Photovoltaic module waste prediction method based on geographic clustering and installation prediction

Abstract

The invention discloses a photovoltaic module waste prediction method based on geographic clustering and installation prediction, belongs to the technical field of photovoltaic waste management, and solves the problem of low precision in predicting failure time and residual life of a photovoltaic array by an existing method. Constructing a sub-region life prediction model according to a geographic clustering analysis result, constructing a sub-region prediction installation model based on resource endowment and a historical growth contribution weighting method, obtaining the amount of newly increased photovoltaic installation waste in a photovoltaic module installation region every year, and predicting the waste accumulation amount of the photovoltaic module waste; according to the method, the dynamic photovoltaic module performance degradation model is constructed in combination with the environmental influence factor data and the module performance degradation data, and the sub-region prediction installation model is constructed in combination with the resource endowment and the historical growth contribution weighting method, so that the prediction precision and the region applicability are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of photovoltaic waste management, and relates to a method for predicting photovoltaic module waste based on geographical clustering and installed capacity prediction. Background Art

[0002] The rapid development of photovoltaic power generation technology has led to a sharp increase in photovoltaic installed capacity in a short period. However, the management and disposal of a large amount of photovoltaic waste that follows have become a major challenge. Improper treatment methods can cause environmental pollution and resource waste. To reduce the negative impact of photovoltaic module waste on the environment, reduce resource waste and manufacturing costs, a large number of studies have carried out waste prediction through the installation and retirement of photovoltaic modules, so as to assist in the recycling management of photovoltaic module waste.

[0003] At present, the prediction of photovoltaic module waste relies on the linear extrapolation of historical installed capacity, and uses a given average annual growth rate to predict the installed capacity in the next 12 years. It does not carry out prediction work by combining multi-dimensional factors such as local resource endowments (such as light resources, available land area, etc.). In addition, in these photovoltaic module waste prediction works, the traditional service life of photovoltaic modules is mostly based on a fixed number of years (such as 25 years), without considering the dynamic impact of environmental factors (such as temperature, humidity, irradiance, etc.) on the performance degradation of the modules. In fact, the performance degradation of the same module varies greatly in different regions.

[0004] Chinese Patent CN119416654A discloses a method, system, device, medium and program for predicting the photovoltaic life. The method includes collecting historical electrical data and meteorological data of a photovoltaic array in a photovoltaic power station, and preprocessing the historical electrical data and meteorological data to obtain an analysis result of the historical degradation trend of the photovoltaic array; constructing a physical degradation model of the photovoltaic power station based on the analysis result of the historical degradation trend of the photovoltaic array; using the physical degradation model of the photovoltaic power station to predict the failure time and remaining life of the photovoltaic array. However, when the existing method uses the physical degradation model of the photovoltaic power station to predict the failure time and remaining life of the photovoltaic array, due to the large difference in the performance degradation trajectories of photovoltaic modules in different regional environments, it is difficult to ensure the applicability of the physical degradation model, resulting in low prediction life accuracy and poor regional adaptability of the distribution of photovoltaic module waste. In view of the above problems, we propose a method for predicting photovoltaic module waste based on geographical clustering and installed capacity prediction. Summary of the Invention

[0005] The object of the present invention is to provide a prediction method for photovoltaic module waste based on geographical clustering and installed capacity prediction in view of the deficiencies of the prior art, and to solve the problems that when the existing method uses the physical degradation model of a photovoltaic power station to predict the failure time and remaining life of a photovoltaic array, due to the large differences in the performance degradation trajectories of photovoltaic modules in different regional environments, it is difficult for the physical degradation model to ensure applicability, resulting in low prediction accuracy of the distribution life of photovoltaic module waste and poor regional adaptability.

[0006] The present invention is implemented as follows. A prediction method for photovoltaic module waste based on geographical clustering and installed capacity prediction, the prediction method for photovoltaic module waste based on geographical clustering and installed capacity prediction includes: Collect environmental impact factor data and component performance degradation data within the regional cycle time, preprocess the environmental impact factor data and component performance degradation data, and construct a photovoltaic module performance degradation model based on the preprocessed environmental impact factor data and component performance degradation data; Load the constructed photovoltaic module performance degradation model, analyze the constructed photovoltaic module performance degradation model, obtain the action weights of environmental impact factors during the component aging process, perform geographical clustering analysis on the photovoltaic module installation areas based on the action weights of environmental impact factors and the clustering algorithm, and construct a sub-region life prediction model according to the geographical clustering analysis results; Collect the historical photovoltaic installed capacity data of each sub-region in the photovoltaic module installation area, construct a sub-region predicted installed capacity model based on the resource endowment and historical growth contribution weighting method, and output the sub-region predicted installed capacity; Load the sub-region life prediction model constructed according to the geographical clustering analysis results, and establish a failure distribution function of the sub-region photovoltaic modules based on the sub-region life prediction model; Based on the historical photovoltaic installed capacity data and sub-region predicted installed capacity of each sub-region, combine with the market supply A model to obtain the annual new photovoltaic installed waste volume in the photovoltaic module installation area; Load the annual new photovoltaic installed waste volume in the photovoltaic module installation area, and predict the cumulative waste volume of photovoltaic module waste based on the annual new photovoltaic installed waste volume in the photovoltaic module installation area.

[0007] When constructing the photovoltaic module performance degradation model, quantify the cumulative damage effects of temperature cycle, humidity, ultraviolet radiation, and mechanical load environmental stresses on the component life. The photovoltaic module performance degradation model is expressed as: Among them, represents the performance value of the photovoltaic module, is the performance loss value of the photovoltaic module, is the intensity factor of performance degradation, 、 and represents the exponential constant, is a metric for the degradation of photovoltaic module performance caused by non-primary influencing factors, is the activation energy, representing the comprehensive energy of the photovoltaic module performance degradation reaction, is the Boltzmann constant, taking 8.62×10 −5 ev / k, is the absolute temperature, is the ambient temperature, where Kelvin temperature is used here , is the relative humidity of the component service environment, and respectively represent the highest temperature and the lowest temperature in a temperature cycle. The highest temperature is the backplane temperature of the component at the highest ambient temperature, and the lowest temperature takes the lowest ambient temperature, is the incident solar irradiance, is the wind speed measured at a standard height of 10 m above the ground, , are empirical coefficients determined by the type of photovoltaic module, is the ultraviolet irradiance reaching the ground, and are the irradiances of long-wave ultraviolet rays and medium-wave ultraviolet rays respectively.

[0008] When performing geographical clustering analysis on the photovoltaic module installation area based on the action weights of environmental impact factors and the clustering algorithm, the clustering algorithm is the K-means or hierarchical clustering algorithm.

[0009] The method for constructing a sub-region prediction installed capacity model based on resource endowment and historical growth contribution weighting method includes: Loading the photovoltaic historical installed capacity data of each sub-region in the photovoltaic module installation area and normalizing the photovoltaic historical installed capacity data of each sub-region; Decomposing the total installed capacity of the target year in the photovoltaic module installation area into sub-region growth targets; Based on the resource endowment and historical growth contribution weighting method, combining the photovoltaic historical installed capacity data of each sub-region with the decomposed sub-region growth targets to construct a sub-region prediction installed capacity model and output the sub-region prediction installed capacity.

[0010] The sub-region prediction installed capacity model is a dynamic K-value Logistic model, and the dynamic K-value Logistic model is expressed as: Among them, is the installed capacity allocation weight of each sub-region under the photovoltaic installed capacity target in the photovoltaic module installation area; is an adjustable parameter used to balance the resource potential proportion weight and the historical growth contribution rate; is the contribution rate of each sub-region to the growth of the installed photovoltaic capacity in the photovoltaic module installation area, and is the proportion weight of the solar energy resource utilization potential.

[0011] When establishing the failure distribution function of the sub-region photovoltaic module based on the sub-region life prediction model, the failure distribution function of the sub-region photovoltaic module is expressed as: wherein, and are the probability density function and the cumulative distribution function; is the time independent variable, representing the time when the module has been put into use; is the shape parameter, reflecting the failure distribution of the photovoltaic module performance degradation model at time; is the scale parameter, reflecting the uniformity and breadth of the degradation on the time scale.

[0012] The annual newly added photovoltaic installed waste volume in the photovoltaic module installation area obtained by combining the market supply A model is calculated by the following formula: wherein, is the photovoltaic waste generation volume in the nth year; is the photovoltaic installed capacity in the ith year before this; is the proportion of the installed capacity in the ith year that fails in the nth year, obtained from the Weibull distribution function.

[0013] When predicting the cumulative waste volume of photovoltaic modules based on the annual newly added photovoltaic installed waste volume in the photovoltaic module installation area, obtaining the evolution of the market share of each technical type of module and the component mass of each technical type of module, the cumulative waste volume is calculated by the following formula: where is the cumulative waste mass of the mth valuable component in the nth year, is the market share of the jth technical type in the ith year, is the mass power conversion ratio of the jth photovoltaic module, is the mass proportion of the mth component of the jth technical type.

