as a set of interval-level data structures containing dual-quantity descriptive features. These structures are used to describe the transient operating characteristics of the transformer under a certain period, and finally, a set of load fluctuation trend factors is obtained.

[0084] Please see Figure 4 The specific steps for obtaining the load anomaly trend early warning period are as follows:

[0085] S301: Based on the load fluctuation trend factor set, the mean current change rate and the power change frequency term are called, and the product of the two is calculated in each cycle to obtain the periodic product value, construct a continuous sequence representing the load disturbance intensity, and generate a periodic integral fluctuation sequence.

[0086] Based on the load fluctuation trend factor set, the mean current change rate and the power change frequency term are called, and multiplication is performed on both in each cycle to obtain the periodic product value, thus constructing a periodic integral fluctuation sequence. Assuming a sampling period of 60 seconds, with 128 points per cycle, a total of 60 cycles of data are recorded. The mean current change rate is the average difference between adjacent current sampling points in each cycle. For example, in the [missing information - likely a specific timeframe or period]... During the period, the sum of the differences in current values ​​at 128 points is: The mean rate of change of current in this period is The corresponding power fluctuation frequency term is obtained by dividing the number of abrupt changes by the number of sampling points in the period. If there are 14 instances in the current period where the power difference exceeds 0.8kW, then the power disturbance frequency for that period is... Multiplying the two together yields the periodic integral fluctuation:

[0087] ;

[0088] By calculating the product of the same type over 60 cycles, a series of periodic integral fluctuations with a length of 60 is generated, and a time trend curve for measuring the intensity of load disturbance is constructed, ultimately generating the series of periodic integral fluctuations.

[0089] S302: Based on the periodic integral fluctuation sequence, calculate the difference between the current period value and the previous period value according to the integral value of each period, form an integral change rate sequence, calculate the load disturbance trend growth rate, and obtain the load disturbance trend growth sequence.

[0090] The specific formula for calculating the growth rate of load disturbance trend is as follows:

[0091] ;

[0092] in, This indicates the growth rate of the cyclical load disturbance trend. Indicates the first The periodic integral fluctuation of each period. Indicates the first The periodic integral fluctuation of each period. Indicates the first The normalized value of the frequency of current jumps within each cycle. Indicates the first Normalized value of the number of power disturbances within a period. Indicates the first Normalized value of the duration of each period, This represents the coefficient indicating the influence of the current disturbance frequency on the weight calculation. This represents the proportionality coefficient by which the duration of the period modifies the outcome. This represents the total number of consecutive periods;

[0093] Based on the integral value of each period in the periodic integral fluctuation sequence, the difference amplitude between adjacent periods is calculated to construct an integral rate of change sequence. A weighted summation operation is then performed on this sequence using the following formula:

[0094] ;

[0095] The calculation logic of this formula is as follows: First, integrate the fluctuation amount between two adjacent periods. and The difference constructs the absolute magnitude of the fluctuation term, reflecting the rate of change of the disturbance intensity per period; this magnitude is then multiplied by a set of weighted expressions, where the logarithmic function... Used for comprehensive adjustment of frequency caused by sudden current changes and power disturbance number Regarding the contribution to the growth trend of disturbances, the square root operation ensures a gradual increase in control even under high disturbance frequencies, while the logarithmic operation further compresses the numerical expansion caused by extreme fluctuations; this combined term is multiplied by the current disturbance frequency weighting coefficient. To control its dominance in the overall trend change; the second item The reciprocal form reflects the counter-regulatory effect of cycle duration on stability; that is, the shorter the cycle, the weaker the inhibition of the growth trend. This is multiplied by a weighting coefficient. To balance influence; the entire summation part is based on The divisor is used to form a periodic mean, thereby obtaining the overall growth rate of the load disturbance trend over the complete time period.

[0096] Calculation results Indicates continuity The average disturbance growth trend over a period of time is used to determine whether the load operation is transitioning from mild fluctuations to high-risk fluctuations. The higher the value, the faster the fluctuation intensity increases. If it exceeds a certain set boundary, the system can be considered to have entered a state of risk accumulation.

