Register for free to
view the full innovation profile.
Foresense is a prognostics tool that solves range of problems involving flows of data and the need to predict future conditions and outcomes. It can forewarn the user in real time.
About
Foresense involves two parts.
One is a fault prediction-forecasting pipeline. The pipeline consists of three primary components:
· Pre-processing – This component extracts data, cleans and transforms it, and converts it into trainable datasets
· Training – This component holds the Temporal Fusion Transformer, that trains on the data by creating a class of the TFT model
· Inference – This component extracts the trained model and runs customised metrics on it to assess its performance
The pipeline can be run on 2 settings:
· On entire data
· On new data only
The pre-processing component automatically checks for new data using incremental load processing and based on the flag returned, runs the script on the selected data.
The other in a user-friendly platform.
The pipeline ingests time-series data from multiple Delta tables that are updated daily. Using a standard cluster data is stored and processed on an Apache Spark based Databricks platform hosted on Azure. The data-processing pipeline is developed to operate on a cluster consisting of 8 CPUs and 1 GPU with a total memory of 56GB across two quad-cores distributed between the worker and driver nodes. Additionally, the Temporal Fusion Transformer Model uses the data iteratively on 12 workers for training and inference.
Key Benefits
Foresense offers significant benefits to target markets in various industries by anticipating potential failures and malfunctions in equipment, making it the ideal solution for manufacturing companies.
· Reduced costs: By predicting faults before they occur, organisations can proactively schedule maintenance and repairs during planned downtimes and minimize unexpected equipment failures reducing overall maintenance costs and increasing operational efficiency.
· Improved reliability: Pre-emptive maintenance of equipment allows organisations to uphold improved reliability and performance. This advocates better product quality and enhanced customer satisfaction.
· Enhanced safety: Proactive fault prediction in critical systems and infrastructure safeguards against safety risks, preventing accidents and hazardous situations. By addressing potential safety issues beforehand, organizations can foster a safer work environment for employees and reduce the chances of regulatory violations or legal repercussions.
· Data-driven decision making: The pre-processing component provides actionable information and valuable insights from historical data to support data-driven decision making. These insights can also be utilised to optimise operational parameters and organise maintenance schedules efficiently.
Applications
Fault prediction can serve a wide array of markets apart from railway companies.
· Manufacturing: Industries in the manufacturing domain including automotive, aerospace, electronics, consumer goods, rely heavily on machinery and hardware-based systems. Predictive maintenance solutions can help manufacturers minimize downtime, optimize production schedules, and ensure consistent product quality.
· Energy: Power plants, oil and gas facilities, water treatment and waste management plants also require continuous monitoring and maintenance to ensure reliable operation. Fault-detection and early fault prediction solutions can successfully prevent costly outages.
· Mining: Mining companies and resource extraction industries rely on heavy machinery and equipment to extract, process, and transport materials. Fault prediction pipelines can help mining companies minimize equipment downtime, optimize production schedules, and improve worker safety in challenging operating environments.
· Infrastructure: Smart city initiatives and infrastructure projects involve the deployment of sensors and IoT devices to monitor and manage urban infrastructure, such as transportation systems, public utilities, and municipal services. Fault prediction pipelines can help municipalities and urban planners optimize infrastructure maintenance, improve service delivery, and enhance the resilience of smart city systems.