Solar power generation integrated management system, device and method based on AI intelligence

Abstract

The invention discloses a solar power generation integrated management system, equipment and method based on AI intelligence, and particularly relates to the technical field of photovoltaic power generation. By analyzing the light-induced attenuation data of the photovoltaic module, judging whether the light-induced attenuation of the photovoltaic module has reversibility or not by using a convolutional neural network; based on the light intensity change data of the cloud layer edge light intensity abrupt change area, the influence degree of the cloud edge nonlinear shielding and reflection gain effect on the photovoltaic array power fluctuation is evaluated; and determining the working state of the photovoltaic module according to the judgment result of whether the light-induced attenuation is reversible and the evaluation result of the nonlinear shielding and reflection gain effect. When the assembly is in an abnormal working state, the non-linear attenuation influence degree of the aerosol retention layer on the light transmittance is evaluated by analyzing the light transmittance change. And whether an operation and maintenance scheme is started or not is judged according to the evaluation result, so that intelligent diagnosis and fine management of the operation state of the photovoltaic module are realized, and the operation efficiency and reliability of solar power generation are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power generation, and more specifically, to a solar power generation integrated management system, device and method based on AI intelligence. Background Art

[0002] Traditional photovoltaic systems usually only rely on monitoring power generation and meteorological data, and it is difficult to accurately identify the impact of light-induced attenuation of components and cloud occlusion on power fluctuations. It is difficult to ensure the timeliness and scientificity of operation and maintenance decisions. In addition, the aerosol retention layer will cause a decrease in light transmittance and exacerbate the attenuation of power generation efficiency, which is often overlooked. Existing management models often lack the ability of multi-dimensional data mining and are difficult to conduct in-depth comprehensive analysis on the factors affecting component performance attenuation, resulting in difficulties in timely handling of faults.

[0003] In order to solve the above problems, a solar power generation integrated management system, device and method based on AI intelligence are provided. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a solar power generation integrated management system, device and method based on AI intelligence to solve the problems raised in the above background art.

[0005] To achieve the above object, the present invention provides the following technical solutions: A solar power generation integrated management method based on AI intelligence, comprising the following steps: By analyzing the light-induced attenuation data of photovoltaic components, judging whether the light-induced attenuation of photovoltaic components is reversible according to a convolutional neural network; Based on the light intensity change data in the light intensity mutation area at the cloud edge, evaluating the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array; According to the judgment result of whether the light-induced attenuation of the photovoltaic component is reversible and the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array, determining the working state of the photovoltaic component; the working state of the photovoltaic component includes a normal working state and an abnormal working state; When the working state of the photovoltaic component is an abnormal working state, by analyzing the change in the light transmittance on the surface of the photovoltaic component, evaluating the influence degree of the non-linear attenuation of the aerosol retention layer on the light transmittance of the photovoltaic component; Conducting comprehensive analysis on the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array and the influence degree of the non-linear attenuation of the aerosol retention layer on the light transmittance of the photovoltaic component, and judging whether to enable the operation and maintenance plan.

[0006] In a preferred embodiment, by analyzing the light-induced attenuation data of a photovoltaic module, it is determined whether the light-induced attenuation of the photovoltaic module is reversible according to a convolutional neural network. Specifically: The light-induced attenuation data of the photovoltaic module are regularly collected by a high-precision sensor; The collected light-induced attenuation data are subjected to normalization processing; A convolutional neural network is used to extract key features from the light-induced attenuation data and identify the spatio-temporal patterns during the attenuation process of the photovoltaic module; The convolutional neural network is trained using a labeled light-induced attenuation data set; The trained convolutional neural network model is used to determine whether the light-induced attenuation of the photovoltaic module is reversible.

[0007] In a preferred embodiment, it is determined whether the light-induced attenuation of the photovoltaic module is reversible. Specifically: A reversible threshold is preset, and the probability value of the light-induced attenuation being reversible is compared with the reversible threshold: When the probability value of the light-induced attenuation being reversible is greater than the reversible threshold, it is determined that the light-induced attenuation of the photovoltaic module is reversible; When the probability value of the light-induced attenuation being reversible is less than or equal to the reversible threshold, it is determined that the light-induced attenuation of the photovoltaic module is irreversible.

[0008] In a preferred embodiment, based on the light intensity change data in the light intensity mutation region at the cloud edge, the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array is evaluated. Specifically: The light intensity change data in the cloud edge region are collected by a high-precision sensor; The light intensity data are analyzed to identify the light intensity mutation region and determine the distribution characteristics of the cloud edge; A mathematical model is proposed to describe the influence of the non-linear occlusion effect caused by the cloud edge on the light intensity; Based on the optical reflection principle, a reflection gain model is constructed to simulate the reflection gain effect of the light intensity; Combined with the light intensity change model, a mathematical model of the power fluctuation of the photovoltaic array is established; According to the modeling results, the influence degree of the non-linear occlusion and reflection gain effects on the power fluctuation is quantified.

