Phase sequence detection and protection circuit and device for power supply protection cabinet and power supply protection cabinet

Abstract

The invention relates to the technical field of power protection cabinets, in particular to a phase sequence detection and protection circuit and device for a power protection cabinet and the power protection cabinet. An acquisition and calculation unit in the circuit acquires a phase sequence parameter sample of a three-phase line, calculates a phase sequence deviation angle, and calculates a normal index by comparing the difference between a single deviation angle and an overall average deviation angle. The sequence construction unit is provided with a plurality of load grades, sorts normal indexes according to the load grades, constructs a normal index sequence of the power supply protection cabinet, and constructs a reference normal index sequence based on a normal index mean value of historical qualified samples. The protection trigger unit calculates the correlation between the normal index sequence and the reference sequence to obtain an abnormal index; and if the abnormal index is greater than a preset threshold value, judging that the phase sequence is abnormal and triggering a protection action. According to the scheme, through multi-dimensional data analysis and historical reference comparison, accurate phase sequence anomaly detection and graded protection are realized, and the safety and reliability of the power supply protection cabinet are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of power protection cabinet technology, and in particular to a phase sequence detection and protection circuit, device, and power protection cabinet for power protection cabinets. Background Technology

[0002] In power systems, power protection cabinets are core devices that ensure the safe operation of electrical equipment. The accuracy of their phase sequence detection directly affects the stable operation of the equipment. If the phase sequence (phase relationship of phases A, B, and C) of a three-phase power supply deviates or becomes abnormal, it may lead to serious problems such as motor reversal, reduced equipment operating efficiency, or even hardware damage. Especially in high-reliability scenarios such as industrial control and power transmission, accurate phase sequence detection and real-time protection are crucial.

[0003] Traditional phase sequence detection methods are mostly based on fixed threshold judgments or simple phase difference calculations, which are difficult to adapt to the dynamic characteristics of power protection cabinets under different load conditions. Existing technologies lack systematic quantitative analysis of phase sequence deviations, relying solely on single indicators or static thresholds for judgment, failing to fully reflect the correlation between phase sequence stability and load level. Furthermore, they do not effectively utilize historical qualified data to construct reference benchmarks, resulting in insufficient sensitivity and accuracy in anomaly detection, easily leading to missed or false detections, and failing to meet protection requirements under complex operating conditions. Summary of the Invention

[0004] The main objective of this invention is to provide a phase sequence detection and protection circuit, device, and power protection cabinet for power protection cabinets. It aims to solve the technical problems of insufficient accuracy, easy omission or misjudgment in the detection of phase sequence abnormalities in power protection cabinets due to the lack of systematic quantitative analysis of phase sequence deviations under different load levels and the lack of a dynamic reference benchmark constructed from historical qualified data in existing phase sequence detection technologies.

[0005] To achieve the above objectives, the present invention provides a phase sequence detection and protection circuit for a power supply protection cabinet, the circuit comprising: an acquisition and calculation unit, a sequence construction unit, and a protection triggering unit;

[0006] The acquisition and calculation unit is connected to the sequence construction unit, the sequence construction unit is connected to the protection triggering unit, and the protection triggering unit is connected to the power protection cabinet.

[0007] The acquisition and calculation unit is used to acquire a set number of phase sequence parameter samples of the three-phase lines in the power protection cabinet and obtain their phase sequence deviation angles, and calculate the normal index of each phase sequence deviation angle. The normal index is obtained by comparing the difference between a single phase sequence deviation angle and the overall average deviation angle of all samples. The phase sequence deviation angle is the sum of the absolute values ​​of the deviations between the actual phase difference and the standard phase difference of phases A, B, and C in the three-phase voltage. The normal index is obtained by comparing the difference between a single phase sequence deviation angle and the overall average deviation angle of all samples, and is used to characterize the degree to which the phase sequence stability of a single sample deviates from the group level.

[0008] The sequence construction unit is used to set multiple load levels for the power protection cabinet, sort the normal index of the power protection cabinet under each load level according to the load level, and construct a normal index sequence for the power protection cabinet; obtain the average normal index of all qualified power protection cabinets in history under all set load levels, and construct a reference normal index sequence.

[0009] The protection triggering unit is used to calculate the abnormal index of the power protection cabinet based on the correlation between the normal index sequence of the power protection cabinet and the reference normal index sequence; when the abnormal index of the power protection cabinet is greater than the preset abnormal threshold, the current phase sequence detection is determined to be abnormal and the protection action is triggered.

[0010] Optionally, the step of collecting a set number of phase sequence parameter samples of the three-phase lines in the power protection cabinet and obtaining their phase sequence deviation angles also includes obtaining the detection time sequence number or equipment running time sequence number corresponding to the collected phase sequence parameter samples.

[0011] The normality index satisfies the following relationship:

[0012] In the formula, Let be the normal index of the i-th phase sequence parameter sample. The phase sequence deviation angle for the first sample. Let be the phase sequence deviation angle of the i-th sample. Let be the phase sequence deviation angle of the m-th sample, and mean() represents the average value.

[0013] Optionally, the normal index further includes calculating a mutation adjustment factor for the normal index corresponding to the phase sequence parameter sample, and using the product of the mutation adjustment factor and the normal index as the normal index after adjustment of the phase sequence parameter sample.

[0014] Optionally, the mutation adjustment factor satisfies the following relationship:

[0015]

[0016] In the formula, Let be the mutation adjustment factor for the normal index corresponding to the i-th phase sequence parameter sample. Let be the phase sequence deviation angle of the i-th sample. For the first Phase sequence deviation angle of each sample Let be the phase sequence deviation angle of the (i+1)th sample; This represents the average number of normally operating samples within the same time period as the abnormal phase sequence parameter samples under the same load conditions. This represents the detection time sequence number or device runtime sequence number corresponding to the i-th sample. This represents the detection time sequence number or device runtime sequence number corresponding to the (i-1)th sample. This is the detection time sequence number or device runtime sequence number corresponding to the (i+1)th sample.

[0017] Optionally, the anomaly index satisfies the following relationship:

[0018] In the formula, Let be the anomaly index of the i-th phase sequence parameter sample. Let i be the normal index sequence of the i-th sample. Let be the reference normal index sequence for the i-th sample, and PPMCC represent the Pearson correlation coefficient.

[0019] Optionally, the step of determining that the current phase sequence detection is abnormal and triggering the protection action further includes:

[0020] The quality evaluation of the power protection cabinet is calculated. When the quality evaluation of the power protection cabinet is less than a preset quality threshold, the phase sequence of the power protection cabinet is determined to be abnormal and a protection action is triggered. The quality evaluation satisfies the following relationship:

[0021] In the formula, For the quality evaluation of the power protection cabinet corresponding to the i-th sample, Let i be the normal index sequence of the i-th sample. Let N be the normal index sequence of the nth qualified sample before the i-th sample, where N is the number of qualified samples before the i-th sample, and cos{} represents the cosine similarity.

[0022] Optionally, the process of obtaining the normal index is as follows:

[0023] Calculate the mean and standard deviation of the phase sequence deviation angle for all collected phase sequence parameter samples, and use the ratio of the difference between the phase sequence deviation angle and the mean to the standard deviation for each sample as the normal index of the phase sequence deviation angle for the corresponding sample.

[0024] Optionally, the quality evaluation also satisfies the following relationship:

[0025] In the formula, For the quality evaluation of the power protection cabinet corresponding to the i-th sample, Let i be the normal index sequence of the i-th sample. Let be the normal index sequence of the nth qualified sample preceding the i-th sample, where N is the number of qualified samples preceding the i-th sample, and cos{} denotes the cosine similarity. Let be the weight of the nth qualified sample relative to the ith sample.

[0026] Furthermore, to achieve the above objectives, the present invention also proposes a phase sequence detection and protection device for a power protection cabinet, wherein the phase sequence detection and protection device for a power protection cabinet includes the phase sequence detection and protection circuit for a power protection cabinet described above.

[0027] In addition, to achieve the above objectives, the present invention also proposes a power protection cabinet, which includes the phase sequence detection and protection circuit for the power protection cabinet described above.

