Method and system for monitoring operation data of photovoltaic power station

Abstract

The invention discloses a photovoltaic power station operation data monitoring method and system, and relates to the technical field of photovoltaic power station equipment monitoring, and the method comprises the steps: collecting equipment operation parameters to form an original data set; constructing an equipment association network by taking the equipment as a node; based on the data set and the association network, time sequence collaborative change and correlation analysis are carried out on the equipment parameters, and a collaborative change model is established; judging that equipment is abnormal through a model, and calling same-type parameters of directly connected nodes to form a to-be-verified set; in combination with the network connection edge weight, calculating the influence degree of the associated parameters on the abnormal equipment by applying a dynamic weight evaluation algorithm, and generating a priority list; and performing iterative optimization on the collaborative change model according to the priority list, and outputting a monitoring report. The system comprises a data acquisition module, a network construction module, an analysis modeling module, an anomaly judgment module, a weight evaluation module, a model optimization module, a data storage module and the like to realize abnormal associated factor identification and monitoring logic optimization.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power station equipment monitoring, and in particular to a method and system for monitoring photovoltaic power station operation data. Background Art

[0002] At present, abnormal monitoring of photovoltaic power station equipment mainly adopts the method of threshold judgment combined with historical data comparison. However, this method has the following defects: it cannot effectively integrate the multi-dimensional operating parameters of the equipment and the equipment correlation relationship, lacks dynamic weight evaluation of abnormal inducements and monitoring logic optimization mechanism, resulting in inaccurate identification of abnormal inducements and the inability of monitoring strategies to adapt to the dynamic changes in power station operating conditions. Summary of the Invention

[0003] In order to solve the technical problems in the existing technology that the multi-dimensional operating parameters of the equipment and the equipment association relationship cannot be effectively integrated, and there is a lack of dynamic weight evaluation of abnormal inducements and monitoring logic optimization mechanism, which leads to inaccurate identification of abnormal inducements and the inability of monitoring strategies to adapt to the dynamic changes in power station operating conditions, the present invention provides a monitoring method and system for photovoltaic power station operating data.

[0004] The technical solution adopted in the present invention is:

[0005] A first aspect of the present application provides a method for monitoring operation data of a photovoltaic power station, comprising the following steps:

[0006] Step 1: Collect the operating parameters of the photovoltaic power station equipment to form the original data set.

[0007] Step 2: Use each device in the PV power station as a node, and the electrical connection relationship, physical location relationship, and functional collaboration relationship between the devices as the connection edge to build a PV power station equipment association network.

[0008] Step 3: Based on the original data set and the photovoltaic power station equipment association network, a time series coordinated change analysis is performed on the operating parameters of each device node in the photovoltaic power station equipment association network, and a coordinated change model between the operating parameters of each device node is established.

[0009] Step 4: Determine whether a device is operating abnormally based on the coordinated change model. When a device is determined to be operating abnormally, retrieve the same type of operating parameters of the node device directly connected to the abnormal fluctuation device to form a parameter set to be verified.

[0010] Step 5: Based on the parameter set to be verified and combined with the weights of each connection edge in the device association network, a dynamic weight evaluation algorithm is used to calculate the abnormal correlation influence of each node device operating parameter on the abnormal device; the abnormal correlation influence of each node device is used as the sorting basis to generate a priority list of abnormal correlation factors from high to low.

[0011] Step 6: Based on the priority list of abnormal correlation factors, iteratively optimize the collaborative change model and output a monitoring report.

[0012] Preferably, step 1 includes the following sub-steps:

[0013] Sub-step 1.1: synchronously collect operating parameters of photovoltaic power station equipment according to a preset sampling frequency; the operating parameters include electrical parameters, mechanical parameters, and thermodynamic parameters.

[0014] Sub-step 1.2: Pre-process the collected electrical parameters, mechanical parameters, and thermodynamic parameters.

[0015] Sub-step 1.3: The pre-processed electrical parameters, mechanical parameters, and thermodynamic parameters are time-stamped in the order of acquisition time and stored in a time series database to form an original data set.

[0016] Preferably, step 2 includes the following sub-steps:

[0017] Sub-step 2.1: Construct a device node set and abstract the inverters, combiner boxes, transformers, PV module strings, switch cabinets, distribution cabinets, cables, sensors, circuit breakers, and disconnectors in the PV power station as nodes in the graph structure.

[0018] Sub-step 2.2: Establish a set of connection edges. The electrical connection edges establish the current transmission path relationship between devices based on the circuit topology. The physical location edges calculate the Euclidean distance and establish spatial association relationships based on the geographic coordinates of the devices within the power station. The functional collaboration edges establish functional collaboration links based on the device workflow.

[0019] Sub-step 2.3: Assign an initial weight to each connection edge. The electrical connection edge weight is determined by the circuit current carrying capacity ratio, the physical location edge weight is determined by the inverse of the device distance, and the functional collaboration edge weight is determined by the frequency of device collaborative work.

[0020] Sub-step 2.4: Use the graph database storage device to associate the network and establish a bidirectional index relationship between nodes and edges.

