SMART Predictive Maintenance for

Industrial Motors (SMART-PDM)

Major Project 1

Presented by: Ngum Breayon Kwe

Vishal Yadav
7th sem 4th Year

Supervision of: Prof. Pallavi Gupta

SSET
 Presentation Highlights
This presentation will expand on:

⮚ Introduction to Predictive Maintenance

⮚ Problem Statement
⮚ Sensor Network Application
⮚ Development
⮚ Advantages of this project
Introduction
to Predictive
Maintenance
01
AI & Machine Learning Model Predictive Maintenance for MEMS
Motors
• Predictive maintenance for motors is a proactive
approach to maintenance that uses data, analytics,
and machine learning to predict when equipment,
particularly electric motors, is likely to fail.
Predictive
• The goal is to perform maintenance activities just
in time, before a failure occurs, optimizing the Maintenance
performance and lifespan of the motor while
minimizing downtime and reducing overall
maintenance costs.
 Condition Monitoring:
Data Collection: Utilize sensors, IoT devices, and other monitoring tools
to collect real-time data on equipment performance and health.
Continuous Monitoring: Regularly monitor the collected data to
establish a baseline for normal operating conditions.
Baseline Establishment: Create a baseline or reference profile based on
historical data to understand the typical behavior of the equipment.

Data Analysis and Predictive Modeling:

Predictive Analytics: Apply advanced analytics, machine learning, and
statistical methods to analyze the collected data.
Predictive Modeling: Develop predictive models that leverage historical
Key
Activities
and real-time data to forecast when equipment is likely to fail.
Fault Detection: Implement algorithms to detect and diagnose faults in
the equipment, providing insights into potential failures before they
occur.

Alerts and Maintenance Action:

Automated Alerts: Implement a system that automatically generates
alerts or notifications when the predictive maintenance system identifies
potential issues.
Severity Assessment: Classify alerts based on the severity of the
predicted issue, helping prioritize maintenance actions.
Problem
Statement
02
AI & Machine Learning Model Predictive Maintenance for MEMS
Motors
 Industrial machinery and equipment are critical
assets for organizations, and unplanned downtime due
to equipment failures poses significant challenges,
leading to production disruptions and increased
maintenance costs.
Traditional preventive maintenance practices often
result in scheduled downtime and maintenance tasks Problem
regardless of the actual condition of the equipment,
leading to inefficiencies and unnecessary costs. Statement
The challenge lies in transitioning from reactive and
preventive maintenance to a predictive maintenance
strategy that utilizes data-driven insights to anticipate
and address potential failures before they occur.
 ⮚Patent US 2021/0049839 A1 - AUTOMOTIVE PREDICTIVE
MAINTENANCE

Shortcomings:

●Insufficient Data Processing: The data processing capabilities are

rudimentary, leading to delayed or inaccurate predictions.
●Lack of Advanced AI Integration: The system does not leverage advanced AI
or machine learning models, resulting in less accurate and less reliable failure
predictions.

⮚Patent No.: US 7.457,763 B1 - PREDICTIVE MAINTENANCE SYSTEM Literature

Shortcomings:
Review
●Basic Anomaly Detection Algorithms: The anomaly detection algorithms are
simplistic and often fail to identify complex patterns indicative of impending
failures.
●Scalability Issues: The system struggles to scale efficiently with increasing
data volumes, making it less suitable for larger MSMEs with extensive equipment
fleets.
●User Interface Limitations: The user interface is not intuitive and lacks
comprehensive visualization tools, making it difficult for maintenance teams to
interpret data and take timely action.
 ⮚Patent Number: US 5,963,884 - PREDICTIVE MAINTENANCE SYSTEM
WITH REMOTE MONITORING:

Literature
Shortcomings:

●Basic Predictive Models: The predictive models used are basic and not

Review
tailored to specific equipment types, reducing the accuracy and reliability of
predictions.
●High Implementation Costs: The system requires significant upfront
investment and maintenance costs, making it less accessible to smaller MSMEs.
●Data Privacy Concerns: The remote monitoring aspect raises concerns about
data privacy and security, which are not adequately addressed in the patent.
 The concept of predictive maintenance has significantly evolved with
the integration of AI and IoT technologies, offering a more efficient
alternative to traditional preventive maintenance strategies. Research [1, 2]
has shown that predictive models can substantially reduce downtime and
maintenance costs, making them increasingly attractive for industrial
applications.

