Distributed photovoltaic equipment detection method and system based on big data analysis

Abstract

The present invention discloses a distributed photovoltaic equipment detection method and system based on big data analysis. By collecting multi-source data, including equipment operation data, status data, power quality data and environmental data, comprehensive working information of the photovoltaic power station can be obtained, providing rich data support for subsequent analysis. The use of data cleaning and standardization technology ensures the quality and consistency of the data, which helps to improve the accuracy of subsequent model training and fault diagnosis. Through the multi-model fusion diagnosis method, the fault type is given in combination with characteristic indicators and detection methods, and a fault knowledge base is constructed, which not only improves the accuracy of fault diagnosis, but also provides valuable experience and data support for future fault prevention and maintenance.

Description

Technical Field

[0001] The present invention relates to the field of electromechanical equipment, and in particular to a distributed photovoltaic equipment detection method and system based on big data analysis. Background Art

[0002] Currently, photovoltaic power plants are characterized by a wide variety of equipment, a large number of components, and distributed deployment. PV modules are the smallest power generating units in a photovoltaic power plant. Due to the unique series-parallel structure of photovoltaic power plants, a single module failure can result in hundreds of times the power loss. This makes it difficult to accurately locate faults through manual operation and maintenance.

[0003] In summary, a distributed photovoltaic equipment detection method and system based on big data analysis is needed to address the shortcomings of the existing technology. Summary of the Invention

[0004] In view of the shortcomings of the existing technology, the present invention provides a distributed photovoltaic equipment detection method and system based on big data analysis, aiming to solve the above problems.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a distributed photovoltaic equipment detection method based on big data analysis, comprising the following steps:

[0006] Step S1: Multi-source data acquisition, collecting equipment operation data, status data, power quality data and environmental data, and transmitting them to the data processing module;

[0007] Step S2: Data processing: cleaning and standardizing the data, extracting basic features, derived features, and time series features;

[0008] Step S3: Use a time series database to store real-time data, manage structured data through clusters, and process unstructured data using HDFS and object storage. Implement data partitioning strategies based on geographic location, PV plant size, and time range.

[0009] Step S4: Real-time detection and anomaly detection: training the LSTM prediction model based on historical normal data, outputting the expected range of power and voltage, and combining quantile regression to perform dynamic threshold alarms;

[0010] Step S5: Use multiple models for fusion diagnosis, combine characteristic indicators and detection methods, give the fault type, and build a fault knowledge base.

[0011] Optionally, the equipment operation data in step S1 includes but is not limited to:

[0012] DC voltage, current and power of PV panels;

[0013] The inverter's AC voltage, current, power, frequency, efficiency, and operating mode;

[0014] Branch current and insulation resistance of the combiner box.

[0015] Optionally, the status data in step S1 includes but is not limited to component status, inverter status and bracket status;

[0016] Component status: hot spot temperature and hidden cracks;

[0017] Inverter status: cooling fan speed, capacitor aging index and fault code;

[0018] Bracket status: inclination and vibration.

[0019] Optionally, the power quality data and environmental data in step S1 include but are not limited to:

[0020] Power quality data: harmonic distortion rate, voltage fluctuation, frequency deviation and three-phase imbalance;

[0021] Environmental data: irradiance, ambient temperature, humidity and wind speed.

[0022] Optionally, the data cleaning in step S2 is performed in the following manner:

[0023] Missing value processing is done by counting the missing rate of each field, identifying missing fields, and filling them through linear interpolation or based on historical data of similar dates;

[0024] Outlier processing uses physical rules and hard boundary statistics to delete or replace outliers with sliding window means.

[0025] Optionally, the data in step S2 is normalized by:

[0026] Data from different sources are associated by device ID and timestamp, and non-uniformly sampled data is linearly resampled.

[0027] Optionally, the basic features, derived features, and time series features in step S2 are:

[0028] Basic features directly extract fields, including raw sensor data and device metadata;

[0029] Derived features include physical relationship features of inverter conversion efficiency, component performance ratio and temperature loss coefficient;

[0030] Time series characteristics include time dimension characteristics, lag characteristics and periodic characteristics.

[0031] Optionally, the multi-model fault diagnosis in step S5 is performed in the following manner:

[0032] Step A1: Select an appropriate single model for training and testing based on different fault types and characteristic indicators;

[0033] Step A2: Fusion the results of multiple single models, using weighted averaging and / or voting methods, combined with design rules to prioritize or conditionally filter the outputs of different models;

[0034] Step A3: Set a threshold for each fault type and determine the fault type based on the probability value output by the model.