[0097] Substitute the sample data as follows (in continuous order) (Taking three cycles as an example):

[0098] Table 1. Example table of growth rate of periodic disturbance trends:

[0099]

[0100] As shown in Table 1, the perturbation changes for three consecutive cycles in the sample are recorded.

[0101] Period 4: , , , ;

[0102] Cycle 5: , , , ;

[0103] Period 6: , , , ;

[0104] Weighting coefficient settings: , ;

[0105] The calculation steps are as follows:

[0106] The difference between period 4 and 5:

[0107] ;

[0108] ;

[0109] ;

[0110] ;

[0111] The difference between period 5 and 6:

[0112] ;

[0113] ;

[0114] ;

[0115] ;

[0116] Summary of cycles 4-6:

[0117] ;

[0118] ;

[0119] The result indicates A negative value indicates that the recent trend of disturbance growth has not formed an upward trend, and the system fluctuation remains stable. If the trend reverses and exceeds the set boundary in subsequent cycles, it indicates a potential problem. If so, the formula value can be used to trigger subsequent early warning mechanisms.

[0120] The advantage of the formula is that it allows for adjustments based on the power perturbation frequency. Square root processing is applied to improve sensitivity to low-to-medium amplitude high-frequency disturbances, while also introducing... , The weighting coefficients adjust the influence of the disturbance source components, thereby constructing the growth curve of the periodic disturbance evolution without relying on absolute value judgment.

[0121] S303: Call the load disturbance trend growth sequence and the periodic integral fluctuation sequence, accumulate the integral change rate over multiple consecutive periods, and jointly judge it with the load disturbance increase boundary value and the load risk accumulation threshold, filter the period numbers that meet the dual threshold conditions, establish a period number list and trend marker, and obtain the load anomaly trend early warning period segment.

[0122] Call the load disturbance trend growth sequence and the periodic integral fluctuation sequence, and calculate the disturbance trend growth rate for each period. Periodic integral fluctuation The disturbance amplification boundary value is compared with the corresponding threshold value. Based on the tolerable slope standard of the transformer during load fluctuation response, it is generally set as follows: A value higher than this indicates that the upward trend has entered a dangerous zone; risk accumulation threshold. The value is set to 0.065, indicating that if the total integral fluctuation exceeds this value, it is considered a high-risk operating cycle, such as the cycle. In the middle, if and If a cycle simultaneously meets both judgment conditions, it is marked as a trend warning cycle, numbered and recorded, and a trend label is generated. The cycle number list and corresponding warning intensity index that meet the conditions are sorted out in turn, and finally the load anomaly trend warning cycle segment is obtained.

[0123] Please see Figure 5 The specific steps for obtaining the characteristics of the segmental abrupt interference signal are as follows:

[0124] S401: Call the load abnormality trend early warning period segment. Based on the active power data of the transformer secondary side section in the corresponding period, divide the power sampling position into continuously arranged monitoring units, extract the average power value of the specified target section and the two adjacent forward and backward sections in each period, and generate a three-segment power comparison group.

[0125] The load anomaly trend early warning cycle segment is invoked. Based on the active power sampling data of the transformer secondary side segment within a specific cycle, all power sampling points within each cycle are divided into monitoring units of equal length according to the physical segment location. Each unit is 5 meters long, and the total segment length is 100 meters, which is divided into 20 monitoring units. Within each cycle, the average sampling value of the target monitoring unit to be analyzed is extracted. At the same time, the average power value of its forward and backward adjacent monitoring units is extracted respectively. A three-segment power comparison combination consisting of the target monitoring unit, the forward monitoring unit, and the backward monitoring unit is established to generate a three-segment segment power comparison group.

[0126] S402: Based on the power comparison group of three segments, perform difference calculations to construct forward gradient and backward gradient according to the power values ​​of the target segment and the power values ​​of the adjacent segments, arrange them in order to form a power gradient sequence, perform absolute difference calculation on the difference between the first and last terms in the sequence, and calculate the power gradient linkage difference degree.

