WO2018028005A1

WO2018028005A1 - Fault detection algorithm for battery panel in large-scale photovoltaic power station

Info

Publication number: WO2018028005A1
Application number: PCT/CN2016/096744
Authority: WO
Inventors: 徐建荣; 周峰; 徐斐; 解玉凤; 周乐成; 包文中
Original assignee: 苏州瑞得恩自动化设备科技有限公司
Priority date: 2016-08-11
Filing date: 2016-08-25
Publication date: 2018-02-15
Also published as: CN107733357B; CN107733357A

Abstract

A fault detection algorithm for a battery panel in a large-scale photovoltaic power station, wherein a manipulation command is directly sent by an upper computer; an acquired current of each branch and voltage of each panel at a current moment and environmental parameters comprising light intensity and temperature are computed after data is input; wherein by using a horizontal comparison algorithm for locating a single failed panel, a historical comparison algorithm for degradation judgement of the whole plant, and a standard value updating algorithm used as reference, and by using a record table of single faults (112) of an auxiliary algorithm, a record table of historical averages (113) and a look-up table of standard values (111), single fault locating and overall degradation or deposition judgement are conducted at the same time. The fault detection algorithm can realise on-line real-time monitoring, is easy to implement, and can avoid misjudgement due to soft errors. Moreover, the look-up table of standard values (111) is built in an adaptive mode, and can be adapted to different geographical and environmental conditions, thereby having greater universality and higher precision.

Description

Fault detection algorithm for battery panel in large photovoltaic power station

Technical field

The invention belongs to the technical field of photovoltaic power generation, and relates to a fault detection algorithm for a battery panel in a large-scale photovoltaic power plant, and particularly relates to a method for detecting fault location and overall aging or dust accumulation in a large-scale photovoltaic array.

Background technique

With the continuous development of new energy technologies, the application of solar photovoltaic power generation has become more and more extensive. The prior art discloses that a photovoltaic array is a core component of a photovoltaic power generation system, and is composed of several photovoltaic cells panels connected in series and in parallel. The data show that the current common photovoltaic power plants are composed of large-scale photovoltaic arrays placed outdoors. Due to environmental factors, photovoltaic panels are likely to have individual failures, aging or ash accumulation. Due to the large number of photovoltaic panels, it is often impossible to detect them in time. The failure cannot be replaced in time, which seriously affects the power generation efficiency of the photovoltaic power station.

Based on the above-mentioned status quo, the technicians in the industry use some fault detection methods of photovoltaic panels to detect the voltage and current parameters of the photovoltaic panel, but the practice shows that the data of the panel parameters are collected, and the conclusion of the fault condition cannot be directly obtained. It is also necessary to conduct reasonable and effective diagnosis and analysis of the data in order to obtain accurate fault location and fault types. The prior art photovoltaic power generation system fault detection algorithm still has the following limitations, such as: one of the categories of technology, through the method of current difference judgment (literature <solar battery array fault diagnosis method research>, the fourteenth National Conference on Equipment Monitoring and Diagnostics, 2010), by sifting adjacent columns with excessive current difference to find the branch with faulty components, and analyzing the voltage values of the PV modules of the branch, positioning Faulty photovoltaic cell assembly; the algorithm can be realized by real-time measurement, and only needs to measure voltage and current information, the calculation amount is small, and easy to implement, but it still has problems: it can not effectively judge soft errors, for temporary occurrence The occlusion and other situations are prone to misjudgment, and only the lateral comparison between the branches is carried out. The lack of analysis of the overall working conditions of the power station combined with environmental parameters does not reflect the overall aging or ash accumulation of the power station; , using known data for training, obtaining a complete database, and then comparing the measured data with the algorithm, Chinese patent No. 201410449777.6) discloses a method for fault diagnosis using fuzzy clustering algorithm, which is characterized by first establishing a fault knowledge base through a large amount of work in the early stage, and collecting the fault alarm information of the photovoltaic power station as a sample to be compared, by comparing the degree of membership, and selecting The fault type with the highest degree of membership is the fault type represented by the sample to be tested. The problem with the algorithm is that it is necessary to determine in advance that a certain group of data is fault data, and then the membership degree algorithm can be used. The electrical parameters such as current and voltage that are easy to test directly determine the fault; and the algorithm requires a large amount of training data in advance to establish a fault knowledge base for comparison. Due to the variety of faults, the fault simulation is complicated and consumes a lot of resources; Moreover, since the collection of the fault knowledge base is limited, and the update of the fault knowledge base takes a long time, it is difficult to adapt to various geographical environments. When the geographical environment changes, the content of the fault knowledge base will face inaccuracy. In the category technology, the neural network model is used to judge the fault. The electrical parameters such as maximum power point current, maximum power point voltage, short circuit current, and open circuit voltage are used as input parameters of the algorithm, and the fault is obtained through the BP neural network. Model output to determine the type of fault (Document "A four-parameter method for online fault diagnosis of photovoltaic modules", Chinese Journal of Electrical Engineering, May 5, 2014, Vol. 34, No. 13), (Document "Based on BP Neural Network Research on Fault Diagnosis of Photovoltaic Arrays, Power System Protection and Control, August 2013 On the 16th, Vol. 41, No. 16), the method can directly determine the type of fault through electrical parameters, but the problem is that the algorithm itself is complex, computationally intensive, and also requires a large amount of fault data for training; The input short-circuit current and open-circuit voltage cannot be measured in real time, and it is necessary to temporarily stop the PV array, affect the power generation efficiency of the power station, and so on.

