Photovoltaic system quality analysis method and system

Abstract

The present invention relates to a photovoltaic system quality analysis method and system. The method collects historical electricity data of a reference photovoltaic power station and a predicted photovoltaic power station as feature data; calculates the Euclidean distance between the reference photovoltaic power station and the predicted photovoltaic power station according to the feature data; collects real-time data and historical data of meteorological data, wherein the meteorological data includes solar radiation, temperature and humidity, and constructs real-time meteorological feature quantities according to the real-time data, and constructs historical meteorological feature quantities according to the historical data; calculates the similarity between the real-time meteorological feature quantities and the historical meteorological feature quantities according to the Euclidean distance, and determines similar days according to the similarity; predicts the output power of the reference photovoltaic power station and the predicted photovoltaic power station according to the similar days and a kernel extreme learning machine algorithm; determines the quality of the photovoltaic system according to the ratio of the output power of the reference photovoltaic power station to the output power of the predicted photovoltaic power station, and effectively improves the model stability and prediction accuracy.

Description

Technical Field

[0001] The invention relates to a photovoltaic system quality analysis method and system, belonging to the technical field of photovoltaic power generation. Background Art

[0002] With the continuous advancement of photovoltaic technology and the continuous growth of global demand for renewable energy, photovoltaic systems, as a clean and renewable energy solution, have been increasingly widely used in various industries, from large ground power stations to distributed rooftop systems. However, with the surge in the number of photovoltaic system installations, the demand for performance evaluation and fault detection of photovoltaic systems has also increased sharply. This is not only related to the power generation efficiency and energy output of the system, but also directly affects the long-term stable operation of the system and the return on investment.

[0003] Traditional photovoltaic system performance evaluation methods mostly rely on the ratio of theoretical calculated values ​​to actual measured values ​​for comparative analysis. This method can provide a certain reference value under ideal conditions, that is, when the data is complete and the environmental conditions are relatively stable. However, in actual applications, due to the influence of various factors, such as sensor failure, data transmission interruption, sudden changes in weather conditions, etc., the actual measurement data is often incomplete or fluctuates greatly. In addition, the environment in which the photovoltaic system is located is complex and changeable. Changes in environmental factors such as light intensity, temperature, and humidity will have a significant impact on system performance. Therefore, in the case of incomplete data or large changes in the environment, the accuracy of traditional methods will be greatly reduced, and it is difficult to meet the needs of accurate evaluation of photovoltaic system performance.

[0004] The patent document with the patent number "CN115276002A" has developed a method for intelligent analysis of power distribution properties based on power load curves. The problem with this method is that the RNN model, especially when processing long sequence data, may encounter the problem of gradient vanishing or gradient exploding, which will increase the complexity and time required for model training. Traditional RNN may not be effective enough when processing long-term time series data because it is difficult to capture long-term dependencies. This prior art requires manual dynamic adjustment of training parameters until the prediction accuracy reaches 95%, which may be very time-consuming and dependent on expert experience. Summary of the invention

[0005] In order to solve the above problems existing in the prior art, the present invention proposes a photovoltaic system quality analysis method and system.

[0006] The technical solution of the present invention is as follows:

[0007] In one aspect, the present invention provides a photovoltaic system quality analysis method, comprising the following steps:

[0008] Selecting at least one photovoltaic power station as a reference photovoltaic power station, selecting at least one photovoltaic power station as a predicted photovoltaic power station, and collecting historical power data of the reference photovoltaic power station and the predicted photovoltaic power station as feature data;

[0009] Calculate the Euclidean distance between the reference photovoltaic power station and the predicted photovoltaic power station based on the characteristic data;

[0010] Collecting real-time data and historical data of meteorological data, wherein the meteorological data includes solar radiation, temperature and humidity, and constructing real-time meteorological characteristic quantities based on the real-time data and constructing historical meteorological characteristic quantities based on the historical data;

[0011] Calculate the similarity between the real-time meteorological feature quantity and the historical meteorological feature quantity according to the Euclidean distance, and determine the similar day according to the similarity;

[0012] Predict the output power of the reference PV power plant and the predicted PV power plant based on similar days and the kernel extreme learning machine algorithm;

[0013] The quality of the PV system is determined by the ratio of the output power of the reference PV plant to the output power of the predicted PV plant.

