An intelligent power grid big data analysis system and method

Abstract

The application relates to the technical field of power grid data analysis, in particular to an intelligent power grid big data analysis system and method; the system comprises the following: a data acquisition edge access unit collects power terminal data, user load data and environment data of a distribution room and a micro grid based on Internet of Things, and performs edge buffer preprocessing; a time series data storage management unit performs hierarchical management and time series down-sampling processing on the power terminal data, the user load data and the environment monitoring data; a distribution room diagnosis monitoring unit introduces a disturbance trajectory permutation entropy index to construct a disturbance evolution trend prediction model, and constructs a trip event causal matrix model to identify a distribution-induced micro grid linkage trip fault; and a micro grid diagnosis monitoring unit uses a user behavior disturbance analysis method to monitor a micro grid rebound risk during island switching. The application judges a trip fault in a linkage scene of a distribution room and a micro grid and early warns a user-side rebound risk in a micro grid island state.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power grid data analysis, and particularly relates to a smart grid big data analysis system and method. BACKGROUND

[0002] In the existing smart grid system, with the wide deployment of distributed energy, energy storage devices and microgrids, the coordinated operation between the distribution network and the microgrid and the power data monitoring and analysis have become an important research direction to ensure the safety and flexibility of the regional power system.

[0003] The existing power grid data analysis mainly relies on single-point fault monitoring systems, such as integrated protection devices, signal screens and trip record information of intelligent micro-break devices, combined with switch state for static alarm judgment. However, due to the large number of fault devices in multi-hop linkage trip-out fault events, the main fault device is difficult to identify, and users do not realize that the microgrid is in an island state, which may lead to difficulties in troubleshooting trip-out event faults and risks such as backlashes after island switching of the microgrid. SUMMARY

[0004] The purpose of the present application is to provide a smart grid big data analysis system and method to solve the problem of the large number of fault devices in multi-hop linkage trip-out fault events, the difficulty in identifying the main fault device, and the fact that users do not realize that the microgrid is in an island state, which may lead to difficulties in troubleshooting trip-out event faults and risks such as backlashes after island switching of the microgrid.

[0005] To achieve the above purpose, the present application provides a smart grid big data analysis system, comprising:

[0006] A data acquisition edge access unit, which collects power terminal data, user load data and environmental data of the distribution room and the microgrid based on the Internet of Things, and performs edge buffer preprocessing;

[0007] A time series data storage management unit, which is used for hierarchical management and time series down-sampling processing of the power terminal data, user load data and environmental monitoring data;

[0008] A distribution room diagnosis and monitoring unit, which constructs a disturbance evolution trend prediction model based on the power terminal data of the distribution room, introduces a disturbance trajectory permutation entropy index, identifies weak faults of power equipment in the distribution room, and constructs a trip-out event causal matrix model to identify distribution-induced microgrid linkage trip-out faults, and feeds back to the microgrid diagnosis and monitoring unit;

[0009] The micro-grid diagnosis monitoring unit is based on micro-grid power terminal data and user load data, constructs a micro-grid operation stability scoring system to monitor micro-grid stability risks, and uses a user behavior disturbance analysis method to monitor micro-grid rebound risks during island switching.

[0010] Preferably, the data acquisition edge access unit includes a multi-source device access acquisition module and an edge transformation message buffer module.

[0011] The multi-source device access acquisition module remotely accesses power terminal devices, user power consumption devices and environmental monitoring devices in the power distribution room and the micro-grid based on the Internet of Things, and acquires power terminal data, user load data and environmental monitoring data of the power distribution room and the micro-grid.

[0012] The power distribution room power terminal data includes voltage, current, frequency and protection action signals; the micro-grid power terminal data includes photovoltaic output power, energy storage SOC energy storage power; the environmental data includes temperature, humidity, weather station wind speed and solar radiation intensity; the user load data includes charging pile power, user side intelligent electric meter and power consumption circuit breaker power.

[0013] The edge transformation message buffer module is used for communication protocol standardization conversion of the collected power terminal data, user load data and environmental monitoring data, and time stamp alignment, validity judgment and weak update elimination.

[0014] Preferably, the time series data storage management unit includes a hierarchical storage module and a dynamic down-sampling module.

[0015] The hierarchical storage module is used for hierarchical storage of power terminal data, user load data and environmental monitoring data of the power distribution room and the micro-grid.

[0016] The dynamic down-sampling module is used for dynamic feature recognition and down-sampling processing of high-frequency time series data.

[0017] Preferably, the power distribution room diagnosis monitoring unit includes an abnormal trend prediction module and a trip causal chain tracing module.

[0018] The abnormal trend prediction module introduces a disturbance trajectory permutation entropy index to construct a disturbance evolution trend prediction model based on power distribution room power terminal data, and identifies early weak faults of power equipment in the power distribution room.

[0019] The disturbance trajectory permutation entropy index is an index for quantifying the disorder degree of short-term symbolization mode in time series, and is used as an index for identifying early weak faults of power equipment in the abnormal trend prediction module.

[0020] The tripping cause chain tracing module constructs a tripping event cause matrix model based on the protection action signal and the tripping record to identify the distribution-induced microgrid linkage tripping fault.

