| Summary: | Annually, a massive number of agricultural by-products of the palm oil extraction process including palm oil fuel ash (POFA) were generated which contributes towards ammonia pollution and emission of nitrogen compounds. Fortunately, both by-products can be utilised as mixing additives in lightweight aggregate concrete manufacturing. The utilisation leads to a more sustainable green environment. Traditional methods for classifying concrete grades in civil engineering are difficult due to the non-linear relationship between the composition of concrete and its strength and require a significant amount of time, material resources, and labour. To address these shortcomings, a technique to classify the compressive strength grades for lightweight aggregate concrete containing POFA using a machine learning algorithm has been developed. In terms of method, concrete mixtures consisting of POFA, cement, sand, superplasticizer and water were prepared and tested to determine the compressive strength. The data from this process were first transformed using min-max normalization and then, analysed using exploratory and descriptive analysis to discover patterns between input variables and concrete grades. Next, the grades of concrete were classified using a machine learning algorithm named k-Nearest Neighbour (k-NN). Lastly, a confusion matrix was used to assess the performance of the k-NN classifier. The results showed that k-NN can classify the grades of concrete with accuracies between 71% and 95% using five nearest neighbours. The accuracies are inversely proportional to the number of nearest neighbours. To conclude, the study succeeds in classifying the compressive strength grades for lightweight aggregate concrete with POFA using k-Nearest Neighbour. It can cut down a significant amount of time, material resources, and labour in determining the grades of compressive strength for POFA-based lightweight concrete. © Published under licence by IOP Publishing Ltd.
|
|---|