A photovoltaic module power loss calculation method and system

Abstract

The present application provides a photovoltaic module power loss calculation method and system thereof, which relate to the field of photovoltaic power generation technology, including: obtaining historical infrared images and current infrared images of photovoltaic modules; comparing the historical infrared images with the current infrared images to determine whether the state change of the photovoltaic modules is greater than a state change threshold; if the state change is greater than the state change threshold and the duration exceeds a preset duration threshold, predicting a first maximum output power point sequence based on the current infrared image; obtaining an actual output power curve of the photovoltaic modules within a response time; determining the power loss of the photovoltaic modules within the response time based on the first maximum output power point sequence and the actual output power curve; the present application determines whether the state of the photovoltaic modules changes through infrared images, and when the state of the photovoltaic modules changes, determines the power loss of the photovoltaic modules from the change to the time when the maximum output power is entered again, thereby improving the accuracy of the calculation.

Description

Technical Field

[0001] The present application relates to the field of photovoltaic power generation technology, and in particular to a photovoltaic module power loss calculation method and system. Background Art

[0002] With the increase in the installed capacity of photovoltaic power generation, the requirements for photovoltaic power generation are getting higher and higher. The power loss of photovoltaic modules refers to the difference between the actual output power of the module and the ideal output power, which is an important factor affecting the efficiency of photovoltaic power generation. In order to improve the efficiency of photovoltaic power generation, it is crucial to accurately evaluate and analyze power loss. The existing calculation methods for the power loss of photovoltaic modules are mostly about the design power loss of photovoltaic modules in the design stage and the fault power loss of photovoltaic modules caused by faults. There is no calculation method for the power loss in the process of MPPT entering the optimization stage and entering the maximum power output state again due to the change of the state of photovoltaic modules. However, with the investment of deep learning models in photovoltaic power generation technology, there is an urgent need for a method to continuously train the model and determine the power loss between the two in the process of adjusting photovoltaic modules through MPPT, so as to provide subsequent analysis and use.

[0003] In view of this, the present application provides a method and system for calculating the power loss of a photovoltaic module, which determines whether the state of the photovoltaic module has changed through infrared images, and promptly determines the first maximum output power point and obtains the actual output power curve when the change in the state of the photovoltaic module will cause the MPPT to be adjusted, so as to obtain the power loss of the photovoltaic module after the change to the time when it re-enters the maximum output power, thereby improving the accuracy of the calculation. Summary of the invention

[0004] An embodiment of the present application provides a method for calculating the power loss of a photovoltaic component, including: acquiring a historical infrared image and a current infrared image of a photovoltaic component; comparing the historical infrared image with the current infrared image to determine whether a state change of the photovoltaic component is greater than a state change threshold; if the state change is greater than the state change threshold and the duration exceeds a preset duration threshold, predicting a first maximum output power point sequence according to the current infrared image; the first maximum output power point refers to the maximum output power point of the photovoltaic component predicted according to the current infrared image; the first maximum output power point sequence refers to one or more of the first maximum output power points predicted in the process of the MPPT adjusting the photovoltaic component to the second maximum output power point after the state of the photovoltaic component changes; acquiring an actual output power curve of the photovoltaic component within a response time; and determining the power loss of the photovoltaic component within the response time based on the first maximum output power point sequence and the actual output power curve.

[0005] Furthermore, determining whether the state change of the photovoltaic component is greater than the state change threshold includes: respectively obtaining historical environmental information and current environmental information of the historical infrared image and the current infrared image; the environmental information includes temperature and light intensity; performing temperature calibration and geometric correction on the infrared image based on the environmental information to obtain a historical infrared processed image and a current infrared processed image; performing image recognition on key areas within the infrared processed image, and marking the key areas to obtain a historical marked infrared image and a current marked infrared image; the key areas include battery cells and junction boxes; processing each battery cell in the marked infrared image to obtain historical features and current features of the photovoltaic component; the features are at least related to average temperature, temperature distribution, hot spot location, temperature change, temperature gradient and uniformity; when the difference between the historical features and the current features is greater than the state change threshold, determining that the state change of the photovoltaic component is greater than the state change threshold.

