Weakly Cemented Sandstone Abrasivity Test Data for Neural Network Modeling

A 9.5 KB dataset containing experimental results from Cerchar abrasivity tests on weakly cemented sandstones in western China. It was created by Yafeng Yao to train an improved fuzzy stochastic RBF neural network model for predicting rock abrasivity indices. The dataset was last updated in April 2026.

Use Cases

• Predict the Cerchar Abrasivity Index (CAI) using input features like hardness, wave velocity, porosity, and equivalent quartz content.
• Predict the Cerchar Abrasivity Ratio (CAR) based on the same physical property inputs.
• Train neural network models to account for abrasivity changes across different physical states (dry, water-saturated, mud-saturated).
• Compare model performance (MSE, MAE, R-squared) of improved RBF networks against MLR, SVR, and traditional RBF models.
• Analyze the relationship between sandstone microstructure and CAI values under varying saturation conditions.

Strengths

• Model trained on the data achieved a prediction error of less than 5% and a Coefficient of Determination (R-squared) of 0.967.
• Data captures specific, quantified abrasivity changes: CAI increased by 10.64% in dry vs. mud-saturated state and 19.23% in dry vs. water-saturated state.

Limitations

• The dataset is very small at 9.5 KB, indicating a limited number of experimental samples or rows.
• Geographic scope is limited to weakly cemented sandstones from western China, which may not generalize to other rock types or regions.

Provenance

Source

Yafeng Yao via figshare.

Collection Method

Data gathered from Cerchar abrasivity characteristic tests conducted on rock samples in a laboratory setting.

Time Range

null

Freshness

Last updated in April 2026.

Geography

Western China.