[0021] Preferably, step 3 includes the following sub-steps:

[0022] Sub-step 3.1: Analyze the time series correlation of the device node operating parameters and generate a correlation coefficient matrix.

[0023] Sub-step 3.2: Construct a parameter association network based on the correlation coefficient matrix to determine the causal relationship between the operating parameters.

[0024] Sub-step 3.3: Combine the initial weights of the connection edges of the device association network and perform weighted processing on the parameter association network to form a coordinated change model between the operating parameters of the device nodes.

[0025] Preferably, step 4 includes the following sub-steps:

[0026] Sub-step 4.1: Input the real-time collected operating parameters of each device into the coordinated change model to obtain the corresponding parameter prediction values ​​output by the model.

[0027] Sub-step 4.2: Calculate the absolute value of the deviation between the real-time value and the predicted value of each device operating parameter; when the absolute value of the deviation of any operating parameter exceeds the preset threshold corresponding to the parameter, determine that the corresponding device has abnormal fluctuations, and record the time of the abnormality, the type of abnormal parameter, and the deviation value.

[0028] Sub-step 4.3: Based on the connection relationship of the device-associated network, retrieve the same type of operating parameters of the node devices directly connected to the abnormal device to form a parameter set to be verified.

[0029] Preferably, step 5 includes the following sub-steps:

[0030] Sub-step 5.1: For each associated device node in the parameter set to be verified, retrieve the initial weight of the edge connecting it to the abnormal device.

[0031] Sub-step 5.2: Combine the initial weights of the connection edges, the normalized values ​​of the parameter fluctuation amplitudes, and the causal relationship between the operating parameters to evaluate the impact of the operating parameters of each associated device on the abnormal device and generate an abnormal correlation impact value for each associated parameter.

[0032] Sub-step 5.3: Sort the abnormal correlation impact values ​​of all correlation parameters and generate a priority list of abnormal correlation factors in descending order.

[0033] Preferably, step 6 includes the following sub-steps:

[0034] Sub-step 6.1: Adjust and optimize the parameter associations and device connection weights involved in the collaborative change model based on the priority list of abnormal correlation factors;

[0035] Sub-step 6.2: Organize the optimized model operation results to form a monitoring report including equipment operation status and abnormal situation analysis and output it.

[0036] A second aspect of the present application provides a photovoltaic power station operation data monitoring system, which applies the above-mentioned photovoltaic power station operation data monitoring method, including:

[0037] The data acquisition module is used to collect operating parameters of photovoltaic power station equipment to form an original data set.

[0038] The device association network construction module is used to construct a photovoltaic power station device association network using each device in the photovoltaic power station as a node and the electrical connection relationship, physical location relationship, and functional collaboration relationship between the devices as connection edges.

[0039] A time series analysis and modeling module is used to perform a time series collaborative change analysis on the operating parameters of each device node in the photovoltaic power station equipment association network based on the original data set and the photovoltaic power station equipment association network, and establish a collaborative change model between the operating parameters of each device node.

[0040] The abnormality determination and parameter retrieval module is used to determine whether a device is operating abnormally based on the collaborative change model. When a device is determined to be operating abnormally, the same type of operating parameters of the node device directly connected to the abnormal fluctuation device are retrieved to form a parameter set to be verified.

[0041] A dynamic weight evaluation and priority generation module is used to calculate the abnormal correlation influence of each node device operating parameter on the abnormal device based on the parameter set to be verified, combined with the weight of each connection edge in the device association network, using a dynamic weight evaluation algorithm, and using the abnormal correlation influence of each node device as the sorting basis to generate a priority list of abnormal correlation factors from high to low.

[0042] The model optimization and monitoring report module is used to iteratively optimize the collaborative change model according to the priority list of abnormal correlation factors and output a monitoring report.

[0043] The beneficial effects of the present invention are at least one of the following:

[0044] By building a device association network and integrating electrical connections, physical locations, and functional collaboration relationships, we can achieve multi-dimensional collaborative analysis of device operating parameters, breaking through the limitations of traditional single threshold judgment and improving the accuracy of identifying abnormal correlation factors.

[0045] Combined with the connection edge weights and parameter fluctuation characteristics of the device association network, it dynamically evaluates the impact of various associated factors on abnormal devices, generates a priority list, provides a troubleshooting path for operation and maintenance personnel, and shortens fault location time.

[0046] Based on the priority list of abnormal correlation factors, the coordinated change model parameters and monitoring thresholds are adjusted in real time, so that the monitoring system can adapt to the dynamic changes in the power plant operating conditions and reduce false alarms and missed alarms. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 This is a schematic diagram of the method flow of embodiment 1 of the present invention;

[0048] Figure 2 This is a system block diagram of embodiment 2 of the present invention. DETAILED DESCRIPTION

[0049] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0050] The first embodiment provides a method for monitoring the operation data of a photovoltaic power station. Figure 1 As shown, the following steps are included:

[0051] Step 1: Collect the operating parameters of the photovoltaic power station equipment to form the original data set.

[0052] In one possible implementation, step 1 includes the following sub-steps:

[0053] Sub-step 1.1: synchronously collect operating parameters of photovoltaic power station equipment according to a preset sampling frequency; the operating parameters include electrical parameters, mechanical parameters, and thermodynamic parameters.

