Semi-supervised state evaluation method and equipment suitable for photovoltaic string

Through the semi-supervised state evaluation method, the probability distribution fitting and data division are used to use the current, voltage and power data of the photovoltaic string to solve the difficult problem of the photovoltaic string state evaluation, improve the fault diagnosis accuracy and the intelligent level of the photovoltaic power station. .

CN118981703AInactive Publication Date: 2024-11-19BEIJING BRANCH OF BEIJING JINGNENG CLEAN ENERGY POWER CO LTD +1
View PDF 7 Cites 0 Cited by

Patent Information

Application Number
CN202410975791.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-19
Publication Date
2024-11-19
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing photovoltaic string state evaluation method has high cost of drones, great impact on the environment in image acquisition, difficult modeling, difficult to characterize actual faults, rely on prior knowledge and labeled samples, resulting in high analysis time cost, and engineering applications face difficulty.

Method used

The semi-supervised state evaluation method is used to calculate the characteristic values ​​of the current, voltage and power operation data of the photovoltaic string, and perform probability distribution fitting. The data sample set is divided using the distribution curve to set 10%, 90%, and 95% confidence to determine the status category. Number and center, use similarity to determine the operating state of the photovoltaic string.

Benefits of technology

The accuracy of photovoltaic string fault diagnosis is improved, the dependence on training sample labels is reduced, and the level of intelligence and economic benefits of photovoltaic power stations are improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118981703A_ABST
    Figure CN118981703A_ABST
Patent Text Reader

Abstract

The invention discloses a semi-supervised state evaluation method and device suitable for a photovoltaic string, and the method comprises the following steps: S1, calculating the current, voltage and power operation data feature values of the photovoltaic string, and carrying out the probability distribution fitting of the operation data feature values; s2, using the distribution curve, setting 10%, 90% and 95% confidence coefficients to divide the data, and determining data sample sets X1, X2, X3 and X4 in different states; s3, performing calculation by using the sample data sets in different states to obtain different state centers ci; and S4, judging the operation state of the photovoltaic string by using the similarity between the new data sample and the state center ci. According to the method, the running current, voltage and power characteristic values of the photovoltaic string are constructed, semi-supervised division is performed on different running state data samples according to the distribution of the characteristic values, and different state centers are determined, so that intelligent division of the running state of the photovoltaic string is realized. By using the method, the fault detection rate of the photovoltaic string can be improved, and the economic benefit is remarkable.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic string operation, and in particular to a semi-supervisory state evaluation method and device suitable for photovoltaic strings. Background Art

[0002] As a key component of the photovoltaic power generation system, the safe and efficient operation of photovoltaic strings directly affects the operation level of the power station. Timely and accurate judgment of the operating status of photovoltaic strings is an important basis for the intelligent operation of photovoltaic power stations.

[0003] The evaluation methods of photovoltaic string status mainly include monitoring methods based on image recognition, simulation methods based on electrical parameters, classification methods based on statistical features, and evaluation methods based on intelligent algorithms. The monitoring method based on image recognition uses the temperature difference under different states to judge the fault through drones or thermal imagers, but the drones required for this method are expensive, and the image acquisition and recognition work is greatly affected by the environment; the method based on the reference model compares the theoretical model to judge the fault type, but the main problem of this method is that the modeling is difficult and the results are difficult to characterize the actual fault; the classification method based on statistical features provides a comprehensive discrimination idea for the classification of photovoltaic string status by constructing multidimensional statistical indicators, but this method requires prior knowledge when establishing the object distribution model, and the results depend on the data distribution characteristics. Evaluation methods based on intelligent algorithms include neural network methods and fuzzy theory methods, but this method has weak physical interpretation and defects in practical application.

