A fault detection method for a photovoltaic micro-inverter

Abstract

The present invention discloses a fault detection method for a photovoltaic micro-inverter, which relates to the technical field of inverter fault detection. The method collects battery status information of a plurality of photovoltaic panels constituting a photovoltaic array, divides the photovoltaic array into different array partitions according to the arrangement positions of the photovoltaic micro-inverters connected to the photovoltaic panels, monitors the inverter operating parameters of the photovoltaic micro-inverters under each array partition, combines the battery status information of the photovoltaic panels under each array partition to generate battery pack status information, performs data processing and feature extraction on the battery pack status information and the inverter operating parameters to generate partition feature information of each array partition and inputs the information into a preset fault preliminary screening library, marks the partition feature information of suspected faults by the fault preliminary screening library, performs pattern recognition on the partition feature information of each array partition suspected of having a fault, and locates the photovoltaic micro-inverters under all array partitions that ultimately have faults.

Description

Technical Field

[0001] The present invention relates to the technical field of inverter fault detection, and in particular to a fault detection method for a photovoltaic micro-inverter. Background Art

[0002] A photovoltaic micro-inverter is a power electronic device used in photovoltaic power generation systems, which include but are not limited to photovoltaic arrays. Its main function is to convert the direct current (DC) generated by photovoltaic panels into alternating current (AC) for subsequent power supply or grid connection. The photovoltaic micro-inverter is a key component in the photovoltaic power generation system, and its operation affects the efficiency of photovoltaic power generation.

[0003] However, during long-term operation, photovoltaic micro-inverters are prone to failure due to various internal or external factors, which directly affects the operation of the photovoltaic micro-inverter and indirectly affects the normal operation of the photovoltaic power generation system. At present, the fault detection methods for photovoltaic micro-inverters often have poor detection effects. Therefore, there is an urgent need for an effective detection method to comprehensively consider the impact of various influencing factors on photovoltaic micro-inverters and efficiently and accurately locate the faults of photovoltaic micro-inverters. Summary of the Invention

[0004] In order to solve the above problems, an object of the present invention is to provide a fault detection method for a photovoltaic micro-inverter.

[0005] The object of the present invention can be achieved by the following technical solution: A fault detection method for a photovoltaic micro-inverter comprises the following steps:

[0006] Step S1: collecting battery status information of a plurality of photovoltaic panels constituting a photovoltaic array, dividing the photovoltaic array into different array partitions according to the arrangement positions of the photovoltaic micro-inverters connected to the photovoltaic panels in the photovoltaic array, and monitoring the inverter operating parameters of the photovoltaic micro-inverters corresponding to each array partition;

[0007] Step S2: combining the battery status information of all photovoltaic panels in each array partition to generate battery pack status information, performing data processing and feature extraction based on the battery pack status information and inverter operating parameters to generate partition feature information for each array partition;

[0008] Step S3: inputting the partition characteristic information of each array partition into a preset fault screening library, and then marking the partition characteristic information of the array partition suspected of having a fault in the fault screening library;

[0009] Step S4: Using a machine learning algorithm, pattern recognition is performed on the partition feature information of each array partition suspected of having a fault, thereby locating the photovoltaic micro-inverters in all array partitions that ultimately have a fault, as well as the specific fault types of the photovoltaic micro-inverters.

[0010] Furthermore, the process of collecting battery status information of a plurality of photovoltaic panels constituting the photovoltaic array and dividing the photovoltaic array into different array partitions according to the arrangement positions of photovoltaic micro-inverters connected to the photovoltaic panels in the photovoltaic array includes:

[0011] The photovoltaic array consists of several photovoltaic panels. Each photovoltaic panel is numbered and denoted as i, where i = 1, 2, 3, ..., n, and n is a natural number greater than 0. Battery status information corresponding to each photovoltaic panel is collected. The battery status information includes battery temperature, operating voltage, operating current, and battery internal resistance.