[0028] In this invention, the acquisition and calculation unit in the phase sequence detection and protection circuit of the power supply protection cabinet uses dual quantification of phase sequence deviation angle and normal index to correlate the phase sequence stability of a single sample with the average level of the group, avoiding the one-sidedness of traditional single threshold detection and accurately capturing abnormal fluctuations in phase sequence deviation. The sequence construction unit constructs independent normal index sequences for different load levels such as light load and heavy load, breaking the limitation of traditional detection ignoring load differences, enabling the detection model to dynamically reflect the phase sequence characteristics of the power supply protection cabinet under different operating conditions; by extracting the average normal index of historical qualified samples to construct a reference sequence, a dynamic benchmark is provided for anomaly detection, solving the defect of existing technology relying on static thresholds, and significantly improving the sensitivity and anti-interference capability of anomaly judgment; the protection triggering unit quantifies the degree of anomaly by calculating the correlation between the normal index sequence and the reference sequence. Compared with the traditional method of simply comparing deviation values, it can more comprehensively assess the overall deviation trend of phase sequence characteristics and reduce the risk of missed detection; by triggering differentiated protection actions by preset anomaly thresholds, a graded response mechanism from early warning to emergency protection is realized, ensuring equipment safety while avoiding over-protection and improving the stability and economy of system operation. Attached Figure Description

[0029] Figure 1 This is a schematic diagram of the first embodiment of the phase sequence detection and protection circuit for power supply protection cabinet of the present invention;

[0030] Figure 2 This is a schematic diagram illustrating the specific execution steps of the phase sequence detection and protection circuit for power supply protection cabinets according to the present invention.

[0031] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0032] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0033] It should be noted that if the embodiments of the present invention involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of the components in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0034] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0035] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application.

[0036] This invention provides a phase sequence detection and protection circuit for a power supply protection cabinet, referring to... Figure 1 As shown, Figure 1 This is a structural block diagram of the first embodiment of the phase sequence detection and protection circuit for a power supply protection cabinet according to the present invention. The phase sequence detection and protection circuit for a power supply protection cabinet of the present invention includes: an acquisition and calculation unit, a sequence construction unit, and a protection triggering unit;

[0037] Among them, the data acquisition and computing unit is connected to the sequence construction unit, the sequence construction unit is connected to the protection triggering unit, and the protection triggering unit is connected to the power protection cabinet;

[0038] The data acquisition and calculation unit is used to acquire a set number of phase sequence parameter samples of the three-phase lines in the power protection cabinet and obtain their phase sequence deviation angles. It calculates the normal index of each phase sequence deviation angle. The normal index is obtained by comparing the difference between a single phase sequence deviation angle and the overall average deviation angle of all samples. The phase sequence deviation angle is the sum of the absolute values ​​of the deviations between the actual phase difference and the standard phase difference of phases A, B, and C in the three-phase voltage. The normal index is obtained by comparing the difference between a single phase sequence deviation angle and the overall average deviation angle of all samples. It is used to characterize the degree to which the phase sequence stability of a single sample deviates from the group level.

[0039] The sequence construction unit is used to set multiple load levels for the power protection cabinet, sort the normal index of the power protection cabinet under each load level according to the load level, and construct the normal index sequence of the power protection cabinet; obtain the average normal index of all qualified power protection cabinets in history under all set load levels, and construct a reference normal index sequence.

[0040] The protection triggering unit is used to calculate the abnormal index of the power protection cabinet based on the correlation between the normal index sequence of the power protection cabinet and the reference normal index sequence; when the abnormal index of the power protection cabinet is greater than the preset abnormal threshold, the current phase sequence detection is determined to be abnormal and the protection action is triggered.

[0041] The acquisition and calculation unit is connected to the three-phase line at its input and to the sequence construction unit at its output. It can be used to acquire phase sequence parameter samples from the three-phase line, calculate the phase sequence deviation angle, and generate a normality index based on the overall average deviation angle to quantify the degree of deviation of a single sample's phase sequence stability relative to the population level. It is understandable that traditional methods only use fixed thresholds to determine whether the phase sequence is abnormal, failing to reflect the relative position of individual samples in the overall distribution, resulting in insensitivity to slight but persistent shifts. In an exemplary embodiment, the acquisition and calculation unit can receive voltage signals and their phase information from phases A, B, and C of the three-phase line, acquire multiple sets of phase sequence parameters through real-time sampling, calculate the phase sequence deviation angle for each set of samples, further calculate the overall average deviation angle for all samples, and compare the single phase sequence deviation angle with this average to obtain the normality index, thus obtaining the phase sequence deviation angle and the corresponding normality index.

[0042] For example, the acquisition and calculation unit may employ, but is not limited to, a sampling and calculation module based on a digital signal processor (DSP), or one or more methods based on a field-programmable gate array (FPGA) to achieve high-speed parallel processing. The sequence construction unit is connected to the acquisition and calculation unit at its input and to the protection trigger unit at its output. It can be used to classify and sort the normal indexes according to a preset load level, construct the normal index sequence of the current power protection cabinet, and simultaneously construct a reference normal index sequence based on historical qualified equipment data as a dynamic detection benchmark.

[0043] It is understandable that existing technologies do not differentiate the impact of load conditions on phase sequence characteristics, uniformly employing a global threshold, which is prone to misjudgment under light or heavy load conditions. In this embodiment, the sequence construction unit may receive the normal index and corresponding load level identifier from the acquisition and calculation unit, and form an ordered sequence by classifying and sorting the normal indices under different load conditions according to their levels. Simultaneously, it may extract the average normal index under the same load level from historical qualified power protection cabinets to form a reference sequence, thereby obtaining the normal index sequence and reference normal index sequence of the current power protection cabinet. Furthermore, the sequence construction unit may adopt, but is not limited to, a hierarchical storage structure based on a microcontroller, or one or more of the following: a tagged database to manage the index sequences under different load levels.

[0044] The input of the protection trigger unit is connected to the sequence construction unit, and the output is connected to the control loop of the power protection cabinet. It can be used to calculate the correlation between the current normal index sequence and the reference normal index sequence to obtain the abnormal index, and decide whether to trigger the protection action based on the preset abnormal threshold. It is understandable that traditional protection logic relies on tripping when the instantaneous deviation exceeds the threshold, which makes it difficult to distinguish between occasional disturbances and real trend faults, resulting in over-response problems.

[0045] In one specific embodiment, the protection triggering unit may receive a normal index sequence and a reference normal index sequence from the current power protection cabinet, and evaluate the similarity between the two sequences using a statistical correlation algorithm. The greater the deviation, the lower the correlation and the higher the abnormal index. Once a threshold is reached, the corresponding protection mechanism is activated, obtaining the abnormal index and the triggered protection action command. For example, the protection triggering unit may be, but is not limited to, integrated into an embedded controller, or implemented through an independent relay module to achieve one or more of multi-level output responses.

[0046] The power protection cabinet connects the three-phase line input terminal and the load output terminal, receiving action commands from the protection trigger unit. It can be used to support three-phase power supply lines, providing electrical isolation and overload protection, and participates in operational status feedback as a phase sequence detection device. Understandably, this is already included in traditional circuits, requiring no additional adjustments. The three-phase line connects the power input terminal and the main circuit of the power protection cabinet, providing voltage sensing signals to the acquisition and calculation unit. It can be used to transmit three-phase AC power, providing A-phase, B-phase, and C-phase voltage signals for phase sequence detection. Understandably, this is already included in traditional circuits, requiring no additional adjustments.

[0047] Phase sequence parameter samples can be used to record the phase relationship data of three-phase voltages at a specific moment, serving as the basic data unit for analyzing phase sequence stability. It is understandable that traditional methods typically only focus on whether the current sample meets the standard phase sequence, lacking a comprehensive evaluation of the consistency and stability among multiple samples. In an exemplary embodiment, the phase sequence parameter samples can be received from voltage waveform sampling points provided in real time by the three-phase line, processed by a data acquisition and calculation unit, and output as a basic dataset for calculating the phase sequence deviation angle. The phase sequence deviation angle can be used to quantify the total deviation between the actual three-phase voltage phase difference and the ideal standard, serving as a basic indicator for evaluating phase sequence accuracy. It is understandable that traditional methods only use whether the deviation exceeds a fixed threshold to determine anomalies, ignoring its fluctuation characteristics in the time series. In this embodiment, the phase sequence deviation angle can be obtained by receiving the actual phase difference of phases A, B, and C and the standard phase difference, subtracting the actual phase difference between each phase from the standard value, taking the absolute value, and summing the results to obtain the total deviation angle, outputting the sum of the absolute values ​​of the deviations between the three phases.

[0048] The normality index can be used to characterize the degree of deviation of the phase sequence stability of a single phase sequence parameter sample from the average level of the overall sample, enhancing the sensitivity of anomaly detection. Understandably, traditional detection methods do not incorporate such relative stability indicators and cannot identify gradual degradation processes. In one specific embodiment, the normality index can receive a single phase sequence deviation angle and the overall average deviation angle, compare the single deviation angle with the overall mean through normalization, amplify small but systematic deviations, and output a dimensionless numerical value representing the relative difference between an individual and the group.

