Time Series Sensor Data for Cutting Fluid Analysis from the SmarTaladrine Project

Abstract

This dataset provides multivariate time series data from sensors monitoring a cutting fluid (taladrina) test tank, collected as part of the "SmarTaladrine" project. The primary monitored variables are pH, Temperature, Concentration, and Conductivity. The data spans from December 17, 2024, to April 1, 2025. The dataset includes a raw, preprocessed version of these four variables, which contains missing values as originally observed. Additionally, for each of the four primary variables, separate files are provided showcasing the results of imputing these missing values using five different methods: a pre-trained MOMENT model, a fine-tuned MOMENT model, an LSTM-based Variational Autoencoder (LSTM-VAE), K-Nearest Neighbors (KNN), and a hybrid KNN-Clustering method (HybridKCL). This dataset is intended for developing and evaluating models for cutting fluid analysis, anomaly detection, predictive maintenance, and for benchmarking time series imputation techniques within industrial machining contexts.