Distributed photovoltaic power supply method and system based on big data interconnection

Abstract

The invention relates to the technical field of data processing, in particular to a distributed photovoltaic power supply method and system based on big data interconnection, and the method comprises the following steps: collecting photovoltaic node time sequence data, recognizing trend turning and abrupt change positions, positioning a trend synchronization point and a time migration range, extracting and sorting resource repetition requests, and recognizing common time abrupt change nodes. And screening the replacement channel, and outputting a path change list. According to the method, a behavior change sequence is constructed through a node change trend and a mutation amplitude, a sequence structure is constructed by combining direction consistency and an offset range to calibrate a time sequence difference, request intensity and time to express a resource concentration feature, continuous mutation time points are extracted, and a time-sharing node group is positioned in a cross manner; the method combines path load screening to replace channels and adjust the path sequence, and achieves the effects of realizing node behavior identification precision, resource request response ordering, time-sharing event positioning high efficiency and path scheduling adaptation dynamic, and improving the response capability and power supply coordination in a multi-node change scene.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a distributed photovoltaic power supply method and system based on big data interconnection. Background Art

[0002] The field of data processing technology primarily involves the collection, organization, analysis, and conversion of raw data, achieving efficient data utilization and intelligent management through various computational methods and logical judgments. Its core issues include big data analysis, distributed computing, data communications, data fusion, and information decision support. This technology is widely used in multiple industries, including industrial control, energy management, smart cities, and smart grids, and is particularly important in processing large-scale heterogeneous data, real-time transmission control, and optimized scheduling. Traditional distributed photovoltaic power supply methods involve local energy distribution and regulation of photovoltaic systems based on fixed control logic and on-site hardware parameters. Preset parameter matching and timing control strategies are typically used to address the problems of poor power supply stability and low energy utilization efficiency in distributed photovoltaic systems. Common methods include achieving voltage and current matching and power regulation through PLC programming or controlling switching quantities through remote SCADA systems issuing commands to coordinate system operations.

[0003] Existing technologies rely on static parameter settings and preset adjustment strategies. When there are trend inflection points in node behavior or sudden data changes, the response mechanism is insufficient, which can easily lead to delayed recognition of state changes between nodes. Resource requests overlap in time but a priority sorting structure cannot be established, resulting in deviations in resource competition processing. The system finds it difficult to capture and collaboratively process sudden synchronous changes in multiple nodes within a short period of time. The load status of each channel in the communication path lacks dynamic perception capabilities, the path settings are fixed and single, and it is difficult to achieve effective adjustment in situations where there are path bottlenecks or concentrated forwarding pressure, affecting the adjustment flexibility and power supply stability of the overall system operation. Summary of the Invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and propose a distributed photovoltaic power supply method and system based on big data interconnection.

[0005] In order to achieve the above objectives, the present invention adopts the following technical solution: a distributed photovoltaic power supply method based on big data interconnection, comprising the following steps: S1: Collect PV node time series data, compare the direction of value changes in consecutive time periods, record trend turning points, construct a difference sequence, determine whether the jump meets the mutation standard, merge the mutation and turning positions, and obtain the node behavior change position; S2: Based on the node behavior change position, extract the change direction of the power generation and load nodes, determine the consistency, locate the trend synchronization point, compare the trend difference, mark the change offset area, combine the edge controller records, and output the time synchronization offset range set; S3: Read the time synchronization offset range set, filter duplicate resource target request groups, extract strength and time values, perform product sorting, and output resource overlap request results; S4: Calling the continuous running data in the resource overlap request result, identifying the numerical mutation point, extracting the mutation time set, and cross-comparing it with the node time set to verify the time overlap dense segment, and obtaining the synchronous mutation node identification group; S5: According to the node paths listed by the synchronous mutation node identification group, the current forwarding channel and load status are retrieved, and the channels to be replaced are screened in combination with the communication routing repeater records, and a distributed photovoltaic power supply path change list is output.

[0006] As a further solution of the present invention, the node behavior change position includes the trend turning time, the difference mutation amplitude, and the trend change direction; the time synchronization offset range set includes the synchronization offset start and end time, the trend difference range, and the direction consistency identifier; the resource overlap request result includes the duplicate resource target, the request strength value, and the reorganization sorting index; the synchronous mutation node identification group includes the mutation occurrence time, the time overlap interval, and the mutation node number; the distributed photovoltaic power supply path change list includes the channel number to be replaced, the adjusted path order, and the current load value of the channel; Definition of mutation criteria: When the mutation amplitude of the node behavior value reaches the preset change threshold, it indicates a value jump; Definition of the change deviation area: The area marked when the direction of node behavior change is inconsistent with the direction of change of power generation and load nodes.

[0007] As a further solution of the present invention, the specific steps of S1 are: S101: Collect continuous time series data from the photovoltaic nodes, compare the increase and decrease relationship of the values at two adjacent time points, record the time point when the value change direction is reversed, and generate a trend change time sequence; S102: Calculate the difference between adjacent values based on the time series data, compare the difference with the previous and next differences one by one, determine whether it exceeds the difference jump amplitude threshold, and generate a sudden change position sequence; S103: Based on the trend change time sequence and the sudden change position sequence, extract all time points in the merged sequence as node behavior change points to generate node behavior change positions; Definition of the difference jump amplitude threshold: the difference amplitude limit used to determine whether the difference in the node behavior value has undergone a sudden change.

[0008] As a further solution of the present invention, the specific steps of S2 are: S201: extracting the direction of value changes of the power generation node and the load node at the same time point based on the node behavior change position, comparing the increase and decrease trend directions one by one, selecting a set of time points with consistent change directions, and generating a time sequence with consistent trends; S202: Based on the time point in the trend consistent time sequence, the numerical difference between the power generation node and the load node at the time point and the preceding and following time periods is called to calculate the upstream and downstream trend change values, and the coefficient is compared with the trend deviation threshold, and the position exceeding the threshold is extracted to obtain the trend offset interval set; S203: Based on the time period position in the trend offset interval set, call the time period record data of the photovoltaic edge node controller, match the start and end time of the controller response record in the interval, extract the time span, and establish a time synchronization offset range set; Definition of increasing and decreasing trend direction: the increasing and decreasing trend of the node value at consecutive moments compared to the previous moment; Definition of trend deviation threshold: It is used to determine whether there is a deviation in the node power generation and load change trend.

