Waking Up On Time – And Going To Sleep in Clean PJs - Thanks to Electric Motors
The electric motor was invented in the 19th century. This was a critical invention because physical work can be converted from electrical energy. If you woke up to a vibrating alarm on your mobile this morning, you can thank a little motor spinning inside your phone to synthesize a vibration in your phone. Another common place that you will find an electric motor in your laundry washing machine and dryer (one of the very first application of the electric motor). The invention of the electric motor was a key component of the industrial revolutions (we are in the 4th), where it impacts manufacturing, logistics, and warehousing. And today, electric motors has become a national interest in the form of electric vehicles, made hugely popular by companies such as Tesla.
Electric Motor Generating Bad Vibes
Electric motors comes in two major flavors: alternating current (AC) and direct current (DC). Both types of motors are used in industrial machinery. The AC variety usually provides power (drilling, compressor, cooling), whilst the DC variant usually provides position (high precision robotics, belt, pick-and-place). Inside the electric motor, a component called the bearing is critical to the performance to the electric motor. This critical component can easily be impacted by 1) manufacturing defects in the bearing 2) defects in mounting the bearing into the electric motor or 3) gradual bearing damage during operation of the electric motor. If the bearing in the electric motor does not perform up to specification, this can lead to vibration. Vibrations are bad because 1) reduction in quality of the product being produced by the machine with a vibrating electric motor 2) noise generated from a vibrating motor distracts & wears down workers and 3) eventual failure of the industrial machine, caused by electric motor vibration. How does one reduce the negative impact of a vibrating motor?
An AC Motor, Sensors, Cloud Gateway, IoT, AI - Industry 4.0 on a Table
In a recent joint demo between Hitachi and Oracle, we showed real life examples of how Oracle’s IoT Cloud & Oracle AI for Manufacturing Cloud, implemented using Hitachi’s Consulting industrial expertise, can provide an Industry 4.0 solution today.
In this demo, we start with an electric motor, extracted from an industrial machine. Since the AC motor is really the heart of the an industrial machine, we wanted to monitor and analyze it for anomaly. We start our Industry 4.0 deployment from there. In the picture below, a 3-axis motion sensor is magnetically attached to the AC motor. The sensor transmits x,y,z data via a built-in WiFi to the gateway. The gateway then sends the sensor data to the cloud (to both Hitachi & Oracle). Once in the cloud, we can use powerful, yet easy to use analytics and AI.
Is Your Machine Vibrating Badly? Monitor with Oracle IoT Asset Monitoring Cloud
The 3-axis vibration sensor is magnetically attached to the AC motor. Being able to quickly attach a sensor (thanks to magnets) is key to make instrumenting an asset (the motor) easy. This particular sensor can sample vibrations from the electric motor at a rate of roughly 4 KHz (4,000 samples per second). The higher the frequency, the better the data. Once the data arrives in the Oracle IoT Asset Monitoring Cloud, how does it handle this high stream data rate? By using the popular Apache Kafka pub/sub framework to accept, process, and store the high bandwidth incoming stream of IoT sensor data. Analytics is handled by Apache Spark, the popular framework for Big Data analytics (in this case, lots of little data that is unstructured) analytics. With these as foundation, combined with the beautiful and easy to use app provided by Oracle IoT Asset Monitoring Cloud, you can see insights on all of your assets instantly.
Anomaly Detection Built Into Oracle IoT Asset Monitoring Cloud
Shown below is a modified time series representation of data from the sensor (a big chunk of the analysis is performed in frequency domain). Oracle IoT Asset Monitoring Cloud not only can accept a high rate of data input, it also has built in analytics to look at the raw sensor data and detect anomaly in the data. For example, if a bearing is bad in the AC motor, it will vibrate in such a manner that the x-axis sensor will trigger an anomaly alert. In order to detect that a certain vibration characteristic is bad, the anomaly detection system needs to be trained. There are two methods to train the anomaly detection model in Oracle IoT Cloud – 1) automatic anomaly training or 2) user-defined anomaly training. Automatic anomaly training is statistics based – just provide a set of normal sensor data collected from the past, define a training window, then set the normal operating standard deviation. User-defined anomaly training is human (usually very seasoned and can hear problems from a mile away) judgement based – the human trainer looks at past sensor data, highlights the data that is anomalous, and the highlighted region becomes the pattern to trigger an anomaly.
Is The Vibration In Your Machine Making Bad Products? Analyze Using Oracle AI Apps for Manufacturing
You know that vibration in industrial machinery can lower quality of the final product, increase noise level in the factory, and be a harbinger of machine failure. But how do you know exactly the impact of vibration to your final products? Using Oracle AI Apps for Manufacturing, you can find out “how many products had poor quality when the vibration level exceeded upper limit”. In this example, the final product is a GB006 Gearbox. Using AI Apps for Manufacturing, 13.42% of the final product (Gearbox GB006) had chaffing issues, which usually leads to poor quality product. Because of the data collected from both the vibration sensor and the quality system, AI Apps for Manufacturing found that vibration is a key contributor, with an 89.90% confidence.
IoT & AI Discovered a Bad Bearing
With IoT (3-axis vibration sensor attached to the AC motor) and AI (collecting data, detecting anomaly), a problem in a deeply embedded bearing can be discovered BEFORE production is unexpectedly halted. Before IoT & AI, it was difficult to detect that an industrial machine is beginning to malfunction. If you suspected that something was wrong, production has to be stopped, the machine taken apart, and hopefully you found the problem. But now with IoT & AI, problems in industrial machine can be monitored in real time and actions taken before product is negatively impacted.
Conclusion
The electric motor is the workhorse and heart of industrial machines, found in both old and state of the art manufacturing facilities. Like a heart, it is critical to monitor it fully to see how it is doing and predict if it will fail. IoT can be used to sample data from the electric motor. AI can be used to detect anomaly and reveal impact of vibrations in the electric motor on quality of the final product. Having the expertise of Hitachi Consulting, building a solution on top of Oracle IoT Cloud & Oracle AI Apps for Manufacturing, will enable you to quickly adopt Industry 4.0 benefits.