A method and system for identifying power generation anomalies in photovoltaic power plants
Through the analysis of K-mean clustering algorithm and historical equivalent full-power generation time data, the abnormality of photovoltaic power station power generation is automatically identified, which solves the problem of low manual detection efficiency and achieves rapid and accurate fault detection and maintenance efficiency improvement.
Patent Information
- Application Number
- CN202010515789.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2020-06-09
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2040-06-09
AI Technical Summary
In the prior art, the working status detection of the power generation unit of photovoltaic power stations relies on manual analysis, which consumes a long time, is inefficient and is prone to missed detection and misjudgment, and lacks automated identification methods.
Through the K-mean clustering algorithm based on historical operation data, the historical equivalent full-power generation time data of the power generation unit in the photovoltaic power station are extracted, and the power generation abnormal units are automatically identified, and the preset K value and threshold value are used for classification to reduce manual intervention. .
It realizes the automatic identification of abnormal power generation in photovoltaic power stations, shortens the troubleshooting time, improves power generation and operation and maintenance efficiency, and reduces labor intensity.
Abstract
Description
Technical Field
[0001] The invention relates to a method and a system for identifying power generation anomalies in a photovoltaic power station, and belongs to the technical field of photovoltaic power generation. Background Art
[0002] In order to effectively use solar energy resources, solar power generation systems equipped with solar photovoltaic panels have been widely used in social production and life. The design life of a solar power generation system is as long as 25 years, so the operation and maintenance of the equipment is extremely important. At present, a specialized industry for the operation and maintenance of photovoltaic power stations has gradually developed. Large-scale photovoltaic power stations include a large number of photovoltaic power generation units composed of inverters and components. When people operate and maintain photovoltaic power stations, they need to quickly find equipment with poor power generation performance from a large number of photovoltaic power generation units, so as to improve the effectiveness of the operation and maintenance of photovoltaic power stations. At present, the conventional practice is that the on-duty personnel analyze the output curve of the photovoltaic power generation unit and then judge the fault. This manual detection method is time-consuming, inefficient, and prone to missed detection and misjudgment. Therefore, it is urgent to provide a method for automatically identifying abnormal power generation in photovoltaic power stations, so as to timely and accurately find power generation units with poor power generation. Summary of the invention
[0003] In view of the deficiencies in the prior art, the purpose of the present invention is to provide a method and system for identifying power generation anomalies in photovoltaic power stations, so as to solve the technical problems in the prior art that manual detection and judgment of the working status of power generation units is time-consuming, inefficient, and prone to missed detections and misjudgments.
[0004] In order to solve the above technical problems, the technical solution adopted by the present invention is:
[0005] A method for identifying power generation anomalies in a photovoltaic power station comprises the following steps:
[0006] Extracting a sample data set based on pre-acquired historical operation data of the power generation unit, wherein the samples in the sample data set include historical equivalent full-power power generation hours data of the power generation unit in the photovoltaic power station when the device status is normal;
[0007] Based on the historical equivalent full-power power generation hours data, each power generation unit in the photovoltaic power station is classified, and the power generation units with abnormal power generation in the photovoltaic power station are extracted.
[0008] Furthermore, the method for acquiring historical operation data of a power generation unit includes:
[0009] Collecting the operating data of the power generation unit according to a preset collection cycle, wherein the operating data includes the power generation amount and the equipment status of the power generation unit;
[0010] The historical operation data of the power generation unit is generated based on the operation data of no less than two consecutive collection cycles.
[0011] Furthermore, after collecting the operating data of the power generation unit, it also includes: writing the collected operating data into a pre-established operating database, the operating database includes a cumulative power generation array and a device status array, and its record structure is as follows:
[0012] E={t,E1,E2,…EN},
[0013] S = {t, S1, S2, ... SN};
[0014] Where, E is the cumulative power generation array, S is the device status array, t is the acquisition time, N is the number of power generation units in the PV power station, E1 is the daily cumulative power generation of the power generation unit No. 1 at the corresponding time, E2 is the daily cumulative power generation of the power generation unit No. 2 at the corresponding time, EN is the daily cumulative power generation of the power generation unit No. N at the corresponding time, S1 is the device status of the power generation unit No. 1 at the corresponding time, S2 is the device status of the power generation unit No. 2 at the corresponding time, and SN is the device status of the power generation unit No. N at the corresponding time.