[0009] As a further solution of the present invention, the upstream and downstream trend change value calculation formula is specifically: ; in, Represents the upstream and downstream trend change value at the i-th time point, represents the power of the power generation node at the i-th time point, represents the load node power at the i-1th time point, represents the power of the power generation node at the i+1th time point, represents the load node power at the i-th time point, Represents the value of the power generation node at the i-th time point, Represents the load node value at the i-th time point.

[0010] As a further solution of the present invention, the specific steps of S3 are: S301: Read the node request content within the time synchronization offset range set, extract the resource target identifier in each request, compare the duplication of the resource target identifier, filter multiple request records with the same resource target, and establish a resource target duplicate request group list; S302: extracting the intensity values and timestamps of the requests in each group according to the resource target repeat request group list, calculating the product index of the intensity and time value, sorting the requests by the product value, constructing a request structure arranged by the product value, and generating a behavior request sequence sequence; S303: Calling the resource target and time information requested in the behavior request sequence, screening the request groups with overlapping time periods, locating the occupation of the same resource by multiple requests in the same time period, and establishing a resource overlapping request result; Definition of multiple request records for the same resource target: a set of requests that point to the same resource identifier within the time synchronization offset range.

[0011] As a further solution of the present invention, the formula for calculating the product index of the intensity and the time value is specifically: ; in, Represents the product of the intensity and time value of request q in resource group r, represents the intensity value of request q in resource group r, represents the intensity value of the kth request in resource group r except request q, Represents the number of requests in resource group r except request q, represents the sum of the strength values of the remaining requests in resource group r, Represents the timestamp of request q in resource group r, represents the average value of all request timestamps in resource group r, Indicates the absolute value of the difference between the timestamp of request q and the average timestamp of resource group r.

[0012] As a further solution of the present invention, the specific steps of S4 are: S401: Calling the continuous operation data in the resource overlap request result, extracting the numerical difference of the node at adjacent time points, comparing the change range of the difference value range item by item, and judging it with the set numerical jump threshold, screening the time points where the jump amplitude exceeds the threshold, and generating a mutation time set; S402: Based on the mutation time set, call the time record information of the corresponding node, match each mutation time point with the time value in the node running time series, identify the segments where multiple node mutation times overlap, and obtain the time overlap dense segment interval; S403: For each time period in the time-intensive overlap interval, extract node identifiers of nodes that experience simultaneous mutation behaviors, remove duplicate nodes, and sort them by time, extract a set of nodes with synchronous mutation characteristics, and establish a synchronous mutation node identifier group; Definition of the range of variation of the difference value: the fluctuation range of the difference between the values of adjacent time points of a node in a short period of time; Definition of numerical jump threshold: the limit of the numerical change amplitude of the numerical mutation; Definition of a node set with synchronous mutation characteristics: a node identification group whose values of multiple nodes mutate within the same time period.

[0013] As a further solution of the present invention, the specific steps of S5 are: S501: extracting the forwarding channel identifiers and corresponding load values currently used by the nodes in the path based on the node path information listed in the synchronous mutation node identifier group, searching for channel and load combinations in sequence according to the node path order, and establishing a channel load mapping table; S502: Based on the channel load value in the channel load mapping table, calling the channel status record of the photovoltaic communication routing repeater, extracting the channel activation status identifier within the same time period, comparing the channel load value and the activation status to see whether they are both within the abnormal range, screening the channel numbers that meet the replacement conditions, and generating a set of replaceable channels; S503: Calling the node path information corresponding to the replaceable channel set, comparing the channel numbering sequence of the path with the replacement numbering position, rearranging the channel sequence, extracting the changed path numbering structure, and establishing a distributed photovoltaic power supply path change list; Definition of abnormal range: The numerical range in which the channel load value and enable status value exceed the normal operating threshold.

[0014] The distributed photovoltaic power supply system based on big data interconnection includes: The node trend extraction module obtains the time series values of PV node power and load, compares the power change direction of adjacent time periods and records the reversal time, calculates the power difference and compares it with the fluctuation identification benchmark value, extracts the time point when the change amplitude exceeds the limit, integrates the reversal and mutation time, and generates the node behavior change location; The behavior offset comparison module extracts the power direction of the power generation and load nodes according to the node behavior change position, determines the direction consistency, selects the trend synchronization point and calculates the power difference, extracts the time period where the difference exceeds the limit and compares it with the edge record to generate a time synchronization offset range set; The duplicate request screening module reads the node requests within the time synchronization offset range set, screens duplicate resource name records and groups them, extracts the request intensity and time of each group, performs product sorting and reconstructs the request order, and generates resource overlapping request results; The synchronous mutation extraction module calls the node power data in the resource overlap request result, compares the adjacent power changes with the mutation identification value, extracts the mutation time point and cross-checks it with the node behavior change position, filters the repeated time mutation nodes, and generates a synchronous mutation node identification group; The path line update module queries the power supply path and channel time consumption data according to the synchronous mutation node identification group, determines whether the time consumption exceeds the limit, selects the replacement channel and updates the path sorting, and generates a distributed photovoltaic power supply path change list.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, a behavior change sequence is constructed through the node change trend and mutation amplitude, the sequence difference is calibrated in combination with the direction consistency and offset range, the request intensity and time are used to construct a sorting structure to express the resource concentration characteristics, the continuous mutation time points are extracted and the synchronous node groups are cross-located, and the path load is combined to filter and replace the channel and adjust the path order. The effect achieved is to realize the precise node behavior identification, the orderly resource request response, the efficient synchronous event positioning and the dynamic path scheduling adaptation, thereby improving the response capability and power supply coordination in multi-node change scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] 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.

[0017] Figure 1 Schematic diagram of the steps of the present invention; Figure 2 It is a system module diagram of the present invention. DETAILED DESCRIPTION

[0018] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0019] 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.

[0020] 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.

[0021] 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.

[0022] 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.

[0023] See also Figure 1 , a distributed photovoltaic power supply method based on big data interconnection, comprising the following steps: S1: Collect time series data from PV nodes, compare the direction of value changes in consecutive time periods, record the turning point of the trend, construct a difference sequence, determine whether the degree of difference jump meets the mutation standard, merge the position that meets the conditions with the trend turning point, and obtain the position of node behavior change; S2: Extract the change direction of power generation and load nodes based on the location of node behavior changes, determine the consistency of the direction, locate the trend synchronization point, compare the trend difference of the synchronization point, mark the change offset area, and establish the time offset range based on the time period record of the photovoltaic edge node controller. Output the time synchronization offset range set; S3: Read the node request content within the time synchronization offset range set, filter out duplicate resource target request groups, extract the request strength and time value of each group, perform product sorting, reorganize the behavior request sequence in order, and output the resource overlapping request results; S4: Call the continuous running data of the resource overlap request result, identify the numerical mutation point, extract the mutation time set, cross-compare it with the node time set, verify the time overlap dense segment, and obtain the synchronous mutation node identification group; S5: Based on the node paths listed by the synchronous mutation node identification group, the current forwarding channel and load status are retrieved, the channel to be replaced is selected in combination with the photovoltaic communication routing repeater record, the path sequence is adjusted, and a distributed photovoltaic power supply path change list is output.

