Operation
❯
Predictive Maintenance
❯Machine learning-driven approach to identify weak spots in the manufacturing of circuit breakers.
Deep learning technology combined with topological data analysis successfully estimates degree of internal damage to bridge infrastructure
Scope:
Estimate and detect the risk of catastrophic collapse of old bridges.
Goal:
Improve Operation Efficiency
Large Companies
Knowledge representation - Bayesian Network
For:
Process Engineers.
Scope:
Applying Bayesian Networks to root cause analysis of industrial processes.
Goal:
Improve Operation Efficiency
Predictive maintenance of public housing lifts
For:
FM company, residents in public housing
Scope:
Build an AI solution that can predict malfunction in a lift
Goal:
Other
Product failure prediction for critical IT infrastructure
For:
QA engineers, manufacturing line technicians, technical sales
Scope:
Building an AI solution to augment QA engineers
Goal:
Other
Automated defect classification on product surfaces
For:
Sanitary industries
Scope:
Image analytics for water taps in sanitary industries.
Goal:
Other
Analysing and predicting acid treatment effectiveness on bottom hole zone
For:
Manufacturer
Scope:
Mining of oil and gas; digital assistant for analysing and predicting the
effectiveness of acid treatments of the bottom hole zone.
Goal:
Other
Machine learning-driven approach to identify weak spots in the manufacturing of circuit breakers.
For:
Manufacturer of high-voltage (HV) circuit breakers
Scope:
Detecting issues in the manufacturing process that lead to early failure of the
circuit breakers through data mining related to the manufacturing process.
Goal:
Other
Active antenna array satellite
For:
Operators of satellite communication systems
Users of satellite communication systems
Regulation authorities
Space agencies
Scope:
Determine optimal spot beam patterns for communication satellites in order to
react to changing geographic distribution and bandwidth requirements of
terminals.
Goal:
Other
Intelligent technology to control manual operations via video Norma
For:
Industrial enterprises, repair enterprises, repair shops, operators of engineering
products.
Scope:
Tooltip visualization technology (augmented reality) based on technological
process and manual operations control in the assembly, maintenance, and repair
of engineering products.
Goal:
Other
Jet engine predictive maintenance service
For:
Airline industry, Jet engine industry, Airline maintenance industry, cloud-based
AI providers, airline insurance industry
Scope:
Use of jet engine telemetry data to train predictive maintenance algorithms
Goal:
Other
Carrier interference detection and removal for satellite communication
For:
Operators of satellite communication systems
Operators of other communication systems (satellite or non-satellite) that are
potential sources of interference
Users of satellite communication systems
Regulation authorities
Space agencies
Scope:
Machine-learning-based detection, classification and removal of interference
signals for satellite communication systems.
Goal:
Other
Large Companies
Manufacturing and Factories – Predictive Maintenance
For:
Operation, R&D
Scope:
Avoiding unplanned shutdowns in Manufacturing using Machine Learning to predict failure states in equipment.
Goal:
Anticipate Risks, Improve Operation Efficiency
Solution to detect signs of failures in wind power generation system
Scope:
Detect signs of malfunction (failure) in wind power generators
Goal:
Improve Operation Efficiency
Machine learning-driven analysis of batch process operation data to identify causes of poor batch performance
For:
Batch manufacturers such as milk pasteurizers, pharmaceutical makers, paint
manufacturers, etc.
Scope:
Detecting issues in a batch manufacturing process that lead to bad quality
products or longer cycle times for batch processing.
Goal:
Other
Machine learning-driven approach to identify weak spots in the manufacturing of circuit breakers.
For:
Manufacturer of high-voltage (HV) circuit breakersGoal:
OtherProblem addressed
To generate actionable intelligence to improve the manufacturing process for
circuit breakers through mining of manufacturing-related data.
Scope of use case
Detecting issues in the manufacturing process that lead to early failure of the
circuit breakers through data mining related to the manufacturing process.
Description
High voltage circuit breakers are a critical component of an
electric circuit. They have a normal lifespan of 30 y to 40 y.
However, for various reasons, some circuit breakers fail
within 0 y to 5 y of operation. A manufacturer of these circuit
breakers has lots of data related to manufacturing. This data
includes information about production lot size, material of
production, design voltages for sub-components, heater
voltages, date of failure, etc. In general, data related to 49
variables are captured for close to 56 000 circuit breakers
over a lifespan of several years. The manufacturer is
interested to know if there are any weak spots in the
manufacturing process that lead to higher failure rates.
Circuit breakers fail not only due to manufacturing defects
but also due to incorrect operation of the circuit breaker in
the field e.g. applying voltages higher than the design values.
However, operational data of the circuit breakers was not
kept with the manufacturer.
Therefore, the key challenge of this project was knowledge
discovery with a partial data set using machine learning
algorithms.
The data scientists applied various machine learning
algorithms such as decision tree, random forest, support
vector machine, Na�ve Bayes classifier, logistic regression
and neural network, and compared the results of one
algorithm verses the other algorithms. Through multiple
numerical experimentations on data selection and algorithm
hyper parameter tuning, the data scientist team selected the
best algorithms and deduced the key weak spots in the
manufacturing process that are generally associated with
high failure rates. In conclusion, the work provided a set of
five actionable rules where the failure rates jumped
drastically from 0,2 % to 7 %, leading to a 35-fold higher
chance of failure.