Power distribution network residual current protection method and system based on distributed photovoltaic power supply

Abstract

The invention relates to the technical field of residual current protection, in particular to a power distribution network residual current protection method and system based on a distributed photovoltaic power supply, and the method comprises the following steps: obtaining a zero-sequence current waveform in a disturbance period, recognizing an abrupt change node, analyzing the fluctuation rhythm and trend, and constructing a response structure; the method comprises the following steps: extracting a sequence relationship between abrupt change nodes and fluctuation rhythms, constructing a response fragment set with trend consistency and rhythm coherence, screening response fragments capable of triggering protection, adjusting a starting point and a sequence, completing action triggering rhythm merging, and generating a residual current protection strategy, in the method, through sequence relationship extraction between the abrupt change nodes and the fluctuation rhythms, constructing a response fragment set with trend consistency and rhythm coherence, and combining a change cooperation relationship between sections; key paragraphs capable of triggering protection are screened out, the triggering sequence and the starting point position are adjusted according to trend continuity, the starting time sequence matching degree is improved, behavior stability is judged through trend coherence, rhythmic merging of response structures is achieved, action strategies with directivity and continuity are output, and the precise control capacity of protection response under variable disturbance is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of residual current protection, and in particular to a residual current protection method and system for a distribution network based on distributed photovoltaic power sources. Background Art

[0002] The field of residual current protection technology involves the protection mechanism used in power systems to detect and cut off residual currents generated by insulation faults or equipment abnormalities. Its core issues include the zero-sequence current detection principle, the setting of residual current action criteria, the triggering mechanism of the protection device, etc. It is an important technical means to ensure the safe operation of the power system and prevent electrical fires and personal electric shock accidents. It is widely used in urban and rural power distribution systems and new energy access scenarios. As photovoltaic power generation is an important component of clean and renewable energy, it plays an increasingly critical role in the transformation of the energy structure. However, the output of photovoltaic power generation is volatile and uncertain. Its widespread access has put higher demands on the operation, management and protection of the distribution network. Especially in distributed access scenarios, traditional protection methods face problems such as insufficient adaptability. Among them, the traditional residual current protection method for distribution networks based on distributed photovoltaic power sources means that after the photovoltaic system is connected, in order to ensure system safety and stable operation, a fixed-value setting strategy based on zero-sequence current detection is generally adopted to realize residual current fault detection. The three-phase current vector and changes are often monitored in real time through a zero-sequence current transformer. Combined with the preset action threshold, the fault is cleared through a relay or electronic switch control unit. This solution constructs protection logic according to standards such as IEC61008 or GB6829. The action judgment generally adopts a static setting method, which does not fully consider the operating condition changes brought about by the dynamic characteristics of photovoltaic power generation.

[0003] Since the criterion setting is based on a fixed threshold and depends on the instantaneous amplitude change of the current, the evolution trend during the disturbance process is not dynamically modeled. As a result, it is difficult to identify the evolution path of the current behavior in scenarios where continuous disturbances overlap with multiple current fluctuations, and there is a lack of the ability to identify and reconstruct the stage structure of current changes. For example, in the continuous current disturbance caused by short-term flicker after photovoltaic access, it is difficult for the protection device to distinguish between normal fluctuations and potential faults based on fixed criteria, which can easily lead to response timing misalignment or trigger priority imbalance, resulting in delayed or false triggering of protection actions. Under the conditions of multiple source access and frequent load changes, the response accuracy and operational reliability of the distribution network protection system are affected. Summary of the Invention

[0004] In order to solve the technical problems existing in the prior art, an embodiment of the present invention provides a residual current protection method for a distribution network based on a distributed photovoltaic power supply, comprising the following steps: In order to achieve the above object, the present invention adopts the following technical solution: a residual current protection method for a distribution network based on a distributed photovoltaic power supply, comprising the following steps: S1: Extract the zero-sequence current waveform within the disturbance period of the distributed photovoltaic access point, calibrate the starting segment, middle segment and ending segment of the change, identify the evolution mode based on the change trend, inflection point conversion and rhythm difference of the three segments, and generate a profile mutation marker sequence; S2: Based on the profile mutation marker sequence, identify the fluctuation pattern of the disturbance response, analyze the rhythm, persistence and trend of the initial, middle and stabilization periods, construct the response evolution structure, and classify it into fluctuation structure segment groups; S3: combining the fluctuation structure segment group and the profile mutation mark sequence, establishing time series associations pair by pair, judging the order and trend consistency, sorting out the relationship between response combinations, screening the response segments that trigger the protection condition threshold, and generating an action trigger segment list; S4: Based on the action trigger segment list, locate the starting point of the response segment, trace the trend of the previous segment, and if consistent, move the starting point forward, re-arrange the order of the response segments, adjust the start priority, and generate an action start position list; S5: Based on the action start position list, determine the stability of the response segment, sort the action response priority, structure and integrate the execution rhythm, and generate a distribution network residual current protection plan.

[0005] As a further solution of the present invention, the profile mutation mark sequence includes a starting change node, a mid-segment fluctuation feature point and an ending trend node; the fluctuation structure segment group includes a disturbance morphology partition, a curve trend segment and a stage-by-stage fluctuation feature; the action trigger segment list includes a trigger starting position, a trend consistency segment and a response collaborative combination; the action start position list includes start sorting information, a trend continuation indicator and a response segment adjustment relationship; the distribution network residual current protection scheme includes a response priority level, an action rhythm structure and a relay output rule.

[0006] As a further solution of the present invention, the definition of the protection condition threshold refers to when the characteristic change of the zero-sequence current waveform reaches the minimum judgment standard required to trigger the protection action in the disturbance response.

[0007] As a further solution of the present invention, the definition of tracing back the previous trend refers to analyzing the fluctuation trend forward after the starting point of the action response segment. If it is consistent with the current trend, the process of moving the response starting point forward is called tracing back the previous trend.

[0008] As a further solution of the present invention, the specific steps of S1 are: S101: Obtain the zero-sequence current waveform within the disturbance period of the distributed photovoltaic power access point, extract the change amplitude, slope value, and first-order difference value within each period, construct a difference index set based on the three parameters, determine the start and end positions of the segments in the waveform where the change rate exceeds the zero-sequence current disturbance threshold, and generate the current change interval boundary pair; S102: Calling the current change interval boundary pair, calculating the first-order and second-order derivatives of the current waveform within the interval, determining the trend inflection point based on the derivative sign and the extreme value position, and numbering the fluctuation starting point, inflection point, and end point in sampling order to generate a current fluctuation feature point sequence; S103: Based on the current fluctuation feature point sequence, identify the arrangement direction of peaks and troughs and the rhythm ratio of adjacent segments, extract the time index and amplitude change of the direction mutation point, construct a segment connection relationship matrix, and generate a contour mutation mark sequence.

