DE19600694C2 - climate system - Google Patents

Description

The present invention relates to a programmable Air conditioning system according to the preamble of patent claim 1 so as a corresponding method according to the preamble of Patent claim 14.

The following description refers by way of example a heating system. However, the system is for control Useful temperatures suitable, for example, the Steue tion of the cooling capacity of an air conditioner.

Central heating systems are known from the prior art, where in a boiler to a flow temperature heated water via lines in individual radiators ge is headed, in which the transported water its heat gives off and with a lower return temperature to Boiler flows back. Typically every room is egg a building with one or more radiators ausgestat tet.

There are also known heating systems, where ever the room is an individual, independent heating element has, for example, an oil stove.

In all types of heating, the problem is the Control or regulation of the room temperature.  

There are known systems in which the user the inflow the heat-transporting medium by means of a controller set directly. These include, for example, the Steue regulation of the inflow of warm water into one Radiator by means of a set screw, or the controller or regulation of the feed of gas or oil in a entspre heating element. Especially with more modern Zen heating systems use thermostatic valves, with which a desired room temperature permanently turned on can be made. After exceeding or falling below the one set room temperature is here a Regelungsmech nism (valve) activated, which corresponds to the heating power reduced or increased. When permanently leaving or entering a room, the mechanism or the Turn valve on or turned off.

As a disadvantage of these known systems are her gro Viewed operating expenses.

So must with a simple control of the heating power with Constantly adjusted up and down by means of a control mechanism because this system is unable to last adherence constant temperature.

Although a thermostatic valve can maintain a constant temperature maintained, but must leave the room of Be down-regulated when heating energy is saved that should. Experience teaches that turning this down Often forgotten by users. Will after leaving the room does not turn off the heating, this leads to useful loose energy consumption, and of course to high co sten.  

Recently, there are processor controlled heaters Systems have become known in which explicit time entered by a user. So can the User the temperature at a peripheral radiator (or at a central heating unit) time-related to adjust.

From DE 40 09 774 C2 is a method for entering Program data in a heating controller for creating ei during the daily life of a heating system ge variable room temperature profile known. in this connection the operator of the system must during a day too set times desired temperature values ei Enter the setting device, with the heating controller based on the set values a daily base profile is created. the This daily base profile will be after a learning period of some Days, if on one of the following days et At the same time, the same temperature value was entered again ben (confirmation) or by another temperature value is replaced (correction), or if after expiration of a be voted time no confirmation or correction was made. Values entered after completion of the learning phase will only become then taken into account to correct the daily base profile, if the controller is given an explicit instruction.

A generic control and regulating device for a Installation for space heating, the at least one central heating for transferring heat energy to a medium and a Variety of radiators, each in one corresponding space are arranged and with the heating medium to be supplied by the central heating is out of the DE 31 21 596 A1. Here are a memory for a lot number of predetermined and to desired room tempera  tures as well as attendance or absence periods the user in the room for a predetermined time room-related basic data for each room, a program unit for changes to the desired values in terms of room temperature and presence or Absence periods of the user in the room related Random data for each room, a processing facility, the base data stored in the memory with the Random data, entered via the programming unit, ver equals and adjusting devices for the passage of Heizme Diums provided by each radiator, over from the Processing device controlled output signals be provided. Finally, EP 0 631 219 A1 is known a method for temperature control and regulation in a individual living or working spaces. According to this Ver should drive constant habits of residents considered, but at the same time a sufficient Kom and how to maximize energy savings. This is achieved according to the local method, that a set schedule based on presence fields recorded attendance on the following days after a fuzzy algorithm to the adjusting Re regularity of use of space.

In conventional methods is considered disadvantageous that from the user or operator of the heating a high on attention is required, as it leads to various, regular It is necessary to make temperature inputs according to which times then a temperature profile is created. Such an Ver driving is also very inflexible, since a change of the Sy stems again associated with programming effort. experi In this case, an already ge stored temperature profile, which the current request  no longer quite the same, accepted to escape this programming effort.

