Time series prediction model based on improved multi-objective differential optimization echo state network
A technology of echo state network and prediction model, which is applied in the direction of biological neural network model, computing model, biological model, etc., can solve problems such as difficult to determine, and achieve the effect of improving prediction accuracy, high robustness, and strong global search ability
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0060] Embodiment 1 The beneficial effect of the present invention is illustrated through the data sets of monthly average temperature and monthly rainfall in Dalian. The monthly average temperature and precipitation data set from January 1951 to December 2017 in Dalian was selected, with a total of 792 samples. The sampling interval is months. The monthly precipitation in Dalian is used as the second dimension of the experimental data to assist in the prediction of the monthly average temperature in Dalian. 75% of the data is used for training and 25% for testing. In the training samples, the states of the first 50 samples are discarded to clean up the influence of initial transients. In order to prove the effectiveness of the present invention, for PSO-ESN, AFSA-ESN, TLBO-ESN, IBEA-ESN, the present invention makes the range of ESN reserve pool parameters the same as shown in Table 1. The population size and maximum number of iterations are also the same. Table 2 shows th...
Embodiment 2
[0067] Preferably, embodiment two adopts Beijing PM from Beijing Air Pollution Index (AQI) 2.5 sequentially. The present invention uses Convergent Cross Mapping (CCM) for Beijing PM 2.5 Auxiliary variables for time series selection, found PM 10 , CO and SO 2 to PM 2.5 big impact. Therefore, the present invention selects PM 10 , CO and SO 2 as a PM 2.5 Auxiliary variables for time series forecasting. The present invention selects from January 2, 2016 to December 31, 2016, a total of 8759 samples are used for simulation. 70% of the total samples are used for training and 30% of the total samples are used for testing. In the training samples, the states of the first 100 samples are discarded to remove the influence of initial transients.
[0068] From Table 1, we can know the range settings of the reserve pool parameters. In order to prove the validity of the model proposed by the present invention, the present invention has carried out the same settings on the AFSA-ES...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More 


