CN111062516A - Fan output prediction method based on GMDH (Gaussian mixture distribution) multivariate processing - Google Patents
Fan output prediction method based on GMDH (Gaussian mixture distribution) multivariate processing Download PDFInfo
- Publication number
- CN111062516A CN111062516A CN201911137888.2A CN201911137888A CN111062516A CN 111062516 A CN111062516 A CN 111062516A CN 201911137888 A CN201911137888 A CN 201911137888A CN 111062516 A CN111062516 A CN 111062516A
- Authority
- CN
- China
- Prior art keywords
- data
- gmdh
- historical
- polynomial
- predicted
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000012545 processing Methods 0.000 title claims abstract description 14
- 239000000203 mixture Substances 0.000 title description 3
- 238000012549 training Methods 0.000 claims abstract description 28
- 238000003062 neural network model Methods 0.000 claims abstract description 11
- 230000005684 electric field Effects 0.000 claims abstract description 4
- 230000008569 process Effects 0.000 claims description 9
- 238000012360 testing method Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000005215 recombination Methods 0.000 claims description 6
- 230000006798 recombination Effects 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 claims description 5
- 230000004907 flux Effects 0.000 claims description 5
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Chemical compound O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 5
- 238000007689 inspection Methods 0.000 claims description 4
- 210000002569 neuron Anatomy 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000005265 energy consumption Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000003912 environmental pollution Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000010248 power generation Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION 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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a method for predicting output of a fan based on GMDH multivariable processing, which comprises the following steps: aiming at a wind electric field area to be predicted, acquiring historical meteorological data of the area, meteorological data of time to be predicted and historical fan output data of the wind power plant area; standardizing and centralizing the historical meteorological data, the meteorological data of the time to be predicted and the historical fan output data, and dividing the historical meteorological data, the meteorological data of the time to be predicted and the historical fan output data into a training set and a checking set; inputting the standardized and centralized data into a GMDH neural network model, performing learning training on the model, and verifying a training result; outputting an optimal weight factor by the GMDH network model after learning training, and determining GMDH network parameters; and predicting the wind power output of the region by using the GMDH network parameters. The method can predict the large-scale wind power more accurately.
Description
Technical Field
The invention relates to the technical field of artificial intelligence data processing, in particular to a method for predicting output of a fan based on GMDH (Gaussian mixture distribution) multivariate processing.
Background
With the rapid development of economy in China, the total energy consumption of society is continuously rising. The problem of environmental pollution caused by energy consumption is more and more prominent, the development of clean energy is paid much attention, and the vigorous development of renewable energy is an important component in the strategic layout of energy in China. The distributed clean energy such as wind power generation can greatly reduce environmental pollution while providing abundant electric energy, and effectively reduce the harm of fossil energy to the environment.
Wind energy has been popularized in many areas as a clean energy source, but due to uncertainty and randomness of wind power, the wind energy has obvious volatility and uncertainty. The characteristics of the system enable the system to face a plurality of new challenges in the aspects of stability, reliability and the like after wind energy grid connection, and one of key technologies of wind energy grid connection and consumption is accurate control and accurate prediction of fan output. Because meteorological factors such as wind power and the like have great randomness, the power generation of the fan is influenced by various uncertain factors such as wind speed, air pressure, rainfall, altitude and the like, and the influence factors have the characteristics of complexity and diversification. Therefore, the prediction of the fan output is a complex nonlinear function solving process, and the accurate analysis of the output rule of the fan under various complex variable environments is the key for realizing the accurate prediction of the fan output. However, due to the huge data scale and the complex nonlinear relationship, the general algorithm cannot quickly fit the relationship between the multivariate factors and the fan output, so that various external influence factors of the fan output cannot be covered, and the complex rule of the fan output is difficult to comprehensively grasp.
GMDH is a neural network with multilayer neuron selection and weight adjustment functions, and can fit complex nonlinear laws. Meanwhile, the input variables can be effectively selected, and the sparsity of input data is endowed, so that the complicated input variables are subjected to general selection and fitting. The characteristics of the GMDH enable the GMDH to have the effect in large-scale fan output prediction, but the technical application precedent of the GMDH technology in the field is not existed at present.
