[quote=“Pramanj, post:20, topic:17796”]
Perhaps some of the SI’s can try to port one of their existing projects to cloud on a pilot basis in parallel and see the performance. [/quote]
You can check this online demo of Ignition. Pay attention to the IIOT demo of Edge devices on cloud for performance.
https://inductiveautomation.com/products/ignition/demo/online
…
On Ubuntu, you have to install JAWAWS to run .jnlp file.
sudo apt install icedtea-netx
Response is not bad for an HMI, In fact a human interface can be slow because humans don't need to see fast changing screens all the time. Just a snapshot of status and the trend whenever he wants is good enough as long as the alarms are notified as soon possible.
The update rate of plant parameters on the Ignition server on the cloud is important. In other words the response of MQTT or PubNub in refreshing the server data base is important as that has to continuously happen in the background. I guess reporting on exception basis and data compression such as bit representation of digital and 16 bit representation for analog is really crutial. Conventionally the analog values are represented with A/D converters of 12 bit accuracy so 16 bits are adequate for analog tags.
PubNub guarantees 250 milli second latency for a message of 250kb size between two points anywhere on the globe! It can be two way communication. So one packet can theoretically represent 250x8x1000 = 2000000 (2 million digital tags) or 250/2x1000 = about 125000 analog tags values on one channel and there can be any number of channels configured. Also by reporting on exception basis where the digital values hardly change in the plant and analog values also remain steady in a normal operation its possible to achieve required response time. Its only in emergency situation the data change is rapid. So we have to see the response for the worst case i.e. reporting full plant data every cycle.
Of course MQTT is proven for Ignition and works on exception basis so we can just get started with it. However PubNub is server-less and works on subscription basis and has an SLA.
I request diat150 to share his opinion on this.
Hosting Ignition on cloud will be a huge advantage to customers as the infrastructure, and its associated maintenance headaches is out of their scope. Even for SI it will be very easy to spin a a new VM for every customer and host Ignition and get started with configuring it. It will be a paradigm shift in SCADA industry with Ignition being the pioneer provided customers are convinced about security issues etc.
If you were to propose an "AI Powered smart infrastructure command control centre" to meet the following requirements, what would you propose?.
Thanks.
What kind of smart infrastructure is it? Is it a smart city project?
It all depends upon the project type.
Please see our NJSCADA product from PRAMANJ Technologies (My company) which is based 100% on JS and NODEJS technology. Its available as an add on to Ignition. It has following modules:
2. ARSCADA - augmented reality interface to SCADA systems
4. RMSCADA - Remote monitoring of SCADA screens over Internet. Demo of RMSCADA.
5. UNISEMS- simulation interface to SCADA thru OPC UA.Demo of UNISEMS
The advantages of NODEJS for SCADA are explained in various presentations above.
Best Regards
PRAMANJ Technologies
Your engineering capability and coding skills are fantastic. If IA is smart, they must invite you to join their team. You can be an excellent brand ambassador for Indian market.
Why not! I am open to any suggestion , I am happy as long as my modules are used by end customer!
Most impressive part of your package is, you have built your own multiplatform pure JS visualization components for Ignition server. This can easily be ported to any other JS framework like React/Angular/VUE etc. As you have built your framework from the ground up, i think you can rewrite in typescript too, if required. Your foundation is very strong and you have opened up a gateway for multi-visualization module. In fact, you have built a pure JS SCADA framework. Brilliant job.
PubNub integration is a master stroke for very large smart infrastructure projects. It provides a high availability redundancy path for MQTT PubSub network.
Is it possible to have Python ML engine in between Ignition and NJSCADA?.