[0009] In a preferred embodiment, according to the judgment result of whether the light-induced attenuation of the photovoltaic module is reversible and the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array, the working state of the photovoltaic module is determined. The working state of the photovoltaic module includes a normal working state and an abnormal working state. Specifically: A volatility coefficient threshold is preset, and the volatility coefficient is compared with the volatility coefficient threshold: When the volatility coefficient is greater than the volatility coefficient threshold, it indicates that the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array is high; When the volatility coefficient is less than or equal to the volatility coefficient threshold, it indicates that the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array is low; When the probability value of photoinduced degradation being reversible is greater than the reversibility threshold and the volatility coefficient is less than or equal to the volatility coefficient threshold, determine the working state of the photovoltaic module as the normal working state; otherwise, determine the working state of the photovoltaic module as the abnormal working state.

[0010] In a preferred embodiment, by analyzing the change in the light transmittance of the surface of the photovoltaic module, evaluate the degree of non-linear attenuation of the aerosol retention layer on the light transmittance of the photovoltaic module. Specifically: Collect the light transmittance data and the aerosol accumulation situation on the surface of the photovoltaic module; Through meteorological sensors and image recognition technology, monitor the aerosol retention layer; Based on the collected light transmittance data and aerosol retention layer data, use the regression analysis method to establish a non-linear regression model between the light transmittance and the aerosol accumulation; Establish a light transmittance attenuation model to evaluate the degree of non-linear attenuation of the aerosol retention layer on the light transmittance of the photovoltaic module: The calculation formula of the attenuation coefficient is: ; where is the attenuation coefficient; is the light transmittance attenuation value at the th light transmittance sampling point; is the weight coefficient at the th light transmittance sampling point; is the number of light transmittance sampling points.

[0011] In a preferred embodiment, comprehensively analyze the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array and the influence degree of the non-linear attenuation of the aerosol retention layer on the light transmittance of the photovoltaic module, and judge whether to enable the operation and maintenance plan. Specifically: Calculate the normalized volatility coefficient and attenuation coefficient to obtain the health score of the photovoltaic module. The calculation formula is: ; where is the health score of the photovoltaic module; is the attenuation coefficient; is the volatility coefficient; is the weight factor of the volatility coefficient; is the weight factor of the attenuation coefficient; is the smoothing factor; Preset a health score threshold and compare the health score with the health score threshold: When the health score is greater than or equal to the health score threshold, there is no need to enable the operation and maintenance plan; When the health score is less than the health score threshold, it is necessary to enable the operation and maintenance plan.

[0012] On the other hand, the present invention provides a comprehensive management system for solar power generation based on AI intelligence, including an attenuation reversibility judgment module, a power fluctuation evaluation module, a working state determination module, a light transmittance analysis module, and a comprehensive analysis module; The attenuation reversibility judgment module analyzes the light-induced attenuation data of the photovoltaic module and judges whether the light-induced attenuation of the photovoltaic module is reversible according to the convolutional neural network; The power fluctuation evaluation module evaluates the influence degree of the non-linear occlusion and reflection gain effects of the cloud edge on the power fluctuation of the photovoltaic array based on the light intensity change data of the light intensity mutation area at the cloud edge; The working state determination module determines the working state of the photovoltaic module according to the judgment result of whether the light-induced attenuation of the photovoltaic module is reversible and the influence degree of the non-linear occlusion and reflection gain effects of the cloud edge on the power fluctuation of the photovoltaic array; the working state of the photovoltaic module includes a normal working state and an abnormal working state; When the working state of the photovoltaic module is an abnormal working state, the light transmittance analysis module evaluates the influence degree of the non-linear attenuation of the aerosol retention layer on the light transmittance of the photovoltaic module by analyzing the change of the light transmittance on the surface of the photovoltaic module; The comprehensive analysis module comprehensively analyzes the influence degree of the non-linear occlusion and reflection gain effects of the cloud edge on the power fluctuation of the photovoltaic array and the influence degree of the non-linear attenuation of the aerosol retention layer on the light transmittance of the photovoltaic module, and judges whether to enable the operation and maintenance plan.

[0013] On the other hand, the present invention provides a comprehensive management device for solar power generation based on AI intelligence, including: a processor, a memory, and a program or instruction stored on the memory and executable on the processor. When the program or instruction is executed by the processor, a comprehensive management method for solar power generation based on AI intelligence is implemented.

[0014] Technical effects and advantages of a comprehensive management system, device, and method for solar power generation based on AI intelligence according to the present invention: By analyzing the light-induced attenuation data of photovoltaic modules through a convolutional neural network, it is possible to quickly determine whether the attenuation of photovoltaic modules is reversible, providing a scientific basis for the maintenance of photovoltaic modules. Secondly, by evaluating the impact of the non-linear occlusion and reflection gain effects at the cloud edge on power fluctuations, predicting the output fluctuation characteristics of photovoltaic modules in advance helps to ensure power generation stability. In addition, when the photovoltaic module is in an abnormal working state, further analyze the change in transmittance to evaluate the non-linear attenuation effect of the aerosol retention layer, revealing the long-term impact of environmental pollution on the performance of photovoltaic modules. Finally, by comprehensively analyzing and judging whether to enable the operation and maintenance plan, the efficient and intelligent operation and maintenance of photovoltaic modules are realized, significantly improving power generation efficiency and stability, while reducing maintenance costs, extending the service life of photovoltaic modules, and improving overall economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 Schematic diagram of a comprehensive solar power generation management method based on AI intelligence according to the present invention; Figure 2 Schematic diagram of the structure of a comprehensive solar power generation management system based on AI intelligence according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention. Embodiment