Disclosure of Invention
In view of this, the embodiment of the present invention provides a method for predicting output of a wind turbine based on GMDH multivariate processing, which solves the above problems in a more reliable manner and can achieve more accurate prediction of large-scale wind power.
In order to achieve the above object, an embodiment of the present invention provides a method for predicting a fan output based on GMDH multivariate processing, including the following steps:
(1) aiming at a wind electric field area to be predicted, acquiring historical meteorological data of the area, meteorological data of time to be predicted and historical fan output data of the wind power plant area;
(2) standardizing and centralizing the historical meteorological data, the meteorological data of the time to be predicted and the historical fan output data, and dividing the historical meteorological data, the meteorological data of the time to be predicted and the historical fan output data into a training set and a checking set;
(3) inputting the standardized and centralized data into a GMDH neural network model, learning and training the model, and evaluating a training result;
(4) outputting an optimal weight factor by the GMDH network model after learning training, and determining GMDH network parameters;
(5) and predicting the wind power output of the region by using the GMDH network parameters.
Specifically, the acquiring historical meteorological data of the region, meteorological data of time to be predicted and historical fan output data of the wind farm region includes:
the required historical meteorological data and the meteorological data of the time to be predicted comprise: the method comprises the following steps that 14 meteorological elements including maximum instantaneous wind speed, ten-tenth wind speed, maximum wind speed, u component of wind, v component of wind, air density, rainfall, maximum rainfall, air temperature, dew point temperature difference, relative humidity, air pressure, water vapor flux and false phase temperature are included, data of each meteorological element is called a characteristic variable, and fan output data corresponding to time is called an observed value;
the time range of the historical meteorological data and the historical fan output data of the area is from the latest day as a starting point to more than 30 days before the latest day; the latest date is the latest date on which the complete data is currently available, i.e., the day before the day to be predicted is the latest date on which the complete data is available.
All data resolutions are time-by-time data sets of the whole-point moment, and the time points of the meteorological data and the fan output data correspond to each other one by one.
Further, the normalizing and centralizing the data, and the dividing into a training set and a test set comprises: the data normalization and centralization process uses the following formula:
wherein x is original data, x' is data after standardization and centralization processing, mu is the average value of the data, and sigma is the standard deviation of the data; and dividing 75% of historical meteorological data and historical fan output data into a training set, and dividing 25% into a checking set.
Further, the inputting data into the GMDH neural network model, performing learning training on the model, and verifying the training result includes the following steps:
(41) inputting the historical meteorological data, the characteristic variable group corresponding to the meteorological data of the time to be predicted and the observed value of the fan output data of the corresponding time into an input layer of the GMDH neural network model;
(42) inputting data of the model input layer into the network intermediate layer, substituting the data into the K-G polynomial, carrying out evolution through the K-G polynomial, and solving a K-G polynomial coefficient a:
where x is the input characteristic variable, ai,aij,aijk… is the K-G polynomial coefficient a,
is the observed value, i.e. the output of each neuron;
(43) K-G polynomial coefficient a (a) output by the layeri,aij,aijk…) into a K-G polynomial,checking by using the checking set data;
(44) calculating the data of the inspection set through an external criterion, and outputting an external criterion judgment result w;
(45) sorting w in descending order, taking the first m characteristic variables, inputting the variables into the next layer of neural network, and simultaneously recording wmin，
(46) Repeating the step (42) to the step (45);
(47) when w isminAnd if the network expansion is not reduced, the network expansion is finished.
Further, the method for determining the GMDH network parameters by outputting the optimal weight factor to the GMDH network model after learning training includes the following features: the judgment result of the last layer external criterion of the output network is wminSubstituting the polynomial coefficient a into the K-G polynomial and outputting the optimal model.
Further, the prediction of the wind power output of the region by using the GMDH network parameters includes the following features: and substituting the output optimal model into the characteristic variable elements of the day to be predicted to predict the output of the fan.
