Analyzing Big Data for Continuous Improvement

Digital technologies are changing the way companies operate and innovate. Now, with advances in artificial intelligence (AI) and machine learning, Internet of things is moving to a new level of possibilities. In the past, predicting what will happen and prescribing what actions to take usually required human interventions. But with AI and machine learning, systems can learn from data with no human interaction. That means the big data you collect from your operations and business systems can be analyzed in real time to make complex, evidence-based decisions. As the amount of your data grows, the “smarter” the analysis and the related machine decisions become.

Manufacturer can reap tremendous advantages when they use AI to predict maintenance and quality issues. For example, data from an asset’s sensors can be collected and compared using analytics and algorithms to predict when a part will fail or how much time is left in an asset’s useful life. Armed with these insights, the system uses AI to recommend repair or replacement before there is a breakdown.

Some of the areas where AI can help improve industrial operations are:

Monitoring

Sensors and applications on individual components generate huge amounts of data about a production asset’s basic health, utilization, availability and key performance indicators (KPIs). With AI and machine learning, this data helps you gain a deeper understanding of your production environments, identify issues and plan for improvements.

Maintenance

The right maintenance at the right time reduces unplanned downtime and extends an asset’s life. Using AI, machine learning, real-time data and predictive algorithms, production anomalies can be automatically identified along with maintenance recommendations.

Quality

Product defects can be costly, especially if they result in recalls. Analytics and AI enable the rapid identification of deviations by comparing live data to standard deviation models, aggregating data for analysis and modeling in a single data lake, and pinpointing the root cause of production problems.

Productivity

AI can be used to identify the root cause of bottlenecks in production processes, predict their direct impacts, help maximize workforce productivity and increase throughput yield of the products being produced.

Planning and Scheduling

Mitigating supply chain risk—including planning for, ordering, and managing raw materials and parts—is essential to maintain high productivity. With AI, you can use dynamic scheduling to proactively optimize production schedules and promptly implement countermeasures when process delays occur.

Energy and Consumables

Renewable energy growth is causing grid volatility and financial challenges. AI can help maximize profits and boost efficiencies from energy generation and trading to distribution and consumption.

Safety and  Compliance

By collecting data from machine sensors and videos, and applying analytics and AI, organizations gain insights that increase the safety and comfort of operators, and ensure environmental and regulatory compliance.

Engineering and Design

AI and machine learning deliver insights relating to maintenance, quality, productivity, energy and safety. These insights can help manufacturers design newer versions of assets with newer sensors that deliver the intelligence required for continuous improvement.

While Internet of things technologies connect everything and create huge pools of data around assets, AI and machine learning transform the data into insights that enable efficient automation of production processes. Please comment/share your thoughts on how you are applying AI in your context!

Artificial Intelligence in Industrial IoT

Industrial IoT Technology choices are changing the way companies operate and innovate. Now, with advances in artificial intelligence (AI) and machine learning, IIoT is creating a new level of possibilities.

In the past, predicting what will happen and prescribing what actions to take usually required human interventions. But with AI and machine learning, IIoT systems can learn from the historical data. That means the big data collected from operations and business systems can be analyzed in near real-time to make complex, evidence-based decisions. As the amount of data grows, the “smarter” the analysis and the related machine decisions become.

Enterprises can reap tremendous advantages when they use AI to predict maintenance and quality issues. For example, data from sensors can be collected and compared using analytics and algorithms to predict when a particular part is likely to fail or determine remaining useful life.

Although AI isn’t suited for every industrial IoT application, There are many, but I have listed three areas where IIoT with AI can help improve operations.

Monitoring:  Sensors and processes generate massive amounts of data about asset operating conditions, asset’s basic health, utilization, availability and key performance indicators (OEE, OPE etc.). With AI and machine learning, this data can help gain a deeper understanding of production environments, identify issues and plan for improvements.

Maintenance: The right maintenance at the right time can reduce unplanned downtime and extends an asset’s life. Using AI, machine learning, real-time data and predictive algorithms, production anomalies can be automatically identified along with maintenance recommendations.

Quality: Product defects can be costly, especially if they result in recalls. Analytics and AI can enable rapid identification of deviations by comparing live data to standard deviation models, aggregating data for analysis and modelling in a single data lake and pinpointing the cause of production problems.

Machines Don’t Have To Break! Operations Don’t Have To Stop! Leverage advances in AI and IoT to remove blind spots in your operations and gain the operational efficiency that you desire.

 

 

Improve OEE with Industrial IoT Analytics

OEE, i.e. Original Equipment Efficiency is the most common measure used to determine manufacturing losses. There are three key contributing variables to OEE:

  • Availability loss (A%)
  • Performance Loss (B%)
  • Quality loss (C%)

An OEE is represented in % (Percentage) and is the product of A, B and C.