[0021] Preferably, the disturbance evolution trend prediction model is realized based on the disturbance trajectory permutation entropy index, the equipment disturbance evolution trend score and the micro-disturbance mutation dynamic score, and is used for the abnormal trend prediction module to identify the early weak fault of the power equipment in the distribution room, and specifically as follows:

[0022] S3.1.1, set a sliding window to segment the power terminal data in the distribution room, extract the relative size order of the power terminal data in the distribution room in each sliding window, construct a symbolic permutation pattern set, and calculate the disturbance trajectory permutation entropy index of the previous sliding window period by counting the frequency distribution of each type of permutation pattern;

[0023] S3.1.2, based on the disturbance trajectory permutation entropy index of each sliding window period, give each sliding window a different time weight, and weight and accumulate the disturbance trajectory permutation entropy indexes of all sliding window periods to calculate the equipment disturbance evolution trend score;

[0024] S3.1.3, compare the change amplitude of the disturbance trajectory permutation entropy index of the current sliding window period and the disturbance trajectory permutation entropy index of the previous sliding window period, and then normalize to obtain the micro-disturbance mutation dynamic score;

[0025] S3.1.4, weight the equipment disturbance evolution trend score and the micro-disturbance mutation dynamic score to finally obtain the comprehensive micro-disturbance risk score;

[0026] Based on the disturbance trajectory permutation entropy index and the comprehensive micro-disturbance risk score, when the disturbance trajectory permutation entropy index of the power terminal shows an upward trend and at the same time triggers the comprehensive micro-disturbance risk score threshold, it is determined that the power terminal is in an early weak fault state.

[0027] Preferably, the tripping event cause matrix model is a two-dimensional structured matrix model representing the possible causal relationship between the tripping behaviors of multiple power equipment, and is used to judge the tripping fault in the tracing distribution room and microgrid linkage scene, and specifically as follows:

[0028] S3.2.1, accept the protection action signal, the tripping flag record and the switch opening and closing state;

[0029] S3.2.2, set a time tolerance, and group the tripping events occurring within the time tolerance into multiple tripping event groups;

[0030] S3.2.3, in each multi-trip event group, a time-causal score matrix between devices is calculated based on the trip timestamp of each power device, and a topology-causal score matrix between devices is calculated based on the electrical path length of each power device;

[0031] S3.2.4, the time-causal score matrix and the topology-causal score matrix between devices are weightedly fused to obtain a joint-causal score matrix between devices, and the device with the largest cumulative-causal score to other trip events is selected as the main fault device of the trip event of the multi-trip event group in the joint-causal score matrix, and the main fault device and the topology link structure of the multi-trip event group are fed back to the micro-grid diagnosis and monitoring unit.

[0032] Preferably, the micro-grid diagnosis and monitoring unit comprises a multi-energy stability evaluation module and an island switching rebound prediction module;

[0033] The multi-energy stability evaluation module calculates a fused stability confidence value containing all micro-grid power terminal data based on micro-grid power terminal data using fuzzy logic rules , and combines the introduced energy disturbance accumulated potential energy to comprehensively construct a micro-grid operation stability score system to calculate a micro-grid operation stability score value to monitor the stability risk of the micro-grid and feed back the micro-grid operation stability score value to the island switching rebound prediction module;

[0034] wherein, is time;

[0035] The energy disturbance accumulated potential energy is the integral of the disturbance variable in the micro-grid power terminal data within a fixed time in history.

[0036] Preferably, the island switching rebound prediction module is used to monitor the rebound risk of the micro-grid after island switching in the off-grid island state of the micro-grid using a user behavior disturbance analysis method;

[0037] The user behavior disturbance analysis method is used to analyze the rebound risk process in the island state of the micro-grid.

[0038] Preferably, the user behavior disturbance analysis method is as follows:

[0039] The user load surge risk value of each user power node is calculated based on user load data , and the micro-grid operation stability score value is combined to calculate a comprehensive user-side disturbance score, and the micro-grid rebound early warning is performed according to the comprehensive user-side disturbance score;

[0040] wherein, is time; indicates the user power node index.

[0041] In another aspect, the present application provides a smart grid big data analysis method for the above-mentioned smart grid big data analysis system, comprising the following steps:

[0042] S10.1, collecting power terminal data, user load data and environmental data of the distribution room and micro-grid based on the Internet of Things, and performing edge buffer preprocessing;

[0043] S10.2, hierarchical management and time series down-sampling processing of the power terminal data, user load data and environmental monitoring data;

[0044] S10.3, based on the power terminal data of the distribution room, introducing a disturbance trajectory permutation entropy index to construct a disturbance evolution trend prediction model, identifying weak faults of power equipment in the monitored distribution room, and constructing a trip event causal matrix model to identify distribution-induced micro-grid linkage trip faults;

[0045] S10.4, based on the micro-grid power terminal data and user load data, constructing a micro-grid operation stability scoring system to monitor the stability risk of the micro-grid, and using a user behavior disturbance analysis method to monitor the micro-grid rebound risk during island switching.

[0046] Compared with the prior art, the above technical solutions of the present application have the following beneficial technical effects:

[0047] 1. In the present application, based on the disturbance trajectory permutation entropy index and the trip causal score matrix, the main fault device in the multi-device linkage trip event is accurately identified, and the early weak fault of the distribution room is predicted.

