Data-Driven Analysis of the Portevin–Le Chatelier Effect in an Al-Mg Alloy Across Temperatures and Strain Rates

Data associated with the research paper:
Quantitative evaluation of the dependence of the Portevin-Le Chatelier effect on temperature and strain rate in an Al–Mg alloy (AA5083-H111): Insights from machine learning

Paper Abstract:
Al–Mg alloys, among others, exhibit the Portevin-Le Chatelier (PLC) effect. In addition to serrations in the stress–strain curve, the PLC effect also manifests itself macroscopically as stretcher strain marks on the workpiece. Therefore, it is of particular interest to predict and quantify the appearance of the PLC effect. In this work, a simple method for calculating the PLC effect strength based on stress–strain curves is presented, which can be used to evaluate the appearance of PLC effect serrations. The influence of different strain rates, temperatures, and holding times on PLC effect serrations is demonstrated using a 5083-H111 alloy. To classify PLC effect occurrence, unsupervised clustering was applied to stress–strain data. Additionally, machine learning models, including Gaussian process regression (GPR) and multilayer perceptron (MLP), were employed to predict the PLC effect based on experimental parameters.

Description of the files:

• Rohdaten.xlsx: This file holds the original, unprocessed data collected from your experiment. It is an Excel file and represents the raw information before any feature analysis or processing was performed.

• data_exploded.csv: This file contains the processed results of your feature analysis. It's structured as a table (Pandas DataFrame) and saved as a CSV file. The data includes information about curves and newly generated features, such as cluster IDs, which were likely created during your analysis. The GIT repository is linked in our file upload.