[0006] Furthermore, the difference between the historical feature and the current feature is related to the power of the maximum power point, and the calculation formula for predicting the maximum power point is:

[0007] ;

[0008] ;

[0009] in, represents the power at the calculated maximum power point; Represents the first parameter; Indicates short-circuit current; Indicates the total open circuit voltage of the battery cell; Represents the second parameter; Indicates the difference between the temperature under standard test conditions and the temperature of the current environment; Represents the third parameter; Indicates the average temperature of the cells in the current photovoltaic module; represents the fourth parameter; Indicates the maximum area of ​​the solar cell in the current photovoltaic module that is blocked; i represents the solar cell variable; n represents the total number of solar cells; represents the weight of the i-th battery cell; Indicates the area of ​​the ith cell that is blocked; Indicates the open circuit voltage under standard test conditions; represents the fifth parameter; represents the average temperature of the i-th cell; represents the sixth parameter; represents the light intensity of the i-th cell; Indicates the light intensity under standard test conditions.

[0010] Furthermore, predicting the first maximum output power point sequence includes: obtaining a first infrared image sequence and its environmental information within a first time period (all future time); the first time period is later than the current time; normalizing the first infrared image sequence to obtain a standard infrared image sequence; inputting the standard infrared image sequence into a feature extraction network to obtain an initial image feature sequence; inputting the initial image feature sequence into an attention network to obtain an attention image feature sequence; splicing the attention image feature sequence and the environmental information, and fusing the spliced ​​features to obtain a fused feature sequence; inputting the fused features into a multi-layer perceptron to obtain the first maximum output power point of the first time period.

[0011] Furthermore, the first time period is determined by the change frequency and change degree of the photovoltaic component, including: obtaining a second infrared image sequence within a second preset time period (historical time); the second preset time period is earlier than the current time; inputting the second infrared image sequence into a change recognition model to obtain a change value within the second preset time period; the change value is related to the change degree and change frequency of the photovoltaic component; comparing the change value to a preset change threshold, and updating a new first time period based on the comparison result and the current first time period.

[0012] Furthermore, the change recognition model is obtained by training the initial change recognition model: the initial change recognition model includes a convolutional neural network and a long short-term memory network; the long short-term memory network is used to process the image sequence output by the convolutional neural network to obtain the change value; the convolutional neural network includes multiple two-dimensional convolutional layers and multiple pooling layers; the pooling layer pools the image based on the degree of pixel change of the image, including a maximum pooling unit, an average pooling unit, a minimum pooling unit and a fusion unit; the second infrared image sequence is input into the first two-dimensional convolutional layer respectively to obtain a first convolutional image sequence; the first convolutional image sequence is input into the first pooling layer respectively, A first pooled image sequence is obtained; the first pooled image sequence is input into the second two-dimensional convolution layer respectively to obtain a second convolution image sequence; the second convolution image sequence is input into the second pooling layer respectively to obtain a second pooled image sequence; the second pooled image sequence is input into the third two-dimensional convolution layer respectively to obtain a third convolution image sequence; the long short-term memory network includes a first network layer and a second network layer; the first network layer processes the first convolution image sequence and the second convolution image sequence to obtain a first feature sequence; the second network layer processes the third convolution image sequence and the first feature sequence to obtain a change value of the second infrared image sequence.

[0013] Furthermore, the loss function for training the initial change recognition model is:

[0014] ;

[0015] ;

[0016] ;

[0017] in, represents the loss function of the initial change recognition model; T represents the total number of sub-time periods in the second preset time period; t represents the time period variable; and respectively represent the true values ​​of the degree of change of the second infrared image at time t and time t-1; and Respectively represent the predicted values ​​of the change degree of the second infrared image at time t and time t-1; represents the first variation parameter; represents the true value of the rate of change of the second infrared image at time t; represents the predicted value of the rate of change of the second infrared image at time t; represents the second variation parameter; represents the L2 norm; H represents the height of the second infrared image; W represents the width of the second infrared image; h represents the ordinate variable; w represents the abscissa variable; and The coordinates of the second infrared image at time t and t-1 are respectively The pixel value of the pixel point; Indicates the time interval for acquiring the second infrared image.

[0018] Furthermore, the feature extraction network, the attention network and the multi-layer perceptron are obtained through joint training, and the joint loss function of the joint training is:

[0019] ;

[0020] ;

[0021] in, represents the joint loss function; represents the maximum output power predicted by the model; Indicates the maximum output power adjusted by MPPT; represents the first joint weight parameter; represents the power at the calculated maximum power point; represents the second joint weight parameter; Indicates the calculation parameters of the calculation formula; represents the third joint weight parameter; i represents the battery cell variable; n represents the total number of battery cells; represents the weight of the i-th battery cell; Represents the first parameter; Represents the second parameter; Indicates the third parameter; represents the fourth parameter; represents the fifth parameter; Indicates the sixth parameter.

[0022] Further, determining the power loss includes: determining a first time period corresponding to a first maximum output power point; subtracting the actual output power curve based on the first time period and the first maximum output power point to obtain a difference curve; integrating the difference curve within the response time to obtain the power loss.

[0023] The embodiment of the present application also provides a photovoltaic component power loss calculation system, including an infrared image acquisition module, a comparison module, a prediction module, an actual output power acquisition module and a power loss determination module; the infrared image acquisition module is used to acquire historical infrared images and current infrared images of the photovoltaic component; the comparison module is used to compare the historical infrared image with the current infrared image to determine whether the state change of the photovoltaic component is greater than the state change threshold; the prediction module is used to predict a first maximum output power point sequence according to the current infrared image when the state change is greater than the state change threshold and the duration exceeds a preset duration threshold; the first maximum output power point refers to the maximum output power point of the photovoltaic component predicted according to the current infrared image; the first maximum output power point sequence refers to one or more of the first maximum output power points predicted in the process of the MPPT adjusting the photovoltaic component to the second maximum output power point after the state of the photovoltaic component changes; the actual output power acquisition module is used to acquire the actual output power curve of the photovoltaic component within the response time; the power loss determination module is used to determine the power loss of the photovoltaic component within the response time based on the first maximum output power point sequence and the actual output power curve.

[0024] The embodiments of the present application determine whether the state of a photovoltaic module has changed through infrared images, and in a case where the change in the state of the photovoltaic module will cause the MPPT to be adjusted, promptly determine the first maximum output power point and obtain the actual output power curve, so as to obtain the power loss after the photovoltaic module changes and when it re-enters the maximum output power, thereby improving the accuracy of the calculation.

[0025] The embodiments of the present application predict the maximum power point through a calculation formula, and identify the state of the photovoltaic module through the maximum power point predicted by the formula, thereby improving the efficiency of state identification and further improving the accuracy of the calculated power loss.

[0026] The embodiment of the present application can more accurately obtain multiple first maximum output power points within the response time by dynamically adjusting the first time period and obtaining the first maximum output power point in the first time period, thereby overcoming the problem of excessive difference between the final first maximum output power point and the second maximum output power point due to multiple state changes of the photovoltaic components during the MPPT adjustment process.

[0027] The embodiment of the present application obtains the change frequency and degree of the photovoltaic components within the second preset time period, updates the first time period, and makes the duration of the obtained first maximum output power point more accurate, thereby improving the accuracy of the subsequent calculated power loss.

[0028] The embodiments of the present application set a maximum pooling unit, an average pooling unit and a minimum pooling unit, and perform weighted fusion on the features output by the pooling units, so that the temperature gradient in the infrared image can be perceived, the data loss caused by single pooling can be avoided, and the accuracy of subsequent predictions can be improved.