[0004] As photovoltaic power generation becomes increasingly important in the world's energy structure, accurately determining the operating status of photovoltaic strings is the key to the economic operation of power stations. There are currently many types of status evaluation methods. The drones required for the evaluation method based on image recognition are expensive, and the image acquisition and recognition work is greatly affected by the environment; the evaluation method based on electrical parameters is difficult to model and the results are difficult to characterize actual faults; the evaluation method based on statistical features requires prior knowledge when establishing the object distribution model, and the results depend on the data distribution characteristics. Existing methods are highly dependent on labeled samples, and the time cost of analyzing a large number of similar samples is high, which makes engineering applications difficult. Summary of the invention

[0005] The present invention provides a semi-supervised state evaluation method, device and storage medium applicable to photovoltaic strings, which can solve at least one of the technical problems in the background technology.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] A semi-supervised state evaluation method applicable to photovoltaic strings comprises the following steps:

[0008] S1: Calculate the characteristic values ​​of the current, voltage and power operation data of the photovoltaic string, and perform probability distribution fitting on the characteristic values ​​of the operation data;

[0009] S2: Use the distribution curve to set 10%, 90%, and 95% confidence levels to divide the data and determine the data sample sets X1, X2, X3, and X4 under different states of the photovoltaic strings;

[0010] S3: Based on the sample data sets X1, X2, X3, and X4 of PV strings in different states, determine the number of state categories and calculate the centers c of different states i ;

[0011] S4: Using new data samples and state center c i The similarity of PV arrays is used to determine the operating status of PV strings.

[0012] Furthermore, the step S1 comprises:

[0013] S11. Collect the current and voltage of all photovoltaic strings, and exclude the night data from 19:00 to 6:00 the next day; record the current as I1, the voltage as U1, and the power as P1;

[0014] S12. Normalize the current, voltage and power:

[0015]

[0016] Among them, I μ ,I σ are the mean and standard deviation of the current, U μ , U σ are the mean and standard deviation of voltage, P μ , P σ are the mean and standard deviation of power, respectively;

[0017] S13. Calculate the characteristic values ​​of current, power and voltage;

[0018]

[0019] U=U2

[0020] Where R is the irradiance;

[0021] S14. The probability density function f(x) of the characteristic value of the string data sample is:

[0022]

[0023] Where h is the bandwidth and K(·) is the kernel function.

[0024] Furthermore, the step S2 comprises:

[0025] S21. Using the distribution curve, set the intervals of 10%, 90%, and 95% to divide the data;

[0026] S22. Set the area below 10% as the faulty data sample set, recorded as X1; the area between 10% and 90% as the inefficient data sample set, recorded as X2; the area between 90% and 95% as the good data sample set, recorded as X3; and the area above 95% as the excellent data sample set, recorded as X4.

[0027] Furthermore, the step S3 comprises:

[0028] S31. Assume the number of state categories is c, where c is set to 4 according to S22, and randomly initialize c state centers:

[0029] S32. Iteratively update c state centers c i :

[0030]

[0031] Where m is a constant greater than 0, x j is a data sample from different data sample sets, taken from the sample set [X1, X2, X3, X4], and the weight coefficient u ij The calculation formula is:

[0032] α is a constant.

[0033] Furthermore, the step S4 comprises:

[0034] S41. For the new data sample, calculate the current characteristic value I of the new sample as described in S1 n , voltage characteristic value U n And power characteristic value P n ;

[0035] S42. Calculate new data samples and various centers c i The similarity w in ;

[0036]

[0037] Among them, I ci , U ci , P ci are the current, voltage and power of the i-th center respectively;

[0038] S43. Compare the similarity between the new data sample and each class, and the class with the maximum similarity is the new sample state.

[0039] On the other hand, the present invention further discloses a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the above method.

[0040] On the other hand, the present invention further discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above method.