[0012] A number of photovoltaic micro-inverters are arranged on the photovoltaic array. Each photovoltaic micro-inverter is used to connect to R photovoltaic panels, where R is an integer from 1 to 3. The photovoltaic micro-inverter is used to convert the direct current of the photovoltaic panel into alternating current. According to the arrangement positions of the photovoltaic micro-inverters connected to the photovoltaic panels in the photovoltaic array, the photovoltaic panels connected to the corresponding photovoltaic micro-inverters are divided into the same array partition on the photovoltaic array, and then a number of array partitions corresponding to the several photovoltaic micro-inverters on the photovoltaic array are divided.

[0013] Furthermore, the process of monitoring the inverter operating parameters of the photovoltaic micro-inverter corresponding to each array partition includes:

[0014] Different types of sensors and power monitoring equipment are arranged in each array partition. The different types of sensors include temperature sensors, humidity sensors, and air pressure sensors. The inverter operating parameters of the corresponding photovoltaic micro-inverter are obtained through the sensors and power monitoring equipment in each array partition.

[0015] The inverter operating parameters include a first operating parameter and a second operating parameter;

[0016] The first operating parameter includes the ambient temperature collected by the temperature sensor, the ambient humidity collected by the humidity sensor, and the ambient atmospheric pressure collected by the pressure sensor;

[0017] The second operating parameters include the inverter DC input voltage, inverter AC output voltage, inverter DC input current, inverter AC output current, inverter DC input power and inverter AC output power collected by the power monitoring equipment.

[0018] Furthermore, the process of combining the battery status information of all photovoltaic panels in each array partition to generate battery pack status information includes:

[0019] The battery status information of each photovoltaic battery group is classified according to the timestamp, and then the battery time sequence status information corresponding to each photovoltaic battery group at several timestamps is generated. The battery time sequence status information consists of the battery temperature, operating voltage, operating current and battery internal resistance at the same timestamp;

[0020] The battery timing status information of all photovoltaic battery groups in the same array partition is combined according to the same timestamp to generate the battery group timing status information corresponding to the photovoltaic arrays at several timestamps. The battery group timing status information corresponding to all timestamps of each photovoltaic array is combined to generate the battery group status information corresponding to each photovoltaic array.

[0021] Furthermore, the process of performing data processing and feature extraction based on the battery pack status information and the inverter operating parameters to generate partition feature information for each array partition includes:

[0022] By performing data processing on the battery pack status information, the battery pack status information is decomposed into battery pack timing status information corresponding to a plurality of timestamps; performing data processing on the first operating parameter and the second operating parameter included in the inverter operating parameter, the inverter operating parameter is decomposed into inverter timing operating parameters corresponding to a plurality of timestamps;

[0023] Construct a one-dimensional data analysis timeline. On the one-dimensional data analysis timeline, perform feature extraction on the battery pack timing state information and inverter timing operating parameters corresponding to each timestamp. This allows obtaining battery pack timing features corresponding to the battery pack timing state information of all photovoltaic panels in the array partition corresponding to each timestamp, as well as inverter timing features corresponding to the inverter timing operating parameters of the photovoltaic microinverters in the array partition corresponding to each timestamp.

[0024] The battery pack timing characteristics and inverter timing characteristics corresponding to each array partition at the same timestamp are integrated to generate the superimposed timing characteristics of the corresponding timestamps. The superimposed timing characteristics of all timestamps of each array partition are summarized to generate the partition characteristic information corresponding to each array partition.

[0025] Furthermore, the partition characteristic information of each array partition is input into a preset fault screening library, and the process of marking the partition characteristic information of the array partition suspected of having a fault by the fault screening library includes:

[0026] A pre-set fault screening library is built based on a machine learning model. This model is generated by processing, mining, and feature processing historical data of the PV array, combined with data modeling techniques. The machine learning model is used to analyze the characteristic information of all partitions corresponding to several PV array partitions when suspected faults occurred in the past.

[0027] The partition characteristic information of each array partition is input into the fault screening library. The fault screening library then marks the partition characteristic information of array partitions suspected of having faults. The partition characteristic information of the array partitions suspected of having faults is associated with a suspected fault identifier, which is recorded as doubt-Er-Sign. The partition characteristic information of other array partitions not suspected of having faults is associated with a safe operation identifier, which is recorded as safe-Sign.