[0014] As a preferred implementation, the Euclidean distance is calculated as follows:

[0015] Reference photovoltaic power station characteristic data P A It is expressed as:

[0016] P A = {P A,1 ...P A,i ...P A,n};

[0017] Predict the characteristic data P of photovoltaic power station B It is expressed as:

[0018] P B = {P B,1 ...P B,i ...P B,n};

[0019] Among them, P A,i represents the i-th element of the characteristic data of the reference PV power station, P B,i Represents the i-th element of the characteristic data of the predicted photovoltaic power station;

[0020] Calculate the Euclidean distance d(P A ,P B )of:

[0021]

[0022] As a preferred implementation, the similarity is calculated as follows:

[0023] Let the Nth historical meteorological characteristic be Y N , the Nth real-time meteorological characteristic is Y P :

[0024]

[0025] Among them, Y N1 , Y P1 represents solar radiation, Y N2 , Y P2 Indicates temperature, Y N3 , Y P3 Indicates humidity;

[0026] Calculate the historical meteorological characteristic quantity Y N and real-time meteorological characteristic quantity Y P The difference Q n (i)

[0027] Q n (i)=|Y Ni -Y pi |;

[0028] Among them, i represents meteorological factors, which represent solar radiation, temperature or humidity respectively, i = 1, 2, 3;

[0029] Based on the difference Q n (i) Calculate the historical meteorological characteristic value Y N and real-time meteorological characteristic quantity Y P The correlation coefficient ε N (i)

[0030]

[0031] Among them, ρ represents the rate factor, Q ′ n (i) represents the normalized difference Q n (i);

[0032] According to the correlation coefficient ε N (i) and the Euclidean distance d(P A ,P B ), calculate the historical meteorological characteristic quantity Y N and real-time meteorological characteristic quantity Y P The similarity F N :

[0033]

[0034] Among them, β represents the reduction factor, t represents the number of days in the interval, and ε N(1) represents the correlation coefficient of solar radiation, ε N (2) represents the temperature correlation coefficient, ε N (3) represents the correlation coefficient of humidity.

[0035] The core extreme learning machine (ELM_k) has fast computational efficiency and strong prediction accuracy in regression prediction, so we choose to use this algorithm to predict the output power of the benchmark solar power plant. Use similar day data (i.e. historical data that is most similar to the current forecast target meteorological parameters) to train the core extreme learning machine (ELM_k) model. In this way, the model can learn the output power change law of photovoltaic power plants under similar meteorological conditions.

[0036] As a preferred implementation, the output power prediction method is:

[0037]

[0038] Among them, f(x) represents the predicted output power, h(x) represents the vector composed of network neurons, N represents the dimension of the neural network input layer, and X ELM represents the kernel function, γ represents the function bandwidth, H represents the hidden layer matrix, H T represents the transposed matrix of H, O represents the target value carrier, C represents a data matrix of m rows and n columns, the data matrix is ​​generated by similar day learning, and ν represents a constant greater than 0;

[0039] Kernel function X ELM Using Gaussian kernel function, it is expressed as:

[0040] X ELM =exp(-γ||x i -x j || 2 );

[0041] Among them, exp() represents the kernel function, γ represents the function bandwidth, and x i represents the i-th input sample, x j represents the jth input sample.

[0042] As a preferred implementation, the output power prediction method further includes a particle swarm optimization algorithm:

[0043] Let particle swarm X = {x1,x i ,…,x n}, the position of the particle group at time t is x id (t), the velocity v of the particle group at time t id (t);

[0044] The optimization method is expressed as:

[0045]

[0046] Among them, t is the number of iterations of the particle swarm, c1 is the acceleration factor, and x id (t+1) represents the position of the particle group at time t+1, v id (t+1) represents the velocity of the particle swarm at time t+1, r1 and r2 are random numbers, a id (t) represents the individual optimal solution of the i-th particle, a gd (t) The global optimal solution of the entire particle swarm;

[0047] The fitness function g(x i ) is expressed as:

[0048] g(x i )=min[lim L→∞ ‖f(x)-f0(x)‖+1] -1 ;

[0049] Among them, f0(x) represents the fitness function.

[0050] The kernel function in the kernel extreme learning machine (ELM_k) is directly used to predict power, while the similar day model provides ELM_k with a historical data set similar to the current meteorological conditions. These data sets are used to train and optimize the ELM_k model so that it can more accurately predict the output power of the photovoltaic power station. The connection between the two is that the similar day model provides training data for the ELM_k model, and the ELM_k model uses this data for learning and prediction. The similar day data set is used to train the ELM_k model to learn the law of output power changes of photovoltaic power stations under similar meteorological conditions. Although the specific formula does not directly mention the similar day data, the training process of the model implies the use of this data.

[0051] As a preferred implementation, the method for determining the similar day is:

[0052] Find n historical dates that are consistent with the season and meteorological conditions of the predicted date, sort the n historical dates in descending order according to their similarity, and select the top k historical dates as similar days.

[0053] The performance evaluation (PE) index of a photovoltaic system mainly evaluates the ratio between the theoretical calculated power generation and the actual measured power generation. The closer this ratio is to 1, the more ideal the photovoltaic system is.

[0054] The actual measured values ​​reflect the true performance of the photovoltaic system during operation, while the predicted values ​​are estimates of system performance based on certain models and methods. By comparing the predicted values ​​with the actual measured values, the accuracy of the model can be evaluated, and then it can be determined whether the performance of the photovoltaic system meets expectations and whether there are potential problems or failures. This method can quantify the overall operating status of the photovoltaic power generation system and can be expressed intuitively in a visual way, which is easy to understand and analyze.

[0055] The actual measured power data is divided by the obtained power prediction data. After obtaining the PPI data, it is plotted as a box plot of the median, mean, and IQR. The box plot helps to see the overall distribution between the data at a glance, so we use the box plot to compare the data. By analyzing the changes in these values, the release characteristics and change patterns of the PPI are obtained. In the final fault detection section, the ideal value of the median and mean is 1, that is, if the median and mean are not 1, the data lacks accuracy. In addition, the accuracy can also be confirmed by the size of the IQR. If the IQR value exceeds twice the minimum IQR, it can be said that the system has an accuracy failure. When the PPI value is 1, it means that the smaller the IQR, the more ideal the photovoltaic system.

[0056] Among them, PPI is the ratio of actual instantaneous power to predicted instantaneous power. It reflects the deviation of the performance of the photovoltaic system in actual operation from the expected performance. IQR represents the degree of dispersion of data and is used to identify the fluctuation of photovoltaic power generation. The stability of system performance can be judged by the change in the size of IQR. In the box plot, the position of the mean and median can reflect the average level of the photovoltaic system power generation status.

[0057] By comparing the normal period with the changes in the mean, median and IQR of the PPI, the solar system can be evaluated based on the quality of the indicator during the fault detection period. Therefore, the performance of the solar system obtained through the analysis can be expressed as a quality indicator, and the moment of problem occurrence can be quantitatively analyzed through the mean, median and IQR size.

[0058] In another aspect, the present invention further provides a photovoltaic system quality analysis system, comprising:

[0059] Data collection module: select at least one photovoltaic power station as a reference photovoltaic power station, select at least one photovoltaic power station as a predicted photovoltaic power station, and collect historical power data of the reference photovoltaic power station and the predicted photovoltaic power station as characteristic data;

[0060] Euclidean distance calculation module: calculates the Euclidean distance between the reference PV power station and the predicted PV power station based on the characteristic data;

[0061] Meteorological feature quantity construction module: Meteorological feature quantity collects real-time data and historical data of meteorological data, wherein the meteorological data includes solar radiation, temperature and humidity, and constructs real-time meteorological feature quantity according to the real-time data, and constructs historical meteorological feature quantity according to the historical data;

[0062] Similarity calculation module: calculates the similarity between the real-time meteorological feature quantity and the historical meteorological feature quantity according to the Euclidean distance, and determines the similar day according to the similarity;

[0063] Output power prediction module: predicts the output power of the reference PV power station and the predicted PV power station based on similar days and the kernel extreme learning machine algorithm;

[0064] Quality analysis module: Determines the quality of the PV system by the ratio of the output power of the reference PV plant and the output power of the predicted PV plant.

