Short-term net load prediction method considering building photovoltaic heat transfer characteristics

By constructing a photovoltaic-roof integrated heat transfer model and using a bidirectional long and short-term memory network, combining architectural and meteorological characteristics, the problem of insufficient net load prediction accuracy in the existing technology is solved, and high-precision short-term net load prediction for building photovoltaic systems is achieved. .

CN118970888AActive Publication Date: 2024-11-15SHANGHAI UNIVERSITY OF ELECTRIC POWER
View PDF 7 Cites 0 Cited by

Patent Information

Application Number
CN202410969852.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-19
Publication Date
2024-11-15
Estimated Expiration
2044-07-19

AI Technical Summary

Technical Problem

The existing net load prediction method considers building photovoltaic systems with single input features and fails to make full use of meteorological and building characteristics, resulting in insufficient prediction accuracy, especially when photovoltaic components are installed on roofs of industrial buildings, the difficulty of load prediction increases.

Method used

By adopting a short-term net load prediction method that considers the heat transfer characteristics of building photovoltaics, the integrated heat transfer model of photovoltaics, including the energy equilibrium equation of photovoltaic modules, the energy equilibrium equation in the ventilation runner, the non-stable thermal conduction equation of the roof and the boundary Conditions, combining meteorological and building feature data, divide seasonal and daily load types, and use a two-way long and short-term memory network for prediction.

Benefits of technology

It significantly improves the accuracy of short-term net load prediction, fully considers the impact of photovoltaic modules on building load, and enhances seasonal prediction accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118970888A_ABST
    Figure CN118970888A_ABST
Patent Text Reader

Abstract

The invention provides a net load short-term prediction method considering photovoltaic heat transfer characteristics of a building roof, and the method comprises the steps: constructing a photovoltaic-roof integrated heat transfer model through a heat balance method for the building roof provided with a parallel overhead photovoltaic module, and calculating the annual heat transfer characteristics of the building photovoltaic roof, and daily load types of all seasons are classified based on the waveform characteristics of the heat transfer capacity. And finally, taking the heat transfer characteristics as input elements, and performing net load short-term prediction of each season. According to the net load short-term prediction method, the roof photovoltaic heat effect having great influence on load fluctuation is considered, the daily load type is subdivided according to the heat transfer characteristics, the relation between the roof photovoltaic and the building load is deeply excavated and subjected to representation application, and the prediction precision of the net load in each season is remarkably improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of building photovoltaic systems, and in particular relates to a short-term net load prediction method considering building photovoltaic heat transfer characteristics. Background Art

[0002] In 2023, the energy consumption of building operation accounts for about 30% of the total global energy use, of which the energy consumption of central air-conditioning systems accounts for about 50% of the total building energy consumption, and is on an upward trend. As the national energy structure transforms towards clean and low-carbon, in order to achieve the goal of "building carbon reduction", building photovoltaics are fully developed to compensate for the power load during peak hours of buildings.

[0003] Although the main demand of photovoltaic systems is power generation, the contribution of building photovoltaic shading and insulation to energy saving has gradually attracted attention. Wang Yiping et al.'s research shows that compared with traditional exposed roofs, BIPV (building integrated photovoltaic) with ventilation gaps can reduce 46% of daily heat gain in summer due to its shading effect and good ventilation performance; in winter, BIPV can reduce 5-20% of indoor heat load peak and heat loss. Ma Zongyao et al. focused on the shading effect of rooftop photovoltaics in summer: after installing photovoltaics on the roof, the roof heat flux decreased by 41.7%, the roof peak temperature decreased by 22.9℃, and the daily heat gain decreased by 74.84%. Bhuvad SS et al. found in the annual effect analysis of photovoltaic shading on heat load that although the photovoltaic shading effect will increase the heat load of buildings in winter in all climate zones, the overall heat load of buildings may be reduced by 22.76-74.07%. The above studies have found and demonstrated that the shading and insulation effect of building photovoltaics has a significant impact on building loads from different regions and seasons, but there is currently a lack of practical application of the energy-saving effect of building photovoltaics.

[0004] Dominguez A et al. pointed out that the shading and heat preservation effect of photovoltaic modules has an effect on improving building energy conservation and thermal comfort, which is particularly evident in low-rise industrial buildings with poor insulation and large surface area. According to statistics, air conditioning cooling and heating loads account for about 50% to 67% of the total electricity consumption of industrial buildings, and industrial buildings are favored by photovoltaic power generation developers because they can provide large roof areas. Espino-Reyes et al. recorded that industrial buildings with photovoltaic systems installed have significant energy-saving effects. For industrial buildings with a large proportion of cooling and heating loads, if photovoltaic modules are installed on their roofs, it will have a huge impact on the size and fluctuation of their loads. This will increase the difficulty of load forecasting for the grid dispatching side that needs to plan and control load demand.

[0005] As the penetration rate of photovoltaic power continues to increase, power companies will need to transition from predicting actual load to predicting net load infiltrated with photovoltaic output, and the difficulty of prediction is increasing. The existing net load prediction methods have many shortcomings, so they still cannot achieve ideal prediction accuracy in actual application scenarios. On the one hand, in the existing net load prediction methods, the consideration of input features is relatively simple. The input features usually only include weather elements and single information of photovoltaic power such as temperature, air pressure, and solar radiation, but do not include meteorological characteristics such as date, week type, and holidays. The impact of these non-meteorological characteristics on net load cannot be ignored; on the other hand, for photovoltaic power installed on the roof of industrial buildings, its installation will also have a greater impact on the load. In addition, it is also necessary to ensure the prediction accuracy in different seasons.

[0006] Therefore, in order to solve the above problems, a net load forecasting method that can further improve the forecasting accuracy is currently needed. Summary of the invention

[0007] The present invention is made to solve the above problems, and aims to provide a short-term net load forecasting method that can further improve the forecasting accuracy. The present invention adopts the following technical solutions:

[0008] The present invention provides a short-term net load prediction method taking into account the photovoltaic heat transfer characteristics of a building, which is used to predict the net load of a building. A photovoltaic module parallel to the roof and overhead is arranged on the roof of the building, and a ventilation channel is formed between the roof and the photovoltaic module. The method has the following technical characteristics, which includes the following steps: step S1, obtaining relevant data required for predicting the net load; step S2, calculating the total irradiance of the slope based on the relevant data; step S3, constructing a photovoltaic-roof integrated heat transfer model of the roof, wherein the model includes energy balance equations of components of the photovoltaic module, energy balance equations of air in the ventilation channel, The one-dimensional non-steady-state heat conduction equation of the roof, the boundary equation of the outer surface of the roof, and the boundary equation of the inner surface of the roof; step S4, dividing the daily load type of each season based on the photovoltaic-roof integrated heat transfer model and the relevant data, and calculating the actual load prediction results of each season in combination with the daily load types; step S5, calculating the photovoltaic output prediction results of each season based on the photovoltaic-roof integrated heat transfer model and the relevant data; step S6, obtaining the net load prediction results of each season based on the actual load prediction results of each season and the photovoltaic output prediction results of each season.

[0009] The short-term net load prediction method considering the photovoltaic heat transfer characteristics of the building provided by the present invention may also have such technical characteristics, wherein, in step S1, the relevant data include horizontal plane total solar irradiance data, direct irradiance data, scattered irradiance data, reflected irradiance data, air temperature data, ground temperature data, wind speed data, installation data of the photovoltaic module, characteristic data of the photovoltaic module, structural data of the building, thermal property data of the roof of the building, thermal property data of the photovoltaic module, and indoor air conditioning setting data of the building; in step S3, the photovoltaic module includes a glass cover plate, a photovoltaic cell and a photovoltaic back plate, and the energy balance equation of the components of the photovoltaic module includes the energy balance equation of the glass cover plate, the energy balance equation of the photovoltaic cell, and the energy balance equation of the photovoltaic back plate.

[0010] The short-term net load prediction method considering the photovoltaic heat transfer characteristics of the building provided by the present invention may also have such a technical feature, wherein in step S3, the energy balance equation of the glass cover plate is:

[0011]

[0012] Where M g is the mass of the glass cover, C g is the specific heat of the glass cover, G t is the total solar irradiance on the slope, α g is the absorption rate of solar radiation by the glass cover, A is the area of ​​the glass cover, ρ o,g is the reflectivity of the glass cover, h c,glass is the convection heat transfer coefficient between the glass cover and the surrounding environment, T a is the ambient temperature, h gc is the heat transfer coefficient between the glass cover and the photovoltaic cell, Φ r,g-s is the radiation heat exchange between the glass cover and the surrounding environment, and the energy balance equation of the photovoltaic cell is:

[0013]

[0014] In the formula, G pv is the irradiance incident on the photovoltaic cell, G pv =G t ·A(1-ρ o,g )(1-α g ), h gc is the heat transfer coefficient between the photovoltaic cell and the glass cover, h tc is the heat transfer coefficient between photovoltaic cells and photovoltaic backplane, P c is the output power of the photovoltaic cell, and the energy balance equation of the photovoltaic backplane is:

[0015]

[0016] In the formula, h tf is the convective heat transfer coefficient between the photovoltaic backsheet and the air in the ventilation channel; ε t , ε e are the emissivity of the photovoltaic back panel and the outer surface of the roof respectively; δ is the Stefan-Boltzmann constant, and the energy balance equation of the air in the ventilation duct is:

[0017]

[0018] In the formula, h fe is the convective heat transfer coefficient between the air in the ventilation duct and the outer surface of the roof, and h fe Convective heat transfer coefficient h between the photovoltaic back panel and the air in the ventilation channel tf Same; m fr , C f are the mass flow rate and specific heat capacity of the air in the flow channel respectively; T fo With T fi are the inlet and outlet temperatures of the flow channel respectively, and T fo =T a , and T f =(T fo +T fi ) / 2, the one-dimensional unsteady heat conduction equation of the roof is:

[0019]

[0020] Where a is the thermal diffusion coefficient of the roof material; the boundary equation of the outer surface of the roof is:

[0021]

[0022] The boundary equation of the inner surface of the roof is:

[0023]

[0024] In the formula, λ e , i are the thermal conductivity of the roof outer surface and the roof inner surface, T he Specify the temperature for indoor air conditioning equipment in the building, h ih is the heat transfer coefficient of the roof inner surface.

[0025] The short-term net load prediction method considering the photovoltaic heat transfer characteristics of the building provided by the present invention may also have such a technical feature, wherein step S4 includes the following sub-steps: step S4-1, calculating the hourly heat transfer of the roof based on the photovoltaic-roof integrated heat transfer model and the relevant data; step S4-2, constructing a feature matrix based on the daily maximum value of the hourly heat transfer, and dividing the daily load types of each season based on the feature matrix using the hierarchical clustering method to obtain the daily load type labels corresponding to each season; step S4-3, constructing the actual load input matrix of each season respectively with the load type labels corresponding to each season, the hourly heat transfer, and the actual load power time series; step S4-4, based on the actual load input matrix corresponding to the season to be predicted, predicting the actual load of each season through a bidirectional long short-term memory network algorithm, and obtaining the actual load prediction result of the season to be predicted.

[0026] The short-term net load prediction method considering the building photovoltaic heat transfer characteristics provided by the present invention may also have such a technical feature, wherein, in step S4-3, the input matrix X during the actual load prediction 1 for:

[0027]

[0028] Where h is the hourly heat transfer of the roof, l is the load type label, S load is the historical actual load power time series, i is the index value of the sample point, and n is the length of the historical actual load power time series.

[0029] The short-term net load prediction method considering the photovoltaic heat transfer characteristics of the building provided by the present invention may also have such a technical feature, wherein, in step S4-2, for spring, the number of clusters is five; for summer, autumn, and winter, the number of clusters is four, and a label is set for each cluster as the load type label.

[0030] The short-term net load prediction method considering the photovoltaic heat transfer characteristics of the building provided by the present invention may also have such technical characteristics, wherein step S5 includes the following sub-steps: step S5-1, calculating the photovoltaic cell temperature of the photovoltaic cell based on the photovoltaic-roof integrated heat transfer model and the relevant data; step S5-2, constructing the photovoltaic output input matrix of the predicted season based on the photovoltaic cell temperature, the total irradiance of the slope, relative humidity, and photovoltaic output time series; step S5-3, based on the photovoltaic output input matrix corresponding to each season, predicting the photovoltaic output of the season to be predicted through a bidirectional long short-term memory network algorithm, and obtaining the photovoltaic output prediction results of each season.

[0031] The short-term net load prediction method considering the building photovoltaic heat transfer characteristics provided by the present invention may also have such a technical feature, wherein in step S5-2, the photovoltaic output input matrix is:

[0032]

[0033] Where, T c is the photovoltaic cell temperature, G t is the total irradiance on the slope, H is the relative humidity; S pv is the historical photovoltaic output time series.

[0034] The short-term net load prediction method considering the photovoltaic heat transfer characteristics of the building provided by the present invention may also have such a technical feature, wherein, in step S2, the Liu&Jordan model is used to calculate the total irradiance of the slope:

[0035]

[0036] In the formula, G t is the total irradiance on the slope, G b is the horizontal direct irradiance, R b is the ratio of direct irradiance on the inclined plane to that on the horizontal plane, G d is the scattered irradiance on the horizontal surface, θ is the inclination angle of the photovoltaic module, G is the total irradiance on the horizontal surface, and ρ is the roof reflectivity.

[0037] Function and Effect of the Invention

[0038] According to the short-term net load prediction method considering the building photovoltaic heat transfer characteristics of the present invention, for the building roof equipped with parallel overhead photovoltaic modules, a photovoltaic-roof integrated heat transfer model is constructed by the heat balance method, including the energy balance equation of the components in the photovoltaic module, the energy balance equation of the air in the ventilation duct, the one-dimensional non-steady-state heat conduction equation of the roof, and the boundary equation of the inner and outer surfaces of the roof. The actual load and photovoltaic output are predicted based on the model to obtain the prediction result of the short-term net load. When predicting the actual load, the daily load types of each season are divided based on the model and related data. That is, the method of this embodiment fully considers the roof photovoltaic thermal effect that has a greater impact on load fluctuations, and subdivides the daily load types according to the season, so it can effectively improve the prediction accuracy of the short-term net load. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is a flow chart of a short-term net load prediction method considering building photovoltaic heat transfer characteristics in an embodiment of the present invention;

[0040] Figure 2 2 is a schematic cross-sectional view of two roof heat transfer models in an embodiment of the present invention;

[0041] Figure 3 is a graph of the internal and external surface temperatures of two roofs in each season in an embodiment of the present invention;

[0042] Figure 4 is a heat transfer comparison diagram of two roof heat transfer models in an embodiment of the present invention;

[0043] Figure 5 is a diagram showing the effect of daily load clustering in each season in an embodiment of the present invention;

[0044] Figure 6 is a graph of actual load forecasting results for each season in an embodiment of the present invention;

[0045] Figure 7 is a graph of the temperature of various components of a photovoltaic module in an embodiment of the present invention;

[0046] Figure 8 is a photovoltaic output prediction result diagram for each season in an embodiment of the present invention;

[0047] Fig. 9 It is a net load prediction result diagram of each season in the embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the short-term net load prediction method considering the photovoltaic heat transfer characteristics of the building of the present invention is specifically described below in combination with the embodiments and drawings.

[0049] <Example>

[0050] Figure 1 It is a flow chart of the short-term net load prediction method considering the photovoltaic heat transfer characteristics of the building in this embodiment.

[0051] like Figure 1 As shown, the method comprises the following steps:

[0052] Step S1, obtaining relevant data required for predicting short-term net load.

[0053] Step S2, calculating the total irradiance of the slope based on the relevant data.

[0054] Step S3, constructing a photovoltaic-roof integrated heat transfer model by heat balance method.

[0055] Step S4, based on the photovoltaic-roof integrated heat transfer model and the above-mentioned related data, the load types of each season are divided, and the actual load prediction result of the season to be predicted is calculated in combination with the corresponding load types.

[0056] Step S5, calculating the photovoltaic output prediction result of the season to be predicted based on the photovoltaic-roof integrated heat transfer model and the above-mentioned related data.

[0057] Step S6, obtaining a net load forecast result for the season to be forecasted based on the actual load forecast result for the season to be forecasted and the photovoltaic output forecast result for the season to be forecasted.

[0058] The above steps will be described in detail below.

[0059] Step S1, obtaining relevant data required for predicting short-term net load.

[0060] In this embodiment, the relevant data include horizontal plane solar irradiance data, air temperature data, ground temperature data, wind speed data, photovoltaic module installation data, photovoltaic module characteristic data, building structure data, building roof thermal property data, photovoltaic module thermal property data, and building indoor air conditioning setting data.

[0061] Among them, the horizontal solar irradiance data includes the total horizontal solar irradiance, direct irradiance, scattered irradiance, and reflected irradiance. The photovoltaic module installation data and photovoltaic module characteristic data include the installation method of the photovoltaic module, the ventilation channel gap, the laying area, the laying length, the materials, thickness, and photoelectric conversion efficiency of each part of the photovoltaic module. The building structure data includes the materials and thickness of each part of the roof. The building roof thermal property data and photovoltaic module thermal property data include the thermal conductivity, density, and specific heat capacity of the materials of each part of the roof and photovoltaic module, the absorptivity and reflectivity of the photovoltaic glass cover of the photovoltaic module, the absorptivity of the photovoltaic cell and the photovoltaic backplane, the absorptivity and reflectivity of the roof, and the emissivity of the photovoltaic glass cover and the roof.

[0062] Step S2, calculating the total irradiance of the inclined surface based on the above related data.

[0063] In this embodiment, the Liu & Jordan model is used to calculate the total irradiance of the inclined surface, and the calculation is specifically based on the following formula:

[0064]

[0065] In the formula, G t is the total irradiance on the slope; G b is the direct irradiance on the horizontal surface; R b is the ratio of direct irradiance on the inclined plane to that on the horizontal plane; G d is the scattered irradiance on the horizontal surface; θ is the inclination angle of the photovoltaic module; G is the total irradiance on the horizontal surface; ρ is the reflectivity of the roof.

[0066] Step S3, constructing a photovoltaic-roof integrated heat transfer model by heat balance method.

[0067] In this embodiment, a rooftop with parallel overhead photovoltaic modules installed is taken as an example for specific description.

[0068] Figure 2 is a schematic cross-sectional view of two types of roofs in this embodiment, Figure 2 (a) shows a bare roof, which corresponds to the roof heat transfer model without photovoltaic modules installed. Figure 2 (b) in the figure shows a roof with parallel overhead photovoltaic modules laid on the roof, corresponding to the parallel overhead photovoltaic-roof integrated heat transfer model.

[0069] like Figure 2 As shown, the roof with parallel overhead photovoltaic modules can be divided into three parts from top to bottom: photovoltaic modules, ventilation ducts (or ventilation gaps), and roofs. There is flowing air in the ventilation ducts. Among them, the photovoltaic modules include a glass cover plate, a photovoltaic cell, and a photovoltaic back plate stacked in sequence. In the formula below, their related parameters are marked with subscripts g, c, and t, respectively. The related parameters of the ventilation duct, the outer surface of the roof, and the inner surface of the roof are marked with subscripts f, e, and i, respectively. The photovoltaic-roof integrated heat transfer model is obtained by modeling the roof structure. The temperature T of each part of the model is calculated by constructing the corresponding heat balance (energy balance) equation group.

[0070] The energy balance equation for the glass cover is:

[0071]

[0072] Where M g is the mass of the glass cover; C g is the specific heat of the glass cover; G t is the total solar irradiance on the slope; α g is the absorption rate of solar radiation by the glass cover; A is the area of ​​the glass cover; ρ o,g is the reflectivity of the glass cover; h c,glass is the convection heat transfer coefficient between the glass cover and the surrounding environment; T a is the ambient temperature; h gc is the heat transfer coefficient between the glass cover and the photovoltaic cell; Φ r,g-s It is the radiation heat exchange between the glass cover and the surrounding environment (including the surrounding air, sky, and ground).

[0073] The energy balance equation for photovoltaic cells is:

[0074]

[0075] In the formula, G pv is the irradiance incident on the photovoltaic cell, G pv =G t ·A(1-ρ o,g)(1-α g );h gc With h tc are the heat transfer coefficients between photovoltaic cells and glass cover plate and photovoltaic back plate respectively; P c is the output power of the photovoltaic cell, let P c =G pv ·η, η is the photoelectric conversion efficiency of the photovoltaic module.

[0076] The energy balance equation of the photovoltaic backplane is:

[0077]

[0078] In the formula, h tf is the convective heat transfer coefficient between the photovoltaic backsheet and the air in the ventilation channel; ε t , ε e are the emissivity of the back panel and the outer surface of the roof, respectively; δ is the Stefan-Boltzmann constant, which is 5.669×10 -8 W / (m 2 ·K 4 ).

[0079] The energy balance equation of the air in the ventilation duct is:

[0080]

[0081] In the formula, h fe is the convective heat transfer coefficient between the air in the flow channel and the outer surface of the roof, and h fe The convection heat transfer coefficient h between the photovoltaic back panel and the air in the ventilation channel tf Same; m fr , C f are the mass flow rate and specific heat capacity of the air in the flow channel respectively; T fo With T fi are the inlet and outlet temperatures of the flow channel respectively, and T fo =T a , and T f =(T fo +T fi ) / 2.

[0082] In the parallel overhead photovoltaic-roof integrated heat transfer model, the boundary equations of the roof outer surface and the roof inner surface are:

[0083]

[0084] In the formula, λ e , i are the thermal conductivity of the roof’s outer surface and inner surface respectively; T he The temperature of the indoor air conditioning equipment in the building is set to 24℃ throughout the year in this embodiment; ihis the heat transfer coefficient of the roof inner surface.

[0085] In this embodiment, Figure 2 The exposed roof shown in (a) and Figure 2 The cross-sectional diagram of the parallel overhead photovoltaic-roof shown in (b) is used for comparative analysis. Therefore, the heat transfer equation of the bare roof heat transfer model is also given below.

[0086] The roof heat transfer model is a one-dimensional unsteady-state heat conduction along the thickness direction of the roof, and its heat transfer equation is:

[0087]

[0088] Where a is the thermal diffusion coefficient of the roof material. For roofs made of composite materials, the thermal diffusion coefficients of the corresponding materials are used at different thicknesses.

[0089] The boundary equations of the outer surface of the exposed roof and the inner surface of the exposed roof are:

[0090]

[0091] Step S4, based on the photovoltaic-roof integrated heat transfer model and the above-mentioned related data, the load types of each season are divided, and the actual load prediction result of the season to be predicted is calculated in combination with the corresponding load types.

[0092] like Figure 1 As shown, step S4 specifically includes the following sub-steps:

[0093] Step S4-1, calculating the hourly heat transfer of the roof based on the photovoltaic-roof integrated heat transfer model and the above-mentioned related data.

[0094] Among them, based on the above model and related data, the temperature of the outer surface of the roof and the temperature of the inner surface of the roof at each time point are calculated respectively, and then the hourly heat transfer of the roof can be calculated. The specific calculation can be based on the following formula: h ih (T i -T he ).

[0095] In this embodiment, the hourly heat transfer of the roof throughout the year is calculated based on relevant data for one year.

[0096] Figure 3 is a graph of the internal and external surface temperatures of the two roofs in each season in this embodiment, Figure 4 3 is a heat transfer comparison diagram of two roof heat transfer models in an embodiment of the present invention.

[0097] like Figure 3 and Figure 4As shown, based on the above model and related data, the temperature of the inner and outer surfaces of the roof at each time point in each day of each season can be calculated, thereby obtaining a curve chart of the inner and outer surface temperatures of the roof. It can be seen that during the period of sunshine, the temperature of the roof with photovoltaic modules installed is significantly lower than the temperature of the exposed roof. Based on the curve chart and parameters such as the heat transfer coefficient of the inner surface of the roof, the hourly heat transfer of the roof can be further calculated. It can be seen that the roof with photovoltaic modules installed is significantly different from the exposed roof in terms of temperature and heat transfer.

[0098] Step S4-2, constructing a characteristic matrix based on the daily maximum value of the hourly heat transfer of the roof, and dividing the daily load types of each season using a hierarchical clustering method based on the characteristic matrix to obtain the daily load type labels corresponding to each season.

[0099] Among them, in the process of constructing the characteristic matrix of load clustering, using the daily maximum value of hourly heat transfer (i.e. the maximum and minimum values ​​in a day) to construct the characteristic matrix can achieve the best classification effect. Load classification is achieved using the hierarchical clustering method, and the classification effect is verified by the daily load mean. Since there are differences in heat transfer characteristics and load requirements in each season, the number of load clusters is also different.

[0100] Figure 5 It is the effect diagram of daily load clustering in each season in this embodiment.

[0101] like Figure 5 As shown in the figure, according to the statistical analysis of the above related data, in spring, the frequency of daytime temperature fluctuation is relatively large. Therefore, the number of load clusters in spring is the largest, reaching five; while the number of clusters in the other three seasons is four each. A label is set for each cluster as a load type label.

[0102] Step S4-3, constructing the actual load input matrix of the season to be predicted with the load type label corresponding to the season to be predicted, the hourly heat transfer of the roof, and the actual load power time series.

[0103] The actual load power time series can be obtained based on the above related data. The input matrix X for actual load prediction 1 as follows:

[0104]

[0105] Where h is the hourly heat transfer of the roof with photovoltaic modules installed; l is the load type label; S load is the historical actual load power time series; i is the index value of the sample point; n is the length of the historical actual load power time series.

[0106] Step S4-4, based on the actual load input matrix corresponding to the season to be predicted, the actual load of the season to be predicted is predicted by the Bi-LSTM (bidirectional long short-term memory network) algorithm to obtain the actual load prediction result of the season to be predicted.

[0107] That is, the above actual load input matrix is ​​input into the trained Bi-LSTM model to obtain the corresponding output.

[0108] Figure 6 is the actual load forecast result diagram for each season in this embodiment, Figure 6 In the figure, the broken line corresponding to True represents the actual load obtained according to the relevant data, Pre1 represents the actual load prediction result with hourly heat transfer and load type label, Pre2 represents the actual load prediction result with hourly heat transfer but no load type label, and Pre3 represents the actual load prediction result without hourly heat transfer and load type label.

[0109] The following Table 1 gives the evaluation indicators of the actual load forecasting results.

[0110] Table 1 Evaluation index table of actual load forecast results

[0111]

[0112] In the table, RMSE is the root mean square error and MAPE is the mean absolute percentage error.

[0113] from Figure 6 From Table 1, we can see that the prediction accuracy of Pre1 is significantly higher than that of Pre2 and Pre3, which means that considering the hourly heat transfer and setting the above load type labels are beneficial to improving the prediction accuracy.

[0114] Step S5, calculating the output prediction results of the photovoltaic components in each season based on the photovoltaic-roof integrated heat transfer model and the above-mentioned related data.

[0115] like Figure 1 As shown, step S5 specifically includes the following sub-steps:

[0116] Step S5-1, calculating the temperature of the photovoltaic cells in the photovoltaic assembly based on the photovoltaic-roof integrated heat transfer model and the above-mentioned related data.

[0117] Figure 7 is a curve diagram of the temperature of each component of the photovoltaic module in this embodiment.

[0118] like Figure 7As shown in the figure, during the sunshine period, the temperature of the photovoltaic cell is very close to the temperature of the glass cover and the temperature of the photovoltaic backplane, while during the night period without sunshine, the temperature of the photovoltaic cell is slightly higher than the temperature of the glass cover and the temperature of the photovoltaic backplane. In addition, the temperature of the photovoltaic cell is quite different in summer and winter.

[0119] Step S5-2: construct the photovoltaic output input matrix X of the season to be predicted based on the photovoltaic cell temperature, the total irradiance of the slope, the relative humidity, and the photovoltaic output time series 2 as follows:

[0120]

[0121] Where, T c is the photovoltaic cell temperature; G t is the total irradiance on the slope; H is the relative humidity; S pv is the historical photovoltaic output time series.

[0122] Step S5-3, based on the photovoltaic output input matrix corresponding to the season to be predicted, the photovoltaic output of the season to be predicted is predicted by using the Bi-LSTM algorithm to obtain the photovoltaic output prediction result of the season to be predicted.

[0123] Figure 8 is a graph showing the photovoltaic output prediction results for each season in this embodiment. Figure 8 In the figure, the line corresponding to True represents the actual PV output obtained based on the relevant data, Pre1 represents the PV output prediction result with the PV cell temperature, and Pre2 represents the PV output prediction result without the PV cell temperature.

[0124] The following Table 2 gives the evaluation indicators of the photovoltaic output prediction results.

[0125] Table 2 Evaluation index table of photovoltaic output prediction results

[0126]

[0127] from Figure 8 It can be seen from Table 2 that the prediction accuracy of Pre1 is significantly higher than that of Pre2, which means that considering the temperature of photovoltaic cells is beneficial to improving the prediction accuracy of photovoltaic output.

[0128] Step S6, obtaining the net load forecast result of each season based on the actual load forecast result of each season and the output forecast result of each season.

[0129] The calculation formula of net load forecast result is as follows:

[0130]

[0131] In the formula, is the actual load forecast result obtained in step S4, is the photovoltaic output prediction result obtained through step S5.

[0132] Fig. 9 is the net load forecast result diagram for each season in this embodiment, Fig. 9 , the broken line corresponding to True represents the actual value of the net load obtained according to the above related data, Pre1 represents the net load prediction result with hourly heat transfer, load type label and photovoltaic cell temperature, Pre2 represents the net load prediction result with hourly heat transfer, no load type label and photovoltaic cell temperature, and Pre3 represents the net load prediction result without hourly heat transfer, load type label and photovoltaic cell temperature.

[0133] The following Table 3 gives the evaluation indicators of the net load prediction results.

[0134] Table 3 Net load forecast result evaluation index table

[0135]

[0136] from Fig. 9 As can be seen from Table 3, the net load prediction result of Pre1 is very close to the actual value, and its prediction accuracy is significantly higher than that of Pre2 and Pre3, that is, the method of this embodiment using hourly heat transfer, load classification labels and photovoltaic cell temperature can effectively improve the prediction accuracy of the net load.

[0137] Functions and Effects of the Embodiments

[0138] According to the short-term net load prediction method considering the building photovoltaic heat transfer characteristics provided by the present embodiment, for the building roof equipped with parallel overhead photovoltaic modules, a photovoltaic-roof integrated heat transfer model is constructed by the heat balance method, including the energy balance equation of the components in the photovoltaic modules, the energy balance equation of the air in the ventilation duct, and the boundary equation of the inner and outer surfaces of the roof. The actual load and photovoltaic output are predicted based on the model to obtain the prediction result of the short-term net load. When predicting the actual load, the daily load types of each season are divided based on the model and related data. That is, the method of the present embodiment fully considers the roof photovoltaic thermal effect that has a great influence on the load fluctuation, and subdivides the daily load types according to the season, so that the prediction accuracy of the short-term net load can be effectively improved.

[0139] In the embodiment, the photovoltaic-roof integrated heat transfer model fully considers the heat transfer conditions of the building roof equipped with parallel overhead photovoltaic modules, including the absorption and reflection of solar radiation by the glass cover, the convection and radiation heat transfer between the glass cover and the surrounding environment, the heat transfer between the glass cover and the photovoltaic cells, the heat transfer between the photovoltaic cells and the photovoltaic back panel, the heat transfer between the photovoltaic back panel and the air in the ventilation duct, etc. The modeling is detailed and complete, so it can improve the accuracy of short-term net load prediction.

[0140] Furthermore, in the method of the embodiment, the heat transfer characteristics of the model throughout the year are calculated based on meteorological data and non-meteorological data, and the daily load types in each season are further classified based on the waveform characteristics of the heat transfer amount. Finally, the heat transfer characteristics are used as input factors to perform short-term prediction of the net load in each season, which comprehensively considers multiple factors and thus significantly improves the prediction accuracy.

[0141] In summary, in the method of the embodiment, the relationship between rooftop photovoltaics and building loads is deeply explored and characterized, which significantly improves the prediction accuracy of short-term net loads in each season.

[0142] The above embodiments are only used to illustrate the specific implementation of the present invention, and the present invention is not limited to the description scope of the above embodiments. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only to illustrate the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A short-term net load prediction method considering the photovoltaic heat transfer characteristics of a building, for predicting the net load of a building, wherein a photovoltaic assembly parallel to the roof and overhead is arranged on the roof of the building, and a ventilation channel is formed between the roof and the photovoltaic assembly, characterized in that: The following steps are involved: Step S1, obtaining relevant data required for predicting the net load; Step S2, calculating the total irradiance of the slope based on the relevant data; Step S3, constructing a photovoltaic-roof integrated heat transfer model of the roof by a heat balance method, wherein the model includes an energy balance equation of components of the photovoltaic module, an energy balance equation of air in the ventilation duct, a one-dimensional unsteady-state heat conduction equation of the roof, a boundary equation of the outer surface of the roof, and a boundary equation of the inner surface of the roof; Step S4, dividing the daily load types of each season based on the photovoltaic-roof integrated heat transfer model and the relevant data, and calculating the actual load prediction result of the season to be predicted in combination with the daily load type; Step S5, calculating and obtaining a photovoltaic output prediction result for the season to be predicted based on the photovoltaic-roof integrated heat transfer model and the relevant data; Step S6, obtaining the net load forecast result of the season to be predicted based on the actual load forecast result of the season to be predicted and the photovoltaic output forecast result of the season to be predicted.

2. The short-term net load forecasting method considering building photovoltaic heat transfer characteristics according to claim 1 is characterized by: in, In step S1, the relevant data includes horizontal surface total solar irradiance data, direct irradiance data, scattered irradiance data, reflected irradiance data, air temperature data, ground temperature data, wind speed data, installation data of the photovoltaic module, characteristic data of the photovoltaic module, structural data of the building, thermal property data of the roof of the building, thermal property data of the photovoltaic module, and indoor air conditioning setting data of the building. In step S3, the photovoltaic module includes a glass cover plate, a photovoltaic cell and a photovoltaic back plate. The energy balance equations of the components of the photovoltaic assembly include the energy balance equation of the glass cover plate, the energy balance equation of the photovoltaic cell, and the energy balance equation of the photovoltaic back plate.

3. The short-term net load forecasting method considering building photovoltaic heat transfer characteristics according to claim 2 is characterized by: in, In step S3, the energy balance equation of the glass cover is: Where M g is the mass of the glass cover, C g is the specific heat of the glass cover, G t is the total solar irradiance on the slope, α g is the absorption rate of solar radiation by the glass cover, A is the area of ​​the glass cover, ρ o,g is the reflectivity of the glass cover, h c,glass is the convection heat transfer coefficient between the glass cover and the surrounding environment, T a is the ambient temperature, h gc is the heat transfer coefficient between the glass cover and the photovoltaic cell, Φ r,g-s The radiation heat exchange between the glass cover and the surrounding environment, The energy balance equation of the photovoltaic cell is: In the formula, G pv is the irradiance incident on the photovoltaic cell, G pv =G t ·A(1-ρ o,g )(1-α g ), h gc is the heat transfer coefficient between the photovoltaic cell and the glass cover, h tc is the heat transfer coefficient between photovoltaic cells and photovoltaic backplane, P c is the output power of the photovoltaic cell, The energy balance equation of the photovoltaic backplane is: In the formula, h tf is the convective heat transfer coefficient between the photovoltaic backsheet and the air in the ventilation channel; ε t , ε e are the emissivity of the photovoltaic back panel and the roof outer surface respectively; δ is the Stefan-Boltzmann constant, The energy balance equation of the air in the ventilation channel is: In the formula, h fe is the convective heat transfer coefficient between the air in the ventilation duct and the outer surface of the roof, and h fe Convective heat transfer coefficient h between the photovoltaic back panel and the air in the ventilation channel tf Same; m fr , C f are the mass flow rate and specific heat capacity of the air in the flow channel respectively; T fo With T fi are the inlet and outlet temperatures of the flow channel respectively, and T fo =T a , and T f =(T fo +T fi ) / 2, The one-dimensional unsteady heat conduction equation of the roof is: Where a is the thermal diffusion coefficient of the roof material, The boundary equation of the roof outer surface is: The boundary equation of the inner surface of the roof is: In the formula, λ e , i are the thermal conductivity of the roof outer surface and the roof inner surface, T he Specify the temperature for indoor air conditioning equipment in the building, h ih is the heat transfer coefficient of the roof inner surface.

4. The short-term net load prediction method considering building photovoltaic heat transfer characteristics according to claim 3, Features: Wherein, step S4 includes the following sub-steps: Step S4-1, calculating the hourly heat transfer of the roof based on the photovoltaic-roof integrated heat transfer model and the relevant data; Step S4-2, constructing a characteristic matrix based on the daily maximum value of the hourly heat transfer, and dividing the daily load types of each season using a hierarchical clustering method based on the characteristic matrix to obtain a daily load type label corresponding to each season; Step S4-3, constructing the actual load input matrix of each season respectively with the load type label, the hourly heat transfer and the actual load power time series corresponding to each season; Step S4-4, based on the actual load input matrix corresponding to the season to be predicted, the actual load of each season is predicted by a bidirectional long short-term memory network algorithm to obtain the actual load prediction result of the season to be predicted.

5. The short-term net load forecasting method considering building photovoltaic heat transfer characteristics according to claim 4 is characterized by: in, In step S4-3, the input matrix X1 for actual load forecasting is: Where h is the hourly heat transfer of the roof, l is the load type label, S load is the historical actual load power time series, i is the index value of the sample point, and n is the length of the historical actual load power time series.

6. The short-term net load forecasting method considering building photovoltaic heat transfer characteristics according to claim 4 is characterized by: in, In step S4-2, for spring, the number of clusters is five. For summer, autumn, and winter, the number of clusters is four. A label is set for each cluster as the load type label.

7. The short-term net load forecasting method considering building photovoltaic heat transfer characteristics according to claim 3 is characterized by: in, Step S5 includes the following sub-steps: Step S5-1, calculating the photovoltaic cell temperature of the photovoltaic cell based on the photovoltaic-roof integrated heat transfer model and the relevant data; Step S5-2, constructing a photovoltaic output input matrix for the season to be predicted based on the photovoltaic cell temperature, the total irradiance of the slope, the relative humidity, and the photovoltaic output time series; Step S5-3, based on the photovoltaic output input matrix corresponding to the season to be predicted, the photovoltaic output of the season to be predicted is predicted by a bidirectional long short-term memory network algorithm to obtain the photovoltaic output prediction results of each season.

8. The short-term net load forecasting method considering building photovoltaic heat transfer characteristics according to claim 7 is characterized by: in, In step S5-2, the photovoltaic output input matrix is: Where, T c is the photovoltaic cell temperature, G t is the total irradiance on the slope, H is the relative humidity; S pv is the historical photovoltaic output time series.

9. The short-term net load forecasting method considering building photovoltaic heat transfer characteristics according to claim 1 is characterized by: in, In step S2, the Liu&Jordan model is used to calculate the total irradiance of the slope: In the formula, G t is the total irradiance on the slope, G b is the horizontal direct irradiance, R b is the ratio of direct irradiance on the inclined plane to that on the horizontal plane, G d is the scattered irradiance on the horizontal surface, θ is the inclination angle of the photovoltaic module, G is the total irradiance on the horizontal surface, and ρ is the roof reflectivity.

Citation Information

Patent Citations

  • Building energy management system optimization control method and device

    CN106527135A

  • Cooperative control system and method of air-conditioning controllable load and photovoltaic energy storage system

    CN110474370A

  • Net zero energy consumption building energy system planning method considering long-term and short-term uncertainty

    CN114818078A

  • Multi-source complementary scheduling method considering incorporation of agricultural photovoltaic heating into central heating pipe network

    CN117553337A

  • Farm power distribution network photovoltaic output short-term prediction method considering load coupling influence

    CN118100152A