[0127] The specific formula for obtaining the power gradient linkage difference is as follows:

[0128] ;

[0129] in, Indicates the first Segment power gradient linkage difference and They represent the first With the Section average power value, Indicates the first Normalized value of section load amplitude Indicates the first Normalized value of power fluctuation amplitude in the section Indicates the first Normalized segment length This is the amplitude adjustment coefficient;

[0130] Based on the average power values ​​of two adjacent directions of the target monitoring section in the three-segment power comparison group, the forward gradient is calculated respectively. With backward gradient Then, the absolute difference between the two gradient endpoints is calculated to form the power fluctuation range, while the load amplitude factor of the target section is introduced. With power fluctuation amplitude factor Both were adjusted using a standard normalization method. It is the ratio of the instantaneous load current amplitude in the current section to the maximum amplitude in the entire section. This is the ratio of the current section's power standard deviation to the maximum standard deviation of the entire section. This is the ratio of the current monitoring unit length to the total monitoring section length. Based on this relationship, the calculation uses the following formula:

[0131] ;

[0132] The calculation logic of this formula is as follows: This represents the absolute difference between the average power values ​​on both sides of the current segment, i.e., the amplitude of power fluctuation across the segment in space. It is the fundamental criterion in this formula, and the subsequent product term... This is a joint adjustment factor that reflects the combined effect of the intensity of load changes and the amplitude of power fluctuations in the current section. This product factor enhances the sensitivity to sudden changes, so that when the load amplitude or power instability increases, the overall response strength also increases accordingly. (Denominator term) This is the normalized result for the segment length, used to balance the amplitude of segments of different lengths, ensuring that differences in physical length do not affect the accuracy of the judgment. The result of this formula... The physical meaning of is the cross-segment power fluctuation intensity index under a unit segment length, adjusted by a disturbance factor, which can be used as a direct reference for detecting abnormal fluctuation boundaries. The power gradient linkage difference degree is calculated.

[0133] During the setup process, The amplitude adjustment coefficient determines the weight of amplitude change in the composite factor. The reference is set as the sensitivity amplification factor of the maximum allowable disturbance value under normal load conditions in the system. In this paper, the load deviation tolerance based on steady-state operation is set to 0.85. This value is adjusted according to the fluctuation trend of the maximum offset value of the load measurement point.

[0134] Let the first The corresponding section: , ,but The load amplitude is 3.7A, and the maximum value is 5.0A. Therefore: The current power standard deviation is 1.8kW, and the maximum standard deviation is 3.0kW. Therefore: The monitoring unit is 5 meters long, and the total length is 20 meters. Therefore: .

[0135] Substitute into the calculation:

[0136] ;

[0137] This result represents the amplification response of the current power disturbance in the context of load fluctuations. The higher the value, the more severe the fluctuations and the closer to or more likely to exceed the abrupt change limit.

[0138] Table 2 Power Fluctuation Factor Sampling Table

[0139]

[0140] Table 2 shows the average power and normalization factor values ​​for typical monitoring sections under a specific period. The values ​​in the table can be directly substituted into the formula to generate multiple [databases / processes]. This is used to further analyze the differences in power disturbances.

[0141] S403: Call the power gradient linkage difference degree to mark the periodic segments that exceed the mutation judgment threshold, store the corresponding segment number and judgment label as a signal set, and obtain the segment mutation interference signal characteristics;

[0142] The power gradient linkage difference is invoked, and it is determined whether it exceeds the mutation detection threshold in each cycle. The threshold is dynamically set under the background of unit segment fluctuation, and the calculation formula is as follows: ,in This is the standard deviation of the average power of all monitoring units within this period segment. Its value is obtained based on the statistical analysis of the average fluctuations of multiple consecutive periods within the system's detection window. For example, if the standard deviation of all power values ​​within a period is 4.0 kW, then the threshold is... .like If the current period corresponds to a segment that is considered to be an abnormal mutation segment, the number, along with the period index and trigger type marker, is stored in the signal set to form an identification record, thereby obtaining the characteristics of the segment mutation interference signal.

[0143] Please see Figure 6 The specific steps for obtaining the load joint anomaly warning identifier sequence are as follows:

[0144] S501: Call the load anomaly trend warning period segment and section sudden interference signal characteristics, extract the anomaly trend identifier number and power sudden identifier number corresponding to each period, compare the period numbers of the two sequences one by one according to the time index, determine whether there is a time overlap interval, and obtain the period anomaly overlap interval information.

[0145] The system retrieves the abnormal trend warning period and segment sudden change interference signal features, extracts the abnormal trend number and power sudden change number marked in the previous results for each period within the current analysis window, and constructs two sets of periodic sequences with timestamp mapping relationships, where the former is the trend abnormal sequence. The latter is a mutated anomalous sequence. Each element is indexed by its corresponding period number and includes a timestamp accurate to the sampling level. Then, using the period timestamp as a comparison benchmark, the period numbers in the two sequences are mapped onto a unified timeline. A dual-pointer comparison method is used to synchronously advance from the beginning. If a period number exists in both sequences simultaneously, or if their time intervals overlap (i.e., the end time of the current period is later than the start time of a period in the other sequence), then the period is considered to have a double anomaly, and the overlapping time interval is recorded. For example, the 7th period... and Simultaneous occurrence, or their starting time is The end time is ,and The 6th cycle is to The time overlap between the two is 0.2 seconds, confirming the overlap status and obtaining information on the abnormal overlap interval of the period.

[0146] S502: Based on the information of periodic abnormal overlapping intervals, extract the periodic index values ​​with dual abnormal identifiers, establish a set of periodic segments corresponding to the index, and perform classification statistics by combining the number of abnormal overlaps and frequency density values ​​to obtain the joint abnormal periodic screening number group.

[0147] Based on the cycle numbers listed in the overlapping interval information of cycle anomalies, extract the index values ​​of all cycle numbers with dual anomaly identification states to construct a joint anomaly cycle set. Subsequently, based on the time continuity criterion, period groups with adjacent period numbers differing by 1 or less are grouped into the same period segment. For example, if the period numbers are {8,9,10,12,13}, then two period segments are formed: {8,9,10} and {12,13}, thus constructing multiple independent groups of abnormally persistent period segments. To improve the accuracy of the screening, a threshold for the number of abnormal overlaps is introduced. The value is set to 1.8 times the average anomaly frequency in this analysis period. In the current example, the total analysis period is 20, and the average frequency is 4. The system counts the cumulative number of abnormal cycles in each periodic segment. If the number of cycles in a segment is greater than or equal to 8, it is included in the screening results. At the same time, the abnormal frequency density value of the periodic segment is calculated, which is the number of abnormal cycles per unit time in the periodic segment. For example, if segment 1 contains 3 cycles and takes 2.4 seconds, the frequency density is 1.25. Finally, the periodic segments that meet the requirement that both the number of cycles and the density exceed the threshold are included in the screening results to obtain the joint abnormal cycle screening number group.

[0148] S503: Call the joint anomaly cycle screening number group, classify the risk of each cycle segment, and perform risk level judgment based on the number of anomaly duration cycles and the power change response amplitude. Perform risk level labeling, store cycle number and corresponding level index in cycle order, and generate load joint anomaly early warning identifier sequence.

[0149] The system invokes the joint abnormal cycle filtering number group, and performs risk level classification for each cycle segment based on its number. First, it counts the number of consecutive cycles contained in each cycle segment. Then, the power abrupt change amplitude value in the corresponding period is extracted from the abrupt interference signal characteristics of the preceding segment and normalized as follows: Both serve as inputs for risk level determination. The risk level labeling formula rules are as follows: When and When marked as "high risk", or The time frame is marked as "medium risk," and all other cases are "low risk." For example, the second period contains four consecutive periods, with a maximum mutation value of 7.6 kW and a normalized baseline value of 10 kW. The value was determined to be "high risk"; the fourth segment contains two cycles, with a maximum mutation value of 3.8kW and a normalized value of 5kW. If the number of cycles and the magnitude reach the medium-high standard respectively, it is still marked as "medium risk". Finally, the judgment results are numbered and recorded in cycle order and mapped to the corresponding level, a sequence structure is constructed and stored in the result set to generate a load joint anomaly early warning identification sequence.

