Application of Artificial Neural Network to Predict Occupational Safety and Health Risk in Shipyards – a Cases Study of 2 Shipyards in Sibu, Sarawak
Occupational safety and health (OSH) risk management research is common in industries such as construction, chemical production, electrical and electronic production. However, to date there is a lack of research performed on the OSH risk management related to shipyard industry. OSH practitioners oft...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
International Journal of Integrated Engineering
2023
|
| Subjects: | |
| Online Access: | http://ir.unimas.my/id/eprint/47422/4/Thesis%20Meng_CALVIN%20CHIN%20YEN%20CHIH.pdf http://ir.unimas.my/id/eprint/47422/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Occupational safety and health (OSH) risk management research is common in industries such as construction, chemical production, electrical and electronic production. However, to date there is a lack of research performed on the OSH risk management related to shipyard industry. OSH practitioners often make preventive decisions based on physical site audit in the shipyard and relying on individual experience. The aforementioned task is difficult to execute due to multiple influencing factors that could affect the decision-making. Therefore, this project aims to propose a risk management framework by utilising a supervised machine learning Artificial Neural Network (ANN) to predict occupational risk in an identified Malaysia shipyard. Optimization and classification with the appropriate architecture of neuron layers were trained for the risk prediction at shipyards. In this research, 10 independent input variables and 1 single layer-dependent output variable (severity of risk) were defined. 7 networks modeling were developed between 19–31 neuron nodes. Accuracy performance of the developed networks were compared by using evaluation criteria, such as correlation of coefficient (R2), mean square error (MSE) and mean absolute percentage error (MAPE), to select the best model. The best selected ANN network no.4 achieves the highest accuracy performance of 90.22%. However, in terms of sensitivity analysis, input factors such as working hours (28.38%), and workplace factors (21.32%), have significant effects onto the OSH risk prediction. Then machine learning was intervened into the proposed framework which entails identification of hazards, risk assessment, safety decision-making, risk control, and risk prediction. With this, a dynamic risk management framework is formed and continuously improved. |
|---|