In view of the limitations of the prior art, the inventors of the present application intend to provide a more universally ubiquitous online real-time monitoring, easy to implement, capable of simultaneous individual fault location and overall aging judgment, and adapt to different geographical and environmental conditions. Saturation and higher accuracy of PV array fault detection algorithms.

Summary of the invention

The object of the present invention is to provide a fault detection algorithm for a battery panel in a large-scale photovoltaic power plant according to the defects of the prior art, and particularly to a method for detecting fault location and overall aging or dust accumulation in a large-scale photovoltaic array. . The method of the invention can realize on-line real-time monitoring, easy implementation, combined with environmental information and electrical parameters, and simultaneously perform individual fault location and overall aging judgment of photovoltaic array fault detection.

Specifically, the fault detection algorithm for a battery panel in a large-scale photovoltaic power plant of the present invention is characterized in that it comprises: directly using a host computer to issue a control command, and the current of each branch at the time acquired by the acquisition device and each block The voltage of the electric board and the environmental parameters (including the light intensity and temperature) obtained by the environmental monitoring equipment are calculated as input data to realize the fault detection of the photovoltaic array at the same time for individual fault location and overall aging judgment; wherein: a single fault panel is adopted The horizontal comparison algorithm for positioning, the historical comparison algorithm for the overall aging judgment of the power station, and the standard value update algorithm used as a comparison, the individual fault record table using the auxiliary algorithm, the historical mean record table, and the standard value lookup table.

In the present invention, the photovoltaic array is a string composed of a plurality of photovoltaic panels, and is additionally equipped with an environmental parameter monitoring device capable of measuring light intensity and temperature;

In the present invention, the upper computer usually adopts a computer that can directly issue a manipulation command;

Wherein, the related table of the detection algorithm is stored in the upper computer:

1) A standard value lookup table that can be changed once: used to compare with the measured data to determine the overall aging or ash accumulation;

The standard value lookup table includes (a) a light intensity-current standard value lookup table for recording corresponding current standard values in various light intensity conditions, (b) a temperature-voltage standard value lookup table, and recording various temperature conditions. The corresponding voltage standard value; wherein, the light intensity-current standard value lookup table and the temperature-voltage standard value lookup table, the initial set standard value is saved in the initial state, wherein each set of data pairs further corresponds to an update flag bit, It is used to obtain a standard current value under the premise of known light intensity value, as the average current of each branch when the power station is normally generating power, compare with the average value of the current measurement, and update the flag to indicate whether the corresponding data pair is In the initial state, when a certain group of data pairs is in an initial state, it can be updated, and after being updated once, its corresponding flag bit is marked, so that the standard value is determined;

2) Individual fault record table: used to record the number of abnormal current or voltage occurrences of a certain board to determine whether a board has a fault, including:

(a) The location of the faulty board, which can be indicated by a number;

(b) The number of fault indications used to record the number of abnormalities in the current or voltage of the board,

(c) the current and voltage values of the board when the fault occurs;

(d) The last update time of the fault indication, indicated by the test sequence number, used to screen out soft errors that may occur;

In an embodiment of the invention, the individual fault record table includes:

The position of the abnormal battery panel can be represented by a one-dimensional number or by two-dimensional coordinates.

