Abstract:Basalt fiber is commonly regarded as an alkaline corrosion-resistant material, capable of being transformed into various products (such as basalt fiber roving, chopped yarn and composite reinforcement). However, the main composition of the basalt fiber network skeleton consists of Si similar to O bonds, which can still react with similar to OH in the alkaline environment. Nevertheless, there are differing viewpoints on the alkaline corrosion of basalt fibers within literatures. Therefore, this study aims to select three representative types of basalt fibers and compare them with two alkaliresistant glass fibers and one regular glass fiber. The attenuation mechanism of basalt fibers were investigated through orthogonal experiments, practical application scenarios and microscopic characterization methods. Additionally, we employed the Weibull model to analyze failure probabilities after erosion on basalt fibers while also applying a machine learning artificial neural network (ANN) model for predicting their alkali resistance. The results indicated that the tensile strength of basalt fiber tended to decrease and the alkali resistance of basalt fiber between alkaliresistant glass fiber and regular glass fiber in four alkaline solutions. The Weibull model also confirms that erosion increases failure probability of basalt fiber and glass fiber. Moreover, Si similar to O, which forms the network reacted with OH-, resulting in a reduction in surface Si and Al elements content. It became possible to predict the alkali resistance of basalt fiber accurately by establishing an ANN model. Our study provide a theoretical foundation for predicting the lifespan of basalt fibers under corrosive conditions while offering guidance for enhancing their alkali resistance.

Alkali Resistance Prediction and Degradation Mechanism of Basalt Fiber: Integrated with Artificial Neural Network Machine Learning Model