[0150] Please see Figure 7 A transformer load anomaly early warning system is used to execute a transformer load anomaly early warning method. The system includes...

[0151] The phase alignment processing module acquires the phasor amplitude curves of the voltage channel within multiple cycles, detects the corresponding sampling time of the peak point of the voltage waveform and the peak point of the current waveform in each cycle, calculates the phase difference between voltage and current, corrects the time axis of the sampling time point of each electrical parameter channel, and generates a phase-aligned electrical parameter sequence group.

[0152] The load fluctuation analysis module calls the phase-aligned electrical parameter sequence group to construct the current change rate sequence and the power change rate sequence. It detects whether the number of numerical changes within the period exceeds the current sudden change threshold frequency and the power disturbance threshold frequency, respectively. It calls the two rate averages and the frequency term to form a factor pair and generates a load fluctuation trend factor set.

[0153] The abnormal trend identification module is based on the load fluctuation trend factor set. If the load disturbance trend growth rate exceeds the disturbance increase boundary value and the cumulative sum exceeds the risk threshold, the period is marked as an abnormal trend period and an abnormal load trend early warning period segment is generated.

[0154] The mutation interference identification module calls the load abnormal trend early warning period segment, extracts the power value of the target monitoring segment and the adjacent units before and after, calculates the power gradient linkage difference, compares it with the segment power mutation identification threshold, and if it exceeds the threshold, it marks the existence of mutation and generates segment mutation interference signal characteristics.

[0155] The abnormal early warning implementation module calls the load abnormal trend early warning period segment and the segment sudden change interference signal characteristics to determine whether there is an overlapping interval in the time period. If both judgments are true in the same period segment, the output period segment is the joint abnormal identification period, and the risk level is marked to generate the load joint abnormal early warning identification sequence.

[0156] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0157] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0158] It should be understood that, in various embodiments of the present invention, the sequence number of each process does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0159] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0160] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the described devices, apparatuses, and units can be referred to the corresponding processes in the method embodiments, and will not be repeated here.