[0054] For example, electrical parameters can be collected using voltage / current sensors to collect information such as the DC voltage on the inverter input side, the AC current on the output side, the current in the combiner box branch, and the open-circuit voltage of the PV panel string. Power sensors can be used to collect the active and reactive power of the transformer. Mechanical parameters can be collected using vibration sensors to collect the vibration frequency of the transformer core and the speed of the inverter cooling fan. Displacement sensors can be used to monitor the closed displacement of the circuit breaker contacts. Thermodynamic parameters can be collected using temperature sensors to collect the temperature of the inverter IGBT module, the temperature of the internal terminal blocks of the combiner box, and the surface temperature of the PV panels. Humidity sensors can be used to collect the ambient humidity within the distribution cabinet.

[0055] For example, a sampling frequency of 1 second / time is used for high-frequency changing electrical parameters (such as current and voltage), a sampling frequency of 1 minute / time is used for slowly changing thermodynamic parameters (such as equipment temperature), and mechanical parameters are set to 5 seconds / time according to the equipment operation cycle to ensure that different types of parameters are synchronized in the time dimension.

[0056] Sub-step 1.2: Pre-process the collected electrical parameters, mechanical parameters, and thermodynamic parameters.

[0057] For example, a sliding window filter is used to remove burrs from high-frequency noise in electrical parameters. Sudden outliers in temperature parameters are calibrated using the mean substitution method for adjacent time points. If a device is missing data for three consecutive sampling periods, the data is filled in based on the average value of the device's historical data for the same period. If a single sampling point is missing, linear interpolation of the data before and after is used to complete the missing data. Analog signals output by different sensors (such as 4-20mA current signals) are converted to digital quantities, with unified units (such as V for voltage and °C for temperature), and dimension normalization is performed.

[0058] Sub-step 1.3: The pre-processed electrical parameters, mechanical parameters, and thermodynamic parameters are time-stamped in the order of acquisition time and stored in a time series database to form an original data set.

[0059] Exemplarily, a millisecond-level timestamp is added to each piece of collected data, and the pre-processed data is stored in a distributed time series database according to a three-level index structure of "device number-parameter type-timestamp".

[0060] Through the above steps, full parameter coverage of the operating status of photovoltaic power station equipment can be collected, forming a structured raw data set with time, equipment, and parameter dimensions. This meets the engineering application requirements of the complex environment of photovoltaic power station sites.

[0061] Step 2: Use each device in the PV power station as a node, and the electrical connection relationship, physical location relationship, and functional collaboration relationship between the devices as the connection edge to build a PV power station equipment association network.

[0062] In one possible implementation, step 2 includes the following sub-steps:

[0063] Sub-step 2.1: Construct a device node set and abstract the inverters, combiner boxes, transformers, PV module strings, switch cabinets, distribution cabinets, cables, sensors, circuit breakers, and disconnectors in the PV power station as nodes in the graph structure.

[0064] For example, the inverters, combiner boxes, transformers, photovoltaic module strings, switch cabinets, distribution cabinets, cables, sensors, circuit breakers, and disconnectors in a photovoltaic power station are abstracted as graph nodes and assigned unique numbers (such as inverter INV-001 and combiner box CB-012). The node attributes include device type, rated parameters, and installation location coordinates.

[0065] Sub-step 2.2: Establish a set of connection edges. The electrical connection edges establish the current transmission path relationship between devices based on the circuit topology. The physical location edges calculate the Euclidean distance and establish spatial association relationships based on the geographic coordinates of the devices within the power station. The functional collaboration edges establish functional collaboration links based on the device workflow.

[0066] Sub-step 2.3: Assign an initial weight to each connection edge. The electrical connection edge weight is determined by the circuit current carrying capacity ratio, the physical location edge weight is determined by the inverse of the device distance, and the functional collaboration edge weight is determined by the frequency of device collaborative work.

[0067] For example, the electrical connection edge is based on the circuit design drawing to sort out the current transmission path (such as photovoltaic module string to junction box to inverter to transformer), and the edge attribute records the current carrying capacity ratio (such as the current carrying capacity of the cable on the inverter input side accounts for 80%).

[0068] Physical location edges use GPS to obtain the coordinates of each device and calculate the Euclidean distance between them. Location edges are established for devices with a distance of 5 meters or less, with an edge weight equal to the inverse of the distance (e.g., a distance of 2 meters results in a weight of 0.5). It should be noted that in PV power plants, similar devices (such as combiner boxes and inverters) are typically installed in clusters. The distance between adjacent devices is typically 1-3 meters (e.g., the distance between combiner boxes on the same rack is approximately 1.5 meters), and the distance between devices across racks is typically 5 meters or less (e.g., the distance between the inverter and the nearest combiner box). For devices exceeding 5 meters, the physical environmental correlation (such as temperature conduction and vibration transmission) between them is significantly reduced, and the direct impact on each other's operating status is negligible. Therefore, 5 meters is set as the threshold for valid physical correlation. Alternatively, the threshold can be set to 10 meters or 3 meters based on site survey results by simply modifying the distance comparison parameters. The inverse of the distance directly reflects this characteristic: the smaller the distance, the greater the weight, indicating that the closer the device, the greater the environmental impact on the target device.

