Image Processing Reference
In-Depth Information
with determinism of wireless communications in the face of interference (as explored in Section .)
usually result in the requirement that wireless time-critical control systems go through extensive and
often costly verification prior to deployment.
23.5.2 Networks for Diagnostics
Diagnostic information that is sent over the network often consists of large amounts of data sent
infrequently. For example, a tool monitoring system may capture spindle current at kHz. he entire
current trace would be sent to the diagnostic system after each part is cut (if the spindle current
is used for real-time control, it could be sent over the network every ms, but this would then be
considered control data). he diagnostic system uses this information for higher-level control, such
as to schedule a tool change or shut down a tool that is underperforming.
Diagnostics networks are thus usually set up to support large amounts of data with the emphasis on
speed over determinism. Ethernet is the dominant network protocol in system diagnostics networks.
As with control, diagnostics is often set up in a multitier hierarchical fashion, with different physical
layer technology (e.g., wireless, broadband, and fiberoptic) utilized at different levels to support the
data volume requirements. Also, a variety of data compression techniques, such as COS reporting
and store and forwarding of diagnostic information on a process “run-by-run” basis, are often used
in communicating diagnostic information up the layers of the network hierarchy [,]. Wireless is
an ideal medium for diagnostics because determinism and high levels of data integrity are usually
not required, and diagnostics systems can often take advantage of the flexibility afforded by the wire-
less media. As an example, a personal digital assistant hand-held system can be designed to collect
diagnostics data wirelessly from a nearby machine and perform system health checks. Maintenance
engineers thus could utilize a single portable unit as a diagnostics tool for all machines in a factory.
As noted in Section ., diagnostics networks enable diagnostics of the networked system rather
than the network itself (with the latter referred to as network diagnostics). Both types of diagnostics
are commonly used in manufacturing systems. Many network protocols have built-in network diag-
nostics. For example, nodes that are configured to send data only when there is a COS may also send
“heartbeat” messages every so often to indicate that they are still on the network.
23.5.3 Networks for Safety
One of the newest applications of networks in manufacturing is safety []. Traditionally, safety
interlocks around manufacturing cells have been hardwired using ultrareliable safety relays to ensure
that the robots and machines in the cell cannot run if the cell door is open or there is an operator inside
the cell. This hardwiring is not easy to reconfigure and can be extremely difficult to troubleshoot if
something goes wrong (e.g., a loose connection). Safety networks allow the safety systems to be recon-
figurable and significantly improve the ability to troubleshoot. hey also allow safety functions to be
more easily coordinated across multiple components, e.g., shutting down all machines in a cell at
the same time, and also coordinating “soft shutdown” where appropriate (safe algorithms for gradual
shutdown of systems without damage to systems and/or scrapping of product). Further, safety net-
work systems often provide better protection against operators bypassing the safety interlocks, and
thus making the overall system safer.
Safety networks have the strongest determinism and jitter requirements of all network function
types. Safety functions must be guaranteed within a defined time; thus, the network must provide
that level of determinism. Further, the network must have a deterministic heartbeat-like capabil-
ity; if network connectivity fails for any reason, the system must revert to a safe state within the
guaranteed defined time; this is often referred to as “fail-to-safe.” CAN-based networks are popular