AI in Predictive Maintenance has been the subject of extensive Literature

Survey
research. For example, [3] examines the application of artificial neural
networks (ANNs) for early fault detection in motors, demonstrating how
machine learning models can be trained on historical vibration data to
accurately predict motor failures. Despite the success of such AI
implementations in large-scale industries [4], there is a notable gap in the
literature concerning the adaptation of these technologies for Micro, Small,
and Medium Enterprises (MSMEs). MSMEs often face unique challenges,
such as limited technical expertise and constrained financial resources,
which are not sufficiently addressed in existing studies.
 Sensor Technologies are critical components of predictive
maintenance systems. Studies [5, 6] have highlighted the effectiveness of
vibration and temperature sensors in detecting motor faults. These sensors
provide the necessary data for predictive models to analyze and forecast
potential failures. However, the focus of much of this research has been on
large enterprises, which typically have the resources to implement and

Literature
maintain advanced sensor networks. This leaves a gap in the literature
regarding the application of sensor technologies in MSMEs, where there is
a need for more cost-effective and energy-efficient solutions.

Given these gaps in the literature, our study proposes a novel Survey
approach that integrates STM32 microcontrollers with NanoEdge AI
technology. This approach is designed to meet the specific needs of
MSMEs by offering a scalable, energy-efficient, and cost-effective
predictive maintenance solution. The proposed system leverages advanced
AI capabilities while being accessible and affordable for smaller
enterprises, thereby addressing the limitations identified in previous
research.
 Key Faults
Key Occurring Faults
Symptoms

Vibrations and Bearing Rotor Faults Cavitation (for Misalignment Mechanical

Noise Failure Pump Motors) Imbalance

Overheating Stator Faulty Contamination Overheating Loose Insulation Voltage

Faults Windings Connections Breakdown Fluctuations
Sensor Network
Application
03
AI & Machine Learning Model Predictive Maintenance for MEMS
Motors
Project
Implementation
Methodology
● In the design of this project, details the use of the STM32
microelectronics board and STM32Cube IDE for sensor interfacing and
programming. Industrial machinery often face faults which results in
noise and vibrations, thus this product is designed to detect the vibrations
of the motor during normal and abnormal working phases.

● This product employs machine learning algorithms using the Artificial

Intelligence Edge Studio, which classified the data gotten from sensor
nodes into Data Sets and categories as Normal and Abnormal data sets.

● The NanaEdgeAI studio provides a User platform for monitoring the state
of the Motor. In case of multiple Motors a User Interface will be built for
monitoring of every motors state.
Development 04
AI & Machine Learning Model Predictive Maintenance for MEMS
Motors
 Software and Hardware
Development

This project will be achieved by analysing data from sensor nodes deployed on the motors through the
integration of multiple AI anomaly detection and classification algorithms. The sensor nodes, equipped
with temperature and vibration detection capabilities, enable real-time anomaly detection and
classification.