[0049] The overall average deviation angle can be used as the statistical center value of the deviation angles of all phase sequence parameter samples, providing a reference benchmark for calculating the normality index. It is understood that traditional methods do not establish a group reference benchmark, relying solely on theoretical standards for judgment, resulting in poor adaptability. In this embodiment, the overall average deviation angle can be the phase sequence deviation angles of all phase sequence parameter samples received in a predetermined number. By calculating the arithmetic mean or weighted average of the deviation angles of all samples, the corresponding mean result is output.

[0050] The single phase sequence deviation angle describes the degree to which the three-phase voltage phase relationship deviates from the ideal state at a given moment, and is a basic input item for constructing the normality index. Understandably, traditional methods view the deviation of each sample in isolation, lacking correlation analysis with the overall trend. In an exemplary embodiment, the single phase sequence deviation angle can be the deviation between the actual phase difference and the standard phase difference between phases AB, BC, and CA under that sample, outputting the corresponding phase sequence deviation angle. The load level can be used to divide the power range of the power protection cabinet during operation, distinguishing the phase sequence response characteristics under different operating conditions.

[0051] In one specific embodiment, the load level can be the current output current or power level of the power supply protection cabinet. This can be defined by segmenting according to the percentage of rated capacity or dynamically classifying the level based on the load current change rate, outputting category identifiers such as light load, medium load, and heavy load. The normal index sequence can be used to organize the current power supply protection cabinet's normal indices into an ordered set according to the load level, reflecting its phase sequence stability pattern under various operating conditions. It is understood that traditional detection methods do not establish a load-related stability model, making it difficult to adapt to varying operating conditions. In this embodiment, the normal index sequence can be multiple normal indices received under the same load level. By centrally sorting the normal indices under the same load conditions, a data structure that can be used for trend comparison is formed, outputting a normal index sequence categorized by load level.

[0052] Historically qualified power protection cabinets have their operational data collected by the sequence construction unit to generate a reference normality index sequence. This sequence provides a validated data source of normally operating equipment, enabling the construction of a statistically representative reference sequence. Understandably, traditional methods do not utilize historical data from a group of qualified equipment as a basis for judgment, lacking an external reference system. The reference normality index sequence can be constructed from the average normality index of historically qualified equipment at various load levels, serving as a dynamic benchmark for current equipment anomaly detection. Understandably, traditional technologies rely on static thresholds and cannot automatically update the judgment criteria as the performance of the equipment group evolves.

[0053] In an exemplary embodiment, the reference normal index sequence can be obtained by receiving the normal indices of multiple historical qualified power protection cabinets under the same load level. Through horizontal aggregation of a large amount of qualified equipment data, stable behavioral characteristics under typical operating conditions are extracted, and a normal index mean sequence corresponding to each load level is output. The anomaly index can be used to quantify the overall deviation between the current power protection cabinet's normal index sequence and the reference normal index sequence, and to determine the abnormal phase sequence state. It is understood that traditional methods only compare the current value with a threshold, and cannot capture sequence-level trend deviations. In this embodiment, the anomaly index can be obtained by receiving the current normal index sequence and the reference normal index sequence, calculating the statistical correlation between the two sequences using the Pearson correlation coefficient or other similarity measurement methods, and outputting a numerical indicator reflecting the consistency of their trends.

[0054] Correlation can be used to measure the consistency of the changing trends between the current normal index sequence and the reference normal index sequence, supporting the generation of abnormal indices. In one specific embodiment, the correlation can be a numerical sequence of two normal index sequences, calculated using a linear correlation coefficient method or a dynamic correlation analysis algorithm based on a sliding window, outputting a correlation coefficient representing the degree of trend matching. An abnormal threshold can be used to set a critical value for determining whether an abnormal index exceeds the allowable range, controlling the triggering timing of protective actions. In an exemplary embodiment, the abnormal threshold can be an empirical value calibrated during the system debugging phase or a learned dynamic boundary, set as a fixed constant threshold or an adaptively adjusted floating threshold based on historical data, outputting a threshold parameter for comparison with the abnormal index. Protective actions can be used to execute corresponding response measures based on the magnitude of the abnormal index, including alarm prompts, output restrictions, or emergency power cutoffs. In one specific embodiment, the protective action can be an output warning signal, a power reduction command, or a cutoff command, configured as a multi-level response mode, such as triggering an audible and visual alarm in the first level, reducing the output voltage in the second level, and cutting off the main circuit in the third level, achieving differentiated responses.

[0055] A voltage sensor connected to the three-phase line input and the data acquisition unit in phase A can be used for one phase of a three-phase power supply, providing a voltage signal for phase sequence detection. Understandably, this is already included in conventional circuits, requiring no additional adjustments. A voltage sensor connected to the three-phase line input and the data acquisition unit in phase B can be used for the middle phase of a three-phase power supply, providing a voltage signal for phase sequence detection. Understandably, this is already included in conventional circuits, requiring no additional adjustments. A voltage sensor connected to the three-phase line input and the data acquisition unit in phase C can be used for the third phase of a three-phase power supply, providing a voltage signal for phase sequence detection. Understandably, this is already included in conventional circuits, requiring no additional adjustments. The standard phase difference can be used to define the phase interval between adjacent phases in an ideal three-phase system, serving as a theoretical benchmark for calculating phase sequence deviation.

[0056] In an exemplary embodiment, the standard phase difference can be the phase difference value specified in the power grid design specifications (typically 120°), outputting a standard value for comparison with the actual phase difference. The actual phase difference can be the real-time phase difference between any two phases measured, reflecting the current phase sequence status of the system. In a specific embodiment, the actual phase difference can be the time difference or angle difference of the voltage waveforms between phases AB, BC, and CA, outputting three actual phase difference values. The sum of the absolute values ​​of the deviations can be used to sum the absolute values ​​of the deviations between the actual phase differences and the standard phase differences of each phase to form the total phase sequence deviation angle. It is understood that although traditional methods involve similar calculations, they are not incorporated into the dynamic sequence analysis framework. In this embodiment, the sum of the absolute values ​​of the deviations can be the absolute values ​​of the deviations between the three phases AB, BC, and CA. By calculating the absolute value of each phase deviation separately and then summing them, the final result is obtained, outputting the total phase sequence deviation angle.

[0057] For example, in a scenario where an industrial motor drive system operates under conditions of frequent load changes, the phase sequence detection and protection circuit for the power protection cabinet in this embodiment can be a large water pump driven by a frequency converter connected to the power protection cabinet, experiencing rapid switching from light load to heavy load during start-up and shutdown. The acquisition and calculation unit continuously acquires three-phase voltage signals, calculates the phase sequence deviation angle at each moment, and generates the corresponding normal index. The sequence construction unit classifies the normal index according to the current load level, constructs the normal index sequence of this equipment under light and heavy loads; at the same time, it calls the historical qualified equipment database to generate a reference normal index sequence for the corresponding load level. The protection triggering unit calculates the correlation between the current sequence and the reference sequence to obtain the abnormal index. When the water pump suddenly experiences a loose phase sequence connection, resulting in slow phase drift, although each individual deviation does not exceed the limit, the sequence trend continuously deviates from the reference mode, the abnormal index gradually increases, and eventually exceeds the preset threshold, triggering a warning signal to remind maintenance personnel to check, thus preventing the motor from reversing.

[0058] This embodiment provides a phase sequence detection and protection circuit for a power supply protection cabinet. The circuit acquires a set number of phase sequence parameter samples from the three-phase lines through a data acquisition and calculation unit, obtains their phase sequence deviation angles, and calculates the normality index of each phase sequence deviation angle. The normality index is obtained by comparing the difference between a single phase sequence deviation angle and the overall average deviation angle of all samples. The phase sequence deviation angle is the sum of the absolute values ​​of the deviations between the actual phase difference and the standard phase difference of phases A, B, and C in the three-phase voltage, used to characterize the degree to which the phase sequence stability of a single sample deviates from the group level. A sequence construction unit sets multiple load levels for the power supply protection cabinet, placing the power supply protection cabinet under each load level... Normal indices under different load levels are sorted according to load level to construct a normal index sequence for the power protection cabinet. The average normal index of all historically qualified power protection cabinets under all set load levels is obtained to construct a reference normal index sequence. Based on the correlation between the normal index sequence and the reference normal index sequence, the protection triggering unit calculates the abnormal index of the power protection cabinet. When the abnormal index of the power protection cabinet exceeds a preset abnormal threshold, the current phase sequence detection is determined to be abnormal, and protection action is triggered. The phase sequence deviation angle is calculated by the acquisition and calculation unit and incorporated into the normal index, thus combining the phase sequence stability of a single sample with its position in the overall sample. The deviation degree is quantitatively correlated, overcoming the limitations of traditional methods that rely solely on fixed thresholds or local deviations for judgment, and improving the sensitivity to abnormal fluctuations. By classifying and sorting normal indices according to multiple preset load levels through sequence construction units, a normal index sequence for the power protection cabinet itself is constructed, enabling the detection model to dynamically adapt to phase sequence characteristics under different operating conditions and avoiding misjudgments due to load changes. Furthermore, by aggregating the average normal index values ​​of all qualified power protection cabinets under the same load level, a statistically representative reference normal index sequence is constructed as a dynamic benchmark, solving the problem that static thresholds in existing technologies cannot adapt to system conditions. The system overcomes the long-term evolutionary deficiencies of traditional methods, significantly enhancing the robustness and universality of detection. By calculating the correlation between the current normal index sequence and the reference sequence through the protection trigger unit, an anomaly index is derived. This index reflects the overall trend deviation rather than a single instantaneous deviation, enabling more comprehensive identification of gradual or complex phase sequence anomalies and reducing the false negative rate. When the anomaly index exceeds a preset threshold, a graded protection action is triggered, achieving differentiated responses from early warning to emergency disconnection. This ensures equipment safety while preventing unnecessary shutdowns caused by over-protection, thereby comprehensively improving the phase sequence detection accuracy, response intelligence, and operational economy of the power supply protection cabinet in complex operating environments.

