1452 Duke Street
Alexandria, Virginia 22314
Phone (703) 836-6727
Fax (703) 836-7491
Email: asnehq@navalengineers.org

ASNE Day 2016 - Technical Paper Session 2 : Wednesday, March 2, 2016 1600-1730

Unmanned and Cyber


Author: LT G. Bredariol (USCG), J. Donnal, P. Lindahl, Faculty Advisor S. B. Leeb

Title: Automatic Watchstander Through NILM Monitoring


Widespread modernization in the Navy, Coast Guard, and commercial maritime industries have placed a large emphasis on electronic control and monitoring of systems. Crew sizes continue to decrease as there is a shift to optimally and minimally manned crews that complete more complex and varied mission sets. Crews rely on sensors and automatic operation to perform a host of duties once completed manually. This places a large amount of reliance on monitoring systems to ensure proper operation of equipment and maintain safety at sea. Modern monitoring systems generally rely on individual sensors at a machinery level. Multiple equipment level sensors are appealing as they can easily monitor each piece of machinery, however these systems require a large infras- tructure of sensors, wires, and intermediate panels. This makes them heavy, expensive, and maintenance intensive when at sea conditions including vibrations, water intrusion, temperature gradients, and other environmental factors are introduced. Because they rely on scores of sensors, communications losses and sensor failures can become common place. With fewer technicians available due to optimum manning there may not be enough man hours available to fix these issues, leaving crews to operate without monitoring equipment.

The Non-Intrusive Load Monitor (NILM) provides a low cost, low-maintenance, rugged, easily installed low impact option for machinery monitoring from a centralized location. NILM systems rely on voltage and current measurements taken at a single centrally located point in a power distribution network. They receive aggregate power usage for all equipment on the system being monitored. Through computer processing techniques individual machinery transient events can be divorced from the stream of power readings to infer human activity and machinery plant status as well as equipment health. Due to their central location and simplicity, NILM systems are ideal replacements for optimized crews as they can provide the same monitoring as multiple equipment level sensors.

This paper presents a real-world application and case study of a NILM system in- stalled on a USCG Famous Class (270 ft) Cutter’s Engine Room Auxiliary Machinery panel monitoring eight key pieces of machinery. Propulsion, power generation, fuel transfer, and support equipment were among those monitored. The NILM system used current transducers and voltage sensors on the power supply cables to the panel and receives combined power readings for the entire panel. The NILM system records data at a sampling frequency of 8 Khz and outputs real and reactive power and harmonics. Individual machinery systems are then characterized by power draw, peak loading at start, and frequency of operation. After characterization, future machinery start and stop events can be disaggregated from total power draw allowing for equipment level monitoring through a single sensor.

Discussed are uses for automatic log generation, machinery health monitoring, and human activity interpolation. During the scaled test and monitoring over a month of machinery operation while in an operational setting, all eight pieces of machinery were consistently identified and monitored. Comparing known machinery status from crew generated logs to retrieved data, the NILM was effective at machinery monitoring. The NILM system was used to automatically generate a watchstander log and identify key human activity such as departing for sea, mooring, entering restricted waters, flight quarters, or other special operations. The NILM system also presented the ability to identify abnormal signatures and suggest machinery failure or improper operation. Unusual power readings indicate loss of flow to pumps, or increased load which are important to maintainers. The NILM was able to generate tracking of hours of operation, fuel burn rate, and crew fatigue measurements. Also discussed are future plans to scale the trial to larger and more complex systems. A similar system is planned to be installed on a ship level at the main power distribution system to monitor a whole ship view. Benefits as a more rugged and easily maintained replacement to current monitoring systems to the USCG, USN, and maritime industry as a whole are discussed.

Results from this real-world month long trial prove the NILM system’s ability to effectively disaggregate equipment level data from cumulative readings. Key metrics for maintainers and operators were able to be gathered through a single point. This system poses the potential of greatly decreasing the workload of an optimized crew, allowing for greater monitoring of equipment remotely and increased autonomous operation. Moreover, the NILM allows for easy modularity on a vessel. Equipment can be changed out without having to install new sensors or monitors, improving a vessels ability to perform multiple mission sets.

Return to ASNE Day Program