The project incorporates both hardware and software signalling mechanisms to monitor the state of the
motors and products:

• Hardware signalling includes visual indications via LEDs and sound alerts for immediate attention.
• Software monitoring is supported by an online database that stores motor and product state
information, and a cloud-based platform for graphical and numerical data visualization using
platforms such as Arduino IoT cloud, AWS IoT or Azure IoT Hub.
Software and Hardware
Development

Prototype Casing
Advantages 06
AI & Machine Learning Model Predictive Maintenance for MEMS
Motors
Advantages

• Cost-Effective: The system is designed to provide advanced predictive maintenance and anomaly
detection at a lower cost compared to traditional solutions, making it accessible for a wider range
of applications and industries.
• High Accuracy: Integration of highly accurate anomaly detection and classification algorithms
ensures reliable and precise identification of potential issues, reducing false positives and
improving maintenance schedules.
• User-Friendly Interface: The solution includes both cloud-based and on-device hardware
signalling for visualization and control, providing users with intuitive and easy-to-use tools for
monitoring and managing motor performance.
• Extendable Product: The system is designed to be extendable, allowing for the integration of
additional sensors for more comprehensive monitoring and broader application possibilities.
• Advanced Communication: Utilizes multiple master to multiple slave wireless communication
between sensor nodes and the central processor, enhancing flexibility and scalability of the system
while maintaining efficient and reliable data transmission.
References 07
AI & Machine Learning Model Predictive Maintenance for MEMS
Motors
 References
1.Aboshosha, Ashraf, Ayman Haggag, Neseem George, and Hisham A. Hamad. "IoT-based data-driven predictive maintenance relying on
fuzzy system and artificial neural networks." Scientific Reports 13, no. 1 (2023): 12186.
2.Yang, Chanseung, Sang Bin Lee, Geunik Jang, Seongnam Kim, Gyukyung Jung, Junghoon Lee, Sangwook Shim, Young Kab Lim,
Jinpyo Kim, and Sungbong Park. "Starting current analysis for condition monitoring of medium voltage induction motors in the steel
industry." In 2017 IEEE Industry Applications Society Annual Meeting, pp. 1-9. IEEE, 2017.
3.Scalabrini Sampaio, Gustavo, Arnaldo Rabello de Aguiar Vallim Filho, Leilton Santos da Silva, and Leandro Augusto da Silva.
"Prediction of motor failure time using an artificial neural network." Sensors 19, no. 19 (2019): 4342.
4.Scalabrini Sampaio, Gustavo, Arnaldo Rabello de Aguiar Vallim Filho, Leilton Santos da Silva, and Leandro Augusto da Silva.
"Prediction of motor failure time using an artificial neural network." Sensors 19, no. 19 (2019): 4342.
5.Alsheryani, Reem M., Sultan S. Alkaabi, Sultan Saif Alkaabi, Ahmed M. Aldhaheri, Fahed I. Khouri, Saeed I. Alharmoodi, Tarik T.
Shadid, and Abdulhadi S. Alhajeri. "Applying artificial intelligence (AI) for predictive maintenance of power distribution networks: A
case study of al ain distribution company." In 2019 International Conference on Electrical and Computing Technologies and Applications
(ICECTA), pp. 1-5. IEEE, 2019.
6.Zhang, Weiting, Dong Yang, and Hongchao Wang. "Data-driven methods for predictive maintenance of industrial equipment: A
survey." IEEE systems journal 13, no. 3 (2019): 2213-2227.
7.Çınar, Zeki Murat, Abubakar Abdussalam Nuhu, Qasim Zeeshan, Orhan Korhan, Mohammed Asmael, and Babak Safaei. "Machine
learning in predictive maintenance towards sustainable smart manufacturing in industry 4.0." Sustainability 12, no. 19 (2020): 8211.
8.Yetgin, Asım Gökhan. "Effects of induction motor end ring faults on motor performance. Experimental results." Engineering Failure
Analysis 96 (2019): 374-383..
9.Glowacz, Adam. "Fault diagnosis of single-phase induction motor based on acoustic signals." Mechanical Systems and Signal
Processing 117 (2019): 65-80.
10.Ojaghi, Mansour, Mahdi Sabouri, and Jawad Faiz. "Analytic model for induction motors under localized bearing faults." IEEE
Transactions on energy conversion 33, no. 2 (2017): 617-626.
Thank You