[0029] The embodiment of the present application can simultaneously update the calculation formula for predicting the maximum power point by setting a joint loss function, so that the maximum power point calculated by the calculation formula is more accurate, thereby improving the accuracy of subsequent detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 An exemplary flow chart of a photovoltaic module power loss calculation method proposed by the present invention;

[0031] Figure 2 This is an exemplary module diagram of a photovoltaic module power loss calculation system proposed by the present invention. DETAILED DESCRIPTION

[0032] Figure 1 This is an exemplary flow chart of a photovoltaic module power loss calculation method proposed by the present invention. Figure 1 As shown, the process of the photovoltaic module power loss calculation method proposed in the present invention includes the following contents:

[0033] Step S1, obtaining a historical infrared image and a current infrared image of a photovoltaic module.

[0034] The historical infrared image refers to the infrared image of the photovoltaic module acquired in the past, for example, the infrared image acquired before the current time. The current infrared image refers to the infrared image of the photovoltaic module acquired at present.

[0035] Step S2, comparing the historical infrared image with the current infrared image to determine whether the state change of the photovoltaic component is greater than a state change threshold.

[0036] A state change refers to a change in the state of a photovoltaic module, which will cause the MPPT (Maximum power point tracking) to adjust the output power of the photovoltaic module to a new maximum output power point. The state change threshold refers to the threshold of the minimum change that causes the MPPT to enter the optimization phase due to the change in the state of the photovoltaic module. For example, the state change threshold can be determined by the ratio of the absolute value of the difference between the current output power and the current maximum output power to the current maximum output power.

[0037] Determining whether the state change of the photovoltaic component is greater than the state change threshold includes: respectively obtaining the historical environmental information and the current environmental information of the historical infrared image and the current infrared image; the environmental information includes temperature and light intensity. Environmental information refers to environmental information related to the operation of the photovoltaic component. For example, the environmental information may include temperature, light intensity, weather conditions, height dimension, etc. The historical environmental information refers to the environmental information when the historical infrared image is obtained. The current environmental information refers to the environmental information when the current infrared image is obtained. Temperature calibration and geometric correction are performed on the infrared image based on the environmental information to obtain a historical infrared processed image and a current infrared processed image. The value of the pixel point in the infrared image is calibrated by temperature. The distortion of the camera can be corrected by geometric correction. Image recognition is performed on the key area in the infrared processed image, and the key area is marked to obtain a historical marked infrared image and a current marked infrared image; the key area includes at least a cell and a junction box. Each cell in the marked infrared image is processed to obtain the historical characteristics and current characteristics of the photovoltaic component; the characteristics are at least related to the average temperature, temperature distribution, hot spot position, temperature change, temperature gradient and uniformity. When the difference between the historical feature and the current feature is greater than the state change threshold, it is determined that the state change of the photovoltaic component is greater than the state change threshold. The difference between the historical feature and the current feature is related to the power of the maximum power point. For example, when the absolute value of the difference between the current output power and the power of the maximum power point is greater than the ratio of the power of the maximum power point to 0.1, it can be considered that the state change of the photovoltaic component is greater than the state change threshold.

[0038] In order to improve the detection efficiency, the power of the maximum power point can be predicted by a faster calculation formula. The calculation formula for predicting the maximum power point is:

[0039] ;

[0040] ;

[0041] in, represents the power at the calculated maximum power point; Represents the first parameter; Indicates short-circuit current; Indicates the total open circuit voltage of the battery cell; Represents the second parameter; Indicates the difference between the temperature under standard test conditions and the temperature of the current environment; Represents the third parameter; Indicates the average temperature of the cells in the current photovoltaic module; represents the fourth parameter; Indicates the maximum area of ​​the solar cell in the current photovoltaic module that is blocked; i represents the solar cell variable; n represents the total number of solar cells; represents the weight of the i-th battery cell; Indicates the area of ​​the ith cell that is blocked; Indicates the open circuit voltage under standard test conditions; represents the fifth parameter; represents the average temperature of the i-th cell; represents the sixth parameter; represents the light intensity of the i-th cell; Indicates the light intensity under standard test conditions.

[0042] Step S3, if the state change is greater than the state change threshold and the duration exceeds the preset duration threshold, predicting the first maximum output power point sequence according to the current infrared image; the first maximum output power point refers to the maximum output power point of the photovoltaic component predicted according to the current infrared image; the first maximum output power point sequence refers to one or more of the first maximum output power points predicted in the process of the MPPT adjusting the photovoltaic component to the second maximum output power point after the state of the photovoltaic component changes.