[0065] The present invention has the following beneficial effects:

[0066] The present invention not only takes into account the historical electricity data of the photovoltaic power station, but also combines the real-time and historical data of meteorological data (solar radiation, temperature, humidity). This comprehensive analysis can more comprehensively evaluate the performance and quality of the photovoltaic system. By calculating the Euclidean distance between the reference photovoltaic power station and the predicted photovoltaic power station, and calculating the similarity based on this, this method can more accurately determine similar days, thereby improving the accuracy of the prediction. The output power is predicted using the kernel extreme learning machine algorithm. This method can handle nonlinear problems and does not require explicit definition of feature space and mapping functions, which improves the generalization ability and stability of the model. The parameters in KELM, such as kernel function bandwidth and regularization coefficient, are optimized by the particle swarm optimization algorithm, which further improves the performance of the model and the accuracy of the prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 The present invention is a flowchart for implementing the method. DETAILED DESCRIPTION

[0068] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0069] It should be understood that the step numbers used in this document are only for convenience of description and are not intended to limit the order in which the steps are executed.

[0070] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0071] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0072] The term "and / or" means and includes any and all possible combinations of one or more of the associated listed items.

[0073] Embodiment 1:

[0074] See also Figure 1 The present invention also provides a photovoltaic system quality analysis method, comprising the following steps:

[0075] Selecting at least one photovoltaic power station as a reference photovoltaic power station, selecting at least one photovoltaic power station as a predicted photovoltaic power station, and collecting historical power data of the reference photovoltaic power station and the predicted photovoltaic power station as feature data;

[0076] Calculate the Euclidean distance between the reference photovoltaic power station and the predicted photovoltaic power station based on the characteristic data;

[0077] Collecting real-time data and historical data of meteorological data, wherein the meteorological data includes solar radiation, temperature and humidity, and constructing real-time meteorological characteristic quantities based on the real-time data and constructing historical meteorological characteristic quantities based on the historical data;

[0078] Calculate the similarity between the real-time meteorological feature quantity and the historical meteorological feature quantity according to the Euclidean distance, and determine the similar day according to the similarity;

[0079] Predict the output power of the reference PV power plant and the predicted PV power plant based on similar days and the kernel extreme learning machine algorithm;

[0080] The quality of the PV system is determined by the ratio of the output power of the reference PV plant to the output power of the predicted PV plant.

[0081] As a preferred implementation, the Euclidean distance is calculated as follows:

[0082] Reference photovoltaic power station characteristic data P A It is expressed as:

[0083] P A = {P A,1 ...PA,i ...P A,n};

[0084] Predict the characteristic data P of photovoltaic power station B It is expressed as:

[0085] P B = {P B,1 ...P B,i ...P B,n};

[0086] Among them, P A,i represents the i-th element of the characteristic data of the reference PV power station, P B,i Represents the i-th element of the characteristic data of the predicted photovoltaic power station;

[0087] Calculate the Euclidean distance d(P A ,P B )of:

[0088]

[0089] As a preferred implementation, the similarity is calculated as follows:

[0090] Let the Nth historical meteorological characteristic be Y N , the Nth real-time meteorological characteristic is Y P :

[0091]

[0092] Among them, Y N1 , Y P1 represents solar radiation, Y N2 , Y P2 Indicates temperature, Y N3 , Y P3 Indicates humidity;

[0093] Calculate the historical meteorological characteristic quantity Y N and real-time meteorological characteristic quantity Y P The difference Q n (i)

[0094] Q n (i)=|Y Ni -Y pi |;

[0095] Among them, i represents meteorological factors, which represent solar radiation, temperature or humidity respectively, i = 1, 2, 3;

[0096] Based on the difference Q n (i) Calculate the historical meteorological characteristic value Y N and real-time meteorological characteristic quantity YP The correlation coefficient ε N (i)

[0097]

[0098] Among them, ρ represents the rate factor, Q ′ n (i) represents the normalized difference Q n (i);

[0099] The normalization method is:

[0100]

[0101] According to the correlation coefficient ε N (i) and the Euclidean distance d(P A ,P B ), calculate the historical meteorological characteristic quantity Y N and real-time meteorological characteristic quantity Y P The similarity F N :

[0102]

[0103] Among them, β represents the reduction factor, t represents the number of days in the interval, and ε N (1) represents the correlation coefficient of solar radiation, ε N (2) represents the temperature correlation coefficient, ε N (3) represents the correlation coefficient of humidity.