The fault flag of the abnormal battery panel records the number of times the battery panel is detected to be abnormal, and is used to determine whether the panel is faulty;

The current and voltage abnormal values are the current and voltage values when the board is detected abnormally, and may include multiple sets of data; the last update time is the corresponding test sequence number when the board is last detected abnormally. , the fault flag clearing process for eliminating soft errors;

In the embodiment of the present invention, the fault flag clearing process may periodically screen the fault flag and the last update time of each group of fault records in the individual fault record table, by comparing the last time and the current time. Automatically delete the fault record whose fault flag has not been added for a long time;

3) Historical mean record table: record the average current and voltage of the whole power station for each measurement, the light intensity during the test And temperature, used to analyze the overall trend of the power generation situation of the power station, to determine whether there is aging and ash, etc., including:

(a) The mean value of the branch current is the average current obtained for the branch current of each faultless string of the entire power station;

(b) The average value of the full board voltage is the average voltage obtained for all the fault-free boards of the entire power station;

(c) test serial number to indicate the time of each test;

In an embodiment of the invention, the historical mean record table includes:

The overall average value of the current, that is, the average value of each branch current obtained per measurement;

The average value of the voltage, that is, the average value of each panel voltage obtained for each measurement;

The light intensity and temperature value, that is, the light intensity value and temperature value recorded for each measurement, as environmental parameters, the corresponding voltage and current standard values can be found in the standard value lookup table;

The test sequence number is used to indicate the time sequence of each test. Among them, each test is automatically added with 1 and a positive integer is used to reflect the sequence relationship of each test result.

In the detection algorithm of the present invention, the main part includes a horizontal comparison algorithm, a standard value update algorithm, and a historical comparison algorithm (the overall aging judgment);

Wherein, the horizontal comparison algorithm compares the currents of the respective series branches, and further compares the voltage values of the abnormal current values to the voltage values, and saves the positions of the photovoltaic components with abnormal voltage values in the individual fault record tables, if If there is a fault record in the location, the fault flag of the board is increased. When the fault flag reaches a certain number, the battery panel may be displayed to the external interface to be faulty and report its position, and the current voltage fault data is displayed. reference;

In the horizontal comparison algorithm, the individual fault record table automatically deletes the board information of the fault flag that has not changed for a long time according to the last update time of the stored element, and prevents misjudgment caused by soft errors such as temporary occlusion;

In an embodiment of the invention, the lateral comparison algorithm for a single fault panel bit includes:

An algorithm input parameter composed of each branch current at the measurement time and each panel voltage of the battery;

The transverse current comparison step compares the difference between the current of each branch and the average value;

a previously set current proportional coefficient, when the ratio of the current difference of a branch to the average value is greater than the proportional coefficient, the branch is considered to be abnormal;

The lateral voltage comparison step compares the difference between the panel voltage and the average value for the abnormal branch;

A previously set voltage scaling factor, when the ratio of the voltage difference between the panel and the average is greater than the proportionality factor, the panel is considered to be abnormal;

In association with an individual fault record table, when a panel abnormality is detected, the individual fault record table needs to be modified accordingly to save the fault record of the board;

A fault flag threshold set in advance. When a fault flag corresponding to a battery panel exceeds the threshold, the panel is considered to be faulty, and the fault information of the panel is outputted outward.

The historical comparison algorithm (the overall aging judgment) excludes the existing battery panel in the individual fault record table for each measurement, obtains the average voltage value for the normal working battery panel, and obtains the current average value for the normal working branch. And stored in the historical mean value record table together with the environmental parameters; when the stored data reaches a certain number, enter the overall aging judgment, and find the corresponding measurement records in the current-light intensity standard value lookup table and the voltage-temperature standard value lookup table. The current and voltage standard values, and calculate the difference between the measured mean and the standard value; fit the time-dependent curve to the calculated difference, and judge whether it is aged by analyzing the change trend; in the embodiment of the present invention, history Comparison algorithms include:

Taking the environmental parameters composed of the light intensity and temperature value at the time of measurement, and the electrical parameters of each branch current and each panel voltage as the input parameters of the algorithm;

The process of association with individual fault record tables, by retrieving the fault records recorded in the individual fault record tables, and eliminating abnormal current and voltage values in the input electrical parameters to increase the accuracy of the algorithm;

The current and voltage are processed as a whole, and the current values of all the branches are combined with the voltage values of all the panels, and the average values of the current and the voltage are used as the overall result of the current measurement;

The process of association with the historical mean record table, the overall result obtained by each measurement is saved in the historical mean record table;

Correlation process with the standard value lookup table, using the light intensity and temperature values as retrieval conditions, obtaining current and voltage standard values in the standard value lookup table, and comparing with the measured values;

The curve fitting process uses the test sequence number as the abscissa, the difference between the current (voltage) and the standard value as the ordinate to fit the curve, indicating the overall aging (or ash) trend.

The standard value update algorithm will be executed after the amount of data in the historical mean record table reaches a certain scale; if the number of occurrences of the same light intensity value reaches the set threshold in the test, the corresponding in the light intensity-current standard value lookup table is found. The point of the light intensity value, if the update flag indicates that the point data has not been updated, the current measurement value corresponding to the light intensity value is averaged, the current value of the point in the original table is replaced, and then the flag bit is modified to indicate the point. The data is not allowed to be updated again. In an embodiment of the present invention, an algorithm for adaptively updating current and voltage standard values during a test according to an environment in which the power station is located and a device used, including:

One-time update feature, that is, each set of standard values can only be updated once after the algorithm is run, and then no longer changes;

The process of association with the historical mean record table, when the data in the table reaches a certain scale, the adaptive update process is started;

a preset current update threshold value, when the number of records having the same light intensity value in the historical mean value record table is greater than the threshold value, updating the standard value of the group;

a preset voltage update threshold, when the number of records having the same temperature value in the historical mean record table is greater than the threshold, updating the set of standard values;

In the process of association with the standard value lookup table, all standard values are stored in the standard value lookup table, and the standard value update algorithm will directly update the data of the table.

In the present invention, the update process of the temperature-voltage standard lookup table is similar to the update process of the light intensity-current standard value lookup table, that is, when the same temperature value appears a certain number of times, the point corresponding to the changed temperature value is found in the lookup table. If the update flag indicates that the point data has not been updated, the voltage value corresponding to the temperature value is averaged, and the voltage value of the point in the original table is replaced, and then the point data is not changed.

In the present invention, the standard value update algorithm makes the two standard value lookup tables more adaptive, does not use fixed data, and can self-correct while detecting and improving the overall operation of the power station for different environments and devices. The accuracy of the state analysis.

The output of the photovoltaic array fault detection algorithm of the present invention includes:

The location of the failed photovoltaic panel and the corresponding current and voltage conditions when the fault occurs;

The result of the overall aging or ash deposition state analysis of the photovoltaic power plant, including the recent work efficiency curve, the overall current and voltage, and the corresponding environmental parameters;

The output will be provided to the display terminal, which includes information of individual faulty unit information, overall aging condition, and power generation curve. At the same time, the photovoltaic array fault detection algorithm of the present invention can accurately judge soft errors and reduce false alarms.

The invention performs actual detection, and the result shows that the photovoltaic array fault detection algorithm of the invention can be monitored online in real time, is easy to implement, can perform individual fault location and overall aging/ashing judgment at the same time, and can avoid misjudgment caused by soft errors. And the standard value lookup table is constructed in an adaptive manner, so that the algorithm can adapt to different geographical and environmental conditions, and thus has stronger universality and higher precision.

For ease of understanding, a large-scale photovoltaic array failure of the present invention will be hereinafter described through specific drawings and embodiments. The detection method is described in detail. It is to be understood that the specific embodiments and the drawings are only for the purpose of illustration and description Variations are also included within the scope of the invention. In addition, the present invention is hereby incorporated by reference in its entirety in its entirety in its entirety in its entirety in its entirety in its entirety in the the the the the the the

DRAWINGS

Figure 1. Overall data structure diagram of the photovoltaic fault detection algorithm of the present invention.

Figure 2. Flow chart of a single panel fault detection algorithm in the photovoltaic fault detection algorithm of the present invention.

Figure 3. Flowchart of an algorithm for eliminating soft errors using flag bit clearing in a single panel fault detection algorithm of the present invention.

Figure 4. Flow chart of the algorithm for judging the overall aging condition in the photovoltaic fault detection algorithm of the present invention.

Figure 5. Flowchart of adaptive update of the standard data lookup table in the photovoltaic fault detection algorithm of the present invention.

Specific implementation method

Example 1

1 is a general data structure diagram of a photovoltaic fault detection algorithm proposed by the present invention, which includes a detection start signal output by the system, and the signal is automatically sent to the data acquisition device according to a preset test time, prompting the collection device to start collecting and measuring the measured signal. Current, voltage, light intensity and temperature information are sent back to the processing system;

It also includes input data 101 sent back by the acquisition device, consisting of the current value of the serial input, the voltage value, and the illumination intensity and temperature at the test time, since the required electrical parameters include the operating current of all serial branches and all photovoltaics. The working voltage of the electric board has a large amount of data, so the serial input mode is adopted, and the input data will first enter the decision system;

It also includes three data tables, which are composed of an individual fault record table 112, a historical mean value record table 113, and a standard value lookup table 111, which are permanently stored in the processing system and used as criteria for the comparison algorithm, wherein the standard value data is used. The table also includes a light intensity-current standard lookup table 114 and a temperature-voltage standard lookup table 115. The two tables are stored in a fixed size, and the light intensity and temperature values are not changed after the setting, and the fault flag record table is a chain structure. Each generation of fault data will be added directly after the original table to save storage space. The historical average record table can hold the fixed amount of data. When the number of test data sets exceeds the capacity of the record table, the earliest measured data is deleted, and the new data is filled in. Enter the corresponding position to basically guarantee the continuity of the test serial number.

Also included is an algorithm body 102 comprised of a horizontal comparison algorithm, a historical comparison algorithm, and a standard value update algorithm. The horizontal comparison algorithm does not need environmental parameter information, only need to compare the currents of each branch, find the branch with abnormal current value, and then compare the voltage values of the electric plates in the branch to find the electric board with abnormal voltage value. And the fault condition is added to the individual fault record table 112. The historical comparison algorithm averages all the branch currents obtained for each measurement, averages all the voltages of the boards, and records the light intensity and temperature during the test. As a set of test data, the test sequence number is added and stored in the historical mean record table 113. When the data amount in the table reaches a certain scale, the historical comparison algorithm extracts the data in the table, and finds the same light intensity in the standard value data table 111. Corresponding current standard value and voltage standard value corresponding to the same temperature, after performing the difference calculation, analyzing the variation law of each group of data according to the chronological order, judging whether there is a problem such as overall aging or ash accumulation,

The utility model further includes output data 103 provided to the external display, which is composed of individual fault unit information, overall aging condition, power generation amount and corresponding environment parameter information, wherein the individual fault unit information includes the position of the board where the fault may occur and the current voltage as the basis for the determination. , generated by a horizontal comparison algorithm. The overall aging situation provides an analysis of whether aging or dust accumulation, and the power generation curve is used as a basis for judgment, and is generated by a historical comparison algorithm.

Example 2

2 is a flow chart of single panel failure detection in a single measurement of the present invention.

First, the algorithm needs to obtain the current value of all test branches and the voltage value of each board (201);

Calculate the average value of all branch currents, and then compare the current value of each branch with the average value to obtain the difference between each branch current and the average current (202);

Using the difference as a criterion for the fault branch, if a branch current exceeds a certain percentage of the average current (the ratio is set in advance), that is, the difference is too large, it is considered that there may be a failure of the board in the branch. And reading the voltage of the electric board to the next step (203), otherwise returning to (202), calculating the next branch;

For a suspicious branch, calculate the average value of the voltage of each board on the branch, and then compare the voltage value of each board with the average value to obtain the difference between each board voltage and the average voltage (205);

Using the difference as a criterion for the faulty board, if a branch voltage exceeds a certain percentage of the average voltage (the ratio is set in advance), that is, the difference is too large, the board may be considered to be malfunctioning (206). Otherwise, return to (204) and calculate the next board;

Searching for the board information in the individual fault record table 112, that is, determining whether the board has been entered into the individual fault record table: if the board has been marked, add 1 to the fault flag, and record the time of the measurement. Board current and voltage The value is used as the fault criterion, and the test sequence number of the current measurement is used as the update time; if not, the board information is added to the individual fault record table 112, including its position information, current and voltage during the current measurement. Value, and update time (test number), and make the fault flag equal to 1 (205);

If the faulty flag of the board is greater than the set threshold, the board is determined to be a faulty board, and the fault information of the board is output to the display end, including the position of the board and each time. The current and voltage values at the time of failure prompt the staff to perform manual inspection (206) in time, otherwise return to (204) and calculate the next electric board;

After all the branches have passed the detection process of (202) to (206), a horizontal comparison (207) is ended.

