Skip to main content
placeholder image

A Multi-dimension Clustering Method for Load Profiles of Australian Local Government Facilities

Conference Paper


Abstract


  • The clustering of historical electricity consumption data is an effective means of developing representative load profiles for long-term energy planning. This paper presents a multi-dimensional approach for clustering, considering scattering and separation metrics and the number of clusters. A novel hybrid approach to solve the clustering function is also proposed: a combination of Invasive Weed Optimization (IWO) and wavelet mutation strategy. The hybrid method is applied to half-hourly metered electricity consumption data from the Civic Centre of a large local (municipal) government in Perth, Western Australia, to create representative seasonal load profiles. The novel clustering approach is then tested against the well-known k-means method using Davies-Bouldin and silhouette indices. In each seasonal clustered profile, the hybrid method is found to outperform the k-means method. The hybrid method has been identified as an effective clustering approach for analyzing the behavior of loads and assisting the identification of suitable energy efficiency initiatives.

Publication Date


  • 2021

Citation


  • Azizivahed, A., Rupf, G. V., Lund, C., Arefi, A., Walia, J., Rahman, M. M., & Islam, M. R. (2021). A Multi-dimension Clustering Method for Load Profiles of Australian Local Government Facilities. In 2021 IEEE 6th International Conference on Computing, Communication and Automation, ICCCA 2021 (pp. 819-825). doi:10.1109/ICCCA52192.2021.9666208

Scopus Eid


  • 2-s2.0-85124800572

Web Of Science Accession Number


Start Page


  • 819

End Page


  • 825

Abstract


  • The clustering of historical electricity consumption data is an effective means of developing representative load profiles for long-term energy planning. This paper presents a multi-dimensional approach for clustering, considering scattering and separation metrics and the number of clusters. A novel hybrid approach to solve the clustering function is also proposed: a combination of Invasive Weed Optimization (IWO) and wavelet mutation strategy. The hybrid method is applied to half-hourly metered electricity consumption data from the Civic Centre of a large local (municipal) government in Perth, Western Australia, to create representative seasonal load profiles. The novel clustering approach is then tested against the well-known k-means method using Davies-Bouldin and silhouette indices. In each seasonal clustered profile, the hybrid method is found to outperform the k-means method. The hybrid method has been identified as an effective clustering approach for analyzing the behavior of loads and assisting the identification of suitable energy efficiency initiatives.

Publication Date


  • 2021

Citation


  • Azizivahed, A., Rupf, G. V., Lund, C., Arefi, A., Walia, J., Rahman, M. M., & Islam, M. R. (2021). A Multi-dimension Clustering Method for Load Profiles of Australian Local Government Facilities. In 2021 IEEE 6th International Conference on Computing, Communication and Automation, ICCCA 2021 (pp. 819-825). doi:10.1109/ICCCA52192.2021.9666208

Scopus Eid


  • 2-s2.0-85124800572

Web Of Science Accession Number


Start Page


  • 819

End Page


  • 825