(12) INTERNATIONAL APPLICATION PUBLISHED UNDER THE PATENT COOPERATION TREATY (PCT)
`
`9
`
`(l
`
`1
`) Wo,ld
`
`",~::~:~0~n;:,::.:';;, o,....n;wtion •
`
`I llll 111111111111111111111111111111111I11111111111111111111111111111111111111111 IIII 11111111
`
`(43) International Publication Date
`15 November 2007 (15.11.2007)
`
`PCT
`
`(10) International Publication Number
`WO 2007/128783 Al
`
`(51) International Patent Classification:
`GOSB 13/02 (2006.01)
`
`(21) International Application Number:
`PCT /EP2007 /0543 23
`
`(22) International Filing Date:
`
`3 May 2007 (03.05.2007)
`
`(25) Filing Language:
`
`(26) Publication Language:
`
`(30) Priority Data:
`S2006/0346
`
`English
`
`English
`
`3 May 2006 (03.05.2006)
`
`IE
`
`(71) Applicant (for all designated States except US): LIGHT(cid:173)
`WAVE TECHNOLOGIES LIMITED [IE/lE]; Inno(cid:173)
`vation Centre, Nova UCD, University College Dublin,
`Belfield, Dublin 4 (IE).
`
`(72) Inventors; and
`(75) Inventors/Applicants (for US only): MCNULTY,
`
`Nicholas [IE/lE]; 172 Holywell, Upper Kilmacud Road,
`Dundrum, Dublin 14 (IE). PACKHAM, Ian [GB/lE];
`18 Hampton Crescent, St Helens Wood, Booterstown,
`Dublin (IE). VANDERSTOCKT, Yann Daniel Edgard
`[FR/FR]; 24 Rue de la Republique, F-59264 Onnaing (FR).
`HAGRAS, Hani [GB/GB]; 21 Caroline Close, Wivenhoe,
`Colchester CO7 9SD (GB). BYRNE, Martin [IE/lE]; 38
`Wainsfort Manor Crescent, Terenure, Dublin 6 West (IE).
`(74) Agents: O'CONNOR, Michael et al.; Unit SA Sandyford
`Business Centre, Sandyford, Dublin 18 (IE).
`
`(81) Designated States (unless otherwise indicated, for every
`kind of national protection available): AE, AG, AL, AM,
`AT, AU, AZ, BA, BB, BG, BH, BR, BW, BY, BZ, CA, CH,
`CN, CO, CR, CU, CZ, DE, DK, DM, DZ, EC, EE, EG, ES,
`Fl, GB, GD, GE, GH, GM, GT, HN, HR, HU, ID, IL, IN,
`IS, JP, KE, KG, KM, KN, KP, KR, KZ, LA, LC, LK, LR,
`LS, LT, LU, LY, MA, MD, MG, MK, MN, MW, MX, MY,
`MZ, NA, NG, NI, NO, NZ, OM, PG, PH, PL, PT, RO, RS,
`RU, SC, SD, SE, SG, SK, SL, SM, SV, SY, TJ, TM, TN,
`TR, TT, TZ, UA, UG, US, UZ, VC, VN, ZA, ZM, ZW.
`
`[Continued on next page}
`---------------------------------------------
`(54) Title: A METHOD OF OPTIMISING ENERGY CONSUMPTION
`
`11
`
`3
`
`9
`
`5
`
`7
`
`1 5~
`
`13
`
`Database
`
`17
`
`21
`
`Management
`Interface
`
`Data Pre-
`Processing
`
`ICE Core
`
`20
`
`19
`
`25
`
`23
`
`27
`
`iiiiiiii
`
`---iiiiiiii
`--!!!!!!!!
`--!!!!!!!! -
`
`iiiiiiii
`!!!!!!!!
`
`!!!!!!!!
`iiiiiiii
`iiiiiiii
`
`~
`Q0
`r--...
`Q0
`M
`1/
`~
`r--... g (57) Abstract: This invention relates to a method and controller (1) for optimising energy consumption in a building. More specifi(cid:173)
`M cally, the present invention describes a method and controller (1) for use in a building having a building management system (BMS)
`0 (3). Typically, the BMS (3) has sensors distributed throughout the building to determine the environmental conditions in the building
`and the BMS controls a heating/cooling system of the building.
`~
`
`[Continued on next page}
`
`GOOGLE 1013
`
`001
`
`