[0104] As a preferred implementation, the output power prediction method is:

[0105]

[0106] Among them, f(x) represents the predicted output power, h(x) represents the vector composed of network neurons, N represents the dimension of the neural network input layer, and X ELM represents the kernel function, γ represents the function bandwidth, H represents the hidden layer matrix, H T represents the transposed matrix of H, O represents the target value carrier, C represents a data matrix of m rows and n columns, the data matrix is ​​generated by similar day learning, and ν represents a constant greater than 0;

[0107] Kernel function X ELM Using Gaussian kernel function, it is expressed as:

[0108] X ELM =exp(-γ||x i -x j || 2 );

[0109] Among them, exp() represents the kernel function, γ represents the function bandwidth, and x i represents the i-th input sample, x j represents the jth input sample.

[0110] As a preferred implementation, the output power prediction method further includes a particle swarm optimization algorithm:

[0111] Let particle swarm X = {x1,x i ,…,x n}, the position of the particle group at time t is x id (t), the velocity v of the particle group at time t id (t);

[0112] The optimization method is expressed as:

[0113]

[0114] Among them, t is the number of iterations of the particle swarm, c1 is the acceleration factor, and x id (t+1) represents the position of the particle group at time t+1, v id (t+1) represents the velocity of the particle swarm at time t+1, r1 and r2 are random numbers, a id (t) represents the individual optimal solution of the i-th particle, a gd (t) The global optimal solution of the entire particle swarm;

[0115] The fitness function g(x i ) is expressed as:

[0116] g(x i )=min[lim L→∞ ‖f(x)-f0(x)‖+1] -1 ;

[0117] Among them, f0(x) represents the fitness function.

[0118] As a preferred implementation, the method for determining the similar day is:

[0119] Find n historical dates that are consistent with the season and meteorological conditions of the predicted date, sort the n historical dates in descending order according to their similarity, and select the top k historical dates as similar days.

[0120] Embodiment 2:

[0121] The present invention also provides a photovoltaic system quality analysis system, comprising:

[0122] Data collection module: select at least one photovoltaic power station as a reference photovoltaic power station, select at least one photovoltaic power station as a predicted photovoltaic power station, and collect historical power data of the reference photovoltaic power station and the predicted photovoltaic power station as characteristic data;

[0123] Euclidean distance calculation module: calculates the Euclidean distance between the reference PV power station and the predicted PV power station based on the characteristic data;

[0124] Meteorological feature quantity construction module: Meteorological feature quantity collects real-time data and historical data of meteorological data, wherein the meteorological data includes solar radiation, temperature and humidity, and constructs real-time meteorological feature quantity according to the real-time data, and constructs historical meteorological feature quantity according to the historical data;

[0125] Similarity calculation module: calculates the similarity between the real-time meteorological feature quantity and the historical meteorological feature quantity according to the Euclidean distance, and determines the similar day according to the similarity;

[0126] Output power prediction module: predicts the output power of the reference PV power station and the predicted PV power station based on similar days and the kernel extreme learning machine algorithm;

[0127] Quality analysis module: Determines the quality of the PV system by the ratio of the output power of the reference PV plant and the output power of the predicted PV plant.

[0128] In the embodiments of the present application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Among them, A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c can be represented by: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, c can be single or multiple.

[0129] Those of ordinary skill in the art will appreciate that the various units and algorithm steps described in the embodiments disclosed herein can be implemented in a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0130] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0131] In several embodiments provided in the present application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory; hereinafter referred to as: ROM), random access memory (Random Access Memory; hereinafter referred to as: RAM), disk or optical disk, and other media that can store program codes.

