Machine Learning In Predictive Maintenance
Arnab Biswas (~arnabbiswas1) |
Our world is full of equipments. For example:
- Aircraft consists of different equipments
- HVAC (Heating, Ventilation, Air Conditioning) equipments consist of various parts
All these equipments or their parts meet failures and hence need maintenance. Thus maintenance is a big industry by itself. Now, predictive maintenance is different from traditional reactive or proactive maintenance. In predictive maintenance, the equipment's future health is predicted based on it's historical health. It avoids unplanned downtime and fully utilizes any part's life.
Although Perdictive Maintenance was done using rule based approach, with advent of IoT, Big Data and Data Science, Machine Learning is being used to address Predictive Maintenance.
In this talk, I am going to discuss how Machine Learning can be used for predictive maintenance. The outline of the talk would be as followed:
- Types of Maintenance (2 Minutes)
- Why and how Predictive Maintenance is better? (1 Minute)
- Goal's of Predictive Maintenance (1 Minute)
- Example use cases for Predictive Maintenance (1 Minute)
- Data Science Life Cycle for Predictive Maintenance
- Introduction (1 Minute)
- Converting Business Problem to a Data Science Problem (4 Minute)
- Data Requirement (1 Minute)
- Data Collection (1 Minute)
- Data Validation (1 Minute)
- Data Pre-Processing (1 Minute)
- Feature Engineering (4 Minutes)
- Relevant ML Algorithms & Architecture (1 Minute)
- Cross Validation (3 Minutes)
- Evaluation Metrics (1 Minute)
- Model Deployment (1 Minute)
- Model Maintenance (1 Minute)
- Questions (5 Minutes)
Along with the different steps in the data science life cycle, associated libraries & tools from Python eco-system will also be discussed.
Note: This talk strictly utilizes information available in the public domain and doesn't use any confidential or proprietary information.
- Basic knowledge of Machine Learning
Arnab Biswas is a Data Scientist working in the HVAC Industry. He has 15 years of experience in Software Development in the Telecom, Networking & HVAC industry. He has developed highly scalable, distributed, fault tolerant softwares to manage/provision IoT devices. For last 3 years he is extensively working in Data Science/Machine Learning.
He has volunteered/worked for different Non Profit Organizations in India helping them address their Data Science related.
He has contributed to StackOverflow, open source softwares like dask, jolokia, hawtio etc.