[0069] The functional collaboration edge determines the data interaction relationship between devices based on the control logic (for example, the monitoring system sends a power adjustment command to the inverter every 10 seconds). The edge weight is the normalized value of the average daily number of interactions (for example, if the inverter and the monitoring system interact 2000 times a day, the weight is set to 1).

[0070] Sub-step 2.4: Use the graph database storage device to associate the network and establish a bidirectional index relationship between nodes and edges.

[0071] Step 3: Based on the original data set and the photovoltaic power station equipment association network, a time series coordinated change analysis is performed on the operating parameters of each device node in the photovoltaic power station equipment association network, and a coordinated change model between the operating parameters of each device node is established.

[0072] In one possible implementation, step 3 includes the following sub-steps:

[0073] Sub-step 3.1: Analyze the time series correlation of the device node operating parameters and generate a correlation coefficient matrix.

[0074] Sub-step 3.2: Construct a parameter association network based on the correlation coefficient matrix to determine the causal relationship between the operating parameters.

[0075] Sub-step 3.3: Combine the initial weights of the connection edges of the device association network and perform weighted processing on the parameter association network to form a coordinated change model between the operating parameters of the device nodes.

[0076] For example, the Pearson correlation coefficient is calculated for the inverter DC voltage and the PV panel string voltage to generate a correlation coefficient matrix (e.g., matrix element [inverter, panel string] = 0.92). 0.92 is the Pearson correlation coefficient, and in this example, 0.92 indicates that the inverter DC input voltage and the panel string output voltage are highly positively correlated in time series. That is, when the panel string voltage increases, the inverter DC voltage generally also increases, and this correlation is very significant.

[0077] For parameters with correlation coefficients greater than 0.7 among the operating parameters of device nodes, directed edges are established in the parameter association network to characterize the direction of coordinated changes between the parameters. The causal relationship between the parameters is determined through Granger causality testing (for example, if the module string voltage changes before the inverter DC voltage changes, it is determined to be "module string to inverter"). The connection edge weight of the device association network (for example, an electrical connection weight of 0.8) is multiplied by the parameter correlation coefficient (for example, 0.92) to obtain the edge weight of the coordinated change model (for example, 0.8 × 0.92 = 0.736), thus forming a coordinated change model.

[0078] In the specific implementation process, in the device association network, the nodes are devices (such as inverters, component strings), and the initial edge weights reflect the inherent connections between devices and represent the static associations at the device level. For example, the initial weight of the connection edge between device i and device j is W ij In this example, device i refers to the inverter and device j refers to the component string, such as the electrical connection weight W between the inverter and the component string. ij =0.8.

[0079] In the parameter association network, the nodes are operating parameters (such as inverter DC voltage, component string voltage), and the parameter correlation coefficient between the parameter p of device i and the parameter q of device j is F pq , represents the dynamic coordination relationship of the parameter layer. For example, the correlation coefficient F between the component string voltage and the inverter DC voltage pq =0.92.

[0080] The steps for constructing the collaborative change model (G) are as follows: using the nodes (devices) and connection edges (electrical / physical / functional relationships) of the device association network to ensure that the network skeleton is consistent with the actual topology of the power station.

[0081] It should be noted that "for each connection edge in the device association network, perform the following operations:

[0082] If there is a correlation coefficient F between the operating parameters of the equipment at both ends of the edgepq >0.7 (i.e., strongly correlated parameter pairs), the collaborative weight of the edge is the initial weight W of the connecting edge. ij Correlation coefficient F pq The product of (W ij ×F pq ) ; Logic: initial weight W of the device association network ij The correlation coefficient (F pq ) integration, and strengthen the credibility of the association of "both device connection and parameter coordination" through multiplication operation.

[0083] If there is no strongly correlated parameter pair, the initial weight W of the edge is retained ij Unchanged. Logic: Keep the device's inherent connection weight W ij , to avoid inflated weights due to insufficient parameter coordination, and to ensure that the model only runs based on verifiable device associations.

[0084] A "parameter pair" refers to a parameter combination consisting of two operating parameters belonging to the devices at both ends of an edge. Specifically:

[0085] The devices at both ends of the edge are device i and device j (device i is an inverter and device j is a module string). The parameter pair consists of an operating parameter of device i (such as the inverter DC voltage) and an operating parameter of device j (such as the module string voltage).

[0086] Parameter pairs are represented by the symbol "(p,q)", where p is the parameter of device i and q is the parameter of device j (e.g., p = inverter DC voltage, q = component string voltage).

[0087] When the parameter correlation coefficient F of the parameter pair (p,q) pq When it is greater than 0.7, it is determined to be a "strongly correlated parameter pair", which means that: p and q have a significant synergistic change trend in the time series (for example, when the component string voltage increases, the inverter DC voltage usually also increases); the strong correlation of this parameter pair can be used to strengthen the synergistic weight of the connection edge between devices i and j (for example, W ij ×F pq ) to quantify the synergistic influence between equipment parameters.