[0161] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0162] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0163] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0164] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0165] 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 transformer load abnormality early warning method, characterized by, The method comprises the following steps: S1: Obtain the phasor amplitude curve in multiple cycles of the voltage channel of the transformer, detect the corresponding sampling time of the peak point of the voltage waveform and the peak point of the current waveform in each cycle, calculate the phase difference between the voltage and the current, correct the time axis of the sampled time point of each electric parameter channel, and generate a sequence group of electric parameters after phase alignment; S2: Call the sequence group of electric parameters after phase alignment, construct the current change rate sequence and the power variation rate sequence, detect whether the number of value changes in the cycle exceeds the current mutation threshold frequency and the power disturbance threshold frequency respectively, call the mean value of the two rates to form a factor pair, and generate a set of load fluctuation trend factors; The acquisition step of the set of load fluctuation trend factors is specifically: S201: Call the sequence group of electric parameters after phase alignment, extract the values of the continuous sampling points in the cycle according to the load current effective value sequence and the active power time sequence, calculate the difference amplitude of each pair of adjacent sampling points in turn according to the sampling order, construct the sequence of the amplitude of the change of the current with time and the sequence of the amplitude of the change of the power with time, and obtain a sequence group of electric parameter change rates; S202: Based on the sequence group of electric parameter change rates, detect the current change frequency and the power variation frequency in the current cycle according to the difference value number of the sequence, respectively judge whether the number of changes exceeds the current mutation threshold frequency and the power disturbance threshold frequency, extract the number of frequencies meeting the conditions and establish a time period marker, and obtain electric parameter mutation frequency statistical information; S203: Call the average value data of the current change rate sequence and the power variation rate sequence in the sequence group of electric parameter change rates, and combine the mutation frequency value recorded in the electric parameter mutation frequency statistical information, combine the two types of values into a factor pair set under the same cycle, and obtain a set of load fluctuation trend factors; S3: Based on the set of load fluctuation trend factors, if the load disturbance trend growth rate exceeds the disturbance amplitude boundary value and the cumulative sum exceeds the risk threshold, mark the cycle as an abnormal trend cycle, and generate a load abnormal trend warning cycle segment; The acquisition step of the load abnormal trend warning cycle segment is specifically: S301: Based on the set of load fluctuation trend factors, call the mean value of the current change rate and the power variation frequency, multiply the two in each cycle respectively to obtain a periodic product value, construct a continuous sequence representing the load disturbance intensity, and generate a cycle integral fluctuation amount sequence; S302: Based on the cycle integral fluctuation amount sequence, calculate the difference amplitude of the current cycle value and the previous cycle value according to each cycle integral value to form an integral change rate sequence, and obtain a load disturbance trend growth sequence by calculating the load disturbance trend growth rate; S303: Call the load disturbance trend growth sequence and the cycle integral fluctuation amount sequence, accumulate the integral change rate in a plurality of continuous cycles, and respectively judge the cycle number meeting the double-threshold conditions by combining the load disturbance amplitude boundary value and the load risk cumulative threshold, establish a cycle sequence number list and a trend marker, and obtain a load abnormal trend warning cycle segment. S4: calling the load abnormal trend early warning period section, extracting the power values in the target monitoring section and the adjacent units before and after, calculating the power gradient linkage difference, comparing with the section power mutation identification threshold to judge, if it exceeds, it is marked that mutation exists, and a section mutation interference signal feature is generated; The acquisition step of the section mutation interference signal feature is specifically: S401: calling the load abnormal trend early warning period section, according to the transformer secondary side section active power data in the corresponding period, dividing the power sampling position into continuous monitoring units, extracting the average power values in the specified target section and the adjacent forward and backward sections in each period, and generating a three-section section power comparison group; S402: based on the three-section section power comparison group, according to the target section power and the adjacent section power values, respectively performing difference calculation to construct forward gradient and backward gradient, arranging in order to form power gradient sequence, performing absolute difference calculation on the difference amplitude between the first term and the last term in the sequence, and obtaining power gradient linkage difference; The formula for obtaining the power gradient linkage difference is specifically: ； wherein, denotes the section power gradient linkage difference, and denotes the and the section average power value, denotes the section load amplitude normalization value, denotes the section power fluctuation amplitude normalization value, denotes the section length normalization value, is an amplitude adjustment coefficient; S403: calling the power gradient linkage difference, marking the period section that exceeds the mutation determination threshold, storing the corresponding section number and determination label as a signal set, and obtaining a section mutation interference signal feature; S5: calling the load abnormal trend early warning period section and the section mutation interference signal feature, judging whether there is an overlapping interval on the time period, if both determinations are true in the same period section, outputting the period section as a joint abnormal identification period, and marking the risk level, and generating a load joint abnormal early warning identification sequence; The load joint abnormal early warning identification sequence specifically refers to an overlapping period identification number, a load interference level index, an abnormal trend mutation period group, a linkage early warning output trigger flag and a period risk state instruction.

2. The transformer load abnormality early warning method according to claim 1, characterized by, The phase-aligned electrical parameter sequence group includes voltage phasor peak point position, current root mean square sampling time node, frequency offset rate correction record, voltage and current phase deviation mapping table and time axis correction synchronization factor, the load fluctuation trend factor set includes current change rate average, power variation amplitude index, mutation frequency count item, total number of disturbance events in the period and trend mutation configuration type, the load abnormal trend early warning period section is specifically a trend integral sudden increase period, a change rate over-limit period, a cumulative disturbance deviation interval, an abnormal trend period number and a fluctuation index superposition level, and the section mutation interference signal feature includes a three-section power difference value, a forward and backward section gradient comparison factor, a section fluctuation jump amplitude, a section response instability label and a monitoring section mutation level identification.

