A grid-connected power quality disturbance identification method based on photovoltaic source-side output change
Patent Information
- Application Number
- CN202611491916.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-23
- Publication Date
- 2026-10-27
Smart Images
Figure CN122992731A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of photovoltaic power generation grid-connected operation and power quality monitoring technology, specifically to a method for identifying grid-connected power quality disturbances based on changes in photovoltaic source-side output. Background Technology
[0002] As the scale of photovoltaic (PV) power generation continues to expand, fluctuations in the output of PV grid-connected systems and changes in inverter control states can impact power quality at the grid connection point. When the active power on the PV source side changes, power quality indicators such as grid connection point voltage, reactive power, harmonic voltage content, and negative sequence voltage components may exhibit corresponding responses. These responses are related to the magnitude, rate, and duration of PV power changes, as well as inverter control states such as maximum power point tracking, power limiting control, reactive power regulation, and current constraints. Therefore, power quality analysis of PV grid-connected systems requires not only determining whether power quality indicators are abnormal but also distinguishing between normal responses caused by PV power changes and genuine abnormal disturbances.
[0003] Fixed threshold methods struggle to reflect the differences in normal responses of power quality indicators under varying photovoltaic power fluctuations and inverter control conditions. This can easily lead to misjudging normal responses as abnormal ones or overlooking abnormal deviations from normal operating patterns. While machine learning or neural network methods can learn abnormal patterns from multivariate time-series data, their results typically focus on anomaly detection or category determination. They struggle to explicitly construct the corresponding normal response range for a specific photovoltaic power fluctuation event and further quantify the degree of deviation of the actual response from the normal response. Furthermore, existing methods often focus on determining whether the power quality indicators themselves are abnormal, lacking a unified event-level correlation analysis mechanism to determine whether the detected abnormal responses are correlated with specific photovoltaic power fluctuation events and the corresponding inverter control state at the time of the anomaly.
[0004] Therefore, it is necessary to provide a method for identifying power quality disturbances in grid-connected photovoltaic systems. This method matches specific photovoltaic power change events and their inverter control states with historical normal operation events to construct a corresponding normal power quality response range. Anomalies are identified by the deviation between the actual response and the normal response. Furthermore, the time, direction, and control conditions of the abnormal response are combined to perform correlation attribution, thereby realizing the identification, quantification, and correlation evaluation of power quality disturbances. Summary of the Invention
[0005] To address the challenges of identifying, quantifying, and evaluating the power quality response at the grid connection point during changes in photovoltaic (PV) source output, this invention provides a method for identifying grid-connected power quality disturbances based on changes in PV source output. This invention combines the characteristics of PV source output changes with the inverter control state to distinguish between normal and abnormal power quality responses corresponding to PV power changes, thereby improving the accuracy and reliability of identifying, attributing, and evaluating grid-connected power quality disturbances.
[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:
[0007] A method for identifying grid-connected power quality disturbances based on changes in photovoltaic source-side output includes the following steps:
[0008] S1. Acquire and synchronize photovoltaic source-side power, grid connection point power quality and inverter operation data to form a synchronized operation dataset under a unified time reference;
[0009] S2. Identify photovoltaic power change events based on the dual threshold continuous criterion of power change rate, and determine the change direction, change magnitude, change rate, duration and change pattern of the event based on multi-dimensional event characteristics;
[0010] S3. Construct the inverter control state during the photovoltaic power change event, identify the control state changes, and segment the photovoltaic power change event according to the control state changes;
[0011] S4. Based on the event characteristics of the photovoltaic power change event and the inverter control state, initially select a set of matching normal reference events from the historical normal operation events;
[0012] S5. Based on the power change characteristics of the current photovoltaic power change event and the historical normal reference event, the historical normal response is normalized and fused to obtain the theoretical normal response and its discrete error of each power quality index of the current event, and then the corresponding normal response range is constructed.
[0013] S6. Compare the actual response of each power quality index within each control condition section with the corresponding normal response range, identify the effective abnormal response section through continuous criteria, and calculate the occurrence time, duration, maximum abnormal deviation and cumulative degree of abnormal response.
[0014] S7. Perform joint attribution of time, direction and control conditions on effective abnormal responses, calculate the attribution confidence, and determine the abnormality index based on the maximum deviation, duration and cumulative degree of abnormal responses that meet the attribution conditions. At the same time, associate the corresponding inverter control state to form an abnormal response set containing abnormal characteristics and associated control conditions.
[0015] S8. Normalize and merge the abnormality of each control condition segment and each power quality indicator, determine the comprehensive power quality disturbance degree of photovoltaic power change events and classify them according to the duration of each segment, and at the same time determine the main disturbance indicators and associated control conditions, and form and output the power quality disturbance evaluation results of each photovoltaic power change event.
[0016] Compared with the prior art, the present invention has the following advantages and technical effects:
[0017] 1. This invention identifies photovoltaic power change events based on a dual-threshold continuous criterion of power change rate, and determines the change direction, amplitude, rate, duration and change pattern of the event by combining the duration, number of power change direction reversals and the proportion of the first-level threshold, which can reduce the impact of instantaneous fluctuations and random noise on event identification and achieve effective differentiation of different types of photovoltaic power change events.
[0018] 2. Based on the characteristics of power change direction, control conditions, power change amplitude, maximum power change rate and duration, this invention selects normal reference events from historical normal operation events, establishes theoretical normal response and normal response range through response normalization and fusion, and determines effective abnormal response segments by combining continuous abnormal response persistence criteria. In this way, it can distinguish between normal dynamic response and abnormal power quality response in the photovoltaic power change process, and reduce misjudgment caused by fixed threshold method.
[0019] 3. This invention combines the time correlation, directional consistency, and control condition matching relationship between abnormal responses and control condition segments to calculate attribution confidence, determine the correlation between abnormal responses and photovoltaic power change events, and further conducts quantitative evaluation based on the maximum deviation, duration, and cumulative deviation of abnormal responses. It also combines different power quality indicators to perform comprehensive disturbance classification, thereby improving the accuracy of disturbance identification, the reliability of attribution, and the interpretability of evaluation results. Attached Figure Description
[0020] Figure 1 This is a flowchart of the overall process for identifying grid-connected power quality disturbances based on changes in photovoltaic source output.
[0021] Figure 2 This is a graph showing the active power at the photovoltaic source side and the voltage at the grid connection point in an embodiment of the present invention.
