A photovoltaic inverter intelligent string diagnosis method and system

Abstract

The present invention provides a method and system for intelligent string diagnosis of photovoltaic inverters, which belongs to the technical field of photovoltaic power generation. The present invention provides a method and system for automatically diagnosing faults of photovoltaic inverters, so that the photovoltaic power generation system has intelligent and digital characteristics, which can save a lot of manpower and material resources for component inspection and reduce the daily maintenance cost of the photovoltaic power generation system; by adopting a deep learning algorithm and optimizing the model, the speed and accuracy of fault detection are improved, early warning can be given, and the fault location and cause can be quickly located; the use of visualization equipment improves the digital level of the photovoltaic power generation system and has better operability.

Description

Technical Field

[0001] The invention belongs to the field of photovoltaic power generation, and in particular relates to an intelligent string diagnosis method for photovoltaic inverters. Background Art

[0002] In recent years, photovoltaic power generation has a good development prospect in my country. In photovoltaic power generation systems, the conversion between solar energy and electrical energy is achieved through inverters. The inverter converts the direct current generated by photovoltaic modules into alternating current, which has an important impact on the stable operation of photovoltaic power stations. Due to the complexity and diversity of the environment in which photovoltaic power stations are located in my country, the string inverter solution shows great advantages. The string inverter adopts a modular design. Each photovoltaic string has a matching inverter for energy conversion, and MPPT tracking can be performed. The individual differences and local shading of different strings will not limit their respective power generation, which has a certain effect on the increase of the power generation of photovoltaic power stations.

[0003] In complex and diverse environments, fault diagnosis and early warning are essential to ensure the stable operation of photovoltaic power generation systems. Early fault diagnosis technology mainly relied on maintenance personnel to make direct judgments and troubleshoot based on instrument measurement data and their own experience. This method is relatively simple and economical and practical in simple equipment. In recent years, with the continuous development of industrial intelligence, traditional fault diagnosis technology has been unable to meet the diagnostic needs of large-scale industrial equipment. Improving the accuracy of fault diagnosis and reducing manpower and material consumption are still one of the important issues facing photovoltaic power generation systems. Summary of the invention

[0004] The purpose of the present invention is to solve the above-mentioned problems existing in the prior art and to provide a method and system for intelligent diagnosis of photovoltaic inverter strings.

[0005] The present invention is achieved through the following technical solutions:

[0006] A photovoltaic inverter intelligent string diagnosis method comprises the following steps:

[0007] 1) Forming a photovoltaic string IV characteristic database: The photovoltaic system includes a plurality of photovoltaic strings, each of which is connected to a string inverter, and the current value and voltage value of each photovoltaic string are collected in real time to form a photovoltaic string IV characteristic database;

[0008] 2) Construct the data matrix X and normalize the data matrix: The data matrix X represents the fault parameters, which is expressed as formula (1),

[0009] X=[V oc I sc P m V m ] (1)

[0010] Among them, V oc Indicates the open circuit voltage, I sc Indicates a short circuit, I m Indicates the maximum power current, V m represents the maximum power voltage, let

[0011]

[0012] Among them, x ij is the initialization output matrix, z ij is the normalized input, x min is the minimum value of each row of matrix X, x max is the maximum value of each row of matrix X, y min is the minimum value of each row in the normalized matrix, y max is the maximum value of each row in the normalized matrix;

[0013] 3) Model training: Use the following function to train the deep learning model.

[0014]

[0015] Linear(x)=kx+c (4)

[0016] Among them, α, k, and c all represent characteristic parameters generated during the function iteration process;

[0017] The output matrix is ​​constructed as:

[0018] Z out =sig(Z in C T ) (5)

[0019] P in =Z out (6)

[0020] P out =f1(P in ) (7)

[0021] L in =P out (8)

[0022] L out =Linear(L in W T ) (9)

[0023] Among them, Z in is the normalized matrix of the input, Z out is the output matrix after applying the function, P in is the input matrix of the optimization layer, P outis the output matrix of the optimization layer, f1 is the specific optimization function, L in is the input matrix of the output layer, L out is the output matrix after applying the function, C T and W T Represent the connection weights of the input layer and the output layer respectively;

[0024] The fitness function is Where j represents the inverter fault state and is a positive integer, i represents the iteration index, and y ij.des is the expected output value, y ij.out is the actual output value;

[0025] 4) Early warning and visualization of faults: Based on the model output results, early warning of line faults is performed, and visualization operations are performed after data rendering.

