USN

Success Stories

Quantum Leap Reasoning for Condition Based Maintenance

Quantum Leap Innovations has a long history of providing solutions to problems that benefit from integrating optimization and reasoning capabilities. Its Quantum Leap Pattern Based Reasoning technology is well-suited to applications that are highly complex (challenging to define with many variables) and dynamic (subject to changes at regular or irregular intervals). Quantum Leap Reasoning has been employed individually and as part of larger solutions, and can be combined with Quantum Leap Adaptive Optimization® for optimizing problems defined using probabilistic terms.

Quantum Leap Reasoning enables solutions that fuse data from users, sensors, or other applications to determine what is causing the data values being reported. Feedback is expressed in terms of probabilities, not absolute answers, and even if some data was missing, the data that is available can still be used productively.

Quantum Leap Reasoning uses Bayesian Belief Networks (BBNs) for developing and running probabilistic reasoning applications. BBNs are commonly used to establish causal relationships between data and outcomes through probabilistic associations. Because relationships between variables in a BBN are defined probabilistically, feedback is expressed probabilistically. This means that trends can be detected and analyzed over a continuous scale, rather than in terms of absolute answers. By knowing the value at one of the nodes in a Bayesian network, one can infer the value of other nodes in the network.

Advantages of BBNs over rule-based systems include:

•Provide probabilistic output
•Can operate with limited sensor data availability
•More flexible relative to engineering development then traditional expert systems
•Used for both data qualification (state recognition) and anomaly reasoning
•Can operate in a centralized or distributed run-time environment
•More in-depth probabilistic calculations incorporating uncertainty are possible, instead of simple true/false conditions
•Dynamic deployment and adjustment of mathematical models to improve accuracy

Quantum Leap Reasoning also supports creating BBNs, automatically identifying important features from data and connecting them so that constructing them doesn’t have to be a laborious process. Quantum Leap Reasoning provides several methods to determine cause and effect: (1) causal relationships can be constructed manually, (2) they can be constructed automatically given a set of data, or (3) a hybrid approach where a subject matter expert can guide the automatic generation.

Quantum Leap Reasoning is already used in condition-based maintenance applications onboard US Navy ships, in manufacturing plants, and in water treatment facilities. It monitors large scale mechanical systems using Bayesian models relating live sensor data to machine status. This enables real-time feedback of probabilities of potential problems, allowing early diagnosis, preventative maintenance, and mean-time-to-failure projections.

Quantum Leap Reasoning Development Environment

•Graphical user interface for specification of Bayesian Belief Networks (BBNs)
•Import and export in standard BBN file formats
•Interface to several different types of data sources including flat-file, eDNA, JDBC, and XML-RPC sources
•Interact with BBN by setting node values and observing their impact on other parts of the network
•Generate optimal strategies by specifying actionable variables for achieving a given set of goals (consisting of objectives and constraints) according to observed variables
•Tools for automated learning of Conditional Probability Tables (CPTs), nodes, states, and connections for a given BBN from a data source, thereby enabling the creation of more accurate Bayesian Networks that can model larger amounts of data more effectively

Quantum Leap Reasoning for Condition Monitoring

•Runs as an invisible service for automated maintenance, monitoring, and diagnostic applications
•Identifies steady state conditions in the monitored machinery and triggers appropriate actions when they are detected
•Provides logging facilities

Quantum Leap Reasoning for Condition-based Maintenance

Quantum Leap Reasoning has been incorporated into The DEI Group's asset management software solution for installation and evaluation in lab and pilot scenarios. It is now deployed and being used in the area of heavy rotating machinery monitoring using existing sensor deployments and middleware collection software to relate live sensor data to machine health. (This version of Quantum Leap Reasoning has been referred to as AutoNet Express)

Quantum Leap Reasoning, integrated with DEI’s PreMa life cycle management technology, has been deployed with both commercial and government partners. This technology combination allows users to monitor and track changes to the operating conditions of critical components.

Quantum Leap Reasoning has been licensed for several commercial deployments across the United States in a variety of industrial environments and applications.

Technologies & Applications