Artificial Intelligence Approach to Energy Management and Control in the HVAC Process An Evaluation Development and Discussion.код для вставкиСкачать
Artificial Intelligence Approach to Energy Management and Control in the HVAC Process: An Evaluation, Development and Discussion Hong Zhou, Ming Rao* and Karl T. Chuang Department of Chemical Engineering, University of Alberta, Edmonton, Alberta, Canada, T6G 2G6 Computer control technology in heating, ventilating and air conditioning (HVAC) systems has evolved in two technical phases: central control and direct digital control (DDC) with central monitoring function. Many control strategies have been developedfor energy conservation in HVAC energy management and control systems (EMCS) and these are reported in the literature. An artificial intelligence approach to the EMCS in HVAC process is proposed in this paper, Two real time expert systems for energy-saving operation and indoor setting are developed and integrated with an adaptive control strategy. Further, a configuration of an integrated intelligent system for intelligent building management is proposed, in which building management functions are coordinated by a meta-system. The applications of artificial intelligence (AI) techniques in a HVAC process give a new avenue for the development of EMCS in the HVAC industry. Introduction Computer control technology has been applied to heating, ventilating, and air conditioning (HVAC) for more than twenty years. During this period, the technology has evolved in two phases: central control and distributed direct digital control (DDC) with central monitoring. In the 1970s, centralised control systems using minicomputers were implemented in the HVAC process. This brought great control flexibility and energy saving opportunities to the HVAC industry, but suffered from poor reliability and the high cost of the computer system. In the early 1980s, a growth in computer applications took place in all HVAC areas because of the introduction of the microcomputer. Direct digital controllers were then distributed throughout the buildings. The functions included at this level could be either occupant comfort control or remote unit control and monitoring. Energy management and control (EMC) are performed by a central microcomputer. With a distributed controller, system reliability was improved and * To whom correspondences should be addressed. Developments in Chemical Engineering, Vol. 1, No. 1, Page 42 Artificial Intelligence Approach 43 computer-based energy management and control systems (EMCS) have recently been accepted as standard control features in large commercial buildings. What advantages could DDC with a central monitoring system achieve by using advanced computer technology?* What is the third technical phase for HVAC control systems? In this paper, we will first briefly review the current energy conservation control strategies, and address the problems encountered within the current EMCS of a HVAC system. Second, an artificial intelligence method for energy management and conservation control is presented, and its applications are outlined. Furthermore, an integrated intelligent system is introduced into intelligent building technology. Finally, the third technical phase in HVAC control systems is discussed. Review of HVAC Energy Conservation Control Strategies The EMCS technology has been utilised in the HVAC industry for more than ten years. Approximately 70% of all energy management systems are used only for the single function of HVAC energy management control. The HVAC energy control strategies may be classified as follows: Time scheduling This is the most effective energy-saving control strategy in the early use of EMCS in HVAC. It turns equipment ON or OFF at preset times. Normally, this strategy turns equipment off when it is not needed or it gives a better match of the loads. For example, scheduling HVAC equipment to run during occupied hours; 'dayhight setback' which sets low temperature when heating is used to maintain temperature, and sets high temperature when cooling is used to maintain temperature at night.2 Seven different equations are compared by Seem el al? to determine the correct time to return from night or weekend setback. It was observed that the return time was linearly related to the outdoor temperature when the initial room temperature was at the setpoint during night or weekend setback. Linear or quadratic equations with a weighting function provided the best forecast of return time. Time scheduling also allows HVAC equipment to be off-duty to avoid electrical peak time. Advanced HVAC process control Optimisation, adaptive control and