Especially in large office buildings, the above be written disadvantages apparent. Since by Un mindfulness the user incurs unnecessary heating costs, is often tried here, fixed settings of a Zen Provide tral heating. Such a firm attitude goes but very inadequate to the modern working world (flexible working hours, etc.). Depending on the start of work of the individual employee is in a room too early, in heated one another too late.

The object of the invention is the creation of an air conditioning system, that is as easy to use, so that the attention ability of a user is claimed as little as possible.

This task is solved by a climate system with the Features of claim 1.

With the climate system according to the invention can the indoor climate up the needs of one or more people set become.

Furthermore, a high level of user-friendliness is ensured. The user does not have to do complicated programming men. Only through concrete user input or -anwei The system is controllable.

Depending on the version of the system are different Benut Inputs specified:  

In the simplest embodiment, only the on instructions "present" or "absent". Entspre The system sets the input to one preset temperature.  

According to a preferred embodiment shown in FIG. 1, "warmer" and "colder" inputs are possible. These instructions may be provided along with the above "present" / "away" instructions, but it is also sufficient to leave only these "warmer" or "colder" statements. The system is capable of generating both a presence forecast and a forecast of a desired temperature profile based solely on these "warmer" and "cold" statements.

The climate system according to the invention is both central heaters, as well as in non-central heating, such as local Radiant heaters, etc., usable. At a central heating can the inventive method both at the central unit (for example, boilers), as well as individual heating bodies are made.

The climate system according to the invention is able indi individual user habits, such as kitchen use only at mealtimes, bath only in the morning and in the evening, sleep Room usage only during the night to be considered.

To be considered particularly advantageous that the erfindungsge suitable system without much effort on conventional heating system can be retrofitted.

Advantageous developments and modifications are counter stood the dependent claims.

Embodiments of the invention will now be described with reference to the beige added drawings explained in more detail.  

It shows:

Fig. 1 shows schematically a preferred embodiment of to the invention OF INVENTION climate system namely the construction of a heating system,

Fig. 2a is a presence prognosis for a typical week day,

FIG. 2b shows a presence prognosis for a holiday,

Fig. 3 shows a desired temperature profile and an actual temperature profile when providing an automatic Her under control of the temperature.

In Fig. 1, the air conditioning system according to the invention is the case of a peripheral heater of a central heating Darge presents.

A heating valve K is controllable by means of a servomotor L. The climate system according to the invention is in the environment of the heating valve K, for example, directly on the radiator, ange introduced.

The system has a single-chip micro-processor A. This microprocessor includes a ROM (program memory) (not shown). Microprocessor A is connected to a timer module B, a RAM module C and AD converters E, F, G. The A / D converters E, F, G each receive signals from a warmer / cold switch H, a temperature sensor I and a threshold value controller (S-value controller) J. The AD converters E, F, G via the lines 1.1 , 1.2 , 1.3 received signals to the microprocessor A via the lines 2.1 , 2.2 , 2.3 weitergege ben. A processed signal is given via line 4.1 to a DA converter D. The signal generated by this DA converter D signal controls the actuator L, which regulates the opening state of the heating valve K.

A new setting of the date or the time is about one Switch O possible.

Optionally, further inputs can be provided. By way of example, FIG. 1 shows a moisture sensor N, which is connected via a line 1.4 to a further AD converter M. The signal generated by this AD converter M is given to the microprocessor A via a line 2.4 .

As another possibility of an optional probe is example an outdoor temperature measuring device mentioned.

The system controls the heating activity with the help of a progno se about the presence of a user.

It will first be described in the following, in which Wei the climate system according to the invention a prognosis over created the presence of a person in the room to be heated.

The state of the heating system is over the time t (eg abso lute time t or date and time of day) and all outside entrances completely described. The time t is here by the timer construction stone B provided. These state parameters will be in the following called "ground state".

