Auxiliary System Sensor Fusion
Navy SBIR 2011.2 - Topic N112-159
ONR - Mrs. Tracy Frost - email@example.com
Opens: May 26, 2011 - Closes: June 29, 2011
N112-159 TITLE: Auxiliary System Sensor Fusion
TECHNOLOGY AREAS: Information Systems, Sensors, Electronics
ACQUISITION PROGRAM: Integrated Condition Assessment System, ACAT IV
OBJECTIVE: Develop methods and algorithms that allow sensor information from disparate auxiliary systems to be intelligently fused to provide enhanced situational awareness.
DESCRIPTION: Currently, auxiliary systems have sensors which are used to understand the state and control the systems of which they are a part. For example, fluid systems have pressure sensors which are used to measure the flow of fluid inside the pipes. It has been shown that these sensors can also provide data to intelligent algorithms to detect and isolate pipe ruptures during damage events. Such an advance is an example of using multiple sensors to gain increased situational awareness and prescribe an appropriate control action. The next logical step is to fuse data from sensors across systems in an effort to gain an enhanced situation awareness of the ship as a whole. Such information can be used to prescribe control actions at the higher levels of hierarchical ship control system, or to perturb resident shipboard models to predict possible future states of the ship. The holistic knowledge gleaned from this sensor fusion can also be useful to lower levels of the control system. For example, a rupture detected in a piping system may be located more precisely if a higher level system passes along the information that it is likely that there is damage in a particular compartment.
The focus of this topic is use sensors from a notional chilled water system, coupled with a notional electrical system to determine the states of the two systems in a manner that is more accurate than can be accomplished using only each system's sensors individually. The notional system that will be used for this project currently exists in both a software simulation and a reduced scale hardware implementation. The simulation will be provided to the investigators of this topic. Using a notional system eliminates issues surrounding releasing current ship designs to personnel outside the Navy. The notional system has been designed such that it resembles an actual ship system closely enough that results on the simulation and reduced scale system will be applicable to actual navy systems. In addition, using the notional system allows this topic to be focused on a specific domain, with a specific design in mind. This will allow the investigators to be focused on a specific instance of the sensor fusion problem that is of interest to the navy, and helps to clarify and focus the problem of interest in what is otherwise a wide open area of investigation. The sensors to be used are also defined as a part of this system, so that the data that the fusion algorithms will use is already defined. Documentation and interface specifications to both the simulation and the hardware implementation of the notional system are available and will be provided to the investigators.
Algorithms of interest include, but are not limited to, Bayesian belief networks, linear and nonlinear classifiers, Kalman filtering, and Dempster-Schafer. These algorithms have not been applied to coupled, distributed shipboard systems in the past. Therefore, this work represents an advance in the state of the art. However, these algorithms have been applied in similar domains, so it seems reasonable that this approach is feasible.
PHASE I: Develop an approach to the fusion of the sensors of multiple shipboard auxiliary systems and investigate fusion algorithms that may apply to this problem.
PHASE II: Evaluate phase one approach and algorithms, develop simulation and perform simulation testing. Refine the algorithms, determine limitations and investigate issues such as algorithm tuning.
PHASE III: Demonstrate the sensor fusion approach in a reduced scale, hardware in the loop model.
PRIVATE SECTOR COMMERCIAL POTENTIAL/DUAL-USE APPLICATIONS: The ability to fuse and reason over sensor data from multiple, interdependent machinery systems for aggregate control and performance increase, while increasing efficiency, capability and reducing cost will is applicable to industries that operate in a like environment. Industries include commercial shipping, manufacturing (process control), utility providers, including information technology.
2. F. Jensen, "An Introduction to Bayesian Networks, UCL Press, London (1996)
3. J Kolodner, "Case Based Reasoning" San Francisco, Ca, Morgan Kaufmann (1993)
KEYWORDS: Machinery; Control; Sensors, Fusion, Reasoning, Efficiency