[0059] In one embodiment, collecting a set number of phase sequence parameter samples of the three-phase lines in the power protection cabinet and obtaining their phase sequence deviation angles also includes obtaining the detection time sequence number or equipment running time sequence number corresponding to the collected phase sequence parameter samples; the normal index satisfies the following relationship:

[0060] In the formula, Let be the normal index of the i-th phase sequence parameter sample. The phase sequence deviation angle for the first sample. Let be the phase sequence deviation angle of the i-th sample. Let be the phase sequence deviation angle of the m-th sample, and mean() represents the average value.

[0061] The detection time sequence number is associated with and stored in the acquisition and calculation unit along with the phase sequence parameter samples. This can be used to identify the absolute time sequence of each phase sequence parameter sample acquisition, establish the arrangement relationship of samples on the time axis, and support time-series-based stability evolution analysis. It is understood that traditional phase sequence detection does not record the sampling time sequence, making it impossible to analyze the trend of deviation changes over time and difficult to identify progressive faults. In an exemplary embodiment, the detection time sequence number can be the sampling time information provided by the system's real-time clock, obtained through an incremental numbering mechanism to obtain an incremental time sequence number or timestamp. For example, the detection time sequence number can include, but is not limited to, one or more of an integer index starting from 0 or a UTC-based timestamp format. The equipment runtime sequence number is synchronously recorded with the phase sequence parameter samples in the acquisition and calculation unit. This can be used to reflect the sampling point sequence number corresponding to the cumulative operating time of the power protection cabinet since startup, and to associate the phase sequence characteristic drift during equipment aging or thermal changes. It is understood that existing technologies do not incorporate equipment runtime into the phase sequence evaluation system, resulting in a lack of awareness of slow degradation phenomena during long-term operation. In this embodiment, the device runtime sequence number can be a segmentation marker of the cumulative running time after the receiving device is started. Through time period segmentation logic, a sequence number label divided by runtime segments is obtained. Furthermore, the device runtime sequence number can be one or more of the following: continuous sequence numbers divided at fixed time intervals, or non-uniform sequence numbers divided according to nodes of changing operating states.

[0062] The normality index calculation formula defines the specific mathematical expression of the normality index. By introducing the initial sample, the current sample, and the overall average deviation angle, and combining it with the mean function to achieve dual deviation quantification, it can be used to output the normality index of the i-th sample. It is understandable that traditional methods only compare the current deviation with a fixed threshold, lacking joint consideration of historical starting points and statistical trends, resulting in insufficient sensitivity. In a specific embodiment, the normality index calculation formula can receive the phase sequence deviation angles of the 1st sample, the i-th sample, and all m samples. By using the initial sample as a benchmark reference and combining it with the population mean for normalization, the normality index simultaneously reflects the degree of deviation relative to the initial state and the overall trend, thus obtaining the normality index of the i-th sample. For example, the normality index calculation formula can include, but is not limited to, one or more of the following: a linear weighted combination form or a nonlinear normalization function structure.

[0063] For example, in a scenario where a power protection cabinet experiences gradual phase drift due to loose terminals during prolonged continuous operation, the phase sequence detection and protection circuit for the power protection cabinet in this embodiment can be configured such that the three-phase voltage phase sequence is normal at the initial startup of the equipment, and the first sample records the ideal phase sequence deviation angle. As the operating time increases, the phase difference slowly shifts due to the gradual loosening of a certain phase terminal caused by thermal expansion and contraction. The acquisition and calculation unit continuously acquires phase sequence parameter samples with duration numbers and calculates the relationship based on a newly defined normal index, comparing the deviation of each sample with the initial state and the overall average value. Although a single deviation does not exceed the traditional limit, the normal index increases significantly due to its continuous deviation from the initial state and the overall upward trend. The sequence construction unit generates a normal index sequence that evolves with the operating time. The protection triggering unit detects that the correlation between this sequence and the reference sequence has decreased, and the abnormal index exceeds the threshold, issuing an early warning to avoid subsequent possible motor reversal or insulation breakdown accidents.

[0064] This embodiment collects a set number of phase sequence parameter samples from the three-phase lines in the power protection cabinet and obtains their phase sequence deviation angles. It also obtains the detection time sequence number or equipment runtime sequence number corresponding to the collected phase sequence parameter samples. The detection time sequence number establishes the arrangement relationship of the samples on the time axis to support stability evolution analysis. The equipment runtime sequence number is associated with the equipment operation process to reflect the phase sequence drift under aging or thermal changes. The normal index calculation formula performs dual deviation quantification based on the initial sample, the current sample, and the overall average deviation angle to improve anomaly sensitivity. This can achieve the technical effects of dynamically tracking the phase sequence stability evolution process, enhancing the detection sensitivity of early minor anomalies, ensuring data comparability and sequence consistency under different operating conditions, upgrading anomaly judgment from static isolation to trend assessment, improving early warning capability and judgment accuracy, and supporting high-sensitivity, low-false-alarm-rate graded protection response.

[0065] In one embodiment, the normal index further includes calculating a mutation adjustment factor for the normal index corresponding to the phase sequence parameter sample, and using the product of the mutation adjustment factor and the normal index as the normal index after adjustment of the phase sequence parameter sample.

[0066] The mutation adjustment factor is used to quantify the rate of change or gradient between adjacent data points in the phase sequence parameter sample sequence, and can be used to dynamically adjust the response strength of the original normal index to mutation trends. It is understood that traditional methods lag in response to rapid phase jumps and have difficulty distinguishing between real faults and transient interference, leading to missed detections or false alarms. In an exemplary embodiment, the mutation adjustment factor can receive the normal index and its temporal difference values ​​corresponding to multiple consecutive phase sequence parameter samples, calculate the degree of mutation based on the slope, second-order difference, or variance change rate within a sliding window, and can be designed as a gain amplification or attenuation suppression function to obtain a dimensionless adjustment coefficient, which is used to multiply with the original normal index. Exemplarily, the mutation adjustment factor can include, but is not limited to, one or more of the following: a normalization factor based on the absolute value of the first-order difference, or a nonlinear compression function based on local variance.

[0067] The normal index, adjusted from the phase sequence parameter samples, is a correction index that integrates the original normal index with the dynamic characteristics of mutations. It can reflect the phase sequence stability level after considering the weighted average of change trends. It is understood that the normal index in the prior art is a static statistical result, which cannot reflect short-term dynamic change characteristics and affects the early identification capability of anomalies. Furthermore, the normal index adjusted from the phase sequence parameter samples can be obtained by receiving the original normal index and the corresponding mutation adjustment factor, and multiplying the mutation adjustment factor by the original normal index to enhance sensitivity to abrupt changes or suppress noise fluctuations, resulting in an adjusted normal index value obtained through product operation. In this embodiment, the normal index adjusted from the phase sequence parameter samples can adopt a linear weighted product form, or introduce a threshold segmentation adjustment mechanism to distinguish one or more of different mutation levels.

[0068] For example, in a scenario where a power protection cabinet experiences a brief phase jump after a lightning strike, the phase sequence detection and protection circuit for the power protection cabinet in this embodiment can be a stable sequence of phase sequence parameters collected during normal system operation. A lightning strike causes a momentary distortion in the B-phase voltage, resulting in a sharp increase in the phase sequence deviation angle. The acquisition and calculation unit calculates that the original normal index of this sample is higher than the mean, but relying solely on historical thresholds may not reach the alarm threshold. At this time, the mutation adjustment factor detects the steep upward gradient of the normal index between the previous and subsequent samples, outputs an amplification coefficient greater than 1, and significantly improves the adjusted normal index after multiplying it by the original normal index. The sequence construction unit incorporates this high value into the normal index sequence under the current load level, causing the overall sequence to show a significant deviation trend. The protection triggering unit calculates that its correlation with the reference sequence has decreased, and the abnormal index quickly exceeds the preset threshold, triggering an early warning action to prompt maintenance personnel to investigate the impact of the transient event and avoid subsequent similar disturbances from causing a chain reaction.