[0043] Duration refers to the time during which the state change continues to be greater than the state change threshold. The preset duration threshold refers to the preset minimum state change duration. The second maximum output power point refers to the maximum power point tracked by the MPPT after it recognizes the change of the photovoltaic module and starts the optimization phase.

[0044] Predicting the first maximum output power point sequence includes: obtaining a first infrared image sequence and its environmental information within a first time period; the first time period is later than the current time. The first time period refers to the time period for predicting the first maximum output power point. Since the photovoltaic component may change multiple times during the process of adjusting the photovoltaic component to the second maximum output power point by MPPT, the photovoltaic component may change multiple times, but due to the response speed of the MPPT component, the final second maximum output power point is different from the photovoltaic component after the first state change. Therefore, in order to obtain more accurate power loss, the response time is adjusted to multiple first time periods, and the state of the photovoltaic component in each first time period is different, corresponding to different first maximum output power points. The first infrared image sequence refers to the infrared image of the photovoltaic component obtained within the first time period. The first infrared image sequence is normalized to obtain a standard infrared image sequence. The standard infrared image sequence is input into a feature extraction network to obtain an initial image feature sequence. The initial image feature sequence is input into an attention network to obtain an attention image feature sequence. The attention image feature sequence and the environmental information are spliced, and the spliced ​​features are fused to obtain a fused feature sequence. The fused features are input into a multi-layer perceptron to obtain the first maximum output power point of the first time period. The feature extraction network, the attention network and the multi-layer perceptron are obtained through joint training, and the joint loss function of the joint training is:

[0045] ;

[0046] ;

[0047] in, represents the joint loss function; represents the maximum output power predicted by the model; Indicates the maximum output power adjusted by MPPT; represents the first joint weight parameter; represents the power at the calculated maximum power point; represents the second joint weight parameter; Indicates the calculation parameters of the calculation formula; represents the third joint weight parameter; i represents the battery cell variable; n represents the total number of battery cells; represents the weight of the i-th battery cell; Represents the first parameter; Represents the second parameter; Represents the third parameter; represents the fourth parameter; represents the fifth parameter; Indicates the sixth parameter.

[0048] In some embodiments, the first time period is determined by the change frequency and change degree of the photovoltaic component, including: obtaining a second infrared image sequence within a second preset time period; the second preset time period is earlier than the current time. The second infrared image sequence refers to the infrared image of the photovoltaic component within the second preset time period from the current time. The second infrared image sequence is input into the change recognition model to obtain the change value within the second preset time period; the change value is related to the change degree and change frequency of the photovoltaic component. The change degree is related to the severity of the temperature change of the photovoltaic component. The change frequency is related to the number of changes of the photovoltaic component. The change value is compared to the preset change threshold, and based on the comparison result and the current first time period, a new first time period is updated. The preset change threshold refers to the minimum change value of the photovoltaic component in the second preset time period designed in advance. When the change value exceeds the preset change threshold, it means that the change of the photovoltaic component is frequent and drastic, so the current first time period is reduced to increase the detection frequency. When the change value is less than the preset change threshold, it means that the change of the photovoltaic component is relatively gentle, so the current first time period is increased to reduce the detection frequency and save resources. Of course, the preset change threshold here can be a range, and when the change value is within this range, the length of the first time period remains unchanged.