[0035] A distributed photovoltaic equipment detection system based on big data analysis, which adopts the distributed photovoltaic equipment detection method based on big data analysis, includes: a multi-source data acquisition module, a data processing module, a data storage and management module, a real-time detection and anomaly detection module, and a fault diagnosis and knowledge base module;

[0036] The multi-source data acquisition module is responsible for collecting the operating data, status data, power quality data, and environmental data of PV panels, inverters, combiner boxes, and other equipment, and transmitting the data to the data processing module;

[0037] Data processing module, used for data cleaning, standardization and feature extraction;

[0038] The data storage and management module uses a time-series database to store real-time data, manages structured data through clusters, and processes unstructured data using HDFS and object storage technologies. It also implements data partitioning strategies based on geographic location, plant size, and time range.

[0039] The real-time detection and anomaly detection module is used to dynamically set thresholds and issue alarms based on the LSTM prediction model and the quantile regression method to monitor whether the output power and voltage parameters of photovoltaic equipment are within the expected range;

[0040] The fault diagnosis and knowledge base module is used to perform fusion diagnosis using multiple models. First, a suitable single model is selected for training and testing. Then, the results of multiple models are fused through weighted averaging or voting. Finally, the specific fault type is determined based on the set threshold and the fault knowledge base is updated.

[0041] Beneficial effects of the present invention:

[0042] 1. In the present invention, comprehensive operating information of the photovoltaic power station can be obtained through multi-source data collection, including equipment operation data, status data, power quality data, and environmental data, providing rich data support for subsequent analysis. The use of data cleaning and standardization technology ensures the quality and consistency of the data, which helps to improve the accuracy of subsequent model training and fault diagnosis.

[0043] 2. In this invention, a time series database, cluster management of structured data, HDFS and object storage are used to process unstructured data. A data partitioning strategy is implemented, enabling the system to effectively manage massive amounts of data and improve data query and analysis efficiency. Dynamic threshold alarms are implemented based on the LSTM prediction model and quantile regression method, enabling timely detection of whether the output power and voltage parameters of photovoltaic equipment deviate from the expected range, thereby quickly locating potential problems.

[0044] 3. In the present invention, the fault type is given by combining characteristic indicators and detection methods through a multi-model fusion diagnosis method, and a fault knowledge base is constructed, which not only improves the accuracy of fault diagnosis, but also provides valuable experience and data support for future fault prevention and maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 The present invention is a flow chart of a method.

[0046] Figure 2 This is an internal flow chart of step S5 of the present invention.

[0047] Figure 3 A schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION

[0048] In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0049] like Figure 1 As shown, a distributed photovoltaic equipment detection method based on big data analysis includes the following steps:

[0050] Step S1: Multi-source data acquisition, collecting equipment operation data, status data, power quality data and environmental data, and transmitting them to the data processing module;

[0051] PV module level: DC sensors, such as Hall effect current sensors, are deployed on each string of modules.

[0052] A temperature sensor, such as a SMD PT100, is installed on the back of the key component.

[0053] Read data through the inverter's built-in communication interface (RS485 / CAN).

[0054] Install the weather station near the PV array to avoid shadows.

[0055] Equipment operation data is transmitted through the smart meter or inverter built-in communication module, including but not limited to:

[0056] DC voltage, current and power of each group or string of PV panels;

[0057] The inverter's AC voltage, current, power, frequency, efficiency, and operating mode;

[0058] Branch current and insulation resistance of the combiner box;

[0059] Status data includes but is not limited to component status, inverter status, and rack status;

[0060] Component status: hot spot temperature and hidden cracks. Hot spot temperature is collected by infrared sensors, and hidden cracks are collected by EL detectors. These need to be collected offline regularly.

[0061] Inverter status: cooling fan speed, capacitor aging index and fault code;

[0062] Bracket status: inclination and vibration, collected by acceleration sensor;

[0063] Power quality data and environmental data include but are not limited to:

[0064] Power quality data: harmonic distortion, voltage fluctuation, frequency deviation, and three-phase imbalance, collected through a power quality analyzer or high-precision meter;

[0065] Environmental data: irradiance, ambient temperature, humidity and wind speed.

