Generated power prediction method and system of photovoltaic power station

By calculating the comprehensive efficiency correction coefficient and building a machine learning model, combining the various influencing factors and environmental conditions of the photovoltaic power station, the refined prediction of the power generation power of the photovoltaic power station is achieved, and the problem of low prediction accuracy in the existing technology is solved.

CN119070273AActive Publication Date: 2024-12-03ECONOMIC & TECH RES INST OF HUBEI ELECTRIC POWER COMPANY SGCC +3

Patent Information

Application Number
CN202410990764.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-23
Publication Date
2024-12-03
Estimated Expiration
2044-07-23

AI Technical Summary

Technical Problem

The power generation power prediction methods of existing photovoltaic power stations fail to fully consider the influence of various factors, resulting in low accuracy of prediction results and large cumulative errors.

Method used

The calculation method of the comprehensive efficiency correction coefficient is adopted, and the cloud motion model and short-term photovoltaic power generation prediction model are constructed in combination with machine learning methods. By comprehensively considering the installation of photovoltaic power stations, line loss and other factors, combined with cloud motion and environmental climatic conditions, the photovoltaic power generation is achieved. Refined prediction of power.

Benefits of technology

The accuracy and reliability of the power prediction of photovoltaic power station power generation is improved, making the prediction results more consistent with the actual situation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119070273A_ABST
    Figure CN119070273A_ABST
Patent Text Reader

Abstract

The invention discloses a photovoltaic power station generation power prediction method and system, and the method comprises the steps: firstly calculating the comprehensive efficiency correction coefficient of a photovoltaic power station, including the type of a photovoltaic module, the inclination angle and azimuth angle of a photovoltaic array, the efficiency of an inverter, line loss, other factors, and the like, and then constructing a cloud layer motion model based on the environment and climate to evaluate the ground irradiation intensity, the method comprises the following steps: establishing a comprehensive efficiency correction coefficient, predicting ultra-short-term power, constructing a short-term photovoltaic power generation power prediction model based on historical power generation power data and weather, predicting short-term power, and finally predicting photovoltaic power generation power in a future period based on the comprehensive efficiency correction coefficient, the ultra-short-term photovoltaic power generation power and the short-term period photovoltaic power generation power. According to the method, the influence of various factors such as installation and loss of the photovoltaic power station on the generated power is comprehensively considered, climate conditions such as cloud layer movement and environment are combined, comprehensive historical data analysis is carried out, refined prediction of the generated power of the photovoltaic power station is achieved, and the timeliness and accuracy of prediction are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a power prediction method, belonging to the field of photovoltaic technology, and in particular to a method and system for predicting the power generation of a photovoltaic power station. Background Art

[0002] With the gradual reduction of oil production and the increasingly serious ecological and environmental problems, people have begun to seek sustainable and clean energy solutions, and solar photovoltaic power generation, as a clean and renewable energy form, has received widespread attention; photovoltaic power stations are affected by a variety of factors during the power generation process, and these factors work together to affect the power generation, causing it to fluctuate; specifically, photovoltaic modules are affected by the intensity of solar radiation and meteorological conditions, and are highly random, volatile and intermittent. Their output voltage and current are directly affected by the intensity of solar radiation and temperature changes, showing dynamic response characteristics, and the photoelectric conversion efficiency of photovoltaic modules, the deviation between the nominal power and the actual power of the modules, the light incidence rate, and the initial light-induced degradation effect of the modules. Factors will affect the power generation efficiency; in addition, due to the limitations of the manufacturing and installation process, even the same model of photovoltaic modules may have manufacturing errors and it is difficult to ensure that they are at the best angle during installation, which further increases the instability of the power generation. At the same time, the static and dynamic tracking losses within the photovoltaic power generation system, the impedance loss of cable power transmission, etc. will comprehensively affect the photovoltaic power generation efficiency; however, in the prior art, the prediction of efficiency is usually obtained by actually testing the conversion efficiency between the power generation and light irradiation of the photovoltaic power station, and the influence of other factors is not fully considered. It has certain limitations, and this method is prone to error accumulation, resulting in a large deviation between the future prediction results and the actual results, and low accuracy. Therefore, there is an urgent need for a means to solve the above-mentioned defects in the prior art. Summary of the invention

[0003] The purpose of the present invention is to overcome the above-mentioned defects and problems existing in the prior art and to provide a method and system for predicting the power generation of a photovoltaic power station with high prediction accuracy.

[0004] To achieve the above objectives, the technical solution of the present invention is: a method for predicting the power generation of a photovoltaic power station, comprising:

[0005] S1. Calculate the comprehensive efficiency correction coefficient of the photovoltaic power station; the calculation formula of the comprehensive efficiency correction coefficient is as follows:

[0006] K=K1×K2×K3×K4×...×Kn;

[0007] Among them: k is the comprehensive efficiency correction coefficient of the photovoltaic power station, K1 is the photovoltaic module type correction coefficient, K2 is the inclination and azimuth correction coefficient of the photovoltaic array, K3 is the inverter efficiency correction coefficient, K4 is the line loss correction coefficient, and Kn is the correction coefficient of other influencing factors;

[0008] S2. Based on the environmental and climatic factors of the photovoltaic power station area, a cloud motion model is constructed using machine learning methods. Then, the cloud motion model is used to simulate cloud motion to evaluate the ground radiation intensity and predict the ultra-short-term photovoltaic power generation power.

[0009] S3. Based on the historical power generation data and weather information of the photovoltaic power station, a short-term photovoltaic power generation prediction model is constructed using a machine learning method. Then, the short-term photovoltaic power generation prediction model is used to predict the short-term period photovoltaic power generation.

[0010] S4. Based on the comprehensive efficiency correction coefficient of the photovoltaic power station, the ultra-short-term photovoltaic power generation power, and the short-term period photovoltaic power generation power, the photovoltaic power generation power of the photovoltaic power station in a certain future period is predicted; the expression for predicting the photovoltaic power generation power in a certain future period is as follows:

[0011] P_cycle1=K×(alpha×P_ultrashort+(1-alpha)×P_short);

[0012] Among them: P_cyclel is the photovoltaic power generation power in a certain cycle in the future, K is the comprehensive efficiency correction coefficient of the photovoltaic power station, P_ultra_short is the ultra-short-term photovoltaic power generation power, P_short is the short-term cycle photovoltaic power generation power, and alpha is the weight coefficient.

[0013] The method for confirming the correction factor of the photovoltaic module type is as follows:

[0014] S11. Determine the material type of the photovoltaic module and collect relevant technical parameters of the photovoltaic module; the relevant technical parameters include conversion efficiency, maximum output power, open circuit voltage, and short circuit current;

[0015] S12. Based on the relevant technical parameters of the photovoltaic module, confirm the actual conversion efficiency of the photovoltaic module; the expression of the actual conversion efficiency is as follows:

[0016] η_actual=(Pmax / (Voc*Isc*FF))*100%;

[0017] Where: η_actual is the actual conversion efficiency of the photovoltaic module, Pmax is the maximum output power, Voc is the open circuit voltage, Isc is the short circuit current, and FF is the fill factor;

[0018] S13. Confirm the rated conversion efficiency of the photovoltaic module under standard test conditions; the expression of the rated conversion efficiency is as follows:

[0019] η_rated=(Pmax / (B*A)))*100%;

[0020] Where: η_rated is the rated conversion efficiency of the photovoltaic module, B is the incident light power, and A is the area of ​​the photovoltaic module;

[0021] S14. Based on the actual conversion efficiency and rated conversion efficiency of the photovoltaic module, determine the photovoltaic module type correction factor; the expression of the photovoltaic module type correction factor is as follows:

[0022] K1=(η_actual / η_rated)×(1-D);

[0023] Where: D is the lifetime decay rate of the photovoltaic module.

