At AquaML, our predictive maintenance approaches employ condition monitoring across different data sets and parameters to predict failure. By combining multiple variables and processing them through our powerful and pre-trained Machine Learning Model, we are able to predict when challenges and failures might occur, with a higher degree of confidence and fewer false positives. On Average, a local government spends $30B on O&M of Water Treatment plants. The majority of the cost is attributed towards the machine breakdowns and repairs
How we solved the problem ?
Analyzed the existing data available and provided by the plant including data from PLC
Augmented the data by using Electronic Signature Analysis and Energy Consumption
Our ML based Feature Engineering to optimize and normalize the data
Leveraged our powerful, pre trained 80 plus ML models (including Classifiers, Regressors, Neural Networks and NLP) for predictions and recommendations
We have technology that allows us to listen and capture the machine vibration and sounds, improving our ability to predict and prevent machine breakdown
Outcome And Impact
- By using 3 months of data, we were able to provide insights into some low hanging fruits around data quality, the co-relation between energy consumption, and sludge and machine breakdown and/or sub optimal performance
- Augmented with near real time data, applied as well as reinforced machine learning, AutoML offered insights on preventive maintenance, reducing energy consumption identifying opportunities to save up to 8% of operating expenses in respective cost category
Given the pressing need to embrace cleaner and more efficient processes, our platform has proven transformative capabilities of AI and ML, alongside the implementation of the autonomous operation through IIOT. The efficacy, impact and value generation of these solutions is dependent on the current automaton of your plant, availability of reliable data towards the full realization of the benefits.