[0014] Compared with the prior art, the embodiments of the present application mainly have the following beneficial effects: In the embodiments of the present invention, by combining environmental impact factor data and component performance degradation data, a dynamic photovoltaic component performance degradation model is constructed. Through geographical clustering analysis combined with the photovoltaic component performance degradation model, the installation areas of photovoltaic components are divided into several sub-regions with similar environmental characteristics, and a differential life prediction model is constructed for each sub-region, which can reflect the performance changes of components in different environments, avoiding the limitations of the fixed-year assumption. At the same time, a sub-region prediction installed capacity model is constructed by combining resource endowment and historical growth contribution weighting method, which not only considers historical data but also combines local resource endowment, improving the prediction accuracy and regional applicability, and can be used to predict the spatio-temporal distribution of crystalline silicon photovoltaic component waste across the country. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 FIG. is a schematic implementation flowchart of a photovoltaic component waste prediction method based on geographical clustering and installed capacity prediction provided by the present invention.

[0016] Figure 2 FIG. shows a schematic flowchart of a particle swarm optimization algorithm.

[0017] Figure 3 FIG. shows a flowchart of a K-Means clustering algorithm. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs; the terms used in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above drawings are intended to cover non-exclusive inclusion. The terms "first", "second", etc. in the specification and claims of this application or the above drawings are used to distinguish different objects and not to describe a specific order.

[0019] When existing methods use the physical degradation model of a photovoltaic power station to predict the failure time and remaining life of a photovoltaic array, due to the large differences in the performance degradation trajectories of photovoltaic modules in different regional environments, it is difficult for the physical degradation model to ensure applicability, resulting in the defects of low prediction accuracy of the distribution prediction life of photovoltaic module waste and poor regional adaptability. To address the above problems, we propose a method for predicting photovoltaic module waste based on geographical clustering and installed capacity prediction. Briefly, when implementing the method, first collect environmental impact factor data and component performance degradation data within the regional cycle time, construct a photovoltaic module performance degradation model based on the preprocessed environmental impact factor data and component performance degradation data, perform geographical clustering analysis on the photovoltaic module installation areas based on the action weights of environmental impact factors and clustering algorithms, construct a sub-region life prediction model according to the results of the geographical clustering analysis, construct a sub-region predicted installed capacity model based on the resource endowment and historical growth contribution weighting method, output the sub-region predicted installed capacity, and establish a failure distribution function for the photovoltaic modules in the sub-region based on the sub-region life prediction model; based on the historical photovoltaic installed capacity data and sub-region predicted installed capacity of each sub-region, combine with the market supply A model to obtain the annual newly added photovoltaic installed waste volume in the photovoltaic module installation area; load the annual newly added photovoltaic installed waste volume in the photovoltaic module installation area, and finally predict the cumulative waste volume of photovoltaic module waste based on the annual newly added photovoltaic installed waste volume in the photovoltaic module installation area. In the embodiments of the present invention, by combining environmental impact factor data and component performance degradation data, a dynamic photovoltaic module performance degradation model is constructed. Through geographical clustering analysis combined with the photovoltaic module performance degradation model, the photovoltaic module installation area is divided into several sub-regions with similar environmental characteristics, and a differentiated life prediction model is constructed for each sub-region, which can reflect the performance changes of the components in different environments, avoiding the limitations of the fixed-year assumption. At the same time, a sub-region predicted installed capacity model is constructed by combining the resource endowment and historical growth contribution weighting method, which not only considers historical data but also combines the local resource endowment, improving the prediction accuracy and regional applicability, and can be used to predict the spatio-temporal distribution of crystalline silicon photovoltaic module waste across the country.

[0020] An embodiment of the present invention provides a method for predicting photovoltaic module waste based on geographical clustering and installed capacity prediction. Figure 1 The schematic diagram of the implementation process of the method for predicting photovoltaic module waste based on geographical clustering and installed capacity prediction is shown. The method for predicting photovoltaic module waste based on geographical clustering and installed capacity prediction includes: S10, collect environmental impact factor data and component performance degradation data within the regional cycle time, preprocess the environmental impact factor data and component performance degradation data, and construct a photovoltaic module performance degradation model based on the preprocessed environmental impact factor data and component performance degradation data; It should be noted that the cycle time can be three years, two years, one year, one quarter, or one month, and the environmental impact factor data includes but is not limited to temperature cycling (temperature has a significant impact on the performance of photovoltaic modules, and high temperature will cause a decrease in module efficiency), humidity (humidity will affect the encapsulation material of the module, which may lead to corrosion and performance degradation), ultraviolet irradiation (ultraviolet irradiation will accelerate the aging of module materials and affect the long-term performance of the module), mechanical load (mechanical loads such as wind and snow load may cause damage to the module structure and affect performance), and light intensity (light intensity directly affects the power generation efficiency of photovoltaic modules, and insufficient light will result in a decrease in power generation).