[0009] As a further solution of the present invention, the specific steps of S2 are: S201: calling the profile mutation marker sequence to detect segments in the zero-sequence current whose fluctuation amplitude exceeds the disturbance response identification threshold, extracting segments in the initial, middle and stabilization stages of the disturbance based on the continuous time and amplitude variation, and obtaining a disturbance stage fluctuation segment set; S202: Based on the disturbance phase fluctuation segment set, extract the duration, fluctuation frequency and rhythm period of the phase segments, calculate the stability coefficient, and determine the continuity state to obtain a phase fluctuation characteristic index group; S203: Calculate the stability difference value between segments based on the stage fluctuation characteristic index group, classify the segments with similar characteristics according to the trend change direction, establish a structure mapping relationship, and generate a fluctuation structure segment group.

[0010] As a further solution of the present invention, the specific steps of S3 are: S301: Calling the fluctuating structure segment group and the contour mutation marker sequence, establishing matches between the structure segments and the mutation points one by one according to the sampling time, determining the time sequence of the starting point and the mutation point of each group of structure segments, screening out segments with consistent time arrangement, and generating a structure mutation matching pair sequence; S302: comparing the direction vectors of the fluctuations before and after the structural segment and the mutation point based on the structural mutation matching pair sequence to determine whether the trend directions are consistent, and marking combinations with consistent trends to obtain a trend consistent combination sequence; S303: Based on the trend-consistent combination sequence, extract the amplitude change and duration of the structural segments in the combination, compare them with the protection condition threshold, screen the response segments that meet the trigger condition, and generate an action trigger segment list.

[0011] As a further solution of the present invention, the specific steps of S4 are: S401: calling the action trigger segment list, locating the starting point of the response segment, tracing back the current direction of the previous change segment, and judging whether it is consistent with the trend of the current segment. If it is consistent, extending the starting point of the response segment forward to generate a trend extension starting point set; S402: extracting the sampling time of the extended response segments based on the trend extension starting point set, sorting them by the starting point time, determining whether the time sequence is misplaced, and unifying the order to obtain a response segment time sequence table; S403: Based on the response segment time sequence table, extract the trend direction, duration and amplitude change, calculate the trend continuation score, adjust the segment priority according to the score, and generate an action start position list.

[0012] As a further solution of the present invention, the specific steps of S5 are: S501: Calling the action start position list, extracting the start sequence and trend continuity parameters of the response segments, judging the trend direction and fluctuation stability of each segment start position, calculating the trend duration and amplitude change, establishing a corresponding stability score matrix, and generating a response segment stability score group; S502: Based on the response segment stability score group and the temporal connection relationship between the action segments, the start and end point index differences between adjacent segments are extracted to determine whether the connection state is continuous, and the response sequence is rearranged based on continuity priority to obtain a response segment priority sequence list; S503: Based on the response segment priority list, the action rhythm change cycle and the response duration are integrated, the trigger segment position and the action rhythm structure are matched, a timing output set with a trigger identifier is formed, and a distribution network residual current protection scheme is generated.

[0013] The residual current protection system for distribution network based on distributed photovoltaic power generation includes: The contour marker extraction module is used to implement S1: extract the zero-sequence current waveform within the disturbance period of the distributed photovoltaic access point, calibrate the starting segment, middle segment, and ending segment of the change, and generate a contour mutation marker sequence based on the change trend, inflection point conversion, and rhythm difference identification evolution mode of the three segments; The fluctuation structure classification module is used to implement S2: based on the profile mutation mark sequence, identify the fluctuation form of the disturbance response, analyze the rhythm, persistence and trend of the initial, middle and stabilization periods, construct the response evolution structure, and classify it into fluctuation structure segment groups; The response segment screening module is used to implement S3: combining the fluctuation structure segment group with the profile mutation mark sequence, establishing time series associations for each pair, judging the order and trend consistency, sorting out the relationship between response combinations, screening response segments that trigger the protection condition threshold, and generating an action trigger segment list; The startup priority adjustment module is used to implement S4: based on the action trigger segment list, locate the starting point of the response segment, trace the trend of the previous segment, move the starting point forward if consistent, re-arrange the order of the response segments, and adjust the startup priority to generate an action startup position list; The trigger rhythm merging module is used to implement S5: based on the action start position list, judge the stability of the response segment, sort the action response priority, structure and integrate the execution rhythm, and generate the distribution network residual current protection plan.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, the sequential relationship between mutation nodes and fluctuation rhythms is extracted to construct a set of response fragments with trend consistency and rhythmic coherence. Combined with the synergistic relationship of changes between segments, the key sections that can trigger protection are screened, and the trigger sequence and starting point position are adjusted according to the trend continuity to improve the matching degree of the startup timing. The behavioral stability is judged by the trend connectivity to achieve the rhythmic merging of the response structure, output an action strategy with directionality and continuity, and enhance the precise control capability of the protection response under variable disturbances. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0016] Figure 1 Schematic diagram of the steps of the present invention; Figure 2 This is a schematic diagram of the refinement of S1 of the present invention; Figure 3 This is a schematic diagram of the refinement of S2 of the present invention; Figure 4 This is a schematic diagram of the refinement of S3 of the present invention; Figure 5 This is a schematic diagram of the refinement of S4 of the present invention; Figure 6 This is a schematic diagram of the refinement of S5 of the present invention; Figure 7 It is a system module diagram of the present invention. DETAILED DESCRIPTION

[0017] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0018] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0019] In the embodiments of the present invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same. The terms "of," "corresponding," and "corresponding" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same.

[0020] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0021] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0022] See also Figure 1 The embodiment of the present invention provides a residual current protection method for a distribution network based on a distributed photovoltaic power source, comprising the following steps: S1: Obtain the zero-sequence current waveform within the disturbance period of the distributed photovoltaic power access point, calibrate the starting segment, middle fluctuation segment, and ending segment of the current drastic change area, identify the morphological evolution mode through the change trend between the three segments, the inflection point conversion trend, and the change rhythm of adjacent segments, incorporate the structural mutation points into the continuous recognition sequence according to the time sequence, establish the contour node relationship, and generate the contour mutation mark sequence; S2: Call the profile mutation marker sequence to identify the fluctuation shape area covered by the disturbance response in the zero-sequence current, analyze the change rhythm, persistence characteristics and stability trend shown in the early, middle and stabilization periods of the disturbance, construct the response evolution structure according to the fluctuation continuity and curve trend changes, and structurally reposition the fluctuation segments in the differentiated stages to generate the fluctuation structure segment group; S3: Call the wave structure segment group and the contour mutation mark sequence, establish associations pair by pair according to the time relationship, arrange and judge the order between the starting point and mutation point of each group, and verify whether there is a consistent trend direction in the structural morphology. Sorting out the synergistic relationship of the change combination in the response mode, screening the response segment that can trigger the protection condition threshold, and generating an action trigger segment list; S4: Call the action trigger segment list, locate the current starting point of the response segment, and trace the current evolution direction of the previous change segment. If the previous segment has the same evolution trend, move the starting point of the current response segment forward, and rearrange the order of multiple response segments. Adjust the start priority relationship by judging the extensibility of the overall trend of the response segment, and generate an action start position list; S5: Call the start sequence and characteristics in the action start position list, judge the stability of the starting behavior of the response segment, and arrange the action response priority according to the segment connection position, structure and merge the action execution rhythm in the segment, organize it into output content with relay triggering properties, and generate the distribution network residual current protection plan.

