Complex Government System Health Monitor & Predictive Tool

Author(s):
Brian Domrese - Life Cycle Engineering Inc.

Industry:
Aerospace/Avionics

Products:
LabVIEW, PXI/CompactPCI

The Challenge:
The National Defense Budget continues to diminish. The resulting reduction in the number of submarine assets and supporting maintenance dollars jeopardize the Navy’s ability to complete their complex mission requirements. The U.S. Navy Headquarters Assistant Program Manager for submarine Towed Systems (PMS 435) contracted LCE and Areté to design, construct and demonstrate a proof-of-concept solution. Based on the Government’s desire to rapidly field the expected solution Fleet-wide, a six month period of performance was negotiated. The currently installed monitoring system for the contracted baseline system provided only limited analog real-time information with no data collection, analysis or prediction capability.

The Solution:
The LCE and Areté Team designed, modeled, demonstrated proof-of-concept, and delivered a Towed System Health Monitor System (THMS) prototype for a OA-9070/TB-23 Baseline configured SSN 688 Class Submarine. The product was developed using a SCXI system, tailored LabVIEW 5.1 interface software, and National Instruments data acquisition hardware and embedded VISUAL C++ Bayesian Belief Network (BBN) "intelligent" code. This demonstrated capability enables the Navy to proactively maintain complex and troubled systems including the OA-9070/TB-23 Towed System. The THMS proof-of-concept demonstrated an effective, greatly enhanced data monitoring collection, assessment, storage and prediction capability while keeping cost far below other alternatives. This delivered THMS product was customized to meet the requirements of a Government client with specific needs. This application can be easily and quickly re-engineered and tailored to meet the need of a large host of clients in the search for improved system and component reliability, reduced corrective maintenance, increased operating time, and improved profits.

Abstract

The U.S. Navy contracted NI Alliance member Life Cycle Engineering, Inc. (LCE) and Areté to design, construct and demonstrate a proof-of-concept solution. The Government customer, PMS 435, prioritized the improvement in submarine thinline towed array reliability and operability. The Fleet Commanders have repeatedly emphasized their support and need of these critical submarine systems and focused on solving current maintenance problems and related component failures. The future life cycle support of these systems requires an ongoing program of monitoring and replacement of weak components prior to causing system failure and potential mission abort. The THMS Introduction User Display pictured below summarizes the main functional segments of the delivered data collection, analysis and prediction capability. Specifically:

• The Total and Sub-Group Health User Screen provides the probability that the Thinline Towed System will operate successfully. The probabilities are calculated for the total system and the four major sub-groups; Signal Path, Mechanical, Electrical, and Hydraulic. The features, options, theory and mechanics for each available THMS user screen are detailed in the delivered Technical Data Package.
• The Maintenance Log feature allows the user to enter maintenance actions that were performed on the OA- 9070/TB-23 System. These actions are entered by selecting the appropriate system Sub-Group and the appropriate maintenance item. The logging of Planned Maintenance is also included in the Maintenance Log.
• The Sensor Panel User Screen displays the real-time values of the 39 channels of THMS inputs. The screen is divided into the Hydraulic, Mechanical, Electrical, and the Signal Path Sub-Groups. For the Baseline THMS, data archiving was not included as part of the original concept. Data logging will be incorporated into the hardware/software suite and is planned for subsequent phases of the THMS for Towed Systems development. The THMS hardware and embedded software is easily upgraded to 512 channels dependent on the customer requirements.
• Government THMS related continuation tasks have been technically approved to enhance, install shipboard and validate the prototype, and
• Additional THMS future enhancements include the incorporation of a Maintenance Expert Module which will facilitate the troubleshooting and repair of identified and predicted-to-fail components. This will include hyperlinks to maintenance manuals and potentially video sequences demonstrating appropriate repair procedures.

To achieve the desired functionality within the short period of performance, the system was developed on a PXI/SCXI system running Windows NT with installed National Instruments IO hardware. The use and tailoring of LabVIEW 5.1 was employed as the interface software due to its flexibility, reduced developmental time, compatibility with the Visual C++ BBN “intelligent” code, and cost effectiveness.

The delivered health monitor provides indications of onboard system and component reliability and directs operators and maintenance personnel to weak links in system effectiveness. Predicted future failures are identifiedto facilitate corrective action and thus minimizing the down time required to support troubleshooting, repair material ordering and receipt, replacement and the re-testing required for problem resolution. THMS is most effective when tailored and applied to systems that have significant consequences if failures occur; i.e. schedule disruption, the need for retrieval of equipment and products, etc.

Introduction

The Thinline Health Monitoring System (THMS) monitors, assesses and predicts the total and sub-group system health and calculates the probability that the system will operate successfully. The prototype THMS currently collects 39 input signals from various onboard Thinline Array and Handling System (TLTAS) components and sensors. In addition, THMS has the capability to monitor and analyze real-time and stored input data and predicted output data.

The THMS hardware and software descriptions are summarized below. The THMS product was customized to meet the requirements of a Government client with specific needs. This application can be easily engineered and tailored to other mechanical, hydraulic, and electrical systems requiring frequent maintenance or systems at risk of costly failures. The customer benefits include improved system and component reliability, reduced corrective maintenance, increased operating time, and improved profits.