OEE of 100% is considered perfect (I don’t think, this exist anywhere). Most equipment operates with the range of 40% to 80%. There are some examples where the OEE is below 40% in some manufacturing plants. OEE should only be treated as an improvement metrics. It may get tempting to aggregate different types of equipment in a shopfloor environment to generate higher level OEE. But this approach can be misleading and should be avoided unless the processes and production are identical.

Most manufacturers apply TPM (Total Productive Maintenance) principles which comprise of Proactive and preventive maintenance to aim for fewer instances of breakdowns, equipment operating performance and machine induced defects. The TPM approach was developed in the 1960s and essentially has 5S as foundation and eight supporting pillars. A simple google search would offer several websites with good literature and examples to learn more about OEE, 5S and TMP.

Industrial IoT analytics has a lot to offer in Manufacturing. Commonly, there is a target OEE, and then processes are improved to achieve that. Now, there lies an opportunity, what should be the target OEE? How does one arrive at that value? What’s important for your business? Is it Quality or availability or performance? Or a mix of these criteria?

Industrial operational technology data can provide greater insights on the efficiency of the production environment; it can help identify patterns which are not obvious to human observations, e.g. is there any window of time when the part rejects are higher? Is there a specific mix which is affecting the quality? Is handling of specific material causing an impact on cycle time? Are right spare parts being used to prevent sudden breakdown? There are many more questions which IoT analytics can attempt to answer at much faster pace than human-centric RCA observations.

There is tremendous power in just knowing these patterns to make informed process improvements, uncover blind spots and maximize existing operating conditions with the help of OT data and analytics. With predictive analytics, there is another dimension of intelligence that opens several new opportunities to calibrate the operating conditions proactively and gain the competitive edge in the marketplace.

Do Mobile Devices Have a Bigger Role in IoT?

As per a recent survey – 84 percent people surveyed across several nations said they couldn’t go a single day without their mobile device in hand – Only money came a close second.

Based on Google search– some interesting facts about mobile device and the usage

pattern –

Out of the ~4.5 billion mobile phone users in the world, 36% are smartphone users

Smartphone platform Android has the higher market share of 87% globally; Apple has ~12% market share.

89% of smartphone users use their smartphones throughout the day

92% of smartphone users use their smartphones to send text messages to other phones. Whereas, 84% of users use their smartphones for browsing the internet

50% of Android Smartphones and 43% of Apple iPhone users are younger than 34 Years.

Average smartphone users use ~1.4GB data.

This speaks volume about the role of a mobile device in our lives. Industrial environments have not embraced the use of smartphones or tablets yet; they continue to use ruggedized custom Windows or Linux based terminals with custom built apps which are difficult to maintain and add new features to.

IoT does not just impacts technology stack choices; It also impacts our work environments. Certainly, there are use cases such as workplace Logistics, Maintenance and tracking, Spare availability, Critical alerts and notifications, and so on which would better serve if the existing mobile device is leveraged without adding the burden of another custom handheld device.

The case of moving Intelligence to the Edge

There is a reason why DCS and SCADA systems in Industrial environments stayed on-prem, it’s the same reason that will drive IoT Intelligence closer to the Edge. There are thousands of scenarios where speed is the operational imperative from Asset management, critical production setup, command and control of high-value assets with predictive intelligence capabilities. The term “Edge” has been around, and architecture has existed. Internet of Things has expanded the scope and blended OT with IT datasets for better insights and outcomes.

Gartner defines edge computing as solutions that facilitate data processing at or near the source of data generation. It’s the same definition that DCS and SCADA systems applied in the past which are still in use to this day.

So what’s the real noise about? It’s the cloud which is perceived as a backbone of all architectures that enable digital transformation. Just that it doesn’t seem to address several core challenges of the Industrial setup:

Reliability- This is the core consideration in the mind of shopfloor manager. Do I have access to my data and insights, here and now in a deterministic way?

Latency – What’s the use of the insight if it’s not actionable in time?

Network Connectivity and bandwidth costs – There are so many blind spots through the connectivity chain and bandwidth costs keep adding up unless there is a way to limit the use and prioritize the data and optimal payloads.

Security – Industrial shopfloor assets and data are susceptible. Most of the devices are not IP-routable. Manufacturers have a higher preference if the data never leaves their premises.

There are several other considerations, but these four are crucial to solving for. Intelligence at the Edge has gained more investments and actions from most IoT vendors. It will continue to grab more mindshare.