Example 3

FIG. 3 is a flow chart of manually removing a faulty power board, or deleting a certain power board from the individual fault record table 112 for soft errors, and (301) to (306) clearing the fault flag after replacing the power board, ( 311) to (314) are cleared after the soft error flag;

After replacing the board, you need to input the position information of the replaced board (301);

In order to improve efficiency and avoid additional test steps, it will wait for the next automatic detection of the system to start, and simultaneously detect the operation of the replaced board in the normal detection process (302);

After the detection algorithm starts, it will enter the single panel fault detection described in Embodiment 2, and after the detection, the individual fault record table 112 (303) will be updated according to the detection result;

After the horizontal comparison algorithm ends, the fault data corresponding to the replaced board position is searched in the individual fault record table (304);

At this time, if the fault flag corresponding to the board is still the same as before the board is replaced, it means that the current and voltage at the position are no longer abnormal after the board is replaced, and the board is considered to be successfully updated, and the position in the individual fault record table 112 is Corresponding fault data clearing (305);

If the fault flag corresponding to the board continues to increase, it indicates that there is still a fault at that place, and further confirmation and troubleshooting are required, the system will prompt through the output that there is still a fault at the position (306);

The interval at which the periodic fault flag is cleared before the system works normally. When the test sequence number is an integer multiple of the value, after the normal detection algorithm ends, the system will additionally enter the flag clearing process (311);

For each set of fault data in the individual fault record table 112, calculate the difference between the last update time (the test number at the time of the last update) and the test number of the current test (312);

If the sequence number difference is greater than the threshold value of the preset difference value, it indicates that the fault identification of the fault data of the group has not increased for a long time, and may be only a soft error caused by partial occlusion, shadow, etc., and not the failure of the power board, indicating the board. The fault data is cleared from the individual fault record table 112 (314).

Example 4

Figure 4 is a flow chart of the overall aging judgment, that is, the history comparison algorithm.

The algorithm includes a pre-processing part of the data obtained from the acquisition device. Since the light intensity and the temperature value in the standard value lookup table 111 are discrete, the light intensity value and the temperature value obtained from the environmental monitoring device need to be approximated to make the environment The parameters are matched with the data stored in the table (401);

Since the overall aging judgment is based on the current and voltage parameters of all the power boards of the power station, the electrical boards of individual faults may have a great influence on the overall judgment result due to the large difference between their electrical parameters and the normal working state. Therefore, according to the individual fault records first, The data stored in Table 112 obtains the electrical boards of possible faults found in several measurements, and excludes them from the input data of the historical comparison algorithm. For other normal working branches, the average current value is obtained. All normal working boards are averaged (402);

The average current and the average voltage obtained by the test, and the pre-processed light intensity and temperature values are taken as a set of data, and after adding the current test number, it is stored in the historical mean value record table 113 (403);

Since it is necessary to analyze the change of the overall working condition of the power station, a certain amount of data is required to perform the next processing, so it is determined whether the valid data existing in the historical mean record table is greater than a preset value (404);

If the amount of data is sufficient for the historical comparison algorithm, for each set of data in the historical mean record table 113, the current standard value corresponding to the same light intensity value is searched in the light intensity-current standard lookup table, and the temperature-voltage standard lookup table is used. Search for the voltage standard value corresponding to the same temperature (405);

The current average value of the data in the historical mean value record table 113 is made to be different from the searched current standard value to obtain a current difference value, and the voltage average value is compared with the searched voltage standard value to obtain a voltage difference value as an overall aging condition. a set of criteria (406);

After a set of current and voltage differences are obtained for each set of data in the historical mean value record table 113, the test sequence number (representing time) is plotted on the abscissa, and the current difference-time and voltage difference-time curve are respectively fitted (407 );

The above curve reflects the change of the current and voltage deviation from the standard value with time. Through the curve, the overall aging condition of the power station can be analyzed. For example, if the curve is found to be in a positive upward trend, it is obvious that the overall working efficiency of the power station is reduced, and overall aging may occur. Ash accumulation requires timely inspection and cleaning. The algorithm finally combines the fitted curves to analyze The possible status is output to an external display so that the manager can more intuitively understand the operation of the plant (408).

Example 5

FIG. 5 shows an adaptive update flow of the standard value lookup table 111. In order to enable the overall aging determination described in Embodiment 4 to be implemented as soon as possible, the table stores initial data before the system is run, and the historical data comparison algorithm can be directly used using the initial data table. Since the initial data is based on prior knowledge and actual experience, in actual application, due to different working conditions of the power station, the initial standard value may be slightly different from the actual standard value. In order to make the overall aging judgment more accurate, adaptive The standard value is continuously improved by the standard value lookup table 111.

The standard table achieves its adaptability by providing an update identifier for each set of data.