Further, the inputting of the feature variable groups and the corresponding observation values into the input layer of the GMDH neural network model in the step (41) includes the following features: carrying out random pairwise cross recombination on the m characteristic variables, generating K characteristic variable groups after recombination, wherein the quantity of the recombined characteristic variable groups is as follows:
inputting the data of the model input layer to the network intermediate layer in the step (42), substituting the data into the K-G polynomial, and evolving through the K-G polynomial to obtain a coefficient a of the K-G polynomial, wherein the coefficient a comprises the following characteristics: because the input layer data is a characteristic variable group, and each characteristic variable group comprises two characteristic variables, the K-G polynomial uses a second-order K-G polynomial, which is specifically as follows:
wherein x isin,xjnFor inputting two of characteristic variable groupsA characteristic variable, a0、a1、a2、a3、a4、a5Is the coefficient a of the K-G polynomial,the observed value is obtained.
The step (44) of calculating the inspection set data by the external criterion and outputting the external criterion judgment result w includes the following features: the expression of the external criterion is shown in the following formula, wherein, ynIn order to be able to take the value of the observation,
is ynCorresponding estimated value, w is the calculation result of the external criterion:
compared with the prior art, the method can realize more accurate prediction on the power of the large-scale fan, and has the main advantages that:
1. the method can cover more external characteristic variables, effectively fit various external factors influencing the output of the fan, and process complex and huge input data, thereby realizing more accurate prediction of the output of the fan.
2. The invention can output results quickly and efficiently while processing large-scale data, and avoids the problem of slow model operation caused by huge data.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
FIG. 1 is a flow chart of a method for predicting output of a wind turbine based on GMDH multivariate processing according to an embodiment of the present invention;
FIG. 2 shows the actual output power and the predicted output power of five wind fields using the prediction method of the present invention;
FIG. 3 is an error rate and average absolute percent error for actual and predicted forces using the prediction method of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In the embodiment, a certain wind power plant in Gansu is taken as a case implementation object, five wind power plants are involved, and the total installed capacity is 1220 MW. And selecting a data set with the resolution of 15min from the time of 1 day of 4 months in 2018 to 1 day of 5 months in 2018.
The factors influencing the output of the fan are divided into two types, one type is the fan attribute factor, and the other type comprises the rated power, the cut-in wind speed, the rated wind speed, the cut-out wind speed and the like of the fan. The factors have obvious influence on the output characteristics of the fan, but the factors do not change along with the time, and are inherent and unchangeable attributes of the fan, so that the attribute factors of the fan are not considered in prediction; the second type is external factors, which mainly consist of the data content of the data set, and comprises the following steps: the maximum instantaneous wind speed, the tenth wind speed, the maximum wind speed, the u component of the wind, the v component of the wind, the air density, the rainfall, the maximum rain intensity, the air temperature, the dew point temperature difference, the relative humidity, the air pressure, the water vapor flux and the false equivalent temperature are 14 meteorological elements, namely 14 characteristic variables. And fan output data, namely an observed value, corresponding to time.
GMDH is a neural network with multilayer neuron selection and weight adjustment functions, and can fit complex nonlinear laws. Meanwhile, the input variables can be effectively selected, and the sparsity of input data is endowed, so that the complicated input variables are subjected to general selection and fitting. These features allow the GMDH to be effective in large scale wind turbine output prediction.
The invention relates to a method for predicting output of a fan based on GMDH multivariable processing, which comprises the following steps:
(1) and aiming at the wind electric field area to be predicted, acquiring historical meteorological data of the area, meteorological data of time to be predicted and historical fan output data of the wind power plant area.
(11) The required historical meteorological data and the meteorological data of the time to be predicted comprise: the maximum instantaneous wind speed, the tenth wind speed, the maximum wind speed, the u component of the wind, the v component of the wind, the air density, the rainfall, the maximum rain intensity, the air temperature, the dew point temperature difference, the relative humidity, the air pressure, the water vapor flux and the false equivalent temperature are 14 meteorological elements, the data of each meteorological element is called a characteristic variable, and the fan output data corresponding to the time is called an observed value.
(12) The time range of the historical meteorological data and the historical fan output data of the region is a data set with the resolution of 15min from 1 day at 4 months and 1 day at 2018 to 1 day at 5 months and 1 day at 2018.
(13) All data resolutions are time-by-time data sets at the integral point moment, the time points of meteorological data and fan output data correspond to each other one by one, and the table 1 shows installed capacities of five wind farms.
TABLE 1 installed capacity of five wind farms
(2) Standardizing and centralizing the data, and dividing the data into a training set and a test set;
(21) the data normalization and centering process adopts the following method, wherein x is original data, x' is data after normalization and centering process, mu is the average value of the data, and sigma is the standard deviation of the data:
after the input data of the model is subjected to standardization and centralization, the original data are converted into data which have the mean value of 0 and the standard deviation of 1 and are subjected to standard normal distribution, and the internal rule of the original data is kept unchanged.
(22) And dividing 75% of historical meteorological data and historical fan output data into a training set, and dividing 25% into a checking set.
(3) Inputting data into a GMDH neural network model, learning and training the model, and evaluating a training result; calculating the training result by using an external criterion standard, wherein the calculation formula is as follows, and judging according to the calculation result, wherein the judgment standard is as follows: the judgment result of the last layer external criterion of the output network is wminSubstituting the polynomial coefficient a into the K-G polynomial and outputting an optimal model corresponding to the coefficient; the smaller the output result w of the outer criteria, the better.
(31) And inputting the historical meteorological data, the characteristic variable group corresponding to the meteorological data of the time to be predicted and the corresponding observed value into an input layer of the GMDH neural network model.
(311) Carrying out random pairwise cross recombination on the m characteristic variables, generating K characteristic variable groups after recombination, wherein the quantity of the recombined characteristic variable groups is as follows:
(32) inputting data of the model input layer into the network intermediate layer, substituting the data into the K-G polynomial, carrying out evolution through the K-G polynomial, and solving a K-G polynomial coefficient a:
wherein x isin,xjnTwo characteristic variables being input characteristic variable groups, a0、a1、a2、a3、a4、a5Is the coefficient a of the K-G polynomial,
is an observed value;
(33) K-G polynomial coefficient a (a) output by the layeri,aij,aijk…) substituting into the K-G polynomial, and testing by using the test set data;
(34) calculating the data of the inspection set through an external criterion, and outputting an external criterion judgment result w;
(341) the expression of the external criterion is shown in formula (4), wherein ynIn order to be able to take the value of the observation,
is ynCorresponding estimated value, w is the calculation result of the external criterion:
(35) sorting w in descending order, taking the first m characteristic variables, inputting the variables into the next layer of neural network, and simultaneously recording wmin；
(36) Repeating the step (32) -the step (35);
(37) when w isminAnd if the network expansion is not reduced, the network expansion is finished.
(4) Outputting an optimal weight factor by the GMDH network model after learning training, and determining GMDH network parameters;
(41) the judgment result of the last layer external criterion of the output network is wminSubstituting the polynomial coefficient a into the K-G polynomial and outputting the optimal model. The model is formed by specific parameters, namely model optimization, namely parameter optimization, and the optimal model is as follows: in the last layer of the network, the result of the external criterion judgment is wminThe parameters of the corresponding model, i.e. a at that time0、a1、a2、a3、a4、a5And (4) the coefficient. And the step of moving the optimal model comprises the steps (31) to (37), and the optimal model is obtained after the step is completed, specifically, a0、a1、a2、a3、a4、a5And (4) the coefficient.
(5) And predicting the wind power output of the region by using the GMDH network parameters.
(51) And substituting the output optimal model into the characteristic variable elements of the day to be predicted to predict the output of the fan. Although the prediction of the fan output is the core of the invention, the core of the prediction is the construction of a prediction model, and the prediction process is simple after the construction of the optimal model is finished, and only data is input to obtain the data; most of the steps are processes about the construction of an optimal model; the specific prediction steps are as follows:
(511) wherein the characteristic variables include: the maximum instantaneous wind speed, the tenth wind speed, the maximum wind speed, the u component of the wind, the v component of the wind, the air density, the rainfall, the maximum rain intensity, the air temperature, the dew point temperature difference, the relative humidity, the air pressure, the water vapor flux and the false equivalent temperature are 14 meteorological elements. The resolution is 1 hour, and the time range is more than one week before the day to be predicted;
(512) standardizing and centralizing the data by adopting the formula;
(513) and inputting the processed data into the optimal model to obtain the predicted fan output.
Through the steps, the predicted points with the number of 180 and the resolution of 15min, namely the predicted value of the fan output in the future 45 hours, are obtained. In order to verify the effect of the technical scheme, the result is measured by using two indexes, namely Mean Absolute Percentage Error (MAPE) and Mean Absolute Error (MAE), and the calculation formula is shown as the following formula, and the evaluation result is shown as table 2 and fig. 1.
TABLE 2 evaluation index
Compared with the prior art, the method can realize more accurate prediction on the power of the large-scale fan, and has the main advantages that:
1. the method can cover more external characteristic variables, effectively fit various external factors influencing the output of the fan, and process complex and huge input data, thereby realizing more accurate prediction of the output of the fan.
2. The invention can output results quickly and efficiently while processing large-scale data, and avoids the problem of slow model operation caused by huge data.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, the system or system embodiments, as they are substantially similar to the method embodiments, will be described more simply below, with reference to the preceding description of the embodiments. The system and the system embodiments described above are merely schematic, wherein the modules described as separate parts may or may not be physically separate, and some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative components and steps have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (9)
1. A method for predicting fan output based on GMDH multivariable processing is characterized by comprising the following steps:
(1) aiming at a wind electric field area to be predicted, acquiring historical meteorological data of the area, meteorological data of time to be predicted and historical fan output data of the wind power plant area;
(2) standardizing and centralizing the historical meteorological data, the meteorological data of the time to be predicted and the historical fan output data, and dividing the historical meteorological data, the meteorological data of the time to be predicted and the historical fan output data into a training set and a checking set;
(3) inputting the standardized and centralized data into a GMDH neural network model, learning and training the model, and evaluating a training result;
(4) outputting an optimal weight factor by the GMDH network model after learning training, and determining GMDH network parameters;
(5) and predicting the wind power output of the region by using the GMDH network parameters.
2. The method of claim 1, wherein obtaining historical meteorological data for the region, meteorological data for a time to be predicted, and historical fan output data for the wind farm region comprises:
the required historical meteorological data and the meteorological data of the time to be predicted comprise: the method comprises the following steps that 14 meteorological elements including maximum instantaneous wind speed, ten-tenth wind speed, maximum wind speed, u component of wind, v component of wind, air density, rainfall, maximum rainfall, air temperature, dew point temperature difference, relative humidity, air pressure, water vapor flux and false phase temperature are included, data of each meteorological element is called a characteristic variable, and fan output data corresponding to time is called an observed value;
the time range of the historical meteorological data and the historical fan output data of the area is from the latest day as a starting point to more than 30 days before the latest day; the latest date is the latest date on which the complete data is currently available, i.e., the day before the day to be predicted is the latest date on which the complete data is available.
All data resolutions are time-by-time data sets of the whole-point moment, and the time points of the meteorological data and the fan output data correspond to each other one by one.
3. The method of claim 2, wherein normalizing and centering the data into a training set and a test set comprises:
the data normalization and centralization process uses the following formula:
wherein x is original data, x' is data after standardization and centralization processing, mu is the average value of the data, and sigma is the standard deviation of the data.
And dividing 75% of historical meteorological data and historical fan output data into a training set, and dividing 25% into a checking set.
4. The method of claim 3, wherein inputting data into the GMDH neural network model, learning and training the model, and evaluating the training result comprises the following steps:
(41) inputting the historical meteorological data, the characteristic variable group corresponding to the meteorological data of the time to be predicted and the observed value of the fan output data of the corresponding time into an input layer of the GMDH neural network model;
(42) inputting data of the model input layer into the network intermediate layer, substituting the data into the K-G polynomial, carrying out evolution through the K-G polynomial, and solving a K-G polynomial coefficient a:
where x is the input characteristic variable, ai,aij,aijk… is the K-G polynomial coefficient a,
is the observed value, i.e. the output of each neuron;
(43) K-G polynomial coefficient a (a) output by the layeri,aij,aijk…) substituting into the K-G polynomial, and testing by using the test set data;
(44) calculating the data of the inspection set through an external criterion, and outputting an external criterion judgment result w;
(45) sorting w in descending order, taking the first m characteristic variables, inputting the variables into the next layer of neural network, and simultaneously recording wmin，
(46) Repeating the step (42) to the step (45);
(47) when w isminAnd if the network expansion is not reduced, the network expansion is finished.
5. The method according to claim 4, wherein the step of outputting the optimal weight factor to the GMDH network model after learning training to determine the GMDH network parameters comprises the following steps:
the judgment result of the last layer external criterion of the output network is wminSubstituting the polynomial coefficient a into the K-G polynomial and outputting the optimal model.
6. The method of claim 5, wherein the predicting the wind power output of the region using the GMDH network parameters comprises the following features:
and substituting the output optimal model into the characteristic variable elements of the day to be predicted to predict the output of the fan.
7. The method of claim 4, wherein the step (41) of inputting the characteristic variable groups and the corresponding observation values into the input layer of the GMDH neural network model comprises the following characteristics:
carrying out random pairwise cross recombination on the m characteristic variables, generating K characteristic variable groups after recombination, wherein the quantity of the recombined characteristic variable groups is as follows:
8. the method of claim 7, wherein the step (42) of inputting the data of the model input layer to the network intermediate layer, substituting the K-G polynomial, and evolving through the K-G polynomial to obtain the coefficient a of the K-G polynomial includes the following features:
because the input layer data is a characteristic variable group, and each characteristic variable group comprises two characteristic variables, the K-G polynomial uses a second-order K-G polynomial, which is specifically as follows:
wherein x isin,xjnTwo characteristic variables being input characteristic variable groups, a0、a1、a2、a3、a4、a5Is the coefficient a of the K-G polynomial,
the observed value is obtained.
9. The method of claim 8, wherein the step (44) of computing the test set data by the external criterion and outputting the external criterion judgment result w comprises the following features:
the expression of the external criterion is shown in the following formula, wherein, ynIn order to be able to take the value of the observation,
is ynCorresponding estimated value, w is the calculation result of the external criterion:
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911137888.2A CN111062516A (en) | 2019-11-20 | 2019-11-20 | Fan output prediction method based on GMDH (Gaussian mixture distribution) multivariate processing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911137888.2A CN111062516A (en) | 2019-11-20 | 2019-11-20 | Fan output prediction method based on GMDH (Gaussian mixture distribution) multivariate processing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111062516A true CN111062516A (en) | 2020-04-24 |
Family
ID=70298532
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911137888.2A Pending CN111062516A (en) | 2019-11-20 | 2019-11-20 | Fan output prediction method based on GMDH (Gaussian mixture distribution) multivariate processing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111062516A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112132260A (en) * | 2020-09-03 | 2020-12-25 | 深圳索信达数据技术有限公司 | Training method, calling method, device and storage medium of neural network model |
| CN112861262A (en) * | 2021-02-08 | 2021-05-28 | 北京理工大学 | Self-organizing data-driven modeling method for trajectory prediction of high-dynamic carrier |
| CN114240003A (en) * | 2022-02-23 | 2022-03-25 | 泰豪软件股份有限公司 | New energy output prediction method, system, storage medium and equipment |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106529713A (en) * | 2016-11-03 | 2017-03-22 | 华北电力大学（保定） | Grey GMDH network combination model-based wind speed prediction method and system |
| CN108229592A (en) * | 2018-03-27 | 2018-06-29 | 四川大学 | Outlier detection method and device based on GMDH neuroids |
| CN108734341A (en) * | 2018-04-27 | 2018-11-02 | 广东电网有限责任公司 | A kind of self-organizing Electric Load Forecasting reporting method based on time series arma modeling |
-
2019
- 2019-11-20 CN CN201911137888.2A patent/CN111062516A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106529713A (en) * | 2016-11-03 | 2017-03-22 | 华北电力大学（保定） | Grey GMDH network combination model-based wind speed prediction method and system |
| CN108229592A (en) * | 2018-03-27 | 2018-06-29 | 四川大学 | Outlier detection method and device based on GMDH neuroids |
| CN108734341A (en) * | 2018-04-27 | 2018-11-02 | 广东电网有限责任公司 | A kind of self-organizing Electric Load Forecasting reporting method based on time series arma modeling |
Non-Patent Citations (3)
| Title |
|---|
| SAIDA MAKHLOUFI等: "Wind Speed and Wind Power Forecasting Using Wavelet Denoising-GMDH Neural Network", 《 2017 5TH INTERNATIONAL CONFERENCE ON ELECTRICAL ENGINEERING - BOUMERDES (ICEE-B)》 * |
| WEI LIU等: "Short-Term Load Forecasting Based on Elastic Net Improved GMDH and Difference Degree Weighting Optimization", 《APPLIED SCIENCES》 * |
| 刘伟: "基于风险备用的主动型配电网优化调度策略研究", 《中国优秀硕士学位论文全文数据库 (工程科技Ⅱ辑)》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112132260A (en) * | 2020-09-03 | 2020-12-25 | 深圳索信达数据技术有限公司 | Training method, calling method, device and storage medium of neural network model |
| CN112861262A (en) * | 2021-02-08 | 2021-05-28 | 北京理工大学 | Self-organizing data-driven modeling method for trajectory prediction of high-dynamic carrier |
| CN114240003A (en) * | 2022-02-23 | 2022-03-25 | 泰豪软件股份有限公司 | New energy output prediction method, system, storage medium and equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108304623B (en) | Probability load flow online calculation method based on stack noise reduction automatic encoder | |
| Wang et al. | Decision tree based online stability assessment scheme for power systems with renewable generations | |
| CN110442833A (en) | One kind assessing Wind turbines health state evaluation method based on various dimensions SCADA data | |
| Hou et al. | Sparse oblique decision tree for power system security rules extraction and embedding | |
| CN111062516A (en) | Fan output prediction method based on GMDH (Gaussian mixture distribution) multivariate processing | |
| CN111237134A (en) | Offshore double-fed wind driven generator fault diagnosis method based on GRA-LSTM-stacking model | |
| CN113298297B (en) | Wind power output power prediction method based on isolated forest and WGAN network | |
| CN112580874A (en) | Short-term wind power prediction method based on random forest algorithm and TCN | |
| CN106650977A (en) | Short-term power prediction method used for newly-built wind farm | |
| CN110991737A (en) | Ultra-short-term wind power prediction method based on deep belief network | |
| CN111900713A (en) | Multi-scene power transmission network planning method considering load and wind power randomness under network source coordination | |
| Chen et al. | Research on wind power prediction method based on convolutional neural network and genetic algorithm | |
| Gupta et al. | Long term estimation of global horizontal irradiance using machine learning algorithms | |
| CN115860797A (en) | Electric quantity demand prediction method suitable for new electricity price reform situation | |
| CN117669960A (en) | New energy power prediction method based on multivariable meteorological factors | |
| CN115689055A (en) | Short-term solar irradiance prediction method and device | |
| Jayashankara et al. | A novel approach for short-term energy forecasting in smart buildings | |
| CN113449847B (en) | Offshore wind power rolling prediction method considering second-level time series wind speed change | |
| Li et al. | An intelligent method for wind power forecasting based on integrated power slope events prediction and wind speed forecasting | |
| Zhou et al. | A Hybrid Method for Ultrashort‐Term Wind Power Prediction considering Meteorological Features and Seasonal Information | |
| Huang et al. | A decomposition‐based multi‐time dimension long short‐term memory model for short‐term electric load forecasting | |
| CN116151799A (en) | BP neural network-based distribution line multi-working-condition fault rate rapid assessment method | |
| CN115577854A (en) | Quantile regression wind speed interval prediction method based on EEMD-RBF combination | |
| CN112232714B (en) | Deep learning-based risk assessment method for distribution network under incomplete structural parameters | |
| Schreiber et al. | Quantifying the influences on probabilistic wind power forecasts |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200424 |