`

`WO 2007/128783 Al
`
`1111111111111111 IIIIII IIIII 11111111111111 I II Ill lllll 111111111111111 lllll 11111111111111111111111
`
`(84) Designated States ( unless otherwise indicated, for every
`kind of regional protection available): ARIPO (BW, GH,
`GM, KE, LS, MW, MZ, NA, SD, SL, SZ, TZ, UG, ZM,
`ZW), Eurasian (AM, AZ, BY, KG, KZ, MD, RU, TJ, TM),
`European (AT, BE, BG, CH, CY, CZ, DE, DK, EE, ES, Fl,
`FR, GB, GR, HU, IE, IS, IT, LT, LU, LV, MC, MT, NL, PL,
`PT, RO, SE, SI, SK, TR), OAPI (BF, BJ, CF, CG, CI, CM,
`GA, GN, GQ, GW, ML, MR, NE, SN, TD, TG).
`Published:
`with international search report
`
`before the expiration of the time limit for amending the
`claims and to be republished in the event of receipt of
`amendments
`
`For two-letter codes and other abbreviations, refer to the "Guid(cid:173)
`ance Notes on Codes and Abbreviations" appearing at the begin(cid:173)
`ning of each regular issue of the PCT Gazette.
`
`The method comprises the steps of gathering weather data relevant to the building, applying a number of intelligent control tech(cid:173)
`niques to the environmental conditions and weather data before determining the accuracy of the intelligent control techniques and
`thereafter determining an appropriate control input for the BMS (3) for subsequent implementation by the BMS. In this way, the
`energy consumption in a building may be minimised by analysing the data in the BMS (3) and suggesting and implementing appro(cid:173)
`priate on/off times, setpoints and other controllable parameters.
`
`002
`
`

`

`WO 2007/128783
`
`PCT /EP2007 /054323
`
`"A Method of optimising energy consumption"
`
`Introduction
`
`5
`
`This invention relates to a method of optimising energy consumption in a building
`
`having a building management system (BMS), the BMS being used to monitor the
`
`environmental conditions of the building and control the heating and/or cooling
`
`system of the building. This invention further relates to a controller for carrying out
`
`1 O
`
`such a method.
`
`Throughout this specification, reference is made to a heating system. However, it will
`
`be understood that the heating system may be used to increase the temperature in a
`
`building and also may be used to decrease the temperature in a building, operating
`
`15
`
`effectively as a cooling system. However, for simplicity, reference is made
`
`predominantly to a heating system and it will be understood that this invention applies
`
`equally to a cooling system and where reference is made to a heating system this is
`
`deemed to include a cooling system also. Furthermore, throughout the specification
`
`the invention is described with respect to a building, however, it will be understood
`
`20
`
`that the invention equally applies to other structures such as ocean liners, cruising
`
`vessels, aircraft and other controlled environments and any reference to a building is
`
`intended to incorporate these other structures.
`
`Building management systems have been in use for some time now and are typically
`
`25
`
`found in a wide variety of buildings ranging in size from skyscrapers down to much
`
`smaller individual office blocks and personal dwellings. These building management
`
`systems are used to control various aspects of the building ranging from security
`
`access to certain areas of the building at certain times, the lighting of the building and
`
`more recently the heating and cooling system of the building. By having such a
`
`30
`
`building management system, an operator will not have to manually turn the lighting
`
`and the heating on and off every day and set the temperature of the heating and
`
`cooling system each and every day. In the case of heating systems in office blocks in
`
`particular, the heating system will normally have to be turned on some time in
`
`advance of the normal working hours in order to ensure that the building is at a
`the employees begin work. By using a building
`
`temperature when
`
`suitable
`
`35
`
`003
`
`

