How big data analytics and artificial intelligence are being used to power predictive operations

Man monitoring equipment analytics.As today’s manufacturers seek greater production efficiency and asset reliability, technology that combines deep analytics with artificial intelligence can help them achieve these gains by facilitating a shift from reactive operations management to a proactive approach capable of preventing costly equipment failures and process disturbances. 

Noel Bell, senior principal software engineer at Emerson, says using artificial intelligence and analytics in operations management for enhanced decision support is essential for manufacturers as they experience increased global competition, stricter regulation and the continued loss of organizational knowledge brought about by retirements. The drive to find and maintain a competitive edge in the face of these issues has necessitated a move away from traditional operations management. 

Traditional operations management, Bell says, is reactive and susceptible to human error because it relies on repetitive procedures with significant human interactions to deal with planned and unplanned events. Mistakes in following procedures or assessing a situation can lead to costly outcomes, and the reactive nature of the approach can make it difficult to diagnose problems and even harder to predict impending events. What’s more, when diagnoses are made, they often require considerable experience and time, and they occur after the fact – when operations have been negatively impacted. 

That’s not the case when artificial intelligence meets analytics, says Bell. 

He’s teaming with Integration Objects – a systems integrator and solutions provider specializing in operational and manufacturing intelligence and advanced analytics – to develop applications using an operations advisory platform that blends artificial intelligence and analytics to execute automatic reasoning by applying real-time data to root cause analysis.  

The platform, called KnowledgeNet (KNET), developed by Integration Objects, enables plant operators, technicians, and maintenance personnel to more effectively identify and diagnose abnormal conditions that lead to costly, complex events – and receive suggested corrective actions to address those issues, based on their specific operating regime, explains Samy Achour, founder and president of Integration Objects. KNet can be applied to both operations and maintenance activities, helping avoid costly operating deviations and equipment failures. 

Through the use of data-driven models combined with expert rules and fault propagation models, KNet provides a method of performing predictive analytics and then deploying them online for real-time KPI calculations and prescriptive analytics that offer an effective solution, enabling operators to respond faster and more effectively as well as take actions that prevent potential issues. KNet is capable of executing forward chaining, recognizing distinct conditions and predicting events based on those conditions. It’s also capable of backward chaining, helping inform personnel about what conditions likely led to a complex event. 

KNet’s intuitive smart dashboard not only provides users with a visual alert after detecting an abnormal condition, it offers a root cause analysis summary, consisting of detected and predicted symptoms, in one quick, glance. The interface also allows users to click on a specific symptom to access the current and potential impact of that symptom as well as suggested corrective actions to restore normal unit operations.  

For example, KNet technology can help prevent column flooding in a debutanizer unit by detecting and alerting operators to early symptoms such as variations in pressure, Bell says. Once KNet determines a specific cause of the symptoms, it offers appropriate recommendations to remedy the issue, which could include reducing reflux or limiting the steam to the column reboiler depending on the cause of the flooding. This improves the operator’s ability to proactively manage the process and results in increased asset availability, reliability and plant safety as well as reduced energy consumption and fewer quality issues, he says. 

“KNet takes the knowledge an organization has, puts that knowledge into code, then applies real-time data to it and tells you what part of a specific knowledge chain is applicable to a situation – and what action you should take,” Bell explains. “The idea is you get early warning signs so that you can take action before big problems manifest.” 

How could your operations benefit from applications that blend analytics with artificial intelligence? Share your comments by replying below.