[0069] In one embodiment, the mutation adjustment factor satisfies the following relationship:

[0070]

[0071] In the formula, Let be the mutation adjustment factor for the normal index corresponding to the i-th phase sequence parameter sample. Let be the phase sequence deviation angle of the i-th sample. For the first Phase sequence deviation angle of each sample Let be the phase sequence deviation angle of the (i+1)th sample; This represents the average number of normally operating samples within the same time period as the abnormal phase sequence parameter samples under the same load conditions. This represents the detection time sequence number or device runtime sequence number corresponding to the i-th sample. This represents the detection time sequence number or device runtime sequence number corresponding to the (i-1)th sample. This is the detection time sequence number or device runtime sequence number corresponding to the (i+1)th sample.

[0072] The mutation adjustment factor of the normal index corresponding to the i-th phase sequence parameter sample can be used as the dynamic correction coefficient of the i-th sample. It is calculated based on the phase sequence deviation angle and time sequence number of its preceding and following adjacent samples, quantifying the degree of abrupt change in phase at that point and participating in the generation of the adjusted normal index. It is understandable that traditional detection methods lack a mutation identification mechanism based on neighborhood difference and time normalization, making it difficult to accurately distinguish between rapid disturbances and noise fluctuations. In an exemplary embodiment, the mutation adjustment factor of the normal index corresponding to the i-th phase sequence parameter sample can be obtained by receiving the phase sequence deviation angle and corresponding time sequence number of the i-1, i, and i+1 samples, as well as the average number of normally operating samples during abnormal periods under the same load conditions. By calculating the rate of change of the phase sequence deviation angle before and after, and normalizing it in conjunction with the time interval, and then introducing statistical adjustment parameters under load conditions, an adaptive weighting of the mutation rate is achieved, resulting in a dimensionless value that reflects the significance of the current sample's change in the local sequence. Furthermore, the mutation adjustment factor for the normal index corresponding to the i-th phase sequence parameter sample can be, but is not limited to, one or more of the following: gradient normalization based on central difference, or an enhanced version introducing variance normalization within a sliding window. The rate of change of phase sequence deviation angle can be used to characterize the speed of change of phase sequence deviation angle between adjacent sampling points, and to depict the dynamic evolution trend of phase sequence state.

[0073] Understandably, existing technologies only focus on whether the current value exceeds the limit, ignoring the rate of change information, resulting in a delayed response to rapid degradation. In this embodiment, the phase sequence deviation angle change rate can be obtained by receiving the phase sequence deviation angles of the (i-1)th and (i+1)th samples, approximating the second-order change characteristics using the central difference method, amplifying the response weight of the sudden shift, and obtaining the difference between the two, i.e., the amount of phase sequence deviation change across the i-th sample. For example, the phase sequence deviation angle change rate can be obtained by direct difference calculation, or by weighted difference to reduce the influence of noise, or one or more of these methods.

[0074] The time sequence number difference ratio can be used to normalize the phase sequence deviation angle change rate in the time dimension, eliminating the risk of misjudgment caused by uneven sampling intervals. It is understandable that traditional methods assume a fixed sampling period, which can easily lead to misjudgments when there are actual timing jitter or non-uniform acquisition. In a specific embodiment, the time sequence number difference ratio can be the detection time sequence number or the device runtime sequence number corresponding to the (i+1)th and (i-1)th samples. By dividing the angle change by the total time span, the equivalent change rate per unit time is obtained, improving the comparability of data under different operating conditions. The difference between the two is used as the denominator for time normalization. Furthermore, the time sequence number difference ratio can include one or more of the following: absolute timestamp difference or integer interval difference of runtime sequence number. The average number of normally operating samples during abnormal periods can be used to reflect the stability level of the system's historical operation under specific load conditions, serving as an adjustment parameter in the mutation adjustment factor to achieve adaptive sensitivity. It is understandable that existing technologies lack a mechanism for dynamically adjusting detection sensitivity based on historical operating characteristics, and a uniform threshold is difficult to accommodate different load scenarios.

[0075] In this embodiment, the average number of normally operating samples during abnormal periods can be a statistical count of samples that remained normal during past abnormal events under the same load level. A larger value indicates overall stability of the load segment and allows for higher tolerance; a smaller value indicates susceptibility to environmental disturbances and a need to improve response sensitivity. A statistical mean is obtained to adjust the gain strength of the mutation adjustment factor. For example, the average number of normally operating samples during abnormal periods can include, but is not limited to, one or more methods such as statistical counting based on historical databases according to load level, or continuously updating the mean through online learning.

[0076] For example, in a scenario where a power protection cabinet experiences a voltage sag during a low-load period at night, causing a phase jump, the phase sequence detection and protection circuit for the power protection cabinet in this embodiment can be operating under light load conditions, with a low sampling frequency and irregular time intervals. A voltage sag causes a sudden increase in the phase sequence deviation angle of the i-th sample, but consecutive offsets are also recorded for the preceding and following samples (i-1 and i+1). The mutation adjustment factor calculation module obtains the phase sequence deviation angles of these three samples and their corresponding detection time numbers, calculates a large rate of change of phase sequence deviation angle, and normalizes it using the time number difference ratio to avoid rate misestimation due to sparse nighttime sampling. Simultaneously, it queries the average number of normally operating samples during historical abnormal periods under this light-load condition. Due to increased interference and poor stability at night, this value is small, therefore the mutation adjustment factor outputs a high gain coefficient. This coefficient, multiplied by the original normal index, significantly improves the adjusted normal index, causing the normal index sequence generated by the sequence construction unit to rapidly deviate from the baseline trend. Based on this, the protection triggering unit calculates low correlation and an abnormal index that rises rapidly and exceeds the threshold, triggering an early warning in a timely manner to remind dispatchers to check the power grid quality, without missing reports due to uneven sampling or special loads.

[0077] This embodiment uses a mutation adjustment factor that satisfies the relationship between the phase sequence deviation angle of the i-th sample and its adjacent samples, the corresponding time index, and the average number of normally operating samples during an abnormal period under the same load conditions. The change rate of the phase sequence deviation angle reflects the dynamic trend of phase change between adjacent samples, and the change rate is normalized in the time dimension by the time index difference ratio to eliminate the influence of non-uniform sampling. Simultaneously, the average number of normally operating samples during an abnormal period is used to achieve condition-adaptive adjustment of the mutation response sensitivity. This achieves the effect of the mutation adjustment factor defined in the new scheme, which, by introducing the phase sequence deviation angle of adjacent samples and their corresponding time index, constructs a dynamic correction mechanism based on local change rate and time interval normalization. This factor not only reflects the instantaneous change trend of the current sample in phase deviation but also performs time normalization processing on the change rate by the time index difference, avoiding misjudgment due to uneven sampling or time jumps. Specifically, the formula introduces "the normal operating sample in the time period where the abnormal phase sequence parameter sample is located under the same load conditions..." Using the "average number of row samples" as an adjustment parameter, the mutation adjustment factor possesses adaptive capabilities under operating conditions. In long-term stable load segments, this value is larger, allowing the system to tolerate minor fluctuations more effectively and suppressing false triggers. Conversely, in load segments with frequent anomalies, this value is smaller, resulting in a more sensitive mutation response and improved early warning capabilities. This design effectively distinguishes between genuine fault mutations and accidental noise. By multiplying this mutation adjustment factor by the original normal index, an adjusted normal index is generated, enabling the stability index output by the acquisition and calculation unit to possess both statistical deviation and dynamic sensitivity. This mechanism significantly enhances the dual identification capability for gradual degradation and sudden disturbances, overcoming the fundamental shortcomings of background technologies that rely solely on static thresholds and lack modeling of abnormal evolution processes. Furthermore, it provides a more time-series-resolved data foundation for the sequence construction unit and enables the correlation analysis of the protection trigger unit to capture early abnormal trends, ultimately achieving a high-precision, low-latency, and interference-resistant graded protection response, comprehensively improving the robustness and intelligence of the phase sequence detection system.

[0078] In one embodiment, the anomaly index satisfies the following relationship:

[0079] In the formula, Let be the anomaly index of the i-th phase sequence parameter sample. Let i be the normal index sequence of the i-th sample. Let be the reference normal index sequence for the i-th sample, and PPMCC represent the Pearson correlation coefficient.