[0022] Figure 3 This is a comparison chart of the actual response range of the event grid connection point voltage and the normal response range in an embodiment of the present invention. Detailed Implementation
[0023] like Figure 1As shown in the figure, this application provides a method for identifying grid-connected power quality disturbances based on changes in photovoltaic source-side output, specifically including the following steps:
[0024] S1. Acquire and synchronize photovoltaic source-side power, grid connection point power quality and inverter operation data to form a synchronized operation dataset under a unified time reference;
[0025] Specifically, the three-phase voltage and three-phase current at the photovoltaic source side are collected by a photovoltaic source-side measurement device, and the active power at the photovoltaic source side is obtained through power calculation. and reactive power .
[0026] The effective value of the grid connection point voltage is collected by a power quality analysis device at the grid connection point. Reactive power at grid connection point Harmonic voltage content and negative sequence voltage components ,in , The highest harmonic order collected or analyzed.
[0027] Read the inverter's operating status flags through the inverter's communication interface. Control mode identifier Active power limit Reactive power limits Current limit and actual output current .
[0028] All types of data are time-aligned according to a unified time base, and a synchronized time series is established. For continuous quantities, interpolation is used for time alignment, while for discrete quantities such as operating status and control modes, nearest neighbor method is used. This ensures that photovoltaic source-side power, grid-connected power quality indicators, and inverter operating data correspond to the same sampling time, and the time-aligned data constitute a synchronized operating dataset. .
[0029] S2. Identify photovoltaic power change events based on the dual threshold continuous criterion of power change rate, and determine the change direction, change magnitude, change rate, duration and change pattern of the event based on multi-dimensional event characteristics;
[0030] From synchronously running datasets Extracting the active power sequence from the photovoltaic source side ,make , Indicates the first Each synchronous sampling time node. The power change rate is calculated based on the active power at the photovoltaic source side of adjacent synchronous time nodes. ,in, This is the time interval between adjacent synchronous sampling time nodes. When, it indicates that the active power on the photovoltaic source side is increasing; when When this occurs, it indicates that the active power on the photovoltaic source side is decreasing.
[0031] Extract power change rate from historical normal operation data And take their absolute values to form a set of absolute values of power change rate. , represented as , This represents the number of power change rate samples in the historical normal operation data. The mean of the set of absolute values of the power change rate is then calculated. and standard deviation Take the confidence coefficient. ,in To obtain the first-level threshold Further, take the proportionality coefficient. ,in Secondary thresholds are obtained. This ensures Preferably, the confidence coefficient Desirable The proportionality coefficient Desirable This balances the detection sensitivity to power change events with the ability to resist random fluctuations. In a preferred embodiment, the confidence coefficient... Pick The proportionality coefficient Pick .in, Used to trigger candidate segments for power change events Used to confirm and maintain power change events.
[0032] When consecutive sampling points satisfy When the corresponding time interval is determined, it is identified as the candidate power change interval, and the sampling time that first meets this condition is determined as the first sampling time interval. The start time of each power change event Within the candidate power variation range, when consecutive sampling points satisfy... Upon confirmation of a power change event, the rate of power change is monitored. Once the event is confirmed, the rate of power change continues to be monitored until consecutive sampling points meet the specified conditions. At that time, determine the end time of the event. And denote the set of synchronous sampling points during the event as .
[0033] For the A confirmed photovoltaic power change event, defined as the event's starting power as... The final power is The change in power is The direction of change is ,in, This indicates an increase in power. This indicates a decrease in power. Further definitions are provided for the power change amplitude, maximum power change rate, and duration. , and .in, This represents the minimum power change threshold used to exclude events with minute power changes. Preferably, the... The setting can be determined based on the rated capacity of the photovoltaic system, the measured noise level, and the actual operating fluctuation range; the preferred value is [value missing] of the rated capacity. .when At that time, the first The event is classified as a minor power change event and will not be included in subsequent power quality disturbance analysis; when When this occurs, it is identified as an effective photovoltaic power change event.
[0034] For effective photovoltaic power change events, the number of times the power change direction reverses during the event period is counted. In this method, direction reversal statistics are performed only on adjacent power change rate sampling points that meet the secondary threshold criterion, in order to reduce the impact of random changes close to zero on the direction reversal judgment. Preferably, when At that time, the changing form of the event It is classified as a fluctuation type to indicate that there are obvious cyclical changes during the event.
[0035] when Then, further calculate the proportion of power change rate sampling points that reached the first-level threshold during the event period. .
[0036] When the event duration meets And the percentage meets At that time, the form will change. It is determined to be a step type, in which, This represents the maximum duration threshold used to determine step-type power change events. Preferably, the... The response speed and sampling period can be set according to the photovoltaic power control, and can be selected. s. In a preferred embodiment, Pick s. For those that do not satisfy the above step-type conditions and Valid events will change form It is determined to be a gradient type.
[0037] S3. Construct the inverter control state during the photovoltaic power change event, identify the control state changes, and segment the photovoltaic power change event according to the control state changes;
[0038] Extract the first A number of effective photovoltaic power change events within the time interval Inverter operating status flags at each sampling time Control mode identifier Active power limit Reactive power limits Current limit and active power on the photovoltaic source side Reactive power at grid connection point and actual output current ;
[0039] According to the inverter communication protocol and Decode the inverter's operating status to obtain the results. and control mode ,in Used to determine whether the inverter is in normal grid-connected operation. This indicates the currently used control mode, including MPPT (Maximum Power Point Tracking) mode, power limiting control mode, etc. Sampling points indicating shutdown, malfunction, or other abnormal operating conditions are marked as abnormal operating points and removed in subsequent processing.
[0040] For sampling points that are in normal grid-connected operation, a limit proximity coefficient is set. , and This is used to determine whether the active power, reactive power regulation, and actual output current are close to the corresponding control limits. The limit proximity coefficient is preferably set to a value that is close to the limit. And can be based on the inverter control accuracy and measurement error within the corresponding limits. Adjustments are made within the specified range. Based on this, the active power constraint flag... according to If and only if ,otherwise Reactive power regulation constraint indicator according to If and only if ,otherwise Current constraint flag according to If and only if ,otherwise Among them, the sign is taken This indicates that the corresponding operating quantity is close to the corresponding control limit, and the flag is set. This indicates that the corresponding control limit has not yet been reached.
[0041] Control mode Combined with the three constraint flags mentioned above, it constitutes the first Control conditions at each sampling time It is used to characterize the actual control mode adopted by the inverter at that sampling moment, as well as the active power, reactive power regulation, and current constraint status.