[0021] The preprocessing methods for the environmental impact factor data and the module performance degradation data can be data cleaning, data denoising, feature extraction, and data normalization processing.

[0022] It should be noted that when constructing the photovoltaic module performance degradation model, the cumulative damage effects of temperature cycling, humidity, ultraviolet irradiation, and mechanical load environmental stresses on the module life are quantified, and the photovoltaic module performance degradation model is expressed as: Wherein, represents the performance value of the photovoltaic module, is the performance loss value of the photovoltaic module, is the intensity factor of performance degradation, 、 and represent exponential constants, is the measure of the performance degradation of the photovoltaic module caused by non-primary influencing factors, is the activation energy, representing the comprehensive energy of the photovoltaic module performance degradation reaction, is the Boltzmann constant, taking 8.62×10 −5 ev / k, is the absolute temperature, is the environmental temperature, and the Kelvin temperature is used here , is the relative humidity of the module service environment, and respectively represent the highest temperature and the lowest temperature in a temperature cycle. The highest temperature is the backplane temperature of the module under the highest environmental temperature, and the lowest temperature is taken as the lowest environmental temperature, is the incident solar irradiance, is the wind speed measured at a standard height of 10m above the ground, 、 are empirical coefficients determined by the type of photovoltaic module, is the ultraviolet irradiance reaching the ground, and They are the irradiances of UVA and UVB respectively.

[0023] S20: Load the established photovoltaic module performance degradation model. Analyze the photovoltaic module performance degradation model to obtain the action weights of environmental impact factors during the component aging process. Based on the action weights of environmental impact factors and the clustering algorithm, conduct geographical clustering analysis on the installation areas of photovoltaic modules. Construct a sub-region life prediction model according to the results of the geographical clustering analysis. When conducting geographical clustering analysis on the installation areas of photovoltaic modules based on the action weights of environmental impact factors and the clustering algorithm, the clustering algorithm is K-means or hierarchical clustering algorithm.

[0024] In the embodiments of the present invention, the particle swarm optimization algorithm can also be introduced into the sub-region life prediction model to optimize the model parameters and the failure distribution function. The particle swarm optimization algorithm (PSO) is a bionic optimization algorithm that uses the velocity and position of particles to simulate the foraging behavior of bird flocks. In the algorithm, each particle searches for the optimal solution in the solution space through the cooperation and information sharing of individuals (particles) in the group, that is, obtains the optimal fitness value corresponding to the objective function. Each particle dynamically adjusts its flight direction and speed according to its own historical optimal position and the group historical optimal position. Figure 2 It shows a schematic diagram of the particle swarm optimization algorithm process. The particle swarm optimization algorithm is introduced into the sub-region life prediction model to optimize the model parameters and the failure distribution function. By continuously adjusting these parameters, the model can more accurately fit the life change law of photovoltaic modules in different sub-regions, thereby improving the accuracy of life prediction and providing a more reliable basis for the maintenance and management of photovoltaic modules. The life of photovoltaic modules is affected by various complex factors, and the environmental conditions and component characteristics in different sub-regions may vary. The particle swarm optimization algorithm can adaptively adjust the search strategy according to different data distributions and problem characteristics, making the model have stronger adaptability and generalization ability, and can effectively predict the life of photovoltaic modules in various complex actual scenarios.

[0025] It should be noted that the installation areas of photovoltaic modules can be the geographical areas of 31 provinces and regions in the country except Hong Kong, Macau and Taiwan of China. The sub-regions constitute the installation areas of photovoltaic modules. The sub-regions can be provinces and cities. Based on the action weights of environmental impact factors and the clustering algorithm, this application conducts geographical clustering analysis on the installation areas of photovoltaic modules, obtains the optimal number of clusters under the clustering index, and divides 31 provinces in the country into several regions with similar environmental characteristics. Each region corresponds to a differentiated regional life prediction model, thereby constructing a sub-region life prediction model, which can overcome the problems of large geographical area, wide longitude and latitude span, multiple climate environments in the country, and large differences in the performance degradation trajectories of photovoltaic modules in different environments.

[0026] Meanwhile, it should be noted that when the clustering algorithm is K-means, the K-Means clustering method is divided into four stages. The first step is initialization: randomly select k initial centroids. The second step is the assignment stage: calculate the Euclidean distance between each data point and all centroids, and assign it to the cluster corresponding to the nearest centroid. The third step is the update stage: recalculate the centroid of each cluster as the mean value of the samples within the current cluster. The fourth step is iteration termination: repeatedly perform the second and third steps until the end condition is met, such as the change in the centroid position is lower than the preset threshold or the maximum number of iterations is reached. The flowchart of the K-Means clustering algorithm is as Figure 3 shown.

[0027] In the embodiment of the present invention, a performance degradation prediction model applicable to this area is constructed based on the performance of existing photovoltaic modules in a power station in Guangdong Province and the set of influencing factors for the performance degradation of local photovoltaic modules. The obtained performance degradation prediction model is as follows. Based on this prediction model, the characteristic life of photovoltaic modules in Guangdong Province reaches 20.58 years. The performance degradation model of photovoltaic modules in this area is expressed as: S30. Collect the historical installed capacity data of each sub-region in the photovoltaic module installation area, construct a sub-region prediction installed capacity model based on the resource endowment and historical growth contribution weighting method, and output the sub-region predicted installed capacity; In the embodiment of the present invention, the method for constructing a sub-region prediction installed capacity model based on the resource endowment and historical growth contribution weighting method includes: S301. Load the historical installed capacity data of each sub-region in the photovoltaic module installation area, and perform normalization processing on the historical installed capacity data of each sub-region; By loading the historical installed capacity data of each sub-region, the input data of the model is ensured to be comprehensive and accurate. The historical installed capacity data is an important basis for predicting future installed capacity and can reflect the installed capacity growth trend of each sub-region. And performing normalization processing on the historical installed capacity data of each sub-region can eliminate the differences in data dimensions and magnitudes of different sub-regions, making the data comparable. The normalized data is more stable, which helps to improve the training effect and prediction accuracy of the model. In this embodiment, the sub-region can be a provincial region.

[0028] S302. Decompose the total installed capacity of the target year in the photovoltaic module installation area into sub-region growth targets; In the embodiment of the present invention, by decomposing the total installed capacity of the target year into sub-region growth targets, the installed capacity growth targets that each sub-region needs to achieve in the future can be clarified. This decomposition method makes the prediction more specific and targeted, which helps each sub-region formulate a reasonable installed capacity plan. The comprehensive weight of each sub-region can be calculated according to the historical installed capacity data and resource endowment of each sub-region, and then the total installed capacity can be allocated to each sub-region according to the weight.

[0029] S303. Based on the resource endowment and historical growth contribution weighting method, combine the historical installed capacity data of photovoltaic in each sub-region with the decomposed sub-region growth target to construct a sub-region prediction installed capacity model, and output the predicted installed capacity of the sub-region.

[0030] By combining the resource endowment and historical growth contribution weighting method, the model can comprehensively consider the resource endowment and historical installed capacity growth of each sub-region, improving the scientificity and accuracy of the prediction. The resource endowment reflects the natural conditions and policy support of each sub-region, while the historical growth contribution reflects the installed capacity growth trend of each sub-region.

[0031] Among them, the sub-region prediction installed capacity model is a dynamic K-value Logistic model, and the dynamic K-value Logistic model is expressed as: Among them, is the installed capacity allocation weight of each sub-region under the photovoltaic installed capacity target of the photovoltaic module installation area; is an adjustable parameter used to balance the resource potential ratio weight and the historical growth contribution rate; is the contribution rate of each sub-region to the growth of the photovoltaic installed capacity in the photovoltaic module installation area, is the solar energy resource utilization potential ratio weight.

[0032] S40. Load the sub-region life prediction model constructed according to the geographical clustering analysis results, and establish the failure distribution function of the sub-region photovoltaic modules based on the sub-region life prediction model; In the embodiment of the present invention, when establishing the failure distribution function of the sub-region photovoltaic modules based on the sub-region life prediction model, the failure distribution function of the sub-region photovoltaic modules is expressed as: Among them, and probability density function and cumulative distribution function; is the time independent variable, representing the time when the component has been put into use; is the shape parameter, reflecting the failure distribution of the photovoltaic module performance degradation model at moment; is the scale parameter, reflecting the uniformity and breadth of the degradation on the time scale.

[0033] In the embodiments of the present invention, the failure distribution function describes the failure probability of a photovoltaic module within a specific time through the probability density function (PDF) and the cumulative distribution function (CDF). The shape parameter and the scale parameter in the failure distribution function can reflect the degradation characteristics of the photovoltaic module: among them, the shape parameter reflects the failure distribution of the photovoltaic module performance degradation model at a specific moment, that is, the shape of the degradation process. Different shape parameters mean different degradation modes, such as linear degradation, exponential degradation, etc. The scale parameter reflects the uniformity and breadth of the degradation on the time scale, that is, the speed of the degradation process. The larger the scale parameter, the slower the degradation process and the longer the service life of the photovoltaic module. Through the failure distribution function, the failure time of the photovoltaic module can be predicted, providing a scientific basis for the replacement and maintenance of the photovoltaic module. For example, the failure probability within a specific time can be calculated, or the time point when the failure probability reaches a certain threshold can be found.

[0034] S50. Based on the historical photovoltaic installation data of each sub-region and the predicted installation volume of the sub-region, combined with the market supply A model, obtain the annual newly added photovoltaic installation waste volume in the photovoltaic module installation area; S60. Load the annual newly added photovoltaic installation waste volume in the photovoltaic module installation area, and predict the cumulative waste volume of the photovoltaic module waste based on the annual newly added photovoltaic installation waste volume in the photovoltaic module installation area.

[0035] In the embodiments of the present invention, by combining the environmental impact factor data and the component performance degradation data, a dynamic photovoltaic module performance degradation model is constructed. Through geographical clustering analysis combined with the photovoltaic module performance degradation model, the photovoltaic module installation area is divided into several sub-regions with similar environmental characteristics, and a differential life prediction model is constructed for each sub-region, which can reflect the performance changes of the components under different environments, avoiding the limitations of the fixed-year assumption. At the same time, a sub-region predicted installation model is constructed by combining the resource endowment and the historical growth contribution weighting method, which not only considers the historical data but also combines the local resource endowment, improving the prediction accuracy and regional applicability, and can be used to predict the spatio-temporal distribution of crystalline silicon photovoltaic module waste across the country.

[0036] In the embodiments of the present invention, the annual newly added photovoltaic installation waste volume in the photovoltaic module installation area obtained by combining the market supply A model is calculated by the following formula: Among them, is the photovoltaic waste generation volume in the nth year; is the photovoltaic installation volume in the ith year before this; is the failure ratio of the installation volume in the ith year in the nth year, obtained from the Weibull distribution function.

[0037] In the embodiments of the present invention, by combining the market supply A model, the amount of newly added photovoltaic installation waste per year can be accurately calculated. The Weibull distribution function can accurately reflect the failure probability of photovoltaic modules, making the prediction results more scientific and reliable. The parameters in the Weibull distribution function can be dynamically adjusted according to actual data to adapt to the performance degradation of photovoltaic modules of different regions and different technology types.

[0038] When predicting the cumulative amount of photovoltaic module waste based on the amount of newly added photovoltaic installation waste per year in the photovoltaic module installation area, obtain the evolution of the market share of each technology type of module and the component mass of each technology type of module. The cumulative amount of waste is calculated by the following formula: Where is the cumulative waste mass of the nth year and the mth valuable component, is the market share of the jth technology type in the ith year, is the mass power conversion ratio of the jth photovoltaic module, is the mass ratio of the mth component of the jth technology type.

[0039] Specifically, based on the panel data of the main influencing factors, the photovoltaic module installation data and the resource endowment data of 31 provincial-level administrative regions in China, the clustering results with consistent performance degradation of photovoltaic modules are as follows, the service life of photovoltaic modules in each provincial-level administrative region, and the prediction of photovoltaic module installation and waste in 2050 are shown in the following table.

[0040] Table 1 Photovoltaic module installation and waste in 31 provincial-level administrative regions of China Provincial administrative region Characteristic life (years) Actual installed capacity in 2024 (GW) Predicted installed capacity in 2050 (GW) Predicted decommissioning quantity due to normal degradation in 2050 (GW) Predicted decommissioning quantity due to early degradation in 2050 (GW) Shandong Province 25.00 76.13 91.83 68.74 63.21 Hebei Province 25.00 72.02 95.75 69.00 64.51 Jiangsu Province 22.75 61.65 86.37 68.31 62.39 Xinjiang Uygur Autonomous Region 22.35 53.46 538.45 102.71 147.31 Inner Mongolia Autonomous Region 23.43 48.11 291.10 65.95 87.99 Zhejiang Province 22.75 47.28 80.26 59.37 55.61 Henan Province 24.00 43.49 68.49 50.53 46.94 Anhui Province 22.75 43.11 96.10 50.79 52.59 Guangdong Province 20.58 41.16 151.55 72.68 78.47 Yunnan Province 19.41 37.23 284.19 120.19 136.66 Qinghai Province 22.35 36.42 284.68 64.27 84.80 Hubei Province 22.75 35.10 147.49 54.59 64.04 Shanxi Province 24.00 34.77 88.15 38.98 42.50 Shaanxi Province 24.00 34.33 118.89 41.27 48.20 Gansu Province 22.35 31.39 166.42 52.10 62.82 Ningxia Hui Autonomous Region 23.43 26.24 52.17 33.76 33.22 Jiangxi Province 20.58 25.64 124.04 52.61 58.85 Guangxi Zhuang Autonomous Region 20.58 20.52 223.98 87.61 104.44 Guizhou Province 21.21 19.86 137.83 50.80 59.77 Hunan Province 20.58 18.73 164.63 59.92 71.94 Fujian Province 20.58 12.58 139.38 59.94 68.90 Liaoning Province 22.88 12.14 87.07 30.21 37.41 Sichuan Province 21.21 10.82 348.46 75.05 115.86 Hainan Province 19.41 7.41 123.01 57.15 63.81 Tianjin Municipality 25.00 7.24 78.58 31.36 39.13 Heilongjiang Province 22.88 7.17 103.56 24.16 34.63 Jilin Province 22.88 5.83 68.00 18.23 24.68 Shanghai Municipality 22.75 4.11 56.31 29.01 31.62 Tibet Autonomous Region 24.39 4.04 470.88 39.72 100.52 Chongqing Municipality 21.21 3.10 125.77 40.45 54.58 Beijing Municipality 25.00 1.30 40.37 11.56 17.15 In the embodiments of the present invention, the cumulative waste amount formula not only considers the amount of newly added photovoltaic installation waste per year, but also combines the market share, mass power conversion ratio and component mass ratio of each technology type of module, and can comprehensively analyze the waste situation of different technology types of modules. Through refined calculations, the cumulative waste mass of each valuable component can be accurately predicted, providing a scientific basis for the recycling and reuse of photovoltaic module waste.

[0041] In summary, the present invention provides a method for predicting photovoltaic module waste based on geographical clustering and installed capacity prediction. In the embodiments of the present invention, a dynamic photovoltaic module performance degradation model is constructed by combining environmental impact factor data and component performance degradation data. Through geographical clustering analysis combined with the photovoltaic module performance degradation model, the installation areas of photovoltaic modules are divided into several sub-regions with similar environmental characteristics, and a differential life prediction model is constructed for each sub-region, which can reflect the performance changes of components in different environments, avoiding the limitations of the fixed-year assumption. At the same time, a sub-region prediction installed capacity model is constructed by combining resource endowment and historical growth contribution weighting method, which not only considers historical data but also combines local resource endowment, improving the prediction accuracy and regional applicability, and can be used to predict the spatio-temporal distribution of crystalline silicon photovoltaic module waste nationwide.

[0042] It should be noted that for the foregoing embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps may be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0043] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the protection scope of the invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on these embodiments, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art can still, without conflict, make combinations, additions, deletions or other adjustments to the features in the embodiments of the present invention according to the situation without creative efforts, so as to obtain different technical solutions that are essentially not divorced from the concept of the present invention, and these technical solutions also belong to the scope of protection of the present invention.

Claims

1. A method for predicting photovoltaic module waste based on geographical clustering and installed capacity prediction, characterized in that, Including: Collecting environmental impact factor data and component performance degradation data within the collection area cycle time, preprocessing the environmental impact factor data and component performance degradation data, and constructing a photovoltaic component performance degradation model based on the preprocessed environmental impact factor data and component performance degradation data; Loading the constructed photovoltaic component performance degradation model, analyzing the photovoltaic component performance degradation model to obtain the action weights of environmental impact factors during the component aging process, and performing geographical clustering analysis on the photovoltaic component installation area based on the action weights of environmental impact factors and the clustering algorithm, and constructing a sub-region life prediction model according to the geographical clustering analysis results; Collecting the photovoltaic historical installed capacity data of each sub-region in the photovoltaic component installation area, constructing a sub-region predicted installed capacity model based on the resource endowment and historical growth contribution weighting method, and outputting the sub-region predicted installed capacity; Loading the sub-region life prediction model constructed according to the geographical clustering analysis results, and establishing a failure distribution function of the sub-region photovoltaic components based on the sub-region life prediction model; Based on the photovoltaic historical installed capacity data and sub-region predicted installed capacity of each sub-region, combining with the market supply A model to obtain the annual newly added photovoltaic installed capacity waste volume in the photovoltaic component installation area; Loading the annual newly added photovoltaic installed capacity waste volume in the photovoltaic component installation area, and predicting the cumulative waste volume of photovoltaic components based on the annual newly added photovoltaic installed capacity waste volume in the photovoltaic component installation area.

2. The photovoltaic module waste prediction method based on geographical clustering and installed capacity prediction according to claim 1, wherein: When constructing the photovoltaic component performance degradation model, quantifying the cumulative damage effects of temperature cycle, humidity, ultraviolet irradiation, and mechanical load environmental stresses on the component life, and the photovoltaic component performance degradation model is expressed as: Among them, represents the performance value of the photovoltaic module, is the performance loss value of the photovoltaic module, is the intensity factor of performance degradation, 、 and represent exponential constants, is the measure of the performance degradation of the photovoltaic module caused by non-primary influencing factors, is the activation energy, representing the comprehensive energy of the performance degradation reaction of the photovoltaic module, is the Boltzmann constant, is the absolute temperature, is the ambient temperature, is the relative humidity of the component service environment, and respectively represent the highest temperature and the lowest temperature in a temperature cycle. The highest temperature is the backplane temperature of the component at the highest ambient temperature, and the lowest temperature is taken as the lowest ambient temperature. is the incident solar irradiance, is the wind speed measured at a standard height of 10 m above the ground, 、 are empirical coefficients determined by the type of photovoltaic module, is the ultraviolet irradiance reaching the ground, and are the irradiances of long-wave ultraviolet rays and medium-wave ultraviolet rays respectively.

3. The photovoltaic module waste prediction method based on geographical clustering and installed capacity prediction according to claim 2, wherein: When performing geographical clustering analysis on the photovoltaic component installation area based on the action weights of environmental impact factors and the clustering algorithm, the clustering algorithm is the K-means or hierarchical clustering algorithm.

4. The photovoltaic module waste prediction method based on geographical clustering and installed capacity prediction according to claim 2, wherein: The method for constructing a sub-region predicted installed capacity model based on the resource endowment and historical growth contribution weighting method includes: Loading the photovoltaic historical installed capacity data of each sub-region in the photovoltaic component installation area, and normalizing the photovoltaic historical installed capacity data of each sub-region; Decomposing the total installed capacity of the target year in the photovoltaic component installation area into sub-region growth targets; Based on the resource endowment and historical growth contribution weighting method, combining the photovoltaic historical installed capacity data of each sub-region with the decomposed sub-region growth targets to construct a sub-region predicted installed capacity model, and outputting the sub-region predicted installed capacity.

5. The photovoltaic module waste prediction method based on geographical clustering and installed capacity prediction according to claim 4, wherein: The sub-region predicted installed capacity model is a dynamic K-value Logistic model, and the dynamic K-value Logistic model is expressed as: Among them, is the installed capacity allocation weight of each sub-region under the photovoltaic installation target in the photovoltaic module installation area, is an adjustable parameter used to balance the weight of the resource potential ratio and the historical growth contribution rate; is the contribution rate of each sub-region to the growth of the photovoltaic installation volume in the photovoltaic module installation area, is the weight of the proportion of solar energy resource utilization potential.

6. The photovoltaic module waste prediction method based on geographical clustering and installed capacity prediction according to claim 5, characterized in that: When establishing the failure distribution function of the sub-region photovoltaic components based on the sub-region life prediction model, the failure distribution function of the sub-region photovoltaic components is expressed as: Among them, and the probability density function and the cumulative distribution function; is the time independent variable, representing the time when the component has been put into use; is the shape parameter; is the scale parameter.

7. The photovoltaic module waste prediction method based on geographical clustering and installed capacity prediction according to claim 6, characterized in that: The calculation formula for obtaining the annual newly added photovoltaic installed capacity waste volume in the photovoltaic component installation area by combining with the market supply A model is as follows: Among them, is the photovoltaic waste generation amount in the nth year; is the photovoltaic installed capacity in the ith year before this; is the proportion of the installed capacity in the ith year that fails in the nth year, obtained from the Weibull distribution function.

8. The photovoltaic module waste prediction method based on geographical clustering and installed capacity prediction according to claim 7, wherein: When predicting the cumulative waste volume of photovoltaic components based on the annual newly added photovoltaic installed capacity waste volume in the photovoltaic component installation area, obtaining the evolution of the market share of each technology type component and the component component quality of each technology type component, and the cumulative waste volume is calculated by the following formula: Among them is the cumulative discarded mass of the m-th valuable component in the n-th year, is the market share of the j-th technology type in the i-th year, is the mass power conversion ratio of the j-th photovoltaic module, is the mass proportion of the m-th component in the j-th technology type.

Images