Photoelectric output sequence prediction method and device, terminal equipment and storage medium

Abstract

The invention discloses a photoelectric output sequence prediction method and device, terminal equipment and a storage medium, and the method comprises the steps: dividing each historical actual power into a historical clearance theoretical output, a historical relative output, and a relative output curve of each historical date; a relative output curve is classified into an output curve data set under a plurality of weather types for fitting analysis and prediction, and reference value probability distribution, offset value probability distribution and fluctuation value probability distribution of photovoltaic equipment under various weather types and a weather prediction sequence of a to-be-predicted date are generated. And predicting a predicted relative output curve of each date to be predicted, and combining the predicted relative output curve with clearance theoretical output to generate a photoelectric output prediction sequence corresponding to each date to be predicted. Therefore, fitting analysis is carried out on the output curve data set of each weather type to explore the photovoltaic output regular change and weather fluctuation characteristics, and then the short-term or medium-and-long-term photovoltaic output sequence is accurately predicted.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power prediction, and particularly to a prediction method, device, terminal device and storage medium for photovoltaic output sequences. Background Art

[0002] Energy is the foundation for the sustainable development of the economy and society, and is an indispensable power guarantee for human production and life. As an important member of renewable energy, in recent years, the development of photovoltaic power generation in China has been rapid, effectively alleviating the energy crisis and environmental pressure. However, the output power of a photovoltaic power generation system is easily affected by environmental factors such as irradiation and temperature, and shows obvious output randomness and uncertainty compared with traditional power sources, which has a greater impact on the peak regulation, frequency modulation, standby, power flow and bus voltage of the power grid. Simulating the photovoltaic output curve for simulation is of great significance for the optimal design of photovoltaic power stations, the configuration planning of power sources and power grids, and the formulation of new energy policies.

[0003] Traditional photovoltaic output curve modeling methods can be divided into two major categories according to the modeling object: The first is the modeling method based on solar irradiance intensity, also known as the indirect method. This method can determine the radiation intensity of the sun on the area where the photovoltaic device is located at a future moment through astronomical algorithms, and then predict the photovoltaic output. Therefore, this method can clearly reflect the physical meaning and better reflect the regular changes. However, the actual output of the photovoltaic device is also affected by other weather factors, and there are often large errors in presetting the photovoltaic output only based on solar radiation, and it is usually difficult to be put into actual use. The second is the modeling method based on power, also known as the direct method, which uses algorithms to find the mathematical relationship between the historical measured photovoltaic output data, establish a mathematical model, and directly simulate a new photovoltaic output sequence. The algorithms mainly used in the direct method include neural networks and machine learning. However, the direct method using neural networks and machine learning can only be used for the prediction modeling of short-term photovoltaic output because it is difficult to simulate the regular changes of photovoltaic output and the characteristics of weather fluctuations, and cannot predict the medium- and long-term photovoltaic output. Summary of the Invention

[0004] Embodiments of the present invention provide a prediction method, device, terminal device and storage medium for photovoltaic output sequences, which can accurately predict short-term or medium- and long-term photovoltaic output sequences by exploring the regular changes of photovoltaic output and the characteristics of weather fluctuations.

[0005] An embodiment of the present invention provides a prediction method for a photovoltaic output sequence, including:

[0006] Obtain the historical actual power, historical radiation intensity of each moment of a photovoltaic device on a number of historical dates, and the future radiation intensity corresponding to each moment of the photovoltaic device on a number of dates to be predicted;

[0007] According to the historical actual power and the historical radiation intensity, split each of the historical actual powers into a historical clear sky theoretical output and a historical relative output;

[0008] According to the historical relative output, generate a relative output curve for each of the historical dates, and cluster the relative output curves, classifying each of the relative output curves into the output curve datasets corresponding to a preset number of weather types;

[0009] Respectively perform fitting analysis and prediction on the relative output curves and the corresponding historical dates in several of the output curve datasets to generate the probability distributions of the benchmark values, offset values, and fluctuation values of several relative output curves of the photovoltaic device under various weather types, and the weather prediction sequences for several dates to be predicted;

[0010] According to the weather prediction sequences, the probability distributions of the benchmark values, the probability distributions of the offset values, and the probability distributions of the fluctuation values, generate a predicted relative output curve for each of the dates to be predicted;

[0011] According to the future radiation intensity, determine the clear sky theoretical output corresponding to each moment under each of the dates to be predicted, and according to the clear sky theoretical output and the predicted relative output curve, generate the photovoltaic output prediction sequence corresponding to each of the dates to be predicted.

[0012] Further, the obtaining the historical radiation intensity of the photovoltaic device at each moment on several historical dates and the future radiation intensity corresponding to each moment of the photovoltaic device on several dates to be predicted includes:

[0013] Obtain the geographical location where the photovoltaic device is located;

[0014] According to the geographical location, calculate the solar incident angle of the photovoltaic device at each target moment; wherein, the target moment includes: each moment on several of the historical dates and each moment on several of the dates to be predicted;

[0015] According to the solar incident angle, calculate the solar instantaneous direct radiation and the solar diffuse radiation intensity corresponding to the geographical location at each target moment;

[0016] According to the solar instantaneous direct radiation and the solar diffuse radiation intensity, calculate the solar radiation intensity at each target moment; wherein, the solar radiation intensity includes: historical radiation intensity and future radiation intensity.

[0017] Further, the splitting each of the historical actual powers into a historical clear sky theoretical output and a historical relative output according to the historical actual power and the historical radiation intensity includes:

[0018] Obtain the photoelectric conversion coefficient of the photovoltaic device;

[0019] Calculate the historical clear-sky theoretical output according to the historical radiation intensity and the photoelectric conversion coefficient;

[0020] According to the historical clear-sky theoretical output, split each historical actual power into the historical clear-sky theoretical output and the historical relative output through the following formula:

[0021] P(i,t) = P DCI (i,t) * P N (i,t);

[0022] where P(i,t) is the historical actual power at the t-th moment in the i-th historical date, P DCI (i,t) is the historical clear-sky theoretical output at the t-th moment in the i-th historical date, and P N (i,t) is the historical relative output at the t-th moment in the i-th historical date.

[0023] Further, the clustering of the relative output curves and the classification of each relative output curve into the output curve datasets corresponding to a preset number of weather types include:

[0024] Generate the feature vectors corresponding to each relative output curve according to each relative output curve;

[0025] Normalize the feature vectors to generate data points for clustering;

[0026] Calculate the similarity between each data point and the preset neurons, and iteratively adjust the positions and matching radii of each neuron according to the similarity; where each neuron corresponds to a weather type;

[0027] When it is determined that the positions of each neuron are stable, determine the target data points within the matching radii of each neuron;

[0028] Classify the relative output curves corresponding to each target data point into the output curve datasets of the weather types corresponding to each neuron.