[0088] The collaborative change model (G) is a weighted directed graph model with the device association network as the skeleton and the initial weight W of the connecting edge. ij and parameter correlation coefficient F pqThe multiplied synergy weight is the edge, and the direction of the edge is determined by the causal relationship between the parameters (such as "component string to inverter" determined by the Granger causality test). The synergistic change model is used to predict the synergistic change trend of the associated device parameters (such as predicting the inverter DC voltage through the component string voltage) by calculating the weights of the weighted directed edges, thereby quantifying the synergistic change ability between the device parameters.

[0089] Step 4: Determine whether a device is operating abnormally based on the coordinated change model. When a device is determined to be operating abnormally, retrieve the same type of operating parameters of the node device directly connected to the abnormal fluctuation device to form a parameter set to be verified.

[0090] In one possible implementation, step 4 includes the following sub-steps:

[0091] Sub-step 4.1: Input the real-time collected operating parameters of each device into the coordinated change model to obtain the corresponding parameter prediction values ​​output by the model.

[0092] Sub-step 4.2: Calculate the absolute value of the deviation between the real-time value and the predicted value of each device operating parameter; when the absolute value of the deviation of any operating parameter exceeds the preset threshold corresponding to the parameter, determine that the corresponding device has abnormal fluctuations, and record the time of the abnormality, the type of abnormal parameter, and the deviation value.

[0093] Sub-step 4.3: Based on the connection relationship of the device-associated network, retrieve the same type of operating parameters of the node devices directly connected to the abnormal device to form a parameter set to be verified.

[0094] It should be noted that the collaborative change model is essentially a weighted association prediction model. Based on the collaborative weights between devices (reflecting the closeness of device connections and the historical collaborative strength of parameters), the predicted values ​​of the target device parameters are calculated through the following steps:

[0095] Step 1: Determine the equipment and parameters (such as the voltage of the component string) directly associated with the target device (inverter).

[0096] Step 2: Retrieve the synergy weights of the two in the device association network (e.g., the synergy weight of the component string and the inverter is 0.95, indicating the intensity of the impact of the component string voltage change on the inverter DC voltage).

[0097] Step 3: Multiply the parameter value of the associated device by the collaborative weight to directly obtain the predicted value of the target device parameter.

[0098] For example, the real-time measured value of the inverter DC voltage, 350V, and the directly associated device parameters determined through the device association network (for example, the measured voltage value of the upstream component string, 400V) are input into the coordinated change model to perform the following calculations:

[0099] Associated equipment parameters: The measured value of the component string voltage is 400V (directly associated with the inverter through electrical connection, with a unique current transmission path); the synergy weight between the component string and the inverter is 0.95.

[0100] Prediction value calculation: Component string voltage × synergy weight = 400 V × 0.95 = 380 V (380 V is the inverter DC voltage prediction value output by the model);

[0101] Calculation of the absolute value of the inverter DC voltage deviation: |measured value - predicted value| = |350V - 380V| = 30V.

[0102] The preset threshold is twice the standard deviation of the historical data (for example, if the historical standard deviation is 10V, the threshold is set to 20V). When the absolute value of the deviation between the measured value of the equipment operating parameter and the predicted value of the coordinated change model is greater than 20V, it is judged to be abnormal, and the abnormal time, abnormal parameter type and absolute value of the deviation are recorded (for example, at 10:05 on June 15, 2024, the absolute value of the deviation of the inverter DC voltage is 30V).

[0103] The associated parameters are retrieved by querying the devices directly connected to the abnormal inverter INV-001 (such as the combiner box CB-012 and the component string MS-003) through the graph database, and obtaining the same type of parameters of these devices (such as the DC current of the combiner box CB-012 and the input voltage of the component string MS-003) to form a set to be verified.

[0104] It should be noted that in this example, assuming that inverter INV-001 detects a DC voltage anomaly (for example, 350V, deviating from the predicted value of 380V), based on the physical structure of the PV power station:

[0105] Upstream equipment: combiner box CB-012 (which aggregates the DC power of multiple PV module strings and inputs it into the inverter), module string MS-003 (which directly provides DC input for the inverter).

[0106] Downstream equipment: Transformer TR-003 (steps up the AC power output by the inverter and feeds it into the grid. It has no direct connection with DC voltage anomalies and is only used for structural description).

[0107] Adjacent equipment: other inverters in the same distribution box (may be affected by thermodynamic parameters).

[0108] These devices are associated with the abnormal inverter INV-001 through electrical connection, physical location or functional collaboration, and their operating status may directly or indirectly cause the abnormality of the inverter INV-001.

[0109] Extract parameters of the same type from related devices, which are parameters that are physically related to abnormal parameters (such as inverter DC voltage):

[0110] Upstream equipment: DC current / voltage of the combiner box (if the combiner box output is abnormal, it may cause abnormal inverter input).

[0111] Downstream equipment: Transformer input voltage / frequency (if the transformer fails, feedback may affect the inverter output).

[0112] Thermodynamic parameters: The temperature of other inverters in the same location (if the ambient temperature is too high, it may affect the performance of the inverter).

[0113] The purpose of integrating these parameters into a set to be verified is to narrow the scope of investigation and focus only on parameters directly associated with abnormal devices, avoiding the analysis of all devices in the entire site; and to provide an analysis basis for the subsequent step 5.

