IoT, AI and the future of asset management
These days, data is everywhere. With the rise of the Internet of Things (IoT), smart sensors, cameras and devices are collecting and sharing rich pools of data about the condition and behaviour of all sorts of assets, from buildings to transport infrastructure, schools, hospitals, water pipes, energy and much more.
Not too long ago, the challenge was trying to find that data. These days, the challenge is there’s so much of it to make sense of.
US astronomer Clifford Stoll famously once said, “Data is not information. Information is not knowledge. Knowledge is not understanding. Understanding is not wisdom.”
Today, that’s very relevant. The real value in all that data is not in collecting it, but in making sense of it, seeing trends and uncovering insights to help make wiser decisions for the communities we work for.
The opportunities and challenges of IoT
IoT has almost limitless applications across services delivered by government and virtually all industry sectors, with experts predicting it to be the lead driver of innovation in the next 10 years.
When managing assets, sensors and cameras are especially useful in detecting condition-based symptoms, allowing operators to manage by exception rather than manually checking thousands of components. Not only does IoT significantly reduce human intervention and risk mitigation, it can also help asset managers track their carbon performance and create more sustainable outcomes.
And yet, with the rapid growth of IoT, there is a lag in understanding and uncertainty of exactly what benefits it will deliver. Extracting meaningful information from enormous volumes of data can also be challenging if doing so manually.
Take pipe inspections as an example. No engineer wants to spend their day watching hours upon hours of pipe inspection CCTV videos to complete a condition assessment and decide if work needs doing.
So, how can asset managers harness the potential of IoT and make the task of managing data easier and more efficient?
The answer lies in the rapid rise of another technology, Artificial Intelligence (AI) and its subset, machine learning to aggregate the data from IoT and systems, uncover insights and make predictions about the performance of assets over the long term.
Making sense of big data with AI and machine learning
According to Gartner, the top technology trend for IoT is an increase in AI. This is because as the number and complexity of IoT systems grow, the ability to analyse the data collected will exceed human capability.
People used to worry that AI and automation might result in jobs lost. Instead, AI intends to supplement, not replace, human review and quality assurance.
AI uses powerful algorithms to filter through large amounts of data and identify patterns. For a local government, AI can be used to collect and analyse data on a community’s needs or reduce the amount of time and resources required to make optimal decisions about asset maintenance.
Next, machine learning takes the patterns found by AI to create data models that predict the likelihood of an issue occurring based on past events. The more data machine learning can access, the more accurate the predictions. Add in real-time data from sensors, and you have continuously-updated predictions to make evidence-based decisions for greater asset optimisation.
Optimising strategic asset management
Owners of vast asset networks have a lot to deal with – the here and now of day-to-day failures, current or upcoming maintenance and the long-term life cycle management of their assets. On top of this, each year their existing assets are depreciating.
Optimising strategic asset management is key to making decisions and allocating expenditure that balances cost and risk. But where to start?
1) Identify the data you need
First, asset managers need to understand what data they need to achieve their strategic goals.
This includes asset information that doesn’t change, such as the material, size and install date, and the asset information that does, such as the level of maintenance, corrosion or metal fatigue. There is no point in collecting information if you’re not going to use it, so identify the data you need to find the answers you seek.
2) Create a centralised data hub
Next, consider using a central cloud platform to create a single source of truth for all your asset data.
As organisations grow, much of their data becomes siloed in different system folders, personal devices, and legacy environments – even in people’s heads. Making sense of your data can only be achieved when data is centralised, standardised and understood in context with other data.
Another advantage of having a single repository for all asset data is that it can be served up quickly and visually via dashboards, making it easier to monitor activity across asset networks and present evidence-based findings to senior stakeholders.
3) Predict the future
Thirdly, having the ability to map out different scenarios is a highly effective way of comparing cost benefits, rate of asset decay or rehabilitation outcomes.
Machine learning can have a significant impact in helping asset managers predict the future behaviour of assets by analysing historical data in combination with any live data collected from sensors. It allows asset managers to see the consequences of action or inaction with scenario modelling for up to 50 years, taking into account treatment types, intervention points and potential funding required.
Connecting the dots between big data will be critical in the future of asset management. With the help of AI and the predictive intelligence of machine learning, exciting opportunities lie ahead in delivering better, more sustainable outcomes for our communities and customers.
About the author:
Mark Lee is the Business Development Manager for VAPAR. Following a decade as a senior water and sewer engineer in both operations and asset management, Mark made the move from the public utility sector to assist in the development of VAPAR’s software. VAPAR is an Australian software company that is leading the advancement in the use of artificial intelligence and data automation to make CCTV pipe inspections more efficient.