Internet of things and predictive manufacturing – I

As a mechanical/aerospace engineer, manufacturing is obviously in my DNA, but that’s not why I am writing this article. Take a moment to consider a few of the top inventions of mechanical engineering: the development of the automobile, space vehicles, utility (power generation) facilities, agricultural mechanization equipment, the airplane, mass production of integrated – circuits […]

As a mechanical/aerospace engineer, manufacturing is obviously in my DNA, but that’s not why I am writing this article. Take a moment to consider a few of the top inventions of mechanical engineering: the development of the automobile, space vehicles, utility (power generation) facilities, agricultural mechanization equipment, the airplane, mass production of integrated – circuits (computers), heating, ventilation, air-conditioning and refrigeration (HVAC&R) equipment, and the development of codes and standards. You will easily see that these inventions taken together are synonymous with what we usually refer to as human civilization. Even the Internet revolution owes its existence in a critical way to the ability of mechanical engineers (with some help from electrical engineers) to mass produce computers, smart phones, and tablets. The relationship between these inventions and the economic development of a nation needs no discussions, as manufacturing represents the mainstay of any nation: excel at it and you’ll be okay, fail at it and your people will starve!
Towards the foundational elements for predictive manufacturing (PM), in today’s article, I’ll focus on the AIDC technologies, which are crucial elements of the Internet of Things (IOT), followed by the state of the art of IOT in the next article, and then concluding in two weeks, with PM.
 Here is the complete picture: In PM, you manufacture hardware that you sell to your customers, but during the manufacturing process, you embed small gadgets (sensors and actuators – if you will) that will monitor the hardware and create data for you about the hardware as it leaves your manufacturing floor, gets shipped and delivered to your customer, goes through the deployment at the customer’s premises, and begins operation at the customer’s site or wherever the hardware may be located. Data collected on the hardware could include the temperature, humidity, location, some diagnostics on the state of the hardware, and so on. These data, which have to be communicated to you, the manufacturer – most likely via some cloud interactions, are instantaneous in time, meaning that you have an exorbitantly huge amount of data to process (analyze) for business intelligence. Processing the collected data allows you to assess the state of the hardware, communicate with the hardware, and remotely fix impending problems with it, even before the problems occur. By this means, you prevent a downtime at the customer’s site, resulting in a happy customer and presumably better business for you. Thus, in this model, you don’t just sell your customer a piece of hardware and disappear: you continuously and remotely service the product based on the predictive power you acquire from the data you collect; hence, the phrases “Predictive Manufacturing” or “Product as a Service.” General Electric Aviation is one of the companies that have deployed PM – in its aircraft engines.
 So, the elements of PM include: 1) the technology to collect data, 2) the Internet capability and bandwidth to transfer the data to some cloud site, 3) the storage capacity of the cloud servers, 4) the Big Data analytics wherewithal to process petabytes of data, and 5) the feedback to the hardware – using the Internet.
Typically, obtaining the data from the hardware is done via the use of some sensors and actuators in the radio frequency identification (RFID) technique, although data acquisition technologies are collectively referred to as AIDC. AIDC devices collect information from objects, images, sound, or even from persons, without any manual data entry. There are many types of AIDC technologies: biometric devices, one- and two-dimensional barcodes, magnetic strips, RFID devices, and smart cards. In biometrics, certain metrics of a person (such as fingerprints, iris (eye) patterns, or voice characteristics) are captured by the device and compared against stored data, to find a match. A reader or scanner in the device captures the bio attribute, followed by a conversion of the data into a digital format that can be stored and operated upon.
Barcodes contain small images of spaces and lines or bars that identify a particular location, person, or product. A laser-beam-based barcode reader converts information from the images to digital data. In one-dimensional barcodes, information are only stored horizontally, with a limit of only 20 characters. Two-dimensional barcodes can store information both horizontally and vertically, with a capacity for up to 7,089 characters. Magnetic strips are what you have on conventional credit or ATM cards. Here, information is written on iron-based magnetic particles in a tape by magnetizing tiny bars along certain directions. Changes in the magnetic field due to the writing process can be detected by a magnetic strip reader. Smart cards are plastic cards with an embedded chip.
 The holy grail of data acquisition for PM purposes is RFID. In fact, it appears that the RFID community created the phrase “Internet of Things,” which refers to the process of storing the data about products (things) in the cloud to make it available to trading partners. Cartons, pellets, or individual items can be tagged. The RFID technology is more advanced than QR (Quick Response) or barcode technologies, for example, as it allows you to collect data in time as the product moves from place to place. Moreover, data related to the condition of the product can be collected.
 
 The article continues next week.