数字孪生应用的无限领域

HOW SOFTWARE IS EATINGTHE REAL ECONOMY The third shift builds on the previoustwo and is gaining momentum. It isthe virtualization of physical assets,environments, and processes that arecornerstones of the real economy—inmanufacturing, transportation, energy,mining, life sciences, and many othersectors. It’s software and AI that are noweating up production lines, machines, andchemical reactions by representing themthrough digits and pushing to the newfrontier in autonomy and optimization. Since Marc Andreessen coined thecatchphrase “Software is eating theworld” in 2011, we witnessed severaltectonic shifts in the digital world that setthe stage for the next decade and beyond. First, cloud has become dominantover on-prem solutions, especially forprocessing and analysis of big dataworkloads. This shift has enabled theircustomers with richer functionality ata faster pace and at a lower cost. Secondly, advancements in artificialintelligence (AI), particularly in thefield of deep learning, enabledautomation of complex scenariospreviously attainable only by humans.Although general AI remains anaspiration, specialized applicationsalready fill a large part of our lives. The leitmotif of this trend is quitestraightforward: businesses can achievemuch more in terms of productivity andinnovation when their physical assetsand processes are represented throughdigital forms (aka digital twins). Previouslyexpensive and available only to spaceprograms at NASA, now digital twins,powered by modern technologies, promisea giant leap for a reasonable cost. DIGITAL TWIN TYPES Areas of digital twin (DT) applications are boundless and are leveraged continuouslyto simplify asset management, maintenance, education, and communicationprocesses by replicating accurate twins of the on-site equipment/spaces. Below,you can see a taxonomy about what a digital twin is, its key aspects with a commonvocabulary across the teams. You can also learn where it can be applied, and why. As you can see, DTs can be distinguished by their applications and hierarchy. Variouscombinations of those define the types of digital twins. Let’s take a look at a few real-lifeexamples. Example 1. Digital Twins in Industrial Context The life of any electric motor starts fromits design. Engineers create a prototypethrough physical modeling to test itsparameters through simulation—we callit the Simulation Twin of a Future Asset(an electric motor). Once the motor isproduced and installed, we need tomonitor its status. To achieve it, themotor sensors read torque, voltage,temperature, vibration, and other measuresto be linked to a dashboard or a 3D visualrepresentation of the motor. It representsthe Status Twin of a physical instance. Throughout this process, sensor-generateddata is collected from a motor or evenmany motors of the same type. For thisreason, it is possible to build a modelthat predicts failures and fosters OEE(Overall Equipment Effectiveness) byreducing downtime through predictivemaintenance. We can call it the OperationalTwin of an Asset (an electric motor).Now, let’s assume that the motor is apart of a larger industrial system thatautonomously executes certain actions. Insuch a case, it constitutes the AutonomyTwin at the System hierarchical level. Example 2. Supply Chain Simulation Twin typically offerssimulation and optimization models toperform what-if analysis, mitigate risks,or suggest the best supplier based oncost, time-to-market, or other criteria. Every digital supply chain initiative startsfrom collecting bits of information aboutsupply, inventory, production, orders, andconsumers to get real-time visibility of an actualsupply chain and to forecast supply chaindynamics. Such data foundation representedthrough dashboards and forecastingmodels define Status and Operation Twinsof a supply chain at the Organizationor Process level. On the other hand, The more complex scenario is,the more you can benefit fromleveraging digital twins. DIGITAL TWINS SPECTRUM Here is an extended view on digital twin taxonomy to help match industrial needs withappropriate solutions. It is represented as an evolutionary step-by-step approachfrom foundational to more complex scenarios that can generate even more value: Monitoring Analyzing Real-time data collection from sensors andother sources is the first step to buildingoperational intelligence and providingvisibility into a specific process, system,or asset state. Insightful metrics and KPIsare delivered through an alerting system,dashboards, 2D-3D visual models of thephysical world (immersive data) andconsumed by different business functionsfrom field engineers and supervisorsto top managers who need to react tothe changes in real or near real-time. Insights and hidden patterns extractedusing built-in machine learning andself-service analytics foster awarenessand accelerate root-cause analysis.Operational digital twins provide a betterunderstanding of failure reasons or linkdifferent correlated ev