[0015] Furthermore, the record structure of the historical equivalent full-power power generation hours data is as follows:
[0016] Ei={[Ei(t1)-Ei(t0)] / Pi,[Ei(t2)-Ei(t1)] / Pi,[Ei(t3)-Ei(t2)] / Pi,…[Ei(t m )-Ei(t m-1 )] / Pi},
[0017] i=1,2,…N;
[0018] Where Ei is the historical equivalent full-power power generation hours data of the power generation unit numbered i# in the sample data set when its equipment is in normal condition, t0, t1, t2, t3, …, t m-1 ,t m are the m+1 measurement moments arranged in sequence, Ei(t0), Ei(t1), Ei(t2), Ei(t3), Ei(t m-1 ),…,Ei(t m ) is the daily cumulative power generation of the power generation unit numbered i# at the corresponding measurement time when its equipment status is normal, Pi is the power generation capacity of the power generation unit numbered i#, [Ei(t1)-Ei(t0)] / Pi is the equivalent full-power power generation hours of the power generation unit numbered i# during t0~t1, [Ei(t2)-Ei(t1)] / Pi is the equivalent full-power power generation hours of the power generation unit numbered i# during t1~t2, [Ei(t3)-Ei(t2)] / Pi is the equivalent full-power power generation hours of the power generation unit numbered i# during t2~t3, [Ei(tm )-Ei(t m-1 )] / Pi is the number i# of the generating unit at t m-1 ~t m The equivalent full power generation hours during the period.
[0019] Furthermore, the method for classifying each power generation unit in the photovoltaic power station includes a K-means clustering algorithm.
[0020] Furthermore, a method for extracting power generation units with abnormal power generation in a photovoltaic power station includes:
[0021] Based on the preset K value, K samples are determined from the sample data set as the initial clustering centers;
[0022] Extract any sample from the sample data set until all samples in the sample data set are traversed;
[0023] Calculate the distance between the extracted sample and the K initial cluster centers, and classify the extracted sample into the initial cluster center with the smallest distance, so as to obtain K types of samples as K cluster sample sets;
[0024] Extract any one cluster sample set from the K cluster sample sets until the K cluster sample sets are traversed, calculate the average value of the samples in the extracted cluster sample set, and determine it as the new cluster center of the extracted cluster sample set;
[0025] Randomly extract a sample from the K cluster sample sets as the target sample, until all samples in the K cluster sample sets are traversed, and the distance between the extracted target sample and its corresponding new cluster center is calculated as the target distance;
[0026] The target distance is compared with a preset threshold, and the power generation units corresponding to the samples whose target distance is greater than the preset threshold are extracted as the power generation units with abnormal power generation in the photovoltaic power station.
[0027] Furthermore, before arbitrarily extracting a sample from the K cluster sample sets as a target sample, the method further includes:
[0028] Calculate the square root of the sum of squares of the differences between the new cluster centers and the initial cluster centers in the K cluster sample sets, and compare the square root with a preset error threshold;
[0029] If at least any square root is not less than the preset error threshold, the K value is reset.
[0030] Further, the method for setting the K value includes setting it based on a combination of at least any one of the component type, installation method, and inverter type of the power generation unit;
[0031] The component types include monofacial components and bifacial components, the installation methods include bifacial component tracking installation, monofacial component tracking installation, and fixed tilt angle installation, and the inverter types include string inverters and centralized inverters.
[0032] To achieve the above object, the present invention also provides a power generation anomaly identification system for a photovoltaic power station, comprising:
[0033] A sample data extraction module: used to extract a sample data set based on the pre-acquired historical operation data of the power generation unit, wherein the samples in the sample data set include historical equivalent full-power power generation hours data of the power generation unit in the photovoltaic power station when the device status is normal;
[0034] Sample classification and identification module: used to classify the power generation units in the photovoltaic power station based on the historical equivalent full-power power generation hours data, and extract the power generation units with abnormal power generation in the photovoltaic power station.
