Professional Documents
Culture Documents
Instrumentation
1 Introduction
In the current scenario, there are several industries that are located in remote,
precarious areas making it inconvenient for employees. The issues posed include
connectivity to the location, deleterious environment, lack of communication
network, etc. Therefore, it becomes necessary to find an efficient method to tackle
these issues. With the advent of AI (Artificial Intelligence), Cloud Computation
and IoT (Internet of Things), an affordable and robust solution is proposed in
this paper. In the current technology, there exists a DCS (Distributed Control
System) that is responsible for the complete control of a process industry. The
industries have a control room where monitoring of the entire process is handled.
Thus, there is a requirement of hardware engineers to be present on the field.
But, due to lack of safety and effective communication, human presence in such
areas is not advisable.
The solution offered in this paper is lucrative, dependable and inexpensive.
To overcome the lack of connectivity, infrastructure expenses and locomotive
2 Related Work
There exists a virtual lab which provides an HMI through which the controls
engineer can monitor and control the system process. This VM is created on
cloud using IaaS infrastructure. Virtualization is abundantly used in IaaS cloud
in order to integrate physical entities in an ad-hoc manner [1]. Virtual machines
are conventionally used for virtualizing real-world entities and hardware [2][3][4],
which makes them a suitable option for HMI. The virtual lab is developed on
the virtual machine. The emulation of the physical process gives the controls
engineer a better perspective of the operations, making it convenient and easier
to control the industrial process. With the use of this technology, the issues of
safety and employee transportation are solved. The virtual lab is located in a
place appropriate and convenient for the employees to work from. LabVIEW
provides an optimal environment for custom-written software-hardware inter-
facing and GUI (Graphic User Interface), complying to our requirement of HMI
[5]. The GUI presents a logical and visual overview of the essence of the entire
process, makes program development and maintenance effortless, and reduces
programming errors to a large extent. The peripherals of National Instruments
myDAQ proffer input and output streams that are used to acquire data and
reciprocate by sending suitable control signals. This data acquisition device also
caters to our data logging requirements.
LoRa is a long-range transceiver module which can communicate over a range
of 10 to 15 km tested in urbanized locations [6]. It has private network capabili-
ties, hence enabling an ad-hoc network setup [7]. This provides a communication
link between the equipment on site and the controls engineer operating from a
safe and suitable location. LoRa provides a strong 128 bits AES Encryption un-
like other low power long range technologies such as Sigfox [8]. Although there
exist long range technologies such a NB-IoT with high LTE based encryption,
they do not offer private network capability and hence, cannot be setup ad-hoc.
Therefore, LoRa provides a compelling blend of low power, long range, highly
secure and flexible setup as a communication link between the process industry
and the virtual lab. LoRaWAN with repeated stations is deployed to increase the
range of operation. It establishes the networking protocol for LoRa based devices
[9]. LoRa satisfies the need for communication in Cyber-Physical Systems with
added advantages.
Cyber-Physical System, even in its embryonic stage, has many applications
[10][11]. Its potential to diminish other existing and developed technologies
makes it the center of shifting trends in the field of industrial automation. The
integration of computation and communication proficiency with monitoring and
control of industrial processes makes it a niche architecture for automation based
companies [12]. Computing techniques are integrated with real-time objects in
physical world with the use of sensors and actuators which comprise the end
device present on field [13]. This makes it possible for the virtual lab to emu-
late the industrial process giving the controls engineer a better idea so that an
informed decision can be made regarding the control signals to be transmitted.
3 Existing Technologies
There are various technologies in place viz. SCADA (Supervisory Control and
Data Acquisition), DCS and PLC which are generally used for industrial au-
tomation applications. However, these systems lack flexibility which hinders fur-
ther expansion making reconfiguration difficult. SCADA uses complex hardware
units which are expensive and difficult to maintain, thereby decreasing cost-
effectiveness. Also, due to the mission critical nature of SCADA, if one dis-
tributed system fails, all the processes supervised by this system are affected
and there is a considerable probability of loss of sensitive data [14]. Due to the
centralized architecture of SCADA, interoperability is thwarted. Reconfiguration
of the control system is a difficult task and upgrading process is time-consuming.
Often, the customer is confined to a specific hardware manufacturer for legacy
device migration to newer versions. Due to the lack of options, the purchase of
newer equipment is made expensive by the few available manufacturers [15].
