Photovoltaic equipment temperature anomaly prediction method and device based on digital twinborn model

Abstract

The invention relates to the field of photovoltaic equipment abnormity early warning, and provides a photovoltaic equipment temperature abnormity prediction method and device based on a digital twinborn model, and the method comprises the steps: obtaining an environment acoustic signal, meteorological data, and the actually measured temperature data of photovoltaic equipment; performing acoustic spectrum feature extraction according to the environmental acoustic signal to obtain a virtual environmental parameter set; sending the meteorological data and the virtual environment parameter set to a preset digital twinborn model for processing to obtain an initial temperature predicted value set; determining a temperature prediction value distribution diagram according to the initial temperature prediction value set and actually measured temperature data of the photovoltaic equipment, and comparing a temperature prediction result with an abnormal temperature threshold value to obtain a temperature abnormal risk point; and processing according to the temperature prediction value distribution diagram, the temperature anomaly risk points and the meteorological data to obtain temperature anomaly early warning information of the photovoltaic equipment, and improving the accuracy and foresight of temperature anomaly prediction of the photovoltaic equipment.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic equipment anomaly early warning, and more specifically, to a method and apparatus for predicting photovoltaic equipment temperature anomalies based on a digital twin model. Background Technology

[0002] In the field of photovoltaic power plant operation and maintenance, accurate monitoring and fault early warning of equipment operating status are crucial to ensuring power generation efficiency and safety. Abnormal increases in photovoltaic module temperature are a major cause of heat spot effects, power degradation, and even fire risks; therefore, accurate prediction and early warning of module temperature anomalies are particularly important. Currently, the mainstream methods in this field mainly rely on direct measurement using temperature sensors or temperature estimation based on limited meteorological data and simplified thermodynamic models. However, existing technologies generally suffer from inherent defects such as insufficient perception of the module surface microenvironment, overly simplistic and idealized model input parameters, and the inability to quantify predictive uncertainties. These defects lead to delayed early warning results, high false alarm rates, or missed alarm risks, making it difficult to meet the operation and maintenance needs of modern large-scale photovoltaic power plants for high-precision, forward-looking prediction and intelligent diagnosis of temperature anomalies. Summary of the Invention

[0003] The purpose of this invention is to provide a method and apparatus for predicting temperature anomalies in photovoltaic equipment based on a digital twin model, so as to improve the above-mentioned problems.

[0004] To achieve the above objectives, the embodiments of this application provide the following technical solutions:

[0005] On the one hand, embodiments of this application provide a method for predicting temperature anomalies in photovoltaic equipment based on a digital twin model, the method comprising:

[0006] Acquire environmental acoustic signals, meteorological data, and measured temperature data of photovoltaic equipment;

[0007] Based on the environmental acoustic signal, acoustic spectrum features are extracted to obtain a virtual environmental parameter set, which includes a virtual wind speed distribution map and a virtual stain distribution map.

[0008] The meteorological data and the set of virtual environmental parameters are sent to a preset digital twin model for processing to obtain an initial set of temperature prediction values;

[0009] A temperature prediction distribution map is determined based on the initial temperature prediction set and the measured temperature data of the photovoltaic equipment. The temperature prediction results are then compared with the abnormal temperature threshold to obtain temperature anomaly risk points.

[0010] Based on the temperature prediction distribution map, the temperature anomaly risk points, and the meteorological data, an early warning information for abnormal temperature of photovoltaic equipment is obtained.

[0011] Secondly, embodiments of this application provide a photovoltaic equipment temperature anomaly prediction device based on a digital twin model, the device comprising:

[0012] The acquisition module is used to acquire environmental acoustic signals, meteorological data, and measured temperature data of photovoltaic equipment.

[0013] The first processing module is used to extract acoustic spectrum features based on the environmental acoustic signal to obtain a virtual environmental parameter set, which includes a virtual wind speed distribution map and a virtual stain distribution map.

[0014] The second processing module is used to send the meteorological data and the virtual environment parameter set to a preset digital twin model for processing to obtain an initial temperature prediction value set;

[0015] The third processing module is used to determine the temperature prediction value distribution map based on the initial temperature prediction value set and the measured temperature data of the photovoltaic equipment, and compare the temperature prediction results with the abnormal temperature threshold to obtain the temperature abnormality risk points.

[0016] The fourth processing module is used to process the temperature prediction distribution map, the temperature anomaly risk points, and the meteorological data to obtain photovoltaic equipment temperature anomaly early warning information.

[0017] Thirdly, embodiments of this application provide a photovoltaic equipment temperature anomaly prediction device based on a digital twin model. The device includes a memory and a processor. The memory stores a computer program; the processor executes the computer program to implement the steps of the aforementioned photovoltaic equipment temperature anomaly prediction method based on a digital twin model.

[0018] Fourthly, embodiments of this application provide a readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described method for predicting temperature anomalies in photovoltaic equipment based on a digital twin model.