#EdgeComputing #IIoT #EdgeAnalytics

Get a 360-Degree View of Your Assets with Asset Avatars

Industry 4.0 holds the promise of helping organizations in manufacturing, transportation, energy, and urban development to digitally transform with IoT-enabled technologies, advanced analytics, artificial intelligence (AI) and machine learning (ML). Underlying these digital technologies are assets providing vast amounts of data through an array of sensors and IoT-connected assets. And the number of IoT-connected assets is expanding at an unprecedented rate, with industry analysts anticipating 20–30 billion IoT-connected assets by 2020. As assets and sensors multiply, the data coming from each asset is exploding in volume and velocity, making the management of that data difficult.

In addition, teams within organizations often have siloed access to data, making management inefficient, and creating incomplete pictures of the health of assets and the entire company. Poor information management contributes to costly unscheduled machine breakdowns, and impacts your bottom line.

To reduce the risk of downtime and make asset management more efficient, organizations need a way to break down data silos, get more insight into their assets, and put information in the hands of the people who need it—when they need it. Asset avatars were designed to provide a 360-degree view of assets, and deliver this information to decision makers across the organization.

An asset avatar is a digital representation of a physical asset. It enables you to view and monitor key performance indicators (KPIs) – such as an asset’s sensors, current and historical state, properties and events – no matter where these assets are geographically located. Asset avatars and the assets they represent have a 1:1 relationship – meaning that a single asset avatar is associated with only one asset. Asset avatars integrate data from many different locations to enable better information management, collaboration and lower operational costs.

 

Asset avatars are also defined by an asset avatar type, which is a digital blueprint for a class of physical assets of the same type. The asset avatar type enriches your asset avatar with information about your asset’s identity. Asset avatar types can also be associated with assets and applied to asset avatars that share the same characteristics. When it comes to mapping, asset avatar types help translate your asset’s raw, unintelligible data to easily understood information.

By providing a 360-degree view of your assets, along with seamless integration to your business systems, asset avatars deliver benefits across the organization. For example, bringing information together in new and meaningful ways with ML and AI can help you to predict and prevent failure of assets, reducing costly unplanned downtime.

With this information, you can also extend the useful life of the machines that run your business by predicting when maintenance is needed and automatically scheduling it when the data dictates, further reducing downtime.

Challenges at the Edge Today!

The explosive growth of IoT isn’t hypothetical; it’s happening today in every sector. Digital transformation is here, and while it is changing industries in new ways, that change doesn’t always come easy. Challenges abound, and when it comes to edge computing, companies are finding some common concerns, including:

·       Latency: Cloud computing is becoming more powerful by the day, but some use cases require data to be processed within milliseconds. Sending data to a backend system, whether it’s located in a public cloud, private cloud or data center, introduces too much latency for these use cases.

·       Perishable data: Not all data is useful in the data center. Sometimes data needs to be acted upon right away with low latency. Perishable data, like the data generated by autonomous vehicles, can’t wait until tomorrow. It must be analyzed at the edge for immediate action.

·       Limited bandwidth: Bandwidth at the edge is often limited, so deciding what data to send can be difficult. Sending everything back to the data center is expensive, time consuming and inefficient. To make the best use of resources and to save time and money, it’s essential to prioritize data and send only what you need.

·       Uninteresting data: Assets and sensors generate huge volumes of data, but not all of it matters to you. You might want to know when an asset is operating outside of normal parameters, but perhaps you don’t need to capture, store, transmit, and analyze the data that’s generated when it’s operating normally. You might also want to reduce the amount of data collected to relieve the burden and overhead of data management.

·       Security: Bringing assets onto the network puts them at risk of being infected with malware. With the growth in smart assets and the fact that cyber attacks are escalating in number and sophistication, securing assets and protecting the integrity of the data and network is vital.

What are some of the other IoT Edge challenges you see in this space?

Industrial IoT is a business imperative, act before your competition does!

What are your immediate business objectives? I don’t like second guessing but here are some scenarios where Industrial Internet of things (IIoT) can make a significant difference:

>Are you stuck in a me-too industry and are looking for a competitive advantage to give you an edge in the marketplace?

>Do you desire improved productivity from your assets and people?

> Do you have blind spots in your operations?

>Do you want to explore new business models that support new revenue opportunities?

>Do you want to move from products to as a services customer engagement models?

>Or do you want to improve your product quality and customer experience?

If one or more of these scenarios resonate with your business needs, then it is time to consider implementing IoT solutions to help you in this journey from edge to outcomes.

IoT might sound futuristic, but the technology itself has existed for a long time. Machine-to-machine networks and control systems, data analytics, remote tracking and guidance systems, enterprise applications etc have been aggregated into an IoT Technology Stack to provide a unified view of operations and control.