Hardware

The THMS implements the power of the PC in a rugged, stand-alone package that is suitable for harsh environments. The physical size and weight requirements for THMS causes insignificant shipboard impact. THMS is based on the PXI (Compact PCI) specification, but adds sophisticated timing and other advanced features, needed for complex monitoring and analysis. THMS uses hardware and embedded software to provide the operator with information related to the health of the TLTAS. The analysis of the raw input information predicts the current and future probability of successful operation. THMS hardware provides the interface between the analog shipboard TLTAS equipment sensors, and the THMS embedded software. The THMS is comprised of the following components:

• PXI Chassis with Real-Time System Integration (RTSI) Back-plane Bus
• Embedded Computer with 4 Gb Hard Drive
• 32 Channel 1.25 Mb A/D Converter expandable to 512 Channels
• Terminal Blocks
• Interconnecting Cables
• Electronic Components for Signal Conditioning and Attenuation
• Flat Screen Monitor
• Keyboard & Mouse, and
• Embedded Baseline THMS Interface and BBN Software Codes

THMS taps the input signals through a high impedance PXI-6979E multifunction I/O card. The architecture of these cards include:

• NI-PGIA gain-independent, fast-setting-time instrumentation amplifier
• DAQ-STC counter/timer
• RTSI multi-board/multi-function synchronization bus
• MITE PCI bus master interface, and
• Shielded, latching metal connectors

The various input signals are paralleled into the THMS system through a breakout connection and terminal block. These signals are feed into the THMS to perform both single and multiple A/D conversions of fixed or infinite number of samples. The THMS can perform multiple A/D conversion operations with programmed I/O, interrupts, or DMA. The embedded THMS interface software provides for operator interface and governs all aspects of interaction between the hardware (computer) and the embedded BBN software.

Software

The THMS interface software is a tailored version of two off-the-shelf products, LabVIEW and MS Visual C++. The software interface module collects and stores real-time input data from selected OA-9070/TB-23 sensors through the THMS Breakout Box. In addition, the interface module facilitates the display of the real-time input data, provides the real-time data to the Bayesian Belief Network (BBN) Decision Tree for analysis, and receives analyzed and resulting metric data from the BBN for display. The input data is collected at a rate of 4 Hertz from various input TLTAS sensors and is displayed on the Sensor Panel User Screen. Currently, each sensor has two THMS channels assigned. This configuration allows the system to collect data in a differential mode reducing signal noise. The sensor data is stored and supplied to the BBN through System and DLL calls at a rate of approximately 4 HZ. The interface software receives TLTAS analyzed metric output data from the BBN and displays the data. THMS provides multiple user screen variations to facilitate input and output displays, data trending and TLTAS health assessment.

The THMS is currently a 32 channel system. For this proof-of-concept phase of development, the individual grounds for each input sensor were tied together which resulted in effective expansion of the Baseline THMS to collect 64 single-ended vice 32 differential channels of input sensor data. The planned follow-on revision to the THMS will increase the channel capacity to 128 differential channels and 20Gb of hard drive space.

The THMS interface software and BBN “intelligent” codes were developed using a modular approach to provide flexibility for easy and quick modifications to both hardware and software. The codes incorporate a multi-threaded process taking full advantage of today’s multi-treed operational systems.

BBNs are particularly useful, because they can predict component failures using measured sensor data and appropriate reliability analysis algorithms with parameters tailored to a particular system. BBNs provide a natural object-oriented framework for the modeling of most classes of systems. Each node in a BBN can be an object that receives input in the form of either measurements or likelihood ratios and produces output in the form of likelihood ratios. The BBN naturally fuses these data using a generalized form of Bayes Rule. The BBN structure, i.e. the links between nodes, also provides a knowledge fusion that relates the causes of each failure mode to their consequences (measurements). BBNs can also calculate the prior probability of a node failure given the evidence of all other nodes.

The BBN uses system and component performance specifications and normal operating parameters or failure probability distributions to form a relationship between each measurement X and the likelihood for each possible hypothesis (e.g., system normal, system not normal). Although BBNs are normally designed to convert measurements into likelihoods, these “measurements” can take on almost any form, from a Data Concentrator (DC) to the system response function of component failure or even alternate Artificial Intelligence sources. The modular nature of a BBN makes it easy to build upon, starting from the simplest components to the most complex plants and platforms.

The embedded BBN or “intelligent code” takes the conditions of the TLTAS as provided by the various sensors and turns them into predictions of successful operation of the TLTAS based on a physical model and understanding of the system. It is these predictions that are presented to the operator on the THMS User System Health Screen. The predictions are supported by various metrics calculated in the physical model and used by the BBN to indicate the condition of the component of interest. The metrics are presented to the user on the Metric Display Screen which provides a trend indication of system performance. The top-level BBN nodes for the TLTAS are provided below. The sketch shows the connection between the probability of operation of each TLTAS subsystem (Signal Path, Mechanical, Hydraulics, and Electrical) and the probability of operation of the TLTAS as a whole. Each ellipse represents a node of the decision tree. Each sub-system is further modeled and coded in its own complex decision tree.

National Instrument’s LabVIEW software proved to be an excellent option in the tailoring of signal monitoring and analysis programs. We will continue to look at LabVIEW as our first choice in developing future cost effective signal monitoring and analysis applications

Conclusion

The use of National Instruments software and hardware and the embedded BBN “intelligent” code enabled us to meet our customer requirements at low cost while delivering our product early. The demonstrated and delivered capability of the THMS Prototype and supporting Technical Data Package exceeded the customer’s initial expectations based on a relatively small amount of funding applied.

The THMS proof-of-concept demonstrated a greatly enhanced data monitoring, collection, assessment, and prediction capability for this U.S. Navy system. This capability was delivered at a cost far below the cost of other alternatives. The THMS product was customized to meet the requirements of a Government client with specific needs. This specific application can be easily and quickly engineered and tailored to meet the need of a large host of clients in the search for improved system and component reliability, reduced corrective maintenance, increased operating time, and improved profits.

Browse All Case Studies »

  Print