[0048] 2. In the present application, the micro-grid operation stability scoring system and the user behavior disturbance analysis method are used to realize the micro-grid rebound risk warning caused by the sudden change of user-side power in the island switching state, and to complete the dynamic identification and warning prompt of the sudden load disturbance behavior. BRIEF DESCRIPTION OF DRAWINGS

[0049] Figure 1 The principle block diagram of an embodiment of the present application is shown in the figure;

[0050] Figure legend: 1, data acquisition edge access unit; 11, multi-source device access collection module; 12, edge transformation message buffer module; 2, time series data storage management unit; 21, hierarchical storage module; 22, dynamic down-sampling module; 3, distribution room diagnosis and monitoring unit; 31, abnormal trend prediction module; 32, trip causal chain tracing module; 4, micro-grid diagnosis and monitoring unit; 41, multi-energy stability evaluation module; 42, island switching rebound prediction module. DETAILED DESCRIPTION

[0051] Example 1, as Figure 1 As shown, a smart grid big data analysis system is provided, including:

[0052] Data acquisition edge access unit 1, which collects power terminal data, user load data and environmental data from the power distribution room and microgrid based on the Internet of Things, and performs edge buffer preprocessing;

[0053] In this embodiment, the data acquisition edge access unit 1 includes a multi-source device access acquisition module 11 and an edge conversion message buffer module 12;

[0054] The multi-source device access acquisition module 11 remotely accesses power terminal equipment, user electrical equipment and environmental monitoring equipment in the power distribution room and microgrid based on the Internet of Things, and collects power terminal data, user load data and environmental monitoring data from the power distribution room and microgrid.

[0055] The power terminal data in the power distribution room includes voltage, current, frequency, and protection action signals; the power terminal data in the microgrid includes photovoltaic output power and energy storage SOC energy storage power; environmental data includes temperature, humidity, wind speed at meteorological stations, and solar radiation intensity; and user load data includes charging pile power, user-side smart meter power, and circuit breaker power.

[0056] In this embodiment, based on the Internet of Things (IoT), communication protocol standards such as Modbus, IEC104, MQTT, and OCPP are used to access power distribution room terminal devices, microgrid terminal devices, and environmental sensors. Each terminal device is assigned a unique device ID and geocode. All collected data is uniformly encapsulated into a five-tuple structure, which consists of [timestamp, location coordinates, device type, sampled value, and communication status]. The sampling frequency supports configuration management from 1000Hz high-frequency waveforms to 1 minute slow variables. All devices are uniformly converted into the internal standard JSON data format through the IoT gateway.

[0057] The power distribution room terminal equipment includes multi-functional meters, integrated protection units, signal panels, DC power supplies, and intelligent micro-circuit switches; the microgrid terminal equipment includes photovoltaic inverters, energy storage systems (BMS), charging piles, and smart meters; and the environmental sensors include temperature and humidity sensors, wind speed sensors, and radiation sensors.

[0058] The edge conversion message buffer module 12 is used to perform communication protocol standardization conversion on the collected power terminal data, user load data and environmental monitoring data, as well as timestamp alignment, validity judgment and weak update elimination.

[0059] In the embodiment, the module introduces a clock synchronization mechanism to align the timestamps of the sampling points, and compares the cached data quintuple structure at the last time. If the cached data quintuple structure at the last time is set to a threshold value, it is marked as a "weak update" to avoid redundancy, and the standard data is pushed to the Kafka message queue. The offline cache retransmission mechanism is used to realize offline cache retransmission, and the buffer size is set to 128-512 MB by default. The message push frequency is set to ensure that high-frequency data does not overflow the queue.

[0060] The time series data storage management unit 2 is further included for hierarchical management and time series down-sampling processing of power terminal data, user load data and environmental monitoring data.

[0061] In the embodiment, the time series data storage management unit 2 includes a hierarchical storage module 21 and a dynamic down-sampling module 22.

[0062] The hierarchical storage module 21 is configured to store power terminal data, user load data and environmental monitoring data of distribution rooms and microgrids in a hierarchical manner.

[0063] In the embodiment, the time series data is divided into hot data, warm data and cold data according to the time span and access frequency of power terminal data, user load data and environmental monitoring data of distribution rooms and microgrids. The specific method is as follows:

[0064] Hot data is data collected within 24 hours. Hot data is written to an in-memory database or NVMe SSD, with a data compression ratio of 10:1, and supports millisecond-level writing and querying. Warm data is data collected within 24 hours to 1 year. It is archived to a distributed time series database TDEngine by minute and hour, supports partitioning by device ID and time, and establishes a composite primary key index according to device number, time interval and label. Cold data is data collected more than 1 year ago. Cold data is archived and stored in a Huawei cloud object storage service database (OBS).

[0065] The dynamic down-sampling module 22 is configured to perform dynamic feature recognition and down-sampling processing on high-frequency time series data.

[0066] In the embodiment, the module is used to perform dynamic feature recognition and down-sampling processing on high-frequency time series data before hierarchical data storage, including anomaly recognition, trend inflection point retention and stable segment compression operation, to realize volume compression of original data while ensuring the integrity of trend information. The down-sampled results are labeled with degradation labels and mapped with original data indexes. The specific method is as follows:

[0067] The key point extraction method using a sliding window combined with a trend identification algorithm retains the slope mutation points, local extreme points and trend reversal points in the waveform to ensure that important feature points of the device behavior are not lost after downsampling; abnormal values such as mutations, distortions and breakpoints are statistically removed, and only data segments with good data continuity and trend representation are retained as main sampling points; the downsampling level is set to three levels: original uncompressed, enhanced compression ratio about 10:1, and lightweight compression ratio up to 100:1, and the appropriate level data can be dynamically loaded according to the query scene; for each group of downsampling data, an index pointer and a hash check value pointing to the original data are retained to realize the reversible mapping and integrity verification of the downsampling and the original data.

[0068] The power distribution room diagnostic monitoring unit 3 introduces a disturbance trajectory permutation entropy index to construct a disturbance evolution trend prediction model based on power terminal data in the power distribution room, identifies weak faults of power equipment in the monitored power distribution room, and constructs a trip event causal matrix model to identify power distribution induced microgrid linkage trip faults, and feeds back to the microgrid diagnostic monitoring unit 4;

[0069] In this embodiment, the power distribution room diagnostic monitoring unit 3 includes an abnormal trend prediction module 31 and a trip causal chain tracing module 32;

[0070] The abnormal trend prediction module 31 introduces a disturbance trajectory permutation entropy index to construct a disturbance evolution trend prediction model based on power terminal data in the power distribution room, and identifies early weak faults of power equipment in the monitored power distribution room;

[0071] The disturbance trajectory permutation entropy index is an index that quantifies the disorder degree of the symbolization mode in a short period of time sequence, and is used as an index for identifying early weak faults of power equipment in the abnormal trend prediction module 31;

[0072] The trip causal chain tracing module 32 constructs a trip event causal matrix model based on protection action signals and trip records to identify power distribution induced microgrid linkage trip faults.

[0073] In this embodiment, the disturbance evolution trend prediction model is realized by fusing the disturbance trajectory permutation entropy index, the device disturbance evolution trend score and the micro-disturbance mutation dynamic score, and is used by the abnormal trend prediction module 31 to identify early weak faults of power equipment in the monitored power distribution room, as follows:

[0074] S3.1.1, set a sliding window to segment the power terminal data in the power distribution room, extract the relative size order of the power terminal data in each sliding window, construct a symbolized permutation pattern set, and calculate the disturbance trajectory permutation entropy index of the previous sliding window period by counting the frequency distribution of each type of permutation pattern;

[0075] ;

[0076] wherein, is the index of the power equipment; is the time; is the power equipment at time corresponding to the permutation entropy of the disturbance trajectory in the sliding window; is the index of the permutation pattern; is the power equipment total number of all valid permutation patterns corresponding to; is the power equipment at time the occurrence probability of the permutation pattern;

[0077] In the formula, the non-fixed embedding dimension and the equal interval sampling are replaced by dynamically adjusting the fluctuation range according to the running state of different equipment ;

[0078] In this embodiment, the permutation entropy of the disturbance trajectory is an index for quantifying the degree of disorder of the short-term symbolic pattern in the time series. The original power terminal data in the distribution room is divided into continuous sliding windows. In each window, the relative size order of the values is investigated. The probability distribution of these order patterns is calculated, and then the degree of disorder of the system running state is calculated. The permutation entropy of the disturbance trajectory is different from the sample entropy and the Shannon entropy, which need to be finely selected and have poor noise resistance. The permutation entropy of the disturbance trajectory is extremely sensitive to small disturbances and short-term mutations. In the distribution room scene, many early failures, such as high-resistance grounding, insulation skin deterioration, and critical hot spots, do not significantly change the mean value or standard deviation, but they can cause slight local disturbance of the waveform and increase the disorder of the data permutation pattern. Therefore, the early failure can be quickly and sensitively reflected by using the permutation entropy of the disturbance trajectory. The permutation entropy of the disturbance trajectory is used as a continuous time series index for modeling, and is used as the basis of the disturbance evolution trend prediction model to realize the early quantitative warning of potential early failures of the electrical system in the distribution room.

[0079] S3.1.2, based on the permutation entropy index of the disturbance trajectory of each sliding window period, giving different time weights to each sliding window, and weighting and accumulating the permutation entropy indexes of all sliding window periods to calculate the disturbance evolution trend score of the equipment;

[0080] ;

[0081] wherein, is the power equipment at time the disturbance evolution trend score of the equipment; is the evolution score sliding window length; is the time; For electric power equipment At time The permutation entropy of the disturbance trajectory in the corresponding sliding window is calculated.

