边缘智能—提高生产力并降低成本

As factories strive to boost productivity and improve operational costs, the demand to deliver new technology that empowers Intelligent Edge is increasing. For those of you asking yourself “what does the edge mean?”, at Analog Devices, we define the edge as where the machine meets or interacts with the real world!

In factory automation, empowering Intelligent Edge means reducing the amount of lost productivity a factory experiences in a year. The biggest contributor to this is factory downtime or when the production line stops causing the company to lose money. In fact, factories experience an average of 800 hours/year of production line shutdowns or an average of 15 hours/week.¹ This is a substantial impact on revenue and profit for a company. Take for instance that a car manufacturer loses close to $22,000 per minute when the factory stops manufacturing. This means it loses $1.3M/hour or close to $20M per week. Empowering intelligence at the edge has already made an impact on the manufacturing line improving productivity by 10% and achieving a 20% savings in maintenance cost. So, the net result of empowering intelligence at the edge in factories works to keep production lines running by preventing costly production line shutdowns.

While it’s clear that empowering intelligence at the edge provides a boost in productivity and reduces operating costs, the real question is “What does it take to empower intelligence at the edge”?

It takes a new way of thinking!

As semiconductor suppliers, we need to deliver solutions that enable intelligent sensors and actuators, support software configurable IO, and provide advanced diagnostics. Let’s review the importance of these three critical elements and the key capabilities they provide in empowering intelligence at the edge.

Sensors are found everywhere! They have become ubiquitous in our everyday lives. In the manufacturing environment, all manufactured products require an array of sensors that work in unison to help machines detect an object, determine the distance to an object, configure the colors and composition of an object, and monitor the temperature and pressure of an object or liquid.

Commissioning new sensors to replace damaged sensors or adapting a piece of equipment to enable the manufacturing of a different product is labor intensive and it contributes a significant cost burden due to loss of productivity. The cost of sending a technician to the factory floor to change a sensor and then recalibrate it to the correct manufacturing parameters impacts factory throughput. If we multiply this same level of maintenance for every sensor across a factory, changing or reconfiguring a sensor is the greatest expense all manufacturing lines incur.

IO-Link^® is an exciting new technology that allows intelligent sensing all the way down to the machines on the factory floor. This new technology enables flexible manufacturing to improve factory throughput and operational efficiency. IO-Link is a technology that converts traditional digital or analog sensors into intelligent sensors by providing bidirectional information exchange with the sensor. This technology adds a new level of intelligence and capability to remotely commission the sensor as well as the ability to react in real time by making on-the- fly adjustments to the sensor parameters. Industrial automation machinery now has a newfound intelligence to dynamically respond to real-time operating conditions based on the health and status of a network of sensors located across the factory floor. By tapping into this rich sea of end-to-end information across a network of intelligent sensors, a facility can create a mapping of its factory floor to provide better real-time information to an overarching AI monitoring solution that can rapidly identify manufacturing bottlenecks, points of failure as well as provide a new capability to optimize the entire factory floor for better operational efficiency.

The way IO-Link technology simplifies the commissioning process and improves factory throughput is by making sensors interchangeable through a common physical interface that uses a protocol stack and an IO device description (IODD) file. This allows technicians to quickly commission a sensor that results in reducing factory downtime and allowing the manufacturing line to be reconfigurable on-the-fly.

The adoption of IO-Link sensors continues to accelerate as companies realize the benefits of having a common interface that makes exchanging various sensors such as pressure, proximity, and temperature as easy as plug and play. According to Research and Markets, the IO-Link market continues to grow, and it is expected to reach $12 billion by 2023 from $3 billion in 2018, at a CAGR of 33.56%.²

While it‘s clear that IO-Link technology is the catalyst behind a set of new intelligent sensors, IO-Link is also providing new opportunities that bring intelligence to the edge through IO-Link hub solutions. These new IO-Link hubs provide a simple way to add analog and digital IO expansion channels as well as the integration of intelligent actuators such as solenoid and motor drives.

The IO-Link hub provides a simple way to expand the type and number of channels needed to support unexpected manufacturing line reconfigurations. These IO expansion hubs provide a solution that leverages all the benefits of IO-Link technology and simplifies the task to add digital and analog IO ports. This new class of products enables the commissioning of the sensors via the IO-Link hub, which reduces factory downtime. Examples of these solutions include Omron’s IO-Link Hub NXR product family where they boast of achieving a 90% reduction in setup and commissioning times.

Software-configurable digital and analog IO solutions allow automation engineers and technicians the convenience of providing a universal IO port that can be commissioned remotely. Comparable to the benefits that IO-Link provides, this new class of digital and analog software-configurable IO products simplifies factories’ wire marshalling burden and provides flexibility to physically connect any digital and analog IO sensors or actuators to any unassigned digital and analog IO port. This software-configurable technology is more cost-effective and increases the channel density on the factory floor.

Higher levels of diagnostic capabilities continue to provide a richer data set that improves real-time, edge-based decision-making to improve productivity and operational integrity on the factory floor. These powerful manufacturing-based AI algorithm platforms are expected to grow from $1B in CY18 revenue to over $17B by CY25, or a CAGR close to 50%.³ During this time, machine learning is expected to be the highest growth segment in AI due to the rapid investments being made to implement smart factories. The driving force behind this growth stems from the abundance of health and status information being generated from a network of IoT-powered devices, algorithms providing predictive analytics, and machine-vision cameras monitoring the quality of products as well as evaluating the status and operational health of the machines.

At the IC level, more and more information is monitored, collected, and communicated via SPI bus to and from a microprocessor. The volume of these IC datagrams continues to multiply as they carry critical information such as the temperature status of a device, overvoltage, overcurrent, open wire detection, short circuit detection, overtemperature warnings, thermal shutdown, and CRC. If we take a step back now and multiply the number of semiconductors providing datagrams across the entire breadth of equipment on a factory floor, it becomes clear that a diagnostic mapping of the factory floor can be achieved to anticipate, identify, and diagnose manufacturing line failures.

One thing is clear, by empowering this new way of thinking, smart factories can take advantage of these new capabilities to improve throughput and increase productivity. As these new technologies continue to mature, the next generation of AI algorithms will become the beneficiaries by leveraging the higher quality of real-time data being generated from these solutions. As a result, these new self-aware capable machines will automatically implement AI-generated solutions to keep a manufacturing line operational until it is repaired or serviced by a technician. This era of self-aware machines will inspire the next big thing in industrial automation.