A performance comparison study on PM2.5 prediction at industrial areas using different training algorithms of feedforward-backpropagation neural network (FBNN).

Presence of particulate matters with aerodynamic diameter of less than 2.5 µm (PM2.5) in the atmosphere is fast increasing in Malaysia due to industrialization and urbanization. Prolonged exposure of PM2.5 can cause serious health effects to human. This research is aimed to identify the most reliable model to predict the PM2.5 pollution using multi-layered feedforward-backpropagation neural network (FBNN). Air quality and meteorological data were collected from Department of Environment (DOE) Malaysia. Six different training algorithms consisting of thirteen various training functions were trained and compared. FBNN model with the highest coefficient correlation (R2) and lowest root mean square error (RMSE), mean absolute error (MAE) and mean absolute percentage error (MAPE) were selected as the best performing model. Levenberg Marquardt (trainlm) is the best performing algorithms compared to other algorithms with R2 value of 0.9834 and the lowest error values for RMSE (2.3981), MAE (1.7843) and MAPE (0.1063).

Identification of Noise Levels for Skill Training Activities, Equipment, Machines and Power Tools at TVET Institutes in Malaysia.

The use of various machines, equipment and power tools at TVET Institute causes the institute's environment to be exposed to noise hazards that are similar to the industry. However, not much data has been published regarding noise exposure at TVET institutes. This study was carried out to document the noise exposure of work activities training in public TVET institutes in Malaysia that implement skill training programs in metal fabrication, furniture manufacturing and automotive maintenance. The identification of excessive noise, task-based noise exposure monitoring and source measurement was conducted. The noise contribution from each work activity to the daily A-weighted noise exposure level and sound pressure level emitted by machines and equipment was documented. The findings of this study recorded 20 activities with task-based noise contribution to the daily A-weighted noise exposure level between 75.3 dB and 95 dB. Based on the findings, the training environment at the TVET institutes has a risk of operating with excessive noise. The documented data can be used in planning the implementation of suitable noise control measures in TVET institutes.

Effectiveness of a Novel Index System in Preventing Early Hearing Loss among Furniture Industry Skills Training Students in Malaysia.

Occupational noise-induced hearing loss (ONIHL) is the most reported occupational disease in Malaysia. ONIHL is aggravated by the presence of early hearing loss amongst the youth prior to entering a real working environment. At technical and vocational education training (TVET) institutions, students may develop early ONIHL because training workshops are designed imitating the industrial working environment to produce skilled workers. The exceeding noise level at workshops and recent risk of non-occupational noise can cause early ONIHL among these students. Therefore, ONIHL must be addressed at the early stage of producing skilled workers. Octa hearing conservation index (OHCI) system is developed as a management and monitoring tool for hearing conservation program (HCP) in TVET institutions. Six existing and two new HCP components were used to build the index system. A pilot test on the effectiveness of the OHCI system was conducted in a selected TVET institution for six months. The post-HCP shows a 52.6% improvement compared to the pre-HCP. The implementation of HCP has shown improved awareness on the hazards of loud noise exposure and active use of hearing protection devices among participants. The OHCI system has a great potential as a tool to improve HCP implementation in TVET institutions, and eventually, industry.

Parametric optimization of polycaprolactone synthesis catalysed by Candida antarctica lipase B using response surface methodology.

A statistical approach with D-optimal design was used to optimize the process parameters for polycaprolactone (PCL) synthesis. The variables selected were temperature (50°C-110°C), time (1-7 h), mixing speed (50-500 rpm) and monomer/solvent ratio (1:1-1:6). Molecular weight was chosen as response and was determined using matrix-assisted laser desorption/ionization time of flight (MALDI TOF). Using the D-optimal method in design of experiments, the interactions between parameters and responses were analysed and validated. The results show a good agreement with a minimum error between the actual and predicted values.