Natural Intelligence Systems (NIS) is proposing the research and development of a Fault Management (FM) machine learning (ML) system for use by NASA, government agencies, and commercial companies for spacecraft, transportation, and industrial applications. In this Phase 1 SBIR proposal NIS will develop and demonstrate the feasibility for using its Neuromorphic Machine Learning (NML) system to detect fault-indicative behavior while monitoring multiple inputs of a system. In Phase 1 NIS will develop a major capability of a Fault Management system, which is the ability to monitor and predict the health of a major subsystem using the system's ML model. This is the foundation for a predictive FM system that reports potential failures before they occur. NIS will use its AWS cloud Platform-as-a-Service NML System product as the development system.
The S5.05 Fault Management Technologies topic is meant to drive the development of new FM technologies. By funding this SBIR, NASA will enable the development of a suite of capabilities that result from the computational and mathematical models of this NML system. The 3rd wave properties that result from this combination of the neuromorphic model and its algorithms, the data representation and the hardware architecture enables the system to learn patterns with minimal data during training. The system does not require huge datasets with all possible failure occurrences to be gathered for training. The system is able to continuously perform unsupervised learning while inferring, thereby enabling new patterns and anomalies to be identified and classified as unknowns. The classifications can be explained, and the explanations traced back to the input features and their ranges, thereby allowing unknown classes to be understood and labeled. The system is insensitive to noisy or missing data which is critical for FM systems as sensors degrade or fail in the space environment. These are FM technology weaknesses NASA seeks to overcome.
A Fault Management system based on the Natural Intelligence Systems Neuromorphic Machine Learning system will enable NASA to address several key Fault Management technology issues and limitations. FM systems implemented using the NML system will improve spacecraft resilience against faults and failures, increase spacecraft autonomy with greater onboard fault estimation and response capability, reduce and mitigate interruptions and faults, and decrease labor and time required to develop and test FM models.
Natural Intelligence Systems is pursuing the development of advanced situational awareness and health management systems with 3rd wave AI properties using a hardware accelerated Neuromorphic Machine Learning system on an AWS PaaS product platform for monitoring applications. Many applications exist in the transportation, aviation, manufacturing, machine tooling, and security industries.