How the Industrial IoT takes production line monitoring BEYOND AUTOMATION

About Webinar

Manufacturers constantly face pressure to continuously improve operational performance, avoid sudden breakdowns of machines on the shop floor, minimize production loss, and reduce operational costs. However, the biggest barrier to operational excellence is the non-uniform standards of machines and equipment which are being used on the shop floor.

In this webinar, we are going to address many questions such as:

• How can a supervisor monitor machine performance remotely?
• How to predict machine failure even before it breaks down?
• Is it possible to have all the information on a single canvas which a floor manager can access from anywhere, anytime?

Our experts will also show how IoT can bring value to the shop floor and production line monitoring.

Webinar agenda

• Pitfalls of traditional manufacturing processes and how to overcome
• What Industry 4.0 really is and why it’s a game changer
• IT/OT Convergence: The key mantras in smart manufacturing
• The role of IoT and data analytics in the manufacturing industry
• Current market trends and value 
• Smart factory: The evolution not revolution
• Demo and review of smart production line monitoring
• Q&A session

Questions & Answers

The following are the answers to the questions that were asked during the live webinar.

Question 1: How are you able to use a Deep Learning model to determine which of the inputs caused a change in the prediction? I've read that this is incredibly difficult to do, since there are too many layers between inputs and predictions.

Answer 1: It is not easy to track which inputs had an effect on the prediction in deep learning specifically because it has numerous non-linear layers between the input layer and output layer. However, by using other ML algorithms we can track the effect of a particular input on the output prediction.

Question 2: What are the specific protocols and communication standards, by name, that are used by the sensors to communicate to the IoT Gateway?

Answer 2: Sensors will not use any protocol. Sensors send analog/digital signal in the form of voltages. A PLC will read digital and analog inputs from various sensors, execute a user defined logic program, and write the resulting digital and analog output values to various output elements like hydraulic and pneumatic actuators, solenoid coils, indication lamps, etc.

A PLC supports protocols and communications standards.
Protocol support for PLC (Sensor/Machine Data) to gateway:
PROFIBUS, Modbus RS-232, RS-485, TCP/UDP, MTConnect, BACnet, serial protocol
Protocol support for gateway to the cloud:
MQTT, CoAP, AMQP, WebSocket, WiFi, LAN

Question 3: Modbus has been "customized" so much how do you handle this w/ all the different machines?

Answer 3: Configuration of Modbus on different Modbus supported machines: Modbus uses RS-485 or RS-232. Modbus TCP uses Ethernet. You will need to pick the model that matches the electrical interface of the equipment you want to connect. When using a gateway to interface, a Modbus device to a non-Modbus network, you need to get documentation from the equipment manufacturer that describes the available registers and how to address them. Modbus protocol does not provide a means for registers to automatically identify themselves. Control solutions cannot determine this information for you. You must consult the equipment manufacturer or online document of Modbus supported device.

Modbus requires you to know or define the baud rate, character format (8 bits no parity, etc.), and slave ID (slave address, unit number, unit ID). A mismatch in any of these will result in no communication.

Modbus TCP requires you to know or define IP addresses on the network.

Question 4: How much historical data is needed for these models to function? Put another way, how long from when I start capturing the data to when it will be useful?

Answer 4: Answer to this question may vary according to data scientist's views on the model so it can be said that there is no right answer to how much amount of data you will need. But you can take data approximately around 4-5 times more than the number of variable (feature) you have.

Question 5: What are the common input sensors used in the Industrial IoT? Is there application for Audio or Voice?

Answer 5: Common input sensors used in the Industrial IoT are as mentioned below:
Temperature, thermistors, thermocouples, RTDs, vibration, gas, light, level, flow, current/voltage detector, rotary encoder, ultrasonic thickness gauge, and many more. There are many use cases of NLP and cognitive services that industries widely use. To know more about different types of sensors used in building a smart factory, click here.