[0114] Step 5: Based on the parameter set to be verified and combined with the weights of each connection edge in the device association network, a dynamic weight evaluation algorithm is used to calculate the abnormal correlation influence of each node device operating parameter on the abnormal device; the abnormal correlation influence of each node device is used as the sorting basis to generate a priority list of abnormal correlation factors from high to low.

[0115] In a possible implementation, sub-step 5.1: for each associated device node in the parameter set to be verified, retrieve the initial weight of the edge connecting it with the abnormal device.

[0116] For example, for each associated device (such as combiner box CB-012), its electrical connection weight (0.8), physical location weight (0.6), and functional collaboration weight (0.7) with the inverter are retrieved, and the average value is taken as the initial weight (0.7).

[0117] Sub-step 5.2: Combine the initial weights of the connection edges, the normalized values ​​of the parameter fluctuation amplitudes, and the causal relationship between the operating parameters to evaluate the impact of the operating parameters of each associated device on the abnormal device and generate an abnormal correlation impact value for each associated parameter.

[0118] For example, for combiner box CB-012:

[0119] Initial weight = (electrical connection weight 0.8 + physical location weight 0.6 + functional collaboration weight 0.7) ÷ 3 = 0.7; in this example, the normalized value of the parameter fluctuation amplitude refers to the current fluctuation amplitude of the combiner box CB-012 + 20%, which is normalized to 1.2.

[0120] Since the module string voltage changes before the combiner box current changes (causal relationship), multiply it by a correction factor of 1.1.

[0121] It should be noted that the Granger causality test is used to determine the causal relationship between parameters. This method is a standard method for determining the causal direction between variables in time series analysis. The specific steps are:

[0122] Check whether the change in the module string voltage is the Granger cause of the change in the combiner box current;

[0123] If the probability value of the test result is less than 0.05 (statistically significant level), it is determined that "the change in the module string voltage precedes the change in the combiner box current", that is, there is a causal relationship (for example, an increase in the module string voltage causes an increase in the combiner box current).

[0124] The basis and purpose of the correction factor of 1.1 are used to quantify the magnitude of the increase in the impact of causal relationships. Specifically, when a clear causal relationship exists between parameters, the abnormal correlation impact value is increased by 10% (i.e., multiplied by 1.1) based on the original calculation. The correction factor is derived from historical abnormal event data. The correction factor can be adjusted based on the significance of the causal relationship.

[0125] Abnormal correlation influence value = 0.7 × 1.2 × 1.1 = 0.924 (rounded to 0.92).

[0126] Sub-step 5.3: Sort the abnormal correlation impact values ​​of all correlation parameters and generate a priority list of abnormal correlation factors in descending order.

[0127] For example, the priority sorting is to sort the abnormal correlation impact values ​​of all associated parameters from high to low, with the first 20% set as first-level factors (abnormal correlation impact values ​​> 0.8), the middle 50% as second-level factors (abnormal correlation impact values ​​between 0.5 and 0.8), and the last 30% as third-level factors (abnormal correlation impact values ​​< 0.5). Among them, (0.5 to 0.8) and (< 0.5) are thresholds for dividing the levels of abnormal correlation factors. Generate a list (such as first-level factors: DC current of combiner box CB-012 (abnormal correlation impact value is 0.84), output voltage of component string MS-003 (abnormal correlation impact value is 0.82)).

[0128] It should be noted that the combiner box current is the operating parameter of the combiner box equipment. It is associated with the inverter through the electrical connection edge, affecting the current on the inverter input side and thus affecting the voltage; the component string voltage is the operating parameter of the component string equipment and is directly related to the inverter DC voltage (the inverter DC voltage is mainly determined by the component string voltage). Both are related parameters of the inverter DC voltage.

[0129] It should be noted that in this example, by integrating the electrical connection weights (reflecting the tightness of current transmission), physical location weights (reflecting the correlation of environmental impact), and functional collaboration weights (reflecting the correlation of control logic) between devices, the inherent connections between devices are converted into computable quantitative indicators (initial weights), and then combined with the normalized value of the real-time parameter fluctuation amplitude to achieve multi-dimensional collaborative analysis of anomalies, avoiding misjudgment or omission due to a single parameter.

[0130] The normalization of the fluctuation amplitude (e.g., converting a +20% fluctuation into a normalized value of 1.2) allows the impact calculation to include both the inherent connection of the equipment (initial weight) and the real-time operating status (fluctuation amplitude), and can capture sudden implicit correlations in the short term (e.g., the impact of insufficient heat dissipation of adjacent equipment on the inverter under rare high temperatures).

[0131] By sorting the abnormal correlation impact and generating a priority list (e.g., the top 20% are first-level factors), operation and maintenance personnel can prioritize the associated factors that have the greatest impact on abnormal equipment (e.g., combiner box current and component string voltage with an impact greater than 0.8), thereby shortening fault location time.

[0132] Step 6: Based on the priority list of abnormal correlation factors, iteratively optimize the collaborative change model and output a monitoring report.

[0133] In a possible implementation, step 6 includes the following sub-steps:

[0134] Sub-step 6.1: Adjust and optimize the parameter associations and device connection weights involved in the collaborative change model based on the priority list of abnormal correlation factors;

[0135] Sub-step 6.2: Organize the optimized model operation results to form a monitoring report including equipment operation status and abnormal situation analysis and output it.

[0136] For example, the weight of the connection edge corresponding to the first-level correlation factor (such as the electrical edge from the combiner box to inverter INV-001) was increased by 10% (from 0.8 to 0.88) to enhance subsequent monitoring sensitivity to this correlation. The threshold range for the abnormal parameter was adjusted: the original threshold of ±20V, combined with the impact of 0.84, was changed to ±20×(1+0.84)=±36.8V to accommodate changing operating conditions. The report content includes: Abnormal device: inverter INV-001, abnormal time: 10:05, current voltage 350V (predicted value 380V); first-level factor: current fluctuation of combiner box CB-012 (impact of 0.84), with the recommendation to "immediately check the CB-012 current sensor"; model optimization: the weight of the combiner box to inverter INV-001 edge was adjusted to 0.88, and the threshold was updated to ±36.8V.

[0137] It should be noted that in this example, the original weight of 0.8 represents the inherent electrical connection tightness between the devices. Combined with the high impact of the current fluctuation of the junction box on the DC voltage of the inverter in the current anomaly (0.84), the weight of the connection edge is actively increased to 0.88, making the priority of this association relationship higher in subsequent monitoring.

[0138] The report clearly points out the associated equipment and parameters that are most likely to cause abnormal inverter DC voltage through the "First-level factor: current fluctuation of combiner box CB-012 (influence degree 0.84)", rather than just prompting "inverter DC voltage abnormality".

[0139] Traditional methods can only detect "low inverter DC voltage" but cannot determine whether it is "abnormal current in the upstream combiner box" or "equipment failure itself." However, this solution uses impact calculation and priority sorting to focus the investigation from "all related devices" to "the top 20% high-impact factors," improving investigation efficiency.

[0140] The second embodiment provides a monitoring system for photovoltaic power station operation data, which applies the above-mentioned monitoring method for photovoltaic power station operation data. Figure 2 As shown, including:

[0141] A data acquisition module, which is used to collect operating parameters of photovoltaic power station equipment to form an original data set;

[0142] The device association network construction module is used to construct a photovoltaic power station device association network using each device in the photovoltaic power station as a node and the electrical connection relationship, physical location relationship, and functional collaboration relationship between the devices as connection edges.

[0143] A time series analysis and modeling module, which is used to perform a time series coordinated change analysis on the operating parameters of each device node in the photovoltaic power station equipment association network based on the original data set and the photovoltaic power station equipment association network, and establish a coordinated change model between the operating parameters of each device node;

[0144] An abnormality determination and parameter retrieval module, which is used to determine whether a device is operating abnormally based on the coordinated change model. When a device is determined to be operating abnormally, the module retrieves the same type of operating parameters of the node devices directly connected to the abnormal fluctuation device to form a parameter set to be verified;

[0145] A dynamic weight evaluation and priority generation module is used to calculate the abnormal correlation influence of each node device operating parameter on the abnormal device based on the parameter set to be verified and the weight of each connection edge in the device association network, and generate a priority list of abnormal correlation factors from high to low using the abnormal correlation influence of each node device as the ranking basis;

[0146] The model optimization and monitoring report module is used to iteratively optimize the collaborative change model according to the priority list of abnormal correlation factors and output a monitoring report.

[0147] The above-described embodiments merely represent specific implementations of the present invention. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, and all such variations and improvements fall within the scope of protection of the present invention.

Claims

1. A method for monitoring photovoltaic power station operation data, characterized in that: The following steps are involved: Step 1: Collect the operating parameters of the photovoltaic power station equipment to form the original data set; Step 2: Use each device in the PV power station as a node and the electrical connection relationship, physical location relationship, and functional collaboration relationship between the devices as the connection edge to build a PV power station equipment association network; Step 3: Based on the original data set and the photovoltaic power station equipment association network, perform a time series coordinated change analysis on the operating parameters of each device node in the photovoltaic power station equipment association network, and establish a coordinated change model between the operating parameters of each device node; Step 4: Determine whether a device is operating abnormally based on the coordinated change model. When a device is determined to be operating abnormally, retrieve the same type of operating parameters of the node devices directly connected to the abnormal fluctuation device to form a parameter set to be verified; Step 5: Based on the parameter set to be verified and the weights of each connection edge in the device association network, a dynamic weight evaluation algorithm is used to calculate the abnormal correlation influence of each node device operating parameter on the abnormal device; the abnormal correlation influence of each node device is used as the ranking basis to generate a priority list of abnormal correlation factors from high to low; Step 6: Based on the priority list of abnormal correlation factors, iteratively optimize the collaborative change model and output a monitoring report.