[0024] The method for confirming the inclination angle and azimuth angle correction coefficients of the photovoltaic array is as follows:

[0025] S15. Confirm the optimal inclination angle and azimuth angle of the photovoltaic array installation position, as well as the inclination angle and azimuth angle during actual installation;

[0026] S16. Taking the power generation of the photovoltaic array at the optimal inclination angle and azimuth angle as the reference condition, and calculating the deviation between the power generation of the photovoltaic array at the inclination angle and azimuth angle during actual installation and the reference condition; the expression of the correction coefficient is as follows:

[0027] K2≈angle optimal / angleA ctual ;

[0028] Where: angle optimal is the power generation under the best inclination and azimuth, angle Actual It is the power generation under the actual installation inclination and azimuth.

[0029] The method for confirming the inverter efficiency correction factor is as follows:

[0030] S17, confirming the ideal efficiency data of the inverter, and selecting the inverter efficiency value closest to the actual situation according to the actual operating conditions of the photovoltaic module to obtain the actual efficiency data;

[0031] S18. Calculate the inverter efficiency correction factor based on the ideal efficiency data and the actual efficiency data; the expression of the inverter efficiency correction factor is as follows:

[0032] K3=K actual / K ideal ;

[0033] Where: K actual is the actual efficiency data, K ideal This is ideal efficiency data.

[0034] The method for confirming the line loss correction factor is as follows:

[0035] S19. Determine the total resistance of the cable based on the resistivity, length and cross-sectional area of ​​the cable in the photovoltaic module; the expression of the total resistance is as follows:

[0036]

[0037] Where: R is the total resistance, ρ is the resistivity of the cable, L is the length, and A is the cross-sectional area;

[0038] S191. Based on the total resistance of the cable and the current passing through it, confirm the actual resistance loss power; the expression of the resistance loss power is as follows:

[0039] P_loss=I 2 *R;

[0040] Where: P_loss is the actual resistance loss power, I 2 is the current passing through the cable, and R is the total resistance of the cable;

[0041] S192. Compare the actual resistance loss power with the theoretical resistance loss power to confirm the line loss correction factor.

[0042] The step S2 specifically includes:

[0043] S21. Collect satellite cloud image data of the photovoltaic power station area, and use machine learning methods to simulate cloud movement and predict the degree of cloud movement blocking solar radiation, thereby evaluating the solar irradiance on the ground;

[0044] S22. Based on the solar irradiance on the ground and the measured irradiance at the current time, calculate the expected irradiance after the solar irradiance decays in the future; the expression of the expected irradiance is as follows:

[0045]

[0046] Where: E t+Δt is the expected irradiance after the future time Δt, E t is the measured irradiance at the current time, k is the cloud attenuation coefficient, C cloud is the proportion of area covered by clouds;

[0047] S23. Calculate the ultra-short-term photovoltaic power generation based on the expected irradiance and the conversion efficiency of the photovoltaic panels in the photovoltaic module; the calculation formula of the ultra-short-term photovoltaic power generation is as follows:

[0048] P=η×E t+△t ×A;

[0049] Where: P is the ultra-short-term photovoltaic power generation power, η is the conversion efficiency of the photovoltaic panel, and A is the area of ​​the photovoltaic panel.

[0050] The step S3 specifically includes:

[0051] S31. Collect historical photovoltaic power generation data and weather data of photovoltaic power stations and build a neural network model;

[0052] S32. Use historical photovoltaic power generation data, weather data, time characteristics, and recent power generation data as input features of the neural network model to make predictions and output short-term photovoltaic power generation in the future time period.

[0053] In step S4, the expression of the weight coefficient is as follows:

[0054]

[0055] Among them: A ultra_short is the accuracy index of ultra-short-term forecast, T factor is the time range factor, W factor is the uncertainty factor of weather change, A short It is an indicator of the accuracy of short-term forecasts;

[0056] T factor =e -λt ;

[0057] Where: λ is the decay rate, and t is the prediction time range.

[0058] The ultra short term in the ultra short term photovoltaic power generation power refers to a time range within 4 hours, and the short term in the short term photovoltaic power generation power refers to a time range within 48 hours.

[0059] A power generation prediction system for a photovoltaic power station, the system is applied to the above method, the system comprising:

[0060] The correction coefficient calculation module is used to calculate the comprehensive efficiency correction coefficient of the photovoltaic power station; the calculation formula of the comprehensive efficiency correction coefficient is as follows:

[0061] K=K1×K2×K3×K4×...×Kn;

[0062] Among them: K is the comprehensive efficiency correction coefficient of the photovoltaic power station, K1 is the photovoltaic module type correction coefficient, K2 is the inclination and azimuth correction coefficient of the photovoltaic array, K3 is the inverter efficiency correction coefficient, K4 is the line loss correction coefficient, and Kn is the correction coefficient of other influencing factors;

[0063] The ultra-short-term power prediction module is used to construct a cloud motion model based on the environmental and climatic factors of the photovoltaic power station area using machine learning methods. The cloud motion model is then used to simulate cloud motion to evaluate the ground radiation intensity and predict the ultra-short-term photovoltaic power generation power.

[0064] The short-term power prediction module is used to construct a short-term photovoltaic power prediction model based on the historical power generation data and weather information of the photovoltaic power station using machine learning methods, and then predict the short-term photovoltaic power generation power through the short-term photovoltaic power prediction model;

[0065] The photovoltaic power generation module is used to predict the photovoltaic power generation power of a photovoltaic power station in a certain future cycle based on the comprehensive efficiency correction coefficient of the photovoltaic power station, the ultra-short-term photovoltaic power generation power, and the short-term period photovoltaic power generation power; the expression for predicting the photovoltaic power generation power in a certain future cycle is as follows:

[0066] P_cycle1=K×(alpha×P_ultra_short+(1-alpha)×P_short);

[0067] Among them: P_cycle1 is the photovoltaic power generation power in a certain cycle in the future, K is the comprehensive efficiency correction coefficient of the photovoltaic power station, P_ultra_short is the ultra-short-term photovoltaic power generation power, P_short is the short-term cycle photovoltaic power generation power, and alpha is the weight coefficient.

[0068] Compared with the prior art, the present invention has the following beneficial effects:

[0069] The invention discloses a power generation prediction method and system for a photovoltaic power station. The method first calculates a comprehensive efficiency correction coefficient of the photovoltaic power station, and the correction coefficient includes correction coefficients of photovoltaic component type, inclination angle and azimuth angle of photovoltaic array, inverter efficiency, line loss and other influencing factors. Then, based on the environmental and climatic factors of the photovoltaic power station area, a cloud movement model is constructed to evaluate the ground radiation intensity, and the ultra-short-term photovoltaic power generation is predicted. Then, based on the historical power generation data and weather information of the photovoltaic power station, a short-term photovoltaic power generation prediction model is constructed, and the short-term period photovoltaic power generation is predicted. Finally, based on the comprehensive efficiency correction coefficient of the photovoltaic power station, the ultra-short-term photovoltaic power generation and the short-term period photovoltaic power generation, the photovoltaic power generation within a certain period in the future of the photovoltaic power station is predicted. In the application of the invention, by comprehensively considering the influence of multiple factors such as the installation of the photovoltaic power station and line loss on the power generation, and combining the climatic conditions such as cloud movement and environment, as well as comprehensive historical data analysis, the refined prediction of the power generation of the photovoltaic power station is realized, so that the prediction result is more consistent with the actual situation, and has higher accuracy and reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 It is a flow chart of the method steps of the present invention.

[0071] Figure 2 It is a schematic diagram of the system structure of the present invention.

[0072] Figure 3 It is a schematic diagram of the device structure of the present invention.

[0073] In the figure: correction coefficient calculation module 1, ultra-short-term power prediction module 2, short-term power prediction module 3, photovoltaic power generation module 4, processor 5, memory 6, computer program code 61. DETAILED DESCRIPTION

[0074] The present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0075] Embodiment 1:

[0076] See also Figure 1 , a method for predicting power generation of a photovoltaic power station, comprising:

[0077] S1. Calculate the comprehensive efficiency correction coefficient of the photovoltaic power station;

[0078] In this embodiment, the comprehensive efficiency correction coefficient takes into account various efficiency losses that may be encountered in the entire process of the photovoltaic power station from receiving solar radiation to converting it into electrical energy, such as inverter efficiency, line loss, component conversion efficiency, etc.; it can be pre-set, for example, a value of 75%-85%; it can also be estimated by calculation.