First, wait for the data in the historical mean value record table 113 to reach a certain scale (500);

The update process for the light intensity-current standard lookup table is as shown in (501) to (506):

Sorting the data of each group in the mean value record table according to the light intensity value, the purpose is to store the data of the same light intensity value together (501);

Sequence traversing the collated data, if it is found that the data having a certain light intensity value reaches a certain number (502), the point corresponding to the light intensity value is found in the light intensity-current standard lookup table (503);

Read the update identifier of the location (the update identifier is 1 when the system first starts running). If the identifier is 1, it indicates that the point data is still the initial data and can be updated (504);

Extracting all the current values corresponding to the light intensity values in the sorted historical mean data, and determining the mean value thereof, and replacing the original current value in the standard lookup table with the mean value (505);

The update identifier of the point modified in the light intensity-current standard lookup table is 0, indicating that the point data has been adaptively processed and cannot be changed (506).

The update process for the temperature-voltage standard lookup table is similar to the above process, as shown in (511) to (516):

Sorting the data of each group in the mean value record table by temperature value, the purpose is to store the data of the same temperature value together (511);

Sequence traversing the collated data, if it is found that the data having a certain temperature value reaches a certain number (512), the point corresponding to the light intensity value is found in the temperature-voltage standard lookup table (513);

Reading the update identifier of the location, if the identifier is 1, it indicates that the point data is still the initial data, and can be updated (514);

Extracting all voltage values corresponding to the temperature value in the sorted historical mean data, and determining the mean value thereof, The mean replaces the original voltage value in the standard value lookup table 111 (505);

The update identifier of the point modified in the temperature-voltage standard lookup table is 0, indicating that the point data has been adaptively processed and cannot be changed (506).

It should be noted that the processes of updating the light intensity-current standard lookup table (501) to (506) in the above embodiment may be interchanged with the processes of the temperature-voltage standard lookup table updating processes (511) to (516), and Does not affect the algorithm results.

In addition, this embodiment is directed to the update process in the case where the initial data exists in the standard value lookup table 111. If the lookup table does not carry the initial data, the adaptive update of the standard value may be performed according to the flow, but the embodiment is delayed. 4 The execution of the overall aging judgment.

Claims

A fault detection algorithm for a battery panel in a large-scale photovoltaic power plant, wherein the method includes: directly using a host computer to issue a control command, collecting current of each branch at each moment of the acquisition device, and voltage of each of the boards, and environmental monitoring The environmental parameters obtained by the equipment, including the light intensity and temperature, are calculated as input data to realize the fault detection of the photovoltaic arrays with simultaneous individual fault location and overall aging judgment;

A horizontal comparison algorithm for positioning a single fault panel, a historical comparison algorithm for the overall aging judgment of the power station, and a standard value update algorithm for comparison, an individual fault record table using the auxiliary algorithm, a historical mean record table, and a standard value lookup table .
The fault detection algorithm for a battery panel in a large-scale photovoltaic power plant according to claim 1, wherein the horizontal comparison algorithm of the single fault panel position comprises:

An algorithm input parameter composed of each branch current at the measurement time and each panel voltage of the battery;

The transverse current comparison step compares the difference between the current of each branch and the average value;

a previously set current proportional coefficient, when the ratio of the current difference of a branch to the average value is greater than the proportional coefficient, the branch is considered to be abnormal;

The lateral voltage comparison step compares the difference between the panel voltage and the average value for the abnormal branch;

A previously set voltage scaling factor, when the ratio of the voltage difference between the panel and the average is greater than the proportionality factor, the panel is considered to be abnormal;

In association with an individual fault record table, when a panel abnormality is detected, the individual fault record table needs to be modified accordingly to save the fault record of the board;

A fault flag threshold set in advance. When a fault flag corresponding to a battery panel exceeds the threshold, the panel is considered to be faulty, and the fault information of the panel is outputted outward.
The fault detection algorithm for a battery panel in a large-scale photovoltaic power plant according to claim 1 or 2, wherein the individual fault record table comprises:

The position of the abnormal battery panel, expressed in one-dimensional number, or in two-dimensional coordinates;

The fault flag of the abnormal battery panel records the number of times the battery panel is detected to be abnormal, and is used to determine whether the panel is faulty;

The current and voltage abnormal values are current and voltage values when the board is detected abnormally, or include multiple sets of data;