[0066] Step S2: Data processing: cleaning and standardizing the data, extracting basic features, derived features, and time series features;

[0067] Detection method for missing value processing:

[0068] Count the missing rate of each field, such as df.isnull().sum(), to identify frequently missing fields.

[0069] Flags consecutive absences due to sensor failure, such as an entire hour without data.

[0070] Numerical data, linear interpolation, suitable for smoothly changing temperature and irradiance.

[0071] Populate based on historical data of similar dates, such as the power value at the same time of the previous week.

[0072] Categorical data, such as fault codes, are marked as Unknown.

[0073] Detection method for outlier processing:

[0074] Threshold method: sets hard boundaries based on physical rules, such as the inability of PV module current to be negative;

[0075] Remove, such as transient noise.

[0076] Replace with sliding window mean, such as sudden temperature data;

[0077] Data normalization is achieved through:

[0078] Correlate data from different sources by device ID and timestamp, and perform linear resampling on non-uniformly sampled data;

[0079] The basic features, derived features, and time series features are:

[0080] Basic features directly extract fields, including raw sensor data and device metadata;

[0081] Derived features include physical relationship features of inverter conversion efficiency, component performance ratio and temperature loss coefficient;

[0082] Time series characteristics include time dimension characteristics, lag characteristics and periodic characteristics;

[0083] Step S3: Use a time series database to store real-time data, manage structured data through clusters, and process unstructured data using HDFS and object storage. Implement data partitioning strategies based on geographic location, PV plant size, and time range.

[0084] Step S4: Real-time detection and anomaly detection: training the LSTM prediction model based on historical normal data, outputting the expected range of power and voltage, and combining quantile regression to perform dynamic threshold alarms;

[0085] Step S5: Use multiple models for fusion diagnosis, combine characteristic indicators and detection methods, give fault types, and build a fault knowledge base;

[0086] like Figure 2 As shown in Figure 2, multi-model fault diagnosis is achieved through the following methods:

[0087] Step A1: Select an appropriate single model for training and testing based on different fault types and characteristic indicators;

[0088] Step A2: Fusion the results of multiple single models, using weighted averaging and / or voting methods, combined with design rules to prioritize or conditionally filter the outputs of different models;

[0089] Step A3: Set a threshold for each fault type and determine the fault type based on the probability value output by the model.

[0090] like Figure 3As shown, a distributed photovoltaic equipment detection system based on big data analysis adopts the distributed photovoltaic equipment detection method based on big data analysis, including: a multi-source data acquisition module, a data processing module, a data storage and management module, a real-time detection and anomaly detection module, and a fault diagnosis and knowledge base module;

[0091] The multi-source data acquisition module is responsible for collecting the operating data, status data, power quality data, and environmental data of PV panels, inverters, combiner boxes, and other equipment, and transmitting the data to the data processing module;

[0092] Data processing module, used for data cleaning, standardization and feature extraction;

[0093] The data storage and management module uses a time-series database to store real-time data, manages structured data through clusters, and processes unstructured data using HDFS and object storage technologies. It also implements data partitioning strategies based on geographic location, plant size, and time range.

[0094] The real-time detection and anomaly detection module is used to dynamically set thresholds and issue alarms based on the LSTM prediction model and the quantile regression method to monitor whether the output power and voltage parameters of photovoltaic equipment are within the expected range;

[0095] The fault diagnosis and knowledge base module is used to perform fusion diagnosis using multiple models. First, a suitable single model is selected for training and testing. Then, the results of multiple models are fused through weighted averaging or voting. Finally, the specific fault type is determined based on the set threshold and the fault knowledge base is updated.

[0096] The present invention acquires comprehensive operating information of photovoltaic power plants through multi-source data collection, including equipment operation data, status data, power quality data, and environmental data, providing rich data support for subsequent analysis. The use of data cleaning and standardization technologies ensures data quality and consistency, which helps improve the accuracy of subsequent model training and fault diagnosis.

[0097] By utilizing a time-series database, cluster management for structured data, HDFS, and object storage to process unstructured data, and implementing a data partitioning strategy, the system can effectively manage massive amounts of data, improving data query and analysis efficiency. Dynamic threshold alarms, based on LSTM prediction models and quantile regression methods, can promptly detect whether photovoltaic equipment output power and voltage parameters deviate from expected ranges, thereby quickly locating potential problems.

