RTXPS
Reference &
User Manual
    RTXPS  On-line Reference Manual

      Release Level 2.0
      Release Date 2012 06
      Revision Level 1.0
    Document was last modified on:   Tuesday, 14-Aug-12 08:42 CEST
    Models: simulation, optimization

    RTXPS can trigger, as part of its repertoire of ACTIONs, complex simulation and optimization models.

    These models can be run "embedded", as foreground functions, or as batch jobs in the background.

    Communication and coordination between the models and the RTXPS inference engine is through the Dynamic Knowledge Base, that provides model inputs, and manages model outputs, interpreted and aggregated as DESCRIPTORS that the RTXPS RULES can process.

    The basic models types available include:

    • simulation with physically based models (scenario analysis, real-time forecasts);
    • statitsical models (regression, time series analysis, CBR, SNN)
    • optimization (multi-criteria, multi-objective, dynamic, non-linear, distributed).

    The models generate an expected or desired (optimal) state of the system to guide the control options or drive communication (warning, alerts, alarms) where applicable. This makes it possible to use pro-active response based on forecast rather than reactive feedback strategies only.

    Optimization, DSS

    The modular architecture of RTXPS can integrate a wide range of optimization algorithms, includsing "claasical" linear and dynamic programming, DO (discrete optimization), NLP (non-linear programmin), etc but also heuristic methods of CBR (Case Based Reasoning).

    The primary toll is a two-step optimization and DSS approach, that combines

    • a multi-criteria satisfizing approach the can use the full resolution of any non-linear simulation model (driven by a combination of Monte-Carlo, adaptive heuristics, machine learning and genetic programming) to generate feasible and pareto-optimal (non-dominated) alternative solutions efficiently.
    • a discrete multi-criteria DSS componenet (automatic or interactive) based on a reference point model. The automatic version uses the normalized distance from UTOPIA as the metric for ranking and selecting the "efficient" solutions from the non-dominated subset of alternatives.

     


© Copyright 1995-2016 by:   ESS   Environmental Software and Services GmbH AUSTRIA | print page