The last update time is the test sequence number corresponding to the last time the board is detected abnormal, and is used to eliminate the soft error fault flag clearing process.
The fault detection algorithm for a battery panel in a large-scale photovoltaic power plant according to claim 3, wherein said fault flag clearing process periodically screens a fault flag and a last update time of each group of fault records in the individual fault record table. By comparing the last time and the current time, the fault record whose fault flag has not been added for a long time is automatically deleted.
The fault detection algorithm for a battery panel in a large-scale photovoltaic power plant according to claim 1, wherein the historical comparison algorithm comprises:

Taking the environmental parameters composed of the light intensity and temperature value at the time of measurement, and the electrical parameters of each branch current and each panel voltage as the input parameters of the algorithm;

The process of association with individual fault record tables, by retrieving the fault records recorded in the individual fault record tables, and eliminating abnormal current and voltage values in the input electrical parameters to increase the accuracy of the algorithm;

The current and voltage are processed as a whole, and the current values of all the branches are combined with the voltage values of all the panels, and the average values of the current and the voltage are used as the overall result of the current measurement;

The process of association with the historical mean record table, the overall result obtained by each measurement is saved in the historical mean record table;

Correlation process with the standard value lookup table, using the light intensity and temperature values as retrieval conditions, obtaining current and voltage standard values in the standard value lookup table, and comparing with the measured values;

The curve fitting process uses the test sequence number as the abscissa, the difference between the current (voltage) and the standard value as the ordinate to fit the curve, indicating the overall aging or dust accumulation trend.
The fault detection algorithm for a battery panel in a large-scale photovoltaic power plant according to claim 1 or 5, wherein

The historical mean record table includes:

The overall average value of the current, that is, the average value of each branch current obtained per measurement;

The average value of the voltage, that is, the average value of each panel voltage obtained for each measurement;

The light intensity and temperature value, that is, the light intensity value and temperature value recorded for each measurement, as environmental parameters, the corresponding voltage and current standard values can be found in the standard value lookup table;

Test sequence number, used to indicate the time sequence of each test.
The fault detection algorithm for a battery panel in a large-scale photovoltaic power plant according to claim 3 or 6, wherein the test serial number is used to indicate time, and is automatically added to each test, and each test result is reflected by a positive integer. The relationship between the two.
The fault detection algorithm for a battery panel in a large-scale photovoltaic power plant according to claim 1, wherein the standard value update algorithm adaptively updates current and voltage during the test according to the environment in which the power station is located and the equipment used. The standard value algorithm includes: a one-time update feature, that is, each set of standard values can only be updated once after the algorithm is run, and then does not change;

The process of association with the historical mean record table, when the data in the table reaches a certain scale, the adaptive update process is started;

a preset current update threshold value, when the number of records having the same light intensity value in the historical mean value record table is greater than the threshold value, updating the standard value of the group;

a preset voltage update threshold, when the number of records having the same temperature value in the historical mean record table is greater than the threshold, updating the set of standard values;

In the process of association with the standard value lookup table, all standard values are stored in the standard value lookup table, and the standard value update algorithm will directly update the data of the table.
The fault detection algorithm for a battery panel in a large-scale photovoltaic power plant according to claim 1 or 5 or 8, wherein said standard value lookup table comprises a light intensity-current standard value lookup table and a temperature-voltage standard value lookup table.
The fault detection algorithm for a battery panel in a large-scale photovoltaic power plant according to claim 9, wherein the light intensity-current standard value lookup table includes a light intensity value, a current standard value, and an update identifier for using the known light intensity Under the premise of the value, a standard current value is obtained as the average current of each branch during normal power generation of the power station, and is compared with the average value of the current measurement.
The fault detection algorithm for a battery panel in a large-scale photovoltaic power plant according to claim 9, wherein said temperature-voltage standard value lookup table includes a temperature value, a voltage standard value, and an update identifier for use in a premise of a known temperature value. Next, obtain a standard voltage value as the average voltage of each battery panel when the power station is normally generating power, and compare it with the average value of the voltage measurement.
The fault detection algorithm for a battery panel in a large-scale photovoltaic power plant according to claim 10 or 11, wherein the update identifier is used to indicate whether a certain group of data in the standard value lookup table has been updated, represented by 0 or 1.
The fault detection algorithm for a battery panel in a large-scale photovoltaic power plant according to claim 9, wherein the standard value lookup table may first set an initial standard value or an empty table before the algorithm starts.
The fault detection algorithm for a battery panel in a large-scale photovoltaic power plant according to claim 1, wherein the individual fault unit information, the overall aging condition, and the power generation amount change curve information provide an output to the display terminal.