[0080] The anomaly index of the i-th phase sequence parameter sample is computationally dependent on the normal index sequence and the reference normal index sequence of the i-th sample. It can be used to characterize the overall deviation between the normal index sequence and the reference normal index sequence corresponding to the current power protection cabinet at the i-th sampling point, and to determine whether the phase sequence operation is abnormal. It is understood that traditional methods only compare the current value with a fixed threshold, failing to identify sequence-level trend deviations and easily missing progressive faults. In an exemplary embodiment, the anomaly index of the i-th phase sequence parameter sample can be obtained by receiving the normal index sequence of the i-th sample and the corresponding reference normal index sequence, calculating the Pearson correlation coefficient between the two sequences, and generating an anomaly index based on the degree of correlation decline. The lower the correlation, the higher the anomaly index, resulting in a quantitative indicator reflecting the level of trend consistency. Furthermore, the anomaly index of the i-th phase sequence parameter sample can be, but is not limited to, one or more of the following: 1 minus the absolute value of the correlation coefficient, or a standardized anomaly score based on z-score transformation.

[0081] The Pearson correlation coefficient (PPMCC), as a statistical metric, participates in the construction of anomaly indices. It can be used to assess the linear correlation between the current normal index sequence and a reference normal index sequence, supporting the generation of anomaly indices. Understandably, existing technologies do not incorporate statistical correlation analysis between sequences, lacking the ability to assess the consistency of overall behavioral patterns. In this embodiment, the Pearson correlation coefficient (PPMCC) can take two equal-length numerical sequences: the normal index sequence of the current device and the reference normal index sequence of historical qualified devices. The similarity in trend between the two sequences is calculated using the ratio of covariance to standard deviation. It is insensitive to amplitude differences but highly responsive to pattern matching, yielding a correlation coefficient value between -1 and 1, representing the consistency of the changing trends of the two sequences. For example, the Pearson correlation coefficient (PPMCC) can include, but is not limited to, one or more methods such as calculating local correlation based on a subset of data within a sliding window, or weighting time-nearest neighbor samples to enhance real-time performance.

[0082] For example, in a scenario where a power protection cabinet experiences slow insulation aging under heavy load conditions, leading to continuous phase drift, the phase sequence detection and protection circuit for the power protection cabinet in this embodiment can be used when the equipment operates under heavy load for a long time. Due to the gradual decline in the insulation performance of a certain phase winding, the phase difference exhibits a slow upward trend. The acquisition and calculation unit records time-series phase sequence parameter samples, generates a normal index, and constructs a normal index sequence categorized by heavy load. The sequence construction unit calls upon the reference normal index sequence of historically qualified equipment under the same heavy load conditions. Although the deviation of a single sample remains within the allowable range, the protection triggering unit uses PPMCC to calculate the correlation between the current sequence and the reference sequence. It finds that the trend shows a systematic upward movement while the reference sequence remains stable, resulting in a significant decrease in the correlation coefficient. The resulting abnormal index subsequently increases, triggering an early warning action when it exceeds a preset threshold. Maintenance personnel then arrange preventative maintenance, avoiding sudden failures and downtime, demonstrating a technological upgrade from instantaneous value judgment to trend consistency assessment.

[0083] In one embodiment, determining that the current phase sequence detection is abnormal and triggering a protection action further includes: calculating the quality evaluation of the power protection cabinet; when the quality evaluation of the power protection cabinet is less than a preset quality threshold, determining that the phase sequence of the power protection cabinet is abnormal and triggering a protection action; the quality evaluation satisfies the following relationship:

[0084]

[0085] In the formula, For the quality evaluation of the power protection cabinet corresponding to the i-th sample, Let i be the normal index sequence of the i-th sample. Let N be the normal index sequence of the nth qualified sample before the i-th sample, where N is the number of qualified samples before the i-th sample, and cos{} represents the cosine similarity.

[0086] The quality evaluation of the power protection cabinet corresponding to the i-th sample can be used to quantify the overall consistency level of the current phase sequence operation status of the power protection cabinet relative to its historical qualified status, serving as a dynamic health indicator for anomaly detection. It can be based on the normal index sequence of the i-th sample, the normal index sequence of the nth qualified sample preceding the i-th sample, and the cumulative number of qualified samples N. By calculating the cosine similarity between the current sequence and the most recent qualified benchmark sequence, and combining it with the historical qualified number for weighted normalization, a quality score evolving over time is formed, resulting in a dimensionless score value reflecting the current health trend of the equipment. It is understandable that traditional methods only compare the current value with a fixed threshold, lacking a mechanism for tracking and comparing the historical behavior patterns of individual equipment, making it difficult to reflect the cumulative characteristics of operational stability. In an exemplary embodiment, the quality evaluation of the power protection cabinet corresponding to the i-th sample can include, but is not limited to, combining similarity with N using a linear weighted form, or introducing an exponential decay factor to highlight the importance of recent qualified records, or one or more of the following:

[0087] Cosine similarity can be used to measure the trend consistency between the current normal index sequence and the historical qualified sequence in multiple dimensions, focusing on the degree of matching of change patterns while ignoring amplitude differences. Cosine similarity can take two equal-length numerical vectors: the current normal index sequence and the historical qualified baseline sequence. The alignment of the two sequences in direction is calculated using the cosine of the angle between the vectors. This is suitable for pattern recognition in high dimensions or under multiple operating conditions, yielding a similarity metric between -1 and 1, with values ​​closer to 1 indicating greater trend consistency. It is understandable that existing technologies do not utilize vector space methods to assess the evolution trend of phase sequence stability, and therefore cannot effectively capture structural deviations in operating modes. Furthermore, cosine similarity can include, but is not limited to, direct calculation based on the standard cosine formula, or using the average similarity of piecewise vectors within a sliding window to enhance robustness, among other methods. The normal index sequence of the nth qualified sample before the i-th sample can be used as a reference baseline for the current equipment, representing the phase sequence stability mode of the power supply protection cabinet most recently confirmed as operating normally. This can be achieved by receiving normal index sequences recorded from historical qualified moments of the same device, and constructing a personalized health reference template by selecting the nth verified qualified normal index sequence closest to the current moment, thus obtaining a benchmark sequence for comparison with the current sequence. It is understood that traditional methods rely on the group mean across devices as the sole benchmark, ignoring the differences and evolutionary trajectories of the normal states of individual devices. In this embodiment, the normal index sequence of the nth qualified sample before the i-th sample can include, but is not limited to, setting n=1 to take the nearest qualified point, or n>1 to take the cluster center of multiple historical qualified points as a robust benchmark, or one or more of the following:

[0088] The number N of qualified samples before the i-th sample can be used to reflect the total number of samples that have been consistently considered normal during the historical operation of the power protection cabinet, and is used to adjust the confidence weight of the quality evaluation. It can be the number of all sampling points that meet the qualification standards from the time the equipment was put into operation until the i-th sample. A larger value indicates long-term stable operation of the equipment, giving the current quality evaluation higher credibility and suppressing misjudgments caused by occasional fluctuations. This is obtained as a positive integer and participates in the normalization weighting process of the quality evaluation. It is understandable that existing technologies do not use historical qualification frequency as an evaluation parameter, leading to the use of the same criteria for frequently fluctuating equipment and stable equipment, lacking the ability to distinguish between them. For example, the number N of qualified samples before the i-th sample can be, but is not limited to, one or more methods such as statistics based on absolute counting or weighted calculation by time window to emphasize recent performance.

[0089] For example, in a scenario where a power protection cabinet that has been operating stably for a long time experiences intermittent phase jitter during recovery after an external disturbance, the phase sequence detection and protection circuit for the power protection cabinet in this embodiment can be based on the fact that the device has previously continuously recorded a large number of qualified samples, resulting in a high N value. After a power grid disturbance, the system experiences a brief anomaly and then enters the recovery phase. The acquisition and calculation unit generates a new normal index sequence, and the sequence construction unit compares it with the normal index sequence of the most recent nth (n=1) qualified sample. Although the current sequence does not significantly exceed the standard, the cosine similarity calculation shows that its trend deviates directionally from the historical good state, manifested as a lag in response at some load switching points. This similarity value is low, and because N is large, the weighting effect of the quality evaluation formula amplifies the impact of this deviation, causing the overall quality evaluation to drop rapidly. When it falls below the preset quality threshold, the protection trigger unit initiates an early warning action. Based on this, maintenance personnel discovered that the aging of the contactor caused the response delay, and timely replacement prevented subsequent fault recurrence. This example demonstrates the advantages of detection logic based on comparison of individual historical behavior in identifying atypical anomalies.

[0090] This embodiment calculates the quality evaluation of the power protection cabinet. When the quality evaluation is less than a preset quality threshold, it determines that the phase sequence is abnormal and triggers protection action. The consistency level between the current operating state and the historical qualified state is quantified by the quality evaluation of the power protection cabinet corresponding to the i-th sample. It receives the current normal index sequence, the historical qualified benchmark sequence, and the total number of qualified samples N. The trend matching degree is calculated using cosine similarity, and a weighted normalization is performed based on N to output a dimensionless score. Cosine similarity focuses on the directional consistency of the sequence change pattern rather than the amplitude difference, improving the sensitivity to structural deviations. The n-th sample before the i-th sample... By establishing a normal index sequence of qualified samples, an individualized health reference template for the equipment is created, avoiding misjudgment bias caused by relying on the group mean. The number N of qualified samples before the i-th sample reflects the long-term stable characteristics of the equipment and enhances the evaluation confidence, suppressing the impact of occasional interference. This can achieve the construction of dynamic health indicators based on the historical behavior evolution trajectory of the equipment itself, realizing the transformation from instantaneous state judgment to historical pattern comparison, improving the accuracy of anomaly detection and early warning capability, effectively reducing missed judgments and false alarms caused by static threshold criteria and ignoring individual differences, and ultimately forming a more robust and intelligent hierarchical protection response technology.

[0091] In one embodiment, the process of obtaining the normality index is as follows:

[0092] Calculate the mean and standard deviation of the phase sequence deviation angle for all collected phase sequence parameter samples, and use the ratio of the difference between the phase sequence deviation angle and the mean to the standard deviation for each sample as the normal index of the phase sequence deviation angle for the corresponding sample.

[0093] The mean of the phase sequence deviation angle can be used to reflect the central tendency of the phase sequence deviation angles across all collected samples, serving as a central reference value in the calculation of the normality index. It is understood that while traditional methods can calculate the average value, they do not use it to construct a relative stability index, relying only on comparisons between instantaneous values ​​and fixed thresholds. In this embodiment, the mean of the phase sequence deviation angle can be the overall average deviation angle obtained by arithmetic averaging the phase sequence deviation angles corresponding to all m phase sequence parameter samples. Furthermore, the mean of the phase sequence deviation angle can be, but is not limited to, one or more of the following: a simple arithmetic mean, or a weighted average adjusted according to sample time weights.

[0094] The standard deviation of the phase sequence deviation angle can be used to quantify the dispersion of the phase sequence deviation angle across multiple samples, reflecting the fluctuation level of phase sequence stability, and is used for normalization. It is understood that existing technologies do not utilize statistical discrete characteristics to assess phase sequence stability, making it difficult to distinguish between scenarios with high fluctuation and small deviation and low fluctuation and large deviation. In an exemplary embodiment, the standard deviation of the phase sequence deviation angle can be obtained by receiving a sequence of squared differences between each phase sequence deviation angle and its mean, and calculating the square root of the variance to obtain the standard deviation value representing the intensity of deviation fluctuation. For example, the standard deviation of the phase sequence deviation angle can be implemented using a formula based on the overall standard deviation, or by combining a sliding window-based dynamically updated standard deviation calculation method, or one or more other methods.

[0095] The difference (the difference between the phase sequence deviation angle and the mean) can be used to represent the deviation of the phase sequence deviation angle of a single sample from the overall average level, and is an intermediate variable in the calculation of the normality index. Understandably, in traditional detection, such differences are usually discarded directly or used only for instantaneous judgment, and are not included in the systematic statistical analysis process. In this embodiment, the difference can be obtained by receiving the phase sequence deviation angle of the i-th sample and the mean of all samples, and then subtracting them to obtain the absolute deviation between the sample and the mean. Furthermore, the difference can include one or more of the following: retaining it as a signed difference to reflect the direction of deviation, or taking the absolute value for undirected analysis.

[0096] The ratio (the ratio of the difference to the standard deviation) can be used to normalize and compare deviations with the level of population fluctuation, forming a dimensionless normality index that characterizes the degree of relative anomaly. Understandably, traditional schemes lack this type of normalization mechanism based on distribution characteristics, leading to problems such as high sensitivity under light loads and sluggishness under heavy loads. In a specific embodiment, this ratio can be the difference between the received phase sequence deviation angle and the mean, and the standard deviation of the phase sequence deviation angle. A division operation is used to achieve scale uniformity, making deviations under different operating conditions comparable, thus obtaining a standardized normality index (in Z-score form). For example, this ratio can be, but is not limited to, used directly as a Z-score, or output after processing by a nonlinear compression function, or one or more of these methods.

[0097] Standard normalization transformation can be used to convert the original phase sequence deviation angle into a relative index that follows a standard normal distribution, eliminating the influence of dimensions and improving the adaptability of anomaly identification. It is understood that the background technology does not incorporate statistical standardization methods, making it unable to dynamically adapt to differences in fluctuation characteristics under different loads. In this embodiment, standard normalization transformation can involve receiving the original phase sequence deviation angle sequence, using the mean and standard deviation to perform a linear transformation on the original data, giving it zero mean and unit variance characteristics, thereby outputting a normalized normal index sequence. Furthermore, standard normalization transformation can include one or more of the following: employing the classic Z-score formula structure, or combining an enhanced normalization algorithm with skewness correction.

[0098] For example, in a scenario where a power protection cabinet experiences brief voltage interference causing phase jitter during light-load operation at night, the phase sequence detection and protection circuit for the power protection cabinet in this embodiment can collect multiple sets of phase sequence parameter samples. The calculated mean of the phase sequence deviation angle is low, but due to interference, the deviation of individual samples suddenly increases, resulting in a small standard deviation and relatively significant fluctuations. By dividing the difference between the deviation and the mean of each sample by the standard deviation, a normalized normal index is obtained. Although the absolute deviation does not reach the traditional threshold, the normal index is significantly higher because it deviates significantly from the group mean and has a small standard deviation. The sequence construction unit generates an index sequence under light-load conditions based on this and compares it with the reference sequence of historical qualified equipment. The correlation calculated by the protection triggering unit decreases, the abnormal index exceeds the warning threshold, triggering an alarm prompt instead of immediate tripping. This achieves sensitive response and reasonable handling of minor disturbances, avoiding unnecessary shutdowns.

[0099] In one embodiment, the quality evaluation also satisfies the following relationship:

[0100]

[0101] In the formula, For the quality evaluation of the power protection cabinet corresponding to the i-th sample, Let i be the normal index sequence of the i-th sample. Let be the normal index sequence of the nth qualified sample preceding the i-th sample, where N is the number of qualified samples preceding the i-th sample, and cos{} denotes the cosine similarity. Let be the weight of the nth qualified sample relative to the ith sample.

[0102] The weight of the nth qualified sample relative to the ith sample is used to adjust the relative importance of historical qualified samples in the quality evaluation calculation. This allows the contribution of cosine similarity to be dynamically weighted based on time, operating conditions, or state distance, improving the timeliness and matching accuracy of the evaluation. It is understandable that traditional methods, when utilizing historical data, do not distinguish the importance of samples at different time points, treating all qualified records equally, leading to outdated or mismatched benchmarks interfering with current judgments. In an exemplary embodiment, the weight of the nth qualified sample relative to the ith sample can be determined by receiving the time interval between the nth qualified sample and the current ith sample, the difference in load level, or the degree of change in equipment operating status. The time or state distance between samples is converted into a weight value based on a preset function (such as exponential decay, piecewise linear, or nonlinear mapping). The greater the distance or the greater the difference, the lower the weight, resulting in a dimensionless coefficient between 0 and 1, representing the influence strength of the historical sample in the current evaluation. For example, the weight of the nth qualified sample relative to the ith sample can include, but is not limited to, one or more of the following: an exponential decay function based on time difference, or a conditional weight allocation mechanism set according to the load level matching degree.

[0103] For example, in a scenario where a power protection cabinet re-enters a stable operating phase after experiencing multiple intermittent faults, the phase sequence detection and protection circuit for the power protection cabinet in this embodiment could be: the equipment previously triggered abnormal judgments multiple times due to external interference, but recently resumed normal operation. The sequence construction unit records multiple historical qualified samples, some of which occurred several days ago under different load conditions. When the i-th new sample arrives, the quality evaluation module calculates the cosine similarity between its normal index sequence and each historical qualified sequence, and introduces the weight of the n-th qualified sample to the current sample: qualified samples that have recently recovered under the same load conditions are given higher weights, while the weights of qualified sequences from earlier periods or under heavy loads are significantly reduced. Although the overall average similarity is not high, the strong match of high-weighted samples dominates the quality score, causing the quality evaluation to recover rapidly. Since this value is still higher than the preset threshold, the protection action is not triggered, and the system determines that the equipment has effectively recovered. This mechanism avoids continuous false alarms caused by relying on outdated or mismatched benchmarks, demonstrating the adaptability advantage of weighted evaluation in dynamic environments.