[0028] Furthermore, the process of performing pattern recognition on the partition feature information of each array partition suspected of having a fault using a machine learning algorithm and then locating the photovoltaic micro-inverters in all array partitions that ultimately have a fault includes:

[0029] An initial pattern recognition model is constructed using a machine learning algorithm. The pattern recognition model is used to perform pattern recognition on the partition feature information of the array partitions suspected of having a fault. All the partition feature information associated with the suspected fault identifier is used as input parameters of the pattern recognition model.

[0030] Obtain a historical fault data set, which records several types of fault problems of photovoltaic micro-inverters at historical times. Synchronize the historical fault data set as input parameters of a pattern recognition model. The pattern recognition model outputs corresponding recognition output results based on the received input parameters. The recognition output results include: the device numbers of photovoltaic micro-inverters in all array partitions that ultimately have faults, and the specific fault types corresponding to the photovoltaic micro-inverters;

[0031] When the photovoltaic micro-inverter under the corresponding array partition is finally located to have a fault, the suspected fault identifier doubt-Er-Sign of the partition characteristic information corresponding to the array partition is replaced with a fault identifier, and the fault identifier is recorded as sure-Er-Sign. Maintenance personnel are arranged to repair all photovoltaic micro-inverters located with specific fault types. When the fault problem corresponding to the specific fault type of the photovoltaic micro-inverter is resolved, the fault identifier sure-Er-Sign is replaced with the safety identifier safe-Sign.

[0032] Furthermore, specific fault types of the photovoltaic micro-inverter include open circuit fault, short circuit fault, over-temperature fault, component aging fault, and capacitor and inductor fault.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] The photovoltaic array is divided into different array partitions according to the layout of the photovoltaic micro-inverters. Data processing and feature extraction are performed based on the inverter operating parameters of the photovoltaic micro-inverters in each array partition and the battery group status information generated by integrating the battery status information of all photovoltaic panels in the array partition to generate partition feature information for each array partition. The partition feature information of each array partition is input into the fault preliminary screening library. The partition feature information of the array partition suspected of having a fault is first marked. Then, pattern recognition is performed on the partition feature information of each array partition suspected of having a fault to locate the photovoltaic micro-inverters in all array partitions that ultimately have a fault and their specific fault types. The relevant battery status information of the photovoltaic panels and the influence of the inverter operating parameters of the photovoltaic micro-inverters themselves during operation on the operation of the photovoltaic micro-inverters are comprehensively considered. The corresponding partition feature information is obtained through feature extraction as the data basis for locating the specific fault type, thereby achieving efficient detection of corresponding faults of several photovoltaic micro-inverters in the photovoltaic array and improving the accuracy of fault detection to a certain extent. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 Flowchart of the present invention. DETAILED DESCRIPTION

[0036] like Figure 1 As shown, a fault detection method for a photovoltaic micro-inverter includes the following steps:

[0037] Step S1: collecting battery status information of a plurality of photovoltaic panels constituting a photovoltaic array, dividing the photovoltaic array into different array partitions according to the arrangement positions of the photovoltaic micro-inverters connected to the photovoltaic panels in the photovoltaic array, and monitoring the inverter operating parameters of the photovoltaic micro-inverters corresponding to each array partition;

[0038] Step S2: combining the battery status information of all photovoltaic panels in each array partition to generate battery pack status information, performing data processing and feature extraction based on the battery pack status information and inverter operating parameters to generate partition feature information for each array partition;

[0039] Step S3: inputting the partition characteristic information of each array partition into a preset fault screening library, and then marking the partition characteristic information of the array partition suspected of having a fault in the fault screening library;

[0040] Step S4: Using a machine learning algorithm, pattern recognition is performed on the partition feature information of each array partition suspected of having a fault, thereby locating the photovoltaic micro-inverters in all array partitions that ultimately have a fault, as well as the specific fault types of the photovoltaic micro-inverters.