[0023] The profile mutation marker sequence includes the starting change node, the middle fluctuation feature point and the ending trend node. The fluctuation structure segment group includes the disturbance morphology partition, the curve trend segment and the stage fluctuation feature. The action trigger segment list includes the trigger starting position, the trend consistency segment and the response synergy combination. The action start position list includes the start sorting information, the trend continuation indicator and the response segment adjustment relationship. The distribution network residual current protection scheme includes the response priority level, the action rhythm structure and the relay output rule.

[0024] See also Figure 2 , the specific steps of S1 are: S101: Obtain the zero-sequence current waveform within the disturbance period of the distributed photovoltaic power access point, extract the change amplitude, slope value, and first-order difference value within each period, construct a difference index set based on the three parameters, determine the start and end positions of the segments in the waveform where the change rate exceeds the zero-sequence current disturbance threshold, and generate the current change interval boundary pair; First, the three-phase current signals need to be synchronously collected at the access point. The sampling frequency is set to 10kHz to ensure that the time domain resolution meets the requirements of dynamic waveform extraction. The sampled three-phase currents A, B, and C are superimposed in sequence to form the zero-sequence current, forming the original waveform data sequence. In the actual sampling scenario, the acquisition cycle is set to 2 seconds, corresponding to 200,000 sampling points. When the sampled data is divided into cycles, the 200,000 points are divided into 100 complete cycles according to the 50Hz power frequency cycle. Each cycle contains 2000 sampling points. Next, within each cycle, the difference between the maximum and minimum values ​​of the zero-sequence current in the cycle is calculated as the change amplitude of the cycle. For example, if the maximum value of the 10th cycle is 3.2A and the minimum value is 1.1A, the amplitude of the cycle is 2.1A. The difference values ​​of consecutive points in the cycle are then solved to extract the first-order difference sequence. For example, if the first sampling point is 1.2A and the second is 1.25A, the difference is 0.05A. After repeating the differential sequence to obtain the differential sequence, the change between adjacent differential points is further calculated. The rate is used as the slope. For example, if the previous difference is 0.05A and the next difference is 0.15A, the slope is 0.1A / 0.1ms, that is, 1000A / s. The change amplitude, maximum difference value, and maximum slope value extracted in each cycle are combined into a period feature triplet to construct a feature set for all cycles. Then, the benchmark threshold for disturbance judgment is set. It is recommended to set the amplitude threshold to 1.5A, the differential threshold to 0.2A, and the slope threshold to 300A / s. The threshold setting is based on the zero-sequence current during stable operation in field measurements. Based on the empirical data that the average current fluctuation does not exceed 0.8A, the difference does not exceed 0.1A, and the slope does not exceed 150A / s, the disturbance threshold is set using 2 times the empirical upper limit. For each cycle, it is determined whether any of the three parameters exceeds the corresponding threshold. If so, it is marked as a disturbance cycle. After traversing 100 cycles in sequence, all disturbance cycle numbers are extracted. For example, if the 12th, 13th, and 14th cycles all meet the disturbance conditions, the continuous interval is formed into a group of boundary pairs (12, 14), and finally the boundary pairs of the current change interval are obtained.

[0025] S102: Calling the boundary pair of the current change interval, calculating the first-order and second-order derivatives of the current waveform within the interval, determining the trend inflection point based on the derivative sign and the extreme value position, and numbering the fluctuation starting point, inflection point, and end point in sampling order to generate a current fluctuation feature point sequence; First, based on the current change interval boundary pairs obtained in the previous step, such as (12, 14) and (28, 31), each corresponding complete data segment is extracted from the original current sampling data. Taking (12, 14) as an example, the corresponding sampling points are from point 22001 to point 28000. In this data segment, the first-order derivative operation is performed on the sampling points in sequence, that is, the difference between the current values ​​of consecutive sampling points is divided by the sampling time interval. The sampling interval is 0.1ms. Therefore, the values ​​for the 24001st point and the 24000th point are 1.2A and 1.1A respectively, and the first-order derivative is 1A / ms. Then, the derivative sequence is subjected to differential processing to obtain the second-order derivative, that is, the difference between the consecutive derivative values ​​is again divided by 0.1ms to obtain the rate of change. For example, the two derivative values ​​are 1. 0A / ms and 1.4A / ms, the corresponding second-order derivative is 4A / ms². All first-order and second-order derivative values ​​are recorded as a vector sequence. On this basis, the sign change point is found. If the first-order derivative changes from positive to negative, it means that the waveform changes from rising to falling, which is determined as the maximum point, and from negative to positive, it is the minimum point. In this way, the sign change is judged point by point, and the number positions of all extreme points in the sampling segment are marked. The area where the derivative value before and after the extreme point changes significantly is found, and the starting point, inflection point and end point of the fluctuation are recorded. For example, the starting point is 24012, the inflection point is 24045, and the end point is 24090. They are numbered in sequence according to the sampling order and recorded as a complete fluctuation feature segment. The above process is repeated for all fluctuation events in the disturbance segment to finally form a current fluctuation feature point sequence.

