Implementation of Particle Swarm Optimization (Pso) Algorithm for Estimating Parameter of Arma Model via Maximum Likelihood Method

Handoyo, Samingun; Efendi, Achmad; Jie, Ferry; Widodo, Agus

doi:10.17654/ms102071337

Cited by 10 publications

(7 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The various steps involved in implementing the dynamic PSO algorithm for the given ARIMA model are given below (Saxena et al, 2015). The DPSO procedure is explained though flowchart shown in Figure 4 (Handoyo et al, 2017;Sankardoss & Geethanjali, 2017) and the pseudo code of the DPSO algorithm is given in Figure 5. Maximum number of gBest stored as history in gBest_h = 10*m, pBest_iter_thresh = gBest_iter_ thresh = 5*m (dimension).…”

Section: Dynamic Particle Swarm Optimization (Dpso) Algorithmmentioning

confidence: 99%

Short-Term Wind Power Prediction Using Hybrid Auto Regressive Integrated Moving Average Model and Dynamic Particle Swarm Optimization

Singh

Negi

2021

International Journal of Cognitive Informatics and Natural Intelligence

View full text Add to dashboard Cite

With the upsurge in restructuring of the power markets, wind power has become one of the key factors in power generation in the smart grids and gained momentum in the recent years. The accurate wind power forecasting is highly desirable for reduction of the reserve capability, enhancement in penetration of the wind power, stability and economic operation of the power system. The time series models are extensively used for the wind power forecasting. The model estimation in the ARIMA model is usually accomplished by maximizing the log likelihood function and it requires to be re-estimated for any change in input value. This degrades the performance of the ARIMA model. In the proposed work, the model estimation of the ARIMA model is done using latest evolutionary algorithm (i.e., dynamic particle swarm optimization [DPSO]). The use of DPSO algorithm eliminates the need for re-estimation of the model coefficients for any change in input value and moreover, it improves the performance of ARIMA model. The performance of proposed DPSO-ARIMA model has been compared to the existing models.

show abstract

Section: Dynamic Particle Swarm Optimization (Dpso) Algorithmmentioning

confidence: 99%

Short-Term Wind Power Prediction Using Hybrid Auto Regressive Integrated Moving Average Model and Dynamic Particle Swarm Optimization

Singh

Negi

2021

International Journal of Cognitive Informatics and Natural Intelligence

View full text Add to dashboard Cite

show abstract

“…With regard to FRB, the cluster center formed as the center (mean parameter) of the Gaussian membership function (MF), but the spread parameter of the Gaussian MF is determined by a formula deterministically that result in the magnitude of spread contradiction with reality. Estimation of spread parameters with optimization techniques has been widely discussed in literature such as by Rao and Bard [21], Pratama and Handoyo [22] estimated the Gaussian MF parameters with genetic algorithms, Handoyo, et al [23] estimated the ARMA model parameters with particle swarm optimization (PSO), and estimation of normal distribution parameters and multiple linear regression model parameters can use the Ordinary Least Square (OLS) method as found in Heaney, and. Poitras [24], Lee [25], and also in Rosipal, and Krämer [26].…”

Section: Figure 1 the Components Of Fuzzy Inference Systemmentioning

confidence: 99%

Generating of Fuzzy Rule Bases with Gaussian Parameters Optimized via Fuzzy C-Mean and Ordinary Least Square

2019

IJRTE

View full text Add to dashboard Cite

The fuzzy rule bases has a central role in the fuzzy inference system. Generating of rule bases based on input-output data pairs by using Fuzzy C-Mean clustering (FCM) requires parameters setting of the membership function (mf). In the Gaussian mf, the mean and spread parameters must be determined. The outputs of FCM clustering include the cluster center and the partition matrix of each object. The value of cluster center can be used as the center parameter, but the spread parameter is usually determined as a constant value. The research proposes a method to determine spread parameter of Gaussian mf by using Ordinary Least Square (OLS) approach with using the partition matrix as the membership degree of each object in the cluster. The magnitude of cluster centers (n) of 3, 5, and 7 are considered as FCM input to cluster the weekly price of soybeans in East Java, Indonesia on the period of January 2014 to December 2017. Based on each cluster center, the optimal spread value of a Gaussian mf is obtained via OLS. This research succeeded in getting a spread values that could approach almost perfectly each element of the partition matrix

show abstract

“…Modeling the transfer function using OLS was done by Kusdarwati and Handoyo [11] and also on the time series modeling based on OLS optimization was conducted by Widodo et al [12]. The performance of the OLS method is also compared with the evolutionary computational method known as particle swarm optimization (PSO) which has been investigated by Handoyo et al [13], and also by Achmad et al [14]. Based on the above studies the OLS method has a very satisfying performance even comparable to the performance of the evolutionary method of the PSO.…”

Section: Introductionmentioning

confidence: 99%

Modeling Academic Performance of The Students by using Partial Least Square

Marji*,

Handoyo,

Handamari

et al. 2020

IJRTE

View full text Add to dashboard Cite

This study aims to model the relationship between predictor variables consisting of learning motivation (LM), parents' socioeconomic status (SS), and school environment (SE) which are all latent variables to academic achievement (AC) which are not latent variables. Modeling is done by the method of partial least square (PLS) which is expected to explore the various effects found in the inner model and also confirm the questionnaire items forming the latent variables. With a real level of 5%, almost all loading values on each latent variable are significant. Likewise, a simple linear relationship consisting of 5 models has a coefficient that has a significant effect. The influence of learning motivation (LM), parents' socioeconomic status (SS), school environment (SE) on academic achievement are 0.270, 0.249, and 0.320, respectively. while learning motivation (LM), socioeconomic status of parents (SS) contributing to academic achievement (AC) were 17.7% and 3.13%, respectively..

show abstract

Implementation of Particle Swarm Optimization (Pso) Algorithm for Estimating Parameter of Arma Model via Maximum Likelihood Method

Cited by 10 publications

References 0 publications

Short-Term Wind Power Prediction Using Hybrid Auto Regressive Integrated Moving Average Model and Dynamic Particle Swarm Optimization

Short-Term Wind Power Prediction Using Hybrid Auto Regressive Integrated Moving Average Model and Dynamic Particle Swarm Optimization

Generating of Fuzzy Rule Bases with Gaussian Parameters Optimized via Fuzzy C-Mean and Ordinary Least Square

Modeling Academic Performance of The Students by using Partial Least Square

Contact Info

Product

Resources

About