The user enters via the switch H to the system as desired, usually at irregular intervals, instructions. In Ausfüh approximately example according to FIG. 1, only the inputs "warmer" or "colder" are possible. The system thereby contains information regarding the input time and the desired temperature control.

In a simplified embodiment, it is also possible to Enter "presence of the user" or "absence of the user to be provided.

It is possible that the time thus entered is discrete time interval to allocate. For example, if you divide the 24 hours of the Day in 96 15 minute intervals, will be an entered time each associated with such a 15 minute interval. The Inter Valleys do not necessarily have the same length. It is For example, in the early morning to provide two-hour intervals.

Additional parameters, from the ground state for the forecast basis are, for example, time of day, year time, weekday or the presence of a Sunday or holiday.

The derived parameters along with the ground state become hereinafter referred to as "derived state".

Alternatively, the derived parameters can be used as coefficients of Fourier functions, trigonometric functions, polynomials or Taylor series. This proves to be an advantage liable if a periodic behavior with a period duration from Z. For example, a day should be forecast.

The system stores in RAM C the user inputs together men with the current ground state, that is with respect to the User input becomes an n-tuple, consisting of the user and the parameters time and temperature. A  such "memory unit" consisting of user instruction and Basic state is called "instruction" in the following. The total melten instructions form the memory of the "learning" Heating control system and provide the basis for more Calculations.

At overflow of memory, the oldest instructions be deleted or overwritten.

On the basis of all stored instructions becomes the forecast posed about the user's presence. This presence forecast is recreated or updated with each user input lisiert. It takes place in the background, that is, almost parallel (Multitasking).

The microprocessor A now calculates based on the stored secure instructions for the following time interval (eg 1-5 hours), for example, minute-by-minute discretized, the property probability of a user. This becomes a true calculated in the interval from 0 to 1.

It can also have a general presence function yet continuous presence probabilities determined who which, however, are scanned discreetly again.

The interval length of the forecasting interval should be at least be the same length of time as the heater for the high heat to a maximum temperature on average (heating leading period).

A preferred method for creating the prognosis is the Regression analysis. This method is from the field of Sta and forecasting technology well known.  

As an alternative to the regression analysis, elements of the new ronal networks, varianzreduzierende method or a fuzzy Logic method can be used.

It is in this context regarding the combination of Basic parameters (in general, but not necessarily, a linear combination), only important that their weights be estimated so that they are as good as possible the target size (Anwe senheit of the user).

The regression analysis determines the weights of the parameters of the derived state, so that this weighted combines the Estimate the probability of presence.

This will be explained in detail by means of an example:

It should in the following example about a week's presence in a typical law office can be predicted.

The following terms are introduced:
t: time t on an infinite time axis,
f (t): Presence function at time t, where f (t) = 0 indicates that the user is absent at a time t, and f (t) = 1 indicates that a user is present at time t
R (t): current room temperature at time t of a user input

The ground state of the system is represented by the values t and R (t) certainly.

An "instruction" is defined by a tuple (t, R (t), f (t)). That is, an instruction tuple consists of the time t, the tempe C and the position of the switch H (f (t) = 0 or 1) to Time t.

In the case of further sensor inputs, z. B. the humid nometer, the instruction tuple is extended accordingly. In the case of an additional air moisture meter N it would then: (t, R (t), f (t), N (t)).

From a given ground state at time t, the derivative state consisting of a set F ₁ (t) to F _n (t) is now generated.

The functions F ₁ to F _n can be either discrete (workday yes / no, holiday yes / no, etc.) or continuous (sine function, cosine function, etc.).