[0042] Compare the control conditions of adjacent sampling times within the event interval in the order of sampling time. At that time, the current sampling time will be... The time when control conditions switch is determined. Based on the event start time. The switching times of each control condition and the end times of the events. As segmentation boundaries, events are divided into A continuous control condition segment, and represented as ,in, Indicates the first The first photovoltaic power change event The start and end times of each control condition segment. This indicates the control conditions corresponding to this segment. If the control conditions remain unchanged throughout the event, then... The event is classified as a single control condition event; if the control mode or any constraint flag changes once or multiple times, then... The event is identified as a multi-control condition event, and each control condition segment and its corresponding control condition are recorded separately.
[0043] S4. Based on the event characteristics of the photovoltaic power change event and the inverter control state, initially select a set of matching normal reference events from the historical normal operation events;
[0044] For the first determined by S2 For each valid photovoltaic power change event, a historical normal event with the same change direction and control conditions as the current event is selected from a pre-determined historical normal operation event, and a normal reference event matching the current event is determined based on the power change amplitude, maximum power change rate, and duration.
[0045] For the first condition that meets the above conditions For each historical normal event, calculate the power change characteristic distance based on the power change amplitude, maximum power change rate, and duration of the current event relative to that historical normal event:
[0046]
[0047] in, , and These represent the magnitude of the power change, the maximum rate of power change, and the duration of the current event, respectively. , and The first The magnitude of power change, the maximum rate of power change, and the duration of each historical normal event; , and These are the scale parameters for the corresponding features; , and These are the weights of the corresponding features, and Preferably, the feature weights are set to , and These are used to characterize the relative importance of power change amplitude, maximum power change rate, and duration in power change feature matching, respectively.
[0048] Historical normal events that meet the same direction of change and the same control conditions are classified according to their power change characteristics and distance. Sort the historical normal events in ascending order, and select the fewest historical normal events with the smallest distance as normal reference events for the current event based on the number of available historical normal events. The number of normal reference events is preferably 5 to 10. When the number of available historical normal events is less than 5, select all available historical normal events. The final result is the [number of events]. A set of normal reference events corresponding to each current photovoltaic power change event. .
[0049] S5. Based on the power change characteristics of the current photovoltaic power change event and the historical normal reference event, the historical normal response is normalized and fused to obtain the theoretical normal response and its discrete error of each power quality index of the current event, and then the corresponding normal response range is constructed.
[0050] For the current number A set of normal reference events obtained from S4 for each photovoltaic power change event. Determine the normal response corresponding to the current event. For grid connection point voltage... Reactive power at grid connection point Harmonic voltage content and negative sequence voltage components Normal response calculations were performed separately, using any power quality index as an example. This indicates that the values include the grid connection point voltage. Reactive power at grid connection point Harmonic voltage content and negative sequence voltage components .
[0051] Starting from the current event time The previous stable operating range is used as the benchmark range, and the indicators within this range are... The average value is used as the baseline value for the current event. The timeframe is set as the start time of the event. Representing the time offset relative to the start time of the event, the actual response to the current event is then expressed as: For the normal reference event set obtained by S4 The first in For each normal reference event, its baseline value and actual response are calculated in the same way to obtain the actual response of historical normal events. Meanwhile, based on the power change determined by S2 for this historical normal event... The actual response is based on the power change of the current event. Normalization is performed to obtain the historical normal response corresponding to the current power change amplitude. in This represents the power change of the current event. For the first The power change of each normal reference event is determined according to S2.
[0052] Historical normal responses corresponding to each normal reference event Perform equal-weighted fusion to obtain the current event index. Theoretical normal response ,in Represents the normal reference event set The number of normal reference events included. Further, the normal response error is determined based on the dispersion between the historical normal response and the theoretical normal response for each normal reference event.
[0053] .
[0054] Determine the normal response range of the current event based on the theoretical normal response and the normal response error. ,in This is a preset range coefficient used to determine the width of the normal response range.
[0055] S6. Compare the actual response of each power quality index within each control condition section with the corresponding normal response range, identify the effective abnormal response section through continuous criteria, and calculate the occurrence time, duration, maximum abnormal deviation and cumulative degree of abnormal response.
[0056] Specifically, for the current number The first photovoltaic power change event Each control condition section, and the power quality index of any grid connection point corresponding to the control condition section. Obtain the normal response range determined by S5. and the actual response corresponding to the current control condition segment. .
[0057] The actual response is compared with the normal response range determined by S5. When the actual response exceeds the upper boundary of the normal response range or falls below the lower boundary of the normal response range, the corresponding sampling time is determined as the index. The abnormal response point; when the actual response is within the normal response range, the corresponding sampling time is determined as the normal response point. Specifically, when ,or When the sampling time is determined, the corresponding sampling time is identified as the abnormal response point.
[0058] For a given abnormal response point, the abnormal deviation amount is determined based on the degree of deviation of the actual response from the boundary of the normal response range. When the actual response exceeds the upper boundary of the normal response, When the actual response is below the lower boundary of the normal response, When the actual response is within the normal response range, .
[0059] The duration of the abnormal response is determined based on the time range corresponding to the consecutive abnormal response points. The maximum absolute value of the abnormal deviation during the abnormal response period is determined as the maximum degree of abnormal deviation, i.e. The maximum value is obtained only for abnormal response points. Simultaneously, the absolute value of the abnormal deviation corresponding to each abnormal response point is accumulated over time to obtain the degree of abnormal response accumulation. The summation is performed only for abnormal response points.
[0060] For a current photovoltaic power change event that includes multiple control condition segments, the above-mentioned abnormal response identification is performed in each control condition segment to obtain the abnormal occurrence time, abnormal duration, maximum abnormal deviation degree, and abnormal response accumulation degree corresponding to each control condition segment and each power quality index.
[0061] S7. Perform joint attribution of time, direction and control conditions on effective abnormal responses, calculate the attribution confidence, and determine the abnormality index based on the maximum deviation, duration and cumulative degree of abnormal responses that meet the attribution conditions. At the same time, associate the corresponding inverter control state to form an abnormal response set containing abnormal characteristics and associated control conditions.
[0062] Specifically, for the abnormal response segment determined by S6, the direction of photovoltaic power change corresponding to the abnormal response and the inverter control conditions for the corresponding time period are obtained, and the starting time of the corresponding control condition segment is used as the starting time. As a time reference point.
[0063] The time correlation is determined based on the time offset between the start time of the abnormal response and the start time of the corresponding control condition segment. The time offset between the start time of the abnormal response and the start time of the corresponding control condition segment is defined as... When an abnormal response occurs within the corresponding control condition segment, and Located within the preset response delay range Within the time frame, time-related criteria will be used. Set to 1 otherwise set to 0. This indicates that the abnormal response occurred within the preset time range. This indicates that the abnormal response does not meet the time correlation condition.
