Concrete compressive strength prediction modeling utilizing deep learning long short-term memory algorithm for a...

Area of Science:

• Materials Science
• Civil Engineering
• Data Science

Background:

• Concrete compressive strength is a critical design parameter impacting construction efficiency.
• Environmental factors like weather and humidity can influence concrete strength development.
• Accurate prediction of concrete strength can lead to significant cost and time savings.

Purpose of the Study:

• To develop and compare predictive models for concrete compressive strength.
• To evaluate the performance of deep learning (LSTM) and machine learning (SVM) algorithms for this task.
• To identify the most effective method for reliable concrete strength prediction.

Main Methods:

• Utilized a comprehensive dataset from published studies on concrete properties.
• Developed predictive models using Long Short-Term Memory (LSTM) deep learning.
• Implemented a Support Vector Machine (SVM) machine learning algorithm for comparison.
• Evaluated model performance using R², MAE, and RMSE statistical indices.

Main Results:

• The LSTM model demonstrated superior performance in predicting concrete compressive strength.
• LSTM achieved an R² of 0.98, MAE of 1.861, and RMSE of 2.36.
• The SVM model yielded an R² of 0.78, MAE of 6.152, and RMSE of 7.93.
• Input variables included cement, slag, fly ash, water, superplasticizer, aggregates, and age.

Conclusions:

• The proposed LSTM model offers a reliable method for measuring high-performance concrete (HPC) compressive strength.
• Deep learning approaches show significant potential for enhancing concrete design and quality control.
• Accurate strength prediction facilitates optimized material usage and construction timelines.