In order to be able to predict a recurring daily course, the Fourier basis functions (sinx, cosx, sin2x, cos2x,...) Are available with a transformation of the time of day from 0:00 to 24:00 on the interval 0 to 2π. If more regularities are forecasted, for example over a week (month, year), the corresponding interval is mapped to (0-2π), ie around Monday 0 ⁰⁰ to 0 and Saturday 24 ⁰⁰ to 2 π. Otherwise, provide for a more extensive forecast of user presence, which is not alone by the time of day gege ben, such as the use of a room on weekdays (Monday to Friday), and limited on Saturday, but not on Sundays or public holidays, etc. offer discrete functions derived from the date (holiday Yes / No with corresponding function values 1 or 0), which are superimposed with the continuous functions.

The following example predicts the following derived functions, where t 'is the transformation of the time of day interval (0:00, 24:00) to (0, 2π):

Weekday Yes / No (1, 0) F ₁ (t) Holiday Yes / No (1, 0) F ₂ (t) Sin t ' F ₃ (t) Cos t ' F ₄ (t) Sin 2t ' ₅ F (t) Cos 2t ' F ₆ (t) AL = L '. , , Sin 11t ' F ₂₃ (t) Cos 11t ' F ₂₄ (t) 1 F ₂₅ (t) (constant value)

With the derived functions F ₁ (t) to F _n (t) and the property fs (t), the linear regression model described below is built according to the "least squares method".

Table 1 shows the times during which a user is scheduled to log in System enters a statement marked with crosses. In the Table 1 can be viewed half an hour from the user zer have been entered via switch H. But that's just it if possible, that the user only at the beginning and the At the end of his presence, switch H is always labeled "Warmer / Käl instructs or informs the system that it is present / absent is. In such a case, the time between these two points in a freely selectable manner, not  necessarily equidistant, discretized, and it becomes each generated discrete time t of the associated destination radio tion value or presence value f (t) generated in such a way that f (t) = 1 when t is in the presence phase and others if it is zero.

In the concrete example, the late evening and night hours (from 20 o'clock) hourly, and the remaining periods half-hourly dis cretized (see Table 1).

In the case of generated instructions (direct user instructions or "intermediate statements" generated by the system), a matrix F with the elements F _i (t _j ) with i = 1 _,. , , n and j = 1,. , , m, ie a matrix with m rows and n columns.

The result is the target vector f with the elements f (t _j ) with j = 1,. , , m. The result is also the vector with the weights of the individual F _i , namely X with the elements x _i (i = 1,... N).

In the present example, the system shown in Table 1 entered visible attendance times, z. On Mondays Presence 8: 30-11: 30 and 13: 30-17: 30 clock.

Based on these data, the weights x _i , i = 1,. , , n the presence prediction function

P (t _j ): = Σ x _i F _i (t _j ) with i = 1,. , , n and j = 1,. , , n

determined so that the target function or presence function f (t) by P (t) is predicted as well as possible. This is done according to the least squares method (Gaussian method, also called linear regression analysis). It is necessary to solve the following linear system of equations:

F. F ^T. X = f,

where F ^{T is} the transpose of the matrix F defined. The result is the following solution:

X = (F, F ^T ) ^-1 . f

In the example chosen, the values given in the uppermost row of Table 2a, ie z. Eg x ₁ = 0.19, x ₂ = -0.18 etc.

The concrete forecast value at time t then results as the result value of the prediction function P (t) with the calculated x _i .

FIGS. 2 a and 2 b show the presence forecasts determined with the calculated weights for a typical day of the week ( FIG. 2 a) or a holiday ( FIG. 2 b).

The forecast values are in the column "Forecast" in the respective Tables 2a and 2b explicitly stated.

With the introduction of a threshold (S-value) can now determine when to forecast presence d. H. for which times to heat.

Is about, as in the present example (see Fig. 2a and 2b), assuming a threshold of 0.5, it is for all time intervals Inter with a predicted presence probability greater than 0.5, the presence, and for the presence of true probabilities less than or equal 0,5 the absence assumed.