[0026] Furthermore, in step 3), formulas (11)-(12) are also used for optimization:

[0027]

[0028]

[0029] Where w is the inertia weight factor, t is the number of iterations, c1 and c2 are system and environment parameters, r1 and r2 are random numbers with values ​​between 0 and 1, represents the optimal position of the i-th particle in the current iteration, represents the best historical position of the i-th particle, represents the current position of the i-th particle in the current iteration, represents the next iteration position of the ith particle, represents the current velocity of the ith particle, Represents the next iteration rate of the i-th particle.

[0030] Furthermore, the PV string IV characteristic database in step 1) includes PV string IV characteristic data in a normal state and PV string IV characteristic data in a fault state.

[0031] Furthermore, the visualization operation in step 4) can display the fault location, fault type and fault parameters on the display device.

[0032] In addition, the present invention also provides a photovoltaic inverter intelligent string diagnosis system, comprising:

[0033] 1) Data acquisition component: The photovoltaic system includes a plurality of photovoltaic strings, each of which is connected to a string inverter, and the current value and voltage value of each photovoltaic string are collected in real time to form a photovoltaic string IV characteristic database;

[0034] 2) Data preprocessing component: construct the data matrix X and normalize the data matrix: the data matrix X represents the fault parameters, which is expressed as formula (1),

[0035] X=[V oc I sc P m V m ] (1)

[0036] Among them, V oc Indicates the open circuit voltage, I sc Indicates a short circuit, I m Indicates the maximum power current, V m represents the maximum power voltage, let

[0037]

[0038] Among them, x ij is the initialization output matrix, z ij is the normalized input, x min is the minimum value of each row of matrix X, x max is the maximum value of each row of matrix X, y min is the minimum value of each row in the normalized matrix, y max is the maximum value of each row in the normalized matrix;

[0039] 3) Model training component: Use the following functions to train the deep learning model.

[0040]

[0041] Linear(x)=kx+c (4)

[0042] Among them, α, k, and c all represent characteristic parameters generated during the function iteration process;

[0043] The output matrix is ​​constructed as:

[0044] Z out =sig(Z in C T ) (5)

[0045] P in =Z out (6)

[0046] P out =f1(P in ) (7)

[0047] L in =P out (8)

[0048] L out =Linear(L in W T ) (9)

[0049] Among them, Z in is the normalized matrix of the input, Z out is the output matrix after applying the function, P in is the input matrix of the optimization layer, P out is the output matrix of the optimization layer, f1 is the specific optimization function, L in is the input matrix of the output layer, L out is the output matrix after applying the function, C T and W T Represent the connection weights of the input layer and the output layer respectively;

[0050] The fitness function is Where j represents the inverter fault state and is a positive integer, i represents the iteration index, and y ij.des is the expected output value, y ij.out is the actual output value;

[0051] 4) Early warning and visualization components: Based on the model output results, early warning is issued for line faults, and visualization operations are performed after data rendering.

[0052] Furthermore, the model training component is optimized using formulas (11)-(12):

[0053]

[0054]

[0055] Where w is the inertia weight factor, t is the number of iterations, c1 and c2 are system and environment parameters, r1 and r2 are random numbers with values ​​between 0 and 1, represents the optimal position of the i-th particle in the current iteration, represents the best historical position of the i-th particle, represents the current position of the i-th particle in the current iteration, represents the next iteration position of the ith particle, represents the current velocity of the ith particle, Represents the next iteration rate of the i-th particle.

[0056] Furthermore, the PV string IV characteristic database in the data acquisition component includes PV string IV characteristic data in a normal state and PV string IV characteristic data in a fault state.

[0057] Furthermore, the visualization operation can display the fault location, fault type and fault parameters on the display device.

[0058] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention provides a method and system for automatically diagnosing photovoltaic inverter faults, so that the photovoltaic power generation system has intelligent and digital characteristics, which can save a lot of manpower and material resources for component inspection and reduce the daily maintenance costs of the photovoltaic power generation system; by adopting a deep learning algorithm and optimizing the model, the speed and accuracy of fault detection are improved, early warning can be given, and the fault location and cause can be quickly located; the use of visualization equipment improves the digital level of the photovoltaic power generation system and improves operability. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 The present invention provides a flow chart of the photovoltaic inverter intelligent string diagnosis method.