[0017] Figure 1 A comprehensive solar power generation management method based on AI intelligence according to the present invention is given, which includes the following steps: By analyzing the light-induced attenuation data of the photovoltaic module, it is judged whether the light-induced attenuation of the photovoltaic module is reversible according to the convolutional neural network; Based on the light intensity change data in the light intensity mutation area at the cloud edge, the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array is evaluated; According to the judgment result of whether the light-induced attenuation of the photovoltaic module is reversible and the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array, the working state of the photovoltaic module is determined; the working state of the photovoltaic module includes a normal working state and an abnormal working state; When the working state of the photovoltaic module is an abnormal working state, by analyzing the change in the transmittance of the photovoltaic module surface, the influence degree of the non-linear attenuation of the aerosol retention layer on the transmittance of the photovoltaic module is evaluated; Comprehensively analyze the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array and the influence degree of the non-linear attenuation of the aerosol retention layer on the light transmittance of the photovoltaic module, and determine whether to enable the operation and maintenance plan.

[0018] Specifically, by analyzing the light-induced attenuation data of the photovoltaic module, judge whether the light-induced attenuation of the photovoltaic module is reversible according to the convolutional neural network, including: Regularly collect the light-induced attenuation data of the photovoltaic module through high-precision sensors: Light-induced attenuation is the phenomenon that the performance of the photovoltaic module decays over time under long-term illumination conditions. To obtain high-quality data, the sensor should have high time resolution and be able to collect the attenuation data of the photovoltaic module in different environments at an appropriate frequency. The collected data should include but not be limited to information such as the operating temperature and irradiance of the photovoltaic module. The sensors should be configured at various positions of the photovoltaic array to monitor the attenuation of each photovoltaic module and ensure comprehensive data capture of the light-induced attenuation. The data collection frequency and range should be adjusted according to actual needs, usually multiple data collections are carried out at multiple time periods to reflect the change trend of the light-induced attenuation.

[0019] Perform standardization processing on the collected light-induced attenuation data: The collected light-induced attenuation data needs to be standardized to eliminate the influence of dimension differences and environmental factors. The standardization process includes: Remove outliers: Eliminate or correct abnormal data caused by sensor failures or external interferences; Data normalization: Map the data to a unified scale; Time series smoothing: Use moving average or Gaussian smoothing methods to reduce short-term noise and highlight long-term trends; The standardized data provides a reliable input for feature extraction.

[0020] Use a convolutional neural network to extract key features from the light-induced attenuation data and identify the spatio-temporal patterns in the attenuation process of the photovoltaic module: The convolutional neural network can learn local and global feature patterns from the original data and reveal the changing laws in the attenuation process of the photovoltaic module. The input data is the standardized light-induced attenuation data, usually represented in the form of a time series.

[0021] The first convolutional layer: In the convolutional operation, the convolutional kernel extracts local features from the input data, such as the local change trend of power decline. The formula for the convolutional operation is: ; where, is the feature map output by the convolutional layer, and the eigenvalue at time ; is the th weight coefficient in the convolutional kernel; represents the input data at time Value; Indicates the size of the convolutional kernel.

[0022] In the convolution operation, the input data is multiplied point - by - point with the convolutional kernel and accumulated to obtain the output feature values over time. The convolutional layer performs parallel operations with multiple convolutional kernels to extract multiple different feature patterns from the data.

[0023] Pooling layer: Reduces the dimensionality of the features obtained from the convolution operation, reduces the computational amount, and enhances the generalization ability of the model. The commonly used pooling method is max - pooling.

[0024] Fully - connected layer: Transfers the features extracted from the convolution and pooling to the fully - connected layer for further feature fusion and decision - making. Through the adjustment of weights and biases between layers, the network can learn the global patterns in the data.

[0025] After being processed by the convolutional neural network, a set of key features representing the photo - induced attenuation process is obtained, which can provide a decision - making basis for subsequent attenuation evaluation.

[0026] Training the convolutional neural network using the labeled photo - induced attenuation dataset: Train the convolutional neural network using the labeled photo - induced attenuation dataset. Each sample in the labeled dataset contains known attenuation information, including attenuation rate, reversibility, and attenuation amplitude, etc.

[0027] Loss function: Use the mean squared error as the loss function to optimize the parameters of the convolutional neural network. The loss function formula is as follows: ; where represents the value of the loss function; is the number of samples in the dataset; is the attenuation value of the th sample predicted by the convolutional neural network model; is the attenuation value of the true label.

[0028] During the training process, the backpropagation algorithm is used to adjust the weights of the network by optimizing the loss function, so that the model can accurately predict the attenuation characteristics of photovoltaic modules.

[0029] Using the trained convolutional neural network model to determine whether the photo - induced attenuation of photovoltaic modules is reversible: Determine whether the photo - induced attenuation of photovoltaic modules is reversible through the output of the convolutional neural network model. The output of the convolutional neural network model is normalized to the [0, 1] interval using the Sigmoid function, which is used to represent whether the photo - induced attenuation is reversible: ; where is the output of the Sigmoid function, representing the probability value that the photo - induced attenuation is reversible; is the original output of the convolutional neural network model, usually the output of the last layer of the neural network, representing the unprocessed value for the prediction of the reversibility of light-induced degradation; is the base of the natural logarithm.