The determined presence probabilities are in the present example, continuous values between 0 and 1, where a value of 0 represents a safe absence while a 1 means a secure presence. Calculated values that less than zero will also be a safe absence assigned.

In a subsequent step, the presence forecasts discretized, d. H. either fixed at 0 or fixed at 1 det. If the probability value is greater than the threshold, If this probability is mapped to 1, the true is It will be less than or equal to the threshold shown on 0

By presetting the threshold, it is possible to use the Heating characteristic individually, namely rather conservative or rather progressive, to steer.

Setting a low threshold means that even with small probability of presence, d. H. at Suspicion of presence, being heated, while a high coffin lwert a heating process only when very certainly assumed Anwe in motion.

A high threshold value saves money, but brings insofar as inconvenience with, as under certain circumstances the User encounters an unheated environment.

A low threshold, however, brings larger Convenience, because even at relatively low attendance Probably the heating is set in motion. The attitudes A low threshold is of course higher Heating costs connected. The threshold switch is also available as input /  Off switch usable. If the threshold z. B. turned on 0 is already present at a probability of presence from 0, d. H. always, heated.

In Fig. 2a it can be seen that the probability of presence exceeds the threshold, which is set here by way of example to 0.5, only at certain morning or afternoon times. On a public holiday, the probability of presence remains below the threshold, so that the heating does not go up ( FIG. 2b).

The described creation of the Presence Forecast is already done possible with a system which only activates the user input " It is with such a system for example, possible that with predicted presence always regulated to a predetermined temperature. This before set temperature can be either by the user, or too centrally set for a group of heating elements who the.

It is according to another embodiment of the erfindungsge However, climate systems are also possible on the basis of Presence forecast a forecast of a desired tempera to create a structure. For this purpose, it is provided that the user zer over the switch H both "warmer" - and "colder" -Anga enter into the system. This is analogous to the above described embodiment, each such input first interpreted as presence input, and the one described above Invoice carried out.

Based on the presence forecast will match with the heated the supply so that the forecasted property a temperature is reached by a desired temperature  tur function (desired temperature) C (t) is defined.

C (t) is determined analogously to P (t), where for the calculation of C (t) values c (t) are used, which are generated as follows the:

If at a time when the room temperature R (t) is greater is equal to C (t) minus δT (delta T), a "warmer" instruction is given, the new desired temperature value c (t) is determined as C (t) + δT. If a "colder" statement is made, ana log c (t): = C (t) minus δT. Here δT is a freely selectable one Temperature difference with which the previous desired temperature C (t) is corrected.

It is possible C (t) factory, z. B. at a constant 20 ° C before to admit.

If the respective conditions are not fulfilled, then no new c (t) is generated.

The weights y _i , i = 1. , , n the desired temperature forecasting function

C (t _j ): = Σx _i c _i (t _j ) with i = 1. , , n and j = 1. , , m

are to be determined such that the objective function c (t) is predicted as well as possible by C (t). Such a prediction function is advantageously determined analogously to the above-described presence prediction function by means of the least squares method. It is the following linear equation system for this purpose

F. F ^T. Y = c

to solve, where with F ^T, the transpose of the matrix F is defined. The result is the following solution:

Y = (F, F ^T ) ^-1 . f

A concrete calculation example is not given here, it However, it should be noted that such an invoice itself completely analogous to the prognosis of Presence Probation in which only f (t) is replaced by c (t) and appreciated.

In a start-up phase, the system can initially at assumed Presence of the user a preset desired temperature taking. This preset desired temperature can either factory or determined by the user.

In the case of a user instruction "warmer" or "colder", there are two cases:

• 1. The system determines that it has already reached the desired temperature. The user input thus means that the previous desired temperature is either too high or too low.
  In this case, the desired temperature is increased or decreased by a differential amount (for example, between 0.5 and 1 ° C), and stored together with the ground state (absolute time and external inputs).
• 2. In the second possible case, the system determines that it is the previously valid desired temperature has not yet reached. In In this case, from a user instruction of the kind "would "nothing can be closed  User instruction of the type "colder" a new, lower one Desired temperature can be derived at this time.

