Download presentation
Presentation is loading. Please wait.
Published byBathsheba Blankenship Modified over 9 years ago
1
INTELLIGENT BUILDING RESEARCH Incorporating Expert Knowledge and Real-Time Data in Smart House/Building Chia Y. Han ECECS Department University of Cincinnati
2
Components are live – each has its own life cycle Ideal design conditions are assumed to be throughout the entire life cycle Interrelationships change with both time and events No monitoring for detection of deterioration Preventive maintenance – often wasteful & harmful Experts in the field are few Digital devices are still proprietary Interoperability across AEC/FM New open systems are coming Introduction Building
4
Work control
5
Application Data Storage Console Devices Client/Server Architecture Old
7
Interoperability
8
IFC_2X
9
BLIS Building Lifecycle Interoperable Software Project
10
BACnet
11
Analog Input Analog Output Analog Value Averaging Object Binary Input Binary Output Binary Value Calendar Command Device Event Enrollment File Object Group Object Life Safety Point Life Safety Zone Loop Object Multi-state Input Object Multi-state Output Object Multi-state Value Object Notification Class Object Program Object Schedule Object Trend Object BACnet Objects
12
BACnet device
13
BACnet Communication Protocol
14
Web Services
15
Web-based open systems
16
Intelligent behavior Automatic model configuration - based on the currently existing components and their respective forms and functions. Operation adaptive to dynamic environment - real-time data search and access - model-based and rule-based control How smart can it be?
17
Console System Modeling PPCL Programs CAD DXF/DWG VISIO EXPRESS BLIS-XML Layout/Layer Reasoner DB KB Service & Building Selection Service request Building Services Operating Maintenance Repair Monitoring Reporting FDD AEC/Vendor/Regulatory Documents Rule Generator IE Monitoring Diagnosis Message Routing Report Generation Work Control Center Energy IBS Architecture
18
Major building systems HVAC Lighting Energy/power Communication Plumbing Fire Safety Security
19
HVAC Systems
23
Background Rule-based expert systems for fault detection and diagnostics (FDD) applications have been implemented Rules can be defined by carefully analyzing the logic of a DDC program in terms of both the normal and abnormal states of the affected subsystem on a given input. Existence of expert knowledge in mature engineering and technology fields.
24
FDD Decision Tree
25
Portability The rules developed in the past are usually created with the assistance of the maintenance personnel, and they often reflect the specifics of a particular HVAC, such as containing hard coded device names. Scalability The rules are dependent on the target system configuration, but there are many differences in system configuration from building to building. Webability The captured domain knowledge not accessible (interoperable or exchangeable) in networked and distributed environment such as the Internet. Motivation for new solution
26
To provide a framework for automatically extracting knowledge and generating expert systems rules for any particular process related to building systems and integrating the expert systems into the current web-based environment. The Main Goal
27
To use the web services technology, such as the Extensible Markup Language (XML) and the Resource Description Framework (RDF) and the industry standards, such as the Industry Foundation Classes (IFC) from the International Alliance for Interoperability (IAI). Proposed Solution
28
FDD XML/RDF File Server Expert System Engine & Knowledge Base Server Database Server FDD App FDD XML/RDF File Server Expert System Engine Knowledge Base Server FDD Client DB Servers Distributed computing
29
Console System Modeling PPCL Programs CAD DXF/DWG VISIO EXPRESS BLIS-XML Layout/Layer Reasoner DB KB Service & Building Selection Service request Building Services Operating Maintenance Repair Monitoring Reporting FDD AEC/Vendor/Regulatory Documents Rule Generator IE Monitoring Diagnosis Message Routing Report Generation Work Control Center Energy Physical system RT data System Modeler FDD Module
30
Desired procedure: 1. The operator requests information about an HVAC from the BLIS- XML/RDF file server. The server responds with two files: (a) BLIS- XML file of a specified HVAC system and (b) RDF file. 2. From the operator’s computer, the BLIS-XML and RDF file are automatically sent for parsing and storing on the expert system engine server. 3. An agent on the expert system engine server uses the two files to parse and generate expert system rules for each HVAC process based on the process model. 4. The operator uses a browser to make a connection to the expert system engine server to specify for which HVAC system FDD should be performed immediately or according to a schedule. 5. The expert system engine server performs FDD immediately or according to a schedule and sends the results to a computer display, pager, SMTP server, etc.
31
The necessary components for implementing the above procedure of generating expert system rule automatically : (1) A generic fault model for each HVAC process; (2) RDF, as a method for associating an HVAC element with its real-time operational value; (3) IFC, as a standard capable of describing an HVAC system; (4) A software agent that will apply the process model to a given HVAC system to generate expert system rules; and (5) An expert system engine to parse the rules and generate inference results in the form of warnings if faults are present.
33
Control loop model
34
VAV System with constant volume return
35
RDF structure ResourcePropertyValue Mixed_Air/Damper1Percent openURL_1 URL 1Locationwww.uc-dbserver/URL_1 Return_Air/FanRotation speedURL_2 Locationwww.uc-dbserver/URL_2 ………
36
XML-IFC description of the sample process model <IfcSensor XMLID="i19" PredefinedType="HVACSENSOR" ConnectedTo="i20" ObjectType="Humidity sensor for measuring the temperature of the air after it has been mixed, warmed or cooled" /> <IfcController XMLID="i20" ControlElementID="5" ConnectedTo="i21" ObjectType="controls the humidifier valve actuator" />
37
Implementation The system is implemented with the Java-based open source solution. The expert system component consists of Jess (Java Expert System Shell), which is in its core a collection of Java classes Jess provides easy to use mechanisms for creating Java objects, accessing their variables and calling their methods. In addition, it is possible to start the Jess engine from Java code, which gives the programmer the ability to redirect the output of the engine to any valid Java output stream (e.g., TCP/IP socket).
38
Jess also takes advantage of the Java Bean™ technology, a form of inter-process communication, which enables a Java application to implement the event driven model. A Java Bean is a Java object that generates (or “fires” in Java terminology) predetermined events. If a Java Bean has one or more “listening” objects attached to it, they will be able to “hear” the events and take appropriate actions. All “listeners” must first register with the Java Bean to receive fired events. Jess can easily register to become a “listener” with such a class and update the changed information in the appropriate slot of its template (a named entity that contains a list of facts). As soon as Jess modifies a fact on its fact list, it re-runs all the rules in its knowledge base.
39
The FDD rules from the prior expert system shell, M4, were converted into the Jess language and a Java Bean based template was created to retrieve parameters relevant to the process from the database. The connection to the database was established with the help of the JDBC-ODBC Bridge, which provides the necessary API for Java programs to make connection to ODBC sources on Windows based computers and use SQL for querying the tables in question.
40
A simple message exchange protocol of a client-server model was implemented: - A TCP/IP server is launched and listens on a port for applet connections. - As soon as one connection is established, it spawns an object of the class “processor” passing the obtained communication socket as the parameter. - The processor immediately enters a new thread waiting for commands from the applet. - The server meanwhile is able to accept new connections creating a processor object for each case.
41
GUI The last step is to create a Java applet to allow the Internet user to view the output of the FDD. After the server receives the request from the incoming connections from the applet, the output of the expert system engine would be re-routed to the client side of the connection into a text area located in the applet.
43
A real-world application of HVAC FDD system was used. The main steps and the major components needed for design and deployment of web-based intelligent systems are highlighted. For an application in other domains, expert knowledge needs to be extracted, modeled, and represented in knowledge base. The knowledge of experienced domain workers is critical. Overall, we have considered and demonstrated successfully that practical configuration mapping, expert system rule generation and real-time data access can be accomplished with the off-the-shelf web tools and today’s component technologies. Conclusion
44
There is a need for a practical solution to implement intelligent behavior of systems that are networked and digitally controlled. It is important to use expert knowledge and salvage the knowledge in the existing expert systems and readapting it into the distributed computing model of the present time. Expert system technology is still used to let the computer “reason” about what it “sees” based on a generic model for each of the modeled processes and real-time data. Application software development for upcoming building open systems is coming of age.
45
Thank You!
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.