[0041] It can be seen from the above technical scheme that as photovoltaic power generation becomes increasingly important in the world's energy structure, accurately determining the operating status of photovoltaic strings is the key to the economic operation of power stations. The present invention proposes a semi-supervised state evaluation method suitable for photovoltaic strings. First, the characteristic values ​​of the current, voltage and power operating data of the photovoltaic strings are calculated, and the probability distribution fitting of the characteristic values ​​of the operating data is performed; then, the distribution curve is used to set 10%, 90%, and 95% confidence levels to divide the data, and determine the data sample sets X1, X2, X3, and X4 under different states; the sample data sets under different states are used for calculation to obtain different state centers; finally, the operating status of the photovoltaic strings is determined by the similarity between the new data samples and the state centers. The present invention innovatively proposes a semi-supervised state evaluation method suitable for photovoltaic strings based on the distribution characteristics of data samples, which solves the problem of identification difficulties caused by sparse fault samples and improves the fault diagnosis accuracy of photovoltaic strings.

[0042] Specifically, in order to solve the problem of difficult training sample labels in traditional methods, the present invention performs probability distribution fitting on the characteristic values ​​of the operating data to achieve intelligent division of the operating status of photovoltaic strings. The use of this method can improve the fault detection rate of photovoltaic strings, with significant economic benefits. The main innovations of the present invention include setting confidence levels to divide the data, determining the operating status of the strings based on the similarity between the data samples and the status center, solving the problem of difficult string status evaluation, and improving the diagnostic accuracy and the intelligence level of photovoltaic power stations. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 Time series distribution diagram of the operation data of all photovoltaic strings for 20 consecutive days; Figure 2a Current-irradiance diagrams for all PV strings; Figure 2b Current-voltage distribution diagram of all PV strings; Figure 3 is a flow chart of the method of the present invention; Figure 4 is the division result of the embodiment of the present invention; Figure 5 is a typical string current and voltage curve diagram of an embodiment of the present invention; Figure 6It is a comparison diagram of string analysis in different states of the embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments.

[0053] In order to solve the problems raised in the background technology, the present invention proposes a semi-supervised state evaluation method suitable for photovoltaic strings, which accurately and efficiently divides the operating state. The following is described in the following order: the first part analyzes the operating data characteristics of photovoltaic strings and gives the distribution characteristics; the second part introduces the principles of each part of the method and gives the algorithm steps; the third part uses the method to divide the string operating state, and refers to examples to prove the accuracy and efficiency of the proposed method; the fourth part summarizes.

[0054] The following are specific instructions:

[0055] 1. Analysis of photovoltaic string operation characteristics

[0056] This part takes a centralized photovoltaic power station as the analysis object, studies the distribution characteristics of the operating status of photovoltaic strings, and provides a basis for the division of the proposed method.

[0057] 1.1 Analysis of photovoltaic string operation data

[0058] The current and voltage operating curves of the photovoltaic string for 20 consecutive days are as follows: Figure 1 As shown in the figure, the available data of photovoltaic strings in large photovoltaic power stations include current and voltage. The amount is huge, the data show similarity, and the status information and fault information are hidden in the massive operation data. There are many abnormal status conditions, and it is difficult to directly judge their operating status based on the time series information, and it is difficult to classify the status of photovoltaic strings.

[0059] The analysis shows that there are many types of anomalies in the operating parameters of the studied PV strings, but it is difficult to distinguish the status of the PV strings by observation alone. It is necessary to quantitatively analyze the distribution characteristics of different operating parameters.

[0060] 1.2 Distribution characteristics of PV string operation data

[0061] Since PV string status evaluation is usually not performed at night, night-time data is excluded. Draw the current-irradiance and current-voltage distribution diagrams of all strings, such as Figure 2a and 2b As shown in the figure. The color depth of the scatter distribution diagram indicates the density of the operating state, and the curve is the probability density curve of the corresponding parameter. The following conclusions can be drawn: a large number of outliers in the figure should be classified as independent states, but due to the extremely small sample density, it is difficult to effectively identify them only by the probability density curve; the sample size of abnormal states accounts for less than 0.1%, and the sample distribution is extremely uneven, so it is difficult for classical methods to identify abnormal states.