[0026] S103: Based on the current fluctuation feature point sequence, identify the arrangement direction of peaks and troughs and the rhythm ratio of adjacent segments, extract the time index and amplitude change of the direction mutation point, construct the segment connection relationship matrix, and generate the contour mutation mark sequence; For the extracted current fluctuation feature point sequence, such as {24012, 24045, 24090}, {24150, 24185, 24230}, further waveform structure analysis is performed on each group of feature points. The middle point of each group of three points is the extreme point, which needs to be judged as a peak or a trough. If the current value at point 24045 is 3.1A, the current value at the starting point 24012 is 2.2A, and the current value at the ending point 24090 is 2.5A, then this section of fluctuation is a peak. After determining the extreme value properties of all fluctuation segments, record the peak-trough arrangement direction, and then calculate the time interval between the starting and ending points of adjacent fluctuation segments. If the ending point of fluctuation 1 is 24090, the corresponding time is 2.409 seconds, and the starting point of fluctuation 2 is 24150, the corresponding time is 2.415 seconds, then the interval is 0.006 seconds. At the same time, calculate the amplitude change of the two fluctuation segments, for example, fluctuation 1 is 3.1A-2.2A=0.9A, and fluctuation 2 is 2.6A-1 .8A=0.8A. The amplitude ratio between the two segments is 0.9 / 0.8=1.125. According to the set mutation judgment ratio threshold of 1.4, this segment does not constitute a mutation. For another example, a fluctuation amplitude of 2.0A has a ratio of 2.5 compared to the previous segment of 0.8A, which is determined to be a mutation segment. The mutation ratio threshold is set to 1.75 times the maximum observed amplitude mean based on the statistical mean of the previous stable operation amplitude. If the historical mean is 0.6A, the mutation threshold is set to 1.05A. For all fluctuation segments, the interval time and amplitude ratio with the previous segment are determined. Segments that meet the ratio threshold are considered mutation segments, and their inflection point numbers and times are recorded in the mutation mark list. Then, the connection relationship between each segment is established. If the time interval between two segments is less than 0.01 seconds, they are considered connected. A connection matrix is ​​constructed. For all fluctuation segments, a profile mutation mark sequence is generated based on the mutation mark and connection relationship. For example, [0, 1, 0, 1, 1, 0] is a binary sequence indicating whether the corresponding segment is a mutation.

[0027] See also Figure 3 , the specific steps of S2 are: S201: calling the profile mutation marker sequence to detect the segments of the zero-sequence current whose fluctuation amplitude exceeds the disturbance response identification threshold, extracting the segments of the initial, middle and stabilization periods of the disturbance based on the continuous time and amplitude change, and obtaining the disturbance stage fluctuation segment set; Call the contour mutation mark sequence as the input basis, read all the fluctuation segment numbers and their corresponding time indexes with the mutation mark value of 1, obtain the starting and ending time of each segment and the maximum and minimum values ​​of the zero-sequence current waveform in the segment, and calculate the fluctuation amplitude of each segment accordingly. For example, if the starting current in a certain fluctuation segment is 1.2A and the ending current is 3.0A, the fluctuation amplitude of this segment is 1.8A. Then set the disturbance response recognition threshold to determine whether the fluctuation amplitude meets the significant disturbance standard. The threshold is determined based on the mean and maximum value of the fluctuation amplitude under the historical stable operation state. If the historical mean is 0.6A and the maximum value is 1.2A, the disturbance recognition threshold is determined to be 1.5A. For each mutation segment, determine whether its fluctuation amplitude is greater than the threshold. If it is greater than, include the segment as a disturbance segment in the recognition range, and then compare the time intervals of adjacent segments in the time order of each segment. If the time interval is less than the set continuous time threshold ( If the time interval is greater than 0.1 seconds, it is considered a continuous disturbance process in the same stage. If it is greater than 0.1 seconds, the continuity is broken to form a new stage segment group. Within each stage segment group, the fluctuation amplitude and time length of each segment are divided into intervals according to the time series. The time threshold for initial disturbance identification is set to 0.3 seconds, and the time threshold for stabilization period is set to 0.5 seconds. If the first 0.3 seconds after the beginning of a segment group is the initial segment, the middle segment is the subsequent 0.4 seconds, and the last 0.5 seconds is the stabilization period segment, then combined with the current value change characteristics within the segment, it is further judged whether it belongs to the initial stage, whether the current upward trend is established, whether the middle segment fluctuates frequently, and whether the stabilization period shows a downward and convergent trend. For example, in the initial segment, the current continuously rises from 1.2A to 3.5A, and in the stabilization period, the current drops from 3.4A to 2.0A with a fluctuation amplitude less than 0.3A, then it meets the corresponding stage characteristic conditions. All types of segments that meet the conditions are classified and summarized to form a disturbance stage fluctuation segment set.

[0028] S202: Based on the disturbance phase fluctuation segment set, extract the duration, fluctuation frequency and rhythm period of the phase segment, calculate the stability coefficient, and determine the continuity state to obtain the phase fluctuation characteristic index group; Based on the obtained set of fluctuation fragments in the disturbance stage, the start and end time points of each stage fragment are extracted one by one, and its duration is calculated. The duration is the end time minus the start time. For example, the start time of a middle segment is 1.05 seconds and the end time is 1.75 seconds, then the duration is 0.7 seconds. The number of local extreme points appearing in the segment, that is, the number of fluctuations, is determined by the forward difference and the sign change. If 5 extreme points appear within 0.7 seconds, the fluctuation frequency is 5, and the rhythm cycle can be defined as the duration divided by the fluctuation frequency, that is, 0.7 seconds / 5=0.14 seconds. After performing the above operations on each segment, the duration, fluctuation frequency and rhythm cycle of each segment are obtained, and then the stability coefficient of the segment is calculated based on the mean-variance relationship between the fluctuation frequency and the rhythm cycle. The stability coefficient is defined as the ratio of the standard deviation of the rhythm cycle sequence to the mean value. If the rhythm of multiple segments in a stage The periods are 0.13 seconds, 0.14 seconds, and 0.15 seconds respectively, with a mean of 0.14 seconds, a standard deviation of approximately 0.0082 seconds, and a stability coefficient of 0.0082 / 0.14≈0.058. If the coefficient is less than 0.1, it is determined to be a stable segment, otherwise it is an unstable segment. The stability judgment threshold of 0.1 is set based on the fact that the standard deviation of the rhythm period in the field test data is concentrated between 0.005 and 0.012. Based on the above calculated stability coefficient value, the continuity state of all segments is judged. The judgment basis is whether the interval between the end time of the current segment and the start time of the next segment is less than 0.05 seconds and the stability states of the previous and subsequent segments are the same. If so, it is classified as a continuous state, otherwise it is disconnected. Finally, the characteristic indicators of the fluctuation segments in each disturbance stage are constructed, forming a stage fluctuation characteristic indicator group including duration, fluctuation frequency, rhythm period, stability coefficient and continuity state.