[0060] Figure 2 It is a schematic diagram of the photovoltaic inverter intelligent string diagnosis system provided by the present invention. DETAILED DESCRIPTION

[0061] The present invention is further described in detail below in conjunction with the accompanying drawings:

[0062] Figure 1 The figure is a flow chart of the photovoltaic inverter intelligent string diagnosis method used in the present invention. The diagnosis method of the present invention comprises the following steps:

[0063] 1) Forming a photovoltaic string IV characteristic database: The photovoltaic system includes a plurality of photovoltaic strings, each of which is connected to a string inverter, and the current value and voltage value of each photovoltaic string are collected in real time to form a photovoltaic string IV characteristic database;

[0064] 2) Construct the data matrix X and normalize the data matrix: The data matrix X represents the fault parameters, which is expressed as formula (1),

[0065] X=[V oc I sc P m V m ] (1)

[0066] Among them, V oc Indicates the open circuit voltage, I sc Indicates a short circuit, I m Indicates the maximum power current, V m represents the maximum power voltage, let

[0067]

[0068] Among them, xij is the initialization output matrix, z ij is the normalized input, x min is the minimum value of each row of matrix X, x max is the maximum value of each row of matrix X, y min is the minimum value of each row in the normalized matrix, y max is the maximum value of each row in the normalized matrix;

[0069] 3) Model training: Use the following function to train the deep learning model.

[0070]

[0071] Linear(x)=kx+c (4)

[0072] Among them, α, k, and c all represent characteristic parameters generated during the function iteration process;

[0073] The output matrix is ​​constructed as:

[0074] Z out =sig(Z in C T ) (5)

[0075] P in =Z out (6)

[0076] P out =f1(P in ) (7)

[0077] L in =P out (8)

[0078] L out =Linear(L in W T ) (9)

[0079] Among them, Z in is the normalized matrix of the input, Z out is the output matrix after applying the function, P in is the input matrix of the optimization layer, P out is the output matrix of the optimization layer, f1 is the specific optimization function, L in is the input matrix of the output layer, L out is the output matrix after applying the function, C T and W T Represent the connection weights of the input layer and the output layer respectively;

[0080] The fitness function is Where j represents the inverter fault state and is a positive integer, i represents the iteration index, and y ij.des is the expected output value, y ij.out is the actual output value;

[0081] 4) Early warning and visualization of faults: Based on the model output results, early warning of line faults is performed, and visualization operations are performed after data rendering.

[0082] Furthermore, in step 3), formulas (11)-(12) are also used for optimization:

[0083]

[0084]

[0085] Where w is the inertia weight factor, t is the number of iterations, c1 and c2 are system and environment parameters, r1 and r2 are random numbers with values ​​between 0 and 1, represents the optimal position of the i-th particle in the current iteration, represents the best historical position of the i-th particle, represents the current position of the i-th particle in the current iteration, represents the next iteration position of the ith particle, represents the current velocity of the ith particle, Represents the next iteration rate of the i-th particle.

[0086] The PV string IV characteristic database in step 1) includes the PV string IV characteristic data in a normal state and the PV string IV characteristic data in a fault state;

[0087] The visualization operation in step 4) can display the fault location, fault type and fault parameters on the display device.

[0088] In addition, the present invention also provides a photovoltaic inverter intelligent string diagnosis system, comprising:

[0089] 1) Data acquisition component: The photovoltaic system includes a plurality of photovoltaic strings, each of which is connected to a string inverter, and the current value and voltage value of each photovoltaic string are collected in real time to form a photovoltaic string IV characteristic database;

[0090] 2) Data preprocessing component: construct the data matrix X and normalize the data matrix: the data matrix X represents the fault parameters, which is expressed as formula (1),

[0091] X=[V oc I sc P m V m ] (1)

[0092] Among them, V oc Indicates the open circuit voltage, I sc Indicates a short circuit, I m Indicates the maximum power current, V m represents the maximum power voltage, let

[0093]

[0094] Among them, x ij is the initialization output matrix, z ij is the normalized input, x min is the minimum value of each row of matrix X, x max is the maximum value of each row of matrix X, y min is the minimum value of each row in the normalized matrix, y max is the maximum value of each row in the normalized matrix;

[0095] 3) Model training component: Use the following functions to train the deep learning model.