[0035] Furthermore, it also includes a historical operation data acquisition module, and the historical operation data acquisition module includes:
[0036] Operation data acquisition unit: used to collect operation data of the power generation unit according to a preset collection cycle, wherein the operation data includes the power generation amount and equipment status of the power generation unit;
[0037] Historical operation data generating unit: used to generate historical operation data of the power generation unit based on operation data of no less than two consecutive collection cycles.
[0038] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: based on the pre-extracted operation data of each power generation unit in the photovoltaic power station, the data of the power generation unit when its equipment status is abnormal or faulty is eliminated, and the historical operation data when the equipment status is normal is generated; then, based on the historical operation data, the ratio between the power generation and the power generation capacity of the power generation unit is obtained to evaluate the historical equivalent full-power power generation hours of the power generation unit, and the historical equivalent full-power power generation hours data of the power generation unit in the photovoltaic power station when its equipment status is normal is obtained; then, the classification is determined according to the component type, installation method, and inverter type of the power generation unit, and the K-means clustering algorithm is used to classify and identify the historical equivalent full-power power generation hours data, so as to effectively extract the power generation units with abnormal power generation in the photovoltaic power station. The method and system of the present invention can automatically identify the power generation units with abnormal equivalent full-power power generation hours in the photovoltaic power station, assist the operation and maintenance personnel to quickly find faults, which is conducive to shortening the troubleshooting time, increasing power generation, and reducing labor intensity. DETAILED DESCRIPTION
[0039] The present invention is further described below. The following examples are only used to more clearly illustrate the technical solution of the present invention, and are not intended to limit the protection scope of the present invention.
[0040] A specific embodiment of the present invention provides a method for identifying power generation anomalies in a photovoltaic power station, comprising the following steps:
[0041] Step 1: collect the operating data of each power generation unit in the photovoltaic power station according to the preset collection cycle, the operating data is the power generation and equipment status of the power generation unit, and write the collected operating data into the pre-established operation database. The operation database includes the cumulative power generation array and the equipment status array, and its record structure is as follows:
[0042] E={t,E1,E2,…EN},
[0043] S = {t, S1, S2, ... SN};
[0044] In the formula, E is the cumulative power generation array, S is the device status array, t is the acquisition time, N is the number of power generation units in the photovoltaic power station, E1 is the daily cumulative power generation of the power generation unit numbered 1# at the corresponding time, E2 is the daily cumulative power generation of the power generation unit numbered 2# at the corresponding time, EN is the daily cumulative power generation of the power generation unit numbered N# at the corresponding time, S1 is the device status of the power generation unit numbered 1# at the corresponding time, S2 is the device status of the power generation unit numbered 2# at the corresponding time, and SN is the device status of the power generation unit numbered N# at the corresponding time. Let Si (i = 1, 2, 3...N), if Si is 1, it means that the device status of the power generation unit numbered i# at the corresponding time is normal; if Si is 0, it means that the device status is abnormal or faulty.
[0045] In a photovoltaic power station, an inverter and its electrically connected photovoltaic modules constitute a power generation unit. At present, technicians can easily collect the power generation of each inverter. The collection cycle T of the operating data can be pre-set. If T = 5 minutes, data is collected once every five minutes. During data collection, the equipment status and daily cumulative power generation of the power generation unit are collected synchronously. For example:
[0046] At the collection time t0, the data is collected and written into the running database, and the record is:
[0047] {t0,E1(t0),E2(t0),E3(t0),…EN(t0)},
[0048] {t0,S1(t0),S2(t0),S3(t0),…SN(t0)};
[0049] At the collection time t1=t0+T, the data is collected and written into the operation database, and the record is:
[0050] {t1,E1(t1),E2(t1),E3(t1),…EN(t1)},
[0051] {t1,S1(t1),S2(t1),S3(t1),…SN(t1)},
[0052] In t k =t k-1 +T collection time, collect and write data into the running database, record as:
[0053] {t k ,E1(t k ),E2(t k ),E3(t k ),…EN(t k )},
[0054] {t k ,S1(t k ),S2(t k ),S3(t k ),…SN(t k )};
[0055] Based on the operating data of several consecutive collection cycles, historical operating data of the power generation unit can be generated.
[0056] Step 2: extract a sample data set based on the historical operation data of the power generation unit. In the extraction process, the historical operation data with abnormal or faulty equipment status are removed. The samples in the sample data set are the historical equivalent full-power power generation hours data of the power generation unit in the photovoltaic power station when its equipment status is normal. Its record structure is as follows:
[0057] Ei={[Ei(t1)-Ei(t0)] / Pi,[Ei(t2)-Ei(t1)] / Pi,[Ei(t3)-Ei(t2)] / Pi,…[Ei(t m )-Ei(t m-1 )] / Pi}, i=1,2,…N;
[0058] Where Ei is the historical equivalent full-power power generation hours data of the power generation unit numbered i# in the sample data set when its equipment is in normal condition, t0, t1, t2, t3, …, t m-1 ,t m are the m+1 measurement moments arranged in sequence, Ei(t0), Ei(t1), Ei(t2), Ei(t3), Ei(t m-1 ),…,Ei(t m) is the daily cumulative power generation of the power generation unit numbered i# at the corresponding measurement time when its equipment status is normal, Pi is the power generation capacity of the power generation unit numbered i#, [Ei(t1)-Ei(t0)] / Pi is the equivalent full-power power generation hours of the power generation unit numbered i# during t0~t1, [Ei(t2)-Ei(t1)] / Pi is the equivalent full-power power generation hours of the power generation unit numbered i# during t1~t2, [Ei(t3)-Ei(t2)] / Pi is the equivalent full-power power generation hours of the power generation unit numbered i# during t2~t3, [Ei(t m )-Ei(t m-1 )] / Pi is the number i# of the generating unit at t m-1 ~t m The equivalent full power generation hours during the period.
[0059] Step three, based on the historical equivalent full-power power generation hours data, the K-means clustering algorithm is used to classify each power generation unit in the photovoltaic power station. Different power generation units have different equivalent full-power power generation hours due to their different component types, installation methods, and inverter types. In order to avoid misjudgment, they need to be classified and identified according to their own differences. Therefore, the initial K value can be set according to the combination of the component type, installation method, and inverter type of the power generation unit. After the initial K value is set, the power generation units with different component types, installation methods, and inverter types are divided into K categories according to the K value. First, K samples that meet the conditions are correspondingly determined from the sample data set as the initial clustering center. For example, a photovoltaic power station component has two types of components: single-sided components and double-sided components. The double-sided components are installed in a tracking manner, the single-sided components are installed in a tracking manner and at a fixed tilt angle, the double-sided components use string inverters, and the single-sided components use string inverters and centralized inverters. There are 5 types of equipment combinations, that is, 5 types of photovoltaic power sources, so the K value can be preset to 5. Then, the historical equivalent full-power generation hours data corresponding to the power generation units that meet the five equipment combinations are selected from the sample data set as the initial clustering centers.
[0060] Step 4: Define the distance between the samples in the sample data set and the initial cluster center as the square root of the sum of the squares of the corresponding vector component differences, calculate the distance between each sample in the sample data set and each initial cluster center, and assign the sample to the initial cluster center with the smallest distance to it. Finally, the samples in the sample data set can be divided into K categories of samples, that is, K cluster sample sets.
[0061] Step 5: According to the classification result in step 4, the average value of all sample data in the i-th cluster sample set (i=1, 2, ...K) is obtained as the new cluster center of the cluster sample set.
[0062] Step 6: Calculate the square root of the sum of squares of the differences between the new cluster center and the original initial cluster center in each cluster sample set. If all are less than the set error threshold, proceed to the next step, otherwise return to step 3; if the error threshold is not reached after a certain number of cycles, return to step 3, increase the number of classifications, that is, increase the K value, reselect samples as the initial cluster center, and then proceed to step 4.