[0098] By using a multi-model fusion diagnosis method, combining characteristic indicators and detection methods to give the fault type and build a fault knowledge base, it not only improves the accuracy of fault diagnosis, but also provides valuable experience and data support for future fault prevention and maintenance.

[0099] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions or improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A distributed photovoltaic equipment detection method based on big data analysis, characterized in that: The following steps are involved: Step S1: Multi-source data acquisition, collecting equipment operation data, status data, power quality data and environmental data, and transmitting them to the data processing module; Step S2: Data processing: cleaning and standardizing the data, extracting basic features, derived features, and time series features; Data cleaning is done in the following ways: Missing value processing is done by counting the missing rate of each field, identifying missing fields, and filling them through linear interpolation or based on historical data of similar dates; Outlier processing uses physical rules to hard-bound statistical outliers and deletes or replaces them with the sliding window mean; Data normalization is achieved through: Correlate data from different sources by device ID and timestamp, and perform linear resampling on non-uniformly sampled data; The basic features, derived features, and time series features are: Basic features directly extract fields, including raw sensor data and device metadata; Derived features include physical relationship features of inverter conversion efficiency, component performance ratio and temperature loss coefficient; Time series characteristics include time dimension characteristics, lag characteristics and periodic characteristics; Step S3: Use a time series database to store real-time data, manage structured data through clusters, and process unstructured data using HDFS and object storage. Implement data partitioning strategies based on geographic location, PV plant size, and time range. Step S4: Real-time detection and anomaly detection: training the LSTM prediction model based on historical normal data, outputting the expected range of power and voltage, and combining quantile regression to perform dynamic threshold alarms; Step S5: Use multiple models for fusion diagnosis, combine characteristic indicators and detection methods, give fault types, and build a fault knowledge base; Multi-model fault diagnosis is performed in the following ways: Step A1: Select an appropriate single model for training and testing based on different fault types and characteristic indicators; Step A2: Fusion the results of multiple single models, using weighted averaging and / or voting methods, combined with design rules to prioritize or conditionally filter the outputs of different models; Step A3: Set a threshold for each fault type and determine the fault type based on the probability value output by the model; Through the multi-model fusion diagnosis method, the fault type is given by combining characteristic indicators and detection methods, and a fault knowledge base is constructed.

2. The distributed photovoltaic equipment detection method based on big data analysis according to claim 1 is characterized in that: The equipment operation data in step S1 includes: DC voltage, current and power of PV panels; The inverter's AC voltage, current, power, frequency, efficiency, and operating mode; Branch current and insulation resistance of the combiner box.

3. The distributed photovoltaic equipment detection method based on big data analysis according to claim 1 is characterized in that: The status data in step S1 includes component status, inverter status and bracket status; Component status: hot spot temperature and hidden cracks; Inverter status: cooling fan speed, capacitor aging index and fault code; Bracket status: inclination and vibration.

4. The distributed photovoltaic equipment detection method based on big data analysis according to claim 1 is characterized in that: The power quality data and environmental data in step S1 include: Power quality data: harmonic distortion rate, voltage fluctuation, frequency deviation and three-phase imbalance; Environmental data: irradiance, ambient temperature, humidity and wind speed.

5. A distributed photovoltaic equipment detection system based on big data analysis, using the distributed photovoltaic equipment detection method based on big data analysis as described in any one of claims 1 to 4, characterized in that: include: Multi-source data acquisition module, data processing module, data storage and management module, real-time detection and anomaly detection module, and fault diagnosis and knowledge base module; The multi-source data acquisition module is responsible for collecting the operating data, status data, power quality data, and environmental data of photovoltaic modules, inverters, and combiner boxes, and transmitting the data to the data processing module; Data processing module, used for data cleaning, standardization and feature extraction; The data storage and management module uses a time-series database to store real-time data, manages structured data through clusters, and processes unstructured data using HDFS and object storage technologies. It also implements data partitioning strategies based on geographic location, plant size, and time range. The real-time detection and anomaly detection module is used to dynamically set thresholds and issue alarms based on the LSTM prediction model and the quantile regression method to monitor whether the output power and voltage parameters of photovoltaic equipment are within the expected range; The fault diagnosis and knowledge base module is used to perform fusion diagnosis using multiple models. First, a suitable single model is selected for training and testing. Then, the results of multiple models are fused through weighted averaging or voting. Finally, the specific fault type is determined based on the set threshold and the fault knowledge base is updated.

Images