predictive control have been applied to HVAC p r o c e ~ s e s . ~Since - ~ chiller plants consume a very large amount of energy in HVAC systems, chiller plant optimisation has been the focus of many research studies. Some important optimal control strategies have been proposed for the chiller.g-" B r a ~ n ' ~ investigated ''~ the methodologies for optimal control of chilled water systems without storage, he proposed a component-based nonlinear optimisation algorithm and simpler system-based near-optimal control. Adaptive control techniques are attractive i n HVAC applications because process characteristics change widely due to variations i n weather and building occupancy. It has been demonstrated by Dexter and HavesI6 that adaptive control can cope with many problems encountered in HVAC applications, and requires far less commissioning efforts than conventional controllers. The control performance is superior to that of a manually tuned PID controller. The long term behaviour and energy savings achieved by using adaptive control in HVAC processes are still I49l5 44 Hong Zhou, Ming Rao and Karl T. Chuang under investigation.16 By predicting outdoor and indoor temperatures, as well as building thermal resistance and capacity, we can prevent the loss of energy in preheat or precool cycles, and determine the best recovery time from night ~ e t b a c k ,as~ well as the optimal start/stop sequence. In addition, if an indoor temperature swin is allowed, a large saving can be achieved by the predictive control strategy. 1.B Operations planning There are several alternative air handling processes which achieve an indoor setpoint in a HVAC system. The option of energy conservation operations is based on the following methods:18 (a) Economy-cycle outside-air control: This strategy is to use outside air for cooling whenever possible, thus minimising energy use in the refrigeration cycle. (b) Enthalpy control: This strategy is achieved by calculating the enthalpy of air being processed and choosing the minimum enthalpy cost for air handling operation. The calculation of the enthalpy is not an easy task. Alternatively, we can compare the enthalpy cost of air processing operations on the psychrometric chart in order to select the minimum-enthalpy cost operation. (c) Conversion of dual-duct to variable-air-volume (VAV): Replacing a dual-duct system by a VAV system is a very effective energy conservation technique. Mixing loss and fan horse power savings can be made by adding fan volume control.19.20 (d) Energy-source shutdown: Depending on the season, shutting down heating or cooling sources also provides energy conservation. However, the shutdown should not cause the loss of environmental control. Energy conservation operation planning of the HVAC process mainly relies on the weather, room set, HVAC system structure, and so on. It often confuses operators. Since there are so many alternative HVAC constructions, a global solution may be very difficult to obtain. Comfort technology Comfort research has provided a number of alternatives for energy We believe that 'comfort' and 'energy' could be simultaneously optimised through the use of operation strategies which consider the dynamics of comfort and control system. "Dynamic control" is a recently developed energy conservation control strategy which utilises comfort technology, having been original1 conceived back in the mid-1970s. Hartman23 and Colburn and Harman2' investigated this advanced control strategy and pointed out that seven key variables had a significant effect on human comfort. These were dry bulb temperature, relative humidity, mean radiant temperature, air velocity, clothing, activity (metabolic rate) and exposure time. Compared to other control strategies, dynamic control allows user interaction and uses all seven comfort variables. It combines three concepts: system-free cooling, interior temperature drifting and thermal flywheeling. This dynamic-control strategy introduced two new control concepts into HVAC control. First, it integrates all controlled HVAC components into a coordinated, but continually changing control strategy. Second, it continuously anticipates upcoming weather and occu ancy conditions in order to develop the control law. A case study by Spratt et al.