[0041] It should be further explained that, in a specific implementation process, the process of collecting battery status information of a plurality of photovoltaic panels constituting a photovoltaic array and dividing the photovoltaic array into different array partitions according to the arrangement positions of photovoltaic micro-inverters connected to the photovoltaic panels in the photovoltaic array includes:

[0042] The photovoltaic array is composed of a plurality of photovoltaic panels, each of which is numbered and denoted as i, where i=1, 2, 3, ..., n, where n is a natural number greater than 0. Battery status information corresponding to each of the plurality of photovoltaic panels is collected, wherein the battery status information includes battery temperature, operating voltage, operating current, and battery internal resistance;

[0043] The photovoltaic array is provided with a plurality of photovoltaic micro-inverters, each photovoltaic micro-inverter is used to connect R photovoltaic panels, where R is an integer from 1 to 3, and the photovoltaic micro-inverter is used to convert the direct current (DC) power of the photovoltaic panels into alternating current (AC), which is then used to power the photovoltaic array or connect to the grid.

[0044] According to the layout positions of several photovoltaic micro-inverters connected to the photovoltaic panels in the photovoltaic array, the photovoltaic panels connected to the corresponding photovoltaic micro-inverters are divided into the same array partition on the photovoltaic array, and then several array partitions corresponding to the several photovoltaic micro-inverters on the photovoltaic array are divided.

[0045] It should be noted that there is a one-to-one correspondence between each photovoltaic micro-inverter and an array partition, and each photovoltaic micro-inverter is used to convert DC power into AC power for 1 to 3 photovoltaic panels under the corresponding array partition.

[0046] It should be further explained that, in a specific implementation process, the process of monitoring the inverter operating parameters of the photovoltaic micro-inverter corresponding to each array partition includes:

[0047] Arrange different types of sensors and power monitoring equipment in each array partition, including temperature sensors, humidity sensors, and air pressure sensors, and obtain inverter operating parameters of corresponding photovoltaic micro-inverters through the sensors and power monitoring equipment in each array partition;

[0048] The inverter operating parameters include a first operating parameter and a second operating parameter;

[0049] The first operating parameter includes the ambient temperature collected by the temperature sensor, the ambient humidity collected by the humidity sensor, and the ambient atmospheric pressure collected by the pressure sensor;

[0050] The second operating parameters include the inverter DC input voltage, inverter AC output voltage, inverter DC input current, inverter AC output current, inverter DC input power, and inverter AC output power collected by the power monitoring equipment;

[0051] The first operating parameter is marked as D1, and the ambient temperature, ambient humidity and ambient atmospheric pressure included in the first operating parameter are marked as C 温度 , SD and Pa, and the first operating parameter is expressed as: D1= <C 温度 , SD, Pa>;

[0052] The second operating parameter is marked as D2, and the inverter DC input voltage and inverter AC output voltage included in the second operating parameter are respectively marked as V 直流输入 and V 交流输出 The inverter DC input current and inverter AC output current are respectively recorded as A 直流输入 and A 交流输出 ,

[0053] The DC input power of the inverter and the AC output power of the inverter are respectively recorded as W 直流输入 and W 交流输出 , and the second operating parameter is expressed as follows:

[0054] D2= <V 直流输入 , V 交流输出 , A 直流输入 , A 交流输出 , W 直流输入 , W 交流输出 >.

[0055] It should be further explained that, in a specific implementation process, the process of combining the battery status information of all photovoltaic panels in each array partition to generate battery pack status information includes:

[0056] Set the information combination period and record it as T 组合 , then T 组合 =[t1, t2], where t1 is the start time of the information combination period and t2 is the end time of the information combination period. During the information combination period, the battery status information of all photovoltaic panels in each array partition is combined as follows:

[0057] The battery status information of each photovoltaic battery group is classified according to the timestamp, and then the battery time sequence status information corresponding to a plurality of timestamps of each photovoltaic battery group is generated. The battery time sequence status information is composed of the battery temperature, operating voltage, operating current and battery internal resistance at the same timestamp;

[0058] The battery timing status information of all photovoltaic battery groups in the same array partition is combined according to the same timestamp to generate the battery group timing status information corresponding to the photovoltaic arrays at several timestamps. The battery group timing status information corresponding to all timestamps of each photovoltaic array is combined to generate the battery group status information corresponding to each photovoltaic array.

