MOTOR ANOMALY DETECTOR
BOOTH NO : I2
Category : Manufacturing Technologies
Predictive maintenance (PdM) systems have the potential to autonomously detect underlying motor issues at early stages. Most of the PdM systems that have been proposed so far are based on supervised learning and require hours of manual data collections and annotations. Furthermore, they are mostly made to tackle a single motor issue instead of the multiple motor issues that may occur and are unable to adapt to varying motor speed and load conditions. Thus, they are not viable for industrial implementations. This product presents an unsupervised LSTM autoencoder-based anomaly detection system for electric motors. The system analyzes the vibration and current consumption data from motors to detect anomalies, which is sufficient to account for various motor defects. The system comes with a variety of features that allows users to autonomously collect data, train models and deploy models. This system is easily implementable, and users can keep track of the motor's conditions remotely.
In the manufacturing industry, due to high demand from users, electric motors operate continuously for 24 hours every day. This can cause motors to develop a multitude of defects, which if not treated in time, could lead to machine breakdown, production downtime, and financial losses. Therefore, industries are increasingly looking into predictive maintenance (PdM) systems for motors to overcome this critical issue. PdM systems are both less exhaustive and less prone-to-errors, unlike their traditional counterparts. Aside from that, they have the potential to function autonomously. The majority of the PdM systems that are available require manual real-time data collections and annotations before a fault prediction model can be trained. This can be time-consuming for companies. Furthermore, most of the proposed systems are meant to tackle a single type of motor fault, whereas multiple faults, such as bearing fault, rotor fault, stator issues, etc. may occur in an electric motor.
One of the inventors did an internship program at JCY HDD Technologies. There is clearly a need there in the manufacturing industry.
JCY uses induction motors in their brushing machines and conveyor belts. These motors breaks down quite often and causes production downtime. Companies like JCY could truly use the presented motor anomaly detection system.
This type of anomaly detection system is certainly needed in the manufacturing industry where electric motors are used in many machines. Due to continuous usage, they may suffer from a variety of issues.
As for the solution to the market needs, the single board computers on which the anomaly detection algorithms will be running can be directly attached to the motors of machines.
The system can adapt to varying speed and torque conditions. The users would not need to do any manual data collection or annotation as the system would be unsupervised.
As mentioned earlier, the system comes with a web UI that can allow people to remotely monitor the conditions of motors.
Should the rate of anomalies rise, users will receive a notification regarding their motor’s condition and maintain it before things get worse.
The motor fault prediction system in easier to implement than proposed fault detection systems, it will be cheaper for companies to deploy this system.
Companies such as Siemens, Google and Amazon already have some predictive maintenance systems on the market.
These systems however, are only for data monitoring purposes. They do not detect anomalies like the presented system.
We intend to partner with manufacturing, automobile or robotics companies.
Possible industrial partners:
1. JCY HDD Technologies
2. Alcon
3. DF Automations
KUMERESAN A/L A.DANAPALASINGAM
FAKULTI KEJURUTERAAN ELEKTRIK