[0082] The formula is different from the traditional moving average or local extreme value judgment method, emphasizes the directional evolution of disturbance trend, and not only judges whether there is an abnormality at present, but also quantifies whether the abnormality is intensifying or weakening;

[0083] S3.1.3, compare the change amplitude of the disturbance trajectory permutation entropy index of the current sliding window period and the disturbance trajectory permutation entropy index of the previous sliding window period, and then normalize to obtain the micro-disturbance mutation dynamic score;

[0084]

[0085] Wherein, For electric power equipment At time The micro-disturbance mutation dynamic score; The mutation detection window; The constant; in the formula, the normalized difference form is used to detect the instantaneous disturbance jump;

[0086] S3.1.4, weighted calculation of the disturbance evolution trend score and the micro-disturbance mutation dynamic score, and finally obtain the comprehensive micro-disturbance risk score;

[0087] Based on the disturbance trajectory permutation entropy index and the comprehensive micro-disturbance risk score, when the disturbance trajectory permutation entropy index of the power terminal shows an upward trend and at the same time triggers the comprehensive micro-disturbance risk score threshold, it is determined that the power terminal is in an early weak fault state.

[0088] In this embodiment, the joint judgment based on the disturbance trajectory permutation entropy index and the comprehensive micro-disturbance risk score can be refined into the following three cases:

[0089] Case one: the disturbance trajectory permutation entropy index is high, but the comprehensive micro-disturbance risk score is low, which represents that the disturbance has no obvious trend change or is short-term fluctuation, and does not constitute a risk only record;

[0090] Case two: the disturbance trajectory permutation entropy index continues to rise, and the comprehensive micro-disturbance risk score is medium to high, indicating that there is a gradual deterioration trend, which should be set to an observation state and added to the intensive monitoring queue;

[0091] Case three: the disturbance trajectory permutation entropy index is high, and the comprehensive micro-disturbance risk score exceeds the dynamic risk threshold, indicating that the current electric power equipment has both continuous disturbance and mutation risk, and is determined as an early weak fault state, which needs to be recorded immediately and an early warning is issued.

[0092] ​In the embodiment, the trip event causal matrix model is a two-dimensional structured matrix model representing possible causal relationships between trip behaviors among a plurality of power devices, used to determine trip faults in a power distribution room and micro-grid linkage scenario, and specifically as follows:

[0093] S3.2.1, accept protection action signals, trip flag records, and switch opening and closing states;

[0094] According to the device number, a trip event log queue is established, and a trip event timestamp, a trip type including overload, short circuit and frequency anomaly, a switch number, and a power supply path are extracted.

[0095] S3.2.2, set a time tolerance, and group trip events occurring within the time tolerance into a multi-trip event group;

[0096] Determine the action sequence within the multi-trip event group, and mark the suspected main trip device and the suspected response device;

[0097] S3.2.3, in each multi-trip event group, calculate a time causal score matrix between devices based on the trip timestamp of each power device, and calculate a topological causal score matrix between devices based on the electrical path length of each power device.

[0098] In the embodiment, based on the protection action timestamp of each device in the trip record, a time causal score matrix is defined. For any two power devices that have a trip event, the time causal score between them is as follows:

[0099] ;

[0100] Wherein, is the time causal score between power device and power device ; is the trip timestamp of power device ; is the trip timestamp of power device ; is a time decay factor;

[0101] The time causal score between all two devices in the multi-trip event group is calculated to construct a time causal score matrix.

[0102] Secondly, combined with the electrical topology structure of the power distribution and micro-grid system, the dependency relationship between the power supply paths of the devices is modeled. For power devices and , if the power supply path dependency relationship exists between the power devices ​Power supply path of power equipment The topology causal score between two power equipment is calculated as follows:

[0103] ;

[0104] wherein, represents the topology causal score between power equipment and power equipment ; represents the power supply path distance between power equipment and power equipment in topology connection node hops;

[0105] If there is no power supply dependence between two power equipment, the topology causal score between them is 0;

[0106] The topology causal scores between all two equipment in the multi-hop event group are calculated, and a topology causal score matrix is constructed;

[0107] S3.2.4, the time causal score matrix and the topology causal score matrix between the equipment are weightedly fused to obtain a joint causal score matrix between the equipment, and the equipment with the largest cumulative causal score to other trip events is selected in the joint causal score matrix as the main fault equipment of the trip event of the multi-hop event group, and the main fault equipment and the topology link structure of the multi-hop event group are fed back to the microgrid diagnosis and monitoring unit 4.

[0108] In this embodiment, the fusion weight coefficient is set to 0.5, the time causal score matrix and the topology causal score matrix between the equipment are weightedly fused to obtain a joint causal score matrix between the equipment, and the specific process is as follows:

[0109] ;

[0110] wherein, represents the joint causal score between power equipment and power equipment ; is a fusion weight coefficient;

[0111] The topology causal scores between all two equipment in the multi-hop event group are calculated, and a topology causal score matrix is constructed;

[0112] Secondly, the main cause equipment identification process is performed on the constructed trip causal score matrix to obtain the main fault equipment with the largest trip causal influence, and the specific process is as follows:

[0113] ;

[0114] wherein, the main fault device in the multi-hop event group;

[0115] feedback the main fault device in the multi-hop event group and the topology link structure to the micro-grid diagnosis monitoring unit 4.

[0116] Further comprising a micro-grid diagnosis monitoring unit 4, which constructs a micro-grid operation stability scoring system based on micro-grid power terminal data and user load data, monitors micro-grid stability risks, and uses a user behavior disturbance analysis method to monitor micro-grid rebound risks during island switching;

[0117] In this embodiment, the micro-grid diagnosis monitoring unit 4 includes a multi-energy stability evaluation module 41 and an island switching rebound prediction module 42;

[0118] The multi-energy stability evaluation module 41 calculates a fused stability confidence value containing all micro-grid power terminal data using fuzzy logic rules based on micro-grid power terminal data and combines the introduced energy disturbance accumulated potential energy to comprehensively construct a micro-grid operation stability scoring system, and calculates a micro-grid operation stability score to monitor micro-grid stability risks and feedback the micro-grid operation stability score to the island switching rebound prediction module 42;

[0119] wherein, is time;

[0120] The energy disturbance accumulated potential energy is the integral of the disturbance variable in the micro-grid power terminal data over a fixed period of time.

[0121] In this embodiment, the micro-grid stability risks include: power fluctuation risks, which are manifested as photovoltaic mutations and load disturbances, leading to serious imbalance between supply and demand; insufficient energy storage response risks, which are manifested as SOC being too low or the power curve being unable to meet the instantaneous power regulation demand; frequency / voltage stability risks, which are manifested as frequency drift when there is no inertia support in the self-supply state of the micro-grid; and low system recovery ability, which is manifested as the system being difficult to recover to a stable point after a small disturbance, showing lagging recovery;

[0122] The micro-grid power terminal data includes photovoltaic output power , energy storage SOC , energy storage power , and charging pile power The principles for selecting disturbance variables in the micro-grid power terminal data are as follows:

[0123] The micro-grid power terminal data has strong time fluctuation; the rapid change of the micro-grid power terminal data will directly cause voltage frequency fluctuation or energy storage imbalance; in the island operation state, the micro-grid power terminal data is most sensitive to its response capability;

[0124] The power terminal number meeting one of the above principles is the disturbance variable in the micro-grid power terminal data;

[0125] The historical fixed time refers to the time interval of the disturbance variable time series data collected in the fixed length period of time back to the current evaluation time, and the fixed time length is set by human;

[0126] This module introduces the energy disturbance accumulation potential energy to represent the energy disturbance accumulation degree of the disturbance variable in the historical fixed time, and the specific calculation method is as follows:

[0127] For each disturbance variable, the change rate in the time interval is calculated, and the energy disturbance accumulation potential energy is calculated by weighted integral:

[0128] ;

[0129] Wherein, is the fixed length time back to the future; is the disturbance variable index; is the total number of disturbance variables; is the weight of the disturbance variable ; is the disturbance variable ; is the integral variable; is the disturbance enhancement factor, which is used to amplify the effect of high frequency oscillation;

[0130] On the basis of the fusion stability confidence value of the micro-grid power terminal data , combined with the energy disturbance accumulation potential energy , the final micro-grid operation stability score value is calculated:

[0131] ;

[0132] Wherein, is the S-type soft threshold function; is the disturbance tolerance threshold; is the micro-grid operation stability score value;

[0133] When the energy disturbance accumulation potential energy accumulates less, the micro-grid operation stability score value remains unchanged, that is, the micro-grid power equipment runs stably; when the energy disturbance accumulation potential energy When the rapid accumulation occurs, the micro-grid operation stability score value is dynamically reduced, so as to reflect that the power equipment of the micro-grid is disturbed or the anti-disturbance ability is reduced.

[0134] In the embodiment, the island switching rebound prediction module 42 is configured to monitor the rebound risk of the micro-grid after the island switching in the off-grid island state of the micro-grid by using a user behavior disturbance analysis method.

[0135] The user behavior disturbance analysis method is configured to analyze the rebound risk process in the island state of the micro-grid.

[0136] In the embodiment, the module receives the main fault equipment in the multi-jump event group fed back from the trip causal chain tracing module 32, judges whether the cause of the current micro-grid island switching behavior is from the main trip event in the main grid trip chain path, and if it is confirmed that the switching is caused by the response to the external power distribution system fault, marks the current running state as a passive island response state.

[0137] The micro-grid rebound risk includes: energy storage rebound risk, which is that the energy storage needs to instantly bear all the loads after the island switching, and if the charging load is not synchronously reduced, the energy storage is instantaneously overcurrent;

[0138] User load mutation, which is that the user does not perceive the system switching, and continues to start high-power equipment such as a charging pile;

[0139] Frequency disturbance impact risk, which is that the power supply lacks synchronous support, and the inertia is insufficient to cause frequency jump and trigger low-frequency off-grid;

[0140] PV oscillation or unstable response risk, which is that there is no reference voltage source after the micro-grid switching, and the photovoltaic inverter appears output oscillation or shutdown;

[0141] Based on the historical load behavior data of the user side and the charging load response characteristics, a user behavior disturbance analysis method is constructed, the rebound risk such as the load power sudden rise and the energy storage power response delay caused by the switching is quantified, a dynamic adjustment basis is provided for the system scheduling strategy, a dynamic sequence of the user load behavior is modeled, and a predictive rebound index is formed in combination with the context of the micro-grid switching event, and the lack of consideration of the user response risk in the existing island switching strategy is filled.

[0142] In the embodiment, the user behavior disturbance analysis method is specifically as follows:

[0143] Based on the user load data, the user load sudden rise risk value of each user power consumption node is calculated , and the comprehensive user side disturbance score is calculated in combination with the micro-grid operation stability score value , and the micro-grid rebound early warning is performed according to the comprehensive user side disturbance score.