[0132] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A photovoltaic system quality analysis method, characterized in that: The following steps are involved: Selecting at least one photovoltaic power station as a reference photovoltaic power station, selecting at least one photovoltaic power station as a predicted photovoltaic power station, and collecting historical power data of the reference photovoltaic power station and the predicted photovoltaic power station as feature data; Calculate the Euclidean distance between the reference photovoltaic power station and the predicted photovoltaic power station based on the characteristic data; Collecting real-time data and historical data of meteorological data, wherein the meteorological data includes solar radiation, temperature and humidity, and constructing real-time meteorological characteristic quantities based on the real-time data and constructing historical meteorological characteristic quantities based on the historical data; Calculate the similarity between the real-time meteorological feature quantity and the historical meteorological feature quantity according to the Euclidean distance, and determine the similar day according to the similarity; Predict the output power of the reference PV power plant and the predicted PV power plant based on similar days and the kernel extreme learning machine algorithm; The quality of the PV system is determined by the ratio of the output power of the reference PV plant to the output power of the predicted PV plant.

2. The photovoltaic system quality analysis method according to claim 1, characterized in that: The calculation method of the Euclidean distance is: Reference photovoltaic power station characteristic data P A It is expressed as: P A ={P A,1 ...P A,i ...P A,n }; Predict the characteristic data P of photovoltaic power station B It is expressed as: P B ={P B,1 ...P B,i ...P B,n }; Among them, P A,i represents the i-th element of the characteristic data of the reference photovoltaic power station, P B,i Represents the i-th element of the characteristic data of the predicted photovoltaic power station; Calculate the Euclidean distance d(P A ,P B )of:

3. The photovoltaic system quality analysis method according to claim 2, characterized in that: The similarity is calculated as follows: Let the Nth historical meteorological characteristic be Y N , the Nth real-time meteorological characteristic is Y P : Among them, Y N1 , Y P1 represents solar radiation, Y N2 , Y P2 Indicates temperature, Y N3 , Y P3 Indicates humidity; Calculate the historical meteorological characteristic quantity Y N and real-time meteorological characteristic quantity Y P The difference Q n (i) Q n (i)=|Y Ni -Y pi |; Among them, i represents meteorological factors, which represent solar radiation, temperature or humidity respectively, i = 1, 2, 3; Based on the difference Q n (i) Calculate the historical meteorological characteristic value Y N and real-time meteorological characteristic quantity Y P The correlation coefficient ε N (i) Among them, ρ represents the rate factor, Q ′ n (i) represents the normalized difference Q n (i); According to the correlation coefficient ε N (i) and the Euclidean distance d(P A ,P B ), calculate the historical meteorological characteristic quantity Y N and real-time meteorological characteristic quantity Y P The similarity F N : Among them, β represents the reduction factor, t represents the number of days in the interval, and ε N (1) represents the correlation coefficient of solar radiation, ε N (2) represents the temperature correlation coefficient, ε N (3) represents the correlation coefficient of humidity.

4. The photovoltaic system quality analysis method according to claim 1, characterized in that: The output power prediction method is: Among them, f(x) represents the predicted output power, h(x) represents the vector composed of network neurons, N represents the dimension of the neural network input layer, and X ELM represents the kernel function, γ represents the function bandwidth, H represents the hidden layer matrix, H T represents the transposed matrix of H, O represents the target value carrier, C represents a data matrix of m rows and n columns, the data matrix is ​​generated by similar day learning, and ν represents a constant greater than 0; Kernel function X ELM Using Gaussian kernel function, it is expressed as: X ELM =exp(-γ||x i -x j || 2 ); Among them, exp() represents the kernel function, γ represents the function bandwidth, and x i represents the i-th input sample, x j represents the jth input sample.

5. The photovoltaic system quality analysis method according to claim 4, characterized in that: The output power prediction method also includes a particle swarm optimization algorithm: Let particle swarm X = {x1,x i ,…,x n }, the position of the particle swarm at time t is x id (t), the velocity v of the particle group at time t id (t); The optimization method is expressed as: Among them, t is the number of iterations of the particle swarm, c1 is the acceleration factor, and x id (t+1) represents the position of the particle group at time t+1, v id (t+1) represents the velocity of the particle swarm at time t+1, r1 and r2 are random numbers, a id (t) represents the individual optimal solution of the i-th particle, a gd (t) The global optimal solution of the entire particle swarm; The fitness function g(x i ) is expressed as: g(x i )=min[lim L→∞ ‖f(x)-f0(x)‖+1] -1 ; Among them, f0(x) represents the fitness function.

