Smart Sensor and Prognostics for Self-Maintenance Machines

By Advantech eAutomation

National Science Foundation Industry/University Collaborative Research Center for Intelligent Maintenance Systems (IMS)

Maintenance has never been regarded as the most appealing aspect of manufacturing, either by industry or academia as it is often considered a dirty and boring job which interferes with meeting production demands. In traditional maintenance, when a breakdown occurs, the entire production system must be shut down until maintenance personnel can be dispatched to rectify the problem.

In order to prevent problems before they occur, many industries have come to rely on preventative maintenance - minimizing unexpected downtime by performing inspections, parts replacement, and other maintenance activities, at predetermined intervals. This type of maintenance can be compared to the frequent oil changes that we all must receive for our cars. When these oil changes are performed is based on generic performance data that cannot take into account each car’s individual operating conditions, which begs the question; are all of the oil changes we have performed necessary? When you consider the economical impact of tens of millions of oil changes every year, this question becomes more important.

Clearly, the downside to preventative maintenance is cost, even for a problem as common as changing an engine’s oil, there is no definitive evidence, or consensus on the best time to perform maintenance. Many people still stick to long standing belief that an engine’s oil should be changed every 3000 miles, though some are beginning to challenge this.

Is there an alternative method for performing/scheduling maintenance? Imagine having a machine, or a piece of equipment that can tell you its current health status; and the degree to which that status deviates from normal or healthy. This ability for machines to perform self-assessment raises the field of maintenance to more than a low-level topic of study. Knowing the health status of a machine or product reaches beyond the functions of traditional maintenance, and affects a product’s entire life cycle.

This is the question that drives the National Science Foundation Industry/ University Cooperative Research Center for Intelligent Maintenance Systems (IMS). Over the last six years, IMS Center researchers have developed a toolbox of algorithms, dubbed the Watchdog Agent®, which is capable of converting sensor data and operating conditions to useful information that is indicative of a machine’s level of performance and health. Watchdog Agent® monitors degradation rate of components and predicts the likelihood of failure and health of machine. This approach is called predictive maintenance, or prognostics.

Current Gaps and Future Trends
This new paradigm shift to an in-machine prognostics and intelligent maintenance systems methodology relies on three levels of intelligence:
1) Machine intelligence: Many new industrial machines are equipped with remote monitoring, data collection and diagnostic systems—a similar trend is evident in the automotive and aviation industries. For example, with systems such as OnStar®, the car’s health data can be collected and transmitted to a remote site. In addition, GE is currently able to collect a jet engine’s in-flight status for thousands of its engines. However, the capabilities of prognostics, degradation assessment – or what is called ‘health information intelligence’ – in comparison to functional intelligence are missing from today’s maintenance practices. Health intelligence includes health assessment, performance prediction, failure prediction, and the ability to interact with functional intelligence. A machine capable of reconfiguration, compensation, and self-maintenance can be considered to possess both functional and health intelligence.

2) Operations intelligence: When one machine’s health condition is known to be degrading, or to be near failure, maintenance decision making is relatively easy. What should be done when there are hundreds of machines? The ability to utilize prioritization, optimization, and responsive maintenance scheduling in dynamic environments is referred to as operations intelligence. This particular paradigm has yet to be implemented effectively.

3) Synchronization intelligence: The third maintenance gap which directs the IMS Center’s activities is synchronization intelligence. Maintenance needs to be a part of asset utilization and business enterprise systems. Synchronization intelligence will enable automatic flow of information between production and demand.

Do We Need to Monitor Everything?
“Well, who's going to monitor the monitors of the monitors?” A fair question posed by Carla Dean in the movie Enemy of the State. Sensors are costly and not every component or system of a machine or product needs to be monitored. In fact, only the critical components which cause downtime should be monitored. For components with a high frequency of failure and low failure cost, an adequate number of spare parts should be available, while components with a high frequency of failure and a high cost may require a change in design. Low frequency of failure and low cost is where the system should move toward. Often a feasibility study using Computerized Maintenance Management Systems (CMMS) and other techniques determine the critical components which need to be monitored.

In this context, monitoring is more than just data collection. The methodology for determining critical components, as practiced at the IMS Center, is comprised mainly of the following steps:
  • Determine the target machine or component’s current health condition and synchronize that data with the relevant signals collected.
  • Streamline and preprocess the original data - this step eliminates extraneous data, and facilitates concentration on that which is most relevant.
  • Extract features or health indicators. A feature might be multi-dimensional.
  • Using these features, operating conditions, fault history and normal and problematic working zones should be determined.
  • Based on previous information, Confidence Values (CV) can be calculated. A CV near one represents a healthy condition, while a CV near zero is indicative of significant degradation.
  • By establishing the rate of change and trend of CVs, the likelihood of failure in the near future can be estimated.
  • A simple chart displaying the target machine’s health condition, in a clear, understandable form, will be presented to human user, thus transforming data to useable, timely information upon which action can be taken.