[0144] The user load sudden rise risk value of each user power consumption node is calculated based on the user load data. is time; represents a user power consumption node index.

[0145] In the embodiment, the user load surge risk value is a unit power mutation intensity index calculated by analyzing the trend of the load power change with time for each user power consumption node in the micro-grid before and after island switching, which is used to quantify the potential risk intensity of the node in the next period to the micro-grid stability disturbance;

[0146] the user load surge risk value combined with the micro-grid operation stability score value to obtain a comprehensive user-side disturbance score .

[0147] When the micro-grid operation stability score value tends to zero, it means that the micro-grid operation stability is poor, and the user disturbance influence weight rises, amplifying the system risk judgment of the behavior disturbance; when the micro-grid operation stability score value tends to one, it means that the micro-grid operation stability is strong, and the disturbance score is suppressed to prevent misjudgment.

[0148] The comprehensive user-side disturbance score of all user power consumption nodes is subjected to maximum value extraction and average weighted analysis to calculate the rebound risk index of the distribution room and the micro-grid in the island switching state, and according to whether the rebound risk index exceeds the preset threshold value, it is judged whether the micro-grid is currently in a high-risk state, and a micro-grid rebound warning is performed.

[0149] In embodiment two, the present application proposes a smart grid big data analysis method for the smart grid big data analysis system in embodiment one, including the following steps:

[0150] S10.1, based on the Internet of Things, collecting power terminal data, user load data and environmental data of the distribution room and the micro-grid, and performing edge buffer preprocessing;

[0151] S10.2, layered management and time series down-sampling processing are performed on the power terminal data, user load data and environmental monitoring data;

[0152] S10.3, based on the power terminal data of the distribution room, a disturbance trajectory permutation entropy index is introduced to construct a disturbance evolution trend prediction model, to identify weak faults of power equipment in the monitored distribution room, and a trip event causal matrix model is constructed to identify distribution-induced micro-grid linkage trip faults;

[0153] S10.4, based on micro-grid power terminal data and user load data, constructing a micro-grid operation stability scoring system to monitor the stability risk of the micro-grid, and using a user behavior disturbance analysis method to monitor the rebound risk of the micro-grid during island switching.

[0154] The embodiments of the present application are described in detail above with reference to the drawings, but the present application is not limited thereto, and various changes can be made within the knowledge of those skilled in the art without departing from the spirit of the present application.

Claims

1. A smart grid big data analytics system, characterized by, Comprising A data acquisition edge access unit (1) collects power terminal data, user load data and environmental data of a distribution room and a microgrid based on Internet of Things, and performs edge buffer preprocessing; A time series data storage management unit (2) is used for hierarchical management and time series down-sampling processing of power terminal data, user load data and environmental monitoring data; A distribution room diagnosis and monitoring unit (3) constructs a disturbance evolution trend prediction model based on the power terminal data of the distribution room, introduces a disturbance trajectory permutation entropy index, identifies and monitors the early weak faults of the power equipment in the distribution room, and constructs a trip event causal matrix model to identify the distribution-induced microgrid linkage trip fault and feedback to the microgrid diagnosis and monitoring unit (4); The microgrid diagnosis and monitoring unit (4) constructs a microgrid operation stability scoring system based on the power terminal data and user load data of the microgrid to monitor the stability risk of the microgrid, and uses a user behavior disturbance analysis method to monitor the microgrid rebound risk during island switching; The distribution room diagnosis and monitoring unit (3) includes an abnormal trend prediction module (31) and a trip causal chain tracing module (32); The abnormal trend prediction module (31) constructs a disturbance evolution trend prediction model based on the power terminal data of the distribution room, introduces a disturbance trajectory permutation entropy index, and identifies and monitors the early weak faults of the power equipment in the distribution room; The disturbance trajectory permutation entropy index is an index for quantifying the disorder degree of the short-term symbolization mode in the time series, which is used as an index for identifying the early weak faults of the power equipment in the abnormal trend prediction module (31); The trip causal chain tracing module (32) constructs a trip event causal matrix model based on the protection action signal and the trip record to identify the distribution-induced microgrid linkage trip fault; The disturbance evolution trend prediction model is realized by fusing the disturbance trajectory permutation entropy index, the device disturbance evolution trend score and the micro-disturbance mutation dynamic score, and is used for the abnormal trend prediction module (31) to identify and monitor the early weak faults of the power equipment in the distribution room, which is specifically as follows: S3.1.1, set a sliding window to segment the distribution room power terminal data, extract the relative size order of the distribution room power terminal data in each sliding window, construct a symbolized permutation pattern set, and calculate the disturbance trajectory permutation entropy index of the previous sliding window period by counting the frequency distribution of each type of permutation pattern; S3.1.2, based on the disturbance trajectory permutation entropy index of each sliding window period, give each sliding window different time weight, and weight the disturbance trajectory permutation entropy index of all sliding window periods to calculate the device disturbance evolution trend score; S3.1.3, compare the change amplitude of the disturbance trajectory permutation entropy index of the current sliding window period and the disturbance trajectory permutation entropy index of the previous sliding window period, and then normalize to obtain the micro-disturbance mutation dynamic score; S3.1.4, weight the device disturbance evolution trend score and the micro-disturbance mutation dynamic score to finally obtain the comprehensive micro-disturbance risk score; The disturbance trajectory permutation entropy index and the comprehensive micro-disturbance risk score are combined to determine whether the power terminal is in an early weak fault state.