[0063] Step 7: Identify the power generation units with abnormal power generation in the photovoltaic power station according to the classification results of step 6, as follows:
[0064] Randomly extract a sample from the K cluster sample sets as the target sample, until all samples in the K cluster sample sets are traversed, and the distance between the extracted target sample and its corresponding new cluster center is calculated as the target distance; then compare the target distance with the preset threshold. If the target distance is greater than the preset threshold, it is considered that the corresponding power generation unit is abnormal and requires maintenance by the operation and maintenance personnel. In order to comprehensively detect and identify power generation units with abnormal power generation, all power generation units corresponding to samples with target distances greater than the preset threshold are extracted as power generation units with abnormal power generation in the photovoltaic power station.
[0065] Step eight: When encountering seasonal changes, changes in power generation equipment, decreased accuracy in identifying abnormalities, or other related situations, you should return to step three and reclassify the photovoltaic power generation units.
[0066] A specific embodiment of the present invention provides a power generation anomaly identification system for a photovoltaic power station. The above method is implemented based on the system of the present invention, including:
[0067] The historical operation data acquisition module is used to execute the contents of step one in the aforementioned inventive method, specifically including: 1. an operation data acquisition unit, used to collect the operation data of the power generation unit according to a preset collection cycle, the operation data including the power generation and equipment status of the power generation unit; 2. a historical operation data generation unit, used to generate the historical operation data of the power generation unit based on the operation data of no less than two consecutive collection cycles.
[0068] Sample data extraction module: used to execute the contents of step 2 in the aforementioned inventive method, namely: extracting a sample data set based on the pre-acquired historical operating data of the power generation unit, wherein the samples in the sample data set include the historical equivalent full-power power generation hours data of the power generation units in the photovoltaic power station when their equipment status is normal.
[0069] Sample classification and identification module: used to execute the contents of steps three to seven in the aforementioned invention method, namely: classify each power generation unit in the photovoltaic power station based on the historical equivalent full-power power generation hours data, and extract the power generation units with abnormal power generation in the photovoltaic power station.
[0070] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A method for identifying power generation anomalies in a photovoltaic power station, characterized in that: The steps include: Extracting a sample data set based on pre-acquired historical operation data of the power generation unit, wherein the samples in the sample data set include historical equivalent full-power power generation hours data of the power generation unit in the photovoltaic power station when the device status is normal; Based on the historical equivalent full-power power generation hours data, each power generation unit in the photovoltaic power station is classified to extract the power generation units with abnormal power generation in the photovoltaic power station; The record structure of the historical equivalent full-power generation hours data is as follows: No={[No(t1)-No(t0)] / Pi,[No(t2)-No(t1)] / Pi,[No(t3)-No(t2)] / Pi,…[No(t m )-No(t m-1 )] / Pi}, i=1,2,…N; Where Ei is the historical equivalent full-power power generation hours data of the power generation unit numbered i# in the sample data set when its equipment is in normal condition, t0, t1, t2, t3, …, t m-1 ,t m are the m+1 measurement moments arranged in sequence, Ei(t0), Ei(t1), Ei(t2), Ei(t3), Ei(t m-1 ), …, Ei(t m ) is the daily cumulative power generation of the power generation unit numbered i# at the corresponding measurement time when its equipment status is normal, Pi is the power generation capacity of the power generation unit numbered i#, [Ei(t1)-Ei(t0)] / Pi is the equivalent full-power power generation hours of the power generation unit numbered i# during t0~t1, [Ei(t2)-Ei(t1)] / Pi is the equivalent full-power power generation hours of the power generation unit numbered i# during t1~t2, [Ei(t3)-Ei(t2)] / Pi is the equivalent full-power power generation hours of the power generation unit numbered i# during t2~t3, [Ei(t m )-Ei(t m-1 )] / Pi is the number i# of the generating unit at t m-1 ~t m The equivalent full power generation hours during the period; Methods for classifying power generation units in a photovoltaic power station, including a K-means clustering algorithm; A method for extracting a power generation unit with abnormal power generation in a photovoltaic power station includes: Based on the preset K value, K samples are determined from the sample data set as the initial clustering centers; Extract any sample from the sample data set until all samples in the sample data set are traversed; Calculate the distance between the extracted sample and the K initial cluster centers, and classify the extracted sample into the