% illustrated that dynamic control offered less space-temperature control than normal control, and the comfort conditions Artificial Intelligence Approach 45 were improved. Thermal comfort dynamic control methodology has been implemented into a commercial product called "Touchstat". The test results conducted on two residences showed a 55% electricity saving, equivalent to $2.58 m-2 ~ r - ' . ~ ~ Comfort technology has not yet been fully utilised for energy conservation. Many users and designers are still unaware of the opportunities available for energy conservation. Comfort technolo!; education has been suggested for transferring this knowledge to the public. New energy eJjcient equipment Heat-pump air conditioners, solar energy heaters, and inverter control systems are the more recently developed energy-efficient HVAC eq~iprnent.~'Most of these items use microprocessor-based control and significantly reduce the energy consumption.28 They have often been used in practical applications. Heat pumps are used in residential air-conditioning systems with advanced control, i.e. inverter control system, defrost control, setback thermostats, variable-speed fan control, etc., and result in a 20-40% energy ~ a v i n g . ~ ~ - ~ ' From the system science viewpoint, the current energy management and control strategies used in HVAC processes are a collection of remedies or recipes, containing neither coordination nor integration. The reduction i n energy consumption was easy to achieve in the early stage with only a small capital investment. But these simple remedies have now been exhausted. It is very difficult to model the HVAC process. Most problems i n the EMCS are ill-structured, and therefore difficult to handle by a mathematical approach.32733 Artificial Intelligence Approach to the EMCS Recently, attention has been focused on A1 applications in HVAC processes.3341 Many ill-structured engineering problems that deal with non-numeric information and non-algorithmic procedures are suitable for applications of artificial intelligence (AI) technique^.^^ The A1 approach provides a programming methodology for solving ill-structured problems and allows use of heuristics. Expert systems is the main field of artificial intelligence research. An expert system is a specific computer program that acquires and codes the knowledge from the human experts in order to solve ill-structured problems. An expert system usually consists of three components: a data-base to hold the data for the system; a knowledge-base to contain the knowledge about the area of the problem; and an inference-engine to determine which rules are relevant to given data. The structure of an expert system is shown in Figure I . Knowledge is mainly collected from the available literature and is confirmed by human experts. A typical rule is presented as: "IF (condition), THEN (action)." As most of the EMCS problems in an HVAC process are ill-structured, an expert system approach provides a suitable methodology. An expert system approach to the EMCS of a HVAC process is proposed here, it investigates real-time expert systems for energy conservation control in the HVAC process. It aims at improving energy efficiency in new or existing residential houses and commercial buildings. Specifically, it is required to apply A1 techniques to the EMCS for the economic operation of HVAC. The primary objectives of the proposed approach are summarised as follows: 46 Hong Zhou, Ming Rao and Karl i? Chuang --r' KNOWLEDGE Figure 1 Structure of an expert system. (1) to develop a real-time decision making system for comfortable and energy efficient settings of temperature and humidity in a HVAC process; (2) to develop a real-time expert system for energy-saving operation planning in various HVAC systems; (3) to design and implement the real-time energy conservation expert control systems for the HVAC process in residential houses and commercial buildings; (4) to integrate the above systems in an intelligent environment; ( 5 ) to investigate new energy management and control strategies for the HVAC process. A basic framework for the intelligent control system is hierarchically constructed to provide indoor settings, air-processing operation planning, and adaptive control. (1) Development of a real-time decision-making system for indoor temperature and humidity settings. The indoor settings play an important role in both thermal comfort and energy consumption, however, they have not been fully utilised to reduce energy consumption and improve comfort. It is recognised that a range of temperature settings exist for which building occupants feel comfortable. We are currently developing an expert system to provide real-time dccision-making for temperature and humidity setting. The knowledge includes thermal comfort technology and energy management, as well as "dynamic control" strategies. This expert system can also provide off-line decision support to the operators. (2) Development of a real-time expert system for energy-saving operation planning in various HVAC systems. A year-round HVAC system is usually made up of five or six items of control equipment, and up to six air-handling operations: mixing, heating, cooling, humidifying, dehumidifying, and air-washing. In addition, various HVAC systems such as multi-zone, single zone, variable air volume, constant air volume, and different combinations of the above systems are being used. The identification of an optimal energy-saving operation is very complicated. An expert system for Artificial Intelligence Approach 47 energy-saving operation planning in various HVAC systems has been developed in an expert system developing shell, namely PC-plus. The expert system consists of two levels of frames and external numerical computation. It can be used in either real-time decision making or for off-line operation assistance. (3) Implementation of an expert system for energy conservation control in a HVAC process. The decision-making system for indoor setting and the expert system for energy-saving operation planning, as well as adaptive control technique, should be integrated to provide this advanced control technology to the HVAC industry. For a particular HVAC system, the energy-saving operation scheduling should be provided on-line by an expert system based on the weather conditions, the indoor setting, and the supplied air. An adaptive control technique will be used to set suitable requirements for supplied air, and to control air handling equipment. Because of the large differences between commercial buildings and residential houses, the implementation will be carried out separately. This new systematic approach provides an opportunity for functional integration and coordination with the other systems in intelligent building. It will also provide innovative HVAC process control incorporating energy efficient operation and better comfort in buildings. Integrated Intelligent System for Building Management Modern building technology is moving towards "Intelligent Building".4349 An "Intelligent Building" is one that provides a productive and cost-effective environment through optimisation of its four basic elements: structure, s stem, services, and management, as well as the interrelationships among them." The building control system is the brain of intelligent building. Essentially, integration is the key to intelligent building technology. The hierarchical structure for the integration of a building management system is shown in Figure 2. This plan divides system functions into two categories, as either 'integrated automation' or 'integrated communication'. The former contains functions such as fire detection, security, lighting, HVAC, and environment control; while the latter holds the foreground office functions of telephone, telex, electronic mail and data communication. To accomplish such a complicated integration, it is necessary to apply distributed intelligence techniques - an important field among artificial intelligence techniques. To facilitate integration for distributed intelligence systems, a "meta-system" concept was suggested by Rao et al. .42,5' A meta-system performs the knowledge integration and management, and provides the following integration functions: (1) integration of knowledge from different disciplinary domains; (2) integration of empirical expertise and analytical knowledge; (3) integration of various objectives, such as research and development, engineering design and implementation, process operation and control; (4) integration of different symbolic processing systems (expert systems); ( 5 ) integration of the symbolic processing system and numerical computation systems; (6) integration of different information, such as symbolic, numerical and graphic information. The meta-system has six important functions in the integrated intelligent system: Hong Zhou, Ming Rao and Karl T. Chuang 48 Integration of building management function Metesystem I Integrated automation 1 ated communication I I I I subsystems I I CCTV Figure 2 Hierarchical structure of the integration of building management. (1) It is the coordinator used to manage all symbolic reasoning systems and numerical computation routines in an integrated intelligent system. (2) It distributes knowledge into separate expert systems and numerical routines so that the integrated intelligent system can be managed effectively. (3) It is the integrator which can easily acquire new knowledge. (4) It can provide a near optimal solution for conflicting solutions and facts among different expert systems. ( 5 ) It provides the possibility of parallel processing in the integrated intelligent system. (6) It can communicate with the measuring devices and the final control elements in the control systems, and transform various non-standard inputloutput signals into the standard communication signals. The meta-system has been implemented in the C environment, namely Meta-C. In Meta-C, the symbolic process follows the progress of the numerical process by receiving posted information from the numerical procedure at several steps during calculation. The symbolic process then has the option of allowing the numerical procedure to continue, or changing some parameters, or aborting the procedure entirely. This process competes favourably with a shallow-coupled process where the heuristic process invokes a numerical routine via a procedure call, supplies the necessary input information, and passively waits for the numerical process to terminate, and then provides the required output. The configuration of the integrated intelligent system has attracted significant attention from both industry and academia, and is expected to lead to a new era in the application of A1 techniques to engineering problems.42 Integration is a key concept in intelligent building technology. The integration of building management functions gives a new avenue for HVAC control systems. Many HVAC professionals throughout the are enthusiastically adopting the concept of intelligent buildings, rather than deprecating it. The application of A1 techniques in HVAC energy management and control systems may give rise to a third technical phase in the HVAC industry. Artificial Intelligence Approach 49 Conclusion C u r r e n t energy conservation control strategies are reviewed in this paper, illustrating a lack of systematic organisation. An artificial intelligence approach is proposed for energy conservation control in HVAC processes. W o real-time expert systems are being developed; one chooses energy saving operations, and the other sets indoor parameters by using comfort technology. Furthermore, an integrated intelligent system is introduced into intelligent building, in which b u i l d i n g m a n a g e m e n t f u n c t i o n s are c o o r d i n a t e d by t h e meta-system. The applications of artificial intelligence techniques provide the new avenue for EMCS development in the HVAC industry. Acknowledgement This project is financially supported by the Natural Sciences and Engineering Research Council of Canada and Hardy Fund. References Payne, P.P. 1988. What distributed controllers bring to the building management system. ASHRAE Trans., 94(1), 1503-1513. Rinehart, R.D. 1981. Energy savings with large set-backs. Energy Engineering. 78(6). 21-38. Seem, J.E., Armstrong, P.R. and Hancock, C.E. 1989. Algorithms for predicting recovery time from night setback. ASHRAE Trans., 95(2), 439-446. Mehta, D.P. 1983. Effects of control dynamics on energy consumption in residential heating systems. Energy Engineering, 80(3), 7-24. Nelser, C.G. 1986. Adaptive control of thermal processes in buildings. IEEE Control System Magazine, 6(4), 9-1 3. Athienitis, A.K. 1988. A predictive control algorithm for massive buildings. ASHRAE Trans., 94(2), 1050-1067. MacArthur, J.W., Grald, E.W. and Konar, A.F. 1989. An effective approach for dynamically compensated adaptive control. ASHRAE Trans., 95(2), 415-423. Maxwell, G.M., Shapiro, H.N. and Westra, D.G. 1989. Dynamics and control of a chilled water coil. ASHRAE Trans., 95(1), 1243-1255. Hachner, R.J., Mitchell, J.W. and Beckman, W.A. 1984. HVAC System and energy use in existing buildings - Part I . ASHRAE Trans., 90(2B), 523-535. 10 Johnson, G.A. 1985. Optimization techniques for a centrifugal chiller plant using a programmable controller. ASHRAE Trans., 91( 1B), 835-847. 1 I Cascia. M.A. 1988. Optimizing chiller plant energy savings using adaptive DDC algorithms. ASHRAE Trans., 94(2), 1937-1946. 12 Braun, J.E., Klein, S.A., Beckman, W.A. and Mitchell, J.W. 1989. Methodologies for optimal control of chilled water systems without storage. ASHRAE Trans.. 95(1), 652-662. 13 Braun, J.E., Klein, S.A., Beckman, W.A. and Mitchell, J.W. 1989. Applications of Optimal control to chilled water system without storage. ASHRAE Trans., 95( I), 663-675. 14 Wepfer, W.J. and Li, X. 1986. Implementation of adaptive control to building HVAC systems. Proc. 18th World Energy Engineering Congress, pp. 143-1 50. 15 Underwood, D.M. 1989. Response of self-tuning single-loop digital controllers to a computer-simulated heating coil. ASHRAE Trans., 95(2), 424-429. 16 Dexter, A.L. and Haves, P. 1989. A robust self-tuning predictive controller for HVAC applications. ASHRAE Trans., 95(2), 431-438. 50 Hong Zhou, Ming Rao and Karl T.Chuang 17 Shapiro, M.M., Yager, A.J. and Ngan, T.H. 1988. Test hut validation of a microcomputer predictive HVAC control. ASHRAE Trans., 94( l), 644-659. 