[0062] 2. Algorithm Principle

[0063] Assume that the sample set of the PV string operation status is X = {x1, x2, ..., x n}, the probability density function of each sample at a certain point is f(x), then:

[0064]

[0065] The method flow chart is as follows Figure 3 As shown, the specific steps include:

[0066] S1: Calculate the characteristic values ​​of the current, voltage and power operation data of the photovoltaic string, and perform probability distribution fitting on the characteristic values ​​of the operation data;

[0067] S2: Use the distribution curve to set 10%, 90%, and 95% confidence levels to divide the data and determine the data sample sets X1, X2, X3, and X4 under different states;

[0068] S3: Use sample data sets under different states to calculate and obtain the centers c of different states i ;

[0069] S4: Using new data samples and state center c i The similarity of PV arrays is used to determine the operating status of PV strings.

[0070] Wherein, the step S1 comprises:

[0071] S11. Collect the current and voltage of all photovoltaic strings and exclude the night data from 19:00 to 6:00 the next day. Record the current as I1, the voltage as U1, and the power as P1.

[0072] S12. Normalize the current, voltage and power:

[0073]

[0074] Among them, I μ ,I σ are the mean and standard deviation of the current, U μ , U σ are the mean and standard deviation of voltage, Pμ , P σ are the mean and standard deviation of power, respectively.

[0075] S13. Calculate the characteristic values ​​of current, power and voltage.

[0076]

[0077] U=U2

[0078] Where R is the irradiance.

[0079] S14. The probability density function f(x) of the characteristic value of the string data sample is:

[0080]

[0081] Where h is the bandwidth and K(·) is the kernel function.

[0082] The step S2 comprises:

[0083] S21. Use the distribution curve to set 10%, 90%, and 95% intervals to divide the data.

[0084] S22. Set the area below 10% as data sample set 1, recorded as X1; the area between 10% and 90% as data sample set 2, recorded as X2; the area between 90% and 95% as data sample set 3, recorded as X3; and the area above 95% as data sample set 4, recorded as X4. Figure 4 shown.

[0085] The step S3 comprises:

[0086] S31. Assume the number of state categories is c, where c is set to 4 according to S22, and randomly initialize c state centers c i :

[0087] S32. Iteratively update c state centers c i :

[0088]

[0089] Where m is a constant greater than 0, x j is a data sample in different data sample sets, taken from the sample set [X1, X2, X3, X4], and the weight coefficient u ij The calculation formula is:

[0090] α is a constant.

[0091] The step S4 comprises:

[0092] S41. For the new data sample, calculate the current characteristic value I of the new sample as described in S1 n , voltage characteristic value U n And power characteristic value P n .

[0093] S42. Calculate new data samples and various centers c i The similarity w in .

[0094]

[0095] in, are the current, voltage and power of the i-th center respectively.

[0096] S43. Compare the similarity between the new data sample and each class, and the class with the maximum similarity is the new sample state.

[0097] The division results are as follows Figure 5 According to the classification results, a typical string electrical characteristic curve is drawn, as shown in Figure 6 As shown; Figure 6 a) Comparison of string efficiency in different states, Figure 6 b) Comparison of string power in different states, Figure 6 c) Comparison of string power generation in different states, Figure 6 d) Comparison of string currents in different states; among them, level A is an excellent string, level B is a good string, level C is an inefficient string, and level D is a faulty string.

[0098] In summary, in order to solve the problem of difficult training sample labels in traditional methods, the present invention performs probability distribution fitting on the characteristic values ​​of operating data to realize intelligent division of the operating status of photovoltaic strings. The use of this method can improve the fault detection rate of photovoltaic strings, with significant economic benefits. The main innovations of the present invention include setting confidence levels to divide data, determining the operating status of strings based on the similarity between data samples and the status center, solving the problem of difficult string status evaluation, and improving the diagnostic accuracy and the intelligence level of photovoltaic power stations.

[0099] On the other hand, the present invention further discloses a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the above method.

[0100] On the other hand, the present invention further discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above method.