[0059] It should be further explained that, in a specific implementation process, data processing and feature extraction are performed based on battery pack status information and inverter operating parameters to generate partition feature information for each array partition, including:

[0060] By performing data processing on the battery pack status information, the battery pack status information is decomposed into battery pack timing status information corresponding to a plurality of timestamps; performing data processing on the first operating parameter and the second operating parameter included in the inverter operating parameter, the inverter operating parameter is decomposed into inverter timing operating parameters corresponding to a plurality of timestamps;

[0061] Construct a one-dimensional data analysis timeline. On the one-dimensional data analysis timeline, perform feature extraction on the battery pack timing state information and inverter timing operating parameters corresponding to each timestamp. This allows obtaining battery pack timing features corresponding to the battery pack timing state information of all photovoltaic panels in the array partition corresponding to each timestamp, as well as inverter timing features corresponding to the inverter timing operating parameters of the photovoltaic microinverters in the array partition corresponding to each timestamp.

[0062] Integrate the battery pack timing characteristics and inverter timing characteristics corresponding to each array partition at the same timestamp to generate the superimposed timing characteristics of the corresponding timestamps. Summarize the superimposed timing characteristics of all timestamps of each array partition to generate the partition feature information corresponding to each array partition.

[0063] It should be noted that the battery pack timing characteristics corresponding to the battery pack timing status information and the inverter timing characteristics corresponding to the inverter timing operating parameters are obtained through feature extraction methods such as time domain feature extraction, frequency domain feature extraction and statistical feature extraction. The specific feature extraction operation is implemented based on historical data and big data technology and is carried out according to actual conditions.

[0064] It should be further explained that, in a specific implementation process, the partition characteristic information of each array partition is input into a preset fault screening library, and the process of the fault screening library marking the partition characteristic information of the array partition suspected of having a fault includes:

[0065] A pre-set fault screening library is built based on a machine learning model. This model is generated by processing, mining, and feature processing historical data of the PV array, combined with data modeling techniques. The model is used to analyze the characteristic information of all partitions corresponding to several PV array partitions when suspected faults occurred in the past.

[0066] The partition characteristic information of each array partition is input into a fault screening library. The fault screening library then labels the partition characteristic information of array partitions suspected of having faults. The partition characteristic information of the array partitions suspected of having faults is associated with a suspected fault identifier, which is recorded as doubt-Er-Sign. The partition characteristic information of other array partitions not suspected of having faults is associated with a safe operation identifier, which is recorded as safe-Sign.

[0067] The following are the steps to determine whether an array partition is suspected of failure:

[0068] Setting battery status information and inverter operating parameters, and a parameter determination set corresponding to partition feature information generated after data processing and feature extraction of the battery status information and inverter operating parameters, wherein the parameter determination set includes a plurality of determination intervals for determining the battery status information and a plurality of determination thresholds for determining the inverter operating parameters;

[0069] The several determination intervals for determining the battery status information include a battery temperature safety interval, an operating voltage safety interval, an operating current safety interval, and a battery internal resistance safety interval. The battery temperature safety interval, the operating voltage safety interval, the operating current safety interval, and the battery internal resistance safety interval are respectively denoted as Ω 安全温度 ,Ω 安全电压 ,Ω 安全电流 and Ω 安全电阻 ;

[0070] The battery temperature and battery internal resistance are respectively denoted as D 电池温度 and D 电池内阻 ;

[0071] The values of the operating voltage and the operating current are respectively denoted as D 工作电压 and D 工作电流 ;

[0072] When any of the following conditions is met, it is determined that the photovoltaic panels in the current array partition may have a fault problem. The conditions are as follows:

[0073] D 电池温度 Ω 安全温度 ;