[0029] S203: Calculate the stability difference value between segments based on the stage fluctuation characteristic index group, classify segments with similar characteristics according to the trend change direction, establish a structure mapping relationship, and generate a fluctuation structure segment group; According to the stage fluctuation characteristic index group obtained above, the difference values ​​of the stability coefficients between all segments are calculated pairwise. The difference value is calculated by subtracting the stability coefficient of the previous segment from the stability coefficient of the latter segment and taking the absolute value. For example, the stability coefficient of segment A is 0.06, and that of segment B is 0.13, then the difference value is 0.07. Then a stability difference judgment threshold is set, for example, 0.08 is used as the distinction boundary. If the difference value is less than the threshold, it is judged as a similar segment, otherwise it is a different segment. The threshold is set according to the tolerance for stability changes. Usually, the difference between adjacent segments under stable operation does not exceed 0.06, so it is set to 0.08 as the upper limit state. The stability difference values ​​between all segments are obtained through the above process, and a difference matrix is ​​formed. Similar segments are merged and processed, and at the same time, it is determined whether their trend change directions are consistent. The judgment is based on whether the change direction of the rhythm cycle sequence remains increasing, decreasing, or approximately constant. For example, if the rhythm period of segment 1 is 0.13 seconds, that of segment 2 is 0.14 seconds, and that of segment 3 is 0.16 seconds, it can be determined to be an increasing trend. Segments with the same stability level and consistent trend direction are classified as similar feature segments. Then, a structural mapping relationship is established based on the classification results, and the connection relationship and sequence characteristics between similar segments are recorded in a mapping matrix. Each unit in the matrix represents the correspondence between the segment number and the class number to which it belongs. For example, segment 3 is mapped to structure class 2, and segment 4 is mapped to structure class 2. Finally, the fluctuation structure segment group under the similar mapping structure is output.

[0030] See also Figure 4 , the specific steps of S3 are: S301: Calling the fluctuating structure segment group and the contour mutation marker sequence, matching the structure segments and mutation points one by one according to the sampling time, determining the time sequence of the starting point and mutation point of each group of structure segments, screening out segments with consistent time arrangement, and generating a structure mutation matching pair sequence; First, read the start time and end time of each structure segment in the structure segment group. For example, structure segment A starts at 2.305 seconds and ends at 2.605 seconds. At the same time, extract the time index of all sampling points marked as mutations from the mutation marker sequence. For example, the time of mutation point P is 2.310 seconds. Compare the mutation point time with the structure segment start time. For each structure segment, find the mutation point closest to its time from the mutation point sequence. Extract the time difference between the mutation point and the structure segment and record it. If the absolute value of the time difference is less than 0.02 seconds and the mutation point time is not later than the structure segment start time, it is considered a matching pair with consistent time arrangement. For example, if the mutation point time is 2.302 seconds and the structure segment start time is 2.305 seconds, then the two are The time difference is 0.003 seconds, which meets the time matching condition and is recorded as a matching pair (structure segment A, mutation point P). We continue to traverse all structure segments and mutation points and perform the above time difference judgment operation to exclude those mutation points that are later than the starting point of the structure segment or whose time difference is greater than the matching threshold. For example, if the starting time of structure segment B is 2.420 seconds and the corresponding mutation point time is 2.430 seconds, the time sequence requirement is not met and the pair is not included in the matching. The matching threshold is set based on the actual photovoltaic response delay characteristics. Usually, the transient disturbance response lag time is about 10ms to 20ms, so a time difference tolerance of 20ms is set. Finally, all matching pairs that meet the time sequence relationship and have a reasonable time interval are collected to form a structure mutation matching pair sequence.

[0031] S302: Based on the structural mutation matching pair sequence, compare the direction vectors of the fluctuations before and after the structural segment and the mutation point to determine whether the trend direction is consistent, and mark the combinations with consistent trends to obtain a trend consistent combination sequence; According to each set of matching data in the above-mentioned structure mutation matching sequence, the starting and ending current values ​​in the structure segment are extracted to form a fluctuation direction vector. For example, the current value of the structure segment A at the starting point 2.305 seconds is 1.8A, and the current value at the ending point 2.605 seconds is 2.9A. The direction vector is positive rising. The trend direction of the fluctuation segment near the corresponding mutation point is extracted by taking 20 sampling points before and after the mutation point P (i.e., 2ms before and after), and calculating the difference between the starting and ending current values. For example, if the front segment is 1.75A and the back segment is 2.10A, the mutation direction is rising. If the fluctuation direction of the structure segment is consistent with the direction of the mutation point, that is, both are rising or both are falling, it is marked as trend consistent. If the structure segment is a downward trend and the mutation point is If the direction is rising, it is marked as inconsistent trend. All matching pairs perform this direction judgment. The judgment standard is made by comparing the current magnitude relationship between the starting point and the end point. If the end point value minus the starting point value is greater than 0.2A, it is defined as an upward trend. If it is less than -0.2A, it is defined as a downward trend. If the amplitude change is between ±0.2A, it is judged as stable or trendless. The threshold setting is derived from the minimum effective fluctuation amount obtained from the actual level of sampling noise and disturbance response statistics. Then, the matching pairs with consistent trends are marked as 1, and the inconsistent or invalid trends are marked as 0. Finally, a trend consistent combination sequence is generated. For example, [1, 0, 1, 1, 0] means there are five groups of matches, of which the 1st, 3rd, and 4th groups have consistent directions. The identification sequence is output as the result.

[0032] S303: Based on the trend-consistent combination sequence, the amplitude change and duration of the structural segments in the combination are extracted and compared with the protection condition threshold, the response segments that meet the trigger condition are screened, and an action trigger segment list is generated; Based on the trend consistent combination sequence, read all the structure segments marked as 1, and extract the amplitude change and duration for each of them. The amplitude change is the current value at the end of the structure segment minus the current value at the starting point. For example, if the starting current of a structure segment is 2.1A and the end current is 3.5A, the amplitude change is 1.4A, and the duration is the end time minus the starting time. If the starting point is 2.400 seconds and the end point is 2.650 seconds, the duration is 0.25 seconds. Then compare these two values ​​with the preset protection condition threshold. It is recommended to set the amplitude threshold to 1.2A and the duration threshold to 0.2 seconds. If the amplitude change is ≥1.2A and the duration is ≥0.2 seconds, then It is determined that the structural segment meets the action trigger conditions, and its number is included in the action trigger list. The setting of the protection threshold refers to the short-term fluctuation amplitude range under normal working conditions, which is usually between 0.5A and 1.0A. Therefore, the amplitude threshold is set slightly higher than the upper limit level, and the duration is set to be no less than 200ms based on the protection action response time. Finally, the amplitude and time dual conditions are judged for the structural segments with consistent trends in turn, and the numbers of all structural segments that meet the trigger conditions are recorded. For example, the structural segments numbers 7, 9, and 12 correspond to amplitudes of 1.5A, 1.3A, and 1.7A respectively, and the duration is all above 0.25 seconds. Then their numbers are used to form the action trigger segment list.