[0096]

[0097] Linear(x)=kx+c (4)

[0098] Among them, α, k, and c all represent characteristic parameters generated during the function iteration process;

[0099] The output matrix is ​​constructed as:

[0100] Z out =sig(Z in C T ) (5)

[0101] P in =Z out (6)

[0102] P out =f1(P in ) (7)

[0103] L in =P out (8)

[0104] L out =Linear(L in W T ) (9)

[0105] Among them, Z in is the normalized matrix of the input, Z out is the output matrix after applying the function, P inis the input matrix of the optimization layer, P out is the output matrix of the optimization layer, f1 is the specific optimization function, L in is the input matrix of the output layer, L out is the output matrix after applying the function, C T and W T Represent the connection weights of the input layer and the output layer respectively;

[0106] The fitness function is Where j represents the inverter fault state and is a positive integer, i represents the iteration index, and y ij.des is the expected output value, y ij.out is the actual output value;

[0107] 4) Early warning and visualization components: Based on the model output results, early warning is issued for line faults, and visualization operations are performed after data rendering.

[0108] Furthermore, the model training component is optimized using formulas (11)-(12):

[0109]

[0110]

[0111] Where w is the inertia weight factor, t is the number of iterations, c1 and c2 are system and environment parameters, r1 and r2 are random numbers with values ​​between 0 and 1, represents the optimal position of the i-th particle in the current iteration, represents the best historical position of the i-th particle, represents the current position of the i-th particle in the current iteration, represents the next iteration position of the ith particle, represents the current velocity of the ith particle, Represents the next iteration rate of the i-th particle.

[0112] Furthermore, the PV string IV characteristic database in the data acquisition component includes PV string IV characteristic data in a normal state and PV string IV characteristic data in a fault state.

[0113] Furthermore, the visualization operation can display the fault location, fault type and fault parameters on the display device.

[0114] The present invention provides a method and system for automatically diagnosing photovoltaic inverter faults, so that the photovoltaic power generation system has intelligent and digital characteristics, which can save a lot of manpower and material resources for component inspection and reduce the daily maintenance cost of the photovoltaic power generation system; by adopting a deep learning algorithm and optimizing the model, the speed and accuracy of fault detection are improved, early warning can be given, and the fault location and cause can be quickly located; the use of visualization equipment improves the digital level of the photovoltaic power generation system and improves operability.

[0115] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "connected" and "connection" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0116] In the description of the present invention, unless otherwise specified, the terms "upper", "lower", "left", "right", "inside", "outside", etc. indicate directions or positional relationships based on the directions or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, they cannot be understood as limitations on the present invention.

[0117] Finally, it should be explained that the above technical solution is only one implementation method of the present invention. For those skilled in the art, it is easy to make various types of improvements or modifications based on the application methods and principles disclosed in the present invention, and it is not limited to the method described in the above specific implementation method of the present invention. Therefore, the method described above is only preferred and does not have a restrictive meaning.

