Specifying the behaviour of building automation systems
3 likes799 views
The document proposes a declarative context description language to specify the behavior of building automation systems. It aims to address the problems with current approaches that are difficult to configure and lack expressivity. The language uses fuzzy logic and temporal logic to define contexts based on sensor values over time. Scenarios are defined as rules with fuzzy conditions and assignments to vary actuator outputs. The proposal is validated through simulations of example scenarios to test functionality. Future work includes improving user personalization, implementing real-world simulations, and extending the language.
Specifying the behaviour of building automation systems
- 1. Specifying the Behaviour of Building Automation Systems João Aguiam September 6th, 2011
- 2. Agenda • Motivation • Problem • Our Proposal • Validation • Conclusions 2
- 3. Agenda • Motivation • Problem • Our Proposal • Validation • Conclusions 3
- 4. Motivation The smart office example Work Desk Door Curtains Window Support Table Luminosity Sensor Luminaries Motion Sensor 4
- 5. Motivation The smart office example • Maximize energy efficiency • Control the curtains and luminaries automatically to: • Reduce the amount of artificial light • Allow the maximum amount of natural light inside • Avoid the excessive glare in the working place of the user • Turn lights off if no one is in the office • Adjust light level according to user’s task and place 5
- 6. Agenda • Motivation • Problem • Our Proposal • Validation • Conclusions 6
- 7. Problem How it is done nowadays? • Crafting behaviour into hardware devices (embedded C programs) • Difficult to develop • Lengthy process • Too far away from the problem domain of BA • Hardware modules with pre-packaged applications • Difficult to configure • Sometimes it is impossible to create the desired behaviour due to limited expressivity
- 8. Problem Statement • How to specify, in an easy way, the logic behind the behaviour of a Building Automation System where the actuators output vary continuously based on sensors input? 8
- 9. Problem Fuzzy controller • Difficult to specify • Require great expertise • Far from the domain level 9
- 10. Agenda • Motivation • Problem • Our Proposal • Validation • Conclusions 10
- 11. Our Proposal Logical level Execution level 11
- 12. Our Proposal Relation with fuzzy controllers Rule-base Fuzzification Inference Defuzzification Logical level Execution level 12
- 13. Our Proposal Context description language • Declarative language based on fuzzy logic and temporal logic • The degree of truth of a context is defined recursively on the value of sensors and other contexts • Some operators reason over a path of past values of sensor readings and contexts 13
- 14. Our Proposal Scenario description • Scenarios are a set of rules which take the form: if antecedent then consequent • The antecedent is is a fuzzy condition defined using fuzzy contexts and fuzzy context operators • The consequent is an assignment to an actuator which uses fuzzy consequent operator • The assignment to an actuator varies with the truth value of the fuzzy condition • The defuzification process is a simple weighted average 14
- 15. Agenda • Motivation • Problem • Our Proposal • Validation • Conclusions 15
- 16. Validation Ideal setting IDE IDE Behaviour Specification Compiler HW User BA Expert IDE 16
- 17. Validation Strategy 1. Illustrative Case Study 2. Excel Model 3. Interactive Simulator 4. Batch Simulator 17
- 18. Validation Results and Discussion Curtains along the day Arriving and leaving Passage of a cloud in the sky Working at desk Working at table Walking around 18
- 19. Agenda • Motivation • Problem • Our Proposal • Validation • Conclusions 19
- 20. Conclusions • We have created an high level declarative language to specify the behaviour of building automation systems • Adapted the notion of fuzzy context from Ambient Intelligence • Created a well defined syntax and semantics that can be compiled and interpreted • Introduced the notion of time from the temporal logic with a new fuzzy Until operator • Adapted the deffuzification process to a less computational one • We have validated the idea through diverse simulators to test the functionality in different scenarios 20
- 21. Conclusions Future work • Learning user preferences • Implementing real world simulation • Create an interpreted development environment • Extend the language expressivity 21
- 22. Thank You João Aguiam September 6th, 2011
- 23. Reserve Slides
- 24. Context-aware Systems • Adapts to the surrounding environment • Environment is abstracted through the notion of context • CAS gathers information related to context and reasons about it to take a certain action • Context is any information that characterizes the situation and is relevant for the interaction between the user and the system 24
- 25. Fuzzy Logic • Propositions may be partially true or partially false and can be seen as fuzzy sets • Fuzzy Sets are represented by a Membership Function (MF) • In Fuzzy Logic the operation conjunction and disjunction are calculated through the operator min and max respectively 1 ColdPlace HotPlace 0.8 Membership Degree 0.6 MildPlace 0.4 0.2 0 0 5 10 15 20 25 30 35 X = Temperature 25
- 26. Fuzzy Logic • Fuzzy logic systems are based on fuzzy rules • Fuzzy control systems must result in a single crisp value to control an object 26
- 27. Temporal Logic • Propositions are qualified in terms of time • Allows the reasoning in a sequence of events along the time • Most common operators: • Next • Eventually • Always • Until • Releases 27
- 28. Expert Systems • Computer program that simulates the reasoning of a human expert to solve problems or give advices to the user • Four topics: • Knowledge acquisition • Knowledge representation • Reasoning control • Solution explanation 28
- 29. Fuzzy Context Operator • Defined in the universe of discourse of a context • Allow a customized evaluation of a context • The input of a fuzzy context operator is a context C and its output a value between 0 and 1 • Defined by a membership function • Examples are: No, Little, Some, Enough, Much, Too Much and Full. 29
- 31. Fuzzy Consequent Operator • Defined in the universe of discourse of an actuator • It is specific for a certain type of actuator • Used in the assignments 31
- 33. Defuzzification 33
- 34. Mandani Fuzzy Inference System 34
- 35. Sugeno Fuzzy Model 35
- 36. Tsukamoto Fuzzy Model 36
- 37. Curtains Along the Day Closed Opened Hours 37
- 38. Arriving and Leaving On Off Time 38
- 39. Passage of a Cloud in the Sky Time 39
- 40. Working at Desk Time 40
- 41. Working at Table Time 41
- 42. Walking Around Time 42
- 43. Limitations • Single user model • Absolute values in the actuator assignments • Set actuators to past values • Reasoning about time intervals • Lack of customization for different users • Validation • Single case study • Scalability • The lack of validation in a real environment 43