AI is disrupting the workplace through digital transformation, resulting in an extensive use of the digital twin, a virtual representation of a physical object or system. The combination of AI with the digital twin results in significantly enhanced productivity; this is quantifiable. AI enhances workforce productivity and improves safety, reliability, quality, and security. This AI-based disruption aids many aspects of the industrial process, from design and engineering to operations to maintenance.
AI itself is not one thing but comprised of a number of technology types, including neural networks, deep learning (a flavour of neural networks), natural language processing, computer vision, unsupervised machine learning, supervised machine learning, reinforcement learning, and transfer learning. The various types of AI are applied in different ways throughout the industrial world to create targeted solutions provided as descriptive, predictive, and prescriptive analytics. A relatively common solution used in a wide range of industries today is predictive analytics in the form of machine learning to identify anomalies with equipment and processes. These anomalies can indicate performance problems or asset health deterioration well in advance of any control system or SCADA warning or alarm. Issues can be identified and corrected quickly, well before they have a major impact on operations. This results in less downtime, better product quality, reduced risk, and increased overall efficiency and profitability.
From an industrial perspective, AI can be broken down into what AVEVA categorizes as the Four P’s of Industrial AI:
- Predictive: Based on machine learning, this is a type of pattern recognition and anomaly detection leveraging industrial big data to create digital signatures of assets and processes.
- Performance: Based on first principles analytics (simulation) and machine learning, it provides early warning detection of pending problems and inefficiencies when compared to actual sensor values.
- Prescriptive: Based on the issues detected in Predictive and Performance analytics, this provides root cause analysis, planning and decision-support, and probabilistic courses of action to best remedy & optimise a given situation.
- Prognostics: Leveraging neural net, deep-learning, and reinforcement learning technologies, this provides a forecast of future events.
Predictive analytics, a bargain for the industry
Predictive analytics in the form of machine learning has become one of the more common advanced technologies used in industry today. Although referred to as “predictive”, it is actually a very effective method of anomaly detection in near-real-time. It is a type of advanced pattern recognition where the digital signatures of the normal behaviour of an asset or process are captured and used as a basis of comparison with incoming, real-time data from SCADA and other control systems. Real-time data from sensors is collected and compared to the expected data signature of a given asset or operational scenario. Deviations from normal behaviour can be detected days, weeks, and even months before a traditional SCADA or control system alarm would trigger. This provides companies with adequate time to take the appropriate actions to rectify the asset or operational problem before it’s too late.
Examples of successful predictive analytics include sophisticated turbine “catches” where there were step changes of vibration reductions (not increases). Each time, the manufacturer told the customer it was OK because it was a reduction in vibration, not an increase. With this particular situation, it turned out to be due to the beginning of blade separation within the turbine stages. The system was nowhere near a control system alarm or warning. However, had it gone on, it would have resulted in a catastrophic failure that could have destroyed the turbine, caused extensive downtime (loss of power production), and a potential for significant injury to personnel. Conservative estimates by the customer showed that costs of over $34 million USD were avoided due to the early warning detection of this issue.
Prescriptive and prognostic analytics, a bridge between AI technologies and humans
It began with condition-based triggers to create a proactive maintenance program, vs calendar-based preventative maintenance. Applied to AI, prescriptive analytics bridge the gap between anomaly detection and the actions needed for resolution. It’s critical to both improved asset maintenance and enhanced operational efficiency; consequently, it has become an increasingly important aspect of an overall Reliability Centered Maintenance (RCM) program.
In order to further enhance predictive and prescriptive analytics, prognostics take AI one step further by forecasting future events, such as operational performance degradation or an asset’s remaining useful life. Prognostics can allow humans to make decisions such as, ”can the system make it to the next planned maintenance outage?”, or “can the asset make it to next week or do we need to call in emergency personnel over the weekend on overtime wages to fix the problem?”. These are critical decisions that impact both risk and costs. Managing risk is a key part of what AI brings to businesses, and it can significantly help improve the bottom line of industrial operations.
However, without a suitable bridge between AI technology and humans, appropriate actions may not be taken, and the value of this advanced technology could be lost. Prescriptive analytics is the key to making this happen in order for businesses to gain maximum value from advanced AI technologies and software investment.
Prescriptive, a requirement to keep up with the competition
Implementing prescriptive analytics is not trivial. It requires extensive, industry-specific fault diagnostic and resolution action databases that are logically defined based on changing sensor values (and other permutations) so that automatic (programmatic) recommendations can be provided to the user. This requires software to encompass vast industry expertise, experience in types of reliability-centred maintenance practices, and predictive analytics. Because of this combined uniqueness, competition in this space is limited and typically targeted to specific industries.
As software continues to evolve, integrated processes become more important. Predictive, prescriptive, and prognostic software will increasingly integrate with Enterprise Asset Management (EAM) systems in order to dynamically create work orders and integrate the forecasted remaining useful life of the asset with recommended prescriptive actions needed to rectify the issue. This will provide automation from issue detection, through root cause analysis, to remediation and rectification. Beyond EAM integration, this type of AI software will also integrate with scheduling systems to recommend the optimal time to perform emergency maintenance within the forecasted remaining useful life window of an asset in order to reduce adverse impacts on operations, minimise overall business risk, and maximise profit. This will then extend to closed-loop, automated process control, where humans merely monitor the fully automated and optimised, end-to-end operations and maintenance processes that are controlled by AI. These technologies exist today, and adoption will increase over time.
Predictive analytics software will continue to be improved and enhanced through prescriptive capabilities in order to detect and prevent problems faster, better maintain industrial operations, optimise scheduling, and enhance process control. From a societal perspective, predictive analytics will be humanised through continued advancements in prescriptive capabilities, in order to better enable and empower the workforce. The combination of these factors will allow them to improve operations, work productivity and safety, as well as the speed of knowledge transfer and learning. An increasing number of industrial companies throughout the world are actively engaged in leveraging artificial intelligence, particularly predictive analytics. This is no longer an option in many industries but often a requirement to keep up with the competition. In order to maximise the benefits, the bridge from predictive AI technology to humans must be as seamless as possible. That is where prescriptive plays a key role, and it is revolutionising the way work is performed.
James H. Chappell – AVEVA