    The Watchdog Agent
    The Watchdog Agent is a toolbox of algorithms that can autonomously asses and predict the performance degradation and remaining life of machines and components. This information can be fed to a closed-loop product life-cycle management system. The Watchdog Agent provides machine level intelligence and is synchronized with the operation and synchronization level intelligence. It includes signal processing and feature extraction, diagnoses, performance prediction, and performance assessment modules. Each module includes several algorithms which can be reconfigured according to application parameters.

    The toolbox consists of tools such as Neural Networks, Fourier Transform, Support Vector Machine, Self-organizing Maps, Fuzzy Logic, Logistic Regression, Hidden Markov Models, Bayesian Belief Networks, Match Matrix, Autoregressive Moving Average, Time-Frequency Analysis, in addition to others.

    The key components are Advantech’s UNO-2160 or UNO-2170, both embedded automation computers with Intel Celeron processor and CeleronR M processor respectively. They both support WindowsR CE 5.0, XP Embedded, and Linux ready solution. Also included is the PCM-3718H, a 12-bit DAS module with programmable gain.

    Watchdog Agent follows an open architecture design. New tools can be added easily and, depending upon the application, some can even be disabled. Based on restrictions such as memory, processing power, and power consumption, tools can be reconfigured to suit many conditions. The user’s input can be captured based on a Quality Function Deployment (QFD) selection tool. Depending on the application, as defined by the user, and any expert knowledge integrated into the QFD selection tool, the best tools from each module will be selected for the defined situation.

    Watchdog Agent Tools

    One of the Watchdog Agent Hardware Platforms

    The Next Step: Self-Maintenance

    The ability of a machine to adjust its own functionality according to its health status is an integral part of a self-maintenance paradigm. Self-maintenance requires both functional intelligence and health intelligence. Functional intelligence provides the current operating condition information for the health assessment system. Health intelligence evaluates the current health status and degradation rate, and predicts the likelihood of failure. This information can be fed into a functional intelligence module – e.g. controller – and the machine’s operation can be adjusted accordingly.

    Adjusting operating parameters of a machine is not the only way in which self-maintenance can be perceived. Activating a self-tuning or self-service function is another way to realize some degree of self-maintenance. The function of a copy machine is a typical example used in self-maintenance literature. Tuner adjustment and calibration can be triggered when performance degradation is detected. Most literature in self-maintenance focuses on adaptive controllers using neuro-fuzzy systems. Self-maintenance does not need to be fully autonomous in the context of the manufacturing industry. The purpose of self-maintenance is to provide enough time for maintenance personnel to become available and proper downtime to be scheduled. In a manufacturing industry setting, the maintenance crew can conceivably fix the underlying problem. In applications, such as in the aerospace industry, in which human intervention is not possible, a higher degree of self-maintenance is desired.

    Sensors have been relied upon, and trusted to monitor systems. But what if a sensor fails? Such a failure could be complete or incipient, which will affect the readings issued from the sensor. There are currently two main approaches to solving this problem: hardware redundancy and analytical redundancy. In the hardware redundancy approach, which is common in safety critical systems such as nuclear power plants or airplanes, additional sensors are used to detect problems in one sensor, or to provide a more accurate reading when one sensor fails by effectively replacing that sensor. The major downsides to this approach are the additional cost and the additional space required for extra sensors. The analytical redundancy approach uses the analytical or experimental model of a system to detect inconsistencies in that system’s behavior, which could be generated by sensor failure. Few sensors are needed to build an analytically redundant system; however, identifying a model for the system is not feasible in every situation.

    Currently the IMS Center has about 30 global company members and sponsors including: DaimlerChrysler, GM, Ford, Toyota, GE Aviation, Harley-Davidson, Omron, Komatsu, Toshiba, Caterpillar, Chevron, Festo, Boeing, P&G, AMD, BorgWarner, Bosch, Parker-Hannifin, McKinsey& Company, etc. The Watchdog Agent has been successfully implemented on several company test-beds. For instance, Harley-Davidson installed the Watchdog Agent on a GROB aluminum cutting machine, allowing the Watchdog Agent® to be able to automatically convert sensory data to health information, and predict machine degradation, as well as its remaining life. In another project, predicting compressor surge contributed to major savings for Toyota in both costs and time. The IMS Center’s practices and test-beds have proven the veracity of predictive maintenance as an emerging field with many promising applications in the future. For more information visit

    Advantech Corporation, eAutomation Group –Founded in 1983, the eAutomation Group of Advantech Corporation is a pioneer in eAutomation technology; combining connectivity, flexibility and ruggedness with today's most reliable PC-based automation technologies. Product offerings include: Open HMI platforms, Industrial Ethernet, Automation Controllers and Software, Fanless Box PCs, Distributed I/O Modules, Plug-in I/O and Industrial Communication for a wide array of industries.

    With more than 2,700 talents, Advantech operates an extensive support, sales and marketing network in 17 countries and 33 major cities to deliver fast time-to-market services to our worldwide customers. Today Advantech offers over 700 products and solutions, adding approximately 30 new products each year.

    US operations for the eAutomation Group are based in Cincinnati, OH (800-205-7940). Advantech Corporation, eAutomation Group, 1320 Kemper Meadow Drive, Suite 500, Cincinnati, OH 45240.
    Website for Advantech customers in North America:
    Email: [email protected]