[0033] See also Figure 5 , the specific steps of S4 are: S401: Call the action trigger segment list, locate the starting point of the response segment, trace the current direction of the previous change segment, and determine whether it is consistent with the current segment trend. If consistent, extend the starting point of the response segment forward to generate a trend extension starting point set; Call the action trigger segment list as input, extract the starting sampling time of each structure segment in the list, for example, the starting time of structure segment A is 3.120 seconds, first locate the index position of the time point in the original sampling sequence, then search forward from this position to the previous identified change segment, search the change segment set for the most recent segment with an end time earlier than 3.120 seconds and a time difference of less than 0.3 seconds with the current structure segment as the previous change segment, extract its end current value and the starting current value of the current structure segment, and determine whether their direction changes are consistent. If the starting current value of the current segment is 2.1A and the ending current value is 3.0A, which is an upward trend, and the ending current value of the previous segment is 1.9A and the starting current value is 1.1A, then The previous segment also has an upward trend. The criterion for determining whether the direction is consistent is that the current difference between the starting point and the end point is greater than 0.5A and the direction is the same. If the direction is consistent, the starting point of the previous segment replaces the starting point of the original structure segment to form a new extension starting point. If the direction is inconsistent or the difference is less than 0.5A, no extension is performed. The minimum amplitude threshold of 0.5A for direction judgment is set based on the distinguishable range of true trend identification under reference noise. The tracing and judgment process is repeated for all structure segments. The starting point time of the structure segment that is successfully extended forward is recorded as the extension starting point, and a trend extension starting point set is formed. For example, the original starting point of structure segment A is 3.120 seconds, and the starting point after extension is 3.010 seconds. Finally, a complete trend extension starting point set is output.

[0034] S402: Extract the sampling time of the extended response segment based on the trend extension starting point set, sort it by the starting point time, determine whether the time sequence is misplaced, and unify the order to obtain a response segment time sequence table; According to the trend extension starting point set, the sampling time corresponding to each extension starting point is extracted. For example, the extension starting points are 2.980 seconds, 3.010 seconds, 3.005 seconds, and 2.990 seconds respectively. These sampling times are arranged in ascending order. The sorting operation is completed by performing ascending order on the starting time. The time sequence after sorting is 2.980 seconds, 2.990 seconds, 3.005 seconds, and 3.010 seconds. Then, whether there is a misalignment between each time point is determined one by one. The judgment standard is whether the time difference between the current time and the starting time of the original structure segment is more than 0.5 seconds. If the extension starting time is earlier than the end time of another structure segment, it is marked as a disordered sequence. For example, the end time of the original structure segment B If the time is 3.000 seconds, and the extension start time is 2.980 seconds, it is considered that the extension point is too early, resulting in a misorder between segments. The time differences of all adjacent segments are compared. If a pair of time sequences is that the extension start point B is 3.010 seconds and the extension start point A is 3.005 seconds, the sorting order should be adjusted to put 3.005 seconds before 3.010 seconds. After unified sorting, a time sequence table without misalignment is formed. The table records the extension start points and their structural segment number sequences arranged in chronological order, for example, [{segment number 3, start point 2.980 seconds}, {segment number 1, start point 2.990 seconds}, {segment number 4, start point 3.005 seconds}, {segment number 2, start point 3.010 seconds}], and finally the response segment time sequence table is obtained.

[0035] S403: Based on the response segment time sequence table, extract the trend direction, duration, and amplitude change, calculate the trend continuation score, adjust the segment priority according to the score, and generate an action start position list; Based on the response segment time sequence table, the trend direction, duration and amplitude change of each segment are read in turn. The trend direction is determined by the relationship between the starting and ending current values ​​within the segment. For example, if the starting point is 2.0A and the ending point is 3.2A, it is in an upward direction. The duration is the ending time minus the starting time, for example, 3.150 seconds - 2.980 seconds = 0.170 seconds. The amplitude change is the ending current minus the starting current, that is, 3.2A - 2.0A = 1.2A. Then the trend continuation score is calculated. The score is set as the sum of the trend consistency weight, amplitude weight, and duration weight. If the trend consistency is consistent with the previous segment, it is scored as 1 point. The amplitude weight is converted proportionally. For every amplitude exceeding 1. 0A is scored as 1 point, and those less than 1.0A are scored proportionally, for example, if the amplitude is 0.8A, it is scored 0.8 points. Every time the duration exceeds 0.2 seconds, it is scored as 1 point. Otherwise, it is scored proportionally, for example, if the duration is 0.17 seconds, it is scored as 0.85 points. If a structural segment has a consistent trend, an amplitude of 1.2A, and a duration of 0.25 seconds, the score is 1+1.2+1=3.2 points. After calculating the score values ​​for all segments separately, the paragraphs are sorted from high to low according to the score to form a priority sequence. For example, if the scores are 3.2, 2.7, 2.5, and 2.0 respectively, the corresponding paragraphs are numbered A, B, C, and D. The final output of the sorted action start position list is [A, B, C, D].

[0036] See also Figure 6 , the specific steps of S5 are: S501: Call the action start position list, extract the start sequence and trend continuity parameters of the response segments, determine the trend direction and fluctuation stability of each segment start position, calculate the trend duration and amplitude change, establish the corresponding stability score matrix, and generate a response segment stability score group; Call the action start position list as the input source, read the start sequence number of each response segment in the list and its corresponding trend continuity parameter, for example, the starting time of the segment numbered 1 is 3.010 seconds, the trend direction is rising, the starting current is 1.9A, and the ending current is 3.0A. Extract the current change direction corresponding to the starting position of the response segment, and judge it as an upward trend by comparing the starting and ending current values. If the direction change is greater than 0.8A, it is judged to be a strong trend. If it is between 0.5A and 0.8A, it is a medium trend, and if it is less than 0.5A, it is a weak trend. This partition is based on the experience of the current disturbance threshold and is obtained in combination with the field fluctuation data. Then, the stability within each fluctuation segment is judged, and the fluctuation amplitude change and the number of local extreme values ​​in the sampled current sequence are extracted. If the number of extreme value points in the segment is less than 5 and the amplitude change is within the range of ±0.2A, the fluctuation is considered stable, otherwise it is judged to be unstable. Combine the trend strength with the stability judgment result, and perform a quantitative analysis on each segment. Calculate the trend duration. For example, if the starting time is 3.010 seconds and the ending time is 3.250 seconds, the duration is 0.24 seconds. Calculate the amplitude change value by subtracting the starting current from the ending current, which is 1.1A here. The trend strength score (strong trend = 1, medium trend = 0.7, weak trend = 0.4) and the stability score (stable = 1, unstable = 0.5) together form a scoring factor. This is then multiplied by the duration (in seconds) and the amplitude change value (in amperes) to form a four-dimensional scoring matrix. For example, if segment 1 is a strong trend + stable, with a duration of 0.24 seconds and an amplitude of 1.1A, the score is [1×0.24, 1×1.1, 0.24, 1.1] = [0.24, 1.1, 0.24, 1.1]. Construct a scoring matrix for all response segment scores, with each row corresponding to a response segment and each column representing the trend score item, stability score item, duration, and amplitude. Finally, output this matrix as the response segment stability score group.

