MUMBAI, India, March 13 -- Intellectual Property India has published a patent application (202631014087 A) filed by C. V. Raman Global University, Bhubaneswar, Orissa, on Feb. 9, for 'an iot and deep learning based system and method for predictive maintenance using time-series sensor data.'
Inventor(s) include Raj Vikram; Gopinath Padhy; Sankalp Saurav Dash; Subhrakant Sahu; Roshan Kumar Dash; and Sanchita Mohanty.
The application for the patent was published on March 13, under issue no. 11/2026.
According to the abstract released by the Intellectual Property India: "A system and method for predictive maintenance of industrial equipment utilizing Internet of Things (IoT) sensors integrated with deep learning neural networks, specifically Long Short-Term Memory (LSTM) architectures combined with edge computing and cloud-based analytics platforms. The invention provides real-time monitoring of critical equipment parameters including temperature, vibration, current, voltage, and rotational speed through multi-modal sensor interfaces, with data acquired via NodeMCU ESP32 microcontrollers executing optimized communication protocols with TLS 1.2 encryption and MQTT messaging standards. The system preprocesses acquired sensor data through normalization, temporal segmentation into sixtytimestep windows, and synthetic minority oversampling techniques to address class imbalance in industrial datasets. The predictive maintenance framework integrates edge intelligence with serverless cloud architecture utilizing Amazon Web Services infrastructure including AWS IoT Core, Amazon Kinesis for real-time streaming at 1000+ messages per second with less than 200 milliseconds latency, AWS Lambda for serverless data processing functions, and Amazon SageMaker for hosting LSTM models with automatic hyperparameter tuning and GPU-accelerated inference capabilities. The innovation achieves 92.71% prediction accuracy with 0.91 precision and 0.86 recall in failure detection compared to conventional machine learning approaches including random forests and support vector machines, while providing human-centric alert notifications with contextual information, severity levels, and actionable maintenance recommendations that reduce unplanned downtime by 41% and deliver $1.2 million annual cost savings for medium-sized industrial plants by shifting from reactive to proactive maintenance paradigms."
Disclaimer: Curated by HT Syndication.