React / Angular /VUE are DOM based frameworks, where as mSCADA is pure SVG based, no DOM required. Just one HTML page having a DIV block where the SVG canvas is invoked! All components including Graphics and Tables are realized using SVG. (Ignition also uses some third party library for UI, I am forgetting its name). However I have plans to use one of the DOM frameworks that you mentioned and my SVG canvas can be drawn on any of the DIV's of an HTML page. So you have a mix of the HTML/CSS and SVG components to design the UI. But traditionally SCADA requires more of graphical displays such as P&ID, dials, trends etc, rather than HTML Text/Table displays. However many of the Client frameworks now a days provide trends / bars/ dash board components which are built using pure HTML/CSS , so theoretically its possible to build SCADA interface using these built in HTML/CSS components using frameworks like Infragesics / Syncfusion/Keno UI etc which use typescript, but I don't know why no SCADA vendor uses these frameworks on client side! Perhaps its due to legacy SCADA systems prevailing in the market. In fact the HTML/CSS frameworks also offer responsive screens that scale/resize according to display size including Mobile phone screens!
Indeed , Imagination is the limit where PubNub can be used! You should see their users list! They also have PubNub functions features which provides server less functionality!
Python is the best language for ML/AI indeed as of now, as many of the libraries are written using Python. With new version of Ignition which will support Jython version 3, it would be possible to access these ML libraries right from within Ignition server or client side!
However hopefully the these ML/AI libraries will also be supported in JS and some NPMs will be available soon so that the complete ML/AI can be implemented in NodeJS/JS as well.
In my opinion use of ML/AI for SCADA is over hyped right now. One needs domain experts to come up with some solid use case!. Digital twin concept is good but I am yet to come across some solid example! Only time will tell. ML/AI is good and has excellent examples like fraud detection in Banking Industry, Imaging applications like motion/object detection, face recognition etc but I am yet to see such an example for Industrial applications. A lot of hype and promises and projections for AI in industrial domain but only time will tell.
Google autonomous vehicle is a very good example of ML powered automation. This kind of automation has NOT yet been achieved in Industrial automation sector. Perhaps, we are stuck with PLC mentality and SCADA software follows PLC religiously. Smart SCADA manufacturers must study Google AV as a benchmark to build ML powered SCADA systems. Otherwise, Industrial automation looks like, it has run out of steam and has come to a dead end.
I am sure, Google and FB are secretly building AI SCADA systems. They have built all the bells and whistles. SCADA manufacturers who remain complacent with their past achievements and waiting for someone else to lead will be thrown out of business very soon.
Driver less cars indeed are good examples of AI implementation but that's in automotive industry, I meant process control and manufacturing industry. Perhaps AI in this field should analogously lead to operator less or unmanned plants! Just set the target setpoint and the plant should automatically run runup/rundown handle normal operation, emergency handling, shut down in most optimum way! I think we are reaching there step by step. First step is reduce manpower, and cost of instrumentation and increase automation.
I don't quite agree as their target is mass end user market which is billions of users. SCADA is a comparatively small Niche market which can only be served by dedicated SCADA manufacturers and automation companies. Although azure and google could platform do have many SCADA like features for IOT, but these are very generic features not Industry specific.
Human like intelligence is known as AGI. Self learning Driverless AV algorithm is close to human intelligence. More than 80% of jobs are repetitive and needs zero innovation. The DLAV algorithm will be embedded into millions of machines and eliminate 80% of jobs. We will enter an age of super productivity and surplus wealth. A federal government of "US+China+India" will take over the world Governance and will declare UBI for all humans to avoid a bloody civil war. Every human will be guaranteed free home+food+cloth+health care. There will be "one world, one government". This will happen by 2025.
Just we have to pull on for 7 more years and after that we all will be happily singing "Hakuna Matata, it means no worries, for the rest of our days.. la la laa.." 
Wonder who will be the president! Trump, Xi Jinping, Modi!
Trump has signed the final executive order to drastically reduce H1B visa workers. Many slave nations are crying that US economy will collapse without their H1B workers. Trump is not an idiot. IBM, Google and FB have demonstrated beyond any doubt and have convinced Trump that US doesn't need H1B manpower any more and their AI machines can do all these jobs far batter than human workers. This is the first step towards "massive decimation of jobs, one world, one government" agenda.
The SCADA will be reduced to just one knob to set the target and one display to show the production!
[quote="Pramanj, post:39, topic:17796, full:true"]
How many million jobs do you think Google Duplex will kill?.