[0024] The location of node behavior changes includes trend turning time, difference mutation amplitude, and trend change direction. The time synchronization offset range set includes synchronization offset start and end time, trend difference range, and direction consistency identifier. The resource overlap request result includes duplicate resource target, request intensity value, and reorganization sorting index. The synchronous mutation node identification group includes mutation occurrence time, time overlap interval, and mutation node number. The distributed photovoltaic power supply path change list includes the channel number to be replaced, the adjusted path order, and the current load value of the channel.

[0025] The specific steps of S1 are: S101: Collect continuous time series data from the photovoltaic nodes, compare the increase and decrease relationship of the values at two adjacent time points, record the time point when the value change direction is reversed, and generate a trend change time sequence; Collect continuous time series data from the photovoltaic node. The specific operation is to collect the current output power data of the node every 10 seconds through the communication module. For example, 60 data points are collected continuously from 10:00 to 10:10. Each data point includes a timestamp and a corresponding power value. The data storage structure is in the form of an array list indexed by time. For example, the first item is [10:00:00, 98W], and the second item is [10:00:10, 102W]. When comparing them in sequence, starting from the second time point, the values are compared with the previous time point in sequence. For example, the second item 102W is compared with the first item 98W, and it is judged to be in an upward direction. The comparison of the third item 106W with the second item 102W is still increasing, until it is found that the power value at a certain time point starts to increase. It becomes smaller. For example, the 5th item 107W is in a downward direction relative to the 4th item 109W. At this time, it changes from continuous rise to fall, and the direction is reversed. The moment of the 4th item is recorded as the trend reversal point. Continue to compare backward. If the 6th item 104W and the 7th item 100W continue to fall, but the 9th item power rises to 101W after the 8th item 96W, then the corresponding time of the 8th item is recorded as a new reversal point. All judgments are based on the numerical relationship between the two adjacent points. The traversal operation is performed in sequence until the last data point. Whenever the trend direction changes from rising to falling or from falling to rising, it is judged that the trend is reversed and the corresponding time point is recorded. Through the above execution process, the trend change moments in the entire time series can be extracted and recorded one by one to form a complete trend change moment sequence.

[0026] S102: Calculate the difference between adjacent values based on the time series data, compare the difference with the previous and next differences one by one, determine whether it exceeds the difference jump amplitude threshold, and generate a sudden change position sequence; After obtaining the continuous time-series power data of the photovoltaic node, the difference between the values of the two adjacent time points is compared item by item, that is, starting from the second item, the first item is subtracted, and so on to obtain the difference sequence of each item. For example, if the input data is [98, 102, 106, 109, 107, 104, 100, 96, 101, 106]W, the difference sequence is [4, 4, 3, -2, -3, -4, -4, 5, 5]. Each difference is compared with the previous and next difference values for amplitude difference. For example, for the fourth difference value -2, the difference between it and the previous difference value 3 and the next difference value -3 is respectively subtracted to obtain the difference change amplitude of 5 and 1. To determine whether there is a sudden change, the difference jump amplitude threshold Tdiff needs to be set. The threshold is set according to the output fluctuation range of the photovoltaic equipment and the on-site interference situation. Usually, the photovoltaic output fluctuation amplitude is small on sunny days. The power level is within ±3W, but can fluctuate dramatically up to ±7W on cloudy days. To avoid misjudgment, Tdiff can be set to 6W as the transition judgment criterion. Points with a difference change greater than or equal to this threshold are considered transition points. For example, the eighth item in the difference sequence above is 5, which differs from the previous difference of -4 by 9 and from the next difference of 5 by 0. Since the difference in only one direction exceeds the threshold, the transition condition is not met. For example, if the power data of a node is [150, 155, 161, 145], the difference sequence is [5, 6, -16]. The third difference of -16 differs from the previous difference of 6 by 22 and is incomparable to the next value. Since the current point is at the boundary, the difference of 22 is much greater than Tdiff = 6, so this point is considered a transition point. This process is repeated for the entire data segment, forming a sudden transition position sequence consisting of all transition points.

[0027] S103: Based on the trend change time sequence and the mutation jump position sequence, extract all time points in the merged sequence as the behavior change points of the node, and generate the node behavior change position; After obtaining the trend change time sequence and the mutation jump position sequence, first load the two sequences into memory to form two list structures. In the example, the trend change time sequence is [10:03, 10:07, 10:12], and the mutation jump position sequence is [10:05, 10:07, 10:15]. Read all the elements of the two lists one by one, and summarize all the time points into a new list. During the execution process, for repeated time points such as 10:07, only record them once. Compare the elements in the two lists one by one through programming traversal, and avoid repeated elements. The elements are appended to the merged list, and the initial set is [10:03, 10:07, 10:12, 10:05, 10:15]. Then the list is sorted in chronological order, and arranged in ascending order by comparing layer by layer by hours, minutes, and seconds. Finally, the set of behavior change points arranged in chronological order is [10:03, 10:05, 10:07, 10:12, 10:15]. This process does not require calling a complex model and can be completed through only three steps of list traversal, deduplication, and sorting, ultimately forming a node behavior change position sequence.

[0028] The specific steps of S2 are: S201: Extract the value change direction of the power generation node and the load node at the same time point based on the node behavior change position, compare the increase and decrease trend directions one by one, select the time point set with consistent change direction, and generate a trend consistent time series; First, call each time point in the previously generated node behavior change position sequence, and traverse and call the power data of the power generation node and the power or current data of the load node corresponding to the time point one by one. Call the current value and the previous moment value of the power generation node and the load node at this time point respectively, and compare the two values. If the current value is greater than the previous value, it is judged to be an upward trend, otherwise it is a downward trend. For a specific example, if the power generation node is 160W at 10:05 and 155W at 10:04, it is considered that the power generation trend is rising. If the corresponding time of the load node is 170W and 175W respectively, the load is a downward trend. The two trend directions are represented by positive and negative signs, for example, rising Use "+" for upward movement and "-" for downward movement, then compare the directions of the two. If the directions are consistent (i.e., both are "+" or "-"), mark the time point as a point with consistent direction. If they are inconsistent, skip it and do not record the time point. Repeat the above operation for all the time points of node behavior change in the traversal process to form a preliminary screening list. Pay attention to processing boundary data in this process. For example, if a certain time point is the first or last position in the time series, it is necessary to determine whether it has data from the previous or next time point to support direction judgment. If the data integrity is not met, the time point will not be judged and will be removed. Finally, all time points that meet the conditions and have consistent directions are aggregated to form a trend consistent time series.

[0029] S202: Based on the time points in the trend-consistent time sequence, the numerical differences between the power generation node and the load node at the time points and the preceding and following time periods are called to calculate the upstream and downstream trend change values. The coefficients are compared with the trend deviation threshold, and the positions exceeding the threshold are extracted to obtain the trend offset interval set. The calculation formula for upstream and downstream trend change values is as follows: ; in, Represents the upstream and downstream trend change value at the i-th time point, represents the power of the power generation node at the i-th time point, represents the load node power at the i-1th time point, represents the power of the power generation node at the i+1th time point, represents the load node power at the i-th time point, Represents the value of the power generation node at the i-th time point, Represents the load node value at the i-th time point; Definition of upstream and downstream trend change values: The upstream and downstream trend change value is an indicator used to measure the power variation characteristics between power generation nodes and load nodes at a specific point in time and in adjacent time periods. This value comprehensively considers the difference in power values between the current point in time and the previous and subsequent points in time, reflecting the dynamic fluctuations between power supply and demand in the power grid within a local time period. This indicator not only reflects the rate of change of power in the temporal dimension but also includes the spatial transmission structure, that is, the corresponding relationship between power generation and load. It is an important indicator of the state changes of the power system in a short period of time.

[0030] This value has the following characteristics: Time correlation: By analyzing the difference in data between adjacent time points, the upward or downward trend is revealed; Spatial contrast: Expresses the degree of imbalance in energy flow through numerical comparisons between upstream and downstream nodes; Sensitivity: The value is sensitive to system fluctuations and can capture small disturbances; Stability control value: used to identify trend deviation areas and determine whether structural changes have occurred.

[0031] Calculation principle of upstream and downstream trend change values: The calculation process is based on the numerical differences between the power generation node and the load node at three consecutive key time points (previous time point, current time point, and next time point). The sum of the differences is used as the dominant quantity of the trend, and a correction factor based on the voltage difference is introduced to normalize the changes in the local power grid state.

[0032] The specific calculation logic is as follows: Local difference summation: Take the current point in time's generated power minus the previous point in time's load power, and the next point in time's generated power minus the current point in time's load power. The sum of these two results represents the net change in power supply and demand in the short term. Normalization: Because voltage level differences in different grid structures and time periods can cause nonlinear responses to power changes, the square root of the voltage difference is used to construct the normalization denominator. A constant term is introduced to prevent abnormal conditions such as division by zero, thereby achieving equivalent normalization of the power differences between different nodes. Take absolute values: avoid trends in different directions offsetting each other and emphasize the quantitative expression of fluctuation amplitude; This calculation principle combines the three dimensions of difference, fluctuation amplitude, and voltage state correction to accurately characterize the degree of trend deviation of the power system on a short-term scale. It is an important basic parameter for extracting trend deviation intervals and identifying abnormal periods. Parameter acquisition and value setting: : The power value of the power generation node at the i-th time point. It is obtained by monitoring the output power of the power generation node. According to the data of the Dinorwig pumped storage power station in the UK, its output power is 1.8GW. In this example, MW.

[0033] : The power value of the load node at the i-1th time point. It is obtained by monitoring the power consumption of the load node. According to the data of the National Renewable Energy Laboratory of the United States, the typical power consumption of the load node is about 1500MW. MW.

[0034] : The power value of the power generation node at the i+1th time point. Obtained through the prediction model or actual monitoring. MW.

[0035] : The power value of the load node at the i-th time point. It is obtained by monitoring the power consumption of the load node. MW.

[0036] : The voltage value of the power generation node at the i-th time point. It is obtained by measuring the voltage sensor. kV.

[0037] : The voltage value of the load node at the i-th time point. It is obtained by measuring the voltage sensor. kV.

[0038] Formula calculation process: Calculate the molecular part: ; Calculate the denominator: ; ; Result description: The results show that the upstream and downstream trend change value at the i-th time point is about 96.15. This value reflects the power change of the power generation node and the load node at adjacent time points, and takes into account the influence of voltage difference. The value may indicate that the power grid has a large power fluctuation at that point in time, and its impact on the grid stability needs to be further analyzed.

[0039] S203: Based on the time period position in the trend offset interval set, call the time period record data of the photovoltaic edge node controller, match the start and end time of the controller response record in the interval, extract the time span, and establish a time synchronization offset range set; According to the position of the time period in the trend offset interval set, the start and end time points in the set are read one by one, and the response log information recorded in the photovoltaic edge node controller is called in the form of timestamps to find all the response records of the controller in these time periods. For each trend offset interval, the timestamp in the controller response record is matched to see if it falls within the interval. If the start time of a response record is after the start point of the offset interval and the end time is before the end point of the interval, it is considered that the response record completely falls within the offset interval. If the start or end time of the response record slightly exceeds the offset interval range but is within 1 minute, it is also considered a partial match. The time span extraction method is to record the start and end time difference of the controller response, and Convert the response duration into minutes or seconds. This process requires addressing time format differences in response records. A unified timestamp format is used for comparison and matching. Extract time periods within all offset intervals where controller response records exist. Record their start time, end time, and duration to form a complete set of time synchronization offset ranges. For example, if an offset interval is from 10:15 to 10:20 and the controller response record is from 10:16 to 10:18, the segment is successfully matched and extracted as a 2-minute time span. This process continues, completing the response matching and time span extraction process for all intervals one by one, and ultimately constructing a set of time synchronization offset ranges.

[0040] The specific steps of S3 are: S301: Read the node request content within the time synchronization offset range set, extract the resource target identifier in each request, compare the duplication of the resource target identifier, filter multiple request records with the same resource target, and establish a resource target duplicate request group list; First, extract the start and end time of each time period from the time synchronization offset range set output in the previous stage, and call the resource request records initiated by the power generation node or load node within the time range, traverse these request data one by one, and extract the resource target identification field recorded in each request. This field is usually a resource unique code or name string, such as "RES_A01", "RES_B05", etc., compare the resource identification values ​​after all records are extracted one by one, and use the string equivalence comparison method to confirm which requests have the same resource target identification. For request records with exactly the same resource identification, record their corresponding original request number and timestamp, and classify them into the same group. Multiple requests are constructed through multiple cycles. The resource target identifiers of all requests within each group are exactly the same. For determining the same resource target, fuzzy matching is not used but strict character equality comparison to ensure the consistency of resource entities in each group. For example, if three request records appear in the offset interval [10:10–10:20] and their resource target fields are all "RES_X01", then these three records are classified as a duplicate request group. If there are two requests in another group and both are "RES_Y01", another group is formed. Resource target requests that appear only once do not form a group and are directly removed from subsequent processing. Finally, a duplicate request group list is established with the resource target as the index. Each element contains the resource identifier, all request numbers in the group and their corresponding time.

[0041] S302: Repeat the request group list according to the resource target, extract the intensity value and time stamp of the request in each group, calculate the product index of the intensity and time value, sort the requests by the product value, construct a request structure arranged by the product value, and generate a behavior request sequence; The specific calculation formula for the product of intensity and time value is: ; in, Represents the product of the intensity and time value of request q in resource group r, represents the intensity value of request q in resource group r, represents the intensity value of the kth request in resource group r except request q, Represents the number of requests in resource group r except request q, represents the sum of the strength values of the remaining requests in resource group r, Represents the timestamp of request q in resource group r, represents the average value of all request timestamps in resource group r, The absolute value of the difference between the timestamp of request q and the average timestamp in resource group r; The product of intensity and time measures the weight of a resource request at a specific point in time. It combines the intensity of a single request with its corresponding timestamp. By integrating the request's temporal position within the resource group and its intensity, it reflects the relative importance of the request within the overall behavior sequence. This metric not only considers the intensity of the request itself but also integrates the timing differences between the request and other requests in the same group.

[0042] The indicator's calculation principle is based on the following logic: Intensity: The intensity value of a single request directly reflects resource consumption and is typically obtained by counting the number of bytes generated by each request response in the server log. A larger value indicates a more resource-intensive request.

[0043] Time layer: The timestamp reflects the time when the request occurs in the entire resource behavior sequence. The difference between the request timestamp and the average timestamp of requests in the same group is calculated to obtain the timing offset of the request within the group, thereby characterizing the degree of synchronization between the request and the behavior pattern within the group.

[0044] Numerator composition: The strength value of the request is added to the sum of the strengths of other requests in the same group to form an extension of the comprehensive expression of the request in the local strength structure, and then multiplied by its timestamp. This allows requests initiated earlier or later to obtain different degrees of product amplification based on their strength performance.

[0045] Denominator adjustment: A timing difference factor is constructed using the absolute value of the difference between the request and the average timestamp of the resource group it belongs to. This factor is then scaled using a square root method. This ensures that requests that are closer in time to other requests in the group have a more significant impact on the final metric. Otherwise, the product weight is reduced, thereby reducing the ranking weight of requests that deviate from the main behavior rhythm.

[0046] Absolute value processing: Finally, the absolute value of the product value is taken to ensure that the product index is uniformly non-negative in the behavioral feature expression, which facilitates subsequent sorting and comparison operations.

[0047] Overall, the product of intensity and time value is a composite metric that integrates intensity scale and temporal location characteristics. It quantifies the significance ranking of each request in resource access behavior through the weighted accumulation of local request intensity structure and synchronous analysis of global behavior rhythm.

[0048] Parameter acquisition and value setting: Request Strength :Get the number of bytes of response to each request by analyzing the web server access log. For example, if the response size of a request is 2048 bytes, then .

[0049] The sum of the strength of other requests in the same group : Sum the response bytes of all requests except the current request in the same resource group. For example, if the response sizes of the other three requests are 1024, 512, and 256 bytes respectively, the sum is byte.

[0050] Request timestamp : Extract the request timestamp from the server log and convert it to the number of seconds since the UNIX epoch. For example, a timestamp of 05:00:00 on June 4, 2025, converts to 1749022800 seconds.

[0051] Resource group average timestamp : Average the timestamps of all requests in the resource group. For example, if the timestamps of four requests in the resource group are 1749022800, 1749022860, 1749022920, and 1749022980 seconds, the average value is Second.

[0052] Formula calculation process: Calculate the molecular part: ; Calculate the denominator: ; Calculate the product value : ; Result description: The result shows that the product value of request q in resource group r is approximately 705264000000. This value is used to sort the requests, build a request structure arranged by product value, and generate a behavioral request sequence.

[0053] S303: Invoke the resource target and time information requested in the behavior request sequence, screen request groups with overlapping time periods, locate the occupation of the same resource by multiple requests in the same time period, and establish a resource overlapping request result; The resource target and time information requested in the call behavior request sequence are first read according to the resource target classification in each request group, and then the time periods of each request in the group are compared. The time period includes the start and end time. If a record only contains a time point, the duration must be deduced according to its request type. For example, the default duration of a normal request is 60 seconds. After obtaining the time period of each request, a pairwise comparison operation is performed on all requests in the group to compare whether there is overlap in their time periods. The specific judgment is: if the start time of one request is less than the end time of another request, and its end time is greater than the start time of another request, it is considered that there is time overlap between the two requests, accounting for 60 seconds. Using the same resource, all request combinations within the group are traversed and processed according to the above judgment logic. For example, if the time period of request 1 is [600, 660] seconds and the time period of request 2 is [630, 690] seconds, since 630≤660 and 600≤690 exist, the overlap condition is met and they are added to the resource overlap tag structure. The structure must include fields such as the resource target identifier, the list of conflicting request numbers, and the conflicting time period range. Conflict sub-results are generated for all overlapping requests in each group. If there are multiple conflicting request pairs within a group, all overlapping combinations are recorded separately. Finally, the conflicting request combinations on all resource targets are merged and sorted to establish the resource overlap request result.

[0054] The specific steps of S4 are: S401: Call the continuous operation data in the resource overlap request result, extract the numerical difference of the node at adjacent time points, compare the change range of the difference value item by item, and judge it with the set numerical jump threshold. Filter the time points where the jump amplitude exceeds the threshold to generate a mutation time set; To call the continuous running data in the resource overlap request result, first extract the relevant node identifiers and their corresponding continuous running data one by one from the established resource overlap request result, and read the running values of the nodes at two adjacent time points in turn. The values can be current, voltage or power, etc., and use the sequential traversal method to start from the second item and calculate the difference between each item and the previous item, that is, the current value is subtracted from the previous value to get the numerical difference, and then compare the obtained difference with the previous and next differences to see the change range. The specific operation is to compare the absolute difference between the current difference and the previous difference, and the absolute difference between the current difference and the next difference. For example, the power values of the node at time points 10:00, 10:01, and 10:02 are 120W, 125W, and 140W, respectively, and the adjacent differences are 5W. The difference between the power consumption and 15W is 10W. This value is compared with the set value jump threshold. If the change amplitude is greater than the jump threshold, the current time point is determined to be a mutation point. The jump threshold is set based on the device measurement accuracy and the actual operation fluctuation pattern. If the normal change of the device is within ±6W, the jump threshold can be set to 10W to identify mutation behaviors that exceed the normal fluctuation range. For example, if the power change suddenly increases from 130W to 160W within a sampling period, the difference is 30W, which is far greater than the threshold of 10W. This point can be marked as a mutation point. The above difference acquisition and difference amplitude comparison process is performed for each time point in the continuous record. If the result is greater than the set threshold, the time point is added to the mutation time set. Finally, the traversal is completed and all mutation time sets that meet the conditions are generated.

[0055] S402: Based on the mutation time set, call the time record information of the corresponding node, match each mutation time point with the time value in the node running time series, identify the segments where multiple node mutation times overlap, and obtain the time overlap dense segment interval; According to the mutation time set, read the time points in the set one by one, and call the node identifier corresponding to each mutation point and the complete time series record information of the node, and match the mutation time point with the node's running time series one by one. The specific method is to compare the mutation time point with the time field in the node running record in timestamp format one by one. If the match is successful, it is recorded that the node has mutation behavior at this time point, and this information is appended to the matching result table. Continue to traverse all nodes and mutation times backward to build a mapping table between all mutation behavior times and nodes, and then aggregate the mapping table by time dimension to count the number of nodes with mutation behavior at each time point. If the number of nodes corresponding to a certain time point is greater than 2, it means that there are multiple nodes that have mutation behavior at the same time point. If there is a mutation, it is considered that there is time overlap at that time point. All time points that meet this condition are collected in sequence, and further judgment is made on whether these time points are continuous. If the interval between consecutive time points is less than the set tolerance value, for example, the interval does not exceed 1 minute, they are merged into one time period to form a time overlap dense segment interval. For example, if there are three time points at 10:02, 10:03, and 10:05 in the mutation set, the first two points can be merged into the [10:02–10:03] segment due to a difference of 1 minute, while 10:05 and 10:03 have a difference of 2 minutes and do not meet the merging condition, so they are used as the starting points of a new segment. Finally, the aggregation and continuity judgment of all mutation times are completed, forming segments with multiple node mutation times overlapping, and constructing a time overlap dense segment interval.

[0056] S403: For each time period in the time-dense overlap interval, extract the node identifiers of nodes that experience simultaneous mutation behaviors, remove duplicate nodes, and sort them by time. Node sets with synchronous mutation characteristics are extracted to establish a synchronous mutation node identifier group. For each time period in the time overlap dense segment, read the start and end time of the segment one by one and query the node mutation mapping table for all node identifiers that have mutated in the segment, extract all node numbers that meet the time range conditions, and form an initial node list. In this list, the node numbers are deduplicated, and after removing duplicate records, a unique node identifier set is obtained. Then, each node mutation record is grouped according to the time dimension to ensure that the mutation behavior of the same node in different time periods is classified separately. For example, if node A mutates at 10:02 and 10:03, the two records are classified into the corresponding time points respectively, and all time periods are grouped. Execute the node clustering operation to obtain a set of independent nodes with mutation behavior in each segment, and then sort the node numbers in each set in ascending order to make the output result structure consistent and easy to compare. For example, the node list in the time period [10:02–10:03] is [A, C, B, A], which is [A, B, C] after deduplication, and then sorted in lexicographic order to [A, B, C]. Finally, a list of nodes with mutation behavior occurring simultaneously in the same time period is obtained. The node collection, deduplication, and sorting operations are completed for all time periods in sequence to construct multiple independent sets of synchronous mutation nodes, which are summarized and output as a synchronous mutation node identification group.

[0057] The specific steps of S5 are: S501: Based on the node path information listed in the synchronous mutation node identifier group, extract the forwarding channel identifier and corresponding load value currently used by the node in the path, search the channel and load combinations in sequence according to the node path order, and establish a channel load mapping table; First, read each group of node numbers in the synchronous mutation node identification group, and call the path data structure of each node at the current moment in turn. The path data structure must clearly record the forwarding channel number sequence used in the communication path to which the node belongs. For example, the path of path node A is [CH_01, CH_03, CH_05], which means that node A sends information outward through three forwarding channels in turn. While obtaining each channel number during the call process, further extract the current load value of the node on the channel from the node operation data. The load value can be the traffic data (such as kbps) or power transmission (such as W) transmitted on the current channel. For each node, the channel number and its corresponding real-time load value are extracted in the order of the path to form a channel and load For example, the path of node B is [CH_02, CH_04], and its corresponding loads are 120 kbps and 190 kbps, respectively. This correspondence is written into the structure. Using "node identifier-channel number-load value" as the basic recording unit, the nodes and their path channel information in all synchronous mutation node identifier groups are traversed in sequence, and the combined information of all relevant channels and their load values is accumulated. In the process, attention should be paid to the situation where there are repeated channels or abnormal values in the path. Channel numbers need to be deduplicated, and channel data with invalid load values should be excluded. For example, if a channel returns a load value of "-1" or "NULL", it should be eliminated. Finally, all valid channel numbers and corresponding load values are recorded and organized into a complete channel-load mapping table.

[0058] S502: Based on the channel load value in the channel load mapping table, the channel status record of the photovoltaic communication routing repeater is called, the channel activation status identifier within the same period is extracted, and the channel load value and the activation status are compared to see whether they are both within the abnormal range. The channel numbers that meet the replacement conditions are selected to generate a set of replaceable channels; First, read the channel number and its corresponding load data in the table one by one, and then call the channel status record data of the photovoltaic communication routing repeater according to the timestamp range of each record. The record must include the enabled status field of each channel in each time period. The field is recorded in Boolean or status code format, such as "0" for disabled, "1" for enabled, "2" for fault termination, etc. For each channel record, first compare its load value to see if it falls into the abnormal load range. The abnormal range needs to be set with the threshold value of the communication system design capacity. For example, if the maximum design bandwidth of a channel is 200kbps, the abnormal load threshold can be set to ≥180kbps. If the current load value of a channel is 185kbps, it is judged to be abnormal, and then compare its enabled state. The status is marked as abnormal. For example, a status code of "2" indicates a communication failure. If the load value exceeds the limit and the enabled status is abnormal at the same time, the channel meets the replacement condition and its channel number is added to the preliminary list of replaceable channels. When setting the abnormality judgment threshold, a safety boundary should be set according to the design tolerance range. For example, the high load threshold is set to 90% × maximum bandwidth. At the same time, the repeater status record should be completely consistent with the load record time period, and time alignment compensation is performed for the inconsistent part. For example, if the repeater status is updated every 5 minutes and the load sampling is every 1 minute, the repeater record time period is used as the basis for aggregation judgment. Finally, all channel numbers that meet the "load abnormality + status abnormality" are recorded to generate a set of replaceable channels.

[0059] S503: Calling the node path information corresponding to the replaceable channel set, comparing the channel number sequence of the path with the replacement number position, rearranging the channel sequence, extracting the changed path number structure, and establishing a distributed photovoltaic power supply path change list; First, read each channel number in the replaceable channel set, and search the original node path structure corresponding to the channel in the synchronous mutation node path record, confirm the position number of each channel one by one, for example, in the path [CH_01, CH_05, CH_07], if CH_05 is a replaceable channel, then its position in the original path is the second item, then call the system's spare channel list, select the replaceable channel number to replace the second position in the original path, and complete the path rearrangement operation, for example, CH_05 is replaced by CH_06, and the changed path is [CH_01, CH_06, CH_07 ], for the path change operation, it is necessary to ensure that the logical sequence of the replaced path number remains unchanged, that is, loops or skipping numbers are not allowed in the path. For example, if the original path number is CH_01→CH_02→CH_03, the replaced number must also be continuously increased or arranged according to the design rules. The same process is performed on all node paths involved in the replacement operation, and the "find → locate → replace → rearrange" operation steps are performed item by item. After the rearrangement is completed, the new path number sequence is output, and the structure before and after the path change is recorded in units of nodes. Finally, a complete list of changed paths is established according to the node number, and the output is a distributed photovoltaic power supply path change list.

[0060] See also Figure 2 , a distributed photovoltaic power supply system based on big data interconnection, including: The node trend extraction module obtains the time series values of PV node power and load, compares the power change direction of adjacent time periods and records the reversal time, calculates the power difference and compares it with the fluctuation identification benchmark value, extracts the time point when the change amplitude exceeds the limit, integrates the reversal and mutation time, and generates the node behavior change location; The behavior offset comparison module extracts the power direction of the power generation and load nodes based on the location of the node behavior change, determines the direction consistency, selects the trend synchronization points and calculates the power difference. It extracts the time period where the difference exceeds the limit and compares it with the edge record to generate a time synchronization offset range set. The duplicate request screening module reads node requests within the time synchronization offset range, screens duplicate resource name records and groups them, extracts the request intensity and time of each group, performs product sorting and reconstructs the request order, and generates resource overlapping request results; The synchronous mutation extraction module calls the node power data in the resource overlap request results, compares adjacent power changes with mutation identification values, extracts mutation time points, cross-references them with the node behavior change locations, filters out nodes with repeated time mutations, and generates synchronous mutation node identification groups. The path line update module queries the power supply path and channel time consumption data based on the synchronous mutation node identification group, determines whether the time consumption exceeds the limit, screens the replacement channel and updates the path sorting, and generates a distributed photovoltaic power supply path change list.

[0061] 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 distributed photovoltaic power supply method based on big data interconnection, characterized in that: The following steps are involved: S1: Collect PV node time series data, compare the direction of value changes in consecutive time periods, record trend turning points, construct a difference sequence, determine whether the jump meets the mutation standard, merge the mutation and turning positions, and obtain the node behavior change position; S2: Based on the node behavior change position, extract the change direction of the power generation and load nodes, determine the consistency, locate the trend synchronization point, compare the trend difference, mark the change offset area, combine the edge controller records, and output the time synchronization offset range set; S3: Read the time synchronization offset range set, filter duplicate resource target request groups, extract strength and time values, perform product sorting, and output resource overlap request results; S4: Calling the continuous running data in the resource overlap request result, identifying the numerical mutation point, extracting the mutation time set, and cross-comparing it with the node time set to verify the time overlap dense segment, and obtaining the synchronous mutation node identification group; S5: According to the node paths listed by the synchronous mutation node identification group, the current forwarding channel and load status are retrieved, and the channels to be replaced are screened in combination with the communication routing repeater records, and a distributed photovoltaic power supply path change list is output.

2. The distributed photovoltaic power supply method based on big data interconnection according to claim 1 is characterized in that: The node behavior change position includes the trend turning time, the difference mutation amplitude, and the trend change direction; the time synchronization offset range set includes the synchronization offset start and end time, the trend difference range, and the direction consistency identifier; the resource overlap request result includes the duplicate resource target, the request strength value, and the reorganization sorting index; the synchronous mutation node identification group includes the mutation occurrence time, the time overlap interval, and the mutation node number; the distributed photovoltaic power supply path change list includes the channel number to be replaced, the adjusted path order, and the current channel load value; Definition of mutation criteria: When the mutation amplitude of the node behavior value reaches the preset change threshold, it indicates a value jump; Definition of the change deviation area: The area marked when the direction of node behavior change is inconsistent with the direction of change of power generation and load nodes.

3. The distributed photovoltaic power supply method based on big data interconnection according to claim 1 is characterized in that: The specific steps of S1 are: S101: Collect continuous time series data from the photovoltaic nodes, compare the increase and decrease relationship of the values at two adjacent time points, record the time point when the value change direction is reversed, and generate a trend change time sequence; S102: Calculate the difference between adjacent values based on the time series data, compare the difference with the previous and next differences one by one, determine whether it exceeds the difference jump amplitude threshold, and generate a sudden change position sequence; S103: Based on the trend change time sequence and the sudden change position sequence, extract all time points in the merged sequence as node behavior change points, and generate node behavior change positions; Definition of the difference jump amplitude threshold: the difference amplitude limit used to determine whether the difference in the node behavior value has undergone a sudden change.

