HyDE (Hybrid Diagnostic Engine)
HyDE (Hybrid Diagnostic Engine) is a model-based diagnosis engine that uses candidate generation and consistency checking to diagnose discrete faults in stochastic hybrid systems. HyDE uses hybrid (combined discrete and continuous) models built by application developers and sensor data from the system being diagnosed to deduce the evolution of the state of the system over time, including changes in state indicative of faults.
The key features of HyDE are:
- Model-based: application developers build models which are input to HyDE's suite of reasoning algorithms.
- Diagnose multiple discrete faults, characterized as instantaneous, unintended persistent changes in the operating mode of the physical system.
- Handle hybrid system behavior, using models that can contain both discrete and continuous variables.
- Handle qualitative (Boolean or finite-domain) and quantitative (floating-point or interval) data.
- Stochastic reasoning, including sensor noise, prior probabilities of faults and posterior probabilities of candidate diagnoses.
- Livingstone 2 (L2) Webpage - L2 is the predecessor to HyDE which allowed only discrete models and reasoning.
- ADAPT Webpage - ADAPT is a power system test bed at NASA Ames Research center and is one of the applications of HyDE