[0079] In this embodiment, the main factors affecting the photovoltaic power station itself are:

[0080] PV module type: Different types of PV modules have different conversion efficiencies.

[0081] Inverter efficiency: When an inverter converts DC power to AC power, there is energy loss.

[0082] Line loss: There will be a certain amount of loss in the current during transmission.

[0083] Module conversion efficiency: The ability of a module to convert solar energy into electrical energy.

[0084] Other factors: such as the array's inclination angle, azimuth angle, system availability, light utilization rate, etc.

[0085] Methods for calculating the comprehensive efficiency coefficient K: Collect the technical specifications and performance parameters of photovoltaic modules, inverters and other equipment used in photovoltaic power stations; understand the layout of photovoltaic power stations, the inclination angle of the array, the azimuth angle and other design parameters; calculate the correction coefficient of each influencing factor based on the collected data;

[0086] Specifically, the calculation formula of the comprehensive efficiency correction coefficient is as follows:

[0087] K=K1×K2×K3×K4×...×Kn;

[0088] Among them: K is the comprehensive efficiency correction coefficient of the photovoltaic power station, K1 is the photovoltaic module type correction coefficient, K2 is the inclination and azimuth correction coefficient of the photovoltaic array, K3 is the inverter efficiency correction coefficient, K4 is the line loss correction coefficient, and Kn is the correction coefficient of other influencing factors;

[0089] Furthermore, the method for confirming the photovoltaic module type correction factor is as follows:

[0090] S11. Determine the material type of the photovoltaic module and collect relevant technical parameters of the photovoltaic module; the relevant technical parameters include conversion efficiency, maximum output power, open circuit voltage, and short circuit current;

[0091] The material types of the photovoltaic modules include monocrystalline silicon, polycrystalline silicon, amorphous silicon, etc. Different types of modules have different physical properties and performances.

[0092] The relevant technical parameters can be obtained by referring to the technical specification sheet of the photovoltaic module or the data manual provided by the manufacturer.

[0093] S12. Based on the relevant technical parameters of the photovoltaic module, confirm the actual conversion efficiency of the photovoltaic module; the expression of the actual conversion efficiency is as follows:

[0094] η_actual=(Pmax / (Voc*Isc*FF)))*100%;

[0095] Where: η_actual is the actual conversion efficiency of the photovoltaic module, Pmax is the maximum output power, Voc is the open circuit voltage, Isc is the short circuit current, and FF is the fill factor;

[0096] The maximum output power Pmax refers to the maximum power that the photovoltaic module can output under STC conditions, in watts (W); the open circuit voltage Voc refers to the voltage of the photovoltaic module in the open circuit state, in volts (V); the short-circuit current Isc refers to the current of the photovoltaic module in the short-circuit state, in amperes (A); the fill factor FF refers to the ratio of the voltage and current product of the photovoltaic module at the maximum power point to the open circuit voltage and short-circuit current product, which is a dimensionless value.

[0097] S13. Confirm the rated conversion efficiency of the photovoltaic module under standard test conditions; the expression of the rated conversion efficiency is as follows:

[0098] η_rated=(Pmax / (B*A))*100%;

[0099] Where: η_rated is the rated conversion efficiency of the photovoltaic module, B is the incident light power, and A is the area of ​​the photovoltaic module;

[0100] In this embodiment, under standard test conditions (STC), it is assumed that the incident light power is 1000 W / m 2 , and the temperature of the photovoltaic module is set to 25°C. Therefore, if the area of ​​the module (in square units) is known, the expression for the rated conversion efficiency is as follows:

[0101] η_rated=(Pmax / (1000*A))*100%;

[0102] Considering the attenuation characteristics of photovoltaic modules during long-term operation, their performance will gradually decay over time due to various reasons such as material aging and environmental factors. The attenuation rate of the module can be accurately determined based on the data provided by the manufacturer or the industry. Finally, the specific value of the photovoltaic module type correction factor can be determined based on the conversion efficiency and attenuation characteristics of the photovoltaic module, as well as other possible influencing factors (such as dust obstruction, ambient temperature changes, etc.).

[0103] S14. Based on the actual conversion efficiency and rated conversion efficiency of the photovoltaic module, determine the photovoltaic module type correction factor; the expression of the photovoltaic module type correction factor is as follows:

[0104] K1=(η_actual / η_rated)×(1-D);

[0105] Where: D is the lifetime decay rate of the photovoltaic module.

[0106] In this embodiment, it is assumed that the relevant technical parameters collected are: the photovoltaic module is monocrystalline silicon, its rated conversion efficiency η_rated is 18%, and the decay rate D within the expected service life is 0.5% per year. Assuming that the conversion efficiency of the module in the first year is the same as the rated conversion efficiency, that is, η_first_year = 18%, the conversion efficiency in the second year is as follows:

[0107] η_second_year=eta_first_year×(1-D)=18%×(1-0.5%)=17.91%;

[0108] The PV module type correction factor K1 is determined as follows:

[0109] K1=(n_second_year / n_rated)=17.91% / 18%≈0.995.

[0110] Furthermore, the method for confirming the tilt angle and azimuth angle correction coefficients of the photovoltaic array is as follows:

[0111] S15. Confirm the optimal inclination angle and azimuth angle of the photovoltaic array installation position, as well as the inclination angle and azimuth angle during actual installation;

[0112] S16. Taking the power generation of the photovoltaic array at the optimal inclination angle and azimuth angle as the reference condition, and calculating the deviation between the power generation of the photovoltaic array at the inclination angle and azimuth angle during actual installation and the reference condition; the expression of the correction coefficient is as follows:

[0113] K2≈angle optimal / angle Actual ;

[0114] Where: angle optimal is the power generation under the best inclination and azimuth, angle Actual It is the power generation under the actual installation inclination and azimuth.

[0115] In this embodiment, the reference conditions are first determined, that is, the optimal inclination and azimuth of the photovoltaic array installation location. Usually, the photovoltaic array faces due south, and the power generation is maximum when the azimuth is 0°; and the inclination is the local optimal inclination, which is usually close to the latitude angle, but may be adjusted due to seasons and specific needs; then the deviation of the inclination and azimuth during actual installation from the reference conditions is calculated. Photovoltaic simulation software (such as PVsyst, SAM, etc.) or empirical formulas can be used to evaluate the impact of these deviations on power generation; based on the above evaluation, the correction factor K2 is calculated, which reflects the reduction in power generation due to the deviation of the inclination and azimuth.

[0116] Assuming that the function representing the relative power generation at a given inclination angle \theta and azimuth angle \phi compared to the baseline condition is (f(\theta, \phi)), K2 can be approximated as:

[0117] K2≈power generation at optimal inclination and azimuth / power generation at actual inclination and azimuth;

[0118] In this embodiment, it is assumed that the photovoltaic system is located at 40° north latitude, and the optimal inclination angle under baseline conditions is 40°, facing due south (azimuth 0°); however, due to site limitations, the actual installation inclination angle is 35° and the azimuth is 5° east.

[0119] Photovoltaic simulation software was then used to simulate the power generation under baseline conditions (40° inclination, 0° azimuth) and actual installation conditions (35° inclination, 5° east azimuth). The simulation results showed that the annual power generation under baseline conditions was 1000kWh / kWp, while the annual power generation under actual installation conditions was 950kWh / kWp.

[0120] The correction coefficient K2 is: K2=950 / 1000=0.95; this coefficient indicates that due to the deviation of the inclination angle and the azimuth angle, the actual power generation is reduced by 5% compared with the power generation under the reference conditions.

[0121] Furthermore, the method for confirming the inverter efficiency correction coefficient is as follows:

[0122] S17, confirming the ideal efficiency data of the inverter, and selecting the inverter efficiency value closest to the actual situation according to the actual operating conditions of the photovoltaic module to obtain the actual efficiency data;

[0123] S18. Calculate the inverter efficiency correction factor based on the ideal efficiency data and the actual efficiency data; the expression of the inverter efficiency correction factor is as follows:

[0124] K3=K actual / K ideal ;

[0125] Where: K actual is the actual efficiency data, K ideal It is the ideal efficiency data or the benchmark efficiency data (usually 100% or a value close to 1).

[0126] In this embodiment, the inverter efficiency generally refers to the ratio of the inverter output power to the input power under specific working conditions. The inverter efficiency is affected by many factors, including input voltage range, load conditions, ambient temperature, etc. The ideal efficiency data of the inverter can be obtained through the technical manual or product specification sheet provided by the inverter manufacturer to obtain the efficiency curve or efficiency data of the inverter. These data may include efficiency values ​​under different load conditions, or an average efficiency value.

[0127] Then, according to the actual operating conditions of the photovoltaic power generation system (such as sunshine intensity, load demand, etc.), select the inverter efficiency value that is closest to the actual situation; if the efficiency data is in the form of a curve, it may be necessary to calculate the efficiency value under specific conditions through interpolation or fitting formulas; if the inverter efficiency is high and close to the ideal value, K3 is close to 1; if the inverter efficiency is low, K3 is less than 1. However, in actual applications, the efficiency value provided by the inverter manufacturer is often used directly as K actual , and the ideal efficiency data K ideal Considered as 1 or a reference value close to 1 is set according to the actual situation), which can be simplified to: K3 = K actual .

[0128] In this embodiment, it is assumed that the efficiency of the inverter used in a photovoltaic system is 95% under standard test conditions, that is, K actual =0.95, and the ideal efficiency data is set to 1 or 100%; in this case, the inverter efficiency correction coefficient K3 is directly equal to the actual efficiency value of the inverter, that is, K3=0.95.

[0129] Furthermore, the method for confirming the line loss correction coefficient is as follows:

[0130] S19. Determine the total resistance of the cable based on the resistivity, length and cross-sectional area of ​​the cable in the photovoltaic module; the expression of the total resistance is as follows:

[0131]

[0132] Where: R is the total resistance, ρ is the resistivity of the cable, L is the length, and A is the cross-sectional area;

[0133] S191. Based on the total resistance of the cable and the current passing through it, confirm the actual resistance loss power; the expression of the resistance loss power is as follows:

[0134] P_loss=I 2 *R;

[0135] Where: P_loss is the actual resistance loss power, I 2 is the current passing through the cable, and R is the total resistance of the cable;

[0136] S192. Compare the actual resistance loss power with the theoretical resistance loss power to confirm the line loss correction factor.

[0137] In this embodiment, the line loss mainly includes resistance loss and electromagnetic induction loss; the resistance loss is proportional to the square of the current and inversely proportional to the cable resistance. Electromagnetic induction loss (such as inductance and capacitance effects) is more significant in high-frequency currents, but is usually not a major factor in photovoltaic systems.

[0138] When calculating the loss, first determine the specifications and performance parameters of the PV modules, the efficiency of the inverter and other power electronic equipment, the specifications, length and material of the cables, the layout and installation conditions of the PV system (such as temperature, humidity, pollution level, etc.). Then use the resistivity, length and cross-sectional area of ​​the cable to calculate the total resistance of the cable, and calculate the resistance loss power based on the current and cable resistance, convert the resistance loss power into energy loss, consider the system's operating time and efficiency, and consider other factors, such as the effect of ambient temperature on cable resistance (usually corrected using a temperature coefficient), the contact resistance and loss of the cable joints, and other non-ideal factors in the system, such as pollution, aging, etc.

[0139] In this embodiment, the theoretical resistance loss power can refer to the following table:

[0140] P_loss K4 [0,10W] 0.99 (10W, 15W] 0.97 >15W 0.95

[0141] In the line loss correction coefficient, theoretically, the resistance loss power is lossless, then the parameter value of K4 is 1. In actual situations, there is usually loss in the line, then the parameter value of K4 should be less than 1. In practice, it can be adjusted according to the actual situation, and the K4 value corresponding to different resistance losses is set by testing the impact on the comprehensive efficiency coefficient, and finally the appropriate values ​​of the actual resistance loss power and the theoretical resistance loss power are determined.

[0142] In this embodiment, it is assumed that the photovoltaic system uses a copper cable with a length of 100 meters, a cable cross-sectional area of ​​2.5 square millimeters, a system operating current of 10A, an ambient temperature of 25°C, and a resistivity of copper at 25°C of approximately 1.71*10-82·m; the calculated total cable resistance is R=1.7*10 -8 *100 / 2.5*10 -4 =00.0682.

[0143] The estimated resistance power loss is as follows:

[0144] P_loss=I 2 *R=10 2 *0.068 = 6.8W;

[0145] By looking up the benchmark resistor power loss table, 6.8W belongs to the [0,10W] range, so the value of K4 is 0.99.

[0146] In summary, taking a photovoltaic power station as an example, the determination process of its comprehensive efficiency coefficient K is as follows:

[0147] Assuming that the type of photovoltaic module used is a high-efficiency crystalline silicon module with a conversion efficiency of 18%, the correction coefficient K1 is 0.95; assuming that the inclination and azimuth of the array have been optimized, the correction coefficient K2 is 0.98 (indicating that the inclination and azimuth settings are relatively reasonable and the efficiency loss is small); assuming that the efficiency of the inverter used is 97%, the correction coefficient K3 is 0.97; taking into account the line loss, the correction coefficient K4 is 0.99 (indicating that the line loss is small); assuming that the sum of the correction coefficients Kn of other influencing factors is 0.98 (indicating that the efficiency loss caused by other factors is small).

[0148] The comprehensive efficiency coefficient K is as follows:

[0149] K=0.95×0.98×0.97×0.99×0.98≈0.87.

[0150] S2. Based on the environmental and climatic factors of the photovoltaic power station area, a cloud motion model is constructed using machine learning methods. Then, the cloud motion model is used to simulate cloud motion to evaluate the ground radiation intensity and predict the ultra-short-term photovoltaic power generation power.

[0151] Further, it specifically includes:

[0152] S21. Collect satellite cloud image data of the photovoltaic power station area, and use machine learning methods to simulate cloud movement and predict the degree of cloud movement blocking solar radiation, thereby evaluating the solar irradiance on the ground;

[0153] S22. Based on the solar irradiance on the ground and the measured irradiance at the current time, calculate the expected irradiance after the solar irradiance decays in the future; the expression of the expected irradiance is as follows:

[0154]

[0155] Where: E t+Δt is the expected irradiance after the future time Δt, E t is the measured irradiance at the current time, k is the cloud attenuation coefficient, C cloud is the proportion of area covered by clouds;

[0156] S23. Calculate the ultra-short-term photovoltaic power generation based on the expected irradiance and the conversion efficiency of the photovoltaic panels in the photovoltaic module; the calculation formula of the ultra-short-term photovoltaic power generation is as follows:

[0157] P=η×E t+Δt ×A;

[0158] Where: P is the ultra-short-term photovoltaic power generation power, η is the conversion efficiency of the photovoltaic panel, and A is the area of ​​the photovoltaic panel.

[0159] In this embodiment, it is first necessary to collect satellite cloud image data, which are usually provided by meteorological satellites and can display information such as the distribution, thickness and movement speed of clouds; then, image processing technology is used to extract key features from the cloud image, such as the coverage area of ​​clouds, the type of clouds (such as cumulus clouds, stratus clouds, etc.), the movement direction and speed of clouds, etc.; then, according to the characteristics of the clouds and the known meteorological data, a cloud motion model is established, such as machine learning-based methods such as convolutional neural networks (CNN) and long short-term memory networks (LSTM), to simulate the movement of clouds over time.

[0160] When collecting satellite cloud image data of the photovoltaic power station area, satellite images or radar image sequences containing cloud movement are collected; these images contain timestamps so that the movement of the clouds can be tracked; and meteorological data corresponding to the image sequence, such as wind speed, wind direction, air pressure, etc., are collected; then the image data needs to be preprocessed as necessary, such as denoising, normalization, enhancement, etc., and the meteorological data is appropriately processed, such as interpolation, standardization, etc.

[0161] After preprocessing the data, CNN or other image processing techniques are used to extract cloud features such as shape, texture, density, etc. from the image to capture the static properties of the clouds. Then, by comparing the cloud positions in consecutive image frames, the motion features of the clouds, such as speed, acceleration, trajectory, etc., are calculated to capture the dynamic properties of the clouds. Finally, meteorological data is input into the model as an additional feature to consider the impact of meteorological conditions on cloud motion.

[0162] When applied, CNN is used to extract the spatial features of clouds from images. The convolutional layer of CNN can learn local patterns in images, while the pooling layer can capture the spatial hierarchy of these patterns. LSTM is used to capture the temporal dependency of cloud motion. LSTM is a special recurrent neural network (RNN) that can process long-term dependencies in sequence data. The LSTM layer can receive the features extracted by CNN and meteorological features as input, and predict the position and state of clouds at the next time step; the outputs of CNN and LSTM are fused together to predict the final degree of cloud shielding from solar radiation, which can be achieved through concatenation, addition or other forms of feature fusion; finally, according to the movement and characteristics of clouds, the degree of cloud shielding from solar radiation is predicted through the cloud motion model, and then the solar irradiance on the ground is predicted, and the irradiance prediction results and the performance parameters of photovoltaic panels (such as conversion efficiency) are combined to predict the photovoltaic power generation.

[0163] S3. Based on the historical power generation data and weather information of the photovoltaic power station, a short-term photovoltaic power generation prediction model is constructed using a machine learning method. Then, the short-term photovoltaic power generation prediction model is used to predict the short-term period photovoltaic power generation.

[0164] Further, it specifically includes:

[0165] S31. Collect historical photovoltaic power generation data and weather data of photovoltaic power stations and build a neural network model;

[0166] S32. Use historical photovoltaic power generation data, weather data, time characteristics, and recent power generation data as input features of the neural network model to make predictions and output short-term photovoltaic power generation in the future time period.

[0167] In this embodiment, historical photovoltaic power generation data and weather data of the photovoltaic power station are collected; the historical photovoltaic power generation data includes: historical photovoltaic power generation data of the photovoltaic power station, timestamp (such as data specific to hours or 15-minute intervals), solar radiation, temperature, humidity and other weather parameters.

[0168] The data is then cleaned, such as processing missing values ​​and outliers to ensure the integrity and accuracy of the data; the data is then standardized / normalized, such as converting data of different dimensions into the same dimension, usually by scaling the data to the range of 0 to 1, to avoid affecting the training effect of the model due to large differences in the numerical ranges of different features.

[0169] Next, select the features to build the neural network model, for example:

[0170] Time features: hour, date, day of the week, month, season, etc., to capture the periodic changes in photovoltaic power generation; weather features: solar radiation, temperature, humidity, wind speed, wind direction, cloud coverage, etc., these features have a direct impact on photovoltaic power generation; historical power data: select power generation data over a period of time (such as a few hours or a day) as input features, because changes in photovoltaic power generation usually have a certain degree of continuity.

[0171] Next, a neural network model is constructed. For short-term prediction, the model structure in this embodiment selects a recursive neural network such as LSTM or GRU because they can capture long-term dependencies in time series data. In the hidden layer, the ReLU (Rectified Linear Unit) activation function is selected to accelerate training and alleviate the gradient vanishing problem. In the output layer, since continuous values ​​(photovoltaic power generation) need to be predicted, no activation function is used or a linear activation function is used. For the loss function, the mean square error (MSE) or mean absolute error (MAE) loss function is selected to measure the accuracy of the model prediction. The optimizer uses Adam, RMSprop or SGD (with momentum) to update the weights of the model during training.

[0172] The neural network model is then trained and evaluated. During training, the data set is first divided into a training set, a validation set, and a test set. The training set is used to train the model, the validation set is used to adjust model parameters and prevent overfitting, and the test set is used to evaluate the generalization ability of the model. The training set data is then used to train the neural network model, and the weights of the model are optimized through the back propagation algorithm and the selected optimizer. The validation set is then used to evaluate the performance of the model, such as prediction accuracy, error rate, etc., and the model parameters or structure are adjusted according to the evaluation results. Finally, the test set is used to test the performance of the model to ensure that the model can maintain good prediction capabilities on unseen data.

[0173] Finally, the weather forecast data for the time period to be predicted is used as input data and input into the trained neural network model to obtain the predicted photovoltaic power generation value. The prediction results are then post-processed as needed, such as denormalization / normalization, smoothing, etc.

[0174] In this embodiment, it is assumed that the data set contains medium- and long-term data such as photovoltaic power generation power, solar radiation, temperature, etc. every hour of every day in the past year, as well as weather forecast data for the next 24 hours. The prediction target is the photovoltaic power generation power per hour in the next 24 hours.

[0175] The photovoltaic power generation, solar radiation, temperature and other data in the historical data are standardized. In addition to the above three features, time features (such as hour, date, day of the week) and power generation data in the past few hours (such as the last 4 hours) are added as input features; a neural network model suitable for processing time series data (such as LSTM) is selected, and appropriate model parameters (such as the number of hidden layers, number of neurons, etc.) are set; then the model is trained using the data of the past year as a training set, and the performance of the model is evaluated using a validation set; then the weather forecast data and other relevant features for the next 24 hours are input into the trained model to obtain the photovoltaic power generation forecast value for each hour in the next 24 hours, and the forecast results are de-standardized and smoothed as much as possible to reduce the impact of noise.

[0176] Through this process, it is possible to use the historical data and weather information of the power station itself or surrounding power stations to train a short-term photovoltaic power generation prediction model through an artificial neural network and make predictions.

[0177] In this embodiment, the time range referred to by the ultra-short term in the ultra-short term photovoltaic power generation power is within 4 hours, and the time range referred to by the short term in the short-term photovoltaic power generation power is within 48 hours.

[0178] S4. Based on the comprehensive efficiency correction coefficient of the photovoltaic power station, the ultra-short-term photovoltaic power generation power, and the short-term period photovoltaic power generation power, the photovoltaic power generation power of the photovoltaic power station in a certain future period is predicted; the expression for predicting the photovoltaic power generation power in a certain future period is as follows:

[0179] P_cyclel=K×(alpha×P_ultra_short+(1-alpha)×P_short);

[0180] Among them: P_cycle1 is the photovoltaic power generation power in a certain cycle in the future, K is the comprehensive efficiency correction coefficient of the photovoltaic power station, P_ultra_short is the ultra-short-term photovoltaic power generation power, P_short is the short-term cycle photovoltaic power generation power, alpha is the weight coefficient, and the weight coefficient alpha is used to adjust the contribution of ultra-short-term and short-term prediction results in the final prediction. This coefficient can be determined based on the accuracy of historical data, the time range of the prediction, the uncertainty of weather changes and other factors.

[0181] In the calculation process of the weight coefficient, an initial alpha value is set (such as 0.5, indicating that the weights of the two are equal), and then adjusted according to historical accuracy; if the historical accuracy of the ultra-short-term forecast is higher than that of the short-term forecast, the alpha value is increased.

[0182] If the prediction time range is taken into consideration, generally speaking, ultra-short-term predictions may be more accurate because they are closer to the actual time point, so they tend to be given a higher weight. If the weather forecast shows that the weather will change significantly in the future, the weight of the short-term prediction may be reduced, that is, the complement of alpha 1-alpha is reduced. At the same time, a dynamic adjustment mechanism can be set to dynamically update the value of alpha based on new prediction results and real-time data.

[0183] The expression of the weight coefficient is as follows:

[0184]

[0185] Among them: A ultra _short is the accuracy index of ultra-short-term forecast, T factor is the time range factor, W factor is the uncertainty factor of weather change, A short is the accuracy index of short-term forecast; ultra_short With A short It refers to the measurement index used to reflect the accuracy of ultra-short-term and short-term forecasts. Its measurement function includes any one of the error rate, mean square error, root mean square error, and mean absolute error.

[0186] T factor =e -t ;

[0187] Where: λ is the decay rate, and t is the prediction time range.

[0188] In this embodiment, Altraa_short is an evaluation based on historical data, such as the inverse of accuracy, the inverse of error, or other indicators of accuracy; Altraa_short is a numerical value that reflects the accuracy and reliability of ultra-short-term forecasts (usually forecasts within 0-6 hours). This indicator can be a function of multiple metrics, including but not limited to error rate, mean square error (MSE), root mean square error (RMSE), mean absolute error (MAE), etc.

[0189] Mean Squared Error (MSE) is the average of the squares of the differences between the predicted values ​​and the true values. It measures the average squared deviation between the predicted values ​​and the actual observed values.

[0190] Root Mean Squared Error (RMSE) RMSE is the square root of MSE and has the same units as MSE, so it is easier to interpret.

[0191] Mean Absolute Error (MAE) is the average of the absolute values ​​of the differences between the predicted values ​​and the true values. It measures the average absolute deviation between the predicted values ​​and the actual observed values.

[0192] A short It is also an evaluation based on historical data and is calculated in the same way as A ultra_short Similarly, it can also be a function of multiple metrics.

[0193] T factor is a factor used to adjust the effect of the forecast time range on the weight; generally speaking, as the forecast time range increases, the accuracy of the forecast may decrease, so T factor It may be a function that decreases as the prediction time increases.

[0194] For example, you can set a base time range (such as 1 hour) and assign it a higher T factor value (such as 1.0), and then gradually reduce T as the prediction time range increases. factor value; such as using exponential decay, linear decay or other suitable decay functions to simulate T factor A decreasing trend with increasing time range, e.g., T factor =e -λt .

[0195] W factor It is used to reflect the impact of weather forecast uncertainty on forecast accuracy; the higher the uncertainty, the higher the W factor The smaller the value may be, the weather change uncertainty factor can be set in advance according to different weather conditions, fixed parameters, such as setting fixed parameters according to solar radiation intensity, temperature, wind speed, etc.

[0196] In the calculation formula of the weight coefficient, the numerator A ultra_short ·T factor ·W factor The “weighted accuracy” of the ultra-short-term forecast is calculated after taking into account the uncertainty of the time horizon and weather changes; the denominator A ultra_short ·T factor · Wfactor +A short ·(1-W factor ) adds the weighted accuracy of the ultra-short-term forecast to the weighted accuracy of the short-term forecast; it should be noted that the weighted accuracy of the short-term forecast is calculated by its accuracy index A short and (1-W factor ), which reflects that when the uncertainty of weather changes is high, the weight of short-term forecasts will be reduced accordingly.

[0197] Finally, the calculation result of the entire formula is: the value of alpha represents the contribution weight of the ultra-short-term prediction in the final prediction result; when the weighted accuracy of the ultra-short-term prediction is much higher than that of the short-term prediction, the value of alpha will be close to 1; conversely, when the weighted accuracy of the short-term prediction is higher, the value of alpha will be close to 0 (but in fact, since the weighted accuracy of the ultra-short-term prediction is always included in the denominator, alpha will not be equal to 0).

[0198] In this embodiment, the weight coefficient calculation formula is a heuristic method for dynamically adjusting the weights of ultra-short-term and short-term predictions in the final prediction based on multiple factors. However, in actual applications, this formula needs to be adjusted and optimized based on specific data and scenarios.

[0199] Assume K = 0.9 (indicates that the overall efficiency of the photovoltaic power station is 90%),

[0200] P_ultra_short = 10000kW (indicating that the ultra-short-term prediction value is 1000 kilowatts), P_short = 1050kW (indicating that the short-term prediction value is 1050 kilowatts), alpha = 0.6 (indicating that a greater weight is given to the ultra-short-term prediction), then the photovoltaic power generation power prediction value in cycle 1 is as follows:

[0201]

[0202] In summary, the photovoltaic power generation project power generation prediction method of the present invention covers a variety of prediction technologies from ultra-short term to medium and long term, combines direct prediction, indirect prediction and other prediction principles, and combines hybrid prediction methods with advanced technologies such as machine learning and satellite remote sensing to improve the accuracy of photovoltaic power generation prediction.

[0203] Embodiment 2:

[0204] See also Figure 2 , a power generation prediction system for a photovoltaic power station, the system is applied to the method described in Example 1, the system comprising:

[0205] The correction coefficient calculation module 1 is used to calculate the comprehensive efficiency correction coefficient of the photovoltaic power station; the calculation formula of the comprehensive efficiency correction coefficient is as follows:

[0206] K=K1×K2×K3×K4×...×Kn;

[0207] Among them: K is the comprehensive efficiency correction coefficient of the photovoltaic power station, K1 is the photovoltaic module type correction coefficient, K2 is the inclination and azimuth correction coefficient of the photovoltaic array, K3 is the inverter efficiency correction coefficient, K4 is the line loss correction coefficient, and Kn is the correction coefficient of other influencing factors;

[0208] Furthermore, the correction coefficient calculation module 1 calculates the coefficients according to the following methods:

[0209] The method for confirming the correction factor of the photovoltaic module type is as follows:

[0210] S11. Determine the material type of the photovoltaic module and collect relevant technical parameters of the photovoltaic module; the relevant technical parameters include conversion efficiency, maximum output power, open circuit voltage, and short circuit current;

[0211] S12. Based on the relevant technical parameters of the photovoltaic module, confirm the actual conversion efficiency of the photovoltaic module; the expression of the actual conversion efficiency is as follows:

[0212] η_actual=(Pmax / (Voc*Isc*FF))*100%;

[0213] Where: η_actual is the actual conversion efficiency of the photovoltaic module, Pmax is the maximum output power, Voc is the open circuit voltage, Isc is the short circuit current, and FF is the fill factor;

[0214] S13. Confirm the rated conversion efficiency of the photovoltaic module under standard test conditions; the expression of the rated conversion efficiency is as follows:

[0215] η_rated=(Pmax / (B*A))*100%;

[0216] Where: η_rated is the rated conversion efficiency of the photovoltaic module, B is the incident light power, and A is the area of ​​the photovoltaic module;

[0217] S14. Based on the actual conversion efficiency and rated conversion efficiency of the photovoltaic module, determine the photovoltaic module type correction factor; the expression of the photovoltaic module type correction factor is as follows:

[0218] K1=(η_actual / η_rated)×(1-D);

[0219] Where: D is the lifetime decay rate of the photovoltaic module.

[0220] The method for confirming the inclination angle and azimuth angle correction coefficients of the photovoltaic array is as follows:

[0221] S15. Confirm the optimal inclination angle and azimuth angle of the photovoltaic array installation position, as well as the inclination angle and azimuth angle during actual installation;

[0222] S16. Taking the power generation of the photovoltaic array at the optimal inclination angle and azimuth angle as the reference condition, and calculating the deviation between the power generation of the photovoltaic array at the inclination angle and azimuth angle during actual installation and the reference condition; the expression of the correction coefficient is as follows:

[0223] K2≈angle optimal / angle Actual ;

[0224] Where: angle optimal is the power generation under the best inclination and azimuth, angle Actual It is the power generation under the actual installation inclination and azimuth.

[0225] The method for confirming the inverter efficiency correction factor is as follows:

[0226] S17, confirming the ideal efficiency data of the inverter, and selecting the inverter efficiency value closest to the actual situation according to the actual operating conditions of the photovoltaic module to obtain the actual efficiency data;

[0227] S18. Calculate the inverter efficiency correction factor based on the ideal efficiency data and the actual efficiency data; the expression of the inverter efficiency correction factor is as follows:

[0228] K3=K actual / K ideal ;

[0229] Where: K actual is the actual efficiency data, K ideal This is ideal efficiency data.

[0230] The method for confirming the line loss correction factor is as follows:

[0231] S19. Determine the total resistance of the cable based on the resistivity, length and cross-sectional area of ​​the cable in the photovoltaic module; the expression of the total resistance is as follows:

[0232]

[0233] Where: R is the total resistance, ρ is the resistivity of the cable, L is the length, and A is the cross-sectional area;

[0234] S191. Based on the total resistance of the cable and the current passing through it, confirm the actual resistance loss power; the expression of the resistance loss power is as follows:

[0235] P_loss=I 2 *R:

[0236] Where: P_loss is the actual resistance loss power, I 2 is the current passing through the cable, and R is the total resistance of the cable;

[0237] S192. Compare the actual resistance loss power with the theoretical resistance loss power to confirm the line loss correction factor.

[0238] Ultra-short-term power prediction module 2 is used to construct a cloud motion model based on the environmental and climatic factors of the photovoltaic power station area using machine learning methods, and then simulate cloud motion through the cloud motion model to evaluate the ground radiation intensity and predict the ultra-short-term photovoltaic power generation power;

[0239] Furthermore, the ultra-short-term power prediction module 2 predicts the ultra-short-term photovoltaic power generation power according to the following steps:

[0240] S21. Collect satellite cloud image data of the photovoltaic power station area, and use machine learning methods to simulate cloud movement and predict the degree of cloud movement blocking solar radiation, thereby evaluating the solar irradiance on the ground;

[0241] S22. Based on the solar irradiance on the ground and the measured irradiance at the current time, calculate the expected irradiance after the solar irradiance decays in the future; the expression of the expected irradiance is as follows:

[0242]

[0243] Where: E t+Δt is the expected irradiance after the future time Δt, E t is the measured irradiance at the current time, k is the cloud attenuation coefficient, C cloud is the proportion of area covered by clouds;

[0244] S23. Calculate the ultra-short-term photovoltaic power generation based on the expected irradiance and the conversion efficiency of the photovoltaic panels in the photovoltaic module; the calculation formula of the ultra-short-term photovoltaic power generation is as follows:

[0245] P=η×E t+△ ×A;

[0246] Where: P is the ultra-short-term photovoltaic power generation power, η is the conversion efficiency of the photovoltaic panel, and A is the area of ​​the photovoltaic panel.

[0247] The short-term power prediction module 3 is used to construct a short-term photovoltaic power prediction model based on the historical power generation data and weather information of the photovoltaic power station by using a machine learning method, and then predict the short-term photovoltaic power generation power through the short-term photovoltaic power prediction model;

[0248] Furthermore, the short-term power prediction module 3 predicts the short-term period photovoltaic power generation power according to the following steps:

[0249] S31. Collect historical photovoltaic power generation data and weather data of photovoltaic power stations and build a neural network model;

[0250] S32. Use historical photovoltaic power generation data, weather data, time characteristics, and recent power generation data as input features of the neural network model to make predictions and output short-term photovoltaic power generation in the future time period.

[0251] Photovoltaic power generation module 4 is used to predict the photovoltaic power generation power of a photovoltaic power station in a certain future cycle based on the comprehensive efficiency correction coefficient of the photovoltaic power station, the ultra-short-term photovoltaic power generation power, and the short-term period photovoltaic power generation power; the expression for predicting the photovoltaic power generation power in a certain future cycle is as follows:

[0252] Pcyclel=K×(alpha×P_ultra_short+(1-alpha)×P_short);

[0253] Among them: P_cclel is the photovoltaic power generation power in a certain period in the future, K is the comprehensive efficiency correction coefficient of the photovoltaic power station, P_ultra_short is the ultra-short-term photovoltaic power generation power, P_short is the short-term photovoltaic power generation power, and alpha is the weight coefficient.

[0254] Furthermore, the photovoltaic power generation module 4 calculates the weight coefficient according to the following expression:

[0255]

[0256] Among them: A ultra_short is the accuracy index of ultra-short-term forecast, T factor is the time range factor, W factor is the uncertainty factor of weather change, A shor t is the accuracy index of short-term forecast; ultra_short With A short It refers to the measurement index used to reflect the accuracy of ultra-short-term and short-term forecasts. Its measurement function includes any one of the error rate, mean square error, root mean square error, and mean absolute error.

[0257] T factor =e -λt ;

[0258] Where: λ is the decay rate, and t is the prediction time range.

[0259] Embodiment 3:

[0260] See also Figure 3 , in this embodiment, it also includes a power generation prediction device for a photovoltaic power station, the device includes a processor 5 and a memory 6;

[0261] The memory 6 is used to store the computer program code 61 and transmit the computer program code 61 to the processor 5;

[0262] The processor 5 is used to execute the method for predicting the power generation of the photovoltaic power station described in Example 1 according to the instructions in the computer program code 61 .

[0263] This embodiment also includes a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed on a computer, the method for predicting the power generation of a photovoltaic power station described in Example 1 is implemented.

[0264] Generally speaking, the computer instructions for implementing the method of the present invention may be carried in any combination of one or more computer-readable storage media. Non-transitory computer-readable storage media may include any computer-readable media, except for the signal itself that is temporarily propagating.

[0265] The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EKROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, device, or device.

[0266] Computer program code for performing the operation of the present invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, SMalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages, in particular, Python suitable for neural network computing and platform frameworks based on TensorFlow, PyTorch, etc. can be used. The program code can be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer or to an external computer (for example, using an Internet service provider to connect via the Internet) through any type of network, including a local area network (LAN) or a wide area network (WAN).

[0267] The above-mentioned device and non-temporary computer-readable storage medium can refer to the specific description of a method for predicting power generation of a photovoltaic power station and its beneficial effects, which will not be repeated here.

[0268] Although the embodiments of the present invention have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A method for predicting power generation of a photovoltaic power station, characterized in that: include: S1. Calculate the comprehensive efficiency correction coefficient of the photovoltaic power station; the calculation formula of the comprehensive efficiency correction coefficient is as follows: K=K1×K2×K3×K4×...×Kn; Among them: K is the comprehensive efficiency correction coefficient of the photovoltaic power station, K1 is the photovoltaic module type correction coefficient, K2 is the inclination and azimuth correction coefficient of the photovoltaic array, K3 is the inverter efficiency correction coefficient, K4 is the line loss correction coefficient, and Kn is the correction coefficient of other influencing factors; S2. Based on the environmental and climatic factors of the photovoltaic power station area, a cloud motion model is constructed using machine learning methods. Then, the cloud motion model is used to simulate cloud motion to evaluate the ground radiation intensity and predict the ultra-short-term photovoltaic power generation power. S3. Based on the historical power generation data and weather information of the photovoltaic power station, a short-term photovoltaic power generation prediction model is constructed using a machine learning method. Then, the short-term photovoltaic power generation prediction model is used to predict the short-term period photovoltaic power generation. S4. Based on the comprehensive efficiency correction coefficient of the photovoltaic power station, the ultra-short-term photovoltaic power generation power, and the short-term period photovoltaic power generation power, the photovoltaic power generation power of the photovoltaic power station in a certain future period is predicted; the expression for predicting the photovoltaic power generation power in a certain future period is as follows: P_cycle1=K×(alpha×P_ultra_short+(1-alpha)×P_short); Among them: P_cycle1 is the photovoltaic power generation power in a certain cycle in the future, K is the comprehensive efficiency correction coefficient of the photovoltaic power station, P_ultra_short is the ultra-short-term photovoltaic power generation power, P_short is the short-term cycle photovoltaic power generation power, and alpha is the weight coefficient.

2. The method for predicting power generation of a photovoltaic power station according to claim 1, characterized in that: The method for confirming the correction factor of the photovoltaic module type is as follows: S11. Determine the material type of the photovoltaic module and collect relevant technical parameters of the photovoltaic module; the relevant technical parameters include conversion efficiency, maximum output power, open circuit voltage, and short circuit current; S12. Based on the relevant technical parameters of the photovoltaic module, confirm the actual conversion efficiency of the photovoltaic module; the expression of the actual conversion efficiency is as follows: η_actual=(Pmax / (Voc*Isc*FF))*100%; Where: η_actual is the actual conversion efficiency of the photovoltaic module, Pmax is the maximum output power, Voc is the open circuit voltage, Isc is the short circuit current, and FF is the fill factor; S13. Confirm the rated conversion efficiency of the photovoltaic module under standard test conditions; the expression of the rated conversion efficiency is as follows: η_rated=(Pmax / (B*A))*100%; Where: η_rated is the rated conversion efficiency of the photovoltaic module, B is the incident light power, and A is the area of ​​the photovoltaic module; S14. Based on the actual conversion efficiency and rated conversion efficiency of the photovoltaic module, determine the photovoltaic module type correction factor; the expression of the photovoltaic module type correction factor is as follows: K1=(η_actual / η_rated)×(1-D); Where: D is the lifetime decay rate of the photovoltaic module.

3. The method for predicting power generation of a photovoltaic power station according to claim 1, characterized in that: The method for confirming the inclination angle and azimuth angle correction coefficients of the photovoltaic array is as follows: S15. Confirm the optimal inclination angle and azimuth angle of the photovoltaic array installation position, as well as the inclination angle and azimuth angle during actual installation; S16. Taking the power generation of the photovoltaic array at the optimal inclination angle and azimuth angle as the reference condition, and calculating the deviation between the power generation of the photovoltaic array at the inclination angle and azimuth angle during actual installation and the reference condition; the expression of the correction coefficient is as follows: K2≈angle optimal / angle Actual ; Where: angle optimal is the power generation under the best inclination and azimuth, angle Actual It is the power generation under the actual installation inclination and azimuth.

4. The method for predicting power generation of a photovoltaic power station according to claim 1, characterized in that: The method for confirming the inverter efficiency correction factor is as follows: S17, confirming the ideal efficiency data of the inverter, and selecting the inverter efficiency value closest to the actual situation according to the actual operating conditions of the photovoltaic module to obtain the actual efficiency data; S18. Calculate the inverter efficiency correction factor based on the ideal efficiency data and the actual efficiency data; the expression of the inverter efficiency correction factor is as follows: K3=K actual / K ideal ; Where: K actual is the actual efficiency data, K ideal This is ideal efficiency data.

5. The method for predicting power generation of a photovoltaic power station according to claim 1, characterized in that: The method for confirming the line loss correction factor is as follows: S19. Determine the total resistance of the cable based on the resistivity, length and cross-sectional area of ​​the cable in the photovoltaic module; the expression of the total resistance is as follows: Where: R is the total resistance, ρ is the resistivity of the cable, L is the length, and A is the cross-sectional area; S191. Based on the total resistance of the cable and the current passing through it, confirm the actual resistance loss power; the expression of the resistance loss power is as follows: P_loss=I 2 *R; Where: P_loss is the actual resistance loss power, I 2 is the current passing through the cable, and R is the total resistance of the cable; S192. Compare the actual resistance loss power with the theoretical resistance loss power to confirm the line loss correction factor.

6. The method for predicting power generation of a photovoltaic power station according to claim 1, characterized in that: The step S2 specifically includes: S21. Collect satellite cloud image data of the photovoltaic power station area, and use machine learning methods to simulate cloud movement and predict the degree of cloud movement blocking solar radiation, thereby evaluating the solar irradiance on the ground; S22. Based on the solar irradiance on the ground and the measured irradiance at the current time, calculate the expected irradiance after the solar irradiance decays in the future; the expression of the expected irradiance is as follows: Where: E t+△t is the expected irradiance after time △t in the future, E t is the measured irradiance at the current time, k is the cloud attenuation coefficient, C cloud is the proportion of area covered by clouds; S23. Calculate the ultra-short-term photovoltaic power generation based on the expected irradiance and the conversion efficiency of the photovoltaic panels in the photovoltaic module; the calculation formula of the ultra-short-term photovoltaic power generation is as follows: P=η×E t+Δt ×A; Where: P is the ultra-short-term photovoltaic power generation power, η is the conversion efficiency of the photovoltaic panel, and A is the area of ​​the photovoltaic panel.

7. The method for predicting power generation of a photovoltaic power station according to claim 1, characterized in that: The step S3 specifically includes: S31. Collect historical photovoltaic power generation data and weather data of photovoltaic power stations and build a neural network model; S32. Use historical photovoltaic power generation data, weather data, time characteristics, and recent power generation data as input features of the neural network model to make predictions and output short-term photovoltaic power generation in the future time period.

8. The method for predicting power generation of a photovoltaic power station according to claim 1, characterized in that: In step S4, the expression of the weight coefficient is as follows: Among them: A ultra_short is the accuracy index of ultra-short-term forecast, T factor is the time range factor, W factor is the uncertainty factor of weather change, A short It is an indicator of the accuracy of short-term forecasts; T factor =e -λt ; Where: λ is the decay rate, and t is the prediction time range.

9. The method for predicting power generation of a photovoltaic power station according to claim 1, characterized in that: The ultra short term in the ultra short term photovoltaic power generation power refers to a time range within 4 hours, and the short term in the short term photovoltaic power generation power refers to a time range within 48 hours.

10. A power generation prediction system for a photovoltaic power station, characterized in that: The system is applied to the method described in any one of claims 1 to 9, and the system comprises: The correction coefficient calculation module (1) is used to calculate the comprehensive efficiency correction coefficient of the photovoltaic power station; the calculation formula of the comprehensive efficiency correction coefficient is as follows: K=K1×K2×K3×K4×...×Kn; Among them: K is the comprehensive efficiency correction coefficient of the photovoltaic power station, K1 is the photovoltaic module type correction coefficient, K2 is the inclination and azimuth correction coefficient of the photovoltaic array, K3 is the inverter efficiency correction coefficient, K4 is the line loss correction coefficient, and Kn is the correction coefficient of other influencing factors; The ultra-short-term power prediction module (2) is used to construct a cloud motion model based on the environmental and climatic factors of the photovoltaic power station area using a machine learning method, and then simulate cloud motion through the cloud motion model to evaluate the ground radiation intensity and predict the ultra-short-term photovoltaic power generation power; A short-term power prediction module (3) is used to construct a short-term photovoltaic power prediction model based on the historical power generation data and weather information of the photovoltaic power station by using a machine learning method, and then predict the short-term photovoltaic power generation power through the short-term photovoltaic power prediction model; The photovoltaic power generation module (4) is used to predict the photovoltaic power generation power of the photovoltaic power station in a certain future cycle based on the comprehensive efficiency correction coefficient of the photovoltaic power station, the ultra-short-term photovoltaic power generation power, and the short-term period photovoltaic power generation power; the expression for predicting the photovoltaic power generation power in a certain future cycle is as follows: P_cycle1=K×(alpha×P_ultra_short+(1-alpha)×P_short); Among them: P_cycle1 is the photovoltaic power generation power in a certain cycle in the future, K is the comprehensive efficiency correction coefficient of the photovoltaic power station, P_ultra_short is the ultra-short-term photovoltaic power generation power, P_short is the short-term cycle photovoltaic power generation power, and alpha is the weight coefficient.

Citation Information

Patent Citations

  • Photovoltaic array power generation capacity deviation anomaly analysis method

    CN116131460A

  • Ultra-short-term prediction method based on image recognition technology

    CN117313939A

  • Short-time photovoltaic output prediction method based on separable convolution and deep gate structure neural network

    CN118213980A

  • Photovoltaic power ultra-short-term multi-step prediction method and device based on time sequence neural hierarchical interpolation model, and medium

    CN118228876A

  • Hybrid photovoltaic power prediction method and system based on multi-source data fusion

    US20220373984A1

Cited By

  • Photovoltaic electric quantity and power prediction method under different time scales

    CN119294618A

  • Photovoltaic panel cleanliness evaluation method based on twin system output

    CN119377554A

  • Composite photovoltaic power generation prediction method based on visual identification

    CN119419803A

  • Low-voltage distributed photovoltaic output prediction and group scheduling and group control method based on big data model

    CN119944651A

  • Low-voltage distributed photovoltaic power output prediction and group control method based on big data model

    CN119944651B