6. The photovoltaic system quality analysis method according to claim 3, characterized in that: The method for determining the similar day is: Find n historical dates that are consistent with the season and meteorological conditions of the predicted date, sort the n historical dates in descending order according to their similarity, and select the top k historical dates as similar days.

7. A photovoltaic system quality analysis system, characterized in that: include: Data collection module: select at least one photovoltaic power station as a reference photovoltaic power station, select at least one photovoltaic power station as a predicted photovoltaic power station, and collect historical power data of the reference photovoltaic power station and the predicted photovoltaic power station as characteristic data; Euclidean distance calculation module: calculates the Euclidean distance between the reference photovoltaic power station and the predicted photovoltaic power station based on the characteristic data; Meteorological feature quantity construction module: Meteorological feature quantity collects real-time data and historical data of meteorological data, wherein the meteorological data includes solar radiation, temperature and humidity, and constructs real-time meteorological feature quantity according to the real-time data, and constructs historical meteorological feature quantity according to the historical data; Similarity calculation module: calculates the similarity between the real-time meteorological feature quantity and the historical meteorological feature quantity according to the Euclidean distance, and determines the similar day according to the similarity; Output power prediction module: predicts the output power of the reference photovoltaic power station and the predicted photovoltaic power station based on similar days and the kernel extreme learning machine algorithm; Quality analysis module: Determines the quality of the PV system by the ratio of the output power of the reference PV plant and the output power of the predicted PV plant.

8. The photovoltaic system quality analysis system according to claim 7, characterized in that: The Euclidean distance calculation module calculates the Euclidean distance by: Reference photovoltaic power station characteristic data P A It is expressed as: P A ={P A,1 ...P A,i ...P A,n }; Predict the characteristic data P of photovoltaic power station B It is expressed as: P B ={P B,1 ...P B,i ...P B,n }; Among them, P A,i represents the i-th element of the characteristic data of the reference photovoltaic power station, P B,i Represents the i-th element of the characteristic data of the predicted photovoltaic power station; Calculate the Euclidean distance d(P A ,P B )of:

9. The photovoltaic system quality analysis system according to claim 8, characterized in that: The similarity calculation module calculates the similarity by: Let the Nth historical meteorological characteristic be Y N , the Nth real-time meteorological characteristic is Y P : Among them, Y N1 , Y P1 represents solar radiation, Y N2 , Y P2 Indicates temperature, Y N3 , Y P3 Indicates humidity; Calculate the historical meteorological characteristic quantity Y N and real-time meteorological characteristic quantity Y P The difference Q n (i) Q n (i)=|Y Ni -Y pi |; Among them, i represents meteorological factors, which represent solar radiation, temperature or humidity respectively, i = 1, 2, 3; Based on the difference Q n (i) Calculate the historical meteorological characteristic value Y N and real-time meteorological characteristic quantity Y P The correlation coefficient ε N (i) Among them, ρ represents the rate factor, Q ′ n (i) represents the normalized difference Q n (i); According to the correlation coefficient ε N (i) and the Euclidean distance d(P A ,P B ), calculate the historical meteorological characteristic quantity Y N and real-time meteorological characteristic quantity Y P The similarity F N : Among them, β represents the reduction factor, t represents the number of days in the interval, and ε N (1) represents the correlation coefficient of solar radiation, ε N (2) represents the temperature correlation coefficient, ε N (3) represents the correlation coefficient of humidity.

10. The photovoltaic system quality analysis system according to claim 7, characterized in that: The output power prediction module, the output power prediction method is: Among them, f(x) represents the predicted output power, h(x) represents the vector composed of network neurons, N represents the dimension of the neural network input layer, and X ELM represents the kernel function, γ represents the function bandwidth, H represents the hidden layer matrix, H T represents the transposed matrix of H, O represents the target value carrier, C represents a data matrix of m rows and n columns, the data matrix is ​​generated by similar day learning, and ν represents a constant greater than 0; Kernel function X ELM Using Gaussian kernel function, it is expressed as: X ELM =exp(-γ||x i -x j || 2 ); Among them, exp() represents the kernel function, γ represents the function bandwidth, and x i represents the i-th input sample, x j represents the jth input sample.

Images