CN108345996B - System and method for reducing wind power assessment electric quantity - Google Patents

System and method for reducing wind power assessment electric quantity Download PDF

Info

Publication number
CN108345996B
CN108345996B CN201810118649.1A CN201810118649A CN108345996B CN 108345996 B CN108345996 B CN 108345996B CN 201810118649 A CN201810118649 A CN 201810118649A CN 108345996 B CN108345996 B CN 108345996B
Authority
CN
China
Prior art keywords
wind power
data
wind
prediction
analysis
Prior art date
Legal status (The legal status 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 status listed.)
Active
Application number
CN201810118649.1A
Other languages
Chinese (zh)
Other versions
CN108345996A (en
Inventor
李伟
韩亚雄
王俊峰
王海挺
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Tianrun Xinneng Investment Co ltd
Original Assignee
Beijing Tianrun Xinneng Investment Co ltd
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 Beijing Tianrun Xinneng Investment Co ltd filed Critical Beijing Tianrun Xinneng Investment Co ltd
Priority to CN201810118649.1A priority Critical patent/CN108345996B/en
Publication of CN108345996A publication Critical patent/CN108345996A/en
Application granted granted Critical
Publication of CN108345996B publication Critical patent/CN108345996B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06QDATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/06Resources, workflows, human or project management, e.g. organising, planning, scheduling or allocating time, human or machine resources; Enterprise planning; Organisational models
    • G06Q10/063Operations research or analysis
    • G06Q10/0639Performance analysis
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06QDATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/04Forecasting or optimisation, e.g. linear programming, "travelling salesman problem" or "cutting stock problem"
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06QDATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
    • G06Q50/06Electricity, gas or water supply
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02EREDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E40/00Technologies for an efficient electrical power generation, transmission or distribution
    • Y02E40/70Smart grids as climate change mitigation technology in the energy generation sector
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02PCLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/80Management or planning
    • Y02P90/82Energy audits or management systems therefor
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04SSYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Abstract

The invention provides a method for reducing wind power assessment electric quantity, which comprises the following steps: (1) analyzing the wind power assessed reason: the method comprises weather forecast analysis, wind power prediction analysis, scheduling analysis, data reporting analysis and manual fault judgment analysis; (2) generating a strategy for reducing wind power assessment: and obtaining a strategy for reducing the wind power assessment electric quantity according to the analyzed reason. The system for reducing the wind power assessment electric quantity correspondingly is further disclosed, the wind power supplier loss caused by assessment of a power grid company on the wind power plant power prediction precision is reduced, a non-examination application is made for assessment of non-wind field reasons, and monthly assessment cost is reduced by using the existing conditions.

Description

System and method for reducing wind power assessment electric quantity
Technical Field
The invention belongs to an electric power system, and particularly relates to a system and a method for reducing wind power assessment electric quantity.
Background
Accurate wind power prediction can help a power grid dispatching department to make dispatching plans of various power supplies, improve the running stability of a power grid, improve the wind power consumption capacity of the power grid, further reduce economic loss brought to wind power developers due to power limitation, increase the return rate of investment of a wind power plant and provide an auxiliary means for the management work of the wind power plant.
The wind power prediction can be carried out in various classification modes, and the method comprises the steps of predicting wind speed firstly according to predicted physical quantities, then predicting output power according to a wind turbine generator or a wind power curve and directly predicting the output power; the method is classified into a continuous prediction method according to a digital model, an ARAM model, namely a differential autoregressive moving average model, and is commonly used in a random time sequence method, Kalman filtering and an intelligent method, such as an artificial neural network; dividing the input data into physical, statistical and comprehensive methods based on time series without data weather forecast data and by data weather forecast; the method is divided into ultra-short-term prediction and short-term prediction according to a time scale.
The wind power prediction system is checked by a certain technical means, and is particularly important, and the core problem of the wind power prediction check is the comparison between a power generation plan reported by a checked wind power plant and the actual wind power output on the day of production. The power generation plan reported by the wind power plant is based on a wind power assessment system, the system predicts a possible wind power curve, and the actual wind power output on the production day is represented by the ground data of a sample computer in the power plant. At present, the wind power prediction level is generally low due to the restriction of factors such as complex terrain, high difficulty in numerical prediction of near-earth wind speed, low precision of numerical weather prediction, late start of prediction technology research, high data quality requirement of a prediction method and the like, and the root mean square error of an all-day prediction result is less than 20% according to the requirement of the national energy agency on a temporary method for wind power plant power prediction and prediction management. The method is characterized in that a Pinglu horizontal longhole wind power plant is put into operation at 2015 year 12 and 20 days, enters an assessment period at 2016 year 7 and is affected by various factors such as maturity of wind power prediction system technology and wind resources, the wind power prediction assessment of the wind power plant at each month is basically in an assessed state, the nuclear power at 2016 year 7 and 7 months is 28.19MWH, the nuclear power at 2016 year 8 and 8 months is 129.16MWH, the nuclear power at 2016 year 9 and 9 months is 13.91MWH, the nuclear power at 2016 year 10 and 10 months is 6.35MWH, and the assessment electric power at 2016 is 177.61 MWH. Therefore, the reason that the accuracy of the wind power prediction system does not meet the power grid assessment requirement needs to be searched, the prediction rule is mastered, the reported prediction output is adjusted, the assessed specific data are analyzed, the examination-free application is provided for the assessment of non-wind field reasons, and the monthly assessment cost is reduced by using the existing conditions.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: the system and the method for reducing the wind power assessment electric quantity overcome the defects of the prior art, reduce the wind power supplier loss caused by assessment of a power grid company on the wind power station power prediction precision, provide a non-examination application for assessment of non-wind field reasons, and reduce monthly assessment cost by using the existing conditions.
Therefore, the invention aims to provide a system for reducing wind power assessment electric quantity, which comprises the following components:
(1) cause analysis system: the system comprises a weather forecast analysis system, a wind power prediction analysis system, a scheduling analysis system, a data reporting analysis system and a manual fault judgment analysis system, wherein the analysis result comprises: (1) the weather forecast is not transmitted successfully and the data is not updated timely; (2) wind power prediction system is imperfect, wind resources are unstable; (3) scheduling system reasons including participation in peak shaving, scheduling data network interruption, untimely data updating and incomplete uploading rate; (4) the accuracy rate of the reported data cannot be counted before reporting; (5) personnel judgment and low fault handling capacity;
(2) the strategy generation system comprises: obtaining a strategy for reducing the wind power assessment electric quantity according to the analysis result of the reason analysis system, wherein the strategy comprises the following steps: (1) the person on duty closely monitors, updates and optimizes the system program in time, and upgrades the system; (2) optimizing a wind power prediction system, adjusting a weather forecast model, preferentially selecting multiple different meteorological sources according to historical actual measurement data of a wind measuring tower on site, and optimizing and adjusting the power prediction model according to historical actual power data to ensure that the recent generation power characteristics can be presented by the prediction data used on site; (3) rechecking the current-month wind power assessment data; (4) checking the data accuracy of the previous day, and performing data supplementary report analysis; (5) the fault problems are summarized and analyzed, the fault elimination level of field personnel is improved, and the fault time is shortened.
Preferably, the wind power prediction system comprises, after being optimized:
(1) the data acquisition server comprises a data acquisition unit, a data acquisition unit and a data processing unit, wherein the data acquisition unit is used for operating data acquisition software, communicating with the wind power station side wind power comprehensive communication management terminal and acquiring data of a fan, a wind measuring tower, wind power station power, numerical weather forecast and local wind power prediction result data of the wind power station;
(2) a database server: the data processing system is used for processing, statistical analysis and storage of data, and in order to ensure reliable storage of the data, the database server is configured with a disk array;
(3) the application workstation comprises: the system comprises PC workstation equipment for completing the functions of modeling, graph generation and display, report making and printing of the system;
(4) wind power prediction server: the method comprises the steps that an entity server is used for operating a wind power prediction module, a neural network comprehensive algorithm based on a weighted least square support vector machine and quantum particle swarm prediction is used based on the collection or the provision of a numerical weather forecast provided by an SCADA system, and the short-term prediction and the ultra-short-term prediction are carried out on the output conditions of a single fan and the whole wind power plant by combining the real-time operation working condition of the fan of the wind power plant at present;
(5) a data interface server: obtaining a numerical weather forecast using an entity server;
(6) the reverse physical isolation equipment comprises a network switch, a network communication accessory, a physical isolation device, a cabinet and an accessory, is used for ensuring the network safety, is arranged at a network boundary, and simultaneously transmits the result of wind power prediction to a data acquisition and monitoring control System (SCADA) and an Energy Management System (EMS).
Preferably, the optimized wind power prediction system further comprises a wind measuring tower and a curve display module, wherein the wind measuring tower is used for measuring and implementing meteorological data, thereby carrying out ultra-short term power prediction, the entity anemometer tower is installed according to the actual physical conditions of the wind field to ensure the accuracy of prediction and is installed in the range of 5km of the wind field, real-time meteorological data of the wind tower are collected through GPRS or optical fibers, the height of the wind tower is not lower than the height of a fan hub, the wind measuring tower is characterized in that the wind measuring tower matched hardware equipment comprises a meteorological data sensor, data acquisition equipment and data transmission equipment, and comprises 3 wind speed sensors, 2 wind direction sensors, 1 atmospheric temperature sensor, 1 atmospheric humidity sensor, 1 atmospheric pressure sensor, a data collector and data transmission equipment, wherein the atmospheric temperature sensor and the atmospheric pressure sensor are arranged at the height of 8 meters; a wind direction sensor is arranged at the height of 10 meters, and one or more wind direction sensors are arranged at other installation heights; the method comprises the steps that one wind speed sensor is respectively installed at 10 meters and 30 meters, the result of wind power prediction is displayed by adopting a curve display module, the curve comprises a prediction curve and an actual curve, the current predicted power and the actual power of each wind field are displayed in a list form, various selections can be carried out through conditions such as a date control, a pull-down list of wind speed layer heights and the like, and the conditions are stored locally for subsequent query.
The invention also aims to provide a method for reducing wind power assessment electric quantity, which comprises the following steps:
(1) analyzing the wind power assessed reason: the method comprises weather forecast analysis, wind power prediction analysis, scheduling analysis, data reporting analysis and manual fault judgment analysis;
(2) generating a strategy for reducing wind power assessment: and obtaining a strategy for reducing the wind power assessment electric quantity according to the analyzed reason.
Preferably, the reason for the wind power being assessed and analyzed includes: (1) the weather forecast is not transmitted successfully and the data is not updated timely; (2) wind power prediction system is imperfect, wind resources are unstable; (3) scheduling system reasons including participation in peak shaving, scheduling data network interruption, untimely data updating and incomplete uploading rate; (4) the accuracy rate of the reported data cannot be counted before reporting; (5) and the personnel judgment and fault processing capacity is low.
Preferably, the policy includes: (1) the person on duty closely monitors, updates and optimizes the system program in time, and upgrades the system; (2) optimizing a wind power prediction system, adjusting a weather forecast model, preferentially selecting multiple different meteorological sources according to historical actual measurement data of a wind measuring tower on site, and optimizing and adjusting the power prediction model according to historical actual power data to ensure that the recent generation power characteristics can be presented by the prediction data used on site; (3) rechecking the current-month wind power assessment data; (4) checking the data accuracy of the previous day, and performing data supplementary report analysis; (5) the fault problems are summarized and analyzed, the fault elimination level of field personnel is improved, and the fault time is shortened.
Preferably, the optimized wind power prediction system adopts a comprehensive algorithm based on neural network prediction combined with a weighted least squares support vector machine and a quantum particle swarm algorithm in the system, and comprises the following steps:
(1) establishing a neural network model to predict meteorological data;
(2) establishing a double neural network model for predicting wind power data according to the obtained meteorological prediction data;
(3) and determining the relation between the correlation point and the wind speed time curve of the wind power plant, and predicting the wind power in the next hours.
Preferably, the step (1) includes: the historical data is preprocessed, namely missing data and dirty data are processed, then meteorological conditions are quantized, finally normalized, and feature vectors forming a certain day are input into a double neural network prediction model for training and prediction.
Preferably, wherein the step (2) comprises:
(2-1) encoding the hyper-parameters, wherein the encoding hyper-parameters comprise regularization parameters and kernel parameters, each particle is replaced by a potential solution to form a hyper-parameter group, and optimal parameters are selected;
(2-2) establishing an adaptive function, and evaluating the generalization performance;
(2-3) after iteration is finished, training a double neural network model, training an RNN (neural network) to be similar to a common neural network, and using a back propagation algorithm, wherein because the parameter at each moment is shared, the gradient of the parameter does not only depend on the output at the current moment but also depends on the previous moment;
(2-4) extension of the dual neural network model: the idea of bidirectional expansion is that the output at the time t depends not only on the previous elements but also on the following elements, and the output depends on the hidden states of the two neural network models, so that the two neural network models are subjected to bidirectional expansion.
Preferably, the adaptive function of the step (2-2) is: the fixed is 1/RMSE (gamma, sigma), wherein RMSE (gamma, sigma) is the root mean square error of the prediction result and changes along with the change of the least square support vector machine parameter pair (gamma, sigma), when the termination iteration criterion is satisfied, the maximum adaptive function corresponds to the optimum parameter of the least square support vector machine, and the termination iteration criterion of the algorithm comprises two modes: the first is that the algorithm stops when the value of the objective function is less than or equal to a given threshold epsilon; the second is to give an iteration number in advance and stop the iteration when this value is reached.
The above and other objects, advantages and features of the present invention will become more apparent to those skilled in the art from the following detailed description of specific embodiments thereof, taken in conjunction with the accompanying drawings.
Drawings
Some specific embodiments of the invention will be described in detail hereinafter, by way of illustration and not limitation, with reference to the accompanying drawings. The same reference numbers in the drawings identify the same or similar elements or components. Those skilled in the art will appreciate that the drawings are not necessarily drawn to scale. The objects and features of the present invention will become more apparent in view of the following description taken in conjunction with the accompanying drawings, in which:
FIG. 1 is a block diagram of a system for reducing wind power assessment electric quantity according to an embodiment of the invention;
FIG. 2 is a block diagram of an optimized wind power prediction system according to an embodiment of the present invention;
FIG. 3 is a flow chart of a method for reducing wind power assessment electric quantity according to an embodiment of the invention;
FIG. 4 is a flow chart of an optimized wind power prediction method according to an embodiment of the invention;
fig. 5 is a flow chart of short-term and ultra-short-term prediction of optimized wind power according to an embodiment of the present invention.
Detailed Description
Fig. 1 is a block diagram of a system for reducing wind power assessment electric quantity according to an embodiment of the invention, and the system comprises:
(1) cause analysis system: the system comprises a weather forecast analysis system, a wind power prediction analysis system, a scheduling analysis system, a data reporting analysis system and a manual fault judgment analysis system, wherein the analysis result comprises: (1) the weather forecast is not transmitted successfully and the data is not updated timely; (2) wind power prediction system is imperfect, wind resources are unstable; (3) scheduling system reasons including participation in peak shaving, scheduling data network interruption, untimely data updating and incomplete uploading rate; (4) the accuracy rate of the reported data cannot be counted before reporting; (5) personnel judgment and low fault handling capacity;
(2) the strategy generation system comprises: obtaining a strategy for reducing the wind power assessment electric quantity according to the analysis result of the reason analysis system, wherein the strategy comprises the following steps: (1) the person on duty closely monitors, updates and optimizes the system program in time, and upgrades the system; (2) optimizing a wind power prediction system, adjusting a weather forecast model, preferentially selecting multiple different meteorological sources according to historical actual measurement data of a wind measuring tower on site, and optimizing and adjusting the power prediction model according to historical actual power data to ensure that the recent generation power characteristics can be presented by the prediction data used on site; (3) rechecking the current-month wind power assessment data; (4) checking the data accuracy of the previous day, and performing data supplementary report analysis; (5) the fault problems are summarized and analyzed, the fault elimination level of field personnel is improved, and the fault time is shortened.
Fig. 2 is a diagram of an optimized wind power prediction system according to an embodiment of the present invention, including: (1) the data acquisition server comprises a data acquisition unit, a data acquisition unit and a data processing unit, wherein the data acquisition unit is used for operating data acquisition software, communicating with the wind power station side wind power comprehensive communication management terminal and acquiring data of a fan, a wind measuring tower, wind power station power, numerical weather forecast and local wind power prediction result data of the wind power station; (2) a database server: the data processing system is used for processing, statistical analysis and storage of data, and in order to ensure reliable storage of the data, the database server is configured with a disk array; (3) the application workstation comprises: the system comprises PC workstation equipment for completing the functions of modeling, graph generation and display, report making and printing of the system; (4) wind power prediction server: the method comprises the steps that an entity server is used for operating a wind power prediction module, a neural network comprehensive algorithm based on a weighted least square support vector machine and quantum particle swarm prediction is used based on the collection or the provision of a numerical weather forecast provided by an SCADA system, and the short-term prediction and the ultra-short-term prediction are carried out on the output conditions of a single fan and the whole wind power plant by combining the real-time operation working condition of the fan of the wind power plant at present; (5) a data interface server: obtaining a numerical weather forecast using an entity server; (6) the reverse physical isolation equipment comprises a network switch, a network communication accessory, a physical isolation device, a cabinet and an accessory, is used for ensuring the network safety, is arranged at a network boundary, and simultaneously transmits the result of wind power prediction to a data acquisition and monitoring control System (SCADA) and an Energy Management System (EMS); (7) the entity anemometer tower is used for measuring and implementing meteorological data so as to predict ultra-short-term power, the entity anemometer tower is installed according to actual physical conditions of a wind field so as to guarantee the accuracy of prediction, the entity anemometer tower is installed in the range of 5km of the wind field, real-time meteorological data of the wind tower are collected through GPRS or optical fibers, and the height of the wind meter tower is not lower than the height of a fan hub. The wind measuring tower is matched with hardware equipment which comprises a meteorological data sensor, data acquisition equipment and data transmission equipment, wherein the wind measuring tower comprises 3 wind speed sensors, 2 wind direction sensors, 1 atmospheric temperature sensor, 1 atmospheric humidity sensor, 1 atmospheric pressure sensor, a data acquisition unit and data transmission equipment, and the atmospheric temperature sensor and the atmospheric pressure sensor are arranged at the height of 8 meters; a wind direction sensor is arranged at the height of 10 meters, and one or more wind direction sensors are arranged at other installation heights; the wind speed sensors are respectively arranged at 10 meters and 30 meters; (8) and the curve display module is used for displaying the result of the wind power prediction by adopting the curve display module, the curve comprises a prediction curve and an actual curve, the current predicted power and the actual power of each wind field are displayed in a list form, various selections can be carried out through conditions such as a date control, a pull-down list of the wind speed layer height and the like, and the conditions are stored locally for subsequent query.
FIG. 3 is a flow chart of a method for reducing wind power assessment electric quantity, which includes: (1) analyzing the wind power assessed reason: the method comprises weather forecast analysis, wind power prediction analysis, scheduling analysis, data reporting analysis and manual fault judgment analysis; (2) generating a strategy for reducing wind power assessment: and obtaining a strategy for reducing the wind power assessment electric quantity according to the analyzed reason.
FIG. 4 is a flow chart of a prediction method for optimizing wind power according to an embodiment of the invention. The method is just a problem of the existing wind power prediction system, needs to search the reason that the accuracy of the wind power prediction system does not meet the power grid assessment requirement, grasps the prediction rule, adjusts the reported prediction output, analyzes the assessed specific data, provides a non-examination application for the assessment of non-wind field reasons, and reduces monthly assessment cost by using the existing conditions, thereby providing the wind power prediction method.
The mathematical concept and physical principle related to the wind power prediction method are explained in detail as follows. The least square support vector machine model is an important result of statistics, the training process of the least square support vector machine follows the principle of minimizing the structural risk,the method changes inequality constraint of a vector machine into equality constraint, changes empirical risk from primary power of deviation into quadratic power, and converts the problem of solving quadratic programming into the problem of solving linear equations, thereby avoiding insensitive loss function, greatly reducing calculation complexity, and having higher operation speed than a common support vector machine, thereby greatly facilitating the solution of Lagrange multiplier alpha, wherein the original problem is QP problem, and the LSSVM problem is a problem of solving linear equations. The least squares support vector machine algorithm is described as: for a given training set, the training set is,the linear equation in the feature space is defined as: y isi=wTφ(xi)+bi1,2,. l; the regression problem can be expressed as:s.t.yi=ωTφ(xi)+b+ξi1,2,. l, wherein ξiE.R is an error, C is larger than 0 and is a punishment coefficient, and the error is adjusted.
In wind power system prediction, the time-varying system has a large correlation with data near an operating point when in operation, but has a small correlation with data in a region far away from the operating point. With the change of the system and the collection of new data, the model established by the previous off-line data sample cannot accurately describe the actual condition of the system, and in order to enable the model to accurately reflect the current state of the wind power system, the obtained new data is continuously utilized to establish a new model capable of reflecting the current condition of the system.
The least square support vector machine improves a standard support vector machine model, but loses robustness, and in order to treat different wind power time-varying data differently, the least square support vector machine is considered to be subjected to weighting operation, and the operation comprises the following specific steps: (1) given a set of training data { xk,ykN, finding out optimal parameters, and calculating e according to the optimal parametersk=αkGamma; (2) according to the error ekComputing robust estimates of the distributionValue of(3) By ekAnddetermining the corresponding weight vk(ii) a (4) Liberation of a*And b*Giving the final nonlinear prediction model:
in the quantum particle swarm optimization, the particles have quantum behaviors, and the searching capability is far better than that of the traditional particle swarm optimization. In order to ensure convergence of the quantum-behaved particle swarm optimization, each particle must converge to a respective p-point, where p ═ p (p ═ p)1,p2,...pd),pdIs the value of the particle in the d-dimension.Wherein:the position of the particle is found by the following equation:wherein u is a random number distributed between 0 and 1. A global point, mbest, is introduced to calculate the variable for the next iteration of the particle, which defines the average of the local best positions of all ions:and mbest ═ i (mbest)1,mbest2,...,mbestD),Where β is a contraction and expansion factor, which can be adjusted to control the rate of convergence, β is 0.5+0.5 (t)max-t)/tmax,tmaxIs the maximum number of iterations, M is the size of the population of particles, and finallyThe position of the particle can be written as:
referring to fig. 5, the neural network prediction based on the weighted least squares support vector machine and the quantum-behaved particle swarm algorithm includes: (1) establishing a neural network model to predict meteorological data; (2) according to the obtained meteorological prediction data, a double neural network model for predicting wind power data is reconstructed; (3) and determining the relation between the correlation point and the wind speed time curve of the wind power plant, and predicting the wind power in the next hours. The establishment of the dual neural network model under the weighted least square support vector machine and the quantum particle swarm algorithm comprises three key factors: (1) how to replace the hyper-parameters with the position of the particle, i.e. how to encode, (2) how to define an adaptation function to evaluate the advantages of the particle; (3) how to perform the double neural network modeling. The method comprises the following steps:
(2-1) encoding the hyper-parameters, wherein the encoding hyper-parameters comprise regularization parameters and kernel parameters, each particle is replaced by a potential solution to form a hyper-parameter group, and optimal parameters are selected;
(2-2) establishing an adaptive function, and evaluating the generalization performance, wherein the adaptive function used in the embodiment is as follows:
the fixed is 1/RMSE (γ, σ), where RMSE (γ, σ) is the root mean square error of the prediction, which varies as the least squares support vector machine parameter pair (γ, σ) varies. When the termination iteration criterion is satisfied, the maximum adaptation function corresponds to the optimal parameters of the least squares support vector machine.
The iteration termination criterion of the algorithm comprises two modes: the first is that the algorithm stops when the value of the objective function is less than or equal to a given threshold epsilon; the second is to give an iteration number in advance and stop the iteration when this value is reached.
(2-3) after the iteration is finished, training the dual neural network model, training the RNN, and using a back propagation algorithm similar to the common neural network, because the parameter at each moment is shared, the gradient of the parameter does not only depend on the output at the current moment, but also depends on the previous moment. For example, in order to calculate the gradient of the wind farm t-4 in prediction, we need to forward the error 3 times.
(2-4) extension of the dual neural network model: the idea of bidirectional expansion is that the output at the time t depends not only on the previous elements but also on the following elements, and the output depends on the hidden states of the two neural network models, so that the two neural network models are subjected to bidirectional expansion.
The background provided by the embodiment is based on the Pinglu horizontal longhole wind power plant which is put into operation at 12 and 20 days in 2015, the wind power prediction and examination period in 2016 is in a 7-month assessment period, and is influenced by factors such as maturity of a wind power prediction system technology and wind resources, the wind power prediction and examination of the wind power plant in each month are basically in an assessed state, the nuclear power amount in 7-month assessment in 2016 is 28.19MWH, the nuclear power amount in 8-month assessment in 2016 is 129.16MWH, the nuclear power amount in 9-month assessment in 2016 is 13.91MWH, the nuclear power amount in 10-month assessment in 2016 is 6.35MWH, and the total assessment electric quantity in 2016 is 177.61 MWH. Therefore, a prediction model is established by adopting the load sample and meteorological data of the wind power plant 2017 in spring. The sampling frequency of the wind power data provided by the SCADA system is 12 seconds per point, and the wind power prediction system shown in the attached figure 2 is adopted to predict the wind power points of the wind power plant. Preprocessing historical data, including processing missing data and dirty data, quantifying meteorological conditions (such as rainfall, maximum temperature, minimum temperature, temperature and the like), and finally normalizing the meteorological conditions and the meteorological conditions to form a characteristic vector of a certain day, inputting the characteristic vector into a dual neural network prediction model for training and prediction, wherein the normalization formula is as follows:
wherein L represents the normalized quantity, LiRepresenting the actual value, Lmax,LminRepresenting the actual maximum and minimum values, respectively; [ a, b ]]Indicating a normalized arrival interval.
The prediction results are compared and analyzed through common prediction errors, namely, the root mean square error, the average absolute error and the average absolute percentage error, the wind power of 0.25 hour, 1 hour and 2 hours in the future is predicted, the root mean square error is 2.410, 5.481 and 7.807 respectively, and the average absolute error is 1.018, 3.814 and 4.936 respectively; the mean absolute percentage errors were 15.17, 26, 35 and 28.17, respectively.
Therefore, the system and the method greatly improve the prediction precision, reduce the root mean square error of the prediction, reduce the average absolute error and reduce the average relative error, are actually applied to the ultra-short-term wind power prediction of a large wind field, and the comparative analysis of the prediction results proves that the method is effective, improves the system prediction precision, reduces the wind power supplier loss caused by the assessment of a power grid company on the power prediction precision of the wind field, provides a non-consideration application on the assessment of non-wind field reasons, and reduces the monthly assessment cost by utilizing the existing conditions.
While the present invention has been described with reference to the particular illustrative embodiments, it is not to be restricted by the embodiments but only by the appended claims. It will be understood by those skilled in the art that variations and modifications of the embodiments of the present invention can be made without departing from the scope and spirit of the invention.

Claims (9)

1. A system for reducing wind power assessment electric quantity is characterized by comprising:
cause analysis system: the method comprises a weather forecast analysis system, a wind power prediction system, a scheduling analysis system, a data reporting analysis system and a manual fault judgment analysis system, wherein the analysis result comprises the following steps: the weather forecast is not transmitted successfully and the data is not updated timely; wind power prediction analysis system is imperfect, wind resource is unstable; scheduling system reasons including participation in peak shaving, scheduling data network interruption, untimely data updating and incomplete uploading rate; the accuracy rate of the reported data cannot be counted before reporting; personnel judgment and low fault handling capacity;
the strategy generation system comprises: obtaining a strategy for reducing the wind power assessment electric quantity according to the analysis result of the reason analysis system, wherein the strategy comprises the following steps: the operators on duty closely monitor, timely update and optimize the program of the system for reducing the wind power check electric quantity, and upgrade the system for reducing the wind power check electric quantity; optimizing a wind power prediction system, adjusting a weather forecast model, preferentially selecting multiple different meteorological sources according to historical actual measurement data of a wind measuring tower on site, and optimizing and adjusting the power prediction model according to historical actual power data to ensure that the prediction data used on site can reflect the recent power generation power characteristics; rechecking the current-month wind power assessment data; checking the data accuracy of the previous day, and performing data supplementary report analysis; fault problems are summarized and analyzed, the fault elimination level of field personnel is improved, and the fault time is shortened;
the wind power prediction system comprises the following components after being optimized:
the data acquisition server comprises a data acquisition unit, a data acquisition unit and a data processing unit, wherein the data acquisition unit is used for operating data acquisition software, communicating with the wind power station side wind power comprehensive communication management terminal and acquiring data of a fan, a wind measuring tower, wind power station power, numerical weather forecast and local wind power prediction result data of the wind power station;
a database server: the data processing system is used for processing, statistical analysis and storage of data, and in order to ensure reliable storage of the data, the database server is configured with a disk array;
the application workstation comprises: the system comprises PC workstation equipment for completing the functions of modeling, graph generation and display, report making and printing of the system;
wind power prediction server: the method comprises the steps that an entity server is used for operating a wind power prediction module, a neural network comprehensive algorithm based on a weighted least square support vector machine and quantum particle swarm prediction is used based on the collection or the provision of a numerical weather forecast provided by an SCADA system, and the short-term prediction and the ultra-short-term prediction are carried out on the output conditions of a single fan and the whole wind power plant by combining the real-time operation working condition of the fan of the wind power plant at present;
a data interface server: obtaining a numerical weather forecast using an entity server;
the reverse physical isolation equipment comprises a network switch, a network communication accessory, a physical isolation device, a cabinet and an accessory, is used for ensuring the network safety, is arranged at a network boundary, and simultaneously transmits the result of wind power prediction to a data acquisition and monitoring control System (SCADA) and an Energy Management System (EMS).
2. The system for reducing wind power assessment electric quantity according to claim 1, characterized in that the optimized wind power prediction system further comprises a wind measuring tower and a curve display module, the wind measuring tower is used for measuring and implementing meteorological data so as to perform ultra-short-term power prediction, the wind measuring tower is installed according to the actual physical conditions of a wind field to ensure the accuracy of prediction, the wind measuring tower is installed in the range of 5km of the wind field, the real-time meteorological data of the wind measuring tower is collected through GPRS or optical fibers, the height of the wind measuring tower is not lower than the height of a fan hub, the supporting hardware equipment of the wind measuring tower comprises a meteorological data sensor, data collection equipment and data transmission equipment, the meteorological data sensor comprises 3 wind speed sensors, 2 wind direction sensors, 1 atmospheric temperature sensor, 1 atmospheric humidity sensor and 1 atmospheric pressure sensor, the atmospheric temperature sensor and the atmospheric pressure sensor are arranged at the height of 8 meters; a wind direction sensor is arranged at the height of 10 meters, and another wind direction sensor is arranged at the other installation height; the method comprises the steps that one wind speed sensor is respectively installed at 10 meters and 30 meters, the result of wind power prediction is displayed by adopting a curve display module, the curve comprises a prediction curve and an actual curve, the current predicted power and the actual power of each wind field are displayed in a list form, multiple selections are carried out through a date control and pull-down list conditions of wind speed layer heights, and the pull-down list conditions are stored locally for subsequent query.
3. A method for reducing wind power assessment electric quantity is applied to the system for reducing wind power assessment electric quantity according to any one of claims 1-2, and is characterized by comprising the following steps:
analyzing the wind power assessed reason: the method comprises weather forecast analysis, wind power prediction analysis, scheduling analysis, data reporting analysis and manual fault judgment analysis;
generating a strategy for reducing wind power assessment: and obtaining a strategy for reducing the wind power assessment electric quantity according to the analyzed reason.
4. The method for reducing wind power assessment electric quantity according to claim 3, wherein the reason for wind power assessment analysis includes: the weather forecast is not transmitted successfully and the data is not updated timely; wind power prediction analysis system is imperfect, wind resource is unstable; scheduling system reasons including participation in peak shaving, scheduling data network interruption, untimely data updating and incomplete uploading rate; the accuracy rate of the reported data cannot be counted before reporting; and the personnel judgment and fault processing capacity is low.
5. The method for reducing the wind power assessment electric quantity according to claim 3, wherein the strategy comprises: the operators on duty closely monitor, timely update and optimize the program of the system for reducing the wind power check electric quantity, and upgrade the system for reducing the wind power check electric quantity; optimizing a wind power prediction system, adjusting a weather forecast model, preferentially selecting multiple different meteorological sources according to historical actual measurement data of a wind measuring tower on site, and optimizing and adjusting the power prediction model according to historical actual power data to ensure that the prediction data used on site can reflect the recent power generation power characteristics; rechecking the current-month wind power assessment data; checking the data accuracy of the previous day, and performing data supplementary report analysis; the fault problems are summarized and analyzed, the fault elimination level of field personnel is improved, and the fault time is shortened.
6. The method for reducing the wind power assessment electric quantity according to claim 5, wherein the optimizing the wind power prediction system comprises adopting a comprehensive algorithm based on neural network prediction combined with a weighted least squares support vector machine and a quantum particle swarm algorithm in the system, and comprises the following steps:
(1) establishing a neural network model to predict meteorological data;
(2) establishing a double neural network model for predicting wind power data according to the obtained meteorological prediction data;
(3) and determining the relation between the correlation point and the wind speed time curve of the wind power plant, and predicting the wind power in the next hours.
7. The method for reducing the wind power assessment electric quantity according to claim 6, wherein the step (1) comprises: the historical data is preprocessed, namely missing data and dirty data are processed, then meteorological conditions are quantized, finally normalized, and feature vectors forming a certain day are input into a double neural network prediction model for training and prediction.
8. The method for reducing the wind power assessment electric quantity according to claim 6, wherein the step (2) comprises:
(2-1) encoding the hyper-parameters, wherein the encoding hyper-parameters comprise regularization parameters and kernel parameters, each particle is replaced by a potential solution to form a hyper-parameter group, and optimal parameters are selected;
(2-2) establishing an adaptive function, and evaluating the generalization performance;
(2-3) after iteration is finished, training a double neural network model, training an RNN (neural network) to be similar to a common neural network, and using a back propagation algorithm, wherein because the parameter at each moment is shared, the gradient of the parameter does not only depend on the output at the current moment but also depends on the previous moment;
(2-4) extension of the dual neural network model: the idea of bidirectional expansion is that the output at the time t depends not only on the previous elements but also on the following elements, and the output depends on the hidden states of the two neural network models, so that the two neural network models are subjected to bidirectional expansion.
9. The method for reducing the wind power assessment electric quantity according to claim 8, wherein the adaptive function in the step (2-2) is: the fixed is 1/RMSE (gamma, sigma), wherein RMSE (gamma, sigma) is the root mean square error of the prediction result and changes along with the change of the least square support vector machine parameter pair (gamma, sigma), when the termination iteration criterion is satisfied, the maximum adaptive function corresponds to the optimum parameter of the least square support vector machine, and the termination iteration criterion of the comprehensive algorithm comprises two modes: the first is that the algorithm stops when the value of the objective function is less than or equal to a given threshold epsilon; the second is to give an iteration number in advance and stop the iteration when this value is reached.
CN201810118649.1A 2018-02-06 2018-02-06 System and method for reducing wind power assessment electric quantity Active CN108345996B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810118649.1A CN108345996B (en) 2018-02-06 2018-02-06 System and method for reducing wind power assessment electric quantity

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810118649.1A CN108345996B (en) 2018-02-06 2018-02-06 System and method for reducing wind power assessment electric quantity

Publications (2)

Publication Number Publication Date
CN108345996A CN108345996A (en) 2018-07-31
CN108345996B true CN108345996B (en) 2021-07-20

Family

ID=62960001

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810118649.1A Active CN108345996B (en) 2018-02-06 2018-02-06 System and method for reducing wind power assessment electric quantity

Country Status (1)

Country Link
CN (1) CN108345996B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109242223B (en) * 2018-11-26 2021-09-21 武汉理工光科股份有限公司 Quantum support vector machine evaluation and prediction method for urban public building fire risk
CN111079980B (en) * 2019-11-22 2021-06-29 天合云能源互联网技术(杭州)有限公司 Optical power prediction method based on self-adaptive classification strategy and hybrid optimization SVR

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102736593A (en) * 2012-06-05 2012-10-17 吴光军 Integrated platform system for remote management and control of wind power field cluster
CN102945508A (en) * 2012-10-15 2013-02-27 风脉(武汉)可再生能源技术有限责任公司 Model correction based wind power forecasting system and method
CN103617454A (en) * 2013-11-21 2014-03-05 中能电力科技开发有限公司 Wind power plant power forecast method according to numerical weather forecasts
CN105279582A (en) * 2015-11-20 2016-01-27 中国水利水电第十四工程局有限公司 An ultra-short-term wind electricity power prediction method based on dynamic correlation characteristics
CN105552970A (en) * 2016-02-25 2016-05-04 华北电力科学研究院有限责任公司 Method and apparatus for improving accuracy of predicting power of wind power station
CN106157165A (en) * 2015-04-09 2016-11-23 华电电力科学研究院 A kind of based on laser radar anemometer Wind turbines power curve examination appraisal procedure
CN106326529A (en) * 2016-08-09 2017-01-11 广东电网有限责任公司电力科学研究院 System and method for assessing contribution electric quantity of primary frequency modulation of hydroelectric generating set
CN107480833A (en) * 2017-09-05 2017-12-15 清华大学 A kind of wind-powered electricity generation electricity generation system peak modulation capacity appraisal procedure

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080228553A1 (en) * 2007-03-12 2008-09-18 Airtricity Holdings Limited Method And System For Determination Of An Appropriate Strategy For Supply Of Renewal Energy Onto A Power Grid
WO2016186694A1 (en) * 2015-05-15 2016-11-24 General Electric Company Condition-based validation of performance updates

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102736593A (en) * 2012-06-05 2012-10-17 吴光军 Integrated platform system for remote management and control of wind power field cluster
CN102945508A (en) * 2012-10-15 2013-02-27 风脉(武汉)可再生能源技术有限责任公司 Model correction based wind power forecasting system and method
CN103617454A (en) * 2013-11-21 2014-03-05 中能电力科技开发有限公司 Wind power plant power forecast method according to numerical weather forecasts
CN106157165A (en) * 2015-04-09 2016-11-23 华电电力科学研究院 A kind of based on laser radar anemometer Wind turbines power curve examination appraisal procedure
CN105279582A (en) * 2015-11-20 2016-01-27 中国水利水电第十四工程局有限公司 An ultra-short-term wind electricity power prediction method based on dynamic correlation characteristics
CN105552970A (en) * 2016-02-25 2016-05-04 华北电力科学研究院有限责任公司 Method and apparatus for improving accuracy of predicting power of wind power station
CN106326529A (en) * 2016-08-09 2017-01-11 广东电网有限责任公司电力科学研究院 System and method for assessing contribution electric quantity of primary frequency modulation of hydroelectric generating set
CN107480833A (en) * 2017-09-05 2017-12-15 清华大学 A kind of wind-powered electricity generation electricity generation system peak modulation capacity appraisal procedure

Also Published As

Publication number Publication date
CN108345996A (en) 2018-07-31

Similar Documents

Publication Publication Date Title
Dong et al. Wind power day-ahead prediction with cluster analysis of NWP
CN105811402B (en) A kind of Electric Load Prediction System and its Forecasting Methodology
CN102411729B (en) Wind power prediction method based on adaptive linear logic network
CN108345996B (en) System and method for reducing wind power assessment electric quantity
CN106228278A (en) Photovoltaic power prognoses system
CN105868853A (en) Method for predicting short-term wind power combination probability
CN108388962B (en) Wind power prediction system and method
WO2021213192A1 (en) Load prediction method and load prediction system employing general distribution
CN105184388A (en) Non-linear regression method for urban power load short-period prediction
Xie et al. Short-term power load forecasting model based on fuzzy neural network using improved decision tree
CN103199616B (en) Large regional grid wind-powered electricity generation group service data monitoring analysis system
CN111738854A (en) New energy spot transaction decision-making data service system based on cloud computing
CN112085285A (en) Bus load prediction method and device, computer equipment and storage medium
CN111178585A (en) Fault reporting amount prediction method based on multi-algorithm model fusion
CN110766200A (en) Method for predicting generating power of wind turbine generator based on K-means mean clustering
CN110222882A (en) A kind of prediction technique and device of electric system Mid-long Term Load
CN109165770A (en) Novel photovoltaic power forecasting method based on AFSA-Elman
CN108960522A (en) A kind of photovoltaic power generation quantity prediction analysis method
Hou et al. A novel algorithm for multi-node load forecasting based on big data of distribution network
CN110175705B (en) Load prediction method and memory and system comprising same
Ding et al. A Statistical Upscaling Approach of Region Wind Power Forecasting Based on Combination Model
CN112949207A (en) Short-term load prediction method based on improved least square support vector machine
CN113608975A (en) Method and system for online evaluation of running state of electric energy metering system in electric power spot market
CN112348070A (en) Method and system for forecasting medium and short term loads of smart power grid
Kalkhorani Creating an Energy Model of an Entire University Campus-Part 1: Preliminary Assessment of Building Modeling Techniques

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant