Machine Learning Optimizes Zeolite Dosage for Fireworks Storage

A machine learning decision-support system was suggested to estimate the right amount of zeolite to help achieve safe storage. The study, published in Scientific Reports, addresses the critical safety risks posed by fluctuations in temperature and relative humidity, which can destabilize chemical mixes and lead to explosions in storage facilities. By integrating environmental monitoring with computational classification, the team aimed to provide a data-driven tool for managing these hazardous conditions.

The experimental design involved gathering temperature and relative humidity data from storage inserts over a ten-day period. The researchers tested three specific quantities of zeolite: 5 kg, 10 kg, and 15 kg. To ensure comprehensive data collection, both wet and dry states of the zeolite were evaluated. This experimental setup resulted in the acquisition of 208 distinct records, which formed the dataset for the subsequent machine learning analysis.

For the computational analysis, the study utilized the WEKA environment to test eight different classification algorithms. These algorithms included Logistic Regression, Support Vector Machine (SMO), Random Forest, BayesNet, k-Nearest Neighbor, J48, Multiclass Classifier, and Multilayer Perceptron. To validate the performance of these models, the researchers employed stratified 10-fold cross-validation.

The results indicated that Logistic Regression provided the best performance among the tested algorithms, achieving a classification accuracy of 78.41%. The Support Vector Machine (SMO) algorithm followed closely, with a classification accuracy of 77.24%. The study identified humidity and the state of the zeolite as the most significant factors influencing the classification outcomes.

The authors note that while the results show the possibility of integrating zeolite-based humidity regulation with machine learning systems, the model requires further validation. Specifically, the study highlights the need for testing with larger datasets and deployment in real-world situations to confirm the robustness of the findings.

Despite these limitations, the suggested model provides a useful tool for decision-making in fireworks storage safety.

Subject of Research: Machine learning classification of zeolite quantities for fireworks storage safety

Article Title: Machine learning-based classification of zeolite quantities for humidity and temperature control in fireworks storage safety

Article References: Indumathi, N., Gnanavel, S., Sultanuddin, S. J., & Vijayalakshmi, V. (2026). Machine learning-based classification of zeolite quantities for humidity and temperature control in fireworks storage safety. Scientific Reports. https://doi.org/10.1038/s41598-026-71532-z

Image Credits: AI Generated

DOI: 10.1038/s41598-026-71532-z

Keywords: machine learning, fireworks safety, zeolite, humidity control, temperature monitoring, Scientific Reports, industrial safety, classification algorithms, Machine, learning-based, classification, quantities