[0074] D 电池内阻 Ω 安全电阻 ;

[0075] D 工作电压 Ω 安全电压 ;

[0076] D 工作电流 Ω 安全电流 ;

[0077] Otherwise, it is determined that there is no fault problem with the photovoltaic panels in the current array partition;

[0078] The plurality of determination thresholds for determining the inverter operating parameters include a temperature threshold corresponding to the ambient temperature of the first operating parameter, a humidity threshold corresponding to the ambient humidity, and a pressure threshold corresponding to the ambient atmospheric pressure; and also include a voltage loss threshold for conversion between the inverter DC input voltage and the inverter AC output voltage, a current loss threshold for conversion between the inverter DC input current and the inverter AC output current, and a power loss threshold for conversion between the inverter DC input power and the inverter AC output power, which are included in the second operating parameter;

[0079] The temperature threshold, humidity threshold and pressure threshold are respectively denoted as C 阈值 , SD 阈值 and Pa 阈值 ;

[0080] The voltage loss threshold, current loss threshold and power loss threshold are denoted as V 损耗阈值 、A 损耗阈值 and W 损耗阈值 ;

[0081] When any of the following conditions is met, it is determined that the photovoltaic micro-inverter in the current array partition may have a fault problem. The conditions are as follows:

[0082] C 温度 ≥C 阈值 ;

[0083] SD≥SD 阈值 ;

[0084] Pa≥Pa 阈值 ;

[0085] |V 直流输入 -V 交流输出 |≥V 损耗阈值 ;

[0086] |A 直流输入 -A 交流输出 |≥A 损耗阈值 ;

[0087] |W 直流输入 -W 交流输出 |≥W 损耗阈值 ;

[0088] Otherwise, it is determined that there is no fault problem in the photovoltaic micro-inverter under the current array partition;

[0089] When it is determined that the photovoltaic panels under the current array partition may have a fault problem, or when it is determined that the photovoltaic micro-inverters under the current array partition may have a fault problem, the current array partition is determined to be suspected of having a fault; otherwise, it is determined that the current array partition is not faulty.

[0090] The judgment rule for determining whether an array partition is suspected of failure is used as the rule for the machine learning model to analyze the partition feature information.

[0091] It should be further explained that, in the specific implementation process, the process of using a machine learning algorithm to perform pattern recognition on the partition feature information of each array partition suspected of having a fault, and then locating the PV microinverters in all array partitions that ultimately have a fault, as well as the specific fault types of the PV microinverters, includes the following:

[0092] An initial pattern recognition model is constructed using a machine learning algorithm. The pattern recognition model is used to perform pattern recognition on partition feature information of array partitions suspected of having a fault, and all partition feature information associated with suspected fault identifiers is used as input parameters of the pattern recognition model.

[0093] Acquire a historical fault data set, wherein the historical fault data set records several fault problems of photovoltaic micro-inverters at historical times, and synchronize the historical fault data set as an input parameter of a pattern recognition model;

[0094] The pattern recognition model outputs corresponding recognition output results based on the received input parameters, wherein the recognition output results include: the device numbers of all photovoltaic micro-inverters in all array partitions that ultimately have faults, and the specific fault types corresponding to the photovoltaic micro-inverters;

[0095] When the photovoltaic micro-inverter under the corresponding array partition is finally located to have a fault, the suspected fault identifier doubt-Er-Sign of the partition characteristic information corresponding to the array partition is replaced with a fault identifier, and the fault identifier is recorded as sure-Er-Sign;

[0096] Obtain the recognition accuracy of the initial pattern recognition model and record it as Sc. Set a lower threshold corresponding to the recognition accuracy and record it as τ. When Sc ≥ τ, do not perform any operation. When Sc < τ, increase the amount of data in the historical fault data set and use it as the training set to train the pattern recognition model.

[0097] Arrange maintenance and operation personnel to repair all photovoltaic micro-inverters that have been identified as having specific fault types, and generate corresponding maintenance logs. The maintenance logs record detailed fault information and maintenance and operation information of the photovoltaic micro-inverters. When the fault problem corresponding to the specific fault type of the photovoltaic micro-inverter is resolved, the fault mark sure-Er-Sign is replaced with the safety mark safe-Sign.