initial cluster center with the smallest distance, so as to obtain K types of samples as K cluster sample sets; Extract any one cluster sample set from the K cluster sample sets until the K cluster sample sets are traversed, calculate the average value of the samples in the extracted cluster sample set, and determine it as the new cluster center of the extracted cluster sample set; Randomly extract a sample from the K cluster sample sets as the target sample, until all samples in the K cluster sample sets are traversed, and the distance between the extracted target sample and its corresponding new cluster center is calculated as the target distance; The target distance is compared with a preset threshold, and the power generation unit corresponding to the sample whose target distance is greater than the preset threshold is extracted as the power generation unit with abnormal power generation in the photovoltaic power station; Before randomly extracting a sample from the K cluster sample sets as a target sample, the following steps are also included: Calculate the square root of the sum of squares of the differences between the new cluster centers and the initial cluster centers in the K cluster sample sets, and compare the square root with a preset error threshold; If at least any square root is not less than the preset error threshold, reset the K value; The method for setting the K value includes setting it based on a combination of at least any one of the component type, installation method, and inverter type of the power generation unit; The component types include monofacial components and bifacial components, the installation methods include bifacial component tracking installation, monofacial component tracking installation, and fixed tilt angle installation, and the inverter types include string inverters and centralized inverters.
2. The method for identifying power generation anomalies in a photovoltaic power station according to claim 1, characterized in that: The method for acquiring historical operation data of a power generation unit comprises: Collecting the operating data of the power generation unit according to a preset collection cycle, wherein the operating data includes the power generation amount and the equipment status of the power generation unit; The historical operation data of the power generation unit is generated based on the operation data of no less than two consecutive collection cycles.
3. The method for identifying power generation anomalies in a photovoltaic power station according to claim 2, characterized in that: After collecting the operating data of the power generation unit, the method further includes: writing the collected operating data into a pre-established operating database, wherein the operating database includes an accumulated power generation array and an equipment status array, and its record structure is as follows: E={t,E1,E2,…EN}, S = {t, S1, S2, … SN}; Where, E is the cumulative power generation array, S is the device status array, t is the acquisition time, N is the number of power generation units in the PV power station, E1 is the daily cumulative power generation of the power generation unit No. 1 at the corresponding time, E2 is the daily cumulative power generation of the power generation unit No. 2 at the corresponding time, EN is the daily cumulative power generation of the power generation unit No. N at the corresponding time, S1 is the device status of the power generation unit No. 1 at the corresponding time, S2 is the device status of the power generation unit No. 2 at the corresponding time, and SN is the device status of the power generation unit No. N at the corresponding time.
4. An identification system based on the method for identifying power generation anomalies in a photovoltaic power station according to any one of claims 1 to 3, characterized in that: include: A sample data extraction module: used to extract a sample data set based on the pre-acquired historical operation data of the power generation unit, wherein the samples in the sample data set include historical equivalent full-power power generation hours data of the power generation unit in the photovoltaic power station when the device status is normal; Sample classification and identification module: used to classify each power generation unit in the photovoltaic power station based on the historical equivalent full-power power generation hours data, and extract the power generation units with abnormal power generation in the photovoltaic power station; A historical operation data acquisition module, the historical operation data acquisition module comprising: Operation data acquisition unit: used to collect operation data of the power generation unit according to a preset collection cycle, wherein the operation data includes the power generation amount and equipment status of the power generation unit; Historical operation data generating unit: used to generate historical operation data of the power generation unit based on operation data of no less than two consecutive collection cycles.
Citation Information
Patent Citations
Photovoltaic power station operation state classifying method based on characteristic cluster analysis
CN104915899A
Photovoltaic power station power abnormity data identification method and device
CN106548270A
Fault prediction method for photovoltaic inverter
CN108847686A
Method and system for diagnosing wind wheel imbalance of wind generating set
CN109655200A
Photovoltaic array fault diagnosis method and device based on spatiotemporal distribution characteristics
CN109842373A