[0019] The beneficial effects of this invention are as follows:

[0020] This invention acquires environmental acoustic signals, meteorological data, and measured temperature data of photovoltaic equipment. Based on the environmental acoustic signals, it extracts acoustic spectrum features to obtain a virtual environmental parameter set including a virtual wind speed distribution map and a virtual stain distribution map. The meteorological data and the virtual environmental parameter set are input into a preset digital twin model to obtain an initial temperature prediction value set. The initial temperature prediction value set is combined with the measured temperature data to determine the temperature prediction value distribution map and compare it with abnormal temperature thresholds to obtain temperature anomaly risk points. Finally, based on the temperature prediction value distribution map, temperature anomaly risk points, and meteorological data processing, a photovoltaic equipment temperature anomaly early warning information is generated. This effectively makes up for the shortcomings of existing technologies, such as insufficient perception of the microenvironment on the surface of photovoltaic equipment, single idealized model input parameters, and inability to quantify prediction uncertainties. It improves the accuracy and foresight of photovoltaic equipment temperature anomaly prediction and reduces the lag, false alarm rate, and missed alarm risk of early warning results.

[0021] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a schematic diagram of the photovoltaic equipment temperature anomaly prediction method using a digital twin model as described in this embodiment of the invention.

[0024] Figure 2 This is a schematic diagram of the structure of the photovoltaic equipment temperature anomaly prediction device based on the digital twin model described in this embodiment of the invention.

[0025] The diagram is labeled as follows: 800, Photovoltaic equipment temperature anomaly prediction device based on digital twin model; 801, Processor; 802, Memory; 803, Multimedia component; 804, I / O interface; 805, Communication component. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0027] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0028] Example 1:

[0029] This embodiment provides a method for predicting temperature anomalies in photovoltaic equipment using a digital twin model. It can be understood that this embodiment can be used to set up a scenario, such as: in the operation and maintenance of a large photovoltaic power station, a scenario in which temperature anomalies in photovoltaic equipment are predicted by collecting on-site environmental acoustic signals, meteorological data such as air temperature and sunlight, and the measured temperature of photovoltaic panels.

[0030] See Figure 1 The figure shows that the method includes steps S1-S5.

[0031] Step S1: Acquire environmental acoustic signals, meteorological data, and measured temperature data of the photovoltaic equipment;

[0032] This step differs from traditional methods that rely solely on meteorological data or sparse temperature sensors. It introduces the information dimension of environmental acoustic signals. In the vast and complex environment of a photovoltaic power station, wind, rain, dust, and the vibration of the modules themselves all produce unique sound characteristics. Acquiring these signals allows us to capture the dynamic changes in the microclimate on the module surface, laying the data foundation for subsequent non-contact, large-scale inversion of key environmental parameters. This overcomes the bottleneck of traditional point-based measurements, which struggle to capture spatial heterogeneity.

[0033] Step S2: Extract acoustic spectrum features based on the environmental acoustic signal to obtain a virtual environmental parameter set, which includes a virtual wind speed distribution map and a virtual stain distribution map;

[0034] This step transforms acoustic phenomena in the physical world into a virtual environmental field in digital space. It utilizes the propagation law of sound in the turbulent boundary layer to invert wind speed and estimates the thickness of dirt based on the attenuation characteristics of the interaction between sound waves and dust particles. This enables real-time monitoring of wind speed and dirt on the component surface and generates a high-resolution environmental map that is difficult to obtain with traditional sensor networks.

[0035] Step S2 further includes steps S21-S24, which specifically include:

[0036] Step S21: Enhance the environmental acoustic signal to obtain a feature vector with enhanced acoustic features;

[0037] Independent Component Analysis (ICA) was employed to perform blind source separation on the environmental acoustic signals collected by the acoustic sensor network, effectively separating environmental wind noise as the dominant independent sound source from the mixed signal. Subsequently, wavelet packet transform was performed on the residual signal after separation, which mainly contains information on particle impact and component vibration, to extract energy features from 2kHz to 8kHz, resulting in feature vectors that enhance acoustic features. This step, addressing the characteristics of strong wind noise and weak target acoustic features in the photovoltaic array area, significantly improved the signal-to-noise ratio of acoustic features related to the component surface state through signal separation guided by physical mechanisms, laying a clean signal foundation for subsequent accurate inversion.

[0038] Step S22: Based on the feature vector of the enhanced acoustic features, the surface wind speed of the photovoltaic device is inverted to obtain a virtual wind speed distribution map;

[0039] Step S22 further includes steps S221-S224, which specifically include:

[0040] Step S221: Perform acoustic coherence feature extraction processing on the feature vector of the enhanced acoustic features to obtain a frequency band energy matrix, wherein the frequency band energy matrix includes the spatial coherence corresponding to the frequency band;

[0041] This step first targets the acoustic characteristics generated by the turbulent boundary layer on the photovoltaic panel surface, selecting a specific high-frequency band (2kHz-8kHz) that is sensitive to changes in turbulence scale and less affected by environmental noise. Then, cross-power spectral density analysis is applied to calculate the coherence of microphone signals at different locations within this frequency band in the acoustic sensor network. Coherence characterizes the degree of linear correlation between signals at different locations in a statistical sense, and its calculation is based on the ratio of cross-power spectral density to self-power spectral density obtained from Fourier transform. By analyzing the spatial coherence patterns of the sensor array signals within the characteristic frequency band, a frequency band energy matrix containing spatial coherence information for each frequency band is constructed. By utilizing the high coherence of acoustic signals excited at different locations when the same turbulent structure passes over the photovoltaic panel surface, uncorrelated environmental background noise is effectively suppressed, and robust acoustic features strongly correlated with surface friction wind speed are extracted.