[0098] The specific fault types of the photovoltaic micro-inverter include open circuit fault, short circuit fault, over-temperature fault, component aging fault and capacitor and inductor fault.

[0099] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A fault detection method for photovoltaic micro-inverters, characterized in that: The following steps are involved: Step S1: collecting battery status information of a plurality of photovoltaic panels constituting a photovoltaic array, dividing the photovoltaic array into different array partitions according to the arrangement positions of the photovoltaic micro-inverters connected to the photovoltaic panels in the photovoltaic array, and monitoring the inverter operating parameters of the photovoltaic micro-inverters corresponding to each array partition; Step S2: combining the battery status information of all photovoltaic panels in each array partition to generate battery pack status information, performing data processing and feature extraction based on the battery pack status information and inverter operating parameters to generate partition feature information for each array partition; Step S3: inputting the partition characteristic information of each array partition into a preset fault screening library, and then marking the partition characteristic information of the array partition suspected of having a fault in the fault screening library; Step S4: Using a machine learning algorithm, pattern recognition is performed on the partition feature information of each array partition suspected of having a fault, thereby locating the photovoltaic micro-inverters in all array partitions that ultimately have a fault, as well as the specific fault types of the photovoltaic micro-inverters; The process of performing data processing and feature extraction based on battery pack status information and inverter operating parameters to generate partition feature information for each array partition includes: By performing data processing on the battery pack status information, the battery pack status information is decomposed into battery pack timing status information corresponding to a plurality of timestamps; performing data processing on the first operating parameter and the second operating parameter included in the inverter operating parameter, the inverter operating parameter is decomposed into inverter timing operating parameters corresponding to a plurality of timestamps; Construct a one-dimensional data analysis timeline. On the one-dimensional data analysis timeline, perform feature extraction on the battery pack timing state information and inverter timing operating parameters corresponding to each timestamp. This allows obtaining battery pack timing features corresponding to the battery pack timing state information of all photovoltaic panels in the array partition corresponding to each timestamp, as well as inverter timing features corresponding to the inverter timing operating parameters of the photovoltaic microinverters in the array partition corresponding to each timestamp. The battery pack timing characteristics and inverter timing characteristics corresponding to each array partition at the same timestamp are integrated to generate the superimposed timing characteristics of the corresponding timestamps. The superimposed timing characteristics of all timestamps of each array partition are summarized to generate the partition characteristic information corresponding to each array partition.

2. A fault detection method for photovoltaic micro-inverters according to claim 1, characterized in that: The process of collecting battery status information of several photovoltaic panels that make up a photovoltaic array and dividing the photovoltaic array into different array partitions according to the layout of photovoltaic micro-inverters connected to the photovoltaic panels in the photovoltaic array includes: The photovoltaic array consists of several photovoltaic panels. Each photovoltaic panel is numbered and denoted as i, where i = 1, 2, 3, ..., n, and n is a natural number greater than 0. Battery status information corresponding to each photovoltaic panel is collected. The battery status information includes battery temperature, operating voltage, operating current, and battery internal resistance. A number of photovoltaic micro-inverters are arranged on the photovoltaic array. Each photovoltaic micro-inverter is used to connect to R photovoltaic panels, where R is an integer from 1 to 3. The photovoltaic micro-inverter is used to convert the direct current of the photovoltaic panel into alternating current. According to the arrangement positions of the photovoltaic micro-inverters connected to the photovoltaic panels in the photovoltaic array, the photovoltaic panels connected to the corresponding photovoltaic micro-inverters are divided into the same array partition on the photovoltaic array, and then a number of array partitions corresponding to the several photovoltaic micro-inverters on the photovoltaic array are divided.

