Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Two-stage-based time sequence prediction method and prediction system, terminal and medium

A technology of time series and forecasting methods, applied in forecasting, neural learning methods, instruments, etc., can solve problems such as weak useful information, model accuracy does not increase but decreases, and achieves the effect of improving accuracy and realizing prediction accuracy

Pending Publication Date: 2021-01-15
WUHAN UNIV OF SCI & TECH
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

For the residual, the mixed useful information is weak, and if the model selected for the prediction of the residual is incorrect, the accuracy of the final model may not increase but decrease.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Two-stage-based time sequence prediction method and prediction system, terminal and medium
  • Two-stage-based time sequence prediction method and prediction system, terminal and medium
  • Two-stage-based time sequence prediction method and prediction system, terminal and medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0058] The object of the present invention is to use the forecasting method of two-stage CEEMDAN-PSO-ELM to realize the promotion of forecasting accuracy on the basis of traditional forecasting model, concrete steps are as follows:

[0059] Step 1: Use the CEEMDAN method to decompose the original sequence to get the IMF 1 ,IMF 2 ,...,IMF n and Res;

[0060] Step 2: Put the IMF 1 ,IMF 2 ,...,IMF n Normalize with Res:

[0061]

[0062] Where: x max and x min are the maximum and minimum values ​​in the input data respectively; y is the normalized input value.

[0063] Step 3: Bring each normalized subsequence into the PSO-ELM model for prediction, denormalize the output result, and obtain the prediction result Y 1 ,Y 2 ,...Y n ,Y n+1 ;

[0064] Step 4: Set Y 1 ,Y 2 ,...Y n ,Y n+1 Sum up to get the first stage prediction result Y sum , subtract the actual value from Y sum Get the error sequence E.

[0065] Step 5: Use the CEEMDAN method to decompose the erro...

Embodiment 2

[0071] The embodiment of the present invention adopts the monthly mean value data of sunspots, which are derived from the official website (http: / / sidc.oma.be / silso / datafiles) of the Belgian Royal Astronomical Observatory solar action data analysis center (Solar Influence Data Analysis Center, SIDC) ), its prediction process is as follows figure 2 shown. The first stage: use CEEMDAN to stabilize the sunspot number series of monthly mean value as image 3 shown; ELM models were established for each sub-sequence after decomposition, and the ELM parameters of each sub-model were optimized with the PSO algorithm, as shown in Table 1 and Figure 4 As shown, the prediction results of each component are superimposed to obtain the first-stage prediction results. The second stage: CEEMDAN-PSO-ELM modeling is performed on the residual obtained in the first stage, and the intermediate process is as follows Figure 5 , Table 2 and Figure 6 As shown, the prediction results of the sec...

Embodiment 3

[0083] CEEMDAN decomposition principle

[0084] Huang proposed an Empirical Mode Decomposition (EMD) method that can decompose any signal into Intrinsic Mode Functions (IMF). M.A.Colorminas proposed the CEEMDAN decomposition method based on the research of Huang et al. CEEMDAN utilizes the characteristics of zero-mean Gaussian white noise to make the decomposition effect of signal data more complete. The specific processing process is as follows:

[0085] Step 1: Add standard normal distribution white noise w of different amplitudes to the given target signal x(n) i (n), construct the signal sequence of the i-th experiment as

[0086] x i (n)=x(n)+γw i (n) (i=1,...,I) (1)

[0087] Step 2: In the first stage, use the EMD method to decompose the target signal, obtain the first modal component and take the mean value

[0088]

[0089] The margin signal of the first stage is expressed as:

[0090]

[0091] Step 3: Define E k (·) is the kth IMF component after performi...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention belongs to the technical field of time sequence prediction, and discloses a two-stage-based time sequence prediction method and prediction system, a terminal and a medium. CEEMDAN is used for stabilizing a monthly mean value sequence of sunspots; the method includes establishing an ELM model for each decomposed sub-sequence, optimizing ELM parameters of each sub-model by using a PSOalgorithm, and superposing each component prediction result to obtain a first-stage prediction result; performing CEEMDAN-PSO-ELM modeling on the residual error obtained in the first stage to obtain aprediction result in the second stage; and summing the first-stage prediction result and the second-stage prediction result to obtain a final prediction result. According to the invention, adaptive noise complete aggregation empirical mode decomposition in a multi-scale decomposition method is combined with an extreme learning machine in a neural network algorithm, and the prediction precision ofa single ELM model is further improved on the basis of a traditional prediction model.

Description

technical field [0001] The invention belongs to the technical field of time series prediction, and in particular relates to a two-stage time series prediction method, a prediction system, a terminal and a medium. Background technique [0002] In recent years, nonlinear forecasting models represented by neural networks have been widely used in the forecasting of nonlinear systems due to their extensive adaptability and learning capabilities, and the use of multi-scale decomposition methods can weaken the nonlinearity of time series and non-stationarity, effectively improving the prediction accuracy. However, in the existing prediction algorithms based on neural networks, there are still two factors that have not been considered. One is how to optimize the model and parameters of the network in the prediction model, and the other is that the impact of residuals on prediction accuracy is not considered. In the prediction model based on neural network, the selection of network ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06Q10/04G06N3/00G06N3/04G06N3/08
CPCG06Q10/04G06N3/006G06N3/08G06N3/045
Inventor 余楠王文波童梦
Owner WUHAN UNIV OF SCI & TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com