Claims

1. A photovoltaic inverter intelligent string diagnosis method, comprising the following steps: 1) Forming a photovoltaic string IV characteristic database: The photovoltaic system includes a plurality of photovoltaic strings, each of which is connected to a string inverter, and the current value and voltage value of each photovoltaic string are collected in real time to form a photovoltaic string IV characteristic database; 2) Construct the data matrix X and normalize the data matrix: The data matrix X represents the fault parameters, which is expressed as formula (1), Among them, V oc Indicates the open circuit voltage, I sc Indicates a short circuit, I m Indicates the maximum power current, V m represents the maximum power voltage, let Among them, x ij is the initialization output matrix, z ij is the normalized input, x min is the minimum value of each row of matrix X, x max is the maximum value of each row of matrix X, y min is the minimum value of each row in the normalized matrix, y max is the maximum value of each row in the normalized matrix; 3) Model training: Use the following function to train the deep learning model. Linear(x)=kx+c (4) Among them, α, k, and c all represent characteristic parameters generated during the function iteration process; The output matrix is ​​constructed as: Z out =sig(Z in C T ) (5) P in =Z out (6) P out =f1(P in ) (7) L in =P out (8) L out =Linear(L in W T ) (9) Among them, Z in is the normalized matrix of the input, Z out is the output matrix after applying the function, P in is the input matrix of the optimization layer, P out is the output matrix of the optimization layer, f1 is the specific optimization function, L in is the input matrix of the output layer, L out is the output matrix after applying the function, C T and W T Represent the connection weights of the input layer and the output layer respectively; The fitness function is Where j represents the inverter fault state and is a positive integer, i represents the iteration index, and y ij.des is the expected output value, y ij.out is the actual output value; 4) Early warning and visualization of faults: Early warning of line faults is performed based on the model output results, and visualization operations are performed after data rendering; The PV string IV characteristic database in step 1) includes the PV string IV characteristic data in a normal state and the PV string IV characteristic data in a fault state; The visualization operation in step 4) can display the fault location, fault type and fault parameters on the display device.

2. The photovoltaic inverter intelligent string diagnosis method according to claim 1, characterized in that: In step 3), formulas (11)-(12) are also used for optimization. Where w is the inertia weight factor, t is the number of iterations, c1 and c2 are system and environment parameters, r1 and r2 are random numbers with values ​​between 0 and 1, represents the optimal position of the i-th particle in the current iteration, represents the best historical position of the i-th particle, represents the current position of the i-th particle in the current iteration, represents the next iteration position of the ith particle, represents the current velocity of the ith particle, Represents the next iteration rate of the i-th particle.

3. A photovoltaic inverter intelligent string diagnostic system, comprising: 1) Data acquisition component: The photovoltaic system includes a plurality of photovoltaic strings, each of which is connected to a string inverter, and the current value and voltage value of each photovoltaic string are collected in real time to form a photovoltaic string IV characteristic database; 2) Data preprocessing component: construct the data matrix X and normalize the data matrix: the data matrix X represents the fault parameters, which is expressed as formula (1), X=[V oc I sc P m V m ] (1) Among them, V oc Indicates the open circuit voltage, I sc Indicates a short circuit, I m Indicates the maximum power current, V m represents the maximum power voltage, let Among them, x ij is the initialization output matrix, z ij is the normalized input, x min is the minimum value of each row of matrix X, x max is the maximum value of each row of matrix X, y min is the minimum value of each row in the normalized matrix, y max is the maximum value of each row in the normalized matrix; 3) Model training component: Use the following functions to train the deep learning model. Linear(x)=kx+c (4) Among them, α, k, and c all represent characteristic parameters generated during the function iteration process; The output matrix is ​​constructed as: Z out =sig(Z in C T ) (5) P in =Z out (6) P out =f1(P in ) (7) L in =P out (8) L out =Linear(L in W T ) (9) Among them, Z in is the normalized matrix of the input, Z out is the output matrix after applying the function, P in is the input matrix of the optimization layer, P out is the output matrix of the optimization layer, f1 is the specific optimization function, L in is the input matrix of the output layer, L out is the output matrix after applying the function, C T and W T Represent the connection weights of the input layer and the output layer respectively; The fitness function is Where j represents the inverter fault state and is a positive integer, i represents the iteration index, and y ij.des is the expected output value, y ij.out is the actual output value; 4) Early warning and visualization component: Based on the model output results, early warning of line faults is issued, and visualization operations are performed after data rendering; The PV string IV characteristic database described in the data acquisition component includes the PV string IV characteristic data in a normal state and the PV string IV characteristic data in a fault state; Visual operation can display the fault location, fault type and fault parameters on the display device.

4. The photovoltaic inverter intelligent string diagnostic system according to claim 3, characterized in that: The model training component is also optimized using formulas (11)-(12). Where w is the inertia weight factor, t is the number of iterations, c1 and c2 are system and environment parameters, r1 and r2 are random numbers with values ​​between 0 and 1, represents the optimal position of the i-th particle in the current iteration, represents the best historical position of the i-th particle, represents the current position of the i-th particle in the current iteration, represents the next iteration position of the ith particle, represents the current velocity of the ith particle, Represents the next iteration rate of the i-th particle.

Images