Model Based Prognoser

Description

In model-based prognostics, a model of the system is used for both state estimation and prediction. The generality of the framework allows any appropriate observer algorithm and predictor algorithm to be used for a given system.

For more details on this approach to prognostics, see the following paper:

• Model-based Prognostics with Concurrent Damage Progression Processes

For information on selecting a model of the system, visit the Model Library.

For information on Observers and Predictors, visit the Algorithm Library.

Use

In using a model-based prognoser, the user selects the model, the observer, and the predictor to use, along with configuration parameters specific to the chosen model and algorithms.

Configuration

The configuration parameters for the model-based prognoser are described below:

Parameter	Description
type [Required]	The type of prognoser (ex: modelBasedPrognoser).
name [Required]	The name of the component being prognosed (ex: battery1).
id [Required]	A unique identifier for the piece of hardware being prognosed (ex: Serial Number).
inTags [Required]	A list of tags expected from communicators in format local_name : global_name (e.g., inTags: voltage:volts, power:pwr, temperature:temp).
model [Required]	The model to be used by the modelBasedPrognoser.
observer [Required]	The observer to be used by the modelBasedPrognoser.
predictor [Required]	The predictor to be used by the modelBasedPrognoser.
histPath [Optional]	A path for the history files.
intervalDelay [Optional]	Delay (in milliseconds) between iteration of the prognoser loop.
resetHist [Optional]	A flag to reset the recorded history for the component. Will archive the current history file and start a new one.