[0049] The change recognition model is obtained by training the initial change recognition model: the initial change recognition model includes a convolutional neural network and a long short-term memory network; the long short-term memory network is used to process the image sequence output by the convolutional neural network to obtain the change value. The convolutional neural network includes multiple two-dimensional convolutional layers and multiple pooling layers; the pooling layer pools the image based on the pixel change degree of the image, including a maximum pooling unit, an average pooling unit, a minimum pooling unit and a fusion unit. The second infrared image sequence is respectively input into the first two-dimensional convolutional layer to obtain a first convolutional image sequence. The first convolutional image sequence is respectively input into the first pooling layer to obtain a first pooled image sequence. The first pooled image sequence is respectively input into the second two-dimensional convolutional layer to obtain a second convolutional image sequence. The second convolutional image sequence is respectively input into the second pooling layer to obtain a second pooled image sequence. The second pooled image sequence is respectively input into the third two-dimensional convolutional layer to obtain a third convolutional image sequence. The long short-term memory network includes a first network layer and a second network layer. The first network layer processes the first convolutional image sequence and the second convolutional image sequence to obtain a first feature sequence. The second network layer processes the third convolution image sequence and the first feature sequence to obtain a change value of the second infrared image sequence. The loss function for training the initial change recognition model is:

[0050] ;

[0051] ;

[0052] ;

[0053] in, represents the loss function of the initial change recognition model; T represents the total number of sub-time periods in the second preset time period; t represents the time period variable; and respectively represent the true values ​​of the degree of change of the second infrared image at time t and time t-1; and Respectively represent the predicted values ​​of the change degree of the second infrared image at time t and time t-1; represents the first variation parameter; represents the true value of the rate of change of the second infrared image at time t; represents the predicted value of the rate of change of the second infrared image at time t; represents the second variation parameter; represents the L2 norm; H represents the height of the second infrared image; W represents the width of the second infrared image; h represents the ordinate variable; w represents the abscissa variable; and The coordinates of the second infrared image at time t and t-1 are respectively The pixel value of the pixel point; Indicates the time interval for acquiring the second infrared image.

[0054] Step S4, obtaining the actual output power curve of the photovoltaic module within the response time.

[0055] The response time refers to the time period from when the state of the photovoltaic module changes to when the MPPT adjusts the photovoltaic module to the second maximum output power point. The actual output power curve refers to the curve composed of the output power of the photovoltaic module actually obtained by the sensor device. Obtaining the second maximum output power point includes: collecting the voltage and current data output by the photovoltaic module; constructing a dynamic IV curve based on the voltage data and current data; calculating the actual output power based on the dynamic IV curve to obtain the actual output power curve.

[0056] Step S5: determining the power loss of the photovoltaic assembly within the response time based on the first maximum output power point sequence and the actual output power curve.

[0057] Power loss refers to the reduction in output power due to the delay in MPPT optimization. Determining the power loss includes: determining a first time period corresponding to a first maximum output power point; subtracting the actual output power curve based on the first time period and the first maximum output power point to obtain a difference curve; integrating the difference curve within the response time to obtain the power loss.

[0058] Figure 2 This is an exemplary module diagram of a photovoltaic module power loss calculation system proposed by the present invention. Figure 2 As shown, the photovoltaic module power loss calculation system proposed in the present invention includes an infrared image acquisition module M1, a comparison module M2, a prediction module M3, an actual output power acquisition module M4 and a power loss determination module M5.

[0059] The infrared image acquisition module M1 is used to acquire the historical infrared image and the current infrared image of the photovoltaic module.

[0060] The comparison module M2 is used to compare the historical infrared image with the current infrared image to determine whether the state change of the photovoltaic component is greater than the state change threshold.

[0061] The prediction module M3 is used to predict the first maximum output power point sequence according to the current infrared image when the state change is greater than the state change threshold and the duration exceeds the preset duration threshold; the first maximum output power point refers to the maximum output power point of the photovoltaic component predicted according to the current infrared image; the first maximum output power point sequence refers to one or more of the first maximum output power points predicted in the process of MPPT adjusting the photovoltaic component to the second maximum output power point after the state of the photovoltaic component changes.

[0062] The actual output power acquisition module M4 is used to obtain the actual output power curve of the photovoltaic module within the response time.

[0063] The power loss determination module M5 is used to determine the power loss of the photovoltaic component within the response time based on the first maximum output power point sequence and the actual output power curve.