[0029] Further, the generation of the feature vectors corresponding to each relative output curve according to each relative output curve includes:

[0030] Traverse each relative output curve;

[0031] Calculate the reference value of the currently traversed target relative output curve through the following formula:

[0032]

[0033] where d 1 is the reference value of the target relative output curve, P N (i) is the i-th historical relative output of the target relative output curve, and n is the total number of historical relative outputs included in the target relative output curve;

[0034] Calculate the standard deviation of the currently traversed target relative output curve through the following formula:

[0035]

[0036] where d 2 is the standard deviation of the target relative output curve;

[0037] Calculate the mean absolute value of the first-order difference of the currently traversed target relative output curve through the following formula:

[0038]

[0039] where d 3 is the mean absolute value of the first-order difference of the target relative output curve;

[0040] Calculate the maximum value of the absolute value of the first-order difference of the currently traversed target relative output curve through the following formula:

[0041] d 4 = max|P N (i + 1)-P N (i)|;

[0042] where d 4 is the maximum value of the absolute value of the first-order difference of the target relative output curve;

[0043] At the end of the traversal, according to the reference value, standard deviation, mean absolute value of the first-order difference, and maximum value of the absolute value of the first-order difference of each relative output curve, form the feature vector of each relative output curve.

[0044] Furthermore, respectively perform fitting analysis and prediction on the relative output curves and corresponding historical dates in several output curve datasets to generate the probability distributions of the reference values, offset values, and fluctuation values of several relative output curves of the photovoltaic device under various weather types, and the weather prediction sequences of several dates to be predicted, including:

[0045] Adopt the kernel density estimation method to fit the probability distributions of the reference values, offset values, and fluctuation values of several relative output curves of the photovoltaic device under various weather types;

[0046] According to the historical dates corresponding to the respective relative output curves in each output curve dataset, count the occurrence times and transition probabilities of each weather type in each season;

[0047] According to each of the to-be-predicted dates, the occurrence times, and the transition probabilities, generate the predicted weather type for each to-be-predicted date, and generate the weather prediction sequence according to the predicted weather type.

[0048] Further, the step of generating the predicted weather type for each to-be-predicted date according to each of the to-be-predicted dates, the occurrence times, and the transition probabilities, and generating the weather prediction sequence according to the predicted weather type includes:

[0049] Generate the state transition matrix and the cumulative state transition matrix corresponding to each season according to the occurrence times and transition probabilities of each weather type in each season; wherein, the state transition matrix is used to represent the conditional probability of various weather types occurring on the next day affected by the weather type on the previous day in two adjacent days, and the cumulative state transition matrix is used to represent the cumulative conditional probability of various weather types occurring on the next day affected by the weather type on the previous day in two adjacent days;

[0050] Randomly generate the predicted weather type for each to-be-predicted date according to each of the to-be-predicted dates, the state transition matrix, and the cumulative state transition matrix;

[0051] Generate the weather prediction sequence according to the predicted weather type.

[0052] Further, the step of generating the predicted relative output curve for each to-be-predicted date according to the weather prediction sequence, the reference value probability distribution, the offset value probability distribution, and the fluctuation value probability distribution includes:

[0053] Traverse the to-be-predicted dates, and determine the target predicted weather type from the weather prediction sequence according to the currently traversed target predicted date;

[0054] Sample the relative output reference value and the relative output offset value at each moment in the target predicted date according to the target reference value probability distribution, the target offset value probability distribution, and the target fluctuation value probability distribution corresponding to the target predicted weather type;

[0055] At the end of the traversal, generate the predicted relative output curve for each to-be-predicted date according to the relative output reference value and the relative output offset value at each moment in each to-be-predicted date.

[0056] Further, sampling the relative output reference value and the relative output offset value at each moment in the target prediction date according to the target reference value probability distribution, the target offset value probability distribution, and the target fluctuation value probability distribution corresponding to the target predicted weather type includes:

[0057] Randomly sampling the relative output reference value at each moment in the date to be predicted according to the target reference value probability distribution;

[0058] Traverse each moment in the target prediction date, and sample the initial relative output offset value of the current target traversed moment according to the target offset value probability distribution;

[0059] According to the initial relative output offset value, repeatedly perform the sampling verification operation until the target relative output offset value of the target traversed moment is generated;

[0060] When it is determined that the target relative output offset value is generated, traverse the next moment;

[0061] When the traversal is completed, generate the relative output offset value at each moment in the target prediction date;

[0062] Among them, the sampling verification operation includes:

[0063] Obtain the offset value to be verified; initially, the offset value to be verified is the initial relative output offset value;

[0064] Perform fluctuation value verification on the offset value to be verified by using the target fluctuation value probability distribution;

[0065] When the offset value to be verified passes the fluctuation value verification, use the offset value to be verified as the target relative output offset value of the target traversed moment;

[0066] When the offset value to be verified does not pass the fluctuation value verification, sample a new relative output offset value according to the target offset value probability distribution as the offset value to be verified in the next round of sampling verification operation.

[0067] Further, the performing fluctuation value verification on the offset value to be verified by using the target fluctuation value probability distribution includes:

[0068] Obtain the relative output offset value of the previous moment of the target traversed moment as the verification value;

[0069] Subtract the verification value from the offset value to be verified to obtain the fluctuation value;

[0070] Judge whether the fluctuation value meets the target fluctuation value probability distribution. If so, determine that the offset value to be verified passes the fluctuation value verification;

[0071] Otherwise, it is determined that the to-be-verified offset value fails the fluctuation value verification.

[0072] Further, the prediction method for an optoelectronic output sequence further includes:

[0073] Obtain the historical sunrise time, historical sunrise duration, historical sunset time, and historical sunset duration of a plurality of historical dates;

[0074] According to the historical sunrise time, historical sunrise duration, historical sunset time, and historical sunset duration, determine from the historical relative output the historical sunrise relative output corresponding to the historical sunrise time, the historical sunrise fluctuation output at several moments in the historical sunrise duration, the historical sunset relative output corresponding to the historical sunset time, and the historical sunset fluctuation output at several moments in the historical sunset duration;

[0075] Generate a sunrise reference output distribution according to the ratios of the historical sunrise fluctuation output to the respective historical sunrise relative outputs;

[0076] Generate a sunset reference output distribution according to the ratios of the historical sunset fluctuation output to the respective historical sunset relative outputs.

[0077] Further, the generating a predicted relative output curve for each to-be-predicted date according to the relative output reference value and relative output offset value at each moment in each to-be-predicted date includes:

[0078] Obtain the sunrise time, sunrise duration, sunset time, and sunset duration of each to-be-predicted date;

[0079] According to the sunrise duration and sunset duration, determine several correction moments and several normal moments that need to be corrected in each to-be-predicted date; wherein, the correction moments include: the sunrise time, the sunset time, several sunrise correction moments in the sunrise duration, and several sunset correction moments in the sunrise duration;

[0080] Calculate the initial predicted relative output at each correction moment and the first predicted relative output at each normal moment in the to-be-predicted date according to the relative output reference value and relative output offset value of the to-be-predicted date;

[0081] Correct the initial predicted relative output at the sunrise time and each sunrise correction moment according to the sunrise reference output to generate the second predicted relative output at the sunrise time and each sunrise correction moment;

[0082] Based on the output reference of sunset, correct the initial predicted relative output at the sunset time and each of the sunset correction times to generate the third predicted relative output at the sunset time and each of the sunset correction times;

[0083] Based on the first predicted relative output, the second predicted relative output, and the third predicted relative output of each date to be predicted, generate a predicted relative output curve for each date to be predicted.

[0084] Another embodiment of the present invention provides a prediction device for an optoelectronic output sequence, including:

[0085] A data acquisition module, configured to acquire the historical actual power, historical radiation intensity of a photovoltaic device at each moment on a plurality of historical dates, and the future radiation intensity corresponding to each moment of the photovoltaic device on a plurality of dates to be predicted;

[0086] A power splitting module, configured to split each of the historical actual powers into a historical clear sky theoretical output and a historical relative output according to the historical actual power and the historical radiation intensity;

[0087] A data clustering module, configured to generate a relative output curve for each of the historical dates according to the historical relative output, cluster the relative output curves, and classify the relative output curves into output curve datasets corresponding to a preset number of weather types;

[0088] A weather prediction module, configured to perform fitting analysis and prediction on the relative output curves and corresponding historical dates in a plurality of the output curve datasets respectively, to generate a probability distribution of the reference value, a probability distribution of the offset value, a probability distribution of the fluctuation value of a plurality of relative output curves of the photovoltaic device under various weather types, and a weather prediction sequence of a plurality of dates to be predicted;

[0089] A relative output prediction module, configured to generate a predicted relative output curve for each of the dates to be predicted according to the weather prediction sequence, the probability distribution of the reference value, the probability distribution of the offset value, and the probability distribution of the fluctuation value;

[0090] An output prediction module, configured to determine the clear sky theoretical output corresponding to each moment of each date to be predicted according to the future radiation intensity, and generate an optoelectronic output prediction sequence corresponding to each date to be predicted according to the clear sky theoretical output and the predicted relative output curve.

[0091] Another embodiment of the present invention provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements a prediction method for an optoelectronic output sequence as described in any one of the above embodiments.

[0092] Another embodiment of the present invention provides a storage medium, which includes a stored computer program. When the computer program runs, it controls the device where the storage medium is located to execute a prediction method for a photovoltaic power output sequence as described in any one of the above embodiments.

[0093] By implementing the present invention, the following beneficial effects are achieved:

[0094] The present invention discloses a prediction method, device, terminal device and storage medium for a photovoltaic power output sequence. By according to the radiation intensity of the sun on each region of the earth at each moment, the net space theoretical output of the photovoltaic device without other factors is calculated. Then, each historical actual power is split into historical net space theoretical output and historical relative output, and the relative output curve of each historical date. By clustering the relative output curves, the relative output curves are classified into the output curve datasets under several weather types for fitting analysis and prediction, generating the benchmark value probability distribution, offset value probability distribution, fluctuation value probability distribution of several relative output curves of the photovoltaic device under various weather types, and the weather prediction sequences of several dates to be predicted. Then, the predicted relative output curve of each date to be predicted is predicted, and combined with the future radiation intensity for calculation, generating the photovoltaic power output prediction sequence corresponding to each date to be predicted. Therefore, when predicting the photovoltaic output, the present invention simultaneously considers the influence of solar radiation intensity and weather factors on the output of the photovoltaic device, and explores the regular changes and fluctuation characteristics of the photovoltaic output by fitting analysis in the output curve datasets of each weather type, and can accurately predict the photovoltaic power output sequence in the short term or medium to long term. Description of the Drawings

[0095] Figure 1 It is a flowchart of a prediction method for a photovoltaic power output sequence provided by an embodiment of the present invention.

[0096] Figure 2 It is a structural diagram of a prediction device for a photovoltaic power output sequence provided by an embodiment of the present invention. Detailed Embodiments

[0097] To make the objectives, technical solutions and advantages of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments in the present application belong to the scope of protection of the present application.

[0098] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field to which this application belongs; the terms used herein are for the purpose of describing specific embodiments only and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above drawings are intended to cover non-exclusive inclusion.

[0099] In the description of the embodiments of this application, technical terms such as "first" and "second" are only used to distinguish different objects and cannot be understood as indicating or implying relative importance or implicitly indicating the quantity, specific order or primary-secondary relationship of the indicated technical features. In the description of the embodiments of this application, the meaning of "a plurality" is more than two, unless otherwise specifically defined.

[0100] Referring to "embodiment" herein means that the specific features, structures or characteristics described in connection with the embodiment can be included in at least one embodiment of this application. The phrase appearing at various positions in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0101] In the description of the embodiments of this application, the term "and / or" is merely a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " herein generally represents an "or" relationship between the associated objects before and after.

[0102] In the description of the embodiments of this application, the term "a plurality" refers to more than two (including two). Similarly, "multiple groups" refers to more than two groups (including two groups), and "multiple pieces" refers to more than two pieces (including two pieces).

[0103] In the description of the embodiments of this application, unless otherwise clearly specified and limited, technical terms such as "installation", "connection", "connection", "fixation" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or integrated; it can also be a mechanical connection or an electrical connection; it can be directly connected, or indirectly connected through an intermediate medium, and can be the communication inside two components or the interaction relationship between two components. For those of ordinary skill in the art, the specific meanings of the above terms in the embodiments of this application can be understood according to specific situations.

[0104] See Figure 1 , which is a schematic flowchart of a method for predicting an optoelectronic output sequence provided by an embodiment of the present invention, including:

[0105] S1. Obtain the historical actual power, historical radiation intensity of the photovoltaic device at each moment on a number of historical dates, and the future radiation intensity corresponding to each moment of the photovoltaic device on a number of dates to be predicted;

[0106] Preferably, the obtaining of the historical radiation intensity of the photovoltaic device at each moment on a number of historical dates and the future radiation intensity corresponding to each moment of the photovoltaic device on a number of dates to be predicted includes:

[0107] S11. Obtain the geographical location where the photovoltaic device is located;

[0108] S12. Calculate the solar incident angle of the photovoltaic device at each target moment according to the geographical location; wherein, the target moments include each moment on a number of the historical dates and each moment on a number of the dates to be predicted;

[0109] In a preferred embodiment of the present invention, first, according to the following formula, calculate the radiation intensity accurately reaching the earth's atmosphere directly by the sun at each moment according to astronomical formulas:

[0110]

[0111] In the formula: I 0 is the solar radiation intensity perpendicular to the atmosphere, S 0 is the solar constant. N is the date serial number in a year, starting from January 1st.

[0112] Secondly, calculate the solar incident angle θ of the photovoltaic device at each target moment according to the following formula i :

[0113] cosθ i =

[0114] cosβsinα + sinβsinγcosδsinω + sinβcosγcosδ(sinφcosω - sinδ);

[0115] In the formula: α represents the solar altitude angle, ω represents the solar hour angle, φ represents the local latitude, δ represents the declination angle, β is the tilt angle of the photovoltaic array panel, and γ is the azimuth angle of the photovoltaic array panel.

[0116] S13. Calculate the solar instantaneous direct radiation and solar diffuse radiation intensity corresponding to the geographical location at each target moment according to the solar incident angle;

[0117] S14. Calculate the solar radiation intensity at each target moment according to the solar instantaneous direct radiation and the solar diffuse radiation intensity; wherein, the solar radiation intensity includes historical radiation intensity and future radiation intensity.

[0118] In a preferred embodiment of the present invention, the instantaneous direct solar radiation I is calculated respectively through the following formula b and the solar diffuse radiation intensity I d :

[0119] I b =I 0 τ b cosθ i ;

[0120]

[0121] Wherein: τ b and τ d are respectively the atmospheric transparency coefficients of direct solar radiation and diffuse radiation. k is a parameter related to the air mass. When the air mass is relatively turbid, k takes a value of 0.6 to 0.7; when the air mass is normal, k takes a value of 0.7 to 0.8; when the air mass is relatively good, k takes a value of 0.8 to 0.9.

[0122] The total solar radiation intensity I at each target time in the area where the photovoltaic device is located is obtained t as:

[0123] I t =I b +I d ;

[0124] S2. According to the historical actual power and the historical radiation intensity, each historical actual power is split into historical clear-sky theoretical output and historical relative output;

[0125] Preferably, the splitting of each historical actual power into historical clear-sky theoretical output and historical relative output according to the historical actual power and the historical radiation intensity includes:

[0126] S21. Obtain the photoelectric conversion coefficient of the photovoltaic device;

[0127] S22. Calculate the historical clear-sky theoretical output according to the historical radiation intensity and the photoelectric conversion coefficient;

[0128] S23. According to the historical clear-sky theoretical output, each historical actual power is split into historical clear-sky theoretical output and historical relative output through the following formula:

[0129] P(i,t)=P DCI (i,t)*P N (i,t);

[0130] Among them, P(i,t) is the historical actual power at the t-th moment on the i-th historical date, and P DCI (i,t) is the historical theoretical net output at the t-th moment on the i-th historical date, and P N (i,t) is the historical relative output at the t-th moment on the i-th historical date.

[0131] S3. Generate a relative output curve for each historical date according to the historical relative output, and cluster the relative output curves, and classify each relative output curve into the output curve datasets corresponding to a preset number of weather types;

[0132] Preferably, the clustering of the relative output curves and the classification of each relative output curve into the output curve datasets corresponding to a preset number of weather types include:

[0133] S31. Generate a feature vector corresponding to each relative output curve according to each relative output curve;

[0134] Preferably, the generating a feature vector corresponding to each relative output curve according to each relative output curve includes:

[0135] S311. Traverse each relative output curve;

[0136] S312. Calculate the reference value of the target relative output curve currently traversed through the following formula:

[0137]

[0138] where d 1 is the reference value of the target relative output curve, P N (i) is the i-th historical relative output of the target relative output curve, and n is the total number of historical relative outputs included in the target relative output curve;

[0139] S313. Calculate the standard deviation of the target relative output curve currently traversed through the following formula:

[0140]

[0141] where d 2 is the standard deviation of the target relative output curve;

[0142] S314. Calculate the average value of the absolute value of the first-order difference of the target relative output curve currently traversed through the following formula:

[0143]

[0144] where d 3is the mean value of the absolute value of the first-order difference of the target relative output curve;

[0145] S315. Calculate the maximum value of the absolute value of the first-order difference of the target relative output curve currently traversed through the following formula:

[0146] d 4 = max|P N (i + 1)-P N (i)|;

[0147] where d 4 is the maximum value of the absolute value of the first-order difference of the target relative output curve;

[0148] S316. At the end of the traversal, construct the feature vector of each relative output curve according to the reference value, standard deviation, mean value of the absolute value of the first-order difference, and maximum value of the absolute value of the first-order difference of each relative output curve.

[0149] S32. Normalize the feature vector to generate data points for clustering;

[0150] S33. Calculate the similarity between each data point and a preset neuron, and iteratively adjust the position and matching radius of each neuron according to the similarity; where each neuron corresponds to a weather type;

[0151] S34. When it is determined that the positions of each neuron are stable, determine the target data points within the matching radius of each neuron;

[0152] S35. Classify the relative output curves corresponding to each target data point into the output curve datasets of the weather types corresponding to each neuron.

[0153] In a preferred embodiment of the present invention, the SOM clustering method is adopted, and four characteristic quantities of the output data of one day are selected, namely the reference value, standard deviation, mean value of the absolute value of the first-order difference, and maximum value of the absolute value of the first-order difference, to form a feature vector. After normalization processing, weather clustering analysis is carried out, and the daily photovoltaic output curves are divided into different weather types. Among them, the reference value is used to reflect the all-day output level of the photovoltaic device, which is high on sunny days and low on rainy days; the standard deviation is used to reflect the all-day fluctuation level of the photovoltaic device, which is low on sunny and rainy days and high on cloudy days; the mean value of the absolute value of the first-order difference is used to reflect the temporal fluctuation level of the photovoltaic device output, which is high on cloudy days; the maximum value of the absolute value of the first-order difference is used to reflect the severity of the temporal fluctuation of the photovoltaic device output, such as high on cloudy and sudden change days.

[0154] S4. Respectively perform fitting analysis and prediction on the relative output curves and corresponding historical dates in several of the output curve datasets to generate the benchmark value probability distribution, offset value probability distribution, fluctuation value probability distribution of several relative output curves of the photovoltaic device under various weather types, and the weather prediction sequences for several dates to be predicted;

[0155] Preferably, the step of respectively performing fitting analysis and prediction on the relative output curves and corresponding historical dates in several of the output curve datasets to generate the benchmark value probability distribution, offset value probability distribution, fluctuation value probability distribution of several relative output curves of the photovoltaic device under various weather types, and the weather prediction sequences for several dates to be predicted includes:

[0156] S41. Use the kernel density estimation method to fit the benchmark value probability distribution, offset value probability distribution, and fluctuation value probability distribution of several relative output curves of the photovoltaic device under various weather types;

[0157] In a preferred embodiment of the present invention, the kernel density estimation method (Kernel Density Estimation, KDE) is used to fit the benchmark value probability distribution, offset value probability distribution, and fluctuation value probability distribution of the photovoltaic output every day under various weather types, and the relative output difference of each time period relative to the previous time period, that is, the first-order difference of the offset value, is statistically calculated and called the fluctuation value.

[0158] S42. According to the historical dates corresponding to the relative output curves in each output curve dataset, count the occurrence times and transition probabilities of each weather type in each season;

[0159] S43. According to each date to be predicted, the occurrence times, and the transition probabilities, generate the predicted weather type for each date to be predicted, and generate the weather prediction sequence according to the predicted weather type.

[0160] Preferably, the step of generating the predicted weather type for each date to be predicted according to each date to be predicted, the occurrence times, and the transition probabilities, and generating the weather prediction sequence according to the predicted weather type includes:

[0161] S431. According to the occurrence times and transition probabilities of each weather type in each season, generate the state transition matrix and cumulative state transition matrix corresponding to each season; wherein, the state transition matrix is used to represent the conditional probability of various weather types occurring on the next day affected by the weather type of the previous day in two adjacent days, and the cumulative state transition matrix is used to represent the cumulative conditional probability of various weather types occurring on the next day affected by the weather type of the previous day in two adjacent days;

[0162] S432. Randomly generate the predicted weather types for each predicted date according to each of the predicted dates, the state transition matrix, and the cumulative state transition matrix;

[0163] S433. Generate the weather prediction sequence according to the predicted weather types.

[0164] In a preferred embodiment of the present invention, the number of times and transition probabilities of each weather type are separately counted for the weather clustering results by season to generate the state transition matrix P z and the cumulative state transition matrix Q z :

[0165]

[0166]

[0167] Where: the element p ij =P(Z n+1 =j|Z n =i) represents the conditional probability that tomorrow is of type j when today is of type i. The state transition matrix P z and the cumulative state transition matrix Q z are established according to historical weather data. After that, the weather prediction sequence Z = {Z 1 , Z 2 , …, Z N} can be generated. The weather types of each day are randomly generated, and Z i is the predicted weather type for the i predicted dates.

[0168] S5. Generate the predicted relative output curve for each predicted date according to the weather prediction sequence, the reference value probability distribution, the offset value probability distribution, and the fluctuation value probability distribution;

[0169] Preferably, generating the predicted relative output curve for each predicted date according to the weather prediction sequence, the reference value probability distribution, the offset value probability distribution, and the fluctuation value probability distribution includes:

[0170] S51. Traverse the predicted dates, and determine the target predicted weather type from the weather prediction sequence according to the currently traversed target predicted date;

[0171] S52. Sample the relative output reference value and the relative output offset value at each moment in the target predicted date according to the target reference value probability distribution, the target offset value probability distribution, and the target fluctuation value probability distribution corresponding to the target predicted weather type;

[0172] Preferably, sampling the relative output reference value and the relative output offset value at each moment in the target prediction date according to the target reference value probability distribution, the target offset value probability distribution, and the target fluctuation value probability distribution corresponding to the target predicted weather type includes:

[0173] S521. Randomly sample the relative output reference value at each moment in the date to be predicted according to the target reference value probability distribution;

[0174] S522. Traverse each moment in the target prediction date, and sample the initial relative output offset value at the currently traversed target traversal moment according to the target offset value probability distribution;

[0175] S523. Repeat the sampling verification operation according to the initial relative output offset value until the target relative output offset value at the target traversal moment is generated;

[0176] S524. When it is determined that the target relative output offset value is generated, traverse the next moment;

[0177] S525. When the traversal is completed, generate the relative output offset value at each moment in the target prediction date;

[0178] Among them, the sampling verification operation includes:

[0179] S526. Obtain the offset value to be verified; initially, the offset value to be verified is the initial relative output offset value;

[0180] S527. Perform a fluctuation value verification on the offset value to be verified using the target fluctuation value probability distribution;

[0181] Preferably, performing a fluctuation value verification on the offset value to be verified using the target fluctuation value probability distribution includes:

[0182] S5271. Obtain the relative output offset value at the previous moment of the target traversal moment as the verification value;

[0183] S5272. Subtract the verification value from the offset value to be verified to obtain a fluctuation value;

[0184] S5273. Determine whether the fluctuation value satisfies the target fluctuation value probability distribution. If so, determine that the offset value to be verified passes the fluctuation value verification;

[0185] S5274. If not, determine that the offset value to be verified does not pass the fluctuation value verification.

[0186] S528. When the offset value to be verified passes the verification by the fluctuation value, use the offset value to be verified as the target relative output offset value at the target traversal time;

[0187] S528. When the offset value to be verified fails to pass the verification by the fluctuation value, sample a new relative output offset value according to the target offset value probability distribution and use it as the offset value to be verified for the next round of sampling verification operation.

[0188] S53. At the end of the traversal, generate a predicted relative output curve for each predicted date according to the relative output reference value and the relative output offset value at each moment in each predicted date.

[0189] In a preferred embodiment of the present invention, first, according to a randomly generated weather chain (weather prediction sequence), simply sample the reference value of each day. Furthermore, according to the randomly generated weather chain, simply sample the offset value of each day. Subtract the offset value at the current moment from the previous moment to obtain the fluctuation value z 0 . Denote the probability density distribution function of z 0 as f(z 0 ). Construct a new probability density function q(z 0 ), satisfying kq(z 0 ) > f(z 0 ), where k is a constant. Uniformly sample [0, kq(z 0 )] to obtain u 0 . If u 0 < f(z 0 ), then accept this sampling and enter the next moment; otherwise, reject this sampling and resample the offset value distribution until the sampling is accepted. If the sampling of each moment of the current day has been completed, calculate the photovoltaic relative output of each moment of the current day and enter the next day. Until the relative output reference value and the relative output offset value at each moment in each predicted date are generated.

[0190] S6. According to the future radiation intensity, determine the clear sky theoretical output corresponding to each moment under each predicted date, and generate an optoelectronic output prediction sequence corresponding to each predicted date according to the clear sky theoretical output and the predicted relative output curve.

[0191] Preferably, the method for predicting an optoelectronic output sequence described in the above embodiment further includes:

[0192] S601. Obtain the historical sunrise time, historical sunrise duration, historical sunset time, and historical sunset duration of several historical dates;

[0193] S602. Determine the historical relative output corresponding to the historical sunrise time, the historical sunrise fluctuation output at several moments within the historical sunrise duration, the historical relative output corresponding to the historical sunset time, and the historical sunset fluctuation output at several moments within the historical sunset duration from the historical relative output according to the historical sunrise time, historical sunrise duration, historical sunset time, and historical sunset duration;

[0194] S603. Generate a sunrise reference output distribution according to the ratios of the historical sunrise fluctuation output to the respective historical relative outputs;

[0195] S604. Generate a sunset reference output distribution according to the ratios of the historical sunset fluctuation output to the respective historical relative outputs.

[0196] In a preferred embodiment of the present invention, the historical relative output shows an increasing or decreasing trend within 1 hour after the start of the output and within 1 hour before the end of the output. Therefore, when predicting the relative output curve of the photovoltaic device, this period of time needs to be separately statistically corrected.

[0197] Specifically, find the starting time Trise, ending time Tset of the relative output curve and the corresponding relative output values P Nrise 、P Nset . Set the time resolution of the output data to 10 minutes. Specify the starting time and the subsequent 5 moments as the sunrise period, and the ending time and the subsequent 5 moments as the sunset period. Respectively calculate the ratios of the relative output at each moment in the sunrise period to P Nrise and the ratios of the relative output at each moment in the sunset period to P Nset as the reference output.

[0198] Preferably, the generating of the predicted relative output curve for each predicted date according to the relative output reference value and relative output offset value at each moment in each predicted date includes:

[0199] S611. Obtain the sunrise time, sunrise duration, sunset time, and sunset duration of each predicted date;

[0200] S612. Determine several correction moments and several normal moments that need to be corrected in each predicted date according to the sunrise duration and sunset duration; wherein, the correction moments include: the sunrise time, sunset time, several sunrise correction moments within the sunrise duration, and several sunset correction moments within the sunrise duration;

[0201] S613. Calculate the initial predicted relative output at each correction moment and the first predicted relative output at each normal moment in the to-be-predicted date according to the relative output reference value and the relative output offset value of the to-be-predicted date;

[0202] S614. Correct the initial predicted relative output at the sunrise moment and each of the sunrise correction moments according to the reference output at sunrise to generate the second predicted relative output at the sunrise moment and each of the sunrise correction moments;

[0203] S615. Correct the initial predicted relative output at the sunset moment and each of the sunset correction moments according to the reference output at sunset to generate the third predicted relative output at the sunset moment and each of the sunset correction moments;

[0204] S616. Generate a predicted relative output curve for each to-be-predicted date according to the first predicted relative output, the second predicted relative output, and the third predicted relative output of each to-be-predicted date.

[0205] In a preferred embodiment of the present invention, the probability distribution of the reference output at each moment is fitted and sampled by season, and then the reference output obtained by sampling each moment is restored to the relative output. Finally, the simulated photovoltaic relative output is multiplied by the clear sky output to obtain the actual photovoltaic output.

[0206] This embodiment provides a method for predicting the photovoltaic output sequence. By calculating the clear sky theoretical output of the photovoltaic device without the influence of other factors according to the radiation intensity of the sun on each region of the earth at each moment, and then splitting each historical actual power into the historical clear sky theoretical output and the historical relative output, as well as the relative output curve of each historical date. By clustering the relative output curves, the relative output curves are classified into the output curve datasets under several weather types for fitting analysis and prediction, generating the probability distributions of the reference values, offset values, and fluctuation values of several relative output curves under various weather types of the photovoltaic device, as well as the weather prediction sequences of several to-be-predicted dates. Furthermore, the predicted relative output curve of each to-be-predicted date is predicted, and combined with the future radiation intensity for calculation, generating the photovoltaic output prediction sequence corresponding to each to-be-predicted date. Therefore, when predicting the photovoltaic output, the present invention simultaneously considers the influence of the solar radiation intensity and weather factors on the output of the photovoltaic device, and explores the regular changes and fluctuation characteristics of the photovoltaic output through fitting analysis in the output curve datasets of each weather type, and thus can accurately predict the short-term or medium- and long-term photovoltaic output sequence.

[0207] See Figure 2 , which is a schematic structural diagram of a photovoltaic output sequence prediction device provided by an embodiment of the present invention, including:

[0208] A data acquisition module, configured to acquire the historical actual power, historical radiation intensity of a photovoltaic device at each moment on a plurality of historical dates, and the future radiation intensity corresponding to each moment of the photovoltaic device on a plurality of dates to be predicted;

[0209] A power splitting module, configured to split each of the historical actual powers into a historical net theoretical output and a historical relative output according to the historical actual power and the historical radiation intensity;

[0210] A data clustering module, configured to generate a relative output curve for each of the historical dates according to the historical relative outputs, and cluster the relative output curves, and classify each of the relative output curves into an output curve dataset corresponding to a plurality of preset weather types;

[0211] A weather prediction module, configured to perform fitting analysis and prediction on the relative output curves and the corresponding historical dates in a plurality of the output curve datasets respectively, to generate a reference value probability distribution, an offset value probability distribution, a fluctuation value probability distribution of a plurality of relative output curves of the photovoltaic device under various weather types, and a weather prediction sequence for a plurality of dates to be predicted;

[0212] A relative output prediction module, configured to generate a predicted relative output curve for each of the dates to be predicted according to the weather prediction sequence, the reference value probability distribution, the offset value probability distribution, and the fluctuation value probability distribution;

[0213] An output prediction module, configured to determine the net theoretical output corresponding to each moment on each of the dates to be predicted according to the future radiation intensity, and generate a photovoltaic power output prediction sequence corresponding to each of the dates to be predicted according to the net theoretical output and the predicted relative output curve.

[0214] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in the attached drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that they have a communication connection, which can be specifically implemented as one or more communication buses or signal lines. Those of ordinary skill in the art can understand and implement without creative work.

[0215] Those skilled in the art can clearly understand that for the sake of convenience and conciseness, the specific working process of the device described above can refer to the corresponding process in the foregoing method embodiments, and will not be elaborated herein.

[0216] Another preferred embodiment of the present invention provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements a prediction method for an optoelectronic output sequence as described in any one of the above embodiments.

[0217] The terminal device may be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The terminal device may include, but is not limited to, a processor and a memory.

[0218] The so-called processor may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor is the control center of the terminal device, and connects various parts of the entire terminal device through various interfaces and lines.

[0219] The memory may be used to store the computer program. The processor realizes various functions of the terminal device by running or executing the computer program stored in the memory and calling the data stored in the memory. The memory may mainly include a program storage area and a data storage area. Among them, the program storage area may store an operating system, application programs required for at least one function, etc.; the data storage area may store data created according to the use of the mobile phone, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one magnetic disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0220] Another preferred embodiment of the present invention provides a storage medium, which is a computer-readable storage medium, and the computer program is stored in the computer-readable storage medium. When the computer program is executed by a processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0221] The above is the preferred implementation manner of the present invention. It should be pointed out that for those of ordinary skill in the art of the present technology, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. A method for predicting a photovoltaic output sequence, characterized in that: include: Obtain the historical actual power and historical radiation intensity of the photovoltaic equipment at each moment on several historical dates, and the future radiation intensity of the photovoltaic equipment at each moment on several dates to be predicted; According to the historical actual power and the historical radiation intensity, each of the historical actual powers is split into a historical net air theoretical output and a historical relative output; According to the historical relative output, a relative output curve for each of the historical dates is generated, and the relative output curves are clustered to classify each of the relative output curves into an output curve data set corresponding to a plurality of preset weather types; Perform fitting analysis and prediction on the relative output curves in the output curve data sets and the corresponding historical dates respectively, and generate the baseline value probability distribution, the offset value probability distribution, the fluctuation value probability distribution of the relative output curves of the photovoltaic equipment under various weather types, and the weather forecast sequence of several dates to be predicted; Generate a forecast relative output curve for each of the dates to be forecasted according to the weather forecast sequence, the baseline value probability distribution, the offset value probability distribution, and the fluctuation value probability distribution; According to the future radiation intensity, the net air theoretical output corresponding to each moment of each of the dates to be predicted is determined, and according to the net air theoretical output and the predicted relative output curve, a photovoltaic output prediction sequence corresponding to each of the dates to be predicted is generated.

2. A method for predicting a photovoltaic output sequence according to claim 1, characterized in that: The obtaining of the historical radiation intensity of the photovoltaic device at each moment on a number of historical dates and the corresponding future radiation intensity of the photovoltaic device at each moment on a number of dates to be predicted includes: Obtain the geographical location of the photovoltaic equipment; According to the geographical location, calculating the solar incidence angle of the photovoltaic device at each target time; wherein the target time includes: each time at a number of the historical dates, and each time at a number of the dates to be predicted; Calculate the instantaneous direct solar radiation and the intensity of the scattered solar radiation corresponding to the geographical location at each target time according to the solar incidence angle; The solar radiation intensity at each target time is calculated based on the instantaneous direct solar radiation and the solar scattered radiation intensity; wherein the solar radiation intensity includes: historical radiation intensity and future radiation intensity.

3. A method for predicting a photovoltaic output sequence according to claim 2, characterized in that: The splitting of each of the historical actual powers into a historical net air theoretical output and a historical relative output according to the historical actual power and the historical radiation intensity includes: Obtaining a photoelectric conversion coefficient of the photovoltaic device; Calculating the historical clearance theoretical output according to the historical radiation intensity and the photoelectric conversion coefficient; According to the historical net-space theoretical output, each of the historical actual powers is split into the historical net-space theoretical output and the historical relative output by the following formula: P(i,t)=P DCI (i,t)*P N (i,t); Where P(i,t) is the historical actual power at the tth moment in the i-th historical date, P DCI (i, t) is the historical clearance theoretical output at the t-th moment in the i-th historical date, P N (i,t) is the historical relative output at the tth moment in the i-th historical date.

4. A method for predicting a photovoltaic output sequence as claimed in claim 3, characterized in that: The clustering of the relative output curves to classify each relative output curve into output curve data sets corresponding to a plurality of preset weather types includes: According to each of the relative output curves, generating a characteristic vector corresponding to each of the relative output curves; Normalizing the feature vector to generate data points for clustering; Calculating the similarity between each of the data points and a preset neuron, and iteratively adjusting the position and matching radius of each of the neurons according to the similarity; wherein each of the neurons corresponds to a weather type; When determining that the position of each of the neurons is stable, determining a target data point located within a matching radius of each of the neurons; The relative output curve corresponding to each target data point is classified into the output curve data set of the weather type corresponding to each neuron.

5. A method for predicting a photovoltaic output sequence as claimed in claim 4, characterized in that: The step of generating a characteristic vector corresponding to each relative output curve according to each relative output curve comprises: Traversing each of the relative output curves; The reference value of the target relative output curve currently traversed is calculated by the following formula: Wherein, d1 is the reference value of the target relative output curve, P N (i) is the i-th historical relative output of the target relative output curve, and n is the total number of historical relative outputs included in the target relative output curve; The standard deviation of the target relative output curve currently traversed is calculated using the following formula: Wherein, d2 is the standard deviation of the target relative output curve; The first-order difference absolute value mean of the target relative output curve currently traversed is calculated by the following formula: Wherein, d3 is the absolute value mean of the first-order difference of the target relative output curve; The maximum absolute value of the first-order difference of the target relative output curve currently traversed is calculated using the following formula: d4=max|P N (i+1)-P N (i)|; Wherein, d4 is the maximum absolute value of the first-order difference of the target relative output curve; At the end of the traversal, a characteristic vector of each relative output curve is constructed according to the reference value, standard deviation, first-order difference absolute value mean, and first-order difference absolute value maximum of each relative output curve.

6. A method for predicting a photovoltaic output sequence according to claim 5, characterized in that: The fitting analysis and prediction are performed on the relative output curves in the output curve data sets and the corresponding historical dates respectively, and the baseline value probability distribution, the offset value probability distribution, the fluctuation value probability distribution, and the weather forecast sequence of the relative output curves of the photovoltaic equipment under various weather types are generated, including: Using the kernel density estimation method, the baseline value probability distribution, the offset value probability distribution, and the fluctuation value probability distribution of several relative output curves of the photovoltaic device under various weather types are fitted; According to the historical dates corresponding to the relative output curves in the output curve data sets, the occurrence times and transition probabilities of each weather type in each season are counted; According to each of the dates to be predicted, the number of occurrences, and the transition probability, a predicted weather type for each of the dates to be predicted is generated, and according to the predicted weather type, the weather forecast sequence is generated.

7. A method for predicting a photovoltaic output sequence according to claim 6, characterized in that: The step of generating a predicted weather type for each of the dates to be predicted according to the dates to be predicted, the number of occurrences, and the transition probability, and generating the weather forecast sequence according to the predicted weather type, comprises: According to the number of occurrences and transition probabilities of each weather type in each season, a state transfer matrix and a cumulative state transfer matrix corresponding to each season are generated; wherein the state transfer matrix is ​​used to characterize the conditional probability of each weather type appearing on the next day under the influence of the weather type on the previous day in two consecutive days, and the cumulative state transfer matrix is ​​used to characterize the cumulative conditional probability of each weather type appearing on the next day under the influence of the weather type on the previous day in two consecutive days; Randomly generate the predicted weather type for each date to be predicted according to each date to be predicted, the state transfer matrix, and the accumulated state transfer matrix; The weather forecast sequence is generated according to the predicted weather type.

8. A method for predicting a photovoltaic output sequence according to claim 7, characterized in that: The generating of the forecast relative output curve for each of the to-be-forecasted dates according to the weather forecast sequence, the reference value probability distribution, the offset value probability distribution, and the fluctuation value probability distribution includes: Traversing the to-be-predicted dates, and determining a target predicted weather type from the weather forecast sequence according to the currently traversed target predicted date; According to the target reference value probability distribution, target offset value probability distribution, and target fluctuation value probability distribution corresponding to the target forecast weather type, sampling the relative output reference value and relative output offset value at each moment in the target forecast date; At the end of the traversal, a predicted relative output curve of each of the dates to be predicted is generated according to the relative output reference value and the relative output offset value at each moment in each of the dates to be predicted.

9. A method for predicting a photovoltaic output sequence according to claim 8, characterized in that: The method of sampling the relative output reference value and the relative output offset value at each moment in the target forecast date according to the target reference value probability distribution, the target offset value probability distribution, and the target fluctuation value probability distribution corresponding to the target forecast weather type includes: According to the target benchmark value probability distribution, randomly sampling the relative output benchmark value at each moment in the to-be-predicted date; Traversing each moment in the target prediction date, sampling the initial relative output offset value of the currently traversed target traversal moment according to the target offset value probability distribution; According to the initial relative output offset value, repeatedly performing the sampling verification operation until the target relative output offset value at the target traversal moment is generated; When determining the relative output offset value of the generated target, traverse the next moment; When the traversal is completed, the relative output offset value at each moment in the target forecast date is generated; The sampling verification operation includes: Obtain the offset value to be verified; initially, the offset value to be verified is the initial relative output offset value; Using the target fluctuation value probability distribution to perform fluctuation value verification on the offset value to be verified; When the offset value to be verified passes the verification of the fluctuation value, the offset value to be verified is used as the target relative output offset value at the target traversal moment; When the offset value to be verified fails to pass the fluctuation value verification, a new relative output offset value is sampled according to the target offset value probability distribution as the offset value to be verified for the next round of sampling verification operation.

10. A method for predicting a photovoltaic output sequence according to claim 9, characterized in that: The adopting the target fluctuation value probability distribution to perform fluctuation value verification on the offset value to be verified includes: Obtaining a relative output offset value at a moment before the target traversal moment as a verification value; Subtracting the verification value from the offset value to be verified to obtain a fluctuation value; Determine whether the fluctuation value satisfies the target fluctuation value probability distribution, and if so, determine that the offset value to be verified passes the fluctuation value verification; If not, it is determined that the offset value to be verified has not passed the fluctuation value verification.

11. A method for predicting a photovoltaic output sequence according to claim 10, characterized in that: Also includes: Get the historical sunrise time, historical sunrise duration, historical sunset time and historical sunset duration of several historical dates; According to the historical sunrise time, the historical sunrise duration, the historical sunset time and the historical sunset duration, determine from the historical relative output the historical sunrise relative output corresponding to the historical sunrise time, the historical sunrise fluctuation output at several moments in the historical sunrise duration, the historical sunset relative output corresponding to the historical sunset time, and the historical sunset fluctuation output at several moments in the historical sunset duration; Generate a sunrise reference output distribution according to the ratios of the historical sunrise fluctuation output to each of the historical sunrise relative outputs; A sunset reference output distribution is generated according to the ratios of the historical sunset fluctuation outputs to the historical sunset relative outputs.

12. A method for predicting a photovoltaic output sequence according to claim 11, characterized in that: The generating of the predicted relative output curve of each of the dates to be predicted according to the relative output reference value and the relative output offset value at each moment of each of the dates to be predicted comprises: Obtain the sunrise time, sunrise duration, sunset time and sunset duration of each date to be predicted; According to the sunrise time and sunset time, determine a number of correction times and a number of normal times that need to be corrected in each predicted date; wherein the correction time includes: sunrise time, sunset time, a number of sunrise correction times in the sunrise time, and a number of sunset correction times in the sunrise time; Calculate the initial predicted relative output at each revised moment in the date to be predicted and the first predicted relative output at each normal moment according to the relative output reference value and the relative output offset value of the date to be predicted; According to the sunrise reference output, the initial predicted relative output at the sunrise time and each of the sunrise correction times is corrected to generate a second predicted relative output at the sunrise time and each of the sunrise correction times; According to the sunset reference output, the initial predicted relative output at the sunset time and each of the sunset correction times is corrected to generate a third predicted relative output at the sunset time and each of the sunset correction times; A predicted relative output curve for each date to be predicted is generated according to the first predicted relative output, the second predicted relative output, and the third predicted relative output for each date to be predicted.

13. A device for predicting photovoltaic output sequence, characterized in that: include: A data acquisition module is used to acquire the historical actual power and historical radiation intensity of the photovoltaic equipment at each moment on a number of historical dates, and the future radiation intensity of the photovoltaic equipment at each moment on a number of dates to be predicted; A power splitting module, used for splitting each of the historical actual powers into a historical net air theoretical output and a historical relative output according to the historical actual power and the historical radiation intensity; A data clustering module, for generating a relative output curve for each of the historical dates according to the historical relative output, clustering the relative output curves, and classifying each of the relative output curves into an output curve data set corresponding to a plurality of preset weather types; A weather forecast module is used to perform fitting analysis and forecasting on the relative output curves in the output curve data sets and the corresponding historical dates, and generate a baseline value probability distribution, an offset value probability distribution, a fluctuation value probability distribution, and a weather forecast sequence for several dates to be predicted for the relative output curves of the photovoltaic equipment under various weather types; A relative output prediction module, used for generating a predicted relative output curve for each of the to-be-predicted dates according to the weather prediction sequence, the baseline value probability distribution, the offset value probability distribution, and the fluctuation value probability distribution; The output prediction module is used to determine the net air theoretical output corresponding to each moment of each of the dates to be predicted according to the future radiation intensity, and to generate a photovoltaic output prediction sequence corresponding to each of the dates to be predicted according to the net air theoretical output and the predicted relative output curve.

14. A terminal device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, a method for predicting a photovoltaic output sequence as described in any one of claims 1 to 12 is implemented.

15. A storage medium, characterized in that: The storage medium includes a stored computer program, wherein when the computer program is executed, the device where the storage medium is located is controlled to execute a method for predicting a photovoltaic output sequence as claimed in any one of claims 1 to 12.

Images