soft-sensor

Python package that provides predictive models for fault detection, soft sensing, and process condition monitoring.

Soft sensor modelling using multiple machine learning algorithms

This repository contains code for an LSTM-based soft sensor designed to estimate nitrate concentration in aquaponics ponds, offering real-time insights into water quality for improved management and sustainability.

Pharma 4.0 Bio-Twin console featuring real-time IIoT simulation, MSPC latent soft sensors (PCA), deterministic Groq LLM diagnostic agent loops, and automated 5-minute/instant interrupt SMTP email notification alerting pipelines.

GradeShift PrimePath — uncertainty-aware, human-authorized commercial-disposition decision support for polyolefin grade transitions.

ML soft sensor predicting real-time biomass (OD) in a methanotroph bioreactor from off-gas signals. Gas mass-balance features (OUR, CER, MUR, RQ), XGBoost/Ridge/PLS benchmarked with leave-one-batch-out CV, SHAP interpretability, and live prediction on a running reactor.

MATLAB code for simulating and analyzing the static performance of Variable Stiffness Soft Pneumatic Sensing Chambers (VSSPSCs).

Planta de flotación real (Kaggle, mineral de hierro): qué se puede predecir de la sílice del concentrado y por qué los datos no permiten recetar setpoints de reactivos.

Reproducible deep-learning image-classification pipeline for industrial soft sensing. Originally built for sludge-cake quality monitoring at DC Water Blue Plains AWWTP. Six architectures (FastViT, EfficientNet, MobileNet, EfficientFormerV2, DeepTEN-ResNet, sparse-AE CNN) compared with multi-seed statistics on Modal cloud GPUs.

Soft sensor para una estación de evaporación de múltiple efecto. Python, SQL, SciPy.

Projeto de ciência de dados em construção: soft sensor industrial para estimar pureza de produto via séries temporais e scikit-learn. Desenvolvido módulo a módulo (EDA → features → modelagem → produção).

Effluent BOD/COD soft sensors with conformal prediction intervals and a compliance risk index, tested in time order on the UCI Water Treatment Plant data

Locally-weighted partial least squares (LW-PLS) for tidymodels: just-in-time regression and classification with a fast C++ engine, covariance-based LW-PLS (CbLW-PLS), and parsnip, dials and tune integration.

Machine learning-based soft sensor for cross-plant production estimation, model transfer, and domain shift analysis.

A soft-sensor benchmark for evaluating adaptive scaffolding in LLM-based programming tutors (Sensors, MDPI, 2026)

Laboratorio virtual para una planta de flotación: modelo + agente de turno, evaluado con datos reales. Proyecto personal.

Machine learning-based soft sensor for predicting debutanizer column product composition using industrial process variables and regression models.

Physics-Informed Neural Network (PINN) bioprocess digital twin & gray-box soft sensor for 20L bioreactor state estimation

Industrial ML with physics: fault diagnosis and a soft sensor, trained on validated simulators

Sensor virtual de sílice en flotación de hierro con datos reales de planta: 3 filtraciones cerradas, persistencia como línea base, holdout abierto una vez y bootstrap por días.