[0037] S502: Based on the response segment stability score group and the temporal connection relationship between the action segments, the start and end point index differences between adjacent segments are extracted to determine whether the connection state is continuous. The response sequence is rearranged based on continuity priority to obtain a response segment priority sequence list; According to the scoring results of each response segment in the response segment stability scoring group, combined with the time connection relationship between the action segments, first read the response segment time sequence and obtain the starting and ending sampling indexes of each segment. For example, the starting index of segment A is 30100 and the ending index is 30500. The starting index of segment B is 30505 and the ending index is 30900. Calculate the index difference between the starting and ending points of adjacent segments. If the index difference is less than 10 sampling points (the corresponding time is less than 1ms at 10kHz sampling), it is considered to be temporally continuous. If it is greater than 10, it is discontinuous. The continuous state is recorded as 1 and the discontinuous state is recorded as 0. Compare all segments pairwise to obtain a connection state list. For example, [1, 1, 0, 1] means there is a discontinuity between the second and third segments. Continuity: For each segment, the priority is adjusted based on whether it is in a continuous segment. If the segment is in a continuous area and the sum of the stability score and the trend score in the previous and next scoring matrices is higher than 1.5 (i.e., the combined score is strong), its response order priority is increased. For example, segment A was originally ranked 3rd, but it was promoted to 1st after meeting the continuity and scoring requirements. The order of all response segments is rearranged, with the highest priority placed at the front, to generate a new response segment priority list. The list records the segment number and its corresponding starting time, priority number, and connection status, for example, [{segment 1, 3.010s, priority 1, continuous 1}, {segment 2, 3.250s, priority 2, continuous 1}, {segment 3, 3.600s, priority 3, continuous 0}].

[0038] S503: Based on the response segment priority list, the action rhythm change cycle and the response duration are integrated, the trigger segment position and the action rhythm structure are matched, a time sequence output set with a trigger identifier is formed, and a distribution network residual current protection scheme is generated; Based on the response segment priority list, the action rhythm change cycle and response duration are extracted for each segment. The rhythm cycle is calculated as the difference between the starting time of adjacent response segments. For example, the starting time of segment 1 is 3.010 seconds, segment 2 is 3.250 seconds, the cycle is 0.240 seconds, and the cycle from segment 2 to segment 3 is 0.350 seconds. The response duration is the end time of the current segment minus the starting time. For example, segment 1 is 3.250-3.010=0.240 seconds. The rhythm cycle and duration of each paragraph are recorded, and then the action start position is compared with the rhythm information. If the rhythm cycle change is less than 0.1 seconds and the duration is more than 0.2 seconds, it is determined that the paragraph structure conforms to the action rhythm law and its position is changed. As the effective position for action triggering, if a certain period mutation exceeds 0.3 seconds or the duration is less than 0.1 seconds, it does not meet the rhythm matching conditions. The segment number that meets the conditions is assigned the trigger identifier 1, and the segment that does not meet the conditions is assigned 0. Finally, the trigger identifier, starting time, rhythm period and duration are integrated to form a timing output set, such as [{segment 1, trigger 1, starting point 3.010s, period 0.240s, duration 0.240s}, {segment 2, trigger 1, starting point 3.250s, period 0.350s, duration 0.220s}, {segment 3, trigger 0, starting point 3.600s, period 0.350s, duration 0.100s}], and the set is finally output as the residual current protection scheme for the distribution network.

[0039] See also Figure 7 , the residual current protection system of the distribution network based on distributed photovoltaic power generation includes: The contour marker extraction module is used to implement S1: extract the zero-sequence current waveform within the disturbance period of the distributed photovoltaic access point, calibrate the starting segment, middle segment, and ending segment of the change, and generate a contour mutation marker sequence based on the change trend, inflection point conversion, and rhythm difference identification evolution mode of the three segments; The fluctuation structure classification module is used to implement S2: based on the profile mutation marker sequence, it identifies the fluctuation pattern of the disturbance response, analyzes the rhythm, persistence and trend in the initial, middle and stabilization periods, constructs the response evolution structure, and classifies it into fluctuation structure segment groups; The response segment screening module is used to implement S3: combining the fluctuation structure segment group with the profile mutation marker sequence, establishing time series associations pair by pair, judging the order and trend consistency, sorting out the relationship between response combinations, screening response segments that trigger the protection condition threshold, and generating an action trigger segment list; The startup priority adjustment module is used to implement S4: based on the action trigger segment list, locate the starting point of the response segment, trace the trend of the previous segment, move the starting point forward if it is consistent, reorder the response segments, adjust the startup priority, and generate the action startup location list; The trigger rhythm merging module is used to implement S5: based on the action start position list, judge the stability of the response segment, sort the action response priority, structure and integrate the execution rhythm, and generate the distribution network residual current protection plan.

[0040] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A residual current protection method for a distribution network based on distributed photovoltaic power sources, characterized in that: The following steps are involved: S1: Extract the zero-sequence current waveform within the disturbance period of the distributed photovoltaic access point, calibrate the starting segment, middle segment and ending segment of the change, identify the evolution mode based on the change trend, inflection point conversion and rhythm difference of the three segments, and generate a profile mutation marker sequence; S2: Based on the profile mutation marker sequence, identify the fluctuation pattern of the disturbance response, analyze the rhythm, persistence and trend of the initial, middle and stabilization periods, construct the response evolution structure, and classify it into fluctuation structure segment groups; S3: combining the fluctuation structure segment group and the profile mutation mark sequence, establishing time series associations pair by pair, judging the order and trend consistency, sorting out the relationship between response combinations, screening the response segments that trigger the protection condition threshold, and generating an action trigger segment list; S4: Based on the action trigger segment list, locate the starting point of the response segment, trace the trend of the previous segment, and if consistent, move the starting point forward, re-arrange the order of the response segments, adjust the start priority, and generate an action start position list; S5: Based on the action start position list, determine the stability of the response segment, sort the action response priority, structure and integrate the execution rhythm, and generate a distribution network residual current protection plan.

