Intelligent Rail Summit 2018

“Data only acquires meaning when it is combined with physical observations and models. Then it becomes information”, says Diego Galar, Professor of Condition Monitoring at Luleå University of Technology (LTU). “You need to know the conditions surrounding a fault, such as the weather and the conditions of the track. Then the sector will reap the rewards.”

“In the rail sector, we have for many years tried to analyse ‘classic’ data, but around four years ago we realised that it is no longer enough,” says Galar. He is one of the speakers on 28 November at the Intelligent Rail Summit 2018 in Sweden. There, he will talk about the analysis of big data and the use of Internet of Things technologies to improve track maintenance and use. “We now find ourselves in a very promising phase, but we are not there yet. But now the big players are beginning to invest in this new method of gathering and analysing data.”

Anomalies

Most data that is collected is about all the times that things go well, in other words, normality, but in reality, the only useful data is about those occasions when things go wrong. “You must look for anomalous situations, which actually don’t happen very often.” Furthermore, if you only collect data about normal situations, you do not know which signs you need to look out for that indicate that things c

ould go wrong. According to Galar, the rail sector is far from knowing all those signs: “We have not collected the data from about 80 per cent of all errors.”

So how can the sector ensure that it does do this? On its own, collecting data is not enough. “If you look at other sectors, such as the financial sector, they gather a lot of data about when things are going well or badly with the markets. Their data comprises many different profiles and scenarios, and they draw a wealth of information from it. In the rail sector, the opposite is the case: we have a great amount of data, but no information,” explains Galar. “So there is a huge chasm between what we already have, and what we still need. This is the current challenge for the rail sector.”

Data and observations

Only when it is combined with physical data, based on observations on the track, for example, is the information useful to the rail sector. “You need information about the circumstances around an accident: what was the temperature, what was the weather like, what were the conditions of the track and rolling stock, etc. It is only with this contextual information that the data actually means something.”

Here, innovations that use the Internet of Things can make an important contribution. These measure real-time information about the track, such as the heat of the rails or vibrations. “We need to measure everything.”

Forecasts

Once that is done, and the data becomes valuable information, the rail sector will reap the rewards, as it can anticipate when things threaten to go wrong. “We can do two things here: make predictions, and get in-the-moment insights into the state of the rolling stock and the infrastructure. The rail sector would like to know in advance if their trains can keep running in the future. Everything is about forecasts.”

The most important thing to remember, says Galar, is that this is all very possible. “It is completely doable. Thanks to the combination of data and observations, we can accurately forecast what state the entire infrastructure will be in. This applies to all elements of the track: switches, level crossings, and so on.” And if better forecasts can be made, more effective maintenance can be done. This is very important if the European rail sector wants to take on competitors from countries such as China. “Forecasts of rail maintenance can save a lot of money.”

Marriage

Rail companies must keep exchanging their data with each other, though it may be tricky. “Increasing numbers of rail companies are going to realise that their data is really valuable. In fact, it can be used as currency. But these companies are afraid of sharing their data because of security or privacy concerns, so clear rules have to be established. We must understand each other,” underlines Galar.

“Sharing data is essential. It is like a marriage: you can’t just choose to ignore one party.” This will only lead to a win-win situation, says the professor. “Sharing data provides nothing other than benefits for all parties.”