[0064] Furthermore, based on the changing direction of current photovoltaic power change events... The direction correlation is determined by the inverter control mode and the deviation direction of the abnormal response. When the direction of the abnormal response is consistent with the expected response direction corresponding to the current control mode and the direction of photovoltaic power change, the direction consistency criterion is applied. Set to 1 otherwise set to 0. This indicates that the direction of the abnormal response is consistent with the direction of the expected response. This indicates that the two are inconsistent.
[0065] At the same time, the actual control conditions during the abnormal response period will be compared with the corresponding control conditions determined by S3. The two are compared, and when they match, the control condition matching criterion is applied. Set to 1 otherwise set to 0. This indicates that the control conditions during the abnormal response period are consistent with the corresponding control condition segment. This indicates that the two are inconsistent.
[0066] In one implementation, time correlation, direction correlation, and control condition matching are combined to obtain the attribution confidence of the abnormal response. Its calculation is ,in, , and All are dimensionless criteria that take values of 0 or 1. , and These are the non-negative weights corresponding to time correlation, direction correlation, and control condition matching, respectively. Preferably, in one embodiment, it may be configured as follows: , and This approach highlights the temporal correlation between abnormal responses and photovoltaic power variation events, and combines directional consistency and control condition matching to comprehensively attribute abnormal responses. Therefore, the attribution confidence level... The value range is from 0 to 1. When When an abnormal response is identified, it is determined to be an abnormal power quality response associated with the current photovoltaic power change event; when At that time, it will not be considered as an associated anomaly response to the current photovoltaic power change event. This is the preset attribution confidence threshold.
[0067] For abnormal responses identified as being related to current photovoltaic power change events, their corresponding grid-connected power quality indicators are used. As a disturbance indicator, the anomaly severity index is determined based on the maximum abnormal deviation of the abnormal response, the duration of the abnormality, and the cumulative degree of the abnormal response. In one implementation, the anomaly index is expressed as... ,in , and These are the maximum degree of abnormal deviation, the duration of abnormality, and the cumulative degree of abnormal response determined by S6. , and These are the scale parameters corresponding to the maximum degree of anomaly deviation, the duration of anomaly, and the cumulative degree of anomaly response, respectively, used to normalize anomaly characteristics of different dimensions.
[0068] Furthermore, based on the control conditions corresponding to the occurrence of the abnormal response. Determine the inverter control state associated with the abnormal response, including the control mode and the active power constraint, reactive power regulation constraint, and current constraint states. When the abnormal response mainly occurs within a certain control condition segment, determine the control conditions corresponding to that control condition segment. As an associated control condition for abnormal responses.
[0069] For cases where the same photovoltaic power change event includes multiple control condition segments, the anomaly severity index corresponding to each control condition segment and each power quality index should be determined separately. Record the control condition section, power quality index, anomaly occurrence time, and anomaly duration corresponding to each anomaly response. Maximum abnormal deviation Accumulated abnormal response level And associated control conditions, thereby forming abnormal response information for current photovoltaic power change events.
[0070] S8. Normalize and merge the abnormality of each control condition segment and each power quality indicator, determine the comprehensive power quality disturbance degree of photovoltaic power change events and classify them according to the duration of each segment, and at the same time determine the main disturbance indicators and associated control conditions, and form and output the power quality disturbance evaluation results of each photovoltaic power change event.
[0071] Specifically, for the current number For each photovoltaic power change event, the degree of abnormality corresponding to each control condition segment and power quality index obtained from S7 is considered. The degree of anomaly for different power quality indicators is normalized to eliminate differences in the dimensions and numerical ranges of different indicators. In one implementation, according to formula... The degree of anomaly is normalized, where, Power quality indicators The corresponding preset upper limit of the abnormality level, and .
[0072] For the For each control condition segment, based on the abnormal responses corresponding to each power quality indicator determined in S7, establish a set of associated abnormal indicators corresponding to that control condition segment. , represented as ,in, Indicates the first The first photovoltaic power change event The set of power quality indicators that actually generate associated abnormal responses in each control condition section.
[0073] when When not empty, only the set of associated abnormal indicators is considered. The normalized anomaly levels of various power quality indicators are weighted and fused to determine the overall disturbance level of the control condition section. , represented as ,in, Represents a set of associated abnormal indicators The number of power quality indicators included. When no abnormal response related to the current photovoltaic power change event is identified within the control condition segment, i.e. When empty, the overall disturbance level of the control condition section will be controlled. The value is determined to be 0.
[0074] Furthermore, the segment weights are determined based on the duration of each control condition segment. The segment weight is calculated according to the formula. Confirmed, among which Indicates the first The first photovoltaic power change event The duration of each control condition segment; This represents the number of control condition segments corresponding to the photovoltaic power change event. Based on the overall disturbance level of each control condition segment and its corresponding segment weight, the overall power quality disturbance level of the current photovoltaic power change event is determined. .
[0075] Furthermore, based on the set of associated abnormal indicators corresponding to each control condition segment... and the corresponding anomaly indexes for each power quality indicator. One or more power quality indicators with significant anomalies are identified as the set of main disturbance indicators for the current photovoltaic power change event. And record the control condition sections corresponding to the main disturbance indicators.
[0076] The current photovoltaic power change events are classified according to the preset comprehensive disturbance classification threshold: when When, the current event is identified as a normal response event; when When, the current event is classified as a general power quality disturbance event; when When the current event is identified as a significant power quality disturbance event, then, and The preset comprehensive disturbance classification threshold is, and .
[0077] Simultaneously, based on the control condition segment determined in S3 and the correlation between the abnormal response and control conditions obtained in S7, the associated control conditions corresponding to the current photovoltaic power change event are determined. When the abnormal response is mainly concentrated in one control condition segment, the control condition corresponding to that control condition segment is taken as the main associated control condition; when the abnormal response occurs in multiple control condition segments, the comprehensive disturbance degree, main disturbance index and associated control condition corresponding to each control condition segment are recorded respectively.
[0078] Ultimately, this forms the current [number]. Power quality disturbance evaluation results corresponding to photovoltaic power change events ,in, The event characteristics of photovoltaic power change events include the direction of change. Power variation range Maximum power change rate Duration and its changing forms ; To comprehensively assess the degree of power quality disturbance, For associated control conditions, This is the main set of disturbance indicators. For all effective photovoltaic power change events, steps S3 to S8 are performed respectively to obtain the power quality disturbance evaluation results corresponding to each photovoltaic power change event, and a set of power quality disturbance evaluation results for photovoltaic power change events is formed.
[0080] This embodiment is applied to a certain rated capacity. The monitoring duration for the photovoltaic grid-connected power generation system is... The sampling period is During the monitoring process, active power at the photovoltaic source side, grid-connected voltage, grid-connected reactive power, harmonic voltage content, negative sequence voltage component, actual output current at the grid-connected point, and operating data such as inverter operating status, control mode, active power limit, reactive power limit, and current limit are collected synchronously according to a unified time base to form a synchronized operating dataset. Specific changes are as follows: Figure 2 , 3 As shown.
[0081] First, the active power sequence from the photovoltaic source side is smoothed using a moving average, and the smoothing window width is taken as half. Subsequently, photovoltaic power change events are identified using the aforementioned dual-threshold criterion for power change rate, and the change pattern is determined based on the power change amplitude, maximum power change rate, duration, and reversal of power change direction. In this embodiment, the first-level trigger threshold is set to... The secondary maintenance threshold is taken as follows: The duration of the first-level trigger is taken as The duration of secondary confirmation is taken The duration of the event ends is taken as The time interval for merging adjacent events in the same direction is taken as After event identification and validity screening, two valid photovoltaic power change events were identified. The following explanation uses the first photovoltaic power change event as an example.
[0082] The start time of the first photovoltaic power change event is The end time is Thus, the duration of the event is obtained. The difference between the initial power and the final power of the event is... Therefore, the direction of power change is This means that the active power on the photovoltaic source side shows an increasing trend. Further statistical analysis of the power change rate at each synchronous sampling point during the event yields the maximum power change rate. Based on the characteristics of the power change rate during the event, no significant directional reversal occurred during the event, i.e. Meanwhile, the duration of this event was... The duration exceeds the maximum duration threshold corresponding to a step-type event, therefore the step-type change condition is not met. Thus, the change pattern of the first event is... The event was determined to be a gradual power increase event.
[0083] Furthermore, the control conditions are determined based on the inverter's operating status, control mode, and various constraint states during the event. During the event, the inverter is in MPPT control mode, and the active power constraint flag, reactive power regulation constraint flag, and current constraint flag are not triggered. Therefore, its control conditions are expressed as follows: Since the aforementioned control conditions remained unchanged during the event, only one control condition segment was formed, with a time range of [missing information]. .
[0084] This embodiment pre-builds a database of historical normal operation events, which includes... A number of valid historical normal events are identified. For each event, historical normal events are first screened according to the direction of power change and control conditions. Among the historical normal events that meet the matching conditions, the power change characteristic distance is further calculated based on the power change amplitude, maximum power change rate, and duration, and each feature is then normalized. In this embodiment, the weights corresponding to the three features are respectively... , and Select the first [item] according to the order of increasing comprehensive feature distance. A set of historical normal events as normal reference events. Therefore, eight historical normal events were used as reference samples for normal response modeling.
[0085] This embodiment selects the grid connection point voltage. This is explained as a power quality indicator. The grid connection point voltage is expressed in per-unit values, where... The corresponding grid connection point rated voltage. The voltage reference value is determined based on stable operating data prior to the event's start time, and the actual voltage response during the event is calculated using the event's start time as the zero time point. For the normal reference event set... For each historical normal event, its normal voltage response was calculated using the same method, and the amplitude was normalized according to the power change between historical events. Subsequently, the normalized voltage responses of the eight historical normal events were fused to obtain the theoretical normal voltage response of the event, and the normal response error was determined based on the dispersion between the historical responses.
[0086] For the first event, its specific voltage normal response error is taken as... and adopt Establish a normal response range. In real-world scenarios... At that time, the relative time of the event is Theoretical normal voltage response is The corresponding normal response range is In real time At that time, the theoretical normal voltage response is The corresponding normal response range is .
[0087] Subsequently, the actual grid-connected voltage response during the event was compared point-by-point with the corresponding normal response range. When the actual response exceeded the upper boundary of the normal response range or fell below the lower boundary, the corresponding sampling point was identified as an abnormal response point, and further filtering was performed according to the continuous abnormal response persistence criterion. Calculation results show that the event occurred... The response time consistently exceeds the corresponding normal response range within the specified time period, thus forming an effective voltage anomaly response segment. After continuous screening, the effective voltage anomaly response segment of the event is: The duration of the abnormal response is Further, the abnormal deviation is calculated based on the degree of deviation between the actual response and the boundary of the normal response range, and the maximum absolute value of the abnormal deviation is taken as the maximum degree of abnormal deviation. This embodiment calculates... Simultaneously, the abnormal deviation is accumulated over time to obtain the degree of abnormal response accumulation. .
[0088] Subsequently, attribution analysis was performed on the aforementioned abnormal responses. In this embodiment, the preset response delay interval relative to the start time of the corresponding control condition segment is [missing information]. The weights of the three attribution conditions—time correlation, directional consistency, and control condition matching—are respectively taken as follows: , and Attribution confidence threshold is taken .
[0089] The start time of the control condition segment corresponding to the event is The abnormal voltage response started at the time when Therefore, the time offset of the abnormal response relative to the start time of the control condition segment is .because Located within the preset response delay range Therefore, the time-related criterion is taken. The direction of the power change of the event is... That is, the photovoltaic power changes upward. According to the normal power-voltage response relationship corresponding to the event, the actual abnormal voltage response direction is consistent with the expected response direction corresponding to the power change. Therefore, we take... Meanwhile, during the abnormal response period, the inverter control mode, active power, reactive power regulation, and current constraint states were all consistent with the control conditions corresponding to the event. Therefore, we take... Therefore, the attribution confidence of the event is... .because Therefore, it was determined that the abnormal voltage response was related to the event, and the grid connection point voltage was identified as the associated disturbance index of the event.
[0090] The anomaly severity index is further determined based on the maximum deviation, duration, and cumulative degree of the abnormal response. In this embodiment, the maximum deviation, duration, and cumulative deviation scale parameters corresponding to the grid connection point voltage are respectively taken as follows: , and The voltage anomaly level index corresponding to the event is obtained based on the aforementioned anomaly level calculation method, and then normalized according to the preset anomaly level upper limit of different power quality indicators.
[0091] After calculating the degree of abnormality for each power quality indicator, a comprehensive disturbance evaluation of the event is performed. In this embodiment, the grid connection point voltage and the 5th harmonic voltage content are used as the main monitoring indicators, and their indicator weights are respectively taken as follows: and The comprehensive disturbance classification threshold is taken as follows: and Based on the calculation results, the overall power quality disturbance level of the event is: .because Therefore, the event was classified as a general power quality disturbance event.
[0092] Based on the anomaly response identification and attribution results, only the grid-connected point voltage anomaly response correlated with photovoltaic power changes was identified. The 5th harmonic voltage content and other monitoring indicators did not generate any valid anomaly responses associated with the event. Therefore, the main disturbance indicators for the event are... The set of associated control conditions is .
[0093] Finally, in this embodiment, the start time of the event is... The end time is The duration is The change in power is The direction of change is The maximum power change rate is The change pattern is a gradual power increase event; its associated voltage anomaly response occurs in The duration is The maximum degree of abnormal deviation is The attribution confidence level is The overall power quality disturbance level is It was ultimately determined to be a general power quality disturbance event.
Claims
1. A method for identifying grid-connected power quality disturbances based on changes in photovoltaic source-side output, characterized in that, Includes the following steps: S1. Acquire and synchronize photovoltaic source-side power, grid connection point power quality and inverter operation data to form a synchronized operation dataset under a unified time reference; S2. Identify photovoltaic power change events based on the dual threshold continuous criterion of power change rate, and determine the change direction, change magnitude, change rate, duration and change pattern of the event based on multi-dimensional event characteristics; S3. Construct the inverter control state during the photovoltaic power change event, identify the control state changes, and segment the photovoltaic power change event according to the control state changes; S4. Based on the event characteristics of the photovoltaic power change event and the inverter control state, initially select a set of matching normal reference events from the historical normal operation events; S5. Based on the power change characteristics of the current photovoltaic power change event and the historical normal reference event, the historical normal response is normalized and fused to obtain the theoretical normal response and its discrete error of each power quality index of the current event, and then the corresponding normal response range is constructed. S6. Compare the actual response of each power quality index within each control condition section with the corresponding normal response range, identify the effective abnormal response section through continuous criteria, and calculate the occurrence time, duration, maximum abnormal deviation and cumulative degree of abnormal response. S7. Perform joint attribution of time, direction and control conditions on effective abnormal responses, calculate the attribution confidence, and determine the abnormality index based on the maximum deviation, duration and cumulative degree of abnormal responses that meet the attribution conditions. At the same time, associate the corresponding inverter control state to form an abnormal response set containing abnormal characteristics and associated control conditions. S8. Normalize and merge the abnormality of each control condition segment and each power quality indicator, determine the comprehensive power quality disturbance degree of photovoltaic power change events and classify them according to the duration of each segment, and at the same time determine the main disturbance indicators and associated control conditions, and form and output the power quality disturbance evaluation results of each photovoltaic power change event.
2. The method for identifying grid-connected power quality disturbances based on changes in photovoltaic source-side output as described in claim 1, characterized in that, Step S1 is as follows: The three-phase voltage and three-phase current at the photovoltaic source side are collected by a photovoltaic source-side measurement device, and the active power at the photovoltaic source side is obtained through power calculation. and reactive power ; The effective value of the grid connection point voltage is collected by a power quality analysis device at the grid connection point. Reactive power at grid connection point Harmonic voltage content and negative sequence voltage components ,in , The highest harmonic order collected or analyzed; Read the inverter's operating status flags through the inverter's communication interface. Control mode identifier Active power limit Reactive power limits Current limit and actual output current ; All types of data are time-aligned according to a unified time base, and a synchronized time series is established. For continuous quantities, interpolation is used for time alignment, while for operating status and control modes, discrete quantities are time-aligned using the nearest neighbor method. This ensures that photovoltaic source-side power, grid-connected power quality indicators, and inverter operating data correspond to the same sampling time, and the time-aligned data constitute a synchronized operating dataset. .
3. The method for identifying grid-connected power quality disturbances based on changes in photovoltaic source-side output as described in claim 2, characterized in that, Step S2 is as follows: From synchronously running datasets Extracting the active power sequence from the photovoltaic source side ,make , Indicates the first One synchronous sampling time node; calculate the power change rate based on the active power on the photovoltaic source side of adjacent synchronous time nodes. ,in, The time interval between adjacent synchronous sampling time nodes; when When, it indicates that the active power on the photovoltaic source side is increasing; when When this occurs, it indicates that the active power on the photovoltaic source side is decreasing; Extract power change rate from historical normal operation data And take their absolute values to form a set of absolute values of power change rate. , represented as , This represents the number of power change rate samples in the historical normal operation data; and the mean of the set of absolute values of the power change rate is calculated. and standard deviation ; Take confidence coefficient ,in To obtain the first-level threshold Further, take the proportionality coefficient. ,in To obtain the second-order threshold This ensures ;in, Used to trigger candidate segments for power change events Used to confirm and maintain power change events; When consecutive sampling points satisfy When the corresponding time interval is determined, it is identified as the candidate power change interval, and the sampling time that first meets this condition is determined as the first sampling time interval. The start time of each power change event Within the candidate power variation range, when consecutive sampling points satisfy... When the power change event is confirmed, the power change rate is monitored until consecutive sampling points meet the condition. At that time, determine the end time of the event. And denote the set of synchronous sampling points during the event as ; For the A confirmed photovoltaic power change event, defined as the event's starting power as... The final power is The change in power is The direction of change is ,in, This indicates an increase in power. This indicates a decrease in power; further define the power change amplitude, maximum power change rate, and duration as follows: , and ;in, This represents the minimum power change threshold used to exclude events with minute power changes. when At that time, the first The event is classified as a minor power change event and will not be included in subsequent power quality disturbance analysis; when When this occurs, it is identified as an effective photovoltaic power change event; For effective photovoltaic power change events, the number of times the power change direction reverses during the event period is counted. In this process, only adjacent power change rate sampling points that meet the secondary threshold criterion are statistically analyzed for direction reversal, so as to reduce the impact of random changes close to zero on the direction reversal judgment. when At that time, the changing form of the event It is classified as a fluctuation type to characterize the obvious cyclical changes during the event; when Then, further calculate the proportion of power change rate sampling points that reached the first-level threshold during the event period. ; When the event duration meets And the percentage meets At that time, the form will change. It is determined to be a step type, in which, This represents the maximum duration threshold used to determine step-type power change events; for events that do not meet the above step-type conditions and Valid events will change form It is determined to be a gradient type.
4. The method for identifying grid-connected power quality disturbances based on changes in photovoltaic source-side output as described in claim 3, characterized in that, Step S3 is as follows: Extract the first A number of effective photovoltaic power change events within the time interval Inverter operating status flags at each sampling time Control mode identifier Active power limit Reactive power limits Current limit and active power on the photovoltaic source side Reactive power at grid connection point and actual output current ; According to the inverter communication protocol and Decode the inverter's operating status to obtain the results. and control mode ,in Used to determine whether the inverter is in normal grid-connected operation. This indicates the currently used control mode, including maximum power point tracking mode and power limiting control mode; Sampling points indicating shutdown, malfunction, or other abnormal operating conditions are marked as abnormal operating points and removed in subsequent processing; For sampling points that are in normal grid-connected operation, a limit proximity coefficient is set. , and It is used to determine whether the active power, reactive power regulation and actual output current are close to the corresponding control limits; Active power constraint flag according to If and only if ,otherwise ; Reactive power regulation constraint flag according to If and only if ,otherwise ; Current constraint flag according to If and only if ,otherwise ; Among them, the sign is taken This indicates that the corresponding operating quantity is close to the corresponding control limit, and the flag is set. This indicates that the value is not yet close to the corresponding control limit. Control mode Combined with the three constraint flags mentioned above, it constitutes the first Control conditions at each sampling time This is used to characterize the actual control mode adopted by the inverter at that sampling moment, as well as the active power, reactive power regulation, and current constraint status. Compare the control conditions of adjacent sampling times within the event interval in the order of sampling time. At that time, the current sampling time will be... The control condition switching time is determined by the event start time. The switching times of each control condition and the end times of the events. As segmentation boundaries, events are divided into A continuous control condition segment, and represented as ,in, Indicates the first The first photovoltaic power change event The start and end times of each control condition segment. This indicates the control conditions corresponding to this segment; if the control conditions remain unchanged throughout the event, then... The event is classified as a single control condition event; if the control mode or any constraint flag changes once or multiple times, then... The event is identified as a multi-control condition event, and each control condition segment and its corresponding control condition are recorded separately.
5. The method for identifying grid-connected power quality disturbances based on changes in photovoltaic source-side output as described in claim 4, characterized in that, Step S4 is as follows: For the first determined by S2 For each valid photovoltaic power change event, a historical normal event with the same change direction and control conditions as the current event is selected from a pre-determined historical normal operation event, and a normal reference event matching the current event is determined based on the power change amplitude, maximum power change rate and duration. For the first condition that meets the above conditions For each historical normal event, calculate the power change characteristic distance based on the power change amplitude, maximum power change rate, and duration of the current event relative to that historical normal event: in, , and These represent the magnitude of the power change, the maximum rate of power change, and the duration of the current event, respectively. , and The first The magnitude of power change, the maximum rate of power change, and the duration of each historical normal event; , and These are the scale parameters for the corresponding features; , and These are the weights of the corresponding features, and ; Historical normal events that meet the same direction of change and the same control conditions are classified according to their power change characteristics and distance. Sort the events in ascending order, and select the closest historical normal events as reference events for the current event based on the available number of historical normal events; finally, obtain the [number of events]. A set of normal reference events corresponding to each current photovoltaic power change event. .
6. The method for identifying grid-connected power quality disturbances based on changes in photovoltaic source-side output as described in claim 5, characterized in that, Step S5 is as follows: For the current number A set of normal reference events obtained from S4 for each photovoltaic power change event. Determine the normal response corresponding to the current event; for grid connection point voltage Reactive power at grid connection point Harmonic voltage content and negative sequence voltage components Normal response calculations were performed separately, using any power quality index as an example. This indicates that the values include the grid connection point voltage. Reactive power at grid connection point Harmonic voltage content and negative sequence voltage components ; Starting time of the current event The previous stable operating range is used as the benchmark range, and the indicators within this range are... The average value is used as the baseline value for the current event. The time zero point is set as the start time of the event. Representing the time offset relative to the start time of the event, the actual response to the current event is then expressed as: ; For the normal reference event set obtained by S4 The first in For each normal reference event, its baseline value and actual response are calculated in the same way to obtain the actual response of historical normal events. Simultaneously, based on the power change determined by S2 for this historical normal event. The actual response is based on the power change of the current event. Normalization is performed to obtain the historical normal response corresponding to the current power change amplitude. in This represents the power change of the current event. For the first The power change of each normal reference event is determined according to S2; Historical normal responses corresponding to each normal reference event Perform equal-weighted fusion to obtain the current event index. Theoretical normal response ,in Represents the normal reference event set The number of normal reference events included; The normal response error is further determined based on the dispersion between the historical normal response and the theoretical normal response for each normal reference event: ； Determine the normal response range of the current event based on the theoretical normal response and the normal response error. ,in This is a preset range coefficient used to determine the width of the normal response range.
7. The method for identifying grid-connected power quality disturbances based on changes in photovoltaic source-side output as described in claim 6, characterized in that, Step S6 is as follows: For the current number The first photovoltaic power change event Each control condition section, and the power quality index of any grid connection point corresponding to the control condition section. Obtain the normal response range determined by S5. and the actual response corresponding to the current control condition segment. ; The actual response is compared with the normal response range determined by S5. When the actual response exceeds the upper boundary of the normal response range or falls below the lower boundary of the normal response range, the corresponding sampling time is determined as the index. Abnormal response points; When the actual response is within the normal response range, the corresponding sampling time is determined as the normal response point; that is, when ,or At that time, the corresponding sampling time is determined as the abnormal response point; For a given abnormal response point, the abnormal deviation amount is determined based on the degree of deviation of the actual response from the boundary of the normal response range. ; When the actual response exceeds the upper boundary of the normal response When the actual response is below the lower boundary of the normal response, When the actual response is within the normal response range, ; The duration of the abnormal response is determined based on the time range corresponding to the consecutive abnormal response points. The maximum absolute value of the abnormal deviation during the abnormal response period is determined as the maximum degree of abnormal deviation, i.e. The maximum value is obtained only for abnormal response points; simultaneously, the absolute value of the abnormal deviation corresponding to the abnormal response point is accumulated over time to obtain the degree of abnormal response accumulation. The summation is performed only for abnormal response points; For a current photovoltaic power change event that includes multiple control condition segments, the above-mentioned abnormal response identification is performed in each control condition segment to obtain the abnormal occurrence time, abnormal duration, maximum abnormal deviation degree, and abnormal response accumulation degree corresponding to each control condition segment and each power quality index.
8. The method for identifying grid-connected power quality disturbances based on changes in photovoltaic source-side output as described in claim 7, characterized in that, Step S7 is as follows: For the abnormal response segment determined by S6, obtain the direction of photovoltaic power change corresponding to the abnormal response and the inverter control conditions for the corresponding time period, and use the start time of the corresponding control condition segment. As a time reference point; The time correlation is determined based on the time offset between the start time of the abnormal response and the start time of the corresponding control condition segment; the time offset between the start time of the abnormal response and the start time of the corresponding control condition segment is defined as... ; When an abnormal response occurs within the corresponding control condition section, and Located within the preset response delay range Within the time frame, time-related criteria will be used. Set to 1 otherwise set to 0. This indicates that the abnormal response occurred within the preset time range. This indicates that the abnormal response does not meet the time correlation condition; Based on the changing direction of current photovoltaic power events The direction correlation is determined by the deviation direction of the inverter control mode and the abnormal response; when the direction of the abnormal response is consistent with the expected response direction corresponding to the current control mode and the photovoltaic power change direction, the direction consistency criterion is applied. Set to 1 otherwise set to 0. This indicates that the direction of the abnormal response is consistent with the direction of the expected response. This indicates that the two are inconsistent; Compare the actual control conditions during the abnormal response period with the corresponding control conditions determined in S3. The two are compared, and when they match, the control condition matching criterion is applied. Set to 1 otherwise set to 0. This indicates that the control conditions during the abnormal response period are consistent with the corresponding control condition segment. This indicates that the two are inconsistent; By combining time correlation, direction correlation, and control condition matching, the attribution confidence of abnormal responses is obtained. Its calculation is ,in, , and All are dimensionless criteria that take values of 0 or 1. , and These are the non-negative weights corresponding to time correlation, direction correlation, and control condition matching, respectively. ; Attribution confidence The value range is from 0 to 1; when When an abnormal response is identified, it is determined to be an abnormal power quality response associated with the current photovoltaic power change event; when At that time, it will not be considered as an associated anomaly response to the current photovoltaic power change event. The preset attribution confidence threshold; For abnormal responses identified as being related to current photovoltaic power change events, their corresponding grid-connected power quality indicators are used. As a disturbance indicator, the anomaly severity index is determined based on the maximum abnormal deviation of the abnormal response, the duration of the abnormality, and the cumulative degree of the abnormal response. ; The abnormality index is expressed as ,in , and These are the maximum degree of abnormal deviation, the duration of abnormality, and the cumulative degree of abnormal response determined by S6, respectively. , and These are the scale parameters corresponding to the maximum degree of anomaly deviation, the duration of anomaly, and the cumulative degree of anomaly response, respectively, used to normalize anomaly characteristics of different dimensions; Based on the control conditions corresponding to the occurrence of the abnormal response Determine the inverter control state associated with the abnormal response, including the control mode and the active power constraint, reactive power regulation constraint, and current constraint state; when the abnormal response mainly occurs within a certain control condition segment, determine the control conditions corresponding to that control condition segment. As an associated control condition for abnormal responses; For cases where the same photovoltaic power change event includes multiple control condition segments, the anomaly severity index corresponding to each control condition segment and each power quality index should be determined separately. Record the control condition section, power quality index, anomaly occurrence time, and anomaly duration corresponding to each anomaly response. Maximum abnormal deviation Accumulated abnormal response level And associated control conditions, thereby forming abnormal response information for current photovoltaic power change events.
9. The method for identifying grid-connected power quality disturbances based on changes in photovoltaic source-side output as described in claim 8, characterized in that, Step S8 is as follows: For the current number For each photovoltaic power change event, the degree of abnormality corresponding to each control condition segment and power quality index obtained from S7 is considered. The abnormality levels of different power quality indicators are normalized to eliminate differences in the dimensions and numerical ranges of different indicators. According to the formula The degree of anomaly is normalized, where, Power quality indicators The corresponding preset upper limit of abnormality level, and ; For the For each control condition segment, based on the abnormal responses corresponding to each power quality indicator determined in S7, establish a set of associated abnormal indicators corresponding to that control condition segment. , represented as ,in, Indicates the first The first photovoltaic power change event The set of power quality indicators that actually generate associated abnormal responses in each control condition section; when When not empty, only the set of associated abnormal indicators is considered. The normalized anomaly levels of various power quality indicators are weighted and fused to determine the overall disturbance level of the control condition section. , represented as ,in, Represents a set of associated abnormal indicators The number of power quality indicators included; when no abnormal response related to the current photovoltaic power change event is identified within the control condition segment, i.e. When empty, the overall disturbance level of the control condition section will be controlled. The value is determined to be 0; The segment weight is determined based on the duration of each control condition segment. Segment weights according to formula Confirmed, among which Indicates the first The first photovoltaic power change event The duration of each control condition segment; The number of control condition segments corresponding to the photovoltaic power change event is defined; based on the comprehensive disturbance degree of each control condition segment and the corresponding segment weight, the comprehensive power quality disturbance degree of the current photovoltaic power change event is determined. ; Based on the set of associated abnormal indicators corresponding to each control condition segment and the corresponding anomaly indexes for each power quality indicator. One or more power quality indicators with significant anomalies are identified as the set of main disturbance indicators for the current photovoltaic power change event. And record the control condition sections corresponding to the main disturbance indicators; The current photovoltaic power change events are classified according to the preset comprehensive disturbance classification threshold: when When, the current event is identified as a normal response event; when When, the current event is classified as a general power quality disturbance event; when When the current event is identified as a significant power quality disturbance event, then, and The preset comprehensive disturbance classification threshold is, and ; Based on the control condition segment determined in S3 and the correlation between the abnormal response and control conditions obtained in S7, determine the associated control conditions corresponding to the current photovoltaic power change event. When the abnormal response is mainly concentrated in one control condition segment, the control condition corresponding to that control condition segment is taken as the main associated control condition; when the abnormal response occurs in multiple control condition segments, the comprehensive disturbance degree, main disturbance index and associated control condition corresponding to each control condition segment are recorded respectively. Ultimately forming the current number Power quality disturbance evaluation results corresponding to photovoltaic power change events ,in, The event characteristics of photovoltaic power change events include the direction of change. Power variation range Maximum power change rate Duration and its changing forms ; To comprehensively assess the degree of power quality disturbance, For associated control conditions, This is the set of main disturbance indicators; For all valid photovoltaic power change events, S3 to S8 are executed respectively to obtain the power quality disturbance evaluation results corresponding to each photovoltaic power change event, and a set of power quality disturbance evaluation results for photovoltaic power change events is formed.