2. The residual current protection method for distribution network based on distributed photovoltaic power supply according to claim 1, characterized in that: The profile mutation mark sequence includes a starting change node, a mid-segment fluctuation feature point, and an ending trend node; the fluctuation structure segment group includes a disturbance morphology partition, a curve trend segment, and a stage-by-stage fluctuation feature; the action trigger segment list includes a trigger starting position, a trend consistency segment, and a response synergy combination; the action start position list includes start sorting information, a trend continuation indicator, and a response segment adjustment relationship; and the distribution network residual current protection scheme includes a response priority level, an action rhythm structure, and a relay output rule.

3. The residual current protection method for distribution network based on distributed photovoltaic power supply according to claim 1, characterized in that: The protection condition threshold is defined as the time when the characteristic change of the zero-sequence current waveform reaches the minimum judgment standard required to trigger the protection action in the disturbance response.

4. The residual current protection method for distribution network based on distributed photovoltaic power supply according to claim 1, characterized in that: The definition of tracing back the previous trend refers to analyzing the fluctuation trend forward after the starting point of the action response segment. If it is consistent with the current trend, the process of moving the response starting point forward is called tracing back the previous trend.

5. The residual current protection method for distribution network based on distributed photovoltaic power supply according to claim 1, characterized in that: The specific steps of S1 are: S101: Obtain the zero-sequence current waveform within the disturbance period of the distributed photovoltaic power access point, extract the change amplitude, slope value, and first-order difference value within each period, construct a difference index set based on the three parameters, determine the start and end positions of the segments in the waveform where the change rate exceeds the zero-sequence current disturbance threshold, and generate the current change interval boundary pair; S102: Calling the current change interval boundary pair, calculating the first-order and second-order derivatives of the current waveform within the interval, determining the trend inflection point based on the derivative sign and the extreme value position, and numbering the fluctuation starting point, inflection point, and end point in sampling order to generate a current fluctuation feature point sequence; S103: Based on the current fluctuation feature point sequence, identify the arrangement direction of peaks and troughs and the rhythm ratio of adjacent segments, extract the time index and amplitude change of the direction mutation point, construct a segment connection relationship matrix, and generate a contour mutation mark sequence.

6. The residual current protection method for distribution network based on distributed photovoltaic power supply according to claim 1, characterized in that: The specific steps of S2 are: S201: calling the profile mutation marker sequence to detect segments in the zero-sequence current whose fluctuation amplitude exceeds the disturbance response identification threshold, extracting segments in the initial, middle and stabilization stages of the disturbance based on the continuous time and amplitude variation, and obtaining a disturbance stage fluctuation segment set; S202: Based on the disturbance phase fluctuation segment set, extract the duration, fluctuation frequency and rhythm period of the phase segments, calculate the stability coefficient, and determine the continuity state to obtain a phase fluctuation characteristic index group; S203: Calculate the stability difference value between segments based on the stage fluctuation characteristic index group, classify the segments with similar characteristics according to the trend change direction, establish a structure mapping relationship, and generate a fluctuation structure segment group.

7. The residual current protection method for distribution network based on distributed photovoltaic power supply according to claim 1, characterized in that: The specific steps of S3 are: S301: Calling the fluctuating structure segment group and the contour mutation marker sequence, establishing matches between the structure segments and the mutation points one by one according to the sampling time, determining the time sequence of the starting point and the mutation point of each group of structure segments, screening out segments with consistent time arrangement, and generating a structure mutation matching pair sequence; S302: comparing the direction vectors of the fluctuations before and after the structural segment and the mutation point based on the structural mutation matching pair sequence to determine whether the trend directions are consistent, and marking combinations with consistent trends to obtain a trend consistent combination sequence; S303: Based on the trend-consistent combination sequence, extract the amplitude change and duration of the structural segments in the combination, compare them with the protection condition threshold, screen the response segments that meet the trigger condition, and generate an action trigger segment list.

8. The residual current protection method for distribution network based on distributed photovoltaic power source according to claim 1, characterized in that: The specific steps of S4 are: S401: calling the action trigger segment list, locating the starting point of the response segment, tracing back the current direction of the previous change segment, and judging whether it is consistent with the trend of the current segment. If it is consistent, extending the starting point of the response segment forward to generate a trend extension starting point set; S402: extracting the sampling time of the extended response segments based on the trend extension starting point set, sorting them by the starting point time, determining whether the time sequence is misplaced, and unifying the order to obtain a response segment time sequence table; S403: Based on the response segment time sequence table, extract the trend direction, duration and amplitude change, calculate the trend continuation score, adjust the response segment priority according to the score, and generate an action start position list.

9. The residual current protection method for distribution network based on distributed photovoltaic power source according to claim 1, characterized in that: The specific steps of S5 are: S501: Calling the action start position list, extracting the start sequence and trend continuity parameters of the response segments, judging the trend direction and fluctuation stability of each segment start position, calculating the trend duration and amplitude change, establishing a corresponding stability score matrix, and generating a response segment stability score group; S502: Based on the response segment stability score group and the temporal connection relationship between the action segments, the start and end point index differences between adjacent segments are extracted to determine whether the connection state is continuous, and the response sequence is rearranged based on continuity priority to obtain a response segment priority sequence list; S503: Based on the response segment priority list, the action rhythm change cycle and the response duration are integrated, the trigger segment position and the action rhythm structure are matched, a timing output set with a trigger identifier is formed, and a distribution network residual current protection scheme is generated.

10. A residual current protection system for distribution network based on distributed photovoltaic power sources, characterized in that: The system is used to implement the residual current protection method for a distribution network based on a distributed photovoltaic power source according to any one of claims 1 to 9, and the system includes: The contour marker extraction module is used to implement S1: extract the zero-sequence current waveform within the disturbance period of the distributed photovoltaic access point, calibrate the starting segment, middle segment, and ending segment of the change, and generate a contour mutation marker sequence based on the change trend, inflection point conversion, and rhythm difference identification evolution mode of the three segments; The fluctuation structure classification module is used to implement S2: based on the profile mutation mark sequence, identify the fluctuation form of the disturbance response, analyze the rhythm, persistence and trend of the initial, middle and stabilization periods, construct the response evolution structure, and classify it into fluctuation structure segment groups; The response segment screening module is used to implement S3: combining the fluctuation structure segment group with the profile mutation mark sequence, establishing time series associations for each pair, judging the order and trend consistency, sorting out the relationship between response combinations, screening response segments that trigger the protection condition threshold, and generating an action trigger segment list; The startup priority adjustment module is used to implement S4: based on the action trigger segment list, locate the starting point of the response segment, trace the trend of the previous segment, move the starting point forward if consistent, re-arrange the order of the response segments, and adjust the startup priority to generate an action startup position list; The trigger rhythm merging module is used to implement S5: based on the action start position list, judge the stability of the response segment, sort the action response priority, structure and integrate the execution rhythm, and generate the distribution network residual current protection plan.

Images