Application of metaheuristic optimization algorithms-based three strategies in predicting the energy absorption property of a novel aseismic concrete material

This study aims to predict the energy absorption property of a novel aseismic concrete material made of rubber, sand and cement. To investigate the energy absorption property of this novel aseismic concrete material, the energy transmission rate (ETR) was calculated by using the Split-Hopkinson Pressure Bar (SHPB) device. Furthermore, some prediction models were developed to predict the ETR in order to estimate it in the field and other laboratory environments. Therefore, six metaheuristic optimization algorithms-based three strategies (i.e., evolutionary algorithms: Differential evolution (DE) and Human felicity algorithm (HFA); physical algorithms: Nuclear reaction optimization (NRO) and Lightning search algorithm (LSA); swarm intelligence algorithms: Harris Hawks optimization (HHO) and Tunicate swarm algorithm (TSA)) and random forest (RF) model were combined to generate various hybrid prediction models for forecasting the ETR. The results indicated that the TSA-RF model has the best performance for predicting the ETR in both the training phase (RMSE: 1.5388 and R2: 0.9349) and the testing phase (RMSE: 1.6083 and R2: 0.9165). The sensitive analysis results demonstrated that cement is the most important parameter for predicting the ETR, but the rubber showed the largest negative correlation with the ETR. As a result, the application of artificial intelligence in ETR prediction has been proven to be feasible, this work can provide a novel idea for the development of aseismic materials in tunnel engineering.

Automated summary

This study aims to predict the energy absorption property of a novel aseismic concrete material made of rubber, sand and cement. Various hybrid prediction models, including metaheuristic optimization algorithms and a random forest model, were developed and tested. The TSA-RF model demonstrated the best performance in predicting the energy transmission rate (ETR) of the concrete material, with cement being identified as the most important parameter for ETR prediction. This study provides a feasible application of artificial intelligence in ETR prediction and offers a novel idea for the development of aseismic materials in tunnel engineering.