3. The transformer load abnormality early warning method according to claim 2, characterized by, The acquisition step of the phase-aligned electrical parameter sequence group is specifically: S101: obtaining the phasor amplitude curve in the transformer voltage channel in multiple periods, obtaining the root mean square current curve and the frequency offset sequence collected in multiple periods, positioning the peak point position in the voltage waveform for each period, extracting the sampling time point of the peak value in the current waveform in the corresponding period, calculating the sampling time difference between the voltage peak point and the current peak point, and obtaining the voltage and current phase difference trend sequence; S102: Call the voltage current phase difference trend sequence, according to the phase difference data of each period, difference operation is carried out between the phase difference of the current period and the phase difference of the previous period in the order of period, forming the change sequence between periods, extracting the absolute amplitude of each period difference value and judging the difference with the set sampling offset reference value, obtaining the sampling time offset degree of the current period, obtaining the time reference offset difference value group; S103: According to the time reference offset difference value group, according to the offset value of each period, the time axis translation operation is carried out on the voltage channel, current channel and frequency offset sampling time point in the period, and the unified adjusted time reference point forms the corrected data index sequence, obtains the complete period data set of each sampling channel under the corrected time reference, and generates the phase aligned electric parameter sequence group.

4. The transformer load abnormality early warning method according to claim 1, characterized by, The formula for calculating the load disturbance trend growth rate is: ； wherein, represents the cycle load disturbance trend growth rate, represents the cycle integral fluctuation amount of the th cycle, represents the cycle integral fluctuation amount of the th cycle, represents the normalized value of the current mutation frequency in the th cycle, represents the normalized value of the power disturbance times in the th cycle, represents the normalized value of the cycle duration of the th cycle, represents the influence coefficient of the current disturbance frequency on the weight calculation, represents the proportional coefficient of the cycle duration on the result adjustment, represents the total number of consecutive cycles.

5. The transformer load abnormality early warning method according to claim 1, characterized by, The acquisition step of the load combined abnormal early warning identification sequence is: S501: Call the load abnormal trend early warning period segment and segment mutation disturbance signal feature, extract the abnormal trend identification number and power mutation identification number corresponding to each period, compare the period numbers of the two sequences one by one according to the time index, judge whether there is time overlapping interval, and obtain the period abnormal overlapping interval information; S502: According to the period abnormal overlapping interval information, extract the period index value with double abnormal identification, establish the period segment set corresponding to the index, and classify and count according to the abnormal overlapping number and frequency density value, obtain the combined abnormal period screening number group; S503: Call the combined abnormal period screening number group, risk classification is carried out on each period segment, according to the abnormal duration period number and power mutation response amplitude, the risk grade judgment rule is executed, the risk grade is labeled, and the period number and corresponding grade index are stored in the order of period, and the load combined abnormal early warning identification sequence is generated.

6. A transformer load abnormality early warning system characterized by comprising: The system is used to realize the transformer load abnormal early warning method in any one of claims 1-5, and the system comprises: The phase alignment processing module acquires the phasor amplitude curve in the voltage channel sampling multiple periods, detects the corresponding sampling time of the voltage waveform peak point and the current waveform peak point in each period, calculates the voltage current phase difference, corrects the time axis of each electric parameter channel sampling time point, and generates the phase aligned electric parameter sequence group; The load fluctuation analysis module calls the phase aligned electric parameter sequence group, constructs the current change rate sequence and the power variation rate sequence, detects whether the number of value changes in the period exceeds the current mutation threshold frequency and the power disturbance threshold frequency respectively, calls the two rate means and frequency items to form the factor pair, and generates the load fluctuation trend factor set; The abnormal trend identification module is based on the load fluctuation trend factor set, if the load disturbance trend growth rate exceeds the disturbance amplitude boundary value and the cumulative sum exceeds the risk threshold, the period is marked as an abnormal trend period, and the load abnormal trend early warning period segment is generated. The mutation interference identification module calls the load abnormal trend early warning period section, extracts the power values in the target monitoring section and the adjacent units before and after, calculates the power gradient linkage difference degree, compares and judges with the section power mutation identification threshold, and if it exceeds, marks that the mutation exists, and generates a section mutation interference signal feature; The abnormal early warning implementation module calls the load abnormal trend early warning period section and the section mutation interference signal feature, judges whether there is an overlapping interval on the time period, if both determinations are established in the same period section, outputs the period section as a joint abnormal identification period, and performs risk level labeling, and generates a load joint abnormal early warning identification sequence.

Images