3. A fault detection method for photovoltaic micro-inverters according to claim 2, characterized in that: The process of monitoring the inverter operating parameters of the PV microinverters corresponding to each array partition includes: Different types of sensors and power monitoring equipment are arranged in each array partition. The different types of sensors include temperature sensors, humidity sensors, and air pressure sensors. The inverter operating parameters of the corresponding photovoltaic micro-inverter are obtained through the sensors and power monitoring equipment in each array partition. The inverter operating parameters include a first operating parameter and a second operating parameter; The first operating parameter includes the ambient temperature collected by the temperature sensor, the ambient humidity collected by the humidity sensor, and the ambient atmospheric pressure collected by the pressure sensor; The second operating parameters include the inverter DC input voltage, inverter AC output voltage, inverter DC input current, inverter AC output current, inverter DC input power and inverter AC output power collected by the power monitoring equipment.

4. A fault detection method for photovoltaic micro-inverters according to claim 3, characterized in that: The process of combining the battery status information of all photovoltaic panels in each array partition to generate battery pack status information includes: The battery status information of each photovoltaic battery group is classified according to the timestamp, and then the battery time sequence status information corresponding to each photovoltaic battery group at several timestamps is generated. The battery time sequence status information consists of the battery temperature, operating voltage, operating current and battery internal resistance at the same timestamp; The battery timing status information of all photovoltaic battery groups in the same array partition is combined according to the same timestamp to generate the battery group timing status information corresponding to the photovoltaic arrays at several timestamps. The battery group timing status information corresponding to all timestamps of each photovoltaic array is combined to generate the battery group status information corresponding to each photovoltaic array.

5. A fault detection method for photovoltaic micro-inverters according to claim 4, characterized in that: The process of inputting the partition characteristic information of each array partition into a preset fault screening library, and then using the fault screening library to mark the partition characteristic information of the array partition suspected of having a fault, includes: A pre-set fault screening library is built based on a machine learning model. This model is generated by processing, mining, and feature processing historical data of the PV array, combined with data modeling techniques. The machine learning model is used to analyze the characteristic information of all partitions corresponding to several PV array partitions when suspected faults occurred in the past. The partition characteristic information of each array partition is input into the fault screening library. The fault screening library then marks the partition characteristic information of array partitions suspected of having faults. The partition characteristic information of the array partitions suspected of having faults is associated with a suspected fault identifier, which is recorded as doubt-Er-Sign. The partition characteristic information of other array partitions not suspected of having faults is associated with a safe operation identifier, which is recorded as safe-Sign.

6. A fault detection method for photovoltaic micro-inverters according to claim 5, characterized in that: The process of using a machine learning algorithm to perform pattern recognition on the partition feature information of each array partition suspected of having a fault, and then locating the photovoltaic microinverters in all array partitions with a fault includes: An initial pattern recognition model is constructed using a machine learning algorithm. The pattern recognition model is used to perform pattern recognition on the partition feature information of the array partitions suspected of having a fault. All the partition feature information associated with the suspected fault identifier is used as input parameters of the pattern recognition model. Obtain a historical fault data set, which records several types of fault problems of photovoltaic micro-inverters at historical times. Synchronize the historical fault data set as input parameters of a pattern recognition model. The pattern recognition model outputs corresponding recognition output results based on the received input parameters. The recognition output results include: the device numbers of photovoltaic micro-inverters in all array partitions that ultimately have faults, and the specific fault types corresponding to the photovoltaic micro-inverters; When the photovoltaic micro-inverter under the corresponding array partition is finally located to have a fault, the suspected fault identifier doubt-Er-Sign of the partition characteristic information corresponding to the array partition is replaced with a fault identifier, and the fault identifier is recorded as sure-Er-Sign. Maintenance personnel are arranged to repair all photovoltaic micro-inverters located with specific fault types. When the fault problem corresponding to the specific fault type of the photovoltaic micro-inverter is resolved, the fault identifier sure-Er-Sign is replaced with the safety identifier safe-Sign.

7. A fault detection method for photovoltaic micro-inverters according to claim 6, characterized in that: The specific fault types of the photovoltaic micro-inverter include open circuit fault, short circuit fault, over-temperature fault, component aging fault and capacitor and inductor fault.

Images