WO2023216576A1 - 一种光伏发电短期功率预测方法和系统 - Google Patents

一种光伏发电短期功率预测方法和系统 Download PDF

Info

Publication number
WO2023216576A1
WO2023216576A1 PCT/CN2022/137289 CN2022137289W WO2023216576A1 WO 2023216576 A1 WO2023216576 A1 WO 2023216576A1 CN 2022137289 W CN2022137289 W CN 2022137289W WO 2023216576 A1 WO2023216576 A1 WO 2023216576A1
Authority
WO
WIPO (PCT)
Prior art keywords
value
forecast
power generation
photovoltaic power
meteorological
Prior art date
Application number
PCT/CN2022/137289
Other languages
English (en)
French (fr)
Inventor
李登宣
周海
程序
秦放
马文文
胡思雨
姚虹春
陈卫东
吴骥
丁煌
崔方
居蓉蓉
秦昊
李悦岑
Original Assignee
中国电力科学研究院有限公司
国家电网有限公司
国网新疆电力有限公司电力科学研究院
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中国电力科学研究院有限公司, 国家电网有限公司, 国网新疆电力有限公司电力科学研究院 filed Critical 中国电力科学研究院有限公司
Publication of WO2023216576A1 publication Critical patent/WO2023216576A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/22Matching criteria, e.g. proximity measures
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/04Architecture, e.g. interconnection topology
    • G06N3/044Recurrent networks, e.g. Hopfield networks
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/08Learning methods
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/04Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06Energy or water supply
    • HELECTRICITY
    • H02GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02JCIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00Circuit arrangements for ac mains or ac distribution networks
    • HELECTRICITY
    • H02GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02JCIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00Circuit arrangements for ac mains or ac distribution networks
    • H02J3/004Generation forecast, e.g. methods or systems for forecasting future energy generation