[0101] In another embodiment provided in the present application, a computer program product including instructions is also provided, which, when executed on a computer, enables the computer to execute any of the semi-supervised state evaluation methods applicable to photovoltaic strings in the above embodiments.

[0102] It is understandable that the system, device and storage medium provided in the embodiments of the present invention correspond to the method provided in the embodiments of the present invention, and the explanation, examples and beneficial effects of the relevant contents can refer to the corresponding parts in the above methods.

[0103] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive Solid State Disk (SSD)), etc.

[0104] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0105] Each embodiment in this specification is described in a related manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0106] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A semi-supervised state evaluation method applicable to photovoltaic strings, characterized in that: The following steps are included: S1: Calculate the characteristic values ​​of the current, voltage and power operation data of the photovoltaic string, and perform probability distribution fitting on the characteristic values ​​of the operation data; S2: Use the distribution curve to set 10%, 90%, and 95% confidence levels to divide the data and determine the data sample sets X1, X2, X3, and X4 under different states of the photovoltaic strings; S3: Based on the sample data sets X1, X2, X3, and X4 of PV strings in different states, determine the number of state categories and calculate the centers c of different states i ; S4: Using new data samples and state center c i The similarity of PV arrays is used to determine the operating status of PV strings.

2. The semi-supervised state evaluation method applicable to photovoltaic strings according to claim 1, characterized in that: The step S1 comprises: S11. Collect the current and voltage of all photovoltaic strings, and exclude the night data from 19:00 to 6:00 the next day; record the current as I1, the voltage as U1, and the power as P1; S12. Normalize the current, voltage and power: Among them, I μ ,I σ are the mean and standard deviation of the current, U μ , U σ are the mean and standard deviation of voltage, P μ , P σ are the mean and standard deviation of power, respectively; S13. Calculate the characteristic values ​​of current, power and voltage; U=U2 Where R is the irradiance; S14. The probability density function f(x) of the characteristic value of the string data sample is: Where h is the bandwidth and K(·) is the kernel function.

3. The semi-supervised state evaluation method applicable to photovoltaic strings according to claim 2, characterized in that: The step S2 comprises: S21. Using the distribution curve, set the intervals of 10%, 90%, and 95% to divide the data; S22. Set the area below 10% as the faulty data sample set, recorded as X1; the area between 10% and 90% as the inefficient data sample set, recorded as X2; the area between 90% and 95% as the good data sample set, recorded as X3; and the area above 95% as the excellent data sample set, recorded as X4.

4. The semi-supervised state evaluation method applicable to photovoltaic strings according to claim 3 is characterized in that: The step S3 comprises: S31. Assume the number of state categories is c, where c is set to 4 according to S22, and randomly initialize c state centers: S32. Iteratively update c state centers c i : Where m is a constant greater than 0, x j is a data sample from different data sample sets, taken from the sample set [X1, X2, X3, X4], and the weight coefficient u ij The calculation formula is: α is a constant.

5. The semi-supervised state evaluation method applicable to photovoltaic strings according to claim 4, characterized in that: The step S4 comprises: S41. For the new data sample, calculate the current characteristic value I of the new sample as described in S1 n , voltage characteristic value U n And power characteristic value P n ; S42. Calculate new data samples and various centers c i The similarity w in ; in, are the current, voltage and power of the i-th center respectively; S43. Compare the similarity between the new data sample and each class, and the class with the maximum similarity is the new sample state.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method according to any one of claims 1 to 5.

Citation Information

Patent Citations

  • Photovoltaic power station operation state classifying method based on characteristic cluster analysis

    CN104915899A

  • IGBT health state evaluation method of wind power converter power module

    CN110658437A

  • Photovoltaic array fault diagnosis method based on statistical modeling

    CN114201483A

  • Photovoltaic array state evaluation method, system and device and computer readable storage medium

    CN115776276A

  • Photovoltaic panel pollution degree state evaluation and cleaning decision-making system

    CN116227689A