[0042] Step S222: Perform forward modeling based on the frequency band energy matrix to obtain the forward model;

[0043] This step constructs a physical model based on an improved Curle acoustic analogy. The classic Curle acoustic analogy treats the surface of an object as a sound source, assuming that sound is generated by unsteady pressure fluctuations on the surface. For the specific scenario of photovoltaic modules, the model simplifies the photovoltaic panel as a rigid plate of finite dimensions and considers its boundary layer turbulent pressure fluctuations as the primary sound source. The modeling process specifically includes: defining the geometric dimensions and spatial orientation of the plate, and setting the surface friction wind speed. As a key input parameter, through solving and A statistical model of the relevant boundary layer pressure fluctuations was developed, and the theoretical sound pressure level distribution radiated by these pressure fluctuations at the sensor location was calculated by integration, particularly the sound pressure level in the 2kHz-8kHz range. The resulting forward model is a deterministic function that establishes a physical mapping relationship between the hydrodynamic parameter of surface frictional wind speed and the predicted sound pressure level distribution.

[0044] Step S223: Solve for the surface friction wind speed according to the forward model and the frequency band energy matrix to obtain the surface friction wind speed field;

[0045] This step uses the forward model as the forward operator and assumes that the model error and measurement error follow a Gaussian distribution to construct the likelihood function for the inversion problem. This function describes the probability of observing the current frequency band energy matrix given the surface friction wind speed. Secondly, the spatial smoothness of the surface friction wind speed is introduced as a prior constraint, meaning that the wind speed values ​​at adjacent spatial points should not undergo drastic changes. Then, a Markov chain Monte Carlo sampling method is used to perform random walks in the parameter space to explore and sample the posterior probability distribution. Markov chain Monte Carlo generates a Markov chain, whose stationary distribution is the posterior distribution of the parameters to be determined. Finally, by analyzing a large number of sampling points, the posterior probability distribution of the surface friction wind speed field can be obtained, and its maximum value (maximum a posteriori estimate) is the most likely surface friction wind speed field.

[0046] Step S224: Determine the virtual wind speed distribution map based on the surface friction wind speed field.

[0047] In this step, the basic hydrodynamic parameters obtained from the inversion are converted into standard height wind speeds that are more intuitive and directly applicable in engineering. The specific conversion process is as follows:

[0048]

[0049] In the above formula, Let be the target wind speed, and represent the distance from the component height as . The average wind speed at that location; Indicates surface friction wind speed; This represents the Karman constant, which is typically 0.4. The vertical height representing the target wind speed to be determined; This represents the surface roughness length, used to quantify the aerodynamic roughness of the surface. It is determined based on the underlying surface conditions of the power plant site or selected from a table; for example, 0.0001-0.001 for concrete surfaces and 0.01-0.05 for low-lying grass. Using this formula, the surface friction wind speed at each location is converted into the average wind speed at a certain standard height above the module surface. Finally, a spatial interpolation algorithm is used to interpolate and extrapolate the wind speed estimates at these discrete sensor points, generating a continuous two-dimensional virtual wind speed distribution map covering the entire photovoltaic array.

[0050] Step S23: Based on the virtual wind speed distribution map and the feature vector of the enhanced acoustic features, predict the thickness of the stains on the surface of the photovoltaic equipment to obtain a virtual stain distribution map;

[0051] Step S23 further includes steps S231-S234, which specifically include:

[0052] Step S231: Perform secondary separation processing of particle impact sound and wind speed background sound based on the virtual wind speed distribution map and the feature vector of the enhanced acoustic features to obtain acoustic energy features;

[0053] Since wind speed is the primary factor influencing the intensity of ambient background noise, a virtual wind speed distribution map is used as the explanatory variable, and the eigenvectors enhancing acoustic features are used as the response variable. A linear regression model between wind speed and broadband acoustic signal energy is established using a partial least squares regression algorithm. This algorithm maximizes the covariance between the explanatory and response variables by finding a set of latent variables, thereby fitting the acoustic components most correlated with wind speed. Subtracting this fitted value from the original acoustic signal yields the residual signal, which represents the characteristic acoustic energy generated primarily by dust particle impact and friction, after removing wind speed-related noise. This approach effectively removes background interference and highlights the acoustic signals sensitive to dirt, taking into account the dynamic changes in ambient noise from photovoltaic power plants with wind speed.

[0054] Step S232: Normalize the sound energy characteristics and the virtual wind speed distribution map to obtain the particle impact sound energy distribution after wind speed normalization;

[0055] The normalization function for this step is as follows:

[0056]

[0057] In the above formula, This represents the sound energy of particle impacts after wind speed normalization. This represents the measured sound energy of particle impact; K represents the scaling factor; and U represents the wind speed value at the corresponding sensor location obtained from the virtual wind speed distribution map. This calculation uniformly converts the sound energy measured at different wind speeds to the equivalent sound energy under the same standard wind speed.

[0058] Step S233: Perform inversion processing based on the particle impact acoustic energy distribution after wind speed normalization to obtain stain thickness information;

[0059] This step models the dust layer on the surface of the photovoltaic device as a porous dielectric layer with a specific acoustic impedance. Then, the transfer matrix method, which measures sound wave propagation in a multi-layered medium, is applied to calculate the theoretical attenuation of the sound wave after penetrating the dust layer. The transfer matrix method represents each dielectric layer and its boundary conditions using matrices, and calculates the overall propagation characteristics of the sound wave along the entire path through matrix multiplication. By matching the theoretical attenuation corresponding to dust layers of different thicknesses with the normalized acoustic energy attenuation (obtained by comparing with the baseline acoustic energy under clean conditions in that area), and using optimization algorithms such as least squares to find the optimal solution, the equivalent dirt thickness corresponding to each sensor point can be derived.

[0060] Step S234: Perform spatial interpolation based on the stain thickness information to obtain a virtual stain distribution map.

[0061] Step S24: Construct a virtual environmental parameter set based on the virtual wind speed distribution map and the virtual stain distribution map.

[0062] In this step, principal component analysis (PCA) is used to compress and fuse the features of the obtained two-dimensional field data. PCA can extract the principal component feature vector that best represents the comprehensive microclimate state of the component surface from two highly correlated parameters: wind speed and dirt thickness (e.g., high dirt areas are often accompanied by low wind speeds). The final generated virtual environment parameter set is a multi-dimensional, gridded data structure that not only contains the original field information but also generates a low-dimensional, efficient feature representation that more comprehensively reflects the actual operating environment of the component, providing a high-quality, integrated driving input for the subsequent digital twin model.

[0063] Step S3: Send the meteorological data and the virtual environment parameter set to a preset digital twin model for processing to obtain an initial temperature prediction value set;

[0064] This step involves constructing a ensemble of digital twin models with varying parameters for parallel simulation, where each model represents a possible physical state. The output of this ensemble is no longer a fixed temperature value, but rather a set of predictions containing probability distribution information. This method is particularly well-suited to the dispersed performance characteristics of photovoltaic power plant components, giving the prediction results an inherent measure of their reliability and providing richer evidence for risk assessment.

[0065] Step S3 further includes steps S31-S37, which specifically include:

[0066] Step S31: Perform finite element modeling based on the material properties and structural dimensions of the photovoltaic equipment to obtain the finite element model;

[0067] Establishing a finite element model of photovoltaic equipment is a well-known technical solution in the field of photovoltaics, so it will not be elaborated here.

[0068] Step S32: Perform posterior distribution estimation processing based on the finite element model and historical operating data to obtain the posterior probability distribution of key thermophysical parameters;

[0069] In this step, the finite element model is used as the forward model, and its key thermophysical parameters are treated as unknowns to be calibrated, with reasonable prior probability distributions set for them. Irradiance, ambient temperature, and wind speed monitored during historical operation are used as model inputs, and component temperatures measured at corresponding times are used as observation data. A Markov chain Monte Carlo sampling method is used to perform random walks within the parameter space to explore which parameter combinations best match the model output with historical observation data, thereby obtaining the posterior probability distributions of these key thermophysical parameters. It should be noted that key thermophysical parameters include, but are not limited to, the backplate thermal conductivity and the specific heat capacity of the encapsulation material.

[0070] Step S33: Perform model population sampling processing based on the optimal Latin hypercube experimental design according to the posterior probability distribution of the key thermophysical parameters to obtain a set of virtual component model instances;

[0071] In this step, optimal Latin hypercube sampling is an experimental design method that maximizes the uniformity of a multidimensional parameter space (each dimension corresponding to a parameter to be sampled) with the fewest possible sample points. Here, the sampling space is the posterior distribution of each parameter. Through optimal Latin hypercube sampling, hundreds to thousands of different parameter combinations can be systematically generated, each of which instantiates a component model with distinct parameters. These model instances, when combined, constitute a ensemble model that can fully characterize the dispersion of component performance caused by manufacturing tolerances, batch variations in materials, and different degrees of aging.

[0072] Step S34: Construct a digital twin model based on the set of virtual component model instances;

[0073] Step S35: Solve the temperature field evolution sequence of each model based on the digital twin model, the meteorological data, and the virtual environment parameter set;

[0074] In this step, future weather forecast data (irradiance, ambient temperature) and virtual environmental parameter sets, including virtual wind speed distribution maps and virtual dirt distribution maps, are used as unified input conditions. This drives the constructed digital twin model to perform transient thermal simulations. Each model independently solves its energy conservation equations, with the virtual wind speed distribution map serving as the boundary condition for convective heat dissipation, and the virtual dirt distribution map correcting for solar radiation heat absorption by influencing the optical properties of the component surface. Through parallel computation, the temperature distribution of each model instance at each future time step is quickly obtained, forming its unique temperature field evolution sequence.

[0075] Step S36: Based on the evolution sequence of the body temperature field of each model, perform statistical aggregation processing of the group output to obtain the ensemble mean field and ensemble variance field of the temperature prediction.

[0076] In this step, for each location point within the power plant, at each prediction time point, the predicted temperature values ​​of all individual models at that point are treated as a statistical sample set. The arithmetic mean of this set is calculated to obtain the ensemble mean field, representing the most likely temperature prediction considered by the model group. Simultaneously, the variance of this set is calculated to obtain the ensemble variance field, quantifying the degree of prediction discrepancy due to model parameter uncertainties. Regions with large variance indicate unreliable predictions and high risk; regions with small variance indicate high prediction confidence.

[0077] Step S37: Based on the set mean field and set variance field of the temperature prediction set, perform probabilistic prediction set construction processing to obtain the initial temperature prediction value set and its distribution description.

[0078] This step transforms the determined predicted values ​​into probabilistic predictions. It assumes that at each location and at each time point, the predicted temperature of the model population follows a probability distribution with the ensemble mean as the expectation and the ensemble variance as the variance. Thus, the final output is no longer a single temperature value, but a probability distribution. A specific implementation would be that the prediction result can be expressed as a 90% probability that the temperature at a certain point on a photovoltaic device falls between 55℃ and 62℃.

[0079] Step S4: Determine the temperature prediction value distribution map based on the initial temperature prediction value set and the measured temperature data of the photovoltaic equipment, and compare the temperature prediction results with the abnormal temperature threshold to obtain the temperature anomaly risk points;

[0080] Step S4 further includes steps S41-S44, which specifically include:

[0081] Step S41: Perform dynamic weight calculation based on the initial temperature prediction value set and the measured temperature data of the photovoltaic equipment to obtain the dynamic adaptive weight coefficient;

[0082] In this step, the specific calculation process of the dynamic adaptive weight coefficients includes:

[0083]

[0084] In the above formula, and These represent the dynamic adaptive weight coefficients of the i-th twin model at the current time t and the previous time t-1, respectively; This represents the smoothing factor, with a value ranging from 0.9 to 0.99. Let represent the prediction residual of the i-th twin model at time t; This is an immediate performance evaluation term for the current prediction residuals. When the model's prediction is highly accurate at the current moment, this term is close to 1, which will positively adjust the weights. When the prediction error is large, this term is close to 0, which will significantly reduce the model's weights. Using this formula for dynamic weight allocation enables the system to adapt to environmental changes and photovoltaic equipment aging, automatically selecting the most robust models and giving the digital twin system online self-learning capabilities.

[0085] Step S42: Perform weighted fusion processing based on the dynamic adaptive weight coefficients and the initial temperature prediction value set to obtain a temperature prediction value distribution map;

[0086] Step S43: Calculate the dynamic anomaly threshold based on the temperature prediction value distribution map, the meteorological data, and the rated parameters of the photovoltaic equipment to obtain the dynamic anomaly temperature threshold distribution map;

[0087] In this step, the dynamic anomaly threshold calculation process specifically includes:

[0088]

[0089] In the above formula, Indicates the dynamic abnormal temperature threshold; Indicates ambient temperature; Indicates solar irradiance; This indicates the solar radiation absorption rate of the surface of a photovoltaic device; Indicates photoelectric conversion efficiency; Indicates the combined convective and radiative heat transfer coefficient; This indicates the statistical safety margin. It should be noted that... It is the sum of the convective heat transfer coefficient and the radiative heat transfer coefficient, where the convective heat transfer coefficient is... pass It is determined that a and b are empirical constants whose values ​​are related to the flow regime and the geometric characteristics of the photovoltaic device surface; the radiative heat transfer coefficient... pass Sure, This represents the Stefan-Boltzmann constant. Indicates the emissivity of the surface of a photovoltaic device. The characteristic temperature is the average of the absolute surface temperature of the photovoltaic device and the absolute ambient temperature.

[0090] Understandably, this step calculates the upper limit of temperature that the photovoltaic equipment can reach under normal conditions, given the current irradiance, wind speed, and ambient temperature. The threshold automatically increases under strong midday sunlight and decreases in the early morning or when wind speeds are high. This dynamic threshold effectively overcomes the drawbacks of fixed thresholds, such as high false alarm rates or missed alarms under changing operating conditions.

[0091] Step S44: Determine the temperature anomaly risk points based on the dynamic abnormal temperature threshold distribution map and the temperature prediction value distribution map.

[0092] Understandably, the dynamic abnormal temperature threshold distribution map and the temperature prediction value distribution map are compared element by element. Spatial points where the predicted temperature value exceeds the local dynamic threshold are initially marked as temperature anomaly risk points. In addition, this step introduces a filtering rule: only when the predicted temperature exceeds the threshold for more than 3 time steps can it be confirmed as a valid risk point, in order to avoid misjudgment caused by instantaneous fluctuations.

[0093] Step S5: Process the temperature prediction value distribution map, the temperature anomaly risk points, and the meteorological data to obtain photovoltaic equipment temperature anomaly early warning information.

[0094] Step S5 further includes steps S51-S54, which specifically include:

[0095] Step S51: Correct the temperature prediction distribution map and meteorological data to obtain the corrected temperature prediction distribution map;

[0096] This step refines the prediction results by introducing an equivalent sky cold source model and dew point temperature. Specifically, cloud cover and dew point temperature are obtained from meteorological data. The equivalent sky cold source model treats the sky as a cold source with equivalent radiation temperature, lower than the ambient air temperature, especially on clear nights when this radiative cooling effect is significant. The model calculates long-wave radiation heat transfer between the module surface and the sky, correcting the predicted temperature dominated by solar radiation and convection. This correction more accurately reflects the actual cooling process of the modules, particularly at night or during periods without sunlight. Simultaneously, it determines whether the module surface temperature is close to or below the dew point temperature. If close, it indicates potential condensation on the surface. The evaporation of this condensation absorbs a significant amount of heat, generating an additional cooling effect. This is further corrected by introducing an unsteady-state mass transfer model to quantify this evaporation heat absorption process. This step addresses the complex radiation and humidity conditions in the natural environment of photovoltaic power plants, overcoming the limitations of traditional thermal models that only focus on convection and solar radiation, and significantly improving the accuracy of temperature predictions in specific scenarios such as early morning, nighttime, or high humidity.

[0097] It should be noted that the equivalent sky cold source model is as follows:

[0098]

[0099] In the above formula, Indicates the net radiative heat transfer rate; This represents the Stefan-Boltzmann constant; Indicates the emissivity of the surface of a photovoltaic device; Indicates the thermodynamic temperature of the photovoltaic device surface; It represents the effective radiation temperature of the sky. It is not a constant value; it is significantly affected by factors such as atmospheric water vapor content and cloud cover. This is crucial for the model to adapt to complex weather conditions. The specific calculation process is as follows:

[0100]

[0101] In the above formula, Indicates the effective emissivity of the sky under clear sky conditions, through The calculation is performed, where m and n both represent empirical coefficients. It is the water vapor pressure; Both and n represent empirical coefficients for cloud cover correction, determined through regression analysis using local historical meteorological data and radiation observation data; C represents total cloud cover.

[0102] Step S52: Determine the temperature anomaly area in the corrected temperature prediction distribution map;

[0103] Step S53: Perform anomaly diagnosis based on the temperature anomaly area and the virtual stain distribution map to obtain the diagnosis result;

[0104] In this step, the temperature anomaly area is spatially overlaid with the virtual stain distribution map for analysis. The proportion of pixels that are also in the high stain area within the temperature anomaly area and the proportion of temperature anomalies in the high stain area are calculated. If both are greater than 80%, it can be inferred that the anomaly is highly likely to be caused by local hot spots due to stain accumulation.

[0105] Step S54: Generate an early warning message for abnormal temperature of photovoltaic equipment based on the diagnostic results.

[0106] In this step, the photovoltaic equipment temperature anomaly early warning information is a structured report, which includes the geographical location or number of the abnormal photovoltaic equipment, risk level, predicted temperature value, possible causes, etc. The early warning information realizes a closed loop from prediction, diagnosis to decision support.

[0107] Example 2:

[0108] This embodiment provides a photovoltaic equipment temperature anomaly prediction device based on a digital twin model. The device includes an acquisition module, a first processing module, a second processing module, a third processing module, and a fourth processing module, specifically comprising:

[0109] The acquisition module is used to acquire environmental acoustic signals, meteorological data, and measured temperature data of photovoltaic equipment.

[0110] The first processing module is used to extract acoustic spectrum features based on the environmental acoustic signal to obtain a virtual environmental parameter set, which includes a virtual wind speed distribution map and a virtual stain distribution map.

[0111] The second processing module is used to send the meteorological data and the virtual environment parameter set to a preset digital twin model for processing to obtain an initial temperature prediction value set;

[0112] The third processing module is used to determine the temperature prediction value distribution map based on the initial temperature prediction value set and the measured temperature data of the photovoltaic equipment, and compare the temperature prediction results with the abnormal temperature threshold to obtain the temperature abnormality risk points.

[0113] The fourth processing module is used to process the temperature prediction distribution map, the temperature anomaly risk points, and the meteorological data to obtain photovoltaic equipment temperature anomaly early warning information.

[0114] In one specific embodiment of this disclosure, the first processing module further includes a first processing unit, a second processing unit, a third processing unit, and a fourth processing unit, specifically including:

[0115] The first processing unit is used to enhance the environmental acoustic signal to obtain a feature vector with enhanced acoustic features.

[0116] The second processing unit is used to invert the surface wind speed of the photovoltaic device based on the feature vector of the enhanced acoustic features to obtain a virtual wind speed distribution map.

[0117] The third processing unit is used to predict the thickness of the stains on the surface of the photovoltaic equipment based on the virtual wind speed distribution map and the feature vector of the enhanced acoustic features, so as to obtain the virtual stain distribution map.

[0118] The fourth processing unit is used to construct a virtual environmental parameter set based on the virtual wind speed distribution map and the virtual stain distribution map.

[0119] In one specific embodiment of this disclosure, the second processing unit further includes a fifth processing unit, a sixth processing unit, a seventh processing unit, and an eighth processing unit, specifically including:

[0120] The fifth processing unit is used to perform acoustic coherence feature extraction processing on the feature vector of the enhanced acoustic features to obtain a frequency band energy matrix, wherein the frequency band energy matrix includes the spatial coherence corresponding to the frequency band.

[0121] The sixth processing unit is used to perform forward modeling based on the frequency band energy matrix to obtain a forward model;

[0122] The seventh processing unit is used to solve the surface friction wind speed according to the forward model and the frequency band energy matrix to obtain the surface friction wind speed field.

[0123] The eighth processing unit is used to determine a virtual wind speed distribution map based on the surface friction wind speed field.

[0124] In one specific embodiment of this disclosure, the third processing unit further includes a ninth processing unit, a tenth processing unit, an eleventh processing unit, and a twelfth processing unit, specifically comprising:

[0125] The ninth processing unit is used to perform secondary separation processing of particle impact sound and wind speed background sound based on the virtual wind speed distribution map and the feature vector of the enhanced acoustic features to obtain acoustic energy features;

[0126] The tenth processing unit is used to perform normalization processing based on the acoustic energy characteristics and the virtual wind speed distribution map to obtain the particle impact acoustic energy distribution after wind speed normalization.

[0127] The eleventh processing unit is used to perform inversion processing based on the particle impact acoustic energy distribution after wind speed normalization to obtain stain thickness information.

[0128] The twelfth processing unit is used to perform spatial interpolation processing based on the stain thickness information to obtain a virtual stain distribution map.

[0129] It should be noted that the specific manner in which each module performs its operation in the apparatus described in the above embodiments has been described in detail in the embodiments of the method, and will not be elaborated here.

[0130] Example 3:

[0131] Corresponding to the above method embodiments, this embodiment also provides a photovoltaic equipment temperature anomaly prediction device based on a digital twin model. The photovoltaic equipment temperature anomaly prediction device based on a digital twin model described below and the photovoltaic equipment temperature anomaly prediction method based on a digital twin model described above can be referred to each other.

[0132] Figure 2 This is a block diagram illustrating a photovoltaic device temperature anomaly prediction device 800 based on an exemplary embodiment. Figure 2 As shown, the photovoltaic equipment temperature anomaly prediction device 800 of the digital twin model may include: a processor 801 and a memory 802. The photovoltaic equipment temperature anomaly prediction device 800 of the digital twin model may also include one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.

[0133] The processor 801 controls the overall operation of the photovoltaic equipment temperature anomaly prediction device 800 in the digital twin model to complete all or part of the steps in the aforementioned photovoltaic equipment temperature anomaly prediction method in the digital twin model. The memory 802 stores various types of data to support the operation of the photovoltaic equipment temperature anomaly prediction device 800 in the digital twin model. This data may include, for example, instructions for any application or method operating on the photovoltaic equipment temperature anomaly prediction device 800 in the digital twin model, as well as application-related data such as contact data, sent and received messages, images, audio, video, etc. The memory 802 can be implemented using any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touchscreen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signals may be further stored in the memory 802 or transmitted via the communication component 805. The audio component also includes at least one speaker for outputting audio signals. I / O interface 804 provides an interface between processor 801 and other interface modules, such as keyboards, mice, and buttons. These buttons can be virtual or physical. Communication component 805 is used for wired or wireless communication between the photovoltaic device temperature anomaly prediction device 800 of the digital twin model and other devices. Wireless communication includes, for example, Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, or 4G, or a combination thereof. Therefore, the corresponding communication component 805 may include a Wi-Fi module, a Bluetooth module, and an NFC module.

[0134] In an exemplary embodiment, the photovoltaic equipment temperature anomaly prediction device 800 of the digital twin model can be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the photovoltaic equipment temperature anomaly prediction method of the digital twin model described above.

[0135] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, which, when executed by a processor, implement the steps of the above-described method for predicting temperature anomalies in photovoltaic devices using a digital twin model. For example, the computer-readable storage medium may be the memory 802 including the program instructions, which may be executed by the processor 801 of the photovoltaic device temperature anomaly prediction device 800 using a digital twin model to complete the above-described method for predicting temperature anomalies in photovoltaic devices using a digital twin model.

[0136] Example 4:

[0137] Corresponding to the above method embodiments, this embodiment also provides a readable storage medium. The readable storage medium described below can be referred to in conjunction with the photovoltaic equipment temperature anomaly prediction method based on a digital twin model described above.

[0138] A readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the steps of the photovoltaic equipment temperature anomaly prediction method based on the digital twin model of the above method embodiment are implemented.

[0139] Specifically, the readable storage medium can be a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, or any other readable storage medium capable of storing program code.

[0140] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0141] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for predicting temperature anomalies in photovoltaic equipment based on a digital twin model, characterized in that, include: Acquire environmental acoustic signals, meteorological data, and measured temperature data of photovoltaic equipment; Based on the environmental acoustic signal, acoustic spectrum features are extracted to obtain a virtual environmental parameter set, which includes a virtual wind speed distribution map and a virtual stain distribution map. The meteorological data and the set of virtual environmental parameters are sent to a preset digital twin model for processing to obtain an initial set of temperature prediction values; A temperature prediction distribution map is determined based on the initial temperature prediction set and the measured temperature data of the photovoltaic equipment. The temperature prediction results are then compared with the abnormal temperature threshold to obtain temperature anomaly risk points. Based on the temperature prediction distribution map, the temperature anomaly risk points, and the meteorological data, an early warning information for abnormal temperature of photovoltaic equipment is obtained.

2. The method for predicting temperature anomalies in photovoltaic equipment based on a digital twin model according to claim 1, characterized in that, Based on the environmental acoustic signals, acoustic spectral features are extracted to obtain a virtual environment parameter set, including: The environmental acoustic signal is enhanced to obtain a feature vector with enhanced acoustic features; Based on the feature vector of the enhanced acoustic features, the surface wind speed of the photovoltaic device is inverted to obtain a virtual wind speed distribution map. The thickness of the surface stains on the photovoltaic equipment is predicted based on the virtual wind speed distribution map and the feature vector of the enhanced acoustic features, thus obtaining the virtual stain distribution map; A virtual environmental parameter set is constructed based on the virtual wind speed distribution map and the virtual stain distribution map.

3. The method for predicting temperature anomalies in photovoltaic equipment based on a digital twin model according to claim 2, characterized in that, The surface wind speed of the photovoltaic device is inverted based on the feature vector of the enhanced acoustic features, including: The feature vectors of the enhanced acoustic features are subjected to acoustic coherence feature extraction processing to obtain a frequency band energy matrix, wherein the frequency band energy matrix includes the spatial coherence corresponding to the frequency band; Forward modeling is performed based on the frequency band energy matrix to obtain the forward model; The surface friction wind speed field is obtained by solving the surface friction wind speed according to the forward model and the frequency band energy matrix. A virtual wind speed distribution map is determined based on the surface friction wind speed field.

4. The method for predicting temperature anomalies in photovoltaic equipment based on a digital twin model according to claim 2, characterized in that, The thickness of surface contaminants on photovoltaic equipment is predicted based on the virtual wind speed distribution map and the feature vector of the enhanced acoustic features, including: Based on the virtual wind speed distribution map and the feature vector of the enhanced acoustic features, a secondary separation process is performed between particle impact sound and wind speed background sound to obtain acoustic energy features. Based on the acoustic energy characteristics and the virtual wind speed distribution map, normalization processing is performed to obtain the particle impact acoustic energy distribution after wind speed normalization. The stain thickness information is obtained by inversion processing based on the particle impact acoustic energy distribution after wind speed normalization. Spatial interpolation is performed based on the stain thickness information to obtain a virtual stain distribution map.

5. The method for predicting temperature anomalies in photovoltaic equipment based on a digital twin model according to claim 1, characterized in that, The meteorological data and the virtual environmental parameter set are sent to a preset digital twin model for processing, including: Finite element modeling is performed based on the material properties and structural dimensions of photovoltaic equipment to obtain the finite element model. Based on the finite element model and historical operating data, posterior distribution estimation is performed to obtain the posterior probability distribution of key thermophysical parameters; Based on the posterior probability distribution of the key thermophysical parameters, a model population sampling process based on the optimal Latin hypercube experimental design is performed to obtain a set of virtual component model instances. A digital twin model is constructed based on the set of virtual component model instances; The temperature field evolution sequence of each model is obtained by solving the problem based on the digital twin model, the meteorological data, and the virtual environment parameter set. Based on the evolution sequence of the body temperature field of each model, statistical aggregation of the group output is performed to obtain the ensemble mean field and ensemble variance field of the temperature prediction. Based on the set mean field and set variance field of the temperature prediction, a probabilistic prediction set construction process is performed to obtain the initial temperature prediction value set and its distribution description.

6. The method for predicting temperature anomalies in photovoltaic equipment based on a digital twin model according to claim 1, characterized in that, The temperature prediction result is determined based on the initial set of predicted temperature values ​​and the measured temperature data of the photovoltaic equipment, and the temperature prediction result is compared with the abnormal temperature threshold, including: Based on the initial temperature prediction set and the measured temperature data of the photovoltaic equipment, dynamic weight calculation is performed to obtain dynamic adaptive weight coefficients. A temperature prediction distribution map is obtained by performing a weighted fusion process based on the dynamic adaptive weighting coefficients and the initial temperature prediction value set. Based on the temperature prediction distribution map, the meteorological data, and the rated parameters of the photovoltaic equipment, a dynamic anomaly threshold is calculated to obtain a dynamic anomaly temperature threshold distribution map. Temperature anomaly risk points are determined based on the dynamic abnormal temperature threshold distribution map and the temperature prediction value distribution map.

7. A photovoltaic equipment temperature anomaly prediction device based on a digital twin model, characterized in that, include: The acquisition module is used to acquire environmental acoustic signals, meteorological data, and measured temperature data of photovoltaic equipment. The first processing module is used to extract acoustic spectrum features based on the environmental acoustic signal to obtain a virtual environmental parameter set, which includes a virtual wind speed distribution map and a virtual stain distribution map. The second processing module is used to send the meteorological data and the virtual environment parameter set to a preset digital twin model for processing to obtain an initial temperature prediction value set; The third processing module is used to determine the temperature prediction value distribution map based on the initial temperature prediction value set and the measured temperature data of the photovoltaic equipment, and compare the temperature prediction results with the abnormal temperature threshold to obtain the temperature abnormality risk points. The fourth processing module is used to process the temperature prediction distribution map, the temperature anomaly risk points, and the meteorological data to obtain photovoltaic equipment temperature anomaly early warning information.

8. The photovoltaic equipment temperature anomaly prediction device based on a digital twin model according to claim 7, characterized in that, The first processing module includes: The first processing unit is used to enhance the environmental acoustic signal to obtain a feature vector with enhanced acoustic features. The second processing unit is used to invert the surface wind speed of the photovoltaic device based on the feature vector of the enhanced acoustic features to obtain a virtual wind speed distribution map. The third processing unit is used to predict the thickness of the stains on the surface of the photovoltaic equipment based on the virtual wind speed distribution map and the feature vector of the enhanced acoustic features, so as to obtain the virtual stain distribution map. The fourth processing unit is used to construct a virtual environmental parameter set based on the virtual wind speed distribution map and the virtual stain distribution map.

9. The photovoltaic equipment temperature anomaly prediction device based on a digital twin model according to claim 8, characterized in that, The second processing unit includes: The fifth processing unit is used to perform acoustic coherence feature extraction processing on the feature vector of the enhanced acoustic features to obtain a frequency band energy matrix, wherein the frequency band energy matrix includes the spatial coherence corresponding to the frequency band. The sixth processing unit is used to perform forward modeling based on the frequency band energy matrix to obtain a forward model; The seventh processing unit is used to solve the surface friction wind speed according to the forward model and the frequency band energy matrix to obtain the surface friction wind speed field. The eighth processing unit is used to determine a virtual wind speed distribution map based on the surface friction wind speed field.

10. The photovoltaic equipment temperature anomaly prediction device based on a digital twin model according to claim 8, characterized in that, The third processing unit includes: The ninth processing unit is used to perform secondary separation processing of particle impact sound and wind speed background sound based on the virtual wind speed distribution map and the feature vector of the enhanced acoustic features to obtain acoustic energy features; The tenth processing unit is used to perform normalization processing based on the acoustic energy characteristics and the virtual wind speed distribution map to obtain the particle impact acoustic energy distribution after wind speed normalization. The eleventh processing unit is used to perform inversion processing based on the particle impact acoustic energy distribution after wind speed normalization to obtain stain thickness information. The twelfth processing unit is used to perform spatial interpolation processing based on the stain thickness information to obtain a virtual stain distribution map.

Images