Machine Learning finally makes some waves in tunnelling. I have read a rather fascinating article in the Tunnel Business Magazine. 'Big Data, Writ Large: Technology to Advance Tunnel Boring'. It is a must-read if you do want some window to the future of the tunnelling.

If I were in the management of Herrenknecht, CREG, Robbins, CRCHI, Terratec or even newcomer manufacturer, I would definitely frame the cover of this Tunnel Business Magazine and hang on the walls of the company. I would make every person in the management and in the product development read and understand it.

Data Sciences & Machine Learning -

The article in itself is not revolutionary. It even uses the term 'Big Data' which is frowned upon nowadays and made way to a more modern term of 'Data Sciences'. That said, it is important in the way that it heralds the ongoing Data Sciences, Machine Learning revolution that the world is going through for some time now and the tunnelling world that has long refused to move beyond the iron and hydraulics can not say anymore 'We did not know'. Now you do know, because Tunnel Business Magazine wrote it. 

I know that there have been many academic papers which talked about the use of TBM sensor data to predict the advance rate of TBM in certain geological conditions etc. This is the first time I read it in a respected mainstream tunnel publication. Just as I would expect from the Tunnel Business Magazine (in my opinion, the only tunnelling publication one needs to read. They frequently publish at least one jewel article on every issue. I see less "background noise news" with them).

Automated versus Autonomous -

The Tunnel Business Magazine digs into this subject mainly focusing on it use in Tunnel Boring Machine (TBM) tunnelling. Whether you like TBMs or not, I believe most agree that it is currently the most automated method of tunnelling. Please pay attention that I am not saying 'autonomous', but 'automated'.

Tunnel Boring Machines today generate a lot of data via its sensors and other systems. The TBM operator normally reads such data and then manually makes adjustments to the controls, like thrust etc for better advancement and less wear in the cutterhead.

There are two interrelated issues with this model:

a- With the advancements in TBM technology, the amount of real time data that the TBM and its sensors create increases to such a level that, it is not reasonable to expect a human operator to correctly interpret them real time and take the most optimum decisions. It is just too much.

b- The data generated during boring is certainly correlated to the geological properties of the ground being bored. Here currently we rely on the experience of the operators. Here come the questions. Why rely on the intuition of the operators while a computer can easily recognize the patterns for a given situation and take the most optimum action automatically? The Machine Learning algorithms of today are not rule-based. The algorithms may literally 'learn' which means the most optimum decision can be the one no human ever thought of before.

Both for the first and the second points, TBM producers may come with an intermediate solution: Expert Systems. An expert system is essentially a computer software that interprets the given situation, recognizes the patterns and then proposes one or more actions to the TBM operator. Then the operator takes the action. This model is very common in other industries. It is also a natural step in the progression, though it has its shortcomings. Like the most important one being, the speed. Humans are slower than computers. This is why the Stock Markets like NYSE or Nasdaq are flooded by Flash Trading software. Because a software program can buy and sell stocks at one thousandth of a second. A human trader cannot do that. Human minds are still necessary for a longer term policy making and overall decisions. Expect the same in TBM tunnelling. A computer program can make thousands of decision in a split section adjusting the parameters of a TBM in such a fine scale no human operator can.

Time for a Mind Shift -

If I were in the management of one of those TBM companies, I would read the Tunnel Business Magazine like my life depended on it. As things go, it is the reality. Your survival depends on how much you embrace these game-changing technologies.

It is time to see beyond the iron, hydraulics, pumps and motors. I know, by some experience that, for some people it is impossible to grasp beyond mechanical engineering. Their companies will not go beyond the "mee too" products and will be the casualties of this change.

No doubt some companies already have a group of people looking into this technology. Though this article says, it is time to move from the 'boss wants me have a look into this technology', 'hobby research' phase to 'the competitiveness of my company relies on Data Sciences and Deep Learning' phase. Innovation is the most important tool to stay alive or to beat the legacy companies.

https://tunnelingonline.com/big-data-writ-large-technology-to-advance-tunnel-boring/