IoT has proven to contribute to similar outcomes such as:

  • New business models generating new revenue
  • Remote asset monitoring enabling costs savings
  • Improved processes leading to higher yield and reduced wastage
  • Hidden and Dark Data Leading to new Insights and accelerated innovation

As per a recent Forbes study, 66% of successful companies include external vendors on their IoT Planning team. It would be of tremendous value to have an experienced co-creation partner to support with leading practices.

What are your business imperatives and thoughts on co-creation?

Challenges of IoT Analytics Today

Across every industry, analytics and business intelligence are hot topics that promise greater insight, more efficiency, and a more competitive organization. And analytics can truly deliver on those promises, despite the myriad challenges that many organizations face in getting that value.

One of the most pervasive challenges today is siloed data. Data coming from assets may be used only at the edge, while business applications often have their own separate infrastructure, databases, and sets of information. Getting a complete picture of information across the organization is difficult—often impossible. Without the ability to bring data from disparate systems together in a way that enables fast, smart decision making, business outcomes are poor and costs are high.

Not only are data silos a common problem, but data generation itself can be a challenge, too. Many organizations have not been ready to take advantage of analytics capabilities, and don’t generate enough data to derive insights from. Other organizations may generate large volumes of data in the hope that they can derive some business value from it, but lack the filtering capabilities to make sense of data and put the right information in the hands of people who need it.

Good predictive abilities require good data gathering and filtering processes. By bringing together the right data from across the organization, whether it lives at the edge or in the data center, the results can be better business outcomes. The analytics capabilities of IoT product suite needs to address these challenges and deliver fast time-to-value, so that your outcomes are better, faster.

What are some of other IoT Analytics challenges are you seeing in this space?

IoT Platform Flexibility Drives Smart Solutions for Industries

The internet of things (IoT) is gaining traction across industries where new IT capabilities and connectivity are driving innovation and operational efficiencies. Companies across the Industrial footprint are leading the way with exciting breakthroughs in industrial IoT (IIoT) that use big data, analytics and artificial intelligence (AI) to streamline production, enable predictive maintenance and improve quality across the entire product lifecycle.

Let’s explore how.

Manufacturing

Multiple vehicle recalls have cost automotive companies billions of dollars. Product failures and quality issues are costly and can reduce productivity. The impact is more severe if you’re unable to rapidly discover the root cause.

Lumada solves the problem with predictive quality management. Using IoT sensors and 3D cameras, Lumada simultaneously gathers and analyzes human, machine and business data and measures worker movements. It detects assembly-line failures and shortcomings as they happen, and applies corrective action. Real-time predictive analytics identify potentially defective products prior to shipment, and advanced algorithms uncover deviations in work-related activities.

This aggregation and integration of data gives your operations a single consistent view for analysis and monitoring. Multilevel traceability pinpoints defective products so you can remove them before they are shipped. After implementing Lumada, manufacturers have achieved significant reduction in production costs and improved product quality and worker productivity.

Energy

With world energy consumption expected to rise 28% between 2015 and 2040, utility companies, power producers, and commercial and industrial customers are turning to renewable sources to help meet growing demands.[1] However, renewable energy production fluctuates with weather conditions, making it difficult to predict availability from day to day.

Using sensor data, Lumada connects, measures, monitors and manages energy delivery in real time, while algorithms provide insights into weather conditions that impact availability. Machine-learning models prepare generation and trading plans based on market, capacity, demand, weather and pricing data. Lumada reduces downtime by using predictive maintenance to detect potential asset failures, and continually builds intelligence using machine learning and artificial intelligence.

Transportation

As urban populations continue to grow, leaders around the world look for solutions to traffic congestion, inadequate parking and public transit, and an aging transportation infrastructure. From proactively managing shipping orders to enhancing business intelligence for airlines, to optimizing automobile traffic flows in cities, Lumada is reinventing transportation operations.

Lumada collects data to provide real-time monitoring of train performance, for example, and provides insights to optimize maintenance practices and overall operations. Predictive maintenance identifies malfunctioning door hinges, worn brakes, and transmission problems and other issues, so repairs occur before they impact passenger services.

Unique Flexibility for Industries: On-premises, in the Cloud or Both

Some organizations are not comfortable moving all your applications to the cloud, while others are now 100% cloud-based. It can run on the edge (on-premises), as a hybrid cloud or fully in the cloud. In addition to the examples described here, there are many innovative ways industries can put the power of IoT Platform to work creating smarter, more efficient and productive solutions.

Watch Hitachi’s Lumada IoT Platform in action:

——-

This article was originally published at – Hitachi Vantara Community