18 Haines, R.W., 1984. Retrofitting reheat-type HVAC system for energy conservation. ASHRAE Trans., 90(2B). 185-191. 19 Johnson, G.A. 1984. Retrofit of a constant volume air system for variable speed fan control. ASHRAE Trans., 90(2B), 201-21 1. 20 Teji, D.S. 1987. Controlling air supply for energy conservation. Energy Engineering, 84(3), 4-14. 21 Nelson, T.M., Nilsson, T.H. and Hopkin, G.W. 1987. Thermal comfort: advantages and deviations. ASHRAE Trans.. 93(1), 1039-1047. 22 Doherty, T.J. and Arens, E. 1988. Evaluation of the physiological bases of thermal comfort models. ASHRAE Trans., 94(1), 1371-1378. 23 Hartman, T.B. 1988. Dynamic control: fundamentals and considerations.ASHRAE Trans., 94( l), 599-609. 24 Colburn, B.K. and Harmon, K.S., Jr., 1987. Thermal comfort dynamics - an advanced control strategy. Energy Engineering, 84(4), 30-52. 25 Spratt, D.K., Sadler, G.W. and Moodie, K.R. 1989. Dynamic control - a case study. ASHRAE Trans., 95(2), 193-200. 26 Hayter, R.B. 1987. Comfort education for energy conservation. ASHRAE Trans., 93(1), 1080-1083. 27 Nahar, N.M. and Gupta, J.P. 1989. Energy conservation and pay back periods of collector-cum-storagetype solar water-heaters. Applied Energy, 34(2), 155-163. 28 Cooper, K.W. 1983. Microcomputers in an energy saving residential heat pump. IEEE Trans. on Industry Applications, IA-19(4), 486-490. 29 Imaiida, T., Kojima, S., Aoi, F. and Ohta, M. 1985. Development of Demand Defrost control system for residential heat pumps. ASHRAE Trans., 91(2B), 1527-1535. 30 Shimma, Y., Tateuchi, T. and Sugiura, H. 1985. Inverter control systems in the residential heat pump air conditioner. ASHRAE Trans., 91(2), 1541-1553. 31 Colliver, D.G., White, G.M. and Turner, L.W. 1987. Electrical energy consumption and demand reduction using heat pump controls. ASHRAE Trans., 93( l), 578-587. 32 Jedlicka, A.D. 1985. Improving conservation behavior. Energy Engineering, 82(5), 29-34. 33 Hartman, T.B. 1989. An operator’s control language to improve EMS successes. Energy Engineering, 86(2), 6-1 1. 34 Haberl, J.S. and Claridge, D.E. 1987. An expert system for building energy consumption analysis: prototype results. ASHRAE Trans., 93( I), 979-987. 35 Camejo, P.J. and Hittle, D.C. 1989. An expert system for the design of heating, ventilation, and air-conditioningsystems. ASHRAE Trans., 95(l), 379-386. 36 Brothers, P.W. 1988. Knowledge engineering for HVAC expert systems. ASHRAE Trans., 94(1), 1063-1073. 37 Haberl, J.S., Smith, L.K., Cooney, K.P. and Stern, F.D. 1988. An expert system for building energy consumption analysis: applications at a university campus. ASHRAE Trans., 94(1), 1037-1051. 38 Liu, S.T. and Kelly, G.E. 1988. Knowledge-based front-end input generating program for building system simulation.ASHRAE Trans., 94(I), 1074-1084. 39 Bagby, D.G. and Cormier, R.A. 1989, A heat exchanger expert system. ASHRAE Trans., 95(2), 927-933. 40 Bergt, C. 1989, Creating a PROLOG Knowledge-based system using an existing air-to-refrigerantcomputer model. ASHRAE Trans., 95( l), 1237-1 244. 41 Tuluca, A.N., Turner, G.E., Krarti, M. and Antipa, V. 1989. Expert system for selection of standardized research greenhouses. ASHRAE Trans., 95(2), 938-953. 42 Rao, M., Jiang, T.S. and Tsai, J.P. 1989. Combining symbolic and numerical processing for real time intelligent control. Engineering Applicution of AI, 2 (March), 19-27. 43 Cower, C.J. 1988. Towards germane analysis and reporting by Building Management Systems. Measurement and Control, 21 (September), 213-214. Artificial Intelligence Approach 51 44 Martin, A. 1988. Integration of building management functions. Measurement and Control, 21 (September), 21 1-212. 45 McKinley, A.D. 1988. Modern building services - more than computerized HVAC. ASHRAE Trans., 94( l), 934-942. 46 Petze, J.D. and Reed, D.R. 1988. Artificial intelligence in building control systems. ASHRAE Trans., 94( l), 96C969. 47 Rouse, K. 1988. BEMS energy management systems - an overview. Measurement and Control, 21 (September), 1988. 48 Williams, V.A. 1988. Intelligent Buildings - smarter with DDC. ASHRAE Trans., 94(1), 1314-1317. 49 Boonyatilkam, S.,Arch, D. and Jones, J.R. 1989. Smart building controls strategies: research and application. ASHRAE Trans., 95( l), 557-562. 50 Caffrey, R.J. 1988 The intelligent building - an ASHRAE opportunity. ASHRAE Trans., 94(1), 925-932. 51 Rao, M . , Jiang, T.S. and Tsd, J.P. 1987. A frame work of integrated intelligent system. Proc. 1987 IEEE International Conference on System, Man, and Cybernetics, pp. 1 133-1 137 Received: 24 May 1991; Accepted: 19 September 1991.