Overall, the system thus generates a presence function in Combination with a temperature profile. So if the property probability of the user's default Threshold or probability value has exceeded, the heating is adjusted to a corresponding temperature value gelt, which progno. Based on the stored temperature profile is stigmatized.

It is for the purpose of a flexible heating operation or a fast relearning of the air conditioning system possible, user instructions weighted according to their relevance. So can at For example, the recent user inputs are weighted more heavily the (exponential weighting).

For an effective installation of the air conditioning system Different possibilities are advantageously usable:

First, it is possible to factory already with a Presence probability profile and a temperature pro to provide fil. Standard profiles can be used However, it could also be based on individual information from Benut factory-oriented profiles or factory-made profiles to be created.

It is also possible, a completely unprogrammed or uninitialized system. This naturally leads especially in the start-up phase of the system, that a still completely insufficient prognosis basis exists. The User must then be prompted to make inputs which then build a reliable forecasting system  can.

Is the system off with a presence / absence switch? equipped and able to switch between a presence statement and an absence statement (i.e., in the presence of the user zers) on their own at discrete time intervals generate the forecasting process in a relatively short time provided an adequate statistical basis.

In the case, however, that only concrete inputs of the user the System are available as a basis for forecasting, in an initial phase, generally after installation of the Heating, any forecast of user presence or about the desired temperature with considerable uncertainties or Be subject to fluctuations.

For this purpose, the user, depending on the number of existing instructions, so irritated or excited that he feels obliged to instruct the system more often give.

This happens because the system of its own tempera downgrades, so that the user often the "heat mer "statements, which of course implies the system zit the presence of the user is communicated.

Also in the event that one system is operated by multiple users it is, as it will typically occur in offices, it is meaningful, the presence or desired temperature forecast on to provide a solid statistical basis.

In the following, with a heating interval, that time interval from t ₀ to t ₁ is designated, which starts with the jump of the property safety prognosis from "absent" to "present", and ends with the corresponding jump from "present" to "absent".

T _i denotes a time within such an interval (t _{0 is} less than or equal to t _i less than or equal to t ₁ ), ie t _i is from the range of a contiguous time period for which presence is predicted. t _0, with each new user input at time t _currently re _currently set at time t. The interval t ₀ to t ₁ is correspondingly reduced to the interval t _actual to t ₁ .

To explain the Temperaturherunterregelung the Reizfunk tion Z (t _i ) is introduced:

Z (t _i ) = p. (t _i -t ₀ )

where p is a function of the number of instructions such that Z (t _i ) per hour z. For example, 2.4 degrees C for a small number of instructions, 1.0 degrees C for an average number of instructions, 0.3 degrees C for a large number of instructions, and 0.0 degrees C for a very large number of instructions Instructions increased.

If such a stimulus function is provided in the system, this regulates Effectively not to the calculated desired temperature C (t), but by the temperature decreased by Z (t), d. H. on C (t) - Z (t), where Z (t) then starts again with 0, then when either a new occupancy interval begins, or when the user has given an instruction (warmer or colder).

A greatly simplified, but in principle typical dynamic heating course, taking into account such a user tion is shown in Fig. 3. It is assumed that conditions in the system, only a small total number of user instructions is stored.

According to FIG. 3, the following system parameters are defined:

Because only a small number of stored instructions before The system regulates the desired temperature C (t) per hour 2.4 ° C from. C (t) is initially 20 ° C. ΔT is fixed at 1 ° C placed. The current heating interval ranges from 8:10 to 1:10 pm.

Further, as shown in FIG. 3, user inputs of the "warmer" type occur at times 8:50, 9:30, 9:50, 10:50, 11:30 and 12:40.

Since in this example is very much down regulated, namely at 2.4 ° C / hour, the user is too frequent system instructions coerced.

In Fig. 3, the upper line shows the predicted desired temperature C (t) while the sawtooth line represents the effective temperature history. The user inputs are marked by the squares.

The system regulates the temperature of 15 ° C to 8:10 predicted desired temperature of 20 ° C high, and at 13:10 back to 15 ° C.

If, at the time of an instruction, the temperature is one value greater than or equal to δT is dropped, in the case of an instruction the predicted desired temperature of 20 ° C restored.

However, if, as shown by way of example for the time 9:50 clock in Fig. 3, the temperature is not yet regulated below a value C (t) - δT, because not enough time has passed since the last instruction, a new forecast over the desired temperature is created and the newly calculated desired temperature is calculated at 21 ° C. This process is shown greatly simplified in the present example. For a concrete calculation of the new desired temperature, a calculation according to the least squares method can of course be used advantageously.

As is further apparent from FIG. 3, at the time 9:50 o'clock each user instruction is up-regulated to the new desired temperature of 21 ° C.

According to another embodiment, not shown the temperature in case of user input not on the Desired temperature C (t) raised, but only by one value δT increased. In such an embodiment would u. U., um to raise the heating back to the desired temperature, more once an instruction is entered.

Such an automatic control of the heating is also advantageous in the saving of heating costs used.

The climate of the invention By itself, the system is fully capable of one to carry out efficient and user-friendly regulation.

However, it is of course also possible that invented The system according to the invention together with conventional systems to use, so that may optionally on one of the systems can be used. Is z. For example, that on certain days or at a certain time a completely  Unusual occupancy or use of a room is pending, that can inventive system temporarily turned off who the.

Claims (14)

1. Programmable climate system for at least one room in which the probability of presence of at least one user predictable and depending on the predicted probability of presence Be a system state of the system is adjustable, characterized in that a presence probability representing the presence function as a combination, in particular linear combination, a means Weighting coefficients weighted set of basis functions, in particular Fourier basis functions, is calculable. 2. Air conditioning system according to claim 1, characterized in that the weighting coefficients via an alignment with a Presence function, which already eingge in the system Describes bene presence inputs, are determinable. 3. Air conditioning system according to claim 2, characterized in that used the approximation methods of regression analysis. 4. Air conditioning system according to claim 2, characterized in that used the approximation methods of fuzzy logic. 5. Air conditioning system according to claim 2, characterized in that the alignment with neural networks works.   6. Air conditioning system according to claim 2, characterized in that used the approximation varianzreduzierende method. 7. Air conditioning system according to one of claims 1 to 3 and 5 or 6, characterized in that the determined presence Probability with one over a Threshold adjustable threshold is compared, in Abhän from falling below or exceeding the limit Threshold determines the operating state of the system becomes. 8. Air conditioning system according to claim 6, characterized in that the threshold between 0 and 1 continuously adjustable is. 9. Air conditioning system according to one of the preceding claims, since characterized in that during a start-up phase of Be user presence or absence instructions in the System are entered. 10. Air conditioning system according to claim 9, characterized that the presence or absence instructions on the Threshold value can be entered. 11. Air conditioning system according to one of claims 9 or 10, characterized characterized in that "warmer" statements or "colder" - Instructions can be entered into the system, causing the Sy stem both attendance instructions and instructions with respect to a desired temperature be presented.   12. Air conditioning system according to one of claims 9 to 11, characterized in that the instructions can be entered via keys are. 13. Air conditioning system according to one of the preceding claims, since characterized in that temporally younger user inputs in determining the probability of presence stronger ker are weightable. 14. Method for controlling a climate system for little at least one room where the presence certificate at least one user predicts and in Dependence on the predicted probability of presence a system operating state is set becomes, characterized, that one representing the probability of presence Presence function as combination, especially linear combination, one weighting coefficient gewich set of basic functions, in particular Fourier Basic functions, is calculated.