Six Key Machine Control Challenges Solved

When designing your next-generation machine, cutting-edge technologies in control systems, software architectures, and electromechanical components differentiate your automation system from the competition. This article explores the top software and hardware challenges machine builders face today and offers a performance-driven approach to solving these challenges.

Software Challenges

1. Combining Multiple Software Architectures
Software architecture and constructs best-suited to program your machine depend on the key performance metrics that need optimization. To optimize reaction time to faults, you want a “reactive” event-driven architecture. If your automated inspection system demands analysis of the acquired image, you want an architecture optimized for signal processing functions. A state-machine architecture best suits your batch process or packaging machine. Semiconductor wafer-processing machines requiring advanced model-based control algorithms benefit from an architecture that supports real-time integration solvers. Machines needing high-speed measurements with real-time analysis benefit from dataflow architectures. Simple logic and arithmetic operations are best performed with a sequential architecture, such as with a programmable logic controller (PLC).

Identifying and applying the right combination of software architectures to solve your automation problem is the challenge for the future. IEC 61131-3-based languages, such as ladder logic and functional block diagrams, suit a majority of discrete manufacturing applications that primarily involve on/off operations. However, as modern machines require multiple changeovers and involve predictive maintenance routines, languages such as National Instruments LabVIEW emerge as a single development platform that can combine state machines for defined modes of operation, dataflow for monitoring routines, real-time integration solvers for precision control, events for fault response, and sequential logic for on/off operations.

Figure 1. Different programming models such as state machines, integral solvers, sequential logic, dataflow, and events are best run on a specific target such as a PC, PLC, PAC, DSP, FPGA, and microprocessor.

2. Write Once, Run Anywhere
While the “write once, run anywhere” concept is gaining acceptance in the consumer world with .NET and Java technologies, it is still far from reality in the automation world. An IEC 61131-3-compliant program written in ladder logic for one PLC may not run on a similar PLC from another vendor. Companies are thus forced to standardize on a single vendor to ensure interoperability, in many cases resulting in non-optimal performance and higher total costs for the complete system.

The challenge for the future is to write your control program once and then deploy the same program to a variety of PC, PLC, or embedded targets. The automation engineer needs the ability to choose among PLC, programmable automation controller (PAC), microprocessor, digital signal processor (DSP), or FPGA targets based on the price-performance requirements of the automation system. The NI LabVIEW graphical development platform offers a variety of modules that help you port your code to different platforms. You can develop your program graphically using LabVIEW, then deploy your application on a real-time operating system using LabVIEW Real-Time, port your code to FPGAs using LabVIEW FPGA, deploy code to DSPs using LabVIEW DSP, and use LabVIEW Embedded to transfer code to 32-bit microprocessors.

3. System Validation
Code reviews, which most development processes now include, ensure reliability of the software created. However, the tight coupling of software with hardware in today’s electromechanical systems necessitates complete system validation. Engineers are moving from just executing a “deploy” phase to going through “design-prototype-deploy” phases. The design phase includes simulating mechanical, thermal, and flow characteristics of the hardware components in the system in addition to the algorithms and control logic that would control these components. The prototyping phase involves the virtual or physical prototyping of both mechanical and control designs to help engineers develop a proof of concept before final implementation. The deploy phase involves deploying the control algorithms and logic to PLCs, PACs, or embedded targets and assembling the mechanical components, such as servo actuators, pneumatics, and hydraulics.

The challenge for engineers is to execute on each of the design, prototype, and deploy phases efficiently without losing time porting code to different development tools in each of the phases. LabVIEW offers a single development platform for each of the phases, with interfaces for mechanical (SolidWorks, COSMOSWorks, COSMOSMotion, MSC.ADAMS, MTS I-DEAS), math (MathCAD, Mathematica, The MathWorks, Inc. MATLAB®), electronic (Multisim, Ansoft, Anadigm, SPICE), embedded (Code Composer Studio), and control (MATRIXx, The MathWorks, Inc. Simulink®, CarSim) design tools for the design phase; integrated analog I/O, digital I/O, motion, vision, and communications on the PC platform for the prototyping phase; and a variety of deployment targets (PC, PAC, FPGA, DSP, microprocessor) for the deploy phase, depending on price-performance requirements.

Figure 2. LabVIEW offers a single environment for graphical system design from design to prototyping to deployment of the final system.

Hardware Challenges

Tomorrow’s automation systems will perform complex tasks on a variety of different products, often simultaneously. The hardware challenges in designing such systems are to achieve throughput, yield, and uptime while accomplishing a complex automation task.

1. Throughput
The speed of your machine directly affects throughput. To achieve higher speeds, use mechanical components with less friction, such as a linear motor instead of a ball screw actuator. You can improve control system speed by using embedded technologies, such as FPGAs with loop rates of 1 MHz instead of traditional PLCs with loop rates of 1 kHz. Servo systems continue to dominate machines moving away from traditional gear/cam-based systems.

Programmable automation controllers, such as National Instruments CompactRIO, that include a programmable FPGA and a floating-point processor running a real-time operating system are well-suited for high-throughput applications, such as sorting or assembly.

2. Yield
Reducing waste with higher repeatability is key to achieving higher yields. Programming the machine to follow desired motion control profiles is critical to repeatability. You can achieve this by tuning your motors with short settling times and less overshoot for a step response. For better tuning, use model-based control methods to arrive at the right PID tuning parameters or replace traditional PID algorithms with model-based control algorithms. Technologies, such as automated inspection and RFID, play a large role in sorting rejects, which speeds up the process.

LabVIEW control design and simulation tools combined with the LabVIEW SoftMotion Development Module help you create custom motion controllers with model-based control algorithms, such as linear quadratic regulator (LQR) or H-infinity for better repeatability and better yield. The NI Vision Development Module helps you create automated inspection systems with more than 200 image processing and machine vision functions.

Figure 3. You can implement a variety of control strategies, including PID, MFA, and model-based control, to shorten time-to-market, decrease changeover time, or increase repeatability.

3. Uptime
A modern packaging machine needs to handle more than 10 products on the same manufacturing line. It is not just the reliability of the components in the system, but also changeover times between different products that affect system uptime. You can improve changeover times by setting the control algorithm to adapt to a different set of conditions with a different product on the line. Model-based adaptive control is an emerging area that eliminates tuning and helps the control system adapt to changes in the system. You can improve system reliability by incorporating intelligent monitoring and predictive maintenance as part of the system. Vibration monitoring, data logging, alarming, and enterprise communication play a key role in improving future system reliability.

You can deploy Cybosoft’s Model-Free Adaptive (MFA) control algorithm for LabVIEW on any LabVIEW Real-Time or LabVIEW FPGA target to eliminate tuning and help adapt to load changes in the system. NI Compact FieldPoint and PXI platforms can help you incorporate intelligent monitoring and predictive maintenance solutions with high-speed analog I/O for vibration monitoring, data logging, alarming, and enterprise connectivity.

In the future, the machine control industry will face challenges such as combining multiple software architectures, system validation, and driving throughput, yield, and uptime in a complex automation system. The key to winning in automation today is selecting software and hardware components for your system that are best suited for the task at hand and that can scale in functionality tomorrow.

Rahul Kulkarni
Product Marketing Engineer
rahul.kulkarni@ni.com

Learn how you can build low-cost, industrial machine control systems with NI CompactRIO in this Webcast on Demand.

This article first ran in the April 4, 2006 issue of NI News.


MATLAB® and Simulink® are registered trademarks of The MathWorks, Inc.

16 ratings | 4.31 out of 5

Read in English German | Print

Reader Comments | Submit a comment »