AI · Predictive Maintenance · Academic project

Motor Health Prediction

Predictive maintenance for electric motors from current and voltage signal parameters.

PythonMachine LearningData Analytics

Current & Voltage

Input Signals

Machine Learning

Approach

Predictive Maintenance

Application

Source

Overview

Electric motors are critical components in industrial systems, and early fault detection can significantly reduce downtime and maintenance costs. This project explores the use of machine learning to predict motor health by analyzing electrical parameters such as current and voltage, eliminating the need for additional sensing hardware.

By combining data preprocessing, feature engineering, and supervised learning techniques, the project demonstrates how electrical signal analysis can support predictive maintenance and enable more informed maintenance decisions.

Problem

Unexpected motor failures can lead to costly downtime and reduced operational efficiency. This project investigates whether electrical parameters such as current and voltage can be used to identify motor health conditions using machine learning, providing an economical alternative to traditional vibration- or thermal-based monitoring approaches.

Process

Signal acquisition

Motor current and voltage measurements were collected from different operating conditions to create a dataset representing both healthy and faulty motor behavior.

Feature extraction

Statistical features were extracted from the electrical signals to capture meaningful characteristics that distinguish healthy operating conditions from potential fault scenarios.

Classification

Supervised machine learning models were trained and evaluated to classify motor health conditions based on the extracted electrical features, with performance assessed using standard evaluation metrics.

Maintenance signal output

The predicted motor health status was presented as an interpretable output that could support preventive maintenance planning and operational decision-making.

Results

Successfully demonstrated that motor current and voltage measurements can be used to classify motor health conditions using supervised machine learning techniques.

Feature engineering improved the model's ability to distinguish healthy and faulty operating conditions by extracting meaningful characteristics from electrical signals.

Validated the potential of electrical-signal-based monitoring as a cost-effective approach for predictive maintenance without relying on dedicated vibration or thermal sensors.

Learnings

Electrical parameters contain valuable information for identifying motor health conditions when combined with appropriate feature engineering and machine learning techniques.

Data quality, feature selection, and class balance significantly influence classification performance in predictive maintenance applications.

Building an end-to-end machine learning pipeline reinforced the importance of preprocessing, feature engineering, model evaluation, and result interpretation in engineering applications.