2. The smart grid big data analytics system of claim 1, wherein, The data acquisition edge access unit (1) comprises a multi-source device access and acquisition module (11) and an edge transformation message buffer module (12). The multi-source device access and acquisition module (11) remotely accesses the power terminal devices, user power consumption devices and environmental monitoring devices in the power distribution room and micro-grid based on the Internet of Things, and acquires power terminal data, user load data and environmental monitoring data of the power distribution room and micro-grid. The power distribution room power terminal data includes voltage, current, frequency and protection action signals; the micro-grid power terminal data includes photovoltaic output power, energy storage SOC and energy storage power; the environmental data includes temperature, humidity, weather station wind speed and solar radiation intensity; the user load data includes charging pile power, user-side intelligent electric meter and power consumption circuit breaker power. The edge transformation message buffer module (12) is used for standardizing the communication protocol conversion of the acquired power terminal data, user load data and environmental monitoring data, and for time stamp alignment, validity judgment and weak update elimination.

3. The smart grid big data analytics system of claim 2, wherein, The time series data storage management unit (2) comprises a hierarchical storage module (21) and a dynamic down-sampling module (22). The hierarchical storage module (21) is used for hierarchical storage of the power terminal data, user load data and environmental monitoring data of the power distribution room and micro-grid. The dynamic down-sampling module (22) is used for dynamic feature recognition and down-sampling processing of high-frequency time series data.

4. The smart grid big data analytics system of claim 3, wherein, The trip event causal matrix model is a two-dimensional structured matrix model representing the possible causal relationship between the trip behaviors of multiple power devices, and is used to determine the trip fault in the power distribution room and micro-grid linkage scene, specifically as follows: S3.2.1, accept the protection action signal, trip flag record and switch opening and closing state; S3.2.2, set the time tolerance, and group the trip events occurring within the time tolerance into a multi-trip event group; S3.2.3, in each multi-trip event group, calculate the time causal score matrix between devices based on the trip time stamp of each power device, and calculate the topological causal score matrix between devices based on the electrical path length of each power device; S3.2.4, weighted fusion of the time causal score matrix and the topological causal score matrix between devices to obtain the joint causal score matrix between devices, and select the device with the maximum cumulative causal score to other trip events in the joint causal score matrix as the main fault device of the trip event of the multi-trip event group, and feed back the main fault device and the topological link structure of the multi-trip event group to the micro-grid diagnosis and monitoring unit (4).

5. The smart grid big data analytics system of claim 4, wherein, The micro-grid diagnosis and monitoring unit (4) comprises a multi-energy stability evaluation module (41) and an island switching rebound prediction module (42). The multi-energy stability evaluation module (41) calculates a fused stability confidence value containing all micro-grid power terminal data based on micro-grid power terminal data using fuzzy logic rules , and combines the introduced energy disturbance accumulated potential energy , and comprehensively constructs a micro-grid operation stability score system to calculate a micro-grid operation stability score value , monitors the stability risk of the micro-grid, and feeds the micro-grid operation stability score value back to the island switching recoil prediction module (42). wherein t is time; The energy disturbance accumulation potential is the integral of the disturbance variable in the micro-grid power terminal data within a fixed time.

6. The smart grid big data analytics system of claim 5, wherein, The island switching recoil prediction module (42) is used to monitor the recoil risk of the micro-grid after island switching in the off-grid island state of the micro-grid using a user behavior disturbance analysis method. The user behavior disturbance analysis method is used to analyze the recoil risk process in the island state of the micro-grid.

7. The smart grid big data analytics system of claim 6, wherein, The user behavior disturbance analysis method is as follows: Calculate the user load surge risk value of each user power consumption node based on user load data And combine the micro-grid operation stability score value Calculate the comprehensive user side disturbance score, and give a micro-grid recoil warning according to the comprehensive user side disturbance score; wherein, is time; denotes the user power consumption node index.

8. A method for smart grid big data analysis, used for the smart grid big data analysis system of any one of claims 1-7, characterized in that: The method comprises the following steps: S10.1, collecting power terminal data, user load data and environmental data of the distribution room and the micro-grid based on the Internet of Things, and performing edge buffer preprocessing; S10.2, performing hierarchical management and time series down-sampling processing on the power terminal data, user load data and environmental monitoring data; S10.3, based on the power terminal data of the distribution room, introducing a disturbance trajectory permutation entropy index to construct a disturbance evolution trend prediction model, identifying weak faults of power equipment in the monitored distribution room, and constructing a trip event causal matrix model to identify distribution-induced micro-grid linkage trip faults; S10.4, based on the micro-grid power terminal data and user load data, constructing a micro-grid operation stability scoring system to monitor the stability risk of the micro-grid, and using a user behavior disturbance analysis method to monitor the recoil risk of the micro-grid during island switching.

Images