4 Framework
4.1 Cyber-Physical Systems
Cyber-Physical System incorporates embedded architecture which obtains data
from the end devices. This data is amplified and fed back to the virtual lab via
LoRa-Cloud infrastructure. The virtual lab responds to this signal by transmit-
ting a suitable control signal via the same communication link.
The virtual lab receives signals via LoRa-Cloud infrastructure from the end
devices located on field. It reverts with suitable control signals which are sent
over the same communication link.
AI driven approach is generally data intensive. The data that is acquired from
the plant is used to make the plant smarter. Furthermore, to enhance user expe-
rience, chat-bots like GEs digital twin are used [19]. Analysis and monitoring of
on-site equipment can be taken forward using computer vision techniques which
also help in detecting faults and hazards. For a power generation plant, load-side
forecasting can be implemented, thereby ensuring sturdy and reliable working
of the plant [20].
5 Methodology
system along with multiple sensors collecting necessary data such as electro-
mechanical parameters of the pump, flow and pressure in the pump valves and
water level in the tank. The data acquired by the sensors is processed by the
embedded system and via the SPI (Serial Peripheral Interface) this data is trans-
ferred to the LoRa transceiver which further sends the data to a LoRa gateway.
Since the setup is in a remote location, it becomes necessary to transmit data
over several kilometers. But since LoRa has a maximum range of 15 km, a Lo-
RaWAN can be setup with repeated stations in order to boost the signal along
the path. The data acquired by the LoRa gateway is then pushed to a cloud in-
frastructure. The cloud platform hosts a virtual machine where various processes
are carried out such as data visualization, data analytics, semantic learning and
data logging operations.
The virtual lab is hosted on the virtual machine. It provides UI (User In-
terface) for a controls engineer to build relevant control logic. This logic is re-
sponsible for the complete control and synchronization of the field devices. The
controls engineer can also analyze various plots on the virtual machine and gain
complete insight to the current parameters of the field equipment. This virtual
machine can be accessed by any person with the credentials for the cloud at any
given time.
The feedback mechanism deployed is such that the virtual lab hosted on the
virtual machine sends a signal to trigger the actuators to perform the directed
action. This triggering signal is sent to the actuator via LoRa architecture. Figure
3 shows a virtual lab implemented on LabView through which the control and
monitoring of the process can be performed real-time. Virtual lab is a software
platform that synchronizes all the field processes.
Any hardware upgrades can be implemented easily by just replacing the end
device as opposed to the replacement of the entire system. Upgrades to the
virtual machine are purely of software domain, making it versatile and allowing
it to be hosted on any machine. It can be migrated from one platform to another
with ease. Put together, it forms a reliable Cyber-Physical System.
Fig. 4. (Left) Graph of Water Flow vs Time, (Right) Graph of RPM of the Pump vs
Time
This technology enables an easily accessible remote system which allows con-
trols engineers to supervise a process on site from a location of their choice. As
this replaces the existing expensive and complex mechanism in place, it proves
to be a cost effective and simple solution. Furthermore, it cuts down on labour
due to complete automation in the process industry.
Figure 4, 5 show graphs with water flow, speed (rpm) and current plots
against time with data being logged every 5 minutes. It can be inferred from
the graph that when fault occurs, the current shoots up rapidly due to sudden
increase in rpm. Meanwhile, the water flow in the valve abruptly decreases,
thereby triggering reverse response. Here, the data logging time is shown from
14:15 and fault occurs at 14:20. This is depicted at 300 seconds on the graphs.
When the fault occurs, the data from the previous 5 minutes is saved and the
fault data begins to log. The virtual lab reciprocates by sending a suitable control
signal to turn off the pump, hence preventing any damage to it. Also, Control
engineers can set the time duration for logging data in order to optimize the usage
of memory. If logging time duration before the fault occurs is more, memory
consumed increases. Therefore, it becomes necessary to find a balance between
the amount of data logged and memory consumption.
7 Future Work
References
1. Dillon, Tharam, Chen Wu, and Elizabeth Chang. ”Cloud computing: issues and
challenges.” Advanced Information Networking and Applications (AINA), 2010
24th IEEE International Conference on. Ieee, 2010.
2. Levis, Philip, and David Culler. ”Mat: A tiny virtual machine for sensor networks.”
ACM Sigplan Notices. Vol. 37. No. 10. ACM, 2002.