Testing for knowledge: Application of machine learning techniques for prediction of flashover in a 1/5 scale ISO 13784‐1 enclosure

Abstract

A machine learning algorithm was applied to predict the onset of flashover in archival experiments in a 1/5 scale ISO 13784-1 enclosure constructed with sandwich panels.The experiments were performed to assess whether a small-scale model could provide a better full-scale correlation than the single burning item test. To predict the binary output, a regularized logistic regression model was chosen as ML environment, for which lasso-regression significantly reduced the amount of variance at a negligible increase in bias. With the regularized model, it was possible to discern the predictive variables and determine the decision boundary. In addition, a methodology was put forward on how to use the to update the learning algorithm iteratively. As a result, it was shown how a learning algorithm can be used to facilitate ongoing experimentation. At first as a crude guideline, and in later stages, as an accurate prediction algorithm. It is foreseen that, by iteratively updating the algorithm, by compiling existing and new experiments in databases, and by applying fire safety knowledge, the final learned algorithm will be able to make accurate predictions for unseen samples and test conditions.

Smart performance-based design for building fire safety: Prediction of smoke motion via AI

“…Over the past decade, smart-firefighting techniques of AI algorithm, IoT, and sensors are gradually adopted [25,26], particularly in the fire behaviors and risk assessments [27][28][29], compartment fire [23,[30][31][32][33][34][35][36] and tunnel fire [37,38]. Hodges et al [30,31] used a transpose convolutional neural network (TCNN) and simulating results conducted by FDS to predict the temperature distribution inside compartment rooms.…”

“…So far, most of the AI application in fire engineering focused on the fire detection and forecast. Arjan et al [33] adopted a logistic regression model to predict the occurrence of flashover in a compartment using the information of fuel thickness, burning intensity and duration. The probabilistic methods have also been applied in evaluating the current fire safety design using [42][43][44].…”

Forecasting Flashover in a Compartment Fire by Using a Real-time Moving Average from Temperature Recordings

“…Focusing on real time forecasting, a zone model is reverted for calculation of fire speed by adjusting the estimated parameters to changes as a function of the fire conditions [10]. A machine learning method is adopted to determine the occurrence of flashover under given fire scenarios [11]. Using some sort of inverse modelling mechanism, generally only the constant fire power can be estimated by infering temperature measurements at the boundaries or within the fire compartment, or visual estimations of the smoke layer height [12].…”

Predicting transient building fire based on external smoke images and deep learning

“…Innovative AI tools have also been developed to assist the fire engineering performance-based design for complex buildings [24,25]. For compartment fires, Dexters et al [26] adopted a machine learning method to determine the occurrence of flashover under given fire scenarios. Wang et al [27] proposed a data recovery algorithm, 'P-flash', which can recover the missing data in case of a sensor was destroyed in a multi-room fire.…”