[0104] This embodiment introduces the weight of the nth qualified sample to the i-th sample in quality assessment. It weights and aggregates the cosine similarity between the normal index sequence of the i-th sample and the normal index sequences of the nth qualified samples preceding the i-th sample, and normalizes this by combining N (the number of qualified samples preceding the i-th sample). The weight of the nth qualified sample to the i-th sample models the receiving time interval, load level differences, or changes in equipment operating status. An exponential decay or conditional weighting mechanism generates dimensionless coefficients in the range of 0 to 1, dynamically adjusting the influence of each historical qualified sample in the current quality assessment. This ensures that qualified samples under similar time and operating conditions receive higher attention, thus prioritizing more representative historical benchmarks during equipment recovery. This mechanism enables quality assessment to utilize historical data differentially, quickly identifying trends towards a healthy state after brief equipment disturbances. If a qualified state has a high degree of matching with the current state and is given a high weight, the score will quickly rebound to confirm the effectiveness of the recovery and avoid long-term misjudgment. Conversely, if the historical qualified sample is too far in the past or the operating conditions deviate significantly from the current state, its weight will automatically decrease to prevent outdated benchmarks from misleading the evaluation conclusion. Combining the statistical coverage of the number of qualified samples N, the overall score takes into account both the breadth of historical data and the depth of focus on key samples, realizing the technical evolution from static equal weight comparison to dynamic credibility aggregation. This makes up for the shortcomings of the original method, which did not distinguish the use of historical qualified data and lacked an importance classification mechanism, and enhances the adaptability to complex operating paths of equipment. Finally, the weighted quality evaluation result is compared with the preset threshold to determine whether to trigger the protection action. This allows the system to not only detect the current anomaly, but also to determine whether it has truly recovered to a stable state, thereby effectively reducing the false alarm rate and the risk of missed alarms. While ensuring the safe operation of the power protection cabinet, it also improves the availability and economy of the system.

[0105] Furthermore, to achieve the above objectives, the present invention also proposes a phase sequence detection and protection device for a power protection cabinet, wherein the phase sequence detection and protection device for a power protection cabinet includes the phase sequence detection and protection circuit for a power protection cabinet described above. The specific structure of this phase sequence detection and protection circuit for a power protection cabinet is as described in the above embodiments. Since this phase sequence detection and protection device for a power protection cabinet adopts all the technical solutions of all the above embodiments, it possesses at least all the beneficial effects brought about by the technical solutions of the above embodiments, which will not be elaborated upon here.

[0106] Furthermore, to achieve the above objectives, the present invention also proposes a power protection cabinet, which includes the phase sequence detection and protection circuit for the power protection cabinet described above. The specific structure of this phase sequence detection and protection circuit for the power protection cabinet is as described in the above embodiments. Since this power protection cabinet adopts all the technical solutions of all the above embodiments, it possesses at least all the beneficial effects brought about by the technical solutions of the above embodiments, which will not be elaborated upon here.

[0107] The above are merely preferred embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A phase sequence detection and protection circuit for a power supply protection cabinet, characterized in that, The circuit includes: an acquisition and calculation unit, a sequence construction unit, and a protection triggering unit; The acquisition and calculation unit is connected to the sequence construction unit, the sequence construction unit is connected to the protection triggering unit, and the protection triggering unit is connected to the power protection cabinet. The acquisition and calculation unit is used to acquire a set number of phase sequence parameter samples of the three-phase lines in the power protection cabinet and obtain their phase sequence deviation angles, and calculate the normal index of each phase sequence deviation angle. The normal index is obtained by comparing the difference between a single phase sequence deviation angle and the overall average deviation angle of all samples. The phase sequence deviation angle is the sum of the absolute values ​​of the deviations between the actual phase difference and the standard phase difference of phases A, B, and C in the three-phase voltage. The normal index is obtained by comparing the difference between a single phase sequence deviation angle and the overall average deviation angle of all samples, and is used to characterize the degree to which the phase sequence stability of a single sample deviates from the group level. The sequence construction unit is used to set multiple load levels for the power protection cabinet, sort the normal index of the power protection cabinet under each load level according to the load level, and construct a normal index sequence for the power protection cabinet; obtain the average normal index of all qualified power protection cabinets in history under all set load levels, and construct a reference normal index sequence. The protection triggering unit is used to calculate the abnormal index of the power protection cabinet based on the correlation between the normal index sequence of the power protection cabinet and the reference normal index sequence; when the abnormal index of the power protection cabinet is greater than the preset abnormal threshold, the current phase sequence detection is determined to be abnormal and the protection action is triggered.

2. The phase sequence detection and protection circuit for a power supply protection cabinet as described in claim 1, characterized in that, The process of collecting a set number of phase sequence parameter samples of the three-phase lines in the power protection cabinet and obtaining their phase sequence deviation angle also includes obtaining the detection time sequence number or equipment running time sequence number corresponding to the collected phase sequence parameter samples. The normality index satisfies the following relationship: ； In the formula, Let be the normal index of the i-th phase sequence parameter sample. The phase sequence deviation angle for the first sample. Let be the phase sequence deviation angle of the i-th sample. Let be the phase sequence deviation angle of the m-th sample, and mean() represents the average value.

3. The phase sequence detection and protection circuit for a power supply protection cabinet as described in claim 2, characterized in that, The normal index also includes a mutation adjustment factor for calculating the normal index corresponding to the phase sequence parameter sample, and the product of the mutation adjustment factor and the normal index is used as the normal index after adjustment of the phase sequence parameter sample.

4. The phase sequence detection and protection circuit for a power supply protection cabinet as described in claim 3, characterized in that, The mutation adjustment factor satisfies the following relationship: ； In the formula, Let be the mutation adjustment factor for the normal index corresponding to the i-th phase sequence parameter sample. Let be the phase sequence deviation angle of the i-th sample. For the first Phase sequence deviation angle of each sample Let be the phase sequence deviation angle of the (i+1)th sample; This represents the average number of normally operating samples within the same time period as the abnormal phase sequence parameter samples under the same load conditions. This represents the detection time sequence number or device runtime sequence number corresponding to the i-th sample. This represents the detection time sequence number or device runtime sequence number corresponding to the (i-1)th sample. This is the detection time sequence number or device runtime sequence number corresponding to the (i+1)th sample.

5. The phase sequence detection and protection circuit for a power supply protection cabinet as described in claim 4, characterized in that, The anomaly index satisfies the following relationship: ； In the formula, Let be the anomaly index of the i-th phase sequence parameter sample. Let i be the normal index sequence of the i-th sample. Let be the reference normal index sequence for the i-th sample, and PPMCC represent the Pearson correlation coefficient.

6. The phase sequence detection and protection circuit for a power supply protection cabinet as described in claim 5, characterized in that, The determination that the current phase sequence detection is abnormal and the triggering of protection actions also includes: The quality evaluation of the power protection cabinet is calculated. When the quality evaluation of the power protection cabinet is less than a preset quality threshold, the phase sequence of the power protection cabinet is determined to be abnormal and a protection action is triggered. The quality evaluation satisfies the following relationship: ； In the formula, For the quality evaluation of the power protection cabinet corresponding to the i-th sample, Let i be the normal index sequence of the i-th sample. Let N be the normal index sequence of the nth qualified sample before the i-th sample, where N is the number of qualified samples before the i-th sample, and cos{} represents the cosine similarity.

7. The phase sequence detection and protection circuit for a power supply protection cabinet as described in claim 2, characterized in that, The process of obtaining the normal index is as follows: Calculate the mean and standard deviation of the phase sequence deviation angle for all collected phase sequence parameter samples, and use the ratio of the difference between the phase sequence deviation angle and the mean to the standard deviation for each sample as the normal index of the phase sequence deviation angle for the corresponding sample.

8. The phase sequence detection and protection circuit for a power supply protection cabinet as described in claim 6, characterized in that, The quality evaluation also satisfies the following relationship: ； In the formula, For the quality evaluation of the power protection cabinet corresponding to the i-th sample, Let i be the normal index sequence of the i-th sample. Let be the normal index sequence of the nth qualified sample preceding the i-th sample, where N is the number of qualified samples preceding the i-th sample, and cos{} denotes the cosine similarity. Let be the weight of the nth qualified sample relative to the ith sample.

9. A phase sequence detection and protection device for a power supply protection cabinet, characterized in that, The phase sequence detection and protection device for the power protection cabinet includes the phase sequence detection and protection circuit for the power protection cabinet as described in any one of claims 1 to 8.

10. A power protection cabinet, characterized in that, The power protection cabinet includes the phase sequence detection and protection circuit for the power protection cabinet as described in any one of claims 1 to 8.

Images