JP2006029607A - Heat load predicting device and method for air conditioning heat source facility - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a heat load predicting device predicting a precooling/preheating time for carrying out a peak cut while suppressing wasteful consumption of energy. <P>SOLUTION: The device is provided with a storing part 10 storing past record data before an objective time zone relevant to a heat load of the air conditioning heat source facility, building external environment information, and/or building internal environment information, an acquiring part 12 acquiring prediction data of the objective time zone relevant to the building external environment information and/or the building internal environment information, a storing part 18 storing heat load prediction models coordinated with each of a plurality of precooling or preheating time candidates represented by building external environment variables and/or building internal environment variables, and a prediction part 14 predicting a heat load of the objective time zone on the basis of the past record data, the prediction data, and the heat load prediction models in regard to a precooling/preheating time candidate. In the prediction part, if a peak of the predicted heat load exceeds a threshold, the heat load prediction is carried out again by a more larger precooling/preheating time candidate. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an apparatus and a method for predicting a heat load of an air conditioning heat source facility used for a building such as a building, particularly an air conditioning heat source facility that is stopped at night and started in the morning.

  In recent years, in order to efficiently operate the air conditioning heat source equipment, for example, the heat load to be processed by the equipment for 24 hours every hour on the next day is predicted, and the operation plan of the equipment (start / stop and air conditioning) is determined according to the predicted load. Output setting) is being made. The target time zone for heat load prediction is the time during which air conditioning is on (building use time).

  In the case of a building in which the air conditioning heat source facility is stopped at night, it is known that the peak of the heat load appears around noon in the case of cooling and immediately after the start of air conditioning in the case of heating. When the energy of the air conditioning heat source facility is covered by electricity, it is necessary to suppress the peak of the heat load to a certain threshold value (peak cut), thereby preventing the power used from exceeding the contract power.

  In order to perform peak cut, prior to normal operation (operation to maintain the indoor set temperature (for example, 26 ° C. for the air conditioning heat source equipment)), the air conditioning heat source equipment is activated and precooled (preliminary cooling) or preheated (preliminary) A method of performing heating is known. In the pre-cooling or pre-heating operation, the heat load of the air-conditioning heat source facility is reduced after the normal operation is started by previously cooling or warming the wall or ceiling, and as a result, the peak of the heat load is made to be below the threshold value.

Patent Document 1 discloses a method for predicting precooling or preheating time necessary for a room temperature or discomfort index to reach a target value at a certain time. Patent Document 1 does not relate to a method for predicting precooling or preheating time necessary for performing peak cutting.
Japanese Patent Laid-Open No. 9-229449

  However, if the pre-cooling or pre-heating operation is performed for a long time, the heat will eventually escape and energy will be wasted.

  Then, an object of this invention is to provide the thermal load prediction apparatