CN106874557B - A kind of forecasting wind speed bearing calibration based on ratio distribution - Google Patents
A kind of forecasting wind speed bearing calibration based on ratio distribution Download PDFInfo
- Publication number
- CN106874557B CN106874557B CN201710028104.7A CN201710028104A CN106874557B CN 106874557 B CN106874557 B CN 106874557B CN 201710028104 A CN201710028104 A CN 201710028104A CN 106874557 B CN106874557 B CN 106874557B
- Authority
- CN
- China
- Prior art keywords
- wind speed
- prediction
- actual measurement
- value
- data
- 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
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Aerodynamic Tests, Hydrodynamic Tests, Wind Tunnels, And Water Tanks (AREA)
- Air Conditioning Control Device (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of forecasting wind speed bearing calibrations based on ratio distribution, comprising: carries out sustained wind velocity prediction using BP neural network method according to measured data first, obtains actual measurement wind speed and corresponding prediction of wind speed;Then, it calculates actual measurement prediction of wind speed overall distribution proportional jitter: making the overall distribution ratio chart of actual measurement wind speed and prediction of wind speed, calculate two kinds of wind speed in the overall distribution upper deviation;Correction multinomial is finally fitted according to actual measurement-prediction of wind speed scatter plot, the wind speed of prediction is substituted into multinomial model, corrects prediction of wind speed, effect of optimization under more each percentage determines optimal calibration model.The present invention is based on the methods for the correction prediction of wind speed that actual measurement wind speed and prediction of wind speed overall distribution ratio are established, and this method solve the low problems of BP neural network model prediction wind speed precision, are more nearly prediction of wind speed in overall distribution ratio with actual measurement wind speed.
Description
Technical field
BP neural network prediction of wind speed is used the present invention relates to a kind of, prediction wind is corrected based on wind speed overall distribution ratio
The method of speed, is particularly suitable in engineering structure Wind resistant analysis, the accurate precision for promoting prediction of wind speed.
Background technique
Wind load is an important factor for antenna structure, influences the electrical property and pointing accuracy of antenna.It
Line operating system needs to know that at least following one day wind speed carrys out planning.Forecasting wind speed as a kind of look-ahead, if
It can accomplish the prediction of degree of precision, important references will be made well for the follow-up operation planning of antenna.Forecasting wind speed
Purpose is to predict the wind speed of lower a period of time, because requirement of the different observation missions to antenna-point accuracy is different, thus
Substantially it can determine which observation mission subsequent time period can do substantially.It is accurately realized the short-term wind speed of environment near antenna
Prediction, has very important significance for the security and stability of planning and designing, the operation of antenna.
The main method of wind speed short-term forecast at present has: continuing method, time series method, Kalman filtering method, neural network
Method, combinatorial forecast etc..The average relative error of various methods is about 20%~40% or so.The reason of causing this error
Mainly there are following three aspects:
A. whether the wind speed characteristics of wind field are regular is followed;
B. the selection of prediction model;
C. the length of predicted time.
BP neural network is one of most representational neural network model, has in terms of forecasting wind speed and widely answers
With.After BP neural network prediction of wind speed, is compared using one day or several days measured value and predicted value, verify prediction
Effect.But in fact forecasting wind speed it is accurate whether and several factors have relationship, fluctuations in wind speed is exactly one vital
Factor.Often comparison prediction effect is good in the stable situation of wind speed variation, and prediction is compared in the case that fluctuations in wind speed is strong
Effect is with regard to not so good.Therefore prediction effect comparison in one day or several days is not able to verify that the effect of prediction technique, it should from wind
Prediction effect is compared in fast overall distribution ratio.
Summary of the invention
Based on the above issues, it is an object of the invention to solve the technical solution of realization the object of the invention at present to be,
A kind of method for creating correction prediction of wind speed established based on actual measurement wind speed and prediction of wind speed overall distribution ratio, the technology is very
Solve the problems, such as that BP neural network model prediction wind speed precision is low in big degree, make prediction of wind speed in overall distribution ratio with
Actual measurement wind speed is more nearly.
In view of the above-mentioned problems, the present invention carries out forecasting wind speed using BP neural network according to somewhere measured data first.So
Afterwards, the distribution proportion figure of actual measurement wind speed and prediction of wind speed is made, the equal of actual measurement two distribution curves of wind speed and prediction of wind speed is calculated
Root mean square deviation.Correction multinomial model is finally fitted, the wind speed predicted before each is substituted into multinomial model, is compared
The effect optimized under each multinomial model determines optimal calibration model.Technical solution process used by solving the above problems
It is as follows:
It is as follows that the present invention solves method and step used by its technical problem:
A kind of forecasting wind speed bearing calibration based on ratio distribution, comprising the following steps:
Step 1, air speed data collection is handled:
Wind tower is established, sensor acquisition and recording air speed data is set, it is flat to acquire actual measurement wind speed per hour for data screening processing
Mean value;
Step 2, using BP neural network method prediction of wind speed:
Actual measurement air speed data is normalized, determines network inputs output predicted value and hidden neuron number,
Training network, establishes prediction model, carries out forecasting wind speed, obtains prediction of wind speed value hourly;
Step 3, actual measurement prediction of wind speed overall distribution proportional jitter is calculated:
Using x-axis as wind speed size, y-axis is wind speed profile ratio, draws actual measurement wind speed and prediction of wind speed overall distribution ratio
Figure, and calculate the root-mean-square error r of actual measurement wind speed and prediction of wind speed overall distribution proportional curvea;
Step 4, fitting correction multinomial:
Using x-axis as prediction of wind speed, y-axis is actual measurement wind speed, draws prediction of wind speed-actual measurement wind speed scatter plot, and percentage is arranged
From 1 to 100, corresponding percentage points are determined, using these points of fitting of a polynomial, obtain 100 correction multinomial models;
Step 5, optimal calibration model is selected:
Prediction of wind speed value hourly in step 2 is substituted into 100 multinomial models that step 4 obtains respectively, is obtained
It is inclined in overall distribution ratio to calculate prediction of wind speed and actual measurement wind speed new after correcting again for prediction of wind speed value after correction
Difference, therefrom selects the smallest deviation, and corresponding multinomial model is optimal calibration model.
Further, in step 2, actual measurement air speed data is normalized, is calculate by the following formula to obtain:
In formula, d (t) is the data before network inputs normalization, and X (t) is the data after normalization, and min (d (t)) is
Air speed data minimum value, max (d (t)) are air speed data maximum value.
Further, the output predicted value of network carries out anti-normalization processing:
Y (t)=o (t) * (max (d (t))-min (d (t))+min (d (t))
In formula, Y (t) is that network exports the data after renormalization, and o (t) is the output of neural network.
Further, in the step 2, it is as follows to establish prediction model:
Wherein HjFor hidden layer output, f is general hidden layer excitation function, vijThe weight of input and hidden layer, xiIt is defeated for network
Enter, a1And ajFor threshold value, m is hidden nodes, o1For the output of network, wj1For the weight of hidden layer and output layer.
Further, the root-mean-square error r of actual measurement wind speed and prediction of wind speed distribution curve is calculateda, it is calculate by the following formula to obtain:
U is prediction of wind speed in formula, and v is actual measurement wind speed, and max (u, v) is the maximum value surveyed in wind speed and prediction of wind speed, ti
=(i-1) * 0.1, f (ti) it is actual measurement wind speed profile proportional curve, g (ti) it is prediction of wind speed distribution proportion curve.
Further, in the step 4, multinomial model is as follows:
B=p1an+p2an-1+p3an-2+p4an-3+p5an-4+......+pna1+pn+1
In formula, a is the item that prediction of wind speed needs to substitute into, and b is the prediction of wind speed value after correction, p1、p2、....pn、pn+1Point
Not Wei coefficient value, n be polynomial highest power.
Compared with prior art, the present invention having the following characteristics that
Because to change prediction effect in stable situation good for wind speed, in the case that fluctuations in wind speed is strong, prediction effect is not just
It is good, so prediction effect comparison in one day or several days is not able to verify that the effect of prediction technique.It therefore should be from wind speed entirety
Prediction effect is compared on distribution proportion.The present invention creates a kind of based on actual measurement wind speed and the foundation of prediction of wind speed overall distribution ratio
Correction prediction of wind speed method, this method largely solves that BP neural network model prediction wind speed precision is low to ask
Topic is more nearly prediction of wind speed in overall distribution ratio with actual measurement wind speed.This method can effectively improve prediction of wind speed
Precision has apparent advanced and reliability, thus can more accurately instruct the wind force proofing design of engineering structure.
Detailed description of the invention
Fig. 1 is a kind of correction prediction of wind speed method model Establishing process figure;
Fig. 2 is that the 24 hours prediction effect figures in certain following day are predicted using BP;
Wind speed overall distribution ratio chart before Fig. 3 is correction;
Fig. 4 is actual measurement-prediction scatter plot;
Fig. 5 is P% point matched curve figure;
Wind speed overall distribution ratio chart after Fig. 6 correction.
Specific embodiment
The present invention will be further described with reference to the accompanying drawings and embodiments.
Referring to Fig.1, the present invention depicts a kind of correction prediction of wind speed side for above-mentioned 5 steps of the explanation being more clear
The Establishing process figure of method model, as shown in Figure 1.Now this is introduced so that Qitai 2011-2012 surveys air speed data as an example
The embodiment of patent, specific as follows:
1) air speed data collection is handled:
Wind tower is established, sensor acquisition and recording air speed data is set, handles to acquire through data screening and surveys wind speed per hour
Average value;
The present embodiment with Qitai wind field when 1 day zero January in 2011 to 31 days 24 December in 2012 when every
Collected air speed data is studied within one minute.Measurement method is measured by establishing wind tower, at different height
Air velocity transducer is set to record the air speed data collected in each hour at different height, acquires average value, represents this
The wind speed of hour.
By data processing, the air speed data being located in each hour is acquired into average value, represents the wind speed of this hour.
Later, it selects, reject unqualified data, if 24 hours certain day air speed datas are imperfect, by this daylong air speed data
It deletes.Air speed data, which rounds up, retains one decimal place number.
2) BP neural network method prediction of wind speed is used:
Actual measurement air speed data is normalized, determines network inputs output, hidden neuron number, training net
Network establishes prediction model, carries out forecasting wind speed, obtains prediction of wind speed value hourly;
Data normalization processing:
Since neural network input layer neuron is generally Sigmoid shape, the input of whole network will be compressed in one
Within a lesser range, in order to improve training speed, sensitivity and effective saturation region for avoiding Sigmoid shape function,
Ask the value of input data between 0~1, air speed value is converted into the value in [0,1] section by BP neural network prediction technique, thus
It needs that data are normalized:
The output predicted value of same network will also carry out anti-normalization processing:
Y (t)=o (t) * (max (d (t))-min (d (t))+min (d (t))
D (t) is the data before network inputs normalization in formula, and X (t) is the data after normalization, and min (d (t)) is wind
Fast data minimum value, max (d (t)) are air speed data maximum value, and o (t) is the output of neural network, and the output of Y (t) network is counter to return
Data after one change;
The determination of neural network hidden neuron number:
It is more safe although this method is relatively time-consuming using the method for gradually increasing hidden layer neuron number.
Hidden neuron is gradually chosen from 3-13, comparison prediction effect.The preferably corresponding neuron number of prediction effect is optimal hidden layer
Neuron number.Since the air speed data rule in each month is different, the number of hidden neuron corresponding to prediction model is established
Also different, so all selecting optimal neuron number to every month.
Output layer input layer is chosen for 4 to BP neural network structure: input layer -- hidden layer (one layer) --, and output layer is chosen for
1.By 30 days 720 data training net networks, prediction model is established, then the gradually wind speed of progressive following 24 hours of prediction.
Prediction model is as follows:
Wherein HjFor hidden layer output, f is general hidden layer excitation function, vijThe weight of input and hidden layer, xiIt is defeated for network
Enter, a1And ajFor threshold value, m is hidden nodes, o1For the output of network, wj1For the weight of hidden layer and output layer.
Predict that the effect of some day is as shown in Figure 2 using BP neural network.2 years wind speed of persistence forecasting, for each actual measurement
Hourly average air speed value just has corresponding prediction of wind speed value.
3) calculation of wind speed overall distribution proportional jitter:
Using x-axis as wind speed size, y-axis is wind speed profile ratio, draws actual measurement wind speed and prediction of wind speed overall distribution ratio
Figure, and calculate the root-mean-square error r of actual measurement wind speed and prediction of wind speed distribution curvea:
U is prediction of wind speed in formula, and v is actual measurement wind speed, and max (u, v) is the maximum value surveyed in wind speed and prediction of wind speed, ti
=(i-1) * 0.1, f (ti) it is actual measurement wind speed profile proportional curve, g (ti) it is prediction of wind speed distribution proportion curve.
Actual measurement wind speed and prediction of wind speed overall distribution ratio chart are drawn, as shown in Figure 3.It can be seen that actual measurement wind speed and
Prediction of wind speed has biggish deviation on distribution proportion.Calculating learns that root-mean-square-deviation is 0.1069.
4) fitting correction multinomial:
Using x-axis as prediction of wind speed, y-axis is actual measurement wind speed, draws prediction of wind speed-actual measurement wind speed scatter plot, and percentage is arranged
From 1 to 100, determines and be divided into corresponding percentage points when 0.1 between abscissa, using these points of fitting of a polynomial, obtain more than 100
Item formula model, multinomial model are as follows:
B=p1an+p2an-1+p3an-2+p4an-3+p5an-4+......+pna1+pn+1
In formula, a is the item that prediction of wind speed needs to substitute into, and b is the prediction of wind speed value after correction, p1、p2、....pn、pn+1Point
Not Wei coefficient value, n be polynomial highest power.
Matched curve principle: one group of percentage points (x is giveni, yi), polynomial fitting makes:
It is minimum.
Actual measurement has obtained 2 years air speed datas as unit of hour, rejects to have obtained 17520 actual measurement wind by data
Fast data.Using 17520 predicted values are obtained after BP neural network prediction technique, each actual measurement air speed value is corresponding to it
Prediction of wind speed value, just have 17520 points, each pair of point answers a predicted value and a measured value.It is prediction wind with x-axis
Speed, y-axis are actual measurement wind speed, draw this 17520 points.Prediction-actual measurement wind speed scatter plot is as shown in Figure 4.
X-axis is interval with 0.1, i.e. x=0,0.1,0.2.......Abscissa is respectively the point group of 0,0.1,0.2......
At multiple set, abscissa is consistent for each set.Each set, ordinate arrange from small to large, and fetch bit is in P%
Point, i.e., the ordinate value for being located at P% this point for each set is greater than the point of corresponding set the inside P%.It is true in this way
Multiple percentage points are determined, these points are fitted using multinomial, obtain the coefficient value of polynomial fitting.
P is set from 1 to 100, seven order polynomials are gone out using the cftool Function Fitting in MATLAB software, one of them hundred
Divide more as shown in Figure 5 than fit solution.100 correction multinomial models are thus obtained.
5) optimal calibration model is selected:
Prediction of wind speed value hourly in step 2) is substituted into 100 multinomial models that step 4) obtains respectively, is obtained
Prediction of wind speed value after to correction calculates prediction of wind speed and actual measurement wind speed new after correcting in overall distribution ratio again
Deviation, therefrom selects the smallest deviation, and corresponding multinomial model is just optimal calibration model.
The air speed value of each prediction is substituted into correction multinomial model, has obtained new prediction of wind speed value.Calculate correction
The deviation of new prediction of wind speed and actual measurement wind speed in overall distribution ratio afterwards.Prediction of wind speed and actual measurement wind speed are learnt by calculating
Minimum deflection in overall distribution ratio is 0.0535, and corresponding percentage is 43%.Use optimal calibration model will in this way
The distribution proportion deviation of the two before is reduced to 0.0535 from 0.1069, reduces 49.9%.By optimal calibration model meter
After calculation, the prediction of wind speed and actual measurement wind speed overall distribution ratio after correction are as shown in Figure 6.It is effectively mentioned in wind speed profile ratio
High precision of prediction, is more nearly prediction of wind speed distribution with wind speed profile is surveyed.
The above is only the embodiment of the present invention, not does any restrictions, all technologies according to the present invention to the present invention
Any simple modification made to the above embodiment, change and equivalent structural changes still fall within technical solution of the present invention
In protection scope.
Claims (4)
1. a kind of forecasting wind speed bearing calibration based on ratio distribution, which comprises the following steps:
Step 1, air speed data collection is handled:
Wind tower is established, sensor acquisition and recording air speed data is set, it is average to acquire actual measurement wind speed per hour for data screening processing
Value;
Step 2, using BP neural network method prediction of wind speed:
Actual measurement air speed data is normalized, determines network inputs output predicted value and hidden neuron number, training
Network establishes prediction model, carries out forecasting wind speed, obtains prediction of wind speed value hourly;
It is as follows to establish prediction model:
Wherein HjFor hidden layer output, f is general hidden layer excitation function, vijThe weight of input and hidden layer, xiFor network inputs, a1
And ajFor threshold value, m is hidden nodes, o1For the output of network, wj1For the weight of hidden layer and output layer;
Step 3, actual measurement prediction of wind speed overall distribution proportional jitter is calculated:
Using x-axis as wind speed size, y-axis is wind speed profile ratio, draws actual measurement wind speed and prediction of wind speed overall distribution ratio chart, and
Calculate the root-mean-square error r of actual measurement wind speed and prediction of wind speed overall distribution proportional curvea;
Step 4, fitting correction multinomial:
Using x-axis as prediction of wind speed, y-axis be actual measurement wind speed, draw prediction of wind speed-actual measurement wind speed scatter plot, setting percentage from 1 to
100, it determines corresponding percentage points, using these points of fitting of a polynomial, obtains 100 correction multinomial models;
Multinomial model is as follows:
B=p1an+p2an-1+p3an-2+p4an-3+p5an-4+......+pna1+pn+1
In formula, a is the item that prediction of wind speed needs to substitute into, and b is the prediction of wind speed value after correction, p1、p2、….pn、pn+1Respectively it is
Numerical value, n are polynomial highest power;
Step 5, optimal calibration model is selected:
Prediction of wind speed value hourly in step 2 is substituted into 100 multinomial models that step 4 obtains respectively, is corrected
Prediction of wind speed value afterwards calculates the deviation of prediction of wind speed and actual measurement wind speed new after correcting in overall distribution ratio again,
The smallest deviation is therefrom selected, corresponding multinomial model is optimal calibration model.
2. a kind of forecasting wind speed bearing calibration based on ratio distribution according to claim 1, which is characterized in that the step
In rapid 2, actual measurement air speed data is normalized, is calculate by the following formula to obtain:
In formula, d (t) is the data before network inputs normalization, and X (t) is the data after normalization, and min (d (t)) is wind speed
Data minimum value, max (d (t)) are air speed data maximum value.
3. a kind of forecasting wind speed bearing calibration based on ratio distribution according to claim 2, which is characterized in that network
It exports predicted value and carries out anti-normalization processing:
Y (t)=o (t) * (max (d (t))-min (d (t))+min (d (t))
In formula, Y (t) is that network exports the data after renormalization, and o (t) is the output of neural network.
4. a kind of forecasting wind speed bearing calibration based on ratio distribution according to claim 1, which is characterized in that calculate real
Survey the root-mean-square error r of wind speed and prediction of wind speed distribution curvea, it is calculate by the following formula to obtain:
U is prediction of wind speed in formula, and v is actual measurement wind speed, and max (u, v) is the maximum value surveyed in wind speed and prediction of wind speed, ti=(i-
1) * 0.1, f (ti) it is actual measurement wind speed profile proportional curve, g (ti) it is prediction of wind speed distribution proportion curve.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710028104.7A CN106874557B (en) | 2017-01-12 | 2017-01-12 | A kind of forecasting wind speed bearing calibration based on ratio distribution |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710028104.7A CN106874557B (en) | 2017-01-12 | 2017-01-12 | A kind of forecasting wind speed bearing calibration based on ratio distribution |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106874557A CN106874557A (en) | 2017-06-20 |
CN106874557B true CN106874557B (en) | 2019-07-02 |
Family
ID=59158143
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710028104.7A Active CN106874557B (en) | 2017-01-12 | 2017-01-12 | A kind of forecasting wind speed bearing calibration based on ratio distribution |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106874557B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107977730A (en) * | 2017-09-18 | 2018-05-01 | 沈阳大学 | A kind of wind measurement method of multisensor Data Fusion technology |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101871948A (en) * | 2010-06-09 | 2010-10-27 | 中国科学院深圳先进技术研究院 | System and method for predicting wind speed of wind farm |
CN102542133A (en) * | 2010-12-10 | 2012-07-04 | 中国科学院深圳先进技术研究院 | Short-time wind speed forecasting method and system for wind power plant |
CN102880907A (en) * | 2012-08-24 | 2013-01-16 | 华锐风电科技(集团)股份有限公司 | Method and device for correcting wind speed |
CN104112166A (en) * | 2014-05-22 | 2014-10-22 | 国家电网公司 | Short-term wind speed prediction method and system of wind power felid |
-
2017
- 2017-01-12 CN CN201710028104.7A patent/CN106874557B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101871948A (en) * | 2010-06-09 | 2010-10-27 | 中国科学院深圳先进技术研究院 | System and method for predicting wind speed of wind farm |
CN102542133A (en) * | 2010-12-10 | 2012-07-04 | 中国科学院深圳先进技术研究院 | Short-time wind speed forecasting method and system for wind power plant |
CN102880907A (en) * | 2012-08-24 | 2013-01-16 | 华锐风电科技(集团)股份有限公司 | Method and device for correcting wind speed |
CN104112166A (en) * | 2014-05-22 | 2014-10-22 | 国家电网公司 | Short-term wind speed prediction method and system of wind power felid |
Non-Patent Citations (4)
Title |
---|
Correction method of wind speed prediction system using predicted wind speed fluctuation;Shogo Yoshida等;《Proceedings of the SICE annual conference 2016》;20160923;第1054-1059页 |
一种实时校正的改进BP神经网络超短期风速预测模型;张文等;《电网与清洁能源》;20160930;第32卷(第9期);第118-127页 |
基于RVM与ARMA误差校正的短期风速预测;孙国强等;《电工技术学报》;20120831;第27卷(第8期);第187-193页 |
基于预测误差校正的支持向量机短期风速预测;周松林等;《系统仿真学报》;20120430;第24卷(第4期);第769-773页 |
Also Published As
Publication number | Publication date |
---|---|
CN106874557A (en) | 2017-06-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CA2807984C (en) | Method and system for monitoring a structure | |
CN108388962B (en) | Wind power prediction system and method | |
CN106875037B (en) | Wind power prediction method and device | |
CN108801387B (en) | System and method for measuring remaining oil quantity of airplane fuel tank based on learning model | |
CN110766132B (en) | Orchard yield intelligent prediction system based on Internet of things | |
CN111680870B (en) | Comprehensive evaluation method for quality of target motion trail | |
CN116205310B (en) | Soil water content influence factor sensitive interval judging method based on interpretable integrated learning model | |
CN111833202B (en) | Farmland evapotranspiration short-term prediction method considering crop coefficient dynamic change and rainfall | |
CN112330050A (en) | Power system load prediction method considering multiple features based on double-layer XGboost | |
CN110263429B (en) | Method for adjusting shape and surface precision of cable net reflecting surface antenna based on machine learning | |
CN104048676B (en) | MEMS (Micro Electro Mechanical System) gyroscope random error compensating method based on improved particle filter | |
CN108960334A (en) | A kind of multi-sensor data Weighted Fusion method | |
CN110794485A (en) | Strong convection weather duration forecasting method based on ensemble learning | |
CN108256693B (en) | Photovoltaic power generation power prediction method, device and system | |
CN115687983A (en) | Bridge health state monitoring method and system and electronic equipment | |
CN106874557B (en) | A kind of forecasting wind speed bearing calibration based on ratio distribution | |
CN108710966B (en) | Photovoltaic power generation power prediction method based on multi-cluster ESN neural network | |
CN102323085A (en) | Angle monitoring based progressive method for recognizing damaged cable and supporting seat generalized displacement | |
CN102323086A (en) | Strain monitoring based progressive method for recognizing damaged cable and supporting seat generalized displacement | |
CN109543911B (en) | Sunlight radiation prediction method and system | |
CN109101759A (en) | A kind of parameter identification method based on forward and reverse response phase method | |
CN108345996B (en) | System and method for reducing wind power assessment electric quantity | |
CN114705177B (en) | Fiber-optic gyroscope attitude measurement data processing method based on error analysis | |
CN113108949B (en) | Model fusion-based sonde temperature sensor error prediction method | |
CN109753018B (en) | Error compensation system and dynamic compensation method based on cloud intelligence |
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 |