Semnan University Press Journal of Modeling in Engineering 2008-4854 Articles in Press 2025 09 14 Estimation of noise variance using the weighted EMD coefficients of the noisy signal Estimation of noise variance using the weighted EMD coefficients of the noisy signal 10045 10.22075/jme.2025.34380.2681 FA Majid Dehghanizadeh Electrical Engineering Department -Yazd University MasoudReza Aghabozorgi Electrical Engineering Department- Yazd University Journal Article 2024 06 08 This paper proposes a noise variance estimation method using the Empirical Mode Decomposition (EMD) of signals and the characteristics of the decomposed layers.The main idea is based on the fact that the layers obtained from the EMD decomposition of the noisy signal will contain several layers with some pure signal components, several layers in the form of pure noise, and layers in the form of a combination of signal and noise. According to the definition of signal and noise in this study, the highest signal power is in the final layers, and the highest noise power is in the initial layers of the decomposition. Based on this principle, the noise variance was estimated by calculating the noise energy and the correlation of the noise with different sub-signals from each layer. Based on the simulation results, it is observed that the proposed method in estimating the noise variance has an error reduction of about 1.5, 3, 5 and 1.7 percent, respectively, compared to the Maciej, Elvander, Cai and Wang methods. This paper proposes a noise variance estimation method using the Empirical Mode Decomposition (EMD) of signals and the characteristics of the decomposed layers.The main idea is based on the fact that the layers obtained from the EMD decomposition of the noisy signal will contain several layers with some pure signal components, several layers in the form of pure noise, and layers in the form of a combination of signal and noise. According to the definition of signal and noise in this study, the highest signal power is in the final layers, and the highest noise power is in the initial layers of the decomposition. Based on this principle, the noise variance was estimated by calculating the noise energy and the correlation of the noise with different sub-signals from each layer. Based on the simulation results, it is observed that the proposed method in estimating the noise variance has an error reduction of about 1.5, 3, 5 and 1.7 percent, respectively, compared to the Maciej, Elvander, Cai and Wang methods. https://modelling.semnan.ac.ir/article_10045_df96dc7ffa21cfab415e1de2ddbb0ae0.pdf