Fault Diagnosis in Intelligent Greenhouse Control with Decomposed Neural Models |
|---|
| Peter Eredics, Tadeusz P. Dobrowiecki |
- Abstract:
- The novel approach to the intelligent greenhouse control is based on predictive thermal models and actuator action plans to increase control performance. As operating the actuators essentially changes the thermal dynamics, the overall thermal model is decomposed into sub-models based on the state of the actuators. For the intelligent control it is essential to know the actuator configuration to select the most appropriate sub-model, but unfortunately information available about actuator states is not reliable enough to be directly applicable. This paper proposes a method to detect the difference between the believed and the actual state of the greenhouse and to identify the most likely model for the actual state. The method can be used to support the intelligent control in case of actuator malfunction or other accidents and is also suitable to provide possible reasons of the malfunctions to support the greenhouse personnel to take the necessary repair actions.
- Download:
- IMEKO-TC10-2014-011.pdf
- DOI:
- -
- Event details
- IMEKO TC:
- TC10
- Event name:
- TC10 Workshop on Technical Diagnostics 2014
- Title:
13th IMEKO TC10 Workshop "Advanced measurement tools in technical diagnostics for systems' reliability and safety"
- Place:
- Warsaw, POLAND
- Time:
- 26 June 2014 - 27 June 2014