`

`WO 2007/128783
`
`PCT /EP2007 /054323
`
`-2-
`
`management system, an operator will not have to be on site many hours in advance
`
`of the other workers in order to determine when to start the heating system.
`
`There are however, problems with the known building management systems. First of
`
`5
`
`all, these building management systems are not intelligent systems and require direct
`
`input from an operator in order to operate. Although effective in starting and stopping
`
`the heating system at any given time in response to an operator's input, these
`
`systems by and large do not take account of other factors such as ambient
`
`temperature either inside the building or outside the building, the weather conditions
`
`10
`
`of the day and the most economical way of achieving a particular desired
`
`temperature in the building. However, these can be very important factors and in
`
`many countries where the climate may be changeable from day to day with large
`
`changes in temperature from one day to the next, the known systems become
`
`relatively inefficient. For example, during winter months, in order to heat an office
`
`15
`
`building up to a desired temperature, the building management system may be
`
`programmed to start the heating at 7.00am in the morning. However, this does not in
`
`any way take account of the fact that there may have been heavy snow fall the night
`
`before which will slow down the heating process and therefore the building will not be
`
`at the desired temperature by the time the employees begin their working day.
`
`20
`
`Similarly, if there was a particularly mild winter's day and the ambient temperature
`
`outside the building is higher than normal, the heating may not have had to have
`-been-engaged unttt ·a-tater-time-after-1~ooam-thereby wasting -valtiablen~nergy ana(cid:173)
`resources. This problem is exacerbated by global warming whereby weather is
`
`becoming highly unpredictable and weather conditions that would be considered to
`
`25
`
`be abnormal for a particular time of year are becoming more common.
`
`Another problem with the known building management systems is that they do not
`
`allow the operator of the building management system to evaluate the actual cost of
`
`heating versus the comfort level of the employees. Furthermore, the known systems
`
`30
`
`do not appear to appreciate that different heating requirements may apply in different
`
`floors in a building. For instance, in a tall skyscraper in a very warm climate, the air
`
`conditioning may have to be started earlier on the higher floors of the building than
`
`the lower floors of the building as the sun will affect the higher floors first as it rises
`
`over the horizon. Similarly, certain parts of the building may be exposed to direct
`
`004
`
`

`

`WO 2007/128783
`
`PCT /EP2007 /054323
`
`- 3 -
`
`sunlight at different times of the day requiring a different cooling strategy for those
`
`parts of the building. Currently, it is not possible to take that into account.
`
`It is an object therefore of the present invention to provide a method of optimising
`
`5
`
`energy consumption in a building that overcomes at least some of these difficulties
`
`that is both simple to implement and cost effective to provide.
`
`Statements of Invention
`
`10
`
`According
`
`to the invention there is provided a method of optimising energy
`
`consumption in a building having a building management system (BMS), the BMS
`
`being used to monitor the environmental conditions of the building and control the
`
`heating system of the building, the method comprising the steps of:
`
`15
`
`gathering the building environmental conditions data from the BMS;
`
`gathering weather data relevant to the building;
`
`applying a plurality of
`
`intelligent control
`
`techniques
`
`to
`
`the building
`
`20
`
`environmental conditions data and the weather data to determine a proposed
`
`BMS control input for each intelligent control technique;
`
`determining the accuracy of the proposed BMS control input for each of the
`
`intelligent control techniques and thereafter determining an appropriate
`
`25
`
`control input for the BMS; and
`
`providing
`
`the appropriate control
`
`input
`
`to
`
`the BMS
`
`for subsequent
`
`implementation by the BMS.
`
`30
`
`By having such a method,
`
`it is possible to use information relating
`
`to the
`
`environmental conditions of the building such as the internal temperature along with
`
`weather data such as the outside temperature to determine the thermodynamic
`
`characteristics of the building (how the building behaves under varying external
`
`weather conditions) and in turn build up a thermodynamic profile of the building. It is
`
`005
`
`

