Kwan, Chak Yin (2018) High accuracy anomaly detection for cyber physical system using self-organizing map based algorithm. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)
|
Text (Full Text)
High accuracy anomaly detection for cyber physical system using self-organizing map based algorithm.pdf - Submitted Version Download (3MB) | Preview |
Abstract
In order to embrace challenges of Industry 4.0 (I4.0), failure prediction on the machinery in the Cyber-Physical System (CPS) is important which gives rise to the research in anomaly detection. Recent studies on anomaly detection generally applied to the network security, fraud system and image processing. However, anomaly detection in I4.0 have difficulty in ensuring the designed algorithm is self-adaptive without compromising the accuracy of the prediction. Hence, this project addresses this problem by proposing a habituating SOM-based algorithm to predict the possible failure faced in the system. In the proposed method, the SOM and k-means act as the clustering network for the mechanism, while the habituation function take role as set of habituating synapses that form connection among the network neurons to the output. Weight vector for the neurons is initialized via k-means clustering to ensure reasonable number of cluster and proper distribution of weight vector. Receiver Operating Characteristic (ROC) curve is used to optimize threshold value of Euclidean distance. Accuracy test is carried out by execute the algorithm to different dataset that contain various number of anomalies. The performance of the algorithm is evaluated via the application of confusion matrix in term of accuracy. The proposed algorithm can detect the anomalies occur in the data accurately with minimum accuracy of 98.5% and maximum accuracy of 99.2%. This indicates that there is possibility to use the proposed anomaly detection technique to predict the possible failure in CPS’ machinery.
Item Type: | Final Year Project (Project Report) |
---|---|
Uncontrolled Keywords: | Image processing, Induction evolution |
Subjects: | T Technology > T Technology (General) T Technology > TA Engineering (General). Civil engineering (General) |
Divisions: | Library > Final Year Project > FKE |
Depositing User: | Mohd Hannif Jamaludin |
Date Deposited: | 31 Dec 2019 02:56 |
Last Modified: | 18 Feb 2025 23:59 |
URI: | http://digitalcollection.utem.edu.my/id/eprint/24111 |
Actions (login required)
![]() |
View Item |