[0030] The preset reversibility threshold is used to compare the probability value of light-induced degradation being reversible with the reversibility threshold: When the probability value of light-induced degradation being reversible is greater than the reversibility threshold, it is determined that the light-induced degradation of the photovoltaic module is reversible, indicating that the degree of damage to the photovoltaic module is relatively light and can be restored through repair or specific maintenance measures. Generally, it will not have a significant impact on the long-term operating performance, and the difficulty of restoration is relatively low. After identifying these photovoltaic modules as reversibly degraded, the allocation of resources can be optimized, and the limited maintenance resources can be preferentially applied to the photovoltaic modules with irreversible degradation; When the probability value of light-induced degradation being reversible is less than or equal to the reversibility threshold, it is determined that the light-induced degradation of the photovoltaic module is irreversible, meaning that the degradation process of the photovoltaic module cannot be restored. The degradation is usually caused by irreversible factors such as long-term high-temperature exposure, material aging, the expansion of microcracks, or other serious damages to the photovoltaic module. These damages not only affect the photoelectric conversion efficiency of the module but may also further lead to performance degradation. Modules with irreversible degradation often need to be replaced or undergo more complex repairs. For these photovoltaic modules with irreversible degradation, corresponding emergency measures need to be taken, such as reducing the overall load or considering replacing them with new and more efficient modules.

[0031] Specifically, based on the light intensity change data in the light intensity mutation area at the cloud edge, evaluate the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array, including: Collect the light intensity change data in the cloud edge area using high-precision sensors: The sensors need to be deployed near the photovoltaic array to monitor the light intensity change in the cloud edge area in real time. These sensors should have high sensitivity, wide dynamic range, and low noise characteristics to ensure that sufficient accurate data can be captured in a rapidly changing light environment.

[0032] During the data acquisition process, the light intensity change data recorded by the sensors will contain the light intensity fluctuation information caused by cloud occlusion and reflection. The data collected by the sensors should include the light intensity values in multiple time periods. In particular, the sensors should be set at different vertical and horizontal positions to accurately capture the light intensity change and avoid data loss caused by occlusion.

[0033] Analyze the light intensity data, identify the light intensity mutation area, and determine the distribution characteristics of the cloud edge: The light intensity mutation area is usually related to the cloud edge or the change in cloud thickness. The data analysis method needs to combine time series analysis techniques, such as the fast Fourier transform and wavelet transform methods in signal processing, to accurately detect the positions of the light intensity mutation points.

[0034] After identifying the light intensity mutation points, further analyze the distribution characteristics of the light intensity mutation region to determine the specific position of the cloud edge. The shape of the cloud edge and its relationship with light intensity fluctuations can be inferred by using multi-dimensional data interpolation methods and machine learning techniques in combination with multi-point light intensity data.

[0035] Propose a mathematical model to describe the impact of the non-linear occlusion effect caused by the cloud edge on light intensity: After obtaining the light intensity data and determining the cloud edge characteristics, establish a mathematical model to describe the impact of the non-linear occlusion effect caused by the cloud edge on light intensity. The non-linear occlusion effect means that the occlusion of the cloud does not change linearly with the change of the cloud thickness, but is affected by multiple factors such as cloud type, thickness, angle, and solar radiation direction.

[0036] The established mathematical model should be based on the optical transmission model. Specifically, the following formula can be used to describe the occlusion effect of the cloud on light intensity: ; where, is the light intensity value after cloud occlusion at time point ; is the initial incident light intensity before cloud occlusion; is the transmission coefficient of the cloud; is the incident angle of sunlight; is the thickness of the cloud; is the base of the natural logarithm.

[0037] The transmission coefficient describes the attenuation degree of light passing through the cloud, depends on the thickness, composition of the cloud and the incident angle of sunlight, and needs to be calibrated through experimental data.

[0038] Utilize the optical reflection principle to construct a reflection gain model to simulate the reflection gain effect of light intensity: During the light intensity change process at the cloud edge, in addition to the occlusion effect, the reflection gain effect is also an important factor affecting light intensity fluctuations. The reflection gain effect is caused by the reflection properties of the cloud. Especially under specific angles and cloud surface conditions, the cloud edge area will cause an increase in the reflected light intensity. Establish a reflection gain model to simulate the contribution of cloud reflection to light intensity.

[0039] The reflection gain can be described by the following formula: ; where, is the reflected light intensity at time point ; is the reflection coefficient of the cloud; is the initial incident light intensity before cloud occlusion; is the incident angle of sunlight.

[0040] The reflection coefficient describes the ability of the cloud surface to reflect light, which depends on the surface characteristics of the cloud (such as smoothness, cloud moisture, etc.) and the incident angle of sunlight.

[0041] The reflection gain effect can cause a sudden change in light intensity in a short period of time, which in turn affects the output power of the photovoltaic array. Therefore, the optical transmission model needs to comprehensively consider the dual effects of cloud occlusion and reflection on light intensity changes.

[0042] Combined with the light intensity change model, a mathematical model of the power fluctuation of the photovoltaic array is established: The output power of the photovoltaic array is proportional to the incident light intensity. Therefore, the light intensity change model can directly affect the power output. Define the output power of the photovoltaic array as , which can be expressed as: ; where is the output power at time point ; is the energy conversion efficiency of the photovoltaic array, which represents the ability of the photovoltaic array to convert light energy into electrical energy and is usually a constant; is the effective area of the photovoltaic array, which represents the total area of the photovoltaic array, usually in square meters, and determines the amount of light energy that the photovoltaic array can capture; is the total light intensity at time point , including the sum of the light intensity of cloud occlusion and reflection gain, expressed as: .