`

`WO 2007/128783
`
`PCT /EP2007 /054323
`
`-4-
`
`then possible to accurately determine, using intelligent control techniques, when the
`
`optimised start-time, the optimised setpoint and the optimised stop time of the heating
`
`system should be in order to use the least amount of energy possible in order to
`
`achieve a desired temperature by a particular time. For example, during the summer,
`
`5
`
`the heating system may not be programmed to come on until 8.00am in the morning,
`
`however, if it is a particularly cold morning where the temperature is well below
`
`normal levels for that time of year the method is able to take this into account and the
`
`intelligent control techniques each propose a BMS input, in this case the heating start
`
`time for the heating system at some time earlier than 8.00am. The intelligent control
`
`10
`
`techniques may then be assessed for accuracy and an appropriate control input for
`
`the BMS may be derived therefrom.
`
`In this example, it may be determined that the
`
`heating system must be turned on by 7.42am in order to achieve the desired
`
`temperature by the time the employees begin their working day.
`
`15
`
`The step of gathering the building environmental conditions data from the BMS is
`
`essentially a pre-processing step of discovering the pertinent variables that cause the
`
`environmental changes to the building. It is important to make the distinction between
`
`this pre-processing step and the step of using the intelligent control techniques to
`
`make predictions, however the pre-processing step may itself use some intelligent
`
`20
`
`control techniques. The invention may be summarised in a number of different ways,
`
`firstly in that it provides intelligent control based on historical data, secondly that it
`atso-provides intemgent ·control based on weatnen::5redictioris ~and· n~ence· predictive ···
`control and finally it uses artificial intelligence techniques to establish the influence of
`
`major variables relevant to the proposed control suggestions sent to the BMS.
`
`25
`
`It is envisaged that at certain times of the year certain intelligent control techniques
`
`may be more efficient than others. Therefore, by having a plurality of intelligent
`
`control techniques, each determining an appropriate start time for the heating
`
`system, it is possible to evaluate the intelligent control techniques over time and use
`
`30
`
`the most accurate of all the intelligent control techniques for that particular weather
`
`condition. For example, it may be found that one particular type of intelligent control
`
`technique may be particularly efficient during the winter months due to the various
`
`variables that it takes into account. However, the same intelligent control technique
`
`may be very ineffective during summer months. By having a number of intelligent
`
`006
`
`

`

`WO 2007/128783
`
`PCT /EP2007 /054323
`
`-5-
`
`control techniques, it is also possible to choose the best overall approximation of the
`
`start time, for example, of the heating and provide the appropriate control input for the
`BMS.
`
`5
`
`In another embodiment of the invention there is provided a method of optimising
`
`energy consumption in a building in which the step of applying the plurality of
`
`intelligent control techniques to the building environmental conditions data and the
`
`weather data comprises applying two or more of neural network (NN) techniques,
`
`genetic algorithm (GA) techniques and fuzzy logic (FL) techniques. These are seen
`
`10
`
`as particularly useful intelligent control techniques to use. It is envisaged that by using
`
`these intelligent control techniques that each have a relatively small memory footprint,
`
`they may be implemented with existing building management systems in a relatively
`
`straightforward manner. Intelligent control techniques using NN, GA and FL also have
`
`the advantage that they can find relationships between two or more variables
`
`15
`
`including finding patterns in data which is not possible using traditional BMS
`
`technologies based on Proportional, Integral and Derivative (PIO) Control. FL
`
`systems can also be used to automatically find and generate "energy-saving" rules
`
`which are unique to any particular building and generic "energy-saving" rules that are
`
`general to all building environments.
`
`20
`
`25
`
`30
`
`In one embodiment of the invention there is provided a method of optimising energy
`consumptton tn -a -buttdtng -m -wn1m-•tt1e -step of aelerm,nrng ffie -accuracy of fffe
`intelligent control techniques further comprises the steps of:
`
`comparing the current building environmental conditions data and weather
`
`data with historical data stored in a database;
`
`determining the set of historical data that most closely matches the current
`
`building environmental conditions data and weather data; and
`
`thereafter determining the accuracy of the intelligent control techniques based
`
`on the accuracy of the intelligent control techniques historically.
`
`By carrying out such a method, it is possible to determine which of the intelligent
`
`007
`
`

`

`WO 2007/128783
`
`PCT /EP2007 /054323
`
`-6-
`
`control techniques was most accurate historically for a given set of weather
`
`conditions. It may be found that one particular intelligent control technique was highly
`
`accurate during winter months when snow was forecast. Therefore, this intelligent
`
`control technique may be preferred when the same weather conditions are being
`
`5
`
`experienced.
`
`In a further embodiment of the invention there is provided a method of optimising
`
`energy consumption in which the step of determining the appropriate control input
`
`for the BMS comprises using the intelligent control technique that is determined to
`
`10
`
`be the most accurate for those conditions. Alternatively, the step of determining the
`
`appropriate control input for the building management system comprises generating
`
`a control input from a weighted average of a plurality of the intelligent control
`
`techniques with the weighting based on their historical accuracy. In other words, it is
`
`possible to take either the most accurate intelligent control technique response or to
`
`15
`
`use a weighted average of a plurality of the intelligent control techniques so that an
`
`average result is taken with a high probability of accuracy.
`
`In another embodiment of the invention there is provided a method of optimising
`
`energy consumption in a building in which the step of determining the accuracy of the
`
`20
`
`intelligent control techniques further comprises minimisation of the error of each of
`
`the intelligent control techniques. By this, what is meant is determining the relative
`-accuracy of the-intettigent · control up to ~a ~certain tower bound lo avoid over-fitting or -
`under-fitting of the model. This further avoids the possibility of over-training or under(cid:173)
`
`training the neural networks. This is seen as a particularly efficient way of determining
`
`25
`
`the accuracy of the intelligent control techniques and assisting in the selection of the
`
`appropriate intelligent control technique and hence the appropriate control input for
`the BMS.
`
`In one embodiment of the invention there is provided a method of optimising energy
`
`30
`
`consumption in a building in which the step of providing the appropriate control input
`
`to the BMS further comprises providing one or more of an optimal start time, an
`
`optimal stop time and a setpoint control.
`
`In a further embodiment of the invention there is provided a method of optimising
`
`008
`
`

`

`WO 2007/128783
`
`PCT /EP2007 /054323
`
`- 7 -
`
`energy consumption in a building in which the BMS data and weather data are
`
`received over a network interface. Preferably, the network is the internet. such
`
`Alternatively, the network may be a network such as a Virtual Private Network which
`
`is IP based or a circuit-switched (PSTN) or other packet-switch network such as
`
`5 mobile 3G or GPRS networks. In this way, data may be received from external
`
`sources.
`
`In another embodiment of the invention there is provided a method of optimising
`
`energy consumption in a building in which the intelligent control techniques are
`
`1 0
`
`arranged in a cascaded manner.
`
`In this way, it is possible to have the intelligent
`
`control techniques used to control a large number of different components of the
`BMS. Furthermore, several different intelligent control techniques may be used to
`
`determine a particular control input.
`
`15
`
`In one embodiment of the invention there is provided a method of optimising energy
`
`consumption in a building in which the weather data comprises predicted weather
`
`data. Alternatively, or in addition to this, current weather data may be used.
`
`In this
`
`way, the method incorporates future weather conditions such as those forecast by a
`
`weather forecast service which may be retrieved over the internet or manually input in
`
`20
`
`order to provide a strategy of the BMS and to provide accurate future inputs for the
`BMS.
`
`In one embodiment of the invention there is provided a method of optimising energy
`
`consumption in a building in which the intelligent control techniques use recursive
`
`25
`
`processing to determine control inputs to the BMS. The advantages of recursive
`
`processing are that a simple model can be created that can keep calling itself with
`
`minimal processing time, the number of "synthetic" variables required by a recursive
`
`method is lower than others because the model creates these values during
`
`processing,
`
`therefore
`
`the amount of pre-processing
`
`is also reduced before
`
`30
`
`deployment.
`
`In a further embodiment of the invention there is provided a method of optimising
`
`energy consumption in a building in which the step of determining an appropriate
`
`control input for the BMS from the intelligent control techniques further comprises
`
`009
`
`

`

`WO 2007/128783
`
`PCT /EP2007 /054323
`
`-8-
`
`using an adaptive decider to decide which intelligent control technique is to be used,
`
`the adaptive decider ranking each of the intelligent control techniques periodically.
`
`In another embodiment of the invention there is provided a controller for optimising
`
`5
`
`energy consumption in a building having a heating system monitored and controlled
`by a building management system (BMS), the controller comprising:
`
`10
`
`means for receiving building environmental conditions and weather data
`
`relating to the building in which the controlled heating system operates;
`
`a database for storing the building environmental conditions data and weather
`
`data therein;
`
`a core processor having a plurality of intelligent control technique units, each
`
`15
`
`of the intelligent control techniques units having means to receive building
`
`environmental conditions data and weather data and provide a proposed BMS
`
`control input;
`
`the core processor further comprising means to determine the accuracy of
`
`each of the intelligent control technique units and means to determine an
`appropriate control input for the BMS; and
`
`the controller having means to transmit the appropriate control input to the
`BMS.
`
`20
`
`25
`
`By building plant conditions data what is meant is boiler and chiller set-points, valve
`
`positions, AHP fan speed and the like.
`
`In another embodiment of the invention there is provided a controller for optimising
`
`30
`
`energy consumption in a building in which the plurality of intelligent control technique
`
`units comprise two or more of a fuzzy logic unit, a genetic algorithm unit and a neural
`
`network unit.
`
`In a further embodiment of the invention there is provided a controller for optimising
`
`010
`
`

`

`WO 2007/128783
`
`PCT /EP2007 /054323
`
`-9-
`
`energy consumption in a building in which the means to determine the accuracy of
`
`each of the intelligent control techniques units comprises means to compare the
`
`current set of inputs with historical inputs stored in the database and determine which
`
`of the intelligent control technique units was most accurate historically.
`
`5
`
`In one embodiment of the invention there is provided a controller for optimising
`
`energy consumption in a building in which the core processors means to determine
`
`an appropriate control input for the BMS further comprises an adaptive decider.
`
`10
`
`In another embodiment of the invention there is provided a controller for optimising
`
`energy consumption in a heating system in which the adaptive decider has means to
`determine the most accurate proposed BMS control input received from the intelligent
`
`control technique units and use that control input as the appropriate control input for
`
`the BMS.
`
`15
`
`In a further embodiment of the invention there is provided a controller for optimising
`
`energy consumption in a heating system in which the adaptive decider has means to
`
`determine the accuracy of each of the proposed BMS control inputs received from
`
`the intelligent control technique units and generate an appropriate control input for
`
`20
`
`the BMS based on a weighted average of a plurality of the proposed control inputs of
`
`the BMS.
`
`In another embodiment of the invention there is provided a controller for optimising
`
`energy consumption in a heating system in which the core processor has a data pre-
`
`25
`
`processing unit to rank each of the intelligent control technique units periodically
`
`thereby providing a weighting value to that intelligent control technique unit.
`
`In one embodiment of the invention there is provided a controller for optimising
`
`energy consumption in a heating system in which the core processor is provided with
`
`30
`
`a plurality of adaptive deciders arranged in cascading format.
`
`Detailed Description of the Invention
`
`The invention will be more clearly understood from the following description of
`
`011
`
`

`

`WO 2007/128783
`
`PCT /EP2007 /054323
`
`- 10 -
`
`some embodiments thereof, given by way of example only, with reference to the
`
`accompanying drawings, in which:-
`
`Fig. 1 is a diagrammatic representation of the overall architecture of the
`
`5
`
`controller used to carry out the method according to the invention;
`
`Fig. 2 is a diagrammatic representation of a control panel used with the
`
`controller of the present invention;
`
`10
`
`Fig. 3 is a block diagram of a plurality of adaptive deciders in cascaded
`
`format used by the controller;
`
`Fig. 4 is a flow diagram of the energy prediction and optimisation agents;
`
`15
`
`Fig. 5 is a diagrammatic representation of a predictor/optimiser neural
`
`network with genetic algorithms;
`
`20
`
`Fig. 6 is diagrammatic representation of a predictive recursive optimal control
`
`unit for use with the controller of the present invention; and
`
`Fig. 7 is a diagrammatic representation of a zone in a building in which the
`~method and· controifenrccordlng··to thepresentlnvenlion ··operate.
`
`Referring to the drawings and initially to Fig. 1 thereof there is shown a controller
`
`25
`
`for optimising energy consumption, indicated generally by the reference numeral 1.
`
`The controller 1 operates in a building (not shown) having a heating system
`
`monitored and controlled by a building management system (BMS) 3. The controller
`
`1 comprises a BMS interface 5 and a weather interface 7 for receiving building
`
`environmental conditions data and weather data respectively over a network 9, in
`
`30
`
`this case the internet. The weather data is received by the weather interface over
`
`the network 9 from a weather provider 11. The controller 1 further comprises a
`
`database 13 having a data interface 15, a core processor 17, a supervisor module
`
`(not shown), a task scheduler 19, a management interface 20 and a user interface
`
`21. The core processor 17 further comprises a plurality of intelligent control
`
`012
`
`

`

`WO 2007/128783
`
`PCT /EP2007 /054323
`
`- 11 -
`
`technique units 23 (only one of which is shown), a data preprocessing unit 25 and a
`
`sensor validation unit 27.
`
`In use, the controller 1 gathers building environmental conditions data from the BMS
`system 3 and weather data from the weather provider 11. The building environmental
`
`5
`
`conditions data and weather data are stored in database 13 for subsequent
`
`processing by the core processor 17. The building environmental conditions data
`
`and the weather data are in turn applied to a plurality of intelligent control technique
`
`units 23 which each provide a proposed BMS control input based on the building
`
`1 0
`
`environmental conditions data and the weather data. The core processor 17
`
`determines the accuracy of the proposed BMS control inputs for each of the
`
`intelligent control techniques and thereafter determines an appropriate control input
`
`for the BMS. A response is sent from the core processor 17 to the BMS system 3 via
`
`the BMS interface 5. The BMS system may thereafter operate using the appropriate
`
`15
`
`control input.
`
`In the embodiment shown either current or predicted weather conditions may be
`
`provided from the external source via the internet.
`
`Indeed, the BMS system itself
`
`may also provide data such as the actual current temperature inside a particular floor
`
`20
`
`of the building or the actual temperature outside a particular building. The inside
`
`temperature or any inside variables of the building are not considered "weather data"
`
`· by-the·system.·rtowevenne BMS·may have~sotar1ndex sensors and the like-that-(cid:173)
`
`would be considered to be weather data. The supervisor module (not shown)
`
`monitors and controls system processes. The task scheduler 19 schedules tasks in
`
`25
`
`the controller such as getting the weather for the next time period for the core
`
`processor so that it may carry out calculations on the building environmental
`
`conditions data.
`
`The
`
`intelligent control
`
`techniques comprise neural network
`
`techniques, genetic algorithm techniques and fuzzy logic techniques. Each of these
`
`techniques may be particularly accurate in certain circumstances in environmental
`
`30
`
`conditions and less accurate in other environmental conditions. Therefore, it is
`
`possible to choose the most accurate intelligent control technique for use in that
`
`particular environmental condition. This is achieved by using the data preprocessing
`
`module 25 which monitors the accuracy of the predictions of each of the intelligent
`
`control techniques over time and thereafter may assign a weighting to each intelligent
`
`013
`
`

`

`WO 2007/128783
`
`PCT /EP2007 /054323
`
`- 12 -
`
`control technique so that a weighted average of each of the intelligent control
`
`techniques based on their historical accuracy may be provided.
`
`As an alternative to this it may be preferable to simply provide the most accurate of
`
`5
`
`each of the intelligent control techniques without providing a weighted average. This
`
`would depend on the preferences of the user. Certain intelligent control techniques
`
`may be used to control different parts of the BMS in preference to other intelligent
`
`control techniques. Furthermore, different intelligent control techniques such as
`
`those understood in the art of intelligent control techniques may be implemented also
`
`1 O
`
`in a relatively straightforward manner. Other intelligent control techniques include hill
`
`climbing algorithms such as gradient descent and Tabu search. Also, Bayesian Belief
`
`Networks used for expert systems and other neural networks such as Self-Organising
`
`Maps (SOM) for sensitivity analysis and recurrent neural networks. By storing the
`
`values of the environmental conditions, the weather conditions and the resulting
`
`15
`
`values of the building management system, it is possible to determine, over time,
`
`those techniques that are more successful than others in achieving the desired goal
`
`(of reducing the building energy demand). Furthermore, it is possible to determine
`
`which of the techniques is particularly efficient in one weather condition and those
`
`which are efficient in other weather conditions. Therefore, the historical analysis is
`
`20
`
`particularly useful in this invention.
`
`-nie controtter -11-may use any intelligent algorithm -or -combination -or algorithms to -·
`control any part of the BMS system. For example, there may be a number of states
`
`of the system that can be monitored to optimise energy consumption such as optimal
`
`25
`
`start, optimal stop and optimal set-point control. Certain intelligent control algorithms
`
`may be more effective than others. Furthermore, the data preprocessing module 25
`
`that determines the most pertinent variables for each of the control modules could
`
`itself use any intelligent algorithm such as a genetic algorithm or fuzzy logic
`
`controller. In that way, the data pre-processing system is used to find the most
`
`30
`
`dominant control variable in the system under control using either Fuzzy Logic,
`
`Neural Networks or ReliefF. ReliefF is a common name for relief algorithms that are
`
`general and successful attribute estimators. An adaptive decider (not shown) can
`
`also be used to select the optimal algorithm.
`
`014
`
`

`

`WO 2007/128783
`
`PCT /EP2007 /054323
`
`- 13 -
`
`Referring to Fig. 2 of the drawings there is shown a diagrammatic rep