The Role of Artificial Intelligence in Improving Failure Mode and Effects Analysis (FMEA) Efficiency in Construction Safety Management

This study presents an analysis of the introduction of the artificial intelligence to the FMEA method. The traditional function of the FMEA only evaluates the impact and causes of system failure which takes a long time. Therefore, it was necessary to find a way to develop and improve the performance of FMEA. The goal is to develop FMEA technique into an electronic application in which artificial intelligence is used to identify, calculate, and evaluate risks at the same time. For achieving the set goal, this paper introduces a real field study on how to develop the FMEA from a traditional method to a more developed program in which artificial intelligence is used as an alternative to the work team in calculating and evaluating various risks. In this study, we took construction as a primary field for evaluating the performance of the modern FMEA program, where the work team takes two risks from the field hazards as a sample for use (“Construction project delivery failed on time” and “Workers falling from high floors”). In this study, 5 factors were put as causes for each risk with an evaluation of 50 possible cases for the occurrence of each risk. Our dataset of 50 cases for each risk has been loaded into our software. This program analyzed the dataset for each hazard in the 50 × 8 dimension. In order to reduce the number of outputs for the (D, O, S) computation, our team used multi output regressor outputs instead of single output regressor, and focused on three types (multiple output regressor algorithms), namely: Linear Regression, Training with the Random Forest Regressor Model, Training with the Decision Tree Regression Model. The results showed that the best metric algorithm MAE = 0.36 turned out to be an algorithm Random Forest Regressor. By using the artificial intelligence feature, we found that the features that play a major role in training the model for failure to deliver the Construction project on time is (work suspension due to the Corona pandemic (two weeks) for 20% of the total number of workers) and that the least influential feature in the model is (The acute shortage of funding for raw materials such as iron and concrete leads to a delay in implementation). As for the workers who fall from high floors, the feature that plays a big role is (falling from the stairs) and the feature that has the least impact is (falling from the stairs). The research employs qualitative, analytical, statistical and comparative methods of FMEA. The acquired results can be used by different construction companies to identify points of failure, anticipate risks, and calculate them in an innovative and smooth way. Also, it allows companies to avoid notifications in advance. This contributes to the improvement of institutions’ reputation and the quality of their products. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.