[0043] According to the mathematical model of the power fluctuation of the photovoltaic array, the effects of cloud edge occlusion and reflection gain on the power of the photovoltaic array can be analyzed, so as to predict the amplitude and frequency of power fluctuation.

[0044] According to the modeling results, quantify the influence degree of non-linear occlusion and reflection gain on power fluctuation: Since the occlusion effect and reflection effect of the cloud are usually non-linear, the amplitude of power fluctuation will be quantified by the fluctuation of light intensity change. Define the amplitude of power fluctuation as ; where is the average value of the output power of the photovoltaic array; is the amplitude of power fluctuation.

[0045] The rapid change of light intensity will cause the frequency of power fluctuation to fluctuate as well. Use spectral analysis (such as Fourier transform) to quantify the frequency characteristics of power fluctuation, and the expression is: ; where represents the distribution of power fluctuation in the frequency domain; is the imaginary unit; is the frequency variable, which represents the frequency component of power fluctuation in the frequency domain; is the base of the natural logarithm.

[0046] According to the results of Fourier transform, the contribution of different frequency components to power fluctuation can be quantified by calculating the spectral density of power fluctuation. It can be obtained by the following integral: ;in, is the energy of power fluctuation.

[0047] Since the blocking and reflection effects of clouds are time-varying, they not only affect the light intensity in a short period of time, but also may cause long-term light intensity fluctuations over time. Therefore, it is necessary to consider the time correlation, that is, whether the power fluctuation has autocorrelation characteristics, which is defined as: ;in, is the autocorrelation function of the power fluctuation; Indicates time delay; represents the time averaging operation, i.e., the time averaging of the product of power fluctuations, which is usually used to eliminate randomness and reveal the systematic changes of power fluctuations.

[0048] The volatility coefficient is defined to quantify the influence of the nonlinear shading and reflection gain effect of the cloud edge on the power fluctuation of the photovoltaic array. The calculation formula is: ;in, is the volatility coefficient; is the average value of the autocorrelation function of power fluctuation; is the energy of power fluctuation; is the average output power of the photovoltaic array.

[0049] The larger the volatility coefficient, the more significant the impact of the nonlinear shading and reflection gain effects caused by the cloud edge on the power fluctuation of the photovoltaic array. A larger volatility coefficient indicates that the light intensity changes caused by cloud shading and reflection effects are more drastic and frequent, directly causing significant fluctuations in the output power of the photovoltaic array. This volatility may manifest as strong power fluctuations in a short period of time, resulting in a decrease in the stability of the photovoltaic array, which may increase the complexity of load regulation and reduce the predictability of energy output. Larger volatility coefficients are usually associated with highly nonlinear effects of clouds, such as the complex morphology of cloud edges and the reflection enhancement effect of light. Especially in the short-term response of photovoltaic arrays, these nonlinear effects may have superimposed effects at different scales. Therefore, the increase in the volatility coefficient indicates the impact of more complex meteorological conditions on photovoltaic performance, and more sophisticated scheduling and prediction mechanisms are needed to reduce the instability and power loss caused by these fluctuations.

[0050] Specifically, determine the working state of the photovoltaic module according to the judgment result of whether the light-induced attenuation of the photovoltaic module is reversible and the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array; the working state of the photovoltaic module includes the normal working state and the abnormal working state, including: A preset volatility coefficient threshold is set, and the volatility coefficient is compared with the volatility coefficient threshold: When the volatility coefficient is greater than the volatility coefficient threshold, it indicates that the power fluctuation of the photovoltaic array is not within the controllable range, and the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array is high; at this time, the dynamic change at the cloud edge will cause a sharp fluctuation of the light intensity in a short time, directly resulting in a large fluctuation of the power generation of the photovoltaic array; a larger volatility coefficient usually means that the output power of the photovoltaic module is interfered by unstable factors, such as sudden shadow changes or a sharp increase in light intensity, which makes the power output show a large volatility and may affect the overall stability and power generation efficiency. When the volatility coefficient is less than or equal to the volatility coefficient threshold, it indicates that the power fluctuation of the photovoltaic array is within the controllable range, and the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array is low; at this time, the influence of the cloud layer on the light intensity is relatively gentle, and the power generation of the photovoltaic module remains relatively stable with a small fluctuation range. Due to the weak cloud edge effect, the change in light intensity will not have a significant impact on the power output of the photovoltaic array, which helps to improve the overall power generation efficiency and reduce the risk brought by frequent fluctuations.

[0051] The volatility coefficient threshold is comprehensively set according to the power fluctuation data of the historical photovoltaic array, the cloud distribution characteristics, and the regional climate conditions. By analyzing a large amount of historical data, the statistical characteristics of the power fluctuation of the photovoltaic array under different weather conditions can be calculated, and then the threshold that can effectively reflect the volatility influence can be determined. The setting of the threshold usually takes into account the sensitivity of the photovoltaic array in a specific environment and its tolerance to power fluctuations. For example, if the cloud edge effect in this area is frequent and strong, the threshold of the volatility coefficient may be set lower to identify potential high-volatility risks earlier. In areas with relatively gentle cloud changes, the threshold of the volatility coefficient may be set relatively higher.

[0052] When the probability value of the reversible light-induced attenuation is greater than the reversible threshold and the volatility coefficient is less than or equal to the volatility coefficient threshold, determine the working state of the photovoltaic module as the normal working state; otherwise, determine the working state of the photovoltaic module as the abnormal working state.

[0053] Specifically, by analyzing the change in the light transmittance of the surface of the photovoltaic module, evaluate the non-linear attenuation influence degree of the aerosol retention layer on the light transmittance of the photovoltaic module, including: Collect transmittance data and the accumulation of aerosols on the surface of photovoltaic modules: Collect the transmittance data of the surface of photovoltaic modules. Use a high-precision spectrometer to measure the transmittance of photovoltaic modules in real time. The high-precision spectrometer can provide full-spectrum data from ultraviolet to infrared, and can accurately capture the subtle changes in the transmittance of the surface of photovoltaic modules.

[0054] Monitor the aerosol retention layer through meteorological sensors and image recognition technology: Meteorological sensors dynamically obtain environmental change data affecting aerosol deposition by measuring factors such as humidity and wind speed in the environment. At the same time, image recognition technology can take real-time photos of the surface of photovoltaic modules through a high-definition camera, and use computer vision algorithms to quantitatively analyze the accumulation state of aerosols on the surface of photovoltaic modules. Combine meteorological conditions to analyze the distribution and adhesion characteristics of aerosols.

[0055] Based on the collected transmittance data and aerosol retention layer data, use the regression analysis method to establish a non-linear regression model between transmittance and aerosol accumulation: Based on the transmittance data and aerosol retention layer data, use the regression analysis method to establish a non-linear regression model to analyze the relationship between the thickness of the aerosol retention layer and the transmittance. The regression model considers the accumulation characteristics of aerosols, the uniformity of the distribution of aerosols on the surface of photovoltaic modules, and the influence of environmental factors. The expression of the regression model is: ; where is the transmittance of the photovoltaic module; is the thickness of the aerosol retention layer; is the uniformity of the distribution of aerosols on the surface; is the environmental humidity; is the adhesion coefficient of aerosols on the surface of the photovoltaic module.

[0056] Through the regression analysis of each variable, obtain a non-linear function describing the relationship between transmittance change and aerosol accumulation .

[0057] Establish a transmittance attenuation model to evaluate the non-linear attenuation degree of the aerosol retention layer on the transmittance of photovoltaic modules: Based on the results of regression analysis, establish a transmittance attenuation model. The degree of transmittance attenuation is a non-linear process. The transmittance attenuation model includes the following expressions for calculating the influence degree of aerosols on the transmittance of photovoltaic modules: ; where is the attenuation value of the transmittance; are regression coefficients, respectively representing the influence degrees of the thickness of the aerosol retention layer, the uniformity of the distribution of aerosols on the surface, and environmental humidity on the transmittance attenuation.

[0058] According to the attenuation value of the light transmittance, calculate the attenuation coefficient to evaluate the non-linear attenuation degree of the aerosol retention layer on the light transmittance of the photovoltaic module. The calculation formula is as follows: ; where is the attenuation coefficient; is the light transmittance attenuation value of the th light transmittance sampling point; is the weight coefficient of the th light transmittance sampling point; is the number of light transmittance sampling points.

[0059] The larger the attenuation coefficient, the greater the non-linear attenuation effect of the aerosol retention layer on the light transmittance of the photovoltaic module. A larger attenuation coefficient indicates that the deposition layer formed by the aerosol on the surface of the photovoltaic module is thicker and unevenly distributed, reducing the light intensity entering the photovoltaic module. A larger attenuation coefficient also reflects the dependence of the power generation capacity of the photovoltaic module on environmental conditions. For example, factors such as high humidity, severe pollution, or low wind speed may exacerbate the accumulation and adhesion of aerosols. A larger attenuation coefficient indicates that maintenance measures need to be taken in a timely manner, such as regularly cleaning the surface of the photovoltaic module to remove the accumulated aerosols and restore the light transmittance performance. Through the accurate evaluation of the attenuation coefficient, the managers of photovoltaic power plants can arrange the cleaning and maintenance cycles more scientifically to ensure maximum energy output.

[0060] Specifically, comprehensively analyze the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array and the influence degree of the non-linear attenuation of the aerosol retention layer on the light transmittance of the photovoltaic module, and judge whether to enable the operation and maintenance plan, including: Normalize the volatility coefficient corresponding to the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array and the attenuation coefficient corresponding to the influence degree of the non-linear attenuation of the aerosol retention layer on the light transmittance of the photovoltaic module respectively, and calculate the normalized volatility coefficient and attenuation coefficient to obtain the health score of the photovoltaic module. The calculation formula is as follows: ; where is the health score of the photovoltaic module; is the attenuation coefficient; is the volatility coefficient; is the weight factor of the volatility coefficient; is the weight factor of the attenuation coefficient; is the smoothing factor.

[0061] Preset a health score threshold, and compare the health score with the health score threshold: When the health score is greater than or equal to the health score threshold, it indicates that the current operating state of the photovoltaic module is good, and there is no need to enable the operation and maintenance plan, and the existing operating state can be continued; When the health score is less than the health score threshold, it indicates that the current operating state of the photovoltaic module is poor, and an operation and maintenance plan needs to be enabled, including starting the surface cleaning procedure of the photovoltaic module, such as using an automated cleaning device or manual cleaning to remove the surface aerosol retention layer, dust, and other pollutants, and restoring the cleanliness of the photovoltaic module; detecting the power output characteristics of the photovoltaic module to identify whether there are abnormal electrical characteristics, such as local hot spots or series mismatch problems, and adjusting the operating parameters of the inverter to optimize the power output. In addition, the operation and maintenance plan may also include manual inspections, focusing on checking for physical damage or aging on the surface of the photovoltaic module, and promptly replacing damaged photovoltaic modules. Embodiment

[0062] The difference between Embodiment 2 and Embodiment 1 of the present invention is that this embodiment introduces a comprehensive solar power generation management system based on AI intelligence.

[0063] Figure 2 The structural schematic diagram of a comprehensive solar power generation management system based on AI intelligence of the present invention is given. A comprehensive solar power generation management system based on AI intelligence includes an attenuation reversibility judgment module, a power fluctuation evaluation module, a working state determination module, a light transmittance analysis module, and a comprehensive analysis module; The attenuation reversibility judgment module analyzes the light-induced attenuation data of the photovoltaic module and judges whether the light-induced attenuation of the photovoltaic module is reversible according to the convolutional neural network; The power fluctuation evaluation module evaluates the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array based on the light intensity change data in the light intensity mutation area at the cloud edge; The working state determination module determines the working state of the photovoltaic module according to the judgment result of whether the light-induced attenuation of the photovoltaic module is reversible and the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array; the working state of the photovoltaic module includes a normal working state and an abnormal working state; When the working state of the photovoltaic module is an abnormal working state, the light transmittance analysis module evaluates the influence degree of the non-linear attenuation of the aerosol retention layer on the light transmittance of the photovoltaic module by analyzing the change in the light transmittance on the surface of the photovoltaic module; The comprehensive analysis module comprehensively analyzes the influence degree of the non-linear occlusion and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array and the influence degree of the non-linear attenuation of the aerosol retention layer on the light transmittance of the photovoltaic module, and judges whether to enable the operation and maintenance plan. Embodiment

[0064] A comprehensive solar power generation management device based on AI intelligence includes: a processor, a memory, and a program or instruction stored on the memory and executable on the processor. When the program or instruction is executed by the processor, it implements a comprehensive solar power generation management method based on AI intelligence.

[0065] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the actual situation. The preset parameters and threshold selection in the formulas are set by those skilled in the art according to the actual situation. Where the present invention is not described applies to the prior art.

Claims

1. A solar power generation integrated management method based on AI intelligence, characterized in that: The steps include: By analyzing the light-induced degradation data of photovoltaic modules, the convolutional neural network is used to determine whether the light-induced degradation of photovoltaic modules is reversible. Based on the light intensity variation data in the cloud edge light intensity mutation area, the influence of the nonlinear shading and reflection gain effect at the cloud edge on the power fluctuation of the photovoltaic array is evaluated; The working state of the photovoltaic module is determined according to the judgment result of whether the light-induced attenuation of the photovoltaic module is reversible and the influence degree of the nonlinear shielding and reflection gain effect of the cloud edge on the power fluctuation of the photovoltaic array; the working state of the photovoltaic module includes the normal working state and the abnormal working state; When the working state of the photovoltaic module is abnormal, the nonlinear attenuation effect of the aerosol retention layer on the transmittance of the photovoltaic module is evaluated by analyzing the change of the transmittance on the surface of the photovoltaic module; A comprehensive analysis is conducted on the impact of the nonlinear shading and reflection gain effects of the cloud edge on the power fluctuations of the photovoltaic array and the nonlinear attenuation of the aerosol retention layer on the transmittance of the photovoltaic module to determine whether to activate the operation and maintenance plan.

2. According to claim 1, a solar power generation integrated management method based on AI intelligence is characterized in that: By analyzing the light-induced degradation data of photovoltaic modules, the convolutional neural network is used to determine whether the light-induced degradation of photovoltaic modules is reversible. Specifically: Regularly collect light-induced degradation data of photovoltaic modules through high-precision sensors; Standardize the collected light-induced attenuation data; A convolutional neural network is used to extract key features from light-induced degradation data and identify spatiotemporal patterns in the degradation process of photovoltaic modules; Train a convolutional neural network using an annotated light-induced attenuation dataset; The trained convolutional neural network model is used to determine whether the light-induced degradation of photovoltaic modules is reversible.

3. According to claim 2, a solar power generation integrated management method based on AI intelligence is characterized in that: Determine whether the light-induced degradation of photovoltaic modules is reversible, specifically: The reversible threshold is preset, and the probability value of photoinduced decay being reversible is compared with the reversible threshold: When the probability value of the reversible light-induced degradation is greater than the reversible threshold, the light-induced degradation of the photovoltaic module is determined to be reversible; When the probability value that the photo-induced degradation is reversible is less than or equal to the reversible threshold, it is determined that the photo-induced degradation of the photovoltaic module is irreversible.

4. According to claim 3, a solar power generation integrated management method based on AI intelligence is characterized in that: Based on the light intensity change data in the cloud edge light intensity mutation area, the influence of the nonlinear shading and reflection gain effect at the cloud edge on the power fluctuation of the photovoltaic array is evaluated, specifically: Use high-precision sensors to collect data on light intensity changes in the cloud edge area; Analyze light intensity data, identify areas of sudden changes in light intensity, and determine the distribution characteristics of the cloud edge; A mathematical model is proposed to describe the impact of nonlinear shading effect caused by cloud edge on light intensity; Using the principle of optical reflection, a reflection gain model is constructed to simulate the reflection gain effect of light intensity; Combined with the light intensity variation model, a mathematical model of photovoltaic array power fluctuation is established; Based on the modeling results, the influence of nonlinear shading and reflection gain effects on power fluctuations is quantified.

5. According to claim 4, a solar power generation integrated management method based on AI intelligence is characterized in that: The working state of the photovoltaic module is determined according to the judgment result of whether the light-induced attenuation of the photovoltaic module is reversible and the influence of the nonlinear shading and reflection gain effect of the cloud edge on the power fluctuation of the photovoltaic array; the working state of the photovoltaic module includes the normal working state and the abnormal working state, specifically: Preset the volatility coefficient threshold and compare the volatility coefficient with the volatility coefficient threshold: When the volatility coefficient is greater than the volatility coefficient threshold, it means that the nonlinear shading and reflection gain effect of the cloud edge has a high degree of influence on the power fluctuation of the photovoltaic array; When the volatility coefficient is less than or equal to the volatility coefficient threshold, it means that the nonlinear shading and reflection gain effect of the cloud edge has a low impact on the power fluctuation of the photovoltaic array. When the probability value of the reversible photodegradation is greater than the reversible threshold and the volatility coefficient is less than or equal to the volatility coefficient threshold, the working state of the photovoltaic component is determined to be a normal working state; otherwise, the working state of the photovoltaic component is determined to be an abnormal working state.

6. According to claim 5, a solar power generation integrated management method based on AI intelligence is characterized in that: By analyzing the changes in the transmittance of the photovoltaic module surface, the nonlinear attenuation effect of the aerosol retention layer on the transmittance of the photovoltaic module is evaluated, specifically: Collect transmittance data and aerosol accumulation on the surface of photovoltaic modules; Monitor the aerosol retention layer through meteorological sensors and image recognition technology; Based on the collected transmittance data and aerosol retention layer data, a nonlinear regression model between transmittance and aerosol accumulation was established using regression analysis. A transmittance attenuation model was established to evaluate the nonlinear attenuation of the aerosol retention layer on the transmittance of photovoltaic modules: the attenuation coefficient was calculated as: ;in, is the attenuation coefficient; For the The transmittance attenuation value of the transmittance sampling point; For the The weight coefficient of each transmittance sampling point; is the number of transmittance sampling points.

7. The AI-based integrated management method for solar power generation according to claim 6 is characterized in that: A comprehensive analysis is conducted on the impact of the nonlinear shading and reflection gain effects of the cloud edge on the power fluctuation of the photovoltaic array and the nonlinear attenuation of the light transmittance of the photovoltaic module by the aerosol retention layer to determine whether to enable the operation and maintenance plan, specifically: The normalized volatility coefficient and attenuation coefficient are calculated to obtain the health score of the photovoltaic module. The calculation formula is: ;in, Provide health scores for PV panels; is the attenuation coefficient; is the volatility coefficient; is the weight factor of the volatility coefficient; is the weight factor of the attenuation coefficient; is the smoothing factor; Preset health score threshold and compare the health score to the health score threshold: When the health score is greater than or equal to the health score threshold, there is no need to enable the operation and maintenance solution; When the health score is lower than the health score threshold, the operation and maintenance solution needs to be enabled.

8. An AI-based integrated solar power generation management system, used to implement an AI-based integrated solar power generation management method according to any one of claims 1 to 7, characterized in that: It includes an attenuation reversible judgment module, a power fluctuation assessment module, a working state determination module, a transmittance analysis module and a comprehensive analysis module; The attenuation reversibility judgment module analyzes the light-induced attenuation data of the photovoltaic module and judges whether the light-induced attenuation of the photovoltaic module is reversible based on the convolutional neural network; The power fluctuation assessment module evaluates the influence of the nonlinear shading and reflection gain effects at the cloud edge on the power fluctuation of the photovoltaic array based on the light intensity change data in the light intensity mutation area at the cloud edge. The working state determination module determines the working state of the photovoltaic module according to the judgment result of whether the light-induced attenuation of the photovoltaic module is reversible and the influence degree of the nonlinear shading and reflection gain effect of the cloud edge on the power fluctuation of the photovoltaic array; the working state of the photovoltaic module includes the normal working state and the abnormal working state; When the working state of the photovoltaic module is abnormal, the transmittance analysis module evaluates the nonlinear attenuation effect of the aerosol retention layer on the transmittance of the photovoltaic module by analyzing the transmittance change on the surface of the photovoltaic module; The comprehensive analysis module conducts a comprehensive analysis of the impact of the nonlinear shading and reflection gain effects of the cloud edge on the power fluctuation of the photovoltaic array and the nonlinear attenuation of the aerosol retention layer on the transmittance of the photovoltaic module to determine whether to enable the operation and maintenance plan.

9. An AI-based integrated solar power generation management device, characterized in that: include: A processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein when the program or instruction is executed by the processor, an AI-based integrated management method for solar power generation is implemented as described in any one of claims 1 to 7.

Images