Definitions

  • the invention belongs to the technical field of new energy power generation, and specifically relates to a short-term power prediction method and system for photovoltaic power generation.
  • Photovoltaic short-term power forecast refers to the forecast of the active power of the photovoltaic power station from 0:00 the next day to the next 72 hours, with a time resolution of 15 minutes. Photovoltaic output is volatile and random. The connection of large-scale photovoltaic power generation to the grid will have a great impact on the safe and stable operation of the power system. Accurate short-term power prediction of photovoltaic power generation can help the power dispatching department adjust dispatch in advance according to changes in photovoltaic output. The plan also makes full use of photovoltaic resources and enhances the competitiveness of photovoltaic power generation in the power market, thereby achieving greater economic and social benefits.
  • a short-term power prediction model for photovoltaic power generation can be established through a neural network, and the short-term power prediction value of photovoltaic power generation can be obtained by inputting variable parameters into the model.
  • this technical solution has the following shortcomings: (1) The training speed is slow and the input variables More, the calculation is more complicated; (2) it is easy to fall into local extreme values; (3) it is highly dependent on data and requires a large number of historical samples. It can also collect geographical information of photovoltaic power stations, meteorological resource monitoring data, and photovoltaic module performance parameters. , forecast meteorological elements through a mesoscale numerical weather prediction model, and establish an error correction type for photovoltaic power station forecast elements and a photoelectric conversion model.
  • the present invention proposes a photovoltaic power generation short-term power prediction method, which includes:
  • the Euclidean distance is calculated by combining meteorological elements with the goal of obtaining a moment similar to the characteristics of the current NWP prediction sequence.
  • the Euclidean distance of the meteorological element is calculated based on the absolute error of the meteorological element between each forecast moment and each corresponding historical moment within the forecast period, the pre-calculated standard deviation of the meteorological element and the pre-calculated weight value of the meteorological element, including :
  • the calculation process of the meteorological element weight values includes:
  • the impact change value of the simulated power value is generated;
  • the BP neural network model is trained with the numerical model of meteorological elements as input and the actual power of historical photovoltaic power generation as output.
  • the Euclidean distance between each forecast time within the forecast period and each corresponding historical time meteorological element is calculated as follows:
  • ⁇ m is the weight value of meteorological element m
  • ⁇ m is the standard deviation of meteorological element m in the modeling sample data set
  • m is the meteorological element value
  • the meteorological elements include at least one or more of the following: total irradiance, direct radiation irradiance, scattered radiation irradiance, component temperature, average wind speed, average wind direction, ambient temperature and relative humidity. .
  • the weight value corresponding to the actual power value of photovoltaic power generation is calculated by the following formula:
  • eta s is the weight value corresponding to the actual power value of the s-th photovoltaic power generation
  • t s is the corresponding historical moment of the s-th
  • the value range of s is [1, 2, 3,..., j ]
  • j is the total number of preset quantities
  • the predicted photovoltaic power generation value at each forecast time is calculated by the following formula:
  • t' i is the i-th forecast time
  • eta s is the weight value corresponding to the actual power value of s-th photovoltaic power generation
  • the value range of s is [1 ,2,3,...,j]
  • j is the total number of preset quantities, is the actual power value of photovoltaic power generation at the sth historical moment.
  • the present invention also provides a photovoltaic power generation short-term power prediction system, including:
  • Euclidean distance calculation module photovoltaic power generation forecast power weight calculation module, photovoltaic power generation forecast power calculation module and power generation short-term power forecast module;
  • the Euclidean distance calculation module is used to calculate the Euclidean distance of the meteorological elements based on the absolute error of the meteorological elements at each forecast time and each corresponding historical moment within the forecast period, the pre-calculated standard deviation of the meteorological elements and the pre-calculated weight value of the meteorological elements. distance;
  • the photovoltaic power generation prediction power weight calculation module is used to select a preset number of historical moments with the smallest Euclidean distance for each forecast moment, obtain the actual photovoltaic power generation value corresponding to the historical moment, and calculate the actual photovoltaic power generation The weight value corresponding to the power value;
  • the photovoltaic power generation predicted power calculation module is used to perform a weighted sum of the weight values corresponding to the photovoltaic power generation actual power value and the photovoltaic power generation actual power value to obtain the photovoltaic power generation predicted power value at each forecast time;
  • the short-term power generation power prediction module is used to summarize the photovoltaic power generation forecast power values at each forecast time and generate a short-term photovoltaic power generation power forecast.
  • the Euclidean distance calculation module is specifically used for:
  • the calculation process of meteorological element weight values in the Euclidean distance calculation module includes:
  • the impact change value of the simulated power value is generated;
  • the BP neural network model is trained with the numerical model of meteorological elements as input and the actual power of historical photovoltaic power generation as output.
  • the Euclidean distance calculation module calculates the Euclidean distance between each forecast time within the forecast period and each corresponding historical time meteorological element according to the following formula:
  • ⁇ m is the weight value of meteorological element m
  • ⁇ m is the standard deviation of meteorological element m in the modeling sample data set
  • m is the meteorological element value
  • the meteorological elements of the Euclidean distance calculation module include at least one or more of the following: total irradiance, direct radiation irradiance, scattered radiation irradiance, component temperature, average wind speed, average wind direction, Ambient temperature and relative humidity.
  • the photovoltaic power generation predicted power weight calculation module calculates the weight value corresponding to the actual photovoltaic power generation power value according to the following formula:
  • eta s is the weight value corresponding to the actual power value of the s-th photovoltaic power generation
  • t s is the corresponding historical moment of the s-th
  • the value range of s is [1, 2, 3,...,j ]
  • j is the total number of preset quantities
  • the photovoltaic power generation prediction power calculation module calculates the photovoltaic power generation prediction power value at each forecast time according to the following formula:
  • t' i is the i-th forecast time
  • eta s is the weight value corresponding to the actual power value of s-th photovoltaic power generation
  • the value range of s is [1 ,2,3,...,j]
  • j is the total number of preset quantities, is the actual power value of photovoltaic power generation at the sth historical moment.
  • the present invention also provides a computer device, including: one or more processors;
  • Memory used to store one or more programs
  • the present invention also provides a computer-readable storage medium on which a computer program is stored.
  • a computer program is stored.
  • a photovoltaic power generation short-term power prediction method as described above is implemented.
  • the present invention provides a photovoltaic power generation short-term power prediction method and system, including: based on the absolute error of meteorological elements at each forecast time and each corresponding historical moment within the forecast period, the pre-calculated standard deviation of meteorological elements and the pre-calculated meteorological elements
  • the weight value is used to calculate the Euclidean distance of the meteorological elements; for each forecast time, a preset number of historical moments with the smallest Euclidean distance are selected, the actual photovoltaic power generation value corresponding to the historical moment is obtained, and the photovoltaic power generation is calculated
  • the weight value corresponding to the actual power value performing a weighted summation of the weight value corresponding to the actual photovoltaic power value and the actual photovoltaic power value to obtain the predicted photovoltaic power value at each forecast time; summarizing the photovoltaic power values at each forecast time
  • the power generation forecast power value is used to generate a short-term power forecast for photovoltaic power generation; where the Euclidean distance is calculated
  • Photovoltaic power stations with incomplete conditions are simple and practical, reliable in calculation, easy to implement and deploy on site, and have strong operability and promotion and application value.
  • This invention does not require correction of numerical models of measured meteorological elements, and optimizes the process that may produce forecast errors, helping to improve the accuracy of short-term power prediction of voltaic power generation.
  • Figure 1 is a flow chart of a photovoltaic power generation short-term power prediction method provided by the present invention
  • Figure 2 is a flow chart of a specific example of a photovoltaic power generation short-term power prediction method provided by the present invention
  • Figure 3 is a schematic diagram of short-term power prediction of photovoltaic power generation using a physical method provided by the present invention
  • Figure 4 is a schematic diagram of short-term power prediction of photovoltaic power generation provided by the present invention.
  • Figure 5 is a schematic diagram of a photovoltaic power generation short-term power prediction system provided by the present invention.
  • the present invention provides a short-term power prediction method for photovoltaic power generation, the flow diagram of which is shown in Figure 1, including:
  • Step 1 Calculate the Euclidean distance of the meteorological element based on the absolute error of the meteorological element between each forecast moment and each corresponding historical moment within the forecast period, the pre-calculated standard deviation of the meteorological element and the pre-calculated weight value of the meteorological element;
  • Step 2 For each forecast moment, select a preset number of historical moments with the smallest Euclidean distance, obtain the actual photovoltaic power generation value corresponding to the historical moment, and calculate the weight value corresponding to the actual photovoltaic power generation power value;
  • Step 3 Perform a weighted summation of the weight values corresponding to the actual photovoltaic power value and the actual photovoltaic power value to obtain the predicted photovoltaic power value at each forecast time;
  • Step 4 Summarize the photovoltaic power generation predicted power values at each forecast time to generate a short-term photovoltaic power generation power forecast.
  • step 1 includes:
  • meteorological element m at least includes the following One or more of: ⁇ RG, RD, RF, TM, WS, WD, T, RH,... ⁇ , RG is the total irradiance, RD is the direct radiation irradiance, RF is Scattered radiation irradiance, TM is the module temperature, WS is the average wind speed, WD is the average wind direction, T is the ambient temperature, and RH is the relative humidity.
  • the BP neural network model is trained by establishing a model training data set, excluding the actual power of the photovoltaic power station during abnormal periods such as power cuts, and combining it with the numerical model of meteorological elements to form a multivariate data sequence.
  • the data set is established in the format of a data vector:
  • P t is the actual power of photovoltaic power generation at time t;
  • is the numerical model of meteorological elements at time t.
  • Meteorological elements m include: RG t is the total irradiance at time t, RD t is the direct radiation irradiation at time t degree, RF t is the scattered radiation irradiance at time t, TM t is the component temperature at time t, WS t is the average wind speed at time t, WD t is the average wind direction at time t, T t is the ambient temperature at time t, RH t is the relative humidity at time t, etc.; t is the corresponding historical time, and the resolution is usually 15 minutes.
  • the input is set to the meteorological element numerical model, and the output is set to photovoltaic Actual power generated P t .
  • the adaptive learning ability of the neural network a large number of nonlinear mapping relationships f() between inputs and outputs are trained to build a BP neural network model.
  • the BP neural network model formula is formula (2):
  • P' t is the simulated power value at time t
  • f() is the nonlinear mapping relationship between the numerical model of meteorological elements and the actual power of historical photovoltaic power generation
  • is the numerical model of meteorological elements at time t, for Add a preset percentage to each meteorological element and bring it into the BP neural network model to generate simulated value-added power; subtract the preset percentage for each meteorological element and bring it into the BP neural network model to generate simulated impairment power; use the total irradiation Take the degree RG t as an example: add and subtract the preset percentage to the total irradiance RG t , and the new meteorological element numerical model generated is:
  • RG t + RG t *(1+ ⁇ )
  • RG t - RG t *(1- ⁇ )
  • P t,RG '+ is the simulated added power of the total irradiance at time t
  • P t,RG '- is the simulated depletion power of the total irradiance at time t
  • which is the change value of the simulated power value generated by the change of the total irradiance RG t on P' t
  • the calculation formula for the change value of the simulated power value of RG t is the following formula (3):
  • MIV t,RG is the influence change value of the simulated power value of the total irradiance RG t at time t.
  • the influence change value of the simulated power value of the total irradiance RG t at n forecast times is averaged to obtain the average
  • the calculation formula of the impact change value and the average impact change value is the following formula (4):
  • MIV RG is the average influence change value of the simulated power value of the total irradiance RG at n forecast moments.
  • the direct radiation irradiance RD t the scattered radiation irradiance RF t
  • the component temperature TM The average influence change value corresponding to t , average wind speed WS t , average wind direction WD t , ambient temperature T t , relative humidity RH t and other meteorological elements:
  • the online calculation method can share the error risk of single-point prediction, eliminate large deviations, and improve the accuracy of short-term power prediction of photovoltaic power generation;
  • step 2 includes:
  • eta s is the weight value corresponding to the actual power value of the s-th photovoltaic power generation
  • t s is the corresponding historical moment of the s-th
  • the value range of s is [1, 2, 3,...,j ]
  • j is the total number of preset quantities
  • Photovoltaic power station is the weight value corresponding to the actual power value of the s-th photovoltaic power generation
  • t s is the corresponding historical moment of the s-th
  • the value range of s is [1, 2, 3,...,j ]
  • j is the total number of preset quantities
  • This method relies little on measured meteorological data and has strong adaptability.
  • This method only needs to obtain the actual photovoltaic power generation at historical moments, has low requirements on the amount of historical data, and has good universality. It is especially suitable for areas with incomplete meteorological monitoring data conditions. Photovoltaic power station.
  • step 3 includes:
  • the calculation formula for the predicted photovoltaic power value at each forecast time is formula (7) :
  • step 4 includes:
  • the predicted power value of photovoltaic power generation at each forecast time in the forecast period is calculated, and the predicted power value of photovoltaic power generation at each forecast time is simultaneously output and summarized to generate a short-term power forecast for photovoltaic power generation.
  • This method is easy to implement and has strong operability and promotion and application value.
  • Figure 2 shows the model training data set including time-synchronized measured power and historical NWP forecast data. Based on MIV weight allocation, the Euclidean distance between the historical NWP forecast data and current NWP forecast data in the model training data set is established, and the Euclidean distance between the historical NWP forecast data and the current NWP forecast data is filtered out.
  • the present invention also provides a photovoltaic power generation short-term power prediction system, as shown in Figure 5, including:
  • Euclidean distance calculation module photovoltaic power generation forecast power weight calculation module, photovoltaic power generation forecast power calculation module and power generation short-term power forecast module;
  • the Euclidean distance calculation module is used to calculate the Euclidean distance of the meteorological elements based on the absolute error of the meteorological elements at each forecast time and each corresponding historical moment within the forecast period, the pre-calculated standard deviation of the meteorological elements and the pre-calculated weight value of the meteorological elements. distance;
  • the photovoltaic power generation prediction power weight calculation module is used to select a preset number of historical moments with the smallest Euclidean distance for each forecast moment, obtain the actual photovoltaic power generation value corresponding to the historical moment, and calculate the actual photovoltaic power generation The weight value corresponding to the power value;
  • the photovoltaic power generation predicted power calculation module is used to perform a weighted sum of the weight values corresponding to the photovoltaic power generation actual power value and the photovoltaic power generation actual power value to obtain the photovoltaic power generation predicted power value at each forecast time;
  • the short-term power generation power prediction module is used to summarize the photovoltaic power generation forecast power values at each forecast time and generate a short-term photovoltaic power generation power forecast.
  • the Euclidean distance calculation module is specifically used for:
  • the calculation process of meteorological element weight values in the Euclidean distance calculation module includes:
  • the impact change value of the simulated power value is generated;
  • the BP neural network model is trained with the numerical model of meteorological elements as input and the actual power of historical photovoltaic power generation as output.
  • the Euclidean distance calculation module calculates the Euclidean distance between each forecast time within the forecast period and each corresponding historical time meteorological element as follows:
  • ⁇ m is the weight value of meteorological element m
  • ⁇ m is the standard deviation of meteorological element m in the modeling sample data set
  • m is the meteorological element value
  • the meteorological elements of the Euclidean distance calculation module include at least one or more of the following: total irradiance, direct radiation irradiance, scattered radiation irradiance, component temperature, average wind speed, average wind direction, environment temperature and relative humidity.
  • the photovoltaic power generation predicted power weight calculation module calculates the weight value corresponding to the actual photovoltaic power generation power value according to the following formula:
  • eta s is the weight value corresponding to the actual power value of the s-th photovoltaic power generation
  • t s is the corresponding historical moment of the s-th
  • the value range of s is [1, 2, 3,..., j ]
  • j is the total number of preset quantities
  • the photovoltaic power generation prediction power calculation module calculates the photovoltaic power generation prediction power value at each forecast time according to the following formula:
  • t' i is the i-th forecast time
  • eta s is the weight value corresponding to the actual power value of s-th photovoltaic power generation
  • the value range of s is [1 ,2,3,...,j]
  • j is the total number of preset quantities, is the actual power value of photovoltaic power generation at the sth historical moment.
  • the present invention also provides a computer device.
  • the computer device includes a processor and a memory.
  • the memory is used to store a computer program.
  • the computer program includes program instructions.
  • the processor is used to execute the Program instructions stored on computer storage media.
  • the processor may be a Central Processing Unit (CPU), or other general-purpose processor, Digital Signal Processor (DSP), Application Specific Integrated Circuit (ASIC), or off-the-shelf programmable gate Array (Field-Programmable GateArray, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., which are the computing core and control core of the terminal, are suitable for implementing one or more instructions, and are specifically suitable for Loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding functions to implement the steps of a photovoltaic power generation short-term power prediction method in the above embodiment.
  • CPU Central Processing Unit
  • DSP Digital Signal Processor
  • ASIC Application Specific Integrated Circuit
  • the present invention also provides a storage medium, specifically a computer-readable storage medium (Memory).
  • the computer-readable storage medium is a memory device in a computer device and is used to store programs and data. It can be understood that the computer-readable storage medium here may include a built-in storage medium in the computer device, and of course may also include an extended storage medium supported by the computer device.
  • the computer-readable storage medium provides storage space, and the storage space stores the operating system of the terminal. Furthermore, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space. These instructions may be one or more computer programs (including program codes).
  • the computer-readable storage medium here may be a high-speed RAM memory or a non-volatile memory (non-volatile memory), such as at least one disk memory.
  • One or more instructions stored in the computer-readable storage medium can be loaded and executed by the processor to implement the steps of a photovoltaic power generation short-term power prediction method in the above embodiment.
  • embodiments of the present invention may be provided as methods, systems, or computer program products.
  • the invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects.
  • the invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.
  • These computer program instructions may also be stored in a computer-readable memory that causes a computer or other programmable data processing apparatus to operate in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including the instruction means, the instructions
  • the device implements the functions specified in a process or processes of the flowchart and/or a block or blocks of the block diagram.
  • These computer program instructions may also be loaded onto a computer or other programmable data processing device, causing a series of operating steps to be performed on the computer or other programmable device to produce computer-implemented processing, thereby executing on the computer or other programmable device.
  • Instructions provide steps for implementing the functions specified in a process or processes of a flowchart diagram and/or a block or blocks of a block diagram.

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Economics (AREA)
  • Health & Medical Sciences (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • Data Mining & Analysis (AREA)
  • General Health & Medical Sciences (AREA)
  • General Business, Economics & Management (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Marketing (AREA)
  • Tourism & Hospitality (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mathematical Physics (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Software Systems (AREA)
  • Computational Linguistics (AREA)
  • Computing Systems (AREA)
  • Biophysics (AREA)
  • Public Health (AREA)
  • Development Economics (AREA)
  • Power Engineering (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Game Theory and Decision Science (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Primary Health Care (AREA)
  • Water Supply & Treatment (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Photovoltaic Devices (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本发明提供了一种光伏发电短期功率预测方法和系统,包括:基于预报时段内各预报时刻与各相应历史时刻的气象要素绝对误差、预先计算的气象要素标准差和预先计算的气象要素权重值,计算气象要素的欧氏距离;对各预报时刻,选择预设数量个欧氏距离最小的历史时刻,获取历史时刻对应的光伏发电实际功率值,并计算光伏发电实际功率值对应的权重值;对光伏发电实际功率值和光伏发电实际功率值对应的权重值进行加权求和,获得各预报时刻的光伏发电预测功率值;汇总所述各预报时刻的光伏发电预测功率值,生成光伏发电短期功率预测;本发明对实测气象数据依赖小、适应性强,且简单实用,计算可靠,易于实现和工程现场部署。

Description

一种光伏发电短期功率预测方法和系统
相关申请的交叉引用
本申请基于申请号为202210523546.X、申请日为2022年05月13日的中国专利申请提出,并要求该中国专利申请的优先权,该中国专利申请的全部内容在此以引入方式并入本申请。
技术领域
本发明属于新能源发电技术领域,具体涉及一种光伏发电短期功率预测方法和系统。
背景技术
光伏短期功率预测指光伏电站次日零时起到未来72h有功功率的预测,时间分辨率为15min。光伏出力具有波动性和随机性,大规模光伏发电并网会给电力系统的安全稳定运行造成较大的影响,准确的光伏发电短期功率预测有助于电力调度部门提前根据光伏出力变化及时调整调度计划,同时使得光伏资源得到充分利用,增强光伏发电参与电力市场的竞争力,从而获得更大的经济效益和社会效益。
现有技术中,可以通过神经网络建立光伏发电短期功率预测模型,通过向模型中输入变量参数,获得光伏发电短期功率预测值,但是这个技术方案有以下缺点:(1)训练速度慢,输入变量较多,计算比较复杂;(2)容易陷入局部极值;(3)对数据依赖程度高,需要大量的历史样本,还可以通过收集光伏电站地理信息、气象资源监测数据,及光伏组件性能参数,通过中尺度数值天气预报模式,预报气象要素,并建立光伏发电站预报要素误差校正型和光电转换模型,通过光电转换模型,输出光伏发电短期功率预测值,这个技术方案有以下缺点:(1)误差环节多,气象预报、模式校正、光电转换等环节均会产生误差;(2)适应性不强,对实测气象数据质量要求高,部分现场难以满足建模条件。
发明内容
为克服上述现有技术的不足,本发明提出一种光伏发电短期功率预测方法,包括:
基于预报时段内各预报时刻与各相应历史时刻的气象要素绝对误差、预先计算的气象要素标准差和预先计算的气象要素权重值,计算所述气象 要素的欧氏距离;
对各预报时刻,选择预设数量个欧氏距离最小的历史时刻,获取所述历史时刻对应的光伏发电实际功率值,并计算所述光伏发电实际功率值对应的权重值;
对所述光伏发电实际功率值和光伏发电实际功率值对应的权重值进行加权求和,获得各预报时刻的光伏发电预测功率值;
汇总所述各预报时刻的光伏发电预测功率值,生成光伏发电短期功率预测;
其中,所述欧氏距离是以获得与当前NWP预测序列特征相似的时刻为目标结合气象要素进行计算的。
优选的,所述基于预报时段内各预报时刻与各相应历史时刻的气象要素绝对误差、预先计算的气象要素标准差和预先计算的气象要素权重值,计算所述气象要素的欧氏距离,包括:
设定各预报时刻时间窗,并提取历史各天中各相应历史时刻时间窗;
获取各预报时刻与各相应历史时刻时间窗的气象要素值;
计算各预报时刻和各相应历史时刻时间窗中气象要素绝对误差,结合预先计算的气象要素权重值和气象要素标准差,确定预报时段内的各预报时刻与各相应历史时刻气象要素的欧式距离。
优选的,所述气象要素权重值的计算过程,包括:
将相应历史时刻的气象要素值输入到BP神经网络模型中,生成模拟功率值;
对各相应历史时刻的气象要素值加预设百分比,带入到BP神经网络模型中,生成模拟增值功率;
对各相应历史时刻的气象要素值减预设百分比,带入到BP神经网络模型中,生成模拟减值功率;
基于所述模拟增值功率和模拟减值功率,并结合MIV变量筛选算法,生成模拟功率值的影响变化值;
根据预设个数个预报时刻的影响变化值计算平均影响变化值,并基于所述平均影响变化值,确定气象要素权重值;
其中,所述BP神经网络模型是以气象要素数值模式为输入,以历史光伏发电实际功率为输出进行训练得到的。
优选的,所述预报时段内的各预报时刻与各相应历史时刻气象要素的欧式距离,按下式计算:
Figure PCTCN2022137289-appb-000001
其中,
Figure PCTCN2022137289-appb-000002
为预报时段内第i个预报时刻与相应历史时刻气象要素的欧式距离,t' i为第i个预报时刻,t i为第i个相应历史时刻,
Figure PCTCN2022137289-appb-000003
为第i个预报时刻的气象要素值,
Figure PCTCN2022137289-appb-000004
为第i个相应历史时刻的气象要素值,ω m为气象要素m的权 重值,δ m为建模样本数据集中气象要素m的标准差,m为气象要素值。
优选的,所述气象要素至少包括下述中的一种或多种:总辐照度、直接辐射辐照度、散射辐射辐照度、组件温度、平均风速、平均风向、环境温度和相对湿度。
优选的,所述光伏发电实际功率值对应的权重值,按下式计算:
Figure PCTCN2022137289-appb-000005
其中,η s为第s个光伏发电实际功率值对应的权重值,t s为第s个的相应历史时刻,s的取值范围为[1,2,3,......,j],j为预设数量的总数,
Figure PCTCN2022137289-appb-000006
为预报时段内第s个预报时刻与相应历史时刻气象要素的欧氏距离。
优选的,所述各预报时刻的光伏发电预测功率值,按下式计算:
Figure PCTCN2022137289-appb-000007
其中,
Figure PCTCN2022137289-appb-000008
为预报时段内第i个预报时刻的光伏发电预测功率值,t' i为第i个预报时刻,η s为第s个光伏发电实际功率值对应的权重值,s的取值范围为[1,2,3,......,j],j为预设数量的总数,
Figure PCTCN2022137289-appb-000009
为第s个历史时刻的光伏发电实际功率值。
基于同一发明构思,本发明还提供了一种光伏发电短期功率预测系统,包括:
欧式距离计算模块、光伏发电预测功率权重计算模块、光伏发电预测功率计算模块和发电短期功率预测模块;
所述欧式距离计算模块,用于基于预报时段内各预报时刻与各相应历史时刻的气象要素绝对误差、预先计算的气象要素标准差和预先计算的气象要素权重值,计算所述气象要素的欧氏距离;
所述光伏发电预测功率权重计算模块,用于对各预报时刻,选择预设数量个欧氏距离最小的历史时刻,获取所述历史时刻对应的光伏发电实际功率值,并计算所述光伏发电实际功率值对应的权重值;
所述光伏发电预测功率计算模块,用于对所述光伏发电实际功率值和光伏发电实际功率值对应的权重值进行加权求和,获得各预报时刻的光伏发电预测功率值;
所述发电短期功率预测模块,用于汇总所述各预报时刻的光伏发电预测功率值,生成光伏发电短期功率预测。
优选的,所述欧式距离计算模块,具体用于:
设定各预报时刻时间窗,并提取历史各天中各相应历史时刻时间窗;
获取各预报时刻与各相应历史时刻时间窗的气象要素值;
计算各预报时刻和各相应历史时刻时间窗中气象要素绝对误差,结合预先计算的气象要素权重值和气象要素标准差,确定预报时段内的各预报 时刻与各相应历史时刻气象要素的欧式距离。
优选的,所述欧式距离计算模块中气象要素权重值的计算过程,包括:
将相应历史时刻的气象要素值输入到BP神经网络模型中,生成模拟功率值;
对各相应历史时刻的气象要素值加预设百分比,带入到BP神经网络模型中,生成模拟增值功率;
对各相应历史时刻的气象要素值减预设百分比,带入到BP神经网络模型中,生成模拟减值功率;
基于所述模拟增值功率和模拟减值功率,并结合MIV变量筛选算法,生成模拟功率值的影响变化值;
根据预设个数个预报时刻的影响变化值计算平均影响变化值,并基于所述平均影响变化值,确定气象要素权重值;
其中,所述BP神经网络模型是以气象要素数值模式为输入,以历史光伏发电实际功率为输出进行训练得到的。
优选的,所述欧式距离计算模块按下式计算预报时段内的各预报时刻与各相应历史时刻气象要素的欧式距离:
Figure PCTCN2022137289-appb-000010
其中,
Figure PCTCN2022137289-appb-000011
为预报时段内第i个预报时刻与相应历史时刻气象要素的欧式距离,t' i为第i个预报时刻,t i为第i个相应历史时刻,
Figure PCTCN2022137289-appb-000012
为第i个预报时刻的气象要素值,
Figure PCTCN2022137289-appb-000013
为第i个相应历史时刻的气象要素值,ω m为气象要素m的权重值,δ m为建模样本数据集中气象要素m的标准差,m为气象要素值。
优选的,所述欧式距离计算模块的气象要素至少包括下述中的一种或多种:总辐照度、直接辐射辐照度、散射辐射辐照度、组件温度、平均风速、平均风向、环境温度和相对湿度。
优选的,所述光伏发电预测功率权重计算模块按下式计算光伏发电实际功率值对应的权重值:
Figure PCTCN2022137289-appb-000014
其中,η s为第s个光伏发电实际功率值对应的权重值,t s为第s个的相应历史时刻,s的取值范围为[1,2,3,......,j],j为预设数量的总数,
Figure PCTCN2022137289-appb-000015
为预报时段内第s个预报时刻与相应历史时刻气象要素的欧氏距离。
优选的,所述光伏发电预测功率计算模块按下式计算各预报时刻的光伏发电预测功率值:
Figure PCTCN2022137289-appb-000016
其中,
Figure PCTCN2022137289-appb-000017
为预报时段内第i个预报时刻的光伏发电预测功率值,t' i为第i个预报时刻,η s为第s个光伏发电实际功率值对应的权重值,s的取值范围为[1,2,3,......,j],j为预设数量的总数,
Figure PCTCN2022137289-appb-000018
为第s个历史时刻的光伏发电实际功率值。
基于同一发明构思,本发明还提供了一种计算机设备,包括:一个或多个处理器;
存储器,用于存储一个或多个程序;
当所述一个或多个程序被所述一个或多个处理器执行时,实现如前所述的一种光伏发电短期功率预测方法。
基于同一发明构思,本发明还提供了一种计算机可读存储介质,其上存有计算机程序,所述计算机程序被执行时,实现如前所述的一种光伏发电短期功率预测方法。
与最接近的现有技术相比,本发明具有的有益效果如下:
1.本发明提供了一种光伏发电短期功率预测方法和系统,包括:基于预报时段内各预报时刻与各相应历史时刻的气象要素绝对误差、预先计算的气象要素标准差和预先计算的气象要素权重值,计算所述气象要素的欧氏距离;对各预报时刻,选择预设数量个欧氏距离最小的历史时刻,获取所述历史时刻对应的光伏发电实际功率值,并计算所述光伏发电实际功率值对应的权重值;对所述光伏发电实际功率值和光伏发电实际功率值对应的权重值进行加权求和,获得各预报时刻的光伏发电预测功率值;汇总所述各预报时刻的光伏发电预测功率值,生成光伏发电短期功率预测;其中,所述欧氏距离是以获得与当前NWP预测序列特征相似的时刻为目标结合气象要素进行计算的,通过加权方式在线计算,可以分摊单点预测的误差风险,消除大偏差,提高光伏发电短期功率预测精度,本发明对实测气象数据依赖小、适应性强,对历史数据的数据量要求低,普适性好,尤其适用于气象监测数据条件不完备的光伏电站,且简单实用,计算可靠,易于实现和工程现场部署,具有很强的可操作性和推广应用价值。
2.本发明不需要实测气象要素数值模式进行校正,且优化了可能产生预报误差的流程,有助于提升伏发电短期功率预测精度。
附图说明
图1为本发明提供的一种光伏发电短期功率预测方法流程图;
图2为本发明提供的一种光伏发电短期功率预测方法具体示例的流程图;
图3为本发明提供的一种物理方法光伏发电短期功率预测示意图;
图4为本发明提供的光伏发电短期功率预测示意图;
图5为本发明提供的一种光伏发电短期功率预测系统示意图。
具体实施方式
下面结合附图和案例对本发明的具体实施方式作进一步的详细说明。
本发明提供了一种光伏发电短期功率预测方法,其流程示意图如图1所示,包括:
步骤1:基于预报时段内各预报时刻与各相应历史时刻的气象要素绝对误差、预先计算的气象要素标准差和预先计算的气象要素权重值,计算所述气象要素的欧氏距离;
步骤2:对各预报时刻,选择预设数量个欧氏距离最小的历史时刻,获取所述历史时刻对应的光伏发电实际功率值,并计算所述光伏发电实际功率值对应的权重值;
步骤3对所述光伏发电实际功率值和光伏发电实际功率值对应的权重值进行加权求和,获得各预报时刻的光伏发电预测功率值;
步骤4:汇总所述各预报时刻的光伏发电预测功率值,生成光伏发电短期功率预测。
具体的,步骤1包括:
对于预报时段内某一预报时刻t' i,设定时间窗(t' i-μ,t' i+μ),提取历史各天对应的时间窗(t i-μ,t i+μ),计算其与预报时刻气象要素的欧氏距离序列:
Figure PCTCN2022137289-appb-000019
以预报时刻t i∈(t i-μ,t i+μ)为例,计算预报时刻与历史相应时刻的气象要素欧氏距离,需要先计算预报时刻和相应历史时刻时间窗中气象要素绝对误差,结合预先计算的气象要素权重值和气象要素标准差,确定预报时刻与相应历史时刻气象要素的欧式距离,该气象要素的欧式距离公式为公式(1):
Figure PCTCN2022137289-appb-000020
在公式(1)中,
Figure PCTCN2022137289-appb-000021
为预报时段内第i个预报时刻与相应历史时刻气象要素的欧式距离,t' i为第i个预报时刻,t i为第i个相应历史时刻,
Figure PCTCN2022137289-appb-000022
为第i个预报时刻的气象要素值,
Figure PCTCN2022137289-appb-000023
为第i个相应历史时刻的气象要素值,ω m为气象要素m的权重值,δ m为建模样本数据集中气象要素m的标准差,m为气象要素值,气象要素m至少包括下述中的一种或多种:{RG,RD,RF,TM,WS,WD,T,RH,......},RG为总辐照度,RD为直接辐射辐照度,RF为散射辐射辐照度,TM为组件温度,WS为平均风速,WD为平均风向,T为环境温度,RH为相对湿度。
通过建立模型训练数据集对BP神经网络模型进行训练,剔除限电等异 常时段的光伏发电站实际功率,结合气象要素数值模式,形成多元数据序列,该数据集的建立按照数据向量的格式:
(P t,{RG t,RD t,RF t,TM t,WS t,WD t,T t,RH t,......},t)
其中,P t为t时刻的光伏发电实际功率;{}为t时刻的气象要素数值模式,气象要素m包括:RG t为t时刻的总辐照度、RD t为t时刻的直接辐射辐照度、RF t为t时刻的散射辐射辐照度、TM t为t时刻的组件温度、WS t为t时刻的平均风速、WD t为t时刻的平均风向、T t为t时刻的环境温度、RH t为t时刻的相对湿度等;t为相应历史时刻时间,分辨率通常为15分钟,利用神经网络方法,基于上述模型训练数据集,设定输入为气象要素数值模式,设定输出为光伏发电实际功率P t。通过神经网络自适应学习能力,训练大量的输入与输出之间的非线性映射关系f(),构建BP神经网络模型。
在计算气象要素权重值时,首先将所述气象要素输入到BP神经网络模型中,生成模拟功率值,BP神经网络模型公式为公式(2):
P' t=f({RG t,RD t,RF t,TM t,WS t,WD t,T t,RH t,......},t)       (2)
BP神经网络模型公式中,P' t为t时刻的模拟功率值,f()为气象要素数值模式与历史光伏发电实际功率的非线性映射关系,{}为t时刻的气象要素数值模式,对各气象要素加预设百分比,带入到BP神经网络模型中,生成模拟增值功率;对各气象要素减预设百分比,带入到BP神经网络模型中,生成模拟减值功率;以总辐照度RG t为例:对总辐照度RG t加预设百分比和减预设百分比,生成的新的气象要素数值模式为:
{RG t +,RD t,RF t,TM t,WS t,WD t,T t,RH t,......}
{RG t -,RD t,RF t,TM t,WS t,WD t,T t,RH t,......}
其中,RG t +=RG t*(1+λ),RG t -=RG t*(1-λ),通常的λ=10%,通过BP神经网络模型进行仿真,得到两个仿真结果:
P t,RG '+=f({RG t +,RD t,RF t,TM t,WS t,WD t,T t,RH t,......},t)
P t,RG '-=f({RG t -,RD t,RF t,TM t,WS t,WD t,T t,RH t,......},t)
其中,P t,RG '+为t时刻的总辐照度的模拟增值功率,P t,RG '-为t时刻的总辐照度的模拟减值功率,两个仿真结果的差值|P t,RG '+-P t,RG '-|,即为该总辐照度RG t变动对P' t产生的模拟功率值的影响变化值,RG t模拟功率值的影响变化值的计算公式为下述公式(3):
MIV t,RG=|P t,RG '+-P t,RG '-|          (3)
其中,MIV t,RG为t时刻的总辐照度RG t的模拟功率值的影响变化值,将n个预报时刻的总辐照度RG t的模拟功率值的影响变化值平均,得出平均影响变化值,平均影响变化值计算公式为下述公式(4):
Figure PCTCN2022137289-appb-000024
其中,MIV RG为n个预报时刻的总辐照度RG的模拟功率值的平均影响变化值,同理,可以得出直接辐射辐照度RD t、散射辐射辐照度RF t、组件温度TM t、平均风速WS t、平均风向WD t、环境温度T t、相对湿度RH t等气象要素对应的平均影响变化值:
{MIV RD,MIV RF,MIV TM,MIV WS,MIV WD,MIV T,MIV RH,......}
则,得出各气象要素对光伏发电出力影响的权重:
RGRDRFTMWSWDTRH,......}
以总辐照度权重为例,总辐照度权重的计算公式为公式(5):
Figure PCTCN2022137289-appb-000025
通过计算预报时段内的各个预报时刻与相应历史时刻气象要素的欧式距离,可以根据多气象要素寻找相似特征,在与预报时刻对应的历史时刻中查找与当前NWP预测序列特征相似的时刻,方法简单实用,计算可靠,流程更加精简,本发明不需要实测气象对模式预报结果进行校正,还省略了光电转换过程,优化了可能产生预报误差的流程,有助于提升功率预测精度,通过加权误差的方式在线计算,可以分摊单点预测的误差风险,消除大偏差,提高光伏发电短期功率预测精度;
具体的,步骤2包括:
对各历史时刻与预报时刻t' i气象要素的欧氏距离序列进行排序:
Figure PCTCN2022137289-appb-000026
得到欧式距离最小的前j个值,提取其对应的实际功率值:
Figure PCTCN2022137289-appb-000027
计算各实际功率值的权重系数η,各实际功率值的权重系数η计算公式为公式(6):
Figure PCTCN2022137289-appb-000028
其中,η s为第s个光伏发电实际功率值对应的权重值,t s为第s个的相应历史时刻,s的取值范围为[1,2,3,......,j],j为预设数量的总数,
Figure PCTCN2022137289-appb-000029
为预报时段内第s个预报时刻与相应历史时刻气象要素的欧氏距离。本方法对实测气象数据依赖小、适应性强,本发明仅需获取历史时刻的光伏发电实际功率,对历史数据的数据量要求低,普适性好,尤其适用于气象监测数据条件不完备的光伏电站。
具体的,步骤3包括:
对所述光伏发电实际功率值和光伏发电实际功率值对应的权重值进行 加权求和,获得各预报时刻的光伏发电预测功率值,各预报时刻的光伏发电预测功率值计算公式为公式(7):
Figure PCTCN2022137289-appb-000030
其中,
Figure PCTCN2022137289-appb-000031
为预报时段内第i个预报时刻的光伏发电预测功率值,t' i为第i个预报时刻,η s为第s个光伏发电实际功率值对应的权重值,s的取值范围为[1,2,3,......,j],j为预设数量的总数,
Figure PCTCN2022137289-appb-000032
为第s个历史时刻的光伏发电实际功率值。通过提取预测时刻对应的时刻历史实际功率,按照欧式距离进行加权得到预报时刻的功率预测值,这种方法计算过程少,可以减少误差,提高预测准确度。
具体的,步骤4包括:
同时计算预报时段的各预报时刻的光伏发电预测功率值,并将各预报时刻的光伏发电预测功率值同时输出,进行汇总,生成光伏发电短期功率预测。这种方法易于实现,具有很强的可操作性和推广应用价值。
下面以装机容量为80MW的某光伏发电站为例,通过如图2所示的光伏发电短期功率预测方法具体示例的流程图,验证本发明的预测效果。图2显示了模型训练数据集包括时间同步的实测功率和历史NWP预报数据,基于MIV权重分配建立模型训练数据集中的历史NWP预报数据和当前NWP预报数据之间的欧式距离,并筛选出距离较小的预报序列;根据模型训练数据集中的实测功率和筛选出的距离较小的预报序列,筛选出对应时间序列的实测功率;之后依据欧式距离加权组合,最后进行光伏发电短期功率预测。
选取2021年11月1日-11月30日共计30天的短期预测功率、实际发电功率数据进行验证,预测效果如图3、4和表1所示,算例结果表明,本发明短期预测功率各项误差指标均优于LSTM神经网络、物理方法的短期预测功率,应用本发明的方法可以有效提高光伏发电短期功率预测精度。
表1预测误差指标
Figure PCTCN2022137289-appb-000033
实施例2:
基于同一发明构思,本发明还提供了一种光伏发电短期功率预测系统,如图5所示,包括:
欧式距离计算模块、光伏发电预测功率权重计算模块、光伏发电预测功率计算模块和发电短期功率预测模块;
所述欧式距离计算模块,用于基于预报时段内各预报时刻与各相应历史时刻的气象要素绝对误差、预先计算的气象要素标准差和预先计算的气象要素权重值,计算所述气象要素的欧氏距离;
所述光伏发电预测功率权重计算模块,用于对各预报时刻,选择预设数量个欧氏距离最小的历史时刻,获取所述历史时刻对应的光伏发电实际功率值,并计算所述光伏发电实际功率值对应的权重值;
所述光伏发电预测功率计算模块,用于对所述光伏发电实际功率值和光伏发电实际功率值对应的权重值进行加权求和,获得各预报时刻的光伏发电预测功率值;
所述发电短期功率预测模块,用于汇总所述各预报时刻的光伏发电预测功率值,生成光伏发电短期功率预测。
其中,所述欧式距离计算模块,具体用于:
设定各预报时刻时间窗,并提取历史各天中各相应历史时刻时间窗;
获取各预报时刻与各相应历史时刻时间窗的气象要素值;
计算各预报时刻和各相应历史时刻时间窗中气象要素绝对误差,结合预先计算的气象要素权重值和气象要素标准差,确定预报时段内的各预报时刻与各相应历史时刻气象要素的欧式距离。
其中,所述欧式距离计算模块中气象要素权重值的计算过程,包括:
将相应历史时刻的气象要素值输入到BP神经网络模型中,生成模拟功率值;
对各相应历史时刻的气象要素值加预设百分比,带入到BP神经网络模型中,生成模拟增值功率;
对各相应历史时刻的气象要素值减预设百分比,带入到BP神经网络模型中,生成模拟减值功率;
基于所述模拟增值功率和模拟减值功率,并结合MIV变量筛选算法,生成模拟功率值的影响变化值;
根据预设个数个预报时刻的影响变化值计算平均影响变化值,并基于所述平均影响变化值,确定气象要素权重值;
其中,所述BP神经网络模型是以气象要素数值模式为输入,以历史光伏发电实际功率为输出进行训练得到的。
其中,所述欧式距离计算模块按下式计算预报时段内的各预报时刻与各相应历史时刻气象要素的欧式距离:
Figure PCTCN2022137289-appb-000034
其中,
Figure PCTCN2022137289-appb-000035
为预报时段内第i个预报时刻与相应历史时刻气象要素的欧式距离,t' i为第i个预报时刻,t i为第i个相应历史时刻,
Figure PCTCN2022137289-appb-000036
为第i个预报时刻的气象要素值,
Figure PCTCN2022137289-appb-000037
为第i个相应历史时刻的气象要素值,ω m为气象要素m的权重值,δ m为建模样本数据集中气象要素m的标准差,m为气象要素值。
其中,所述欧式距离计算模块的气象要素至少包括下述中的一种或多种:总辐照度、直接辐射辐照度、散射辐射辐照度、组件温度、平均风速、平均风向、环境温度和相对湿度。
其中,所述光伏发电预测功率权重计算模块按下式计算光伏发电实际功率值对应的权重值:
Figure PCTCN2022137289-appb-000038
其中,η s为第s个光伏发电实际功率值对应的权重值,t s为第s个的相应历史时刻,s的取值范围为[1,2,3,......,j],j为预设数量的总数,
Figure PCTCN2022137289-appb-000039
为预报时段内第s个预报时刻与相应历史时刻气象要素的欧氏距离。
其中,所述光伏发电预测功率计算模块按下式计算各预报时刻的光伏发电预测功率值:
Figure PCTCN2022137289-appb-000040
其中,
Figure PCTCN2022137289-appb-000041
为预报时段内第i个预报时刻的光伏发电预测功率值,t' i为第i个预报时刻,η s为第s个光伏发电实际功率值对应的权重值,s的取值范围为[1,2,3,......,j],j为预设数量的总数,
Figure PCTCN2022137289-appb-000042
为第s个历史时刻的光伏发电实际功率值。
实施例3:
基于同一种发明构思,本发明还提供了一种计算机设备,该计算机设备包括处理器以及存储器,所述存储器用于存储计算机程序,所述计算机程序包括程序指令,所述处理器用于执行所述计算机存储介质存储的程序指令。处理器可能是中央处理单元(Central Processing Unit,CPU),还可以是其他通用处理器、数字信号处理器(Digital Signal Processor、DSP)、专用集成电路(Application SpecificIntegrated Circuit,ASIC)、现成可编程门阵列(Field-Programmable GateArray,FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等,其是终端的计算核心以及控制核心,其适于实现一条或一条以上指令,具体适于加载并执行计算机存储介质内一条或一条以上指令从而实现相应方法流程或相应功能,以实现上述实施例中一种光伏发电短期功率预测方法的步骤。
实施例4:
基于同一种发明构思,本发明还提供了一种存储介质,具体为计算机可读存储介质(Memory),所述计算机可读存储介质是计算机设备中的记忆设备,用于存放程序和数据。可以理解的是,此处的计算机可读存储介质既可以包括计算机设备中的内置存储介质,当然也可以包括计算机设备所支持的扩展存储介质。计算机可读存储介质提供存储空间,该存储空间存储了终端的操作系统。并且,在该存储空间中还存放了适于被处理器加载并执行的一条或一条以上的指令,这些指令可以是一个或一个以上的计算机程序(包括程序代码)。需要说明的是,此处的计算机可读存储介质可以是高速RAM存储器,也可以是非不稳定的存储器(non-volatile memory),例如至少一个磁盘存储器。可由处理器加载并执行计算机可读存储介质中存放的一条或一条以上指令,以实现上述实施例中一种光伏发电短期功率预测方法的步骤。
本领域内的技术人员应明白,本发明的实施例可提供为方法、系统、或计算机程序产品。因此,本发明可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且,本发明可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
本发明是参照根据本发明实施例的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器,使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的计算机可读存储器中,使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品,该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。
这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上,使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理,从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的步骤。
最后应当说明的是:以上实施例仅用于说明本发明的技术方案而非对其保护范围的限制,尽管参照上述实施例对本发明进行了详细的说明,所属 领域的普通技术人员应当理解:本领域技术人员阅读本发明后依然可对申请的具体实施方式进行种种变更、修改或者等同替换,但这些变更、修改或者等同替换,均在申请待批的权利要求保护范围之内。

Claims (16)

  1. 一种光伏发电短期功率预测方法,包括:
    基于预报时段内各预报时刻与各相应历史时刻的气象要素绝对误差、预先计算的气象要素标准差和预先计算的气象要素权重值,计算所述气象要素的欧氏距离;
    对各预报时刻,选择预设数量个欧氏距离最小的历史时刻,获取所述历史时刻对应的光伏发电实际功率值,并计算所述光伏发电实际功率值对应的权重值;
    对所述光伏发电实际功率值和光伏发电实际功率值对应的权重值进行加权求和,获得各预报时刻的光伏发电预测功率值;
    汇总所述各预报时刻的光伏发电预测功率值,生成光伏发电短期功率预测;
    其中,所述欧氏距离是以获得与当前NWP预测序列特征相似的时刻为目标结合气象要素进行计算的。
  2. 如权利要求1所述的方法,其中,所述基于预报时段内各预报时刻与各相应历史时刻的气象要素绝对误差、预先计算的气象要素标准差和预先计算的气象要素权重值,计算所述气象要素的欧氏距离,包括:
    设定各预报时刻时间窗,并提取历史各天中各相应历史时刻时间窗;
    获取各预报时刻与各相应历史时刻时间窗的气象要素值;
    计算各预报时刻和各相应历史时刻时间窗中气象要素绝对误差,结合预先计算的气象要素权重值和气象要素标准差,确定预报时段内的各预报时刻与各相应历史时刻气象要素的欧式距离。
  3. 如权利要求2所述的方法,其中,所述气象要素权重值的计算过程,包括:
    将相应历史时刻的气象要素值输入到BP神经网络模型中,生成模拟功率值;
    对各相应历史时刻的气象要素值加预设百分比,带入到BP神经网络模型中,生成模拟增值功率;
    对各相应历史时刻的气象要素值减预设百分比,带入到BP神经网络模型中,生成模拟减值功率;
    基于所述模拟增值功率和模拟减值功率,并结合MIV变量筛选算法,生成模拟功率值的影响变化值;
    根据预设个数个预报时刻的影响变化值计算平均影响变化值,并基于所述平均影响变化值,确定气象要素权重值;
    其中,所述BP神经网络模型是以气象要素数值模式为输入,以历史光伏发电实际功率为输出进行训练得到的。
  4. 如权利要求2所述的方法,其中,所述预报时段内的各预报时刻与各相应历史时刻气象要素的欧式距离,按下式计算:
    Figure PCTCN2022137289-appb-100001
    其中,
    Figure PCTCN2022137289-appb-100002
    为预报时段内第i个预报时刻与相应历史时刻气象要素的欧式距离,t′ i为第i个预报时刻,t i为第i个相应历史时刻,
    Figure PCTCN2022137289-appb-100003
    为第i个预报时刻的气象要素值,
    Figure PCTCN2022137289-appb-100004
    为第i个相应历史时刻的气象要素值,ω m为气象要素m的权重值,δ m为建模样本数据集中气象要素m的标准差,m为气象要素值。
  5. 如权利要求1所述的方法,其中,所述气象要素至少包括下述中的一种或多种:总辐照度、直接辐射辐照度、散射辐射辐照度、组件温度、平均风速、平均风向、环境温度和相对湿度。
  6. 如权利要求1所述的方法,其中,所述光伏发电实际功率值对应的权重值,按下式计算:
    Figure PCTCN2022137289-appb-100005
    其中,η s为第s个光伏发电实际功率值对应的权重值,t s为第s个的相应历史时刻,s的取值范围为[1,2,3,......,j],j为预设数量的总数,
    Figure PCTCN2022137289-appb-100006
    为预报时段内第s个预报时刻与相应历史时刻气象要素的欧氏距离。
  7. 如权利要求1所述的方法,其中,所述各预报时刻的光伏发电预测功率值,按下式计算:
    Figure PCTCN2022137289-appb-100007
    其中,
    Figure PCTCN2022137289-appb-100008
    为预报时段内第i个预报时刻的光伏发电预测功率值,t′ i为第i个预报时刻,η s为第s个光伏发电实际功率值对应的权重值,s的取值范围为[1,2,3,......,j],j为预设数量的总数,
    Figure PCTCN2022137289-appb-100009
    为第s个历史时刻的光伏发电实际功率值。
  8. 一种光伏发电短期功率预测系统,包括:
    欧式距离计算模块、光伏发电预测功率权重计算模块、光伏发电预测功率计算模块和发电短期功率预测模块;
    所述欧式距离计算模块,用于基于预报时段内各预报时刻与各相应历史时刻的气象要素绝对误差、预先计算的气象要素标准差和预先计算的气象要素权重值,计算所述气象要素的欧氏距离;
    所述光伏发电预测功率权重计算模块,用于对各预报时刻,选择预设数量个欧氏距离最小的历史时刻,获取所述历史时刻对应的光伏发电实际功率值,并计算所述光伏发电实际功率值对应的权重值;
    所述光伏发电预测功率计算模块,用于对所述光伏发电实际功率值和光伏发电实际功率值对应的权重值进行加权求和,获得各预报时刻的光伏发电预测功率值;
    所述发电短期功率预测模块,用于汇总所述各预报时刻的光伏发电预测功率值,生成光伏发电短期功率预测。
  9. 如权利要求8所述的系统,其中,所述欧式距离计算模块,具体用于:
    设定各预报时刻时间窗,并提取历史各天中各相应历史时刻时间窗;
    获取各预报时刻与各相应历史时刻时间窗的气象要素值;
    计算各预报时刻和各相应历史时刻时间窗中气象要素绝对误差,结合预先计算的气象要素权重值和气象要素标准差,确定预报时段内的各预报时刻与各相应历史时刻气象要素的欧式距离。
  10. 如权利要求9所述的系统,其中,所述欧式距离计算模块中气象要素权重值的计算过程,包括:
    将相应历史时刻的气象要素值输入到BP神经网络模型中,生成模拟功率值;
    对各相应历史时刻的气象要素值加预设百分比,带入到BP神经网络模型中,生成模拟增值功率;
    对各相应历史时刻的气象要素值减预设百分比,带入到BP神经网络模型中,生成模拟减值功率;
    基于所述模拟增值功率和模拟减值功率,并结合MIV变量筛选算法,生成模拟功率值的影响变化值;
    根据预设个数个预报时刻的影响变化值计算平均影响变化值,并基于所述平均影响变化值,确定气象要素权重值;
    其中,所述BP神经网络模型是以气象要素数值模式为输入,以历史光伏发电实际功率为输出进行训练得到的。
  11. 如权利要求9所述的系统,其中,所述欧式距离计算模块按下式计算预报时段内的各预报时刻与各相应历史时刻气象要素的欧式距离:
    Figure PCTCN2022137289-appb-100010
    其中,
    Figure PCTCN2022137289-appb-100011
    为预报时段内第i个预报时刻与相应历史时刻气象要素的欧式距离,t′ i为第i个预报时刻,t i为第i个相应历史时刻,
    Figure PCTCN2022137289-appb-100012
    为第i个预报时刻的气象要素值,
    Figure PCTCN2022137289-appb-100013
    为第i个相应历史时刻的气象要素值,ω m为气象要素m的权重值,δ m为建模样本数据集中气象要素m的标准差,m为气象要素值。
  12. 如权利要求8所述的系统,其中,所述欧式距离计算模块的气象要素至少包括下述中的一种或多种:总辐照度、直接辐射辐照度、散射辐射辐照度、组件温度、平均风速、平均风向、环境温度和相对湿度。
  13. 如权利要求8所述的系统,其中,所述光伏发电预测功率权重计算模块按下式计算光伏发电实际功率值对应的权重值:
    Figure PCTCN2022137289-appb-100014
    其中,η s为第s个光伏发电实际功率值对应的权重值,t s为第s个的相应历史时刻,s的取值范围为[1,2,3,......,j],j为预设数量的总数,
    Figure PCTCN2022137289-appb-100015
    为预报时段内第s个预报时刻与相应历史时刻气象要素的欧氏距离。
  14. 如权利要求8所述的系统,其中,所述光伏发电预测功率计算模块按下式计算各预报时刻的光伏发电预测功率值:
    Figure PCTCN2022137289-appb-100016
    其中,
    Figure PCTCN2022137289-appb-100017
    为预报时段内第i个预报时刻的光伏发电预测功率值,t′ i为第i个预报时刻,η s为第s个光伏发电实际功率值对应的权重值,s的取值范围为[1,2,3,......,j],j为预设数量的总数,
    Figure PCTCN2022137289-appb-100018
    为第s个历史时刻的光伏发电实际功率值。
  15. 一种计算机设备,包括:一个或多个处理器;
    存储器,用于存储一个或多个程序;
    当所述一个或多个程序被所述一个或多个处理器执行时,实现如权利要求1至7中任一项所述的光伏发电短期功率预测方法。
  16. 一种计算机可读存储介质,其上存有计算机程序,所述计算机程序被执行时,实现如权利要求1至7中任一项所述的光伏发电短期功率预测方法。
PCT/CN2022/137289 2022-05-13 2022-12-07 一种光伏发电短期功率预测方法和系统 WO2023216576A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210523546.XA CN117117819A (zh) 2022-05-13 2022-05-13 一种光伏发电短期功率预测方法、系统、设备和介质
CN202210523546.X 2022-05-13

Publications (1)

Publication Number Publication Date
WO2023216576A1 true WO2023216576A1 (zh) 2023-11-16

Family

ID=88729557

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/137289 WO2023216576A1 (zh) 2022-05-13 2022-12-07 一种光伏发电短期功率预测方法和系统

Country Status (2)

Country Link
CN (1) CN117117819A (zh)
WO (1) WO2023216576A1 (zh)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117634701A (zh) * 2023-12-13 2024-03-01 国能日新科技股份有限公司 基于区域光伏集合预报的太阳辐射短期预报方法及系统
CN117878943A (zh) * 2023-12-25 2024-04-12 江苏明茂新能源科技有限公司 一种光伏微电网实时充电调度方法及系统
CN117895506A (zh) * 2024-01-29 2024-04-16 重庆千信新能源有限公司 针对动态天气状态进行光伏电力预测方法
CN118472945A (zh) * 2024-07-10 2024-08-09 国网山西省电力公司长治供电公司 高比例分布式光伏实测数据的zip负荷模型建立方法
CN118551168A (zh) * 2024-07-29 2024-08-27 国网浙江省电力有限公司乐清市供电公司 光伏储能设施缺失功率数据重构方法、装置、设备和介质

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117996756B (zh) * 2024-04-02 2024-07-26 国网山东省电力公司青州市供电公司 基于多维影响因子的新能源发电预测方法及系统

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105005825A (zh) * 2015-07-03 2015-10-28 深圳供电局有限公司 一种基于动态神经网络预测光伏功率的方法及系统
JP2016031621A (ja) * 2014-07-29 2016-03-07 日立造船株式会社 太陽光発電量予測方法および太陽光発電量予測システム
CN110619360A (zh) * 2019-09-09 2019-12-27 国家电网有限公司 一种考虑历史样本相似性的超短期风功率预测方法
CN112257941A (zh) * 2020-10-28 2021-01-22 福州大学 基于改进型Bi-LSTM的光伏电站短期功率预测方法
CN113592141A (zh) * 2021-06-28 2021-11-02 中国电力科学研究院有限公司 一种光伏发电功率预测方法和装置

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2016031621A (ja) * 2014-07-29 2016-03-07 日立造船株式会社 太陽光発電量予測方法および太陽光発電量予測システム
CN105005825A (zh) * 2015-07-03 2015-10-28 深圳供电局有限公司 一种基于动态神经网络预测光伏功率的方法及系统
CN110619360A (zh) * 2019-09-09 2019-12-27 国家电网有限公司 一种考虑历史样本相似性的超短期风功率预测方法
CN112257941A (zh) * 2020-10-28 2021-01-22 福州大学 基于改进型Bi-LSTM的光伏电站短期功率预测方法
CN113592141A (zh) * 2021-06-28 2021-11-02 中国电力科学研究院有限公司 一种光伏发电功率预测方法和装置

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117634701A (zh) * 2023-12-13 2024-03-01 国能日新科技股份有限公司 基于区域光伏集合预报的太阳辐射短期预报方法及系统
CN117878943A (zh) * 2023-12-25 2024-04-12 江苏明茂新能源科技有限公司 一种光伏微电网实时充电调度方法及系统
CN117895506A (zh) * 2024-01-29 2024-04-16 重庆千信新能源有限公司 针对动态天气状态进行光伏电力预测方法
CN118472945A (zh) * 2024-07-10 2024-08-09 国网山西省电力公司长治供电公司 高比例分布式光伏实测数据的zip负荷模型建立方法
CN118551168A (zh) * 2024-07-29 2024-08-27 国网浙江省电力有限公司乐清市供电公司 光伏储能设施缺失功率数据重构方法、装置、设备和介质

Also Published As

Publication number Publication date
CN117117819A (zh) 2023-11-24

Similar Documents

Publication Publication Date Title
WO2023216576A1 (zh) 一种光伏发电短期功率预测方法和系统
Wang et al. Sequence transfer correction algorithm for numerical weather prediction wind speed and its application in a wind power forecasting system
Behera et al. Solar photovoltaic power forecasting using optimized modified extreme learning machine technique
CN112285807B (zh) 一种气象信息预测方法及装置
CN114493050B (zh) 多维度融合的新能源功率并联预测方法和装置
CN103117546A (zh) 一种超短期风电功率滑动预测方法
CN110705771B (zh) 区域电网新能源发电功率预测优化方法和装置
WO2023115425A1 (zh) 一种超短时风速预测方法及系统
CN116595394A (zh) 风速修正模型的训练方法及风速预测方法、设备和介质
CN114792156A (zh) 基于曲线特征指标聚类的光伏输出功率预测方法和系统
CN111832800A (zh) 一种光伏电站功率预测方法和装置
CN114493052B (zh) 多模型融合自适应新能源功率预测方法和系统
CN114819374B (zh) 区域新能源超短期功率预测方法及系统
CN111697560B (zh) 一种基于lstm预测电力系统负荷的方法及系统
CN116205377B (zh) 分布式光伏电站出力预测方法、系统、计算机及存储介质
CN112862274A (zh) 一种基于WRF-LES和DeepAR相结合的风功率预测方法
Yan et al. A robust probabilistic wind power forecasting method considering wind scenarios
CN115526429A (zh) 风电功率预测误差解耦分析方法、处理器及存储介质
CN110705768B (zh) 用于风电场的风力发电功率预测优化方法和装置
Zhang et al. Short‐Term Power Prediction of Wind Power Generation System Based on Logistic Chaos Atom Search Optimization BP Neural Network
CN112036672B (zh) 一种基于迭代订正的新能源发电超短期功率预测方法及系统
CN110705769B (zh) 新能源发电功率预测优化方法和装置
CN112215383A (zh) 一种分布式光伏发电功率预测方法和系统
CN111178593A (zh) 一种光伏系统输出功率预测方法及装置
CN110705770B (zh) 用于光伏电站的光伏功率预测优化方法和装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22941511

Country of ref document: EP

Kind code of ref document: A1