Big Data Analytics, Artificial Intelligence and Machine Learning. Some people don’t see a difference in these three terms, and treat them as the same. One must understand that while some concepts of these technologies may overlap, they must be looked upon independently and studied according to their applications. AI suggests automatic generation of insights when applied to Big Data, offering results with little to no effort. The data analytics user experience on the other hand has a user experience that’s different altogether.
A case study was shared by a vendor who claimed that an AI-focused data analytics solution could solve a production issue. A formula of regression was used to solve the problem, which is anything but AI. What was even worse was that the package was branded as AI, and sold in the market, with several companies even taking the bait; all due to lack of understanding of ML, AI and deep learning.
The bitter reality is that a huge chunk of data doesn’t even get analyzed or leveraged by analysts, leaving a significant vacuum for actionable information.
Data Analytic Solutions for Process Industry
Several technologies including AI make up data analytics applications, but the prime focus should always be on the user experience and acceleration of insights. The end solution should be able to cater the requirements set forth by subject matter experts (SMEs) that include process engineers, data analysis experts and so on.
Well-designed data analytics applications act as empowering agents for SMEs and produce actionable results through easy-to-use features set. If its AI or ML or any other available technology required, then the data analytics application should use that. Possible algorithmic functions that can be used include those of digital signal processing, map-reduce models and calling procedures.
Looking Beyond the Algorithm
A vital part of an analytics solution being used in process manufacturing and industrial internet of things (IIoT) is the cognitive computing algorithm. In addition, other aspects to look out for include data wrangling, contextualization, cleansing or in other words preparing it for use. These preliminary steps must be included within any data analytics application so that the result generation process can be accelerated.
The application must also enable users expand and extend analytics to the level they require. Remember, end users would always want to expand the usage of their data analytics applications and come up with the need for specific algorithms, therefore there should always be room for growth and upgradation. These scalability features are readily available in algorithms like REST API, OData, etc.
Power Plant Pollution Control
Plant operators struggle to get information from the plant’s automation system to control the level of damage done to plant equipment from pollution. As a result, there is frequent overdosing of mitigation chemicals just to be on the safe side that leads to added cost.
For such a problem, the data analytics application can process information swiftly to provide accurate control of the pollution abatement system, optimizing the chemical injection so that maximum amounts of harmful chemicals are removed.
Ore Smelting Operations
Energy-intensive processes such as smelting are quite sophisticated and difficult to quantify through conventional systems that mostly employ spreadsheets. As a result, either the cost or the environmental parameters are often compromised.
The solution would be a data analytics program that can balance out multiple variables, and make it easier to study the effect of adjusting one parameter in place of the other, until the best possible combination is reached.
Using the right data analytics application would allow processes to become more efficient, and the overall productivity to rise. Surely, it would take some time for the promises to mature, but nevertheless the Industry 4.0 bandwagon is up and running, and would start bearing fruits soon.