2. A method for monitoring photovoltaic power station operation data according to claim 1, characterized in that: The step 1 includes the following sub-steps: Sub-step 1.1: synchronously collecting operating parameters of photovoltaic power station equipment according to a preset sampling frequency; the operating parameters include electrical parameters, mechanical parameters, and thermodynamic parameters; Sub-step 1.2: Pre-processing the collected electrical parameters, mechanical parameters, and thermodynamic parameters; Sub-step 1.3: The pre-processed electrical parameters, mechanical parameters, and thermodynamic parameters are time-stamped in the order of acquisition time and stored in a time series database to form an original data set.

3. The method for monitoring photovoltaic power station operation data according to claim 1, characterized in that: The step 2 includes the following sub-steps: Sub-step 2.1: Construct a device node set, abstracting the inverters, combiner boxes, transformers, PV panel strings, switchgear, distribution cabinets, cables, sensors, circuit breakers, and disconnectors in the PV power station as nodes in a graph structure; Sub-step 2.2: Establish a set of connection edges. The electrical connection edges establish the current transmission path relationship between devices based on the circuit topology. The physical location edges calculate the Euclidean distance and establish spatial association relationships based on the geographic coordinates of the devices within the power station. The functional collaboration edges establish functional collaboration links based on the device workflow. Sub-step 2.3: Assign initial weights to each connection edge. The electrical connection edge weight is determined by the circuit current carrying capacity ratio, the physical location edge weight is determined by the inverse of the device distance, and the functional collaboration edge weight is determined by the frequency of device collaboration. Sub-step 2.4: Use the graph database storage device to associate the network and establish a bidirectional index relationship between nodes and edges.

4. The method for monitoring photovoltaic power station operation data according to claim 3, characterized in that: Step 3 includes the following sub-steps: Sub-step 3.1: Analyze the time series correlation of the device node operating parameters and generate a correlation coefficient matrix; Sub-step 3.2: Construct a parameter association network based on the correlation coefficient matrix to determine the causal relationship between the operating parameters; Sub-step 3.3: Combine the initial weights of the connection edges of the device association network and perform weighted processing on the parameter association network to form a coordinated change model between the operating parameters of the device nodes.

5. The method for monitoring photovoltaic power station operation data according to claim 4, characterized in that: The step 4 includes the following sub-steps: Sub-step 4.1: Input the real-time collected operating parameters of each device into the coordinated change model to obtain the corresponding parameter prediction values ​​output by the model; Sub-step 4.2: Calculate the absolute value of the deviation between the real-time value and the predicted value of each device operating parameter; when the absolute value of the deviation of any operating parameter exceeds the preset threshold corresponding to that parameter, determine that the corresponding device has an abnormal fluctuation, and record the time of the abnormality, the type of abnormal parameter, and the deviation value; Sub-step 4.3: Based on the connection relationship of the device-associated network, retrieve the same type of operating parameters of the node devices directly connected to the abnormal device to form a parameter set to be verified.

6. A method for monitoring photovoltaic power station operation data according to claim 5, characterized in that: The step 5 includes the following sub-steps: Sub-step 5.1: For each associated device node in the parameter set to be verified, retrieve the initial weight of the edge connecting it to the abnormal device; Sub-step 5.2: Combine the initial weights of the connection edges, the normalized values ​​of the parameter fluctuation amplitudes, and the causal relationship between the operating parameters to evaluate the impact of the operating parameters of each associated device on the abnormal device and generate an abnormal correlation impact value for each associated parameter; Sub-step 5.3: Sort the abnormal correlation impact values ​​of all correlation parameters and generate a priority list of abnormal correlation factors in descending order.

7. The method for monitoring photovoltaic power station operation data according to claim 6, characterized in that: Step 6 includes the following sub-steps: Sub-step 6.1: Adjust and optimize the parameter associations and device connection weights involved in the collaborative change model based on the priority list of abnormal correlation factors; Sub-step 6.2: Organize the optimized model operation results to form a monitoring report including equipment operation status and abnormal situation analysis and output it.

8. A photovoltaic power station operation data monitoring system, characterized in that: The method for monitoring photovoltaic power station operation data according to any one of claims 1 to 7 comprises: A data acquisition module, which is used to collect operating parameters of photovoltaic power station equipment to form an original data set; A device association network construction module is used to construct a photovoltaic power station device association network using each device in the photovoltaic power station as a node and the electrical connection relationship, physical location relationship, and functional collaboration relationship between the devices as connection edges; A time series analysis and modeling module, which is used to perform a time series coordinated change analysis on the operating parameters of each device node in the photovoltaic power station equipment association network based on the original data set and the photovoltaic power station equipment association network, and establish a coordinated change model between the operating parameters of each device node; An abnormality determination and parameter retrieval module, which is used to determine whether a device is operating abnormally based on the coordinated change model. When a device is determined to be operating abnormally, the module retrieves the same type of operating parameters of the node devices directly connected to the abnormal fluctuation device to form a parameter set to be verified; A dynamic weight evaluation and priority generation module is used to calculate the abnormal correlation influence of each node device operating parameter on the abnormal device based on the parameter set to be verified and the weight of each connection edge in the device association network, and generate a priority list of abnormal correlation factors from high to low using the abnormal correlation influence of each node device as the ranking basis; The model optimization and monitoring report module is used to iteratively optimize the collaborative change model according to the priority list of abnormal correlation factors and output a monitoring report.

Images