4. The distributed photovoltaic power supply method based on big data interconnection according to claim 1 is characterized in that: The specific steps of S2 are: S201: extracting the direction of value changes of the power generation node and the load node at the same time point based on the node behavior change position, comparing the increase and decrease trend directions one by one, selecting a set of time points with consistent change directions, and generating a time sequence with consistent trends; S202: Based on the time point in the trend-consistent time sequence, the numerical difference between the power generation node and the load node at the time point and the preceding and following time periods is called to calculate the upstream and downstream trend change values, and the coefficient is compared with the trend deviation threshold, and the position exceeding the threshold is extracted to obtain the trend offset interval set; S203: Based on the time period position in the trend offset interval set, call the time period record data of the photovoltaic edge node controller, match the start and end time of the controller response record in the interval, extract the time span, and establish a time synchronization offset range set; Definition of increasing and decreasing trend direction: the increasing and decreasing trend of the node value at consecutive moments compared to the previous moment; Definition of trend deviation threshold: It is used to determine whether there is a deviation in the node power generation and load change trend.

5. The distributed photovoltaic power supply method based on big data interconnection according to claim 4 is characterized in that: The calculation formula for the upstream and downstream trend change value is specifically: ； in, Represents the upstream and downstream trend change value at the i-th time point, represents the power of the power generation node at the i-th time point, represents the load node power at the i-1th time point, represents the power of the power generation node at the i+1th time point, represents the load node power at the i-th time point, Represents the value of the power generation node at the i-th time point, Represents the load node value at the i-th time point.

6. The distributed photovoltaic power supply method based on big data interconnection according to claim 1 is characterized in that: The specific steps of S3 are: S301: Read the node request content within the time synchronization offset range set, extract the resource target identifier in each request, compare the duplication of the resource target identifier, filter multiple request records with the same resource target, and establish a resource target duplicate request group list; S302: extracting the intensity values and timestamps of the requests in each group according to the resource target repeat request group list, calculating the product index of the intensity and time value, sorting the requests by the product value, constructing a request structure arranged by the product value, and generating a behavior request sequence; S303: calling the resource target and time information requested in the behavior request sequence, screening the request groups with overlapping time periods, locating the occupation of the same resource by multiple requests in the same time period, and establishing a resource overlapping request result; Definition of multiple request records for the same resource target: a set of requests that point to the same resource identifier within the time synchronization offset range.

7. The distributed photovoltaic power supply method based on big data interconnection according to claim 6 is characterized in that: The specific calculation formula for the product index of intensity and time value is: ； in, Represents the product of the intensity and time value of request q in resource group r, represents the intensity value of request q in resource group r, represents the intensity value of the kth request in resource group r except request q, Represents the number of requests in resource group r except request q, represents the sum of the strength values of the remaining requests in resource group r, Represents the timestamp of request q in resource group r, represents the average value of all request timestamps in resource group r, Indicates the absolute value of the difference between the timestamp of request q and the average timestamp of resource group r.

8. The distributed photovoltaic power supply method based on big data interconnection according to claim 1 is characterized in that: The specific steps of S4 are: S401: Calling the continuous operation data in the resource overlap request result, extracting the numerical difference of the node at adjacent time points, comparing the change range of the difference value range item by item, and judging it with the set numerical jump threshold, screening the time points where the jump amplitude exceeds the threshold, and generating a mutation time set; S402: Based on the mutation time set, call the time record information of the corresponding node, match each mutation time point with the time value in the node running time series, identify the segments where multiple node mutation times overlap, and obtain the time overlap dense segment interval; S403: For each time period in the time-intensive overlap interval, extract node identifiers of nodes that experience simultaneous mutation behaviors, remove duplicate nodes, and sort them by time, extract a set of nodes with synchronous mutation characteristics, and establish a synchronous mutation node identifier group; Definition of the range of variation of the difference value: the fluctuation range of the difference between the values of adjacent time points of a node in a short period of time; Definition of numerical jump threshold: the limit of the numerical change amplitude of the numerical mutation; Definition of a node set with synchronous mutation characteristics: a node identification group whose values of multiple nodes mutate within the same time period.

9. The distributed photovoltaic power supply method based on big data interconnection according to claim 1, characterized in that: The specific steps of S5 are: S501: extracting the forwarding channel identifiers and corresponding load values currently used by the nodes in the path based on the node path information listed in the synchronous mutation node identifier group, searching for channel and load combinations in sequence according to the node path order, and establishing a channel load mapping table; S502: Based on the channel load value in the channel load mapping table, calling the channel status record of the photovoltaic communication routing repeater, extracting the channel activation status identifier within the same time period, comparing the channel load value and the activation status to see whether they are both within the abnormal range, screening the channel numbers that meet the replacement conditions, and generating a set of replaceable channels; S503: Calling the node path information corresponding to the replaceable channel set, comparing the channel numbering sequence of the path with the replacement numbering position, rearranging the channel sequence, extracting the changed path numbering structure, and establishing a distributed photovoltaic power supply path change list; Definition of abnormal range: The numerical range in which the channel load value and enable status value exceed the normal operating threshold.

10. A distributed photovoltaic power supply system based on big data interconnection is characterized by: According to any one of claims 1 to 9, the distributed photovoltaic power supply method based on big data interconnection comprises: The node trend extraction module obtains the time series values of PV node power and load, compares the power change direction of adjacent time periods and records the reversal time, calculates the power difference and compares it with the fluctuation identification benchmark value, extracts the time point when the change amplitude exceeds the limit, integrates the reversal and mutation time, and generates the node behavior change location; The behavior offset comparison module extracts the power direction of the power generation and load nodes according to the node behavior change position, determines the direction consistency, selects the trend synchronization point and calculates the power difference, extracts the time period where the difference exceeds the limit and compares it with the edge record to generate a time synchronization offset range set; The duplicate request screening module reads the node requests within the time synchronization offset range set, screens duplicate resource name records and groups them, extracts the request strength and time of each group, performs product sorting and reconstructs the request order, and generates resource overlapping request results; The synchronous mutation extraction module calls the node power data in the resource overlap request result, compares the adjacent power changes with the mutation identification value, extracts the mutation time point and cross-checks it with the node behavior change position, filters the repeated time mutation nodes, and generates a synchronous mutation node identification group; The path line update module queries the power supply path and channel time consumption data according to the synchronous mutation node identification group, determines whether the time consumption exceeds the limit, selects the replacement channel and updates the path sorting, and generates a distributed photovoltaic power supply path change list.

Images