Claims

1. A method for calculating power loss of a photovoltaic module, characterized in that: include: Obtain historical infrared images and current infrared images of photovoltaic modules; Comparing the historical infrared image with the current infrared image to determine whether a state change of the photovoltaic component is greater than a state change threshold; If the state change is greater than the state change threshold and the duration exceeds the preset duration threshold, a first maximum output power point sequence is predicted based on the current infrared image; the first maximum output power point refers to the maximum output power point of the photovoltaic component predicted based on the current infrared image; the first maximum output power point sequence refers to one or more of the first maximum output power points predicted during the process of the MPPT adjusting the photovoltaic component to the second maximum output power point after the state of the photovoltaic component changes; Obtain the actual output power curve of the photovoltaic module within the response time; The power loss of the photovoltaic component within the response time is determined based on the first maximum output power point sequence and the actual output power curve.

2. The photovoltaic module power loss calculation method according to claim 1, characterized in that: Determining whether the state change of the PV module is greater than the state change threshold includes: Respectively acquiring historical environmental information and current environmental information of the historical infrared image and the current infrared image; the environmental information includes temperature and light intensity; Based on the environmental information, the infrared image is temperature calibrated and geometrically corrected to obtain the historical infrared processed image and the current infrared processed image; Performing image recognition on key areas in the infrared processed image and marking the key areas to obtain a historical marking infrared image and a current marking infrared image; the key areas include the battery cell and the junction box; Each cell in the identified infrared image is processed to obtain the historical and current characteristics of the photovoltaic module; the characteristics are related to the average temperature, temperature distribution, hot spot location, temperature change, temperature gradient and uniformity; When the difference between the historical feature and the current feature is greater than the state change threshold, it is determined that the state change of the photovoltaic component is greater than the state change threshold.

3. The photovoltaic module power loss calculation method according to claim 2, characterized in that: The difference between the historical feature and the current feature is related to the power of the maximum power point, and the calculation formula for predicting the maximum power point is: ； ; in, represents the power at the calculated maximum power point; Represents the first parameter; Indicates short-circuit current; Indicates the total open circuit voltage of the battery cell; Represents the second parameter; Indicates the difference between the temperature under standard test conditions and the temperature of the current environment; Represents the third parameter; Indicates the average temperature of the cells in the current photovoltaic module; represents the fourth parameter; Indicates the maximum area of ​​the solar cell in the current photovoltaic module that is blocked; i represents the solar cell variable; n represents the total number of solar cells; represents the weight of the i-th battery cell; Indicates the area of ​​the ith cell that is blocked; Indicates the open circuit voltage under standard test conditions; represents the fifth parameter; represents the average temperature of the i-th cell; represents the sixth parameter; represents the light intensity of the i-th cell; Indicates the light intensity under standard test conditions.

4. The photovoltaic module power loss calculation method according to claim 1, characterized in that: Predicting the first maximum output power point sequence, including: Acquire a first infrared image sequence and its environmental information within a first time period; the first time period is later than the current time; performing normalization processing on the first infrared image sequence to obtain a standard infrared image sequence; Inputting the standard infrared image sequence into a feature extraction network to obtain an initial image feature sequence; Inputting the initial image feature sequence into the attention network to obtain an attention image feature sequence; splicing the attention image feature sequence and the environmental information, and fusing the spliced ​​features to obtain a fused feature sequence; The fusion feature is input into a multi-layer perceptron to obtain a first maximum output power point in the first time period.

5. The photovoltaic module power loss calculation method according to claim 4, characterized in that: The first time period is determined by the change frequency and change degree of the photovoltaic components, including: Acquire a second infrared image sequence within a second preset time period; the second preset time period is earlier than the current time; Inputting the second infrared image sequence into a change recognition model to obtain a change value within the second preset time period; the change value is related to the change degree and change frequency of the photovoltaic module; The change value is compared to a preset change threshold, and based on the comparison result and the current first time period, a new first time period is updated.

6. The photovoltaic module power loss calculation method according to claim 5, characterized in that: The change recognition model is obtained by training the initial change recognition model: The initial change recognition model includes a convolutional neural network and a long short-term memory network; the long short-term memory network is used to process the image sequence output by the convolutional neural network to obtain the change value; The convolutional neural network includes multiple two-dimensional convolutional layers and multiple pooling layers; the pooling layer pools the image based on the degree of pixel change of the image, including a maximum pooling unit, an average pooling unit, a minimum pooling unit and a fusion unit; Inputting the second infrared image sequence into the first two-dimensional convolution layer respectively to obtain a first convolution image sequence; Inputting the first convolution image sequence into the first pooling layer respectively to obtain a first pooling image sequence; Inputting the first pooled image sequence into the second two-dimensional convolution layer respectively to obtain a second convolution image sequence; Inputting the second convolution image sequence into the second pooling layer respectively to obtain a second pooled image sequence; Inputting the second pooled image sequence into the third two-dimensional convolution layer respectively to obtain a third convolution image sequence; The long short-term memory network includes a first network layer and a second network layer; The first network layer processes the first convolution image sequence and the second convolution image sequence to obtain a first feature sequence; The second network layer processes the third convolution image sequence and the first feature sequence to obtain a change value of the second infrared image sequence.

7. The method for calculating photovoltaic module power loss according to claim 6, characterized in that: The loss function for training the initial change recognition model is: ; ; ; in, represents the loss function of the initial change recognition model; T represents the total number of sub-time periods in the second preset time period; t represents the time period variable; and respectively represent the true values ​​of the degree of change of the second infrared image at time t and time t-1; and Respectively represent the predicted values ​​of the change degree of the second infrared image at time t and time t-1; represents the first variation parameter; represents the true value of the rate of change of the second infrared image at time t; represents the predicted value of the rate of change of the second infrared image at time t; represents the second variation parameter; represents the L2 norm; H represents the height of the second infrared image; W represents the width of the second infrared image; h represents the ordinate variable; w represents the abscissa variable; and The coordinates of the second infrared image at time t and t-1 are respectively The pixel value of the pixel point; Indicates the time interval for acquiring the second infrared image.

8. The photovoltaic module power loss calculation method according to claim 4, characterized in that: The feature extraction network, the attention network and the multi-layer perceptron are obtained through joint training, and the joint loss function of the joint training is: ; ; in, represents the joint loss function; represents the maximum output power predicted by the model; Indicates the maximum output power adjusted by MPPT; represents the first joint weight parameter; represents the power at the calculated maximum power point; represents the second joint weight parameter; Indicates the calculation parameters of the calculation formula; represents the third joint weight parameter; i represents the battery cell variable; n represents the total number of battery cells; represents the weight of the i-th battery cell; Represents the first parameter; Represents the second parameter; Indicates the third parameter; represents the fourth parameter; represents the fifth parameter; Indicates the sixth parameter.

9. The photovoltaic module power loss calculation method according to claim 1, characterized in that: Determining the power loss, comprising: Determine a first time period corresponding to a first maximum output power point; Subtracting the actual output power curve based on the first time period and the first maximum output power point to obtain a difference curve; The difference curve within the response time is integrated to obtain the power loss.

10. A photovoltaic module power loss calculation system, characterized in that: It includes an infrared image acquisition module, a comparison module, a prediction module, an actual output power acquisition module and a power loss determination module; The infrared image acquisition module is used to acquire historical infrared images and current infrared images of photovoltaic modules; The comparison module is used to compare the historical infrared image with the current infrared image to determine whether the state change of the photovoltaic component is greater than the state change threshold; The prediction module is used to predict the first maximum output power point sequence according to the current infrared image when the state change is greater than the state change threshold and the duration exceeds the preset duration threshold; the first maximum output power point refers to the maximum output power point of the photovoltaic component predicted according to the current infrared image; the first maximum output power point sequence refers to one or more of the first maximum output power points predicted in the process of the MPPT adjusting the photovoltaic component to the second maximum output power point after the state of the photovoltaic component changes; The actual output power acquisition module is used to acquire the actual output power curve of the photovoltaic module within the response time; The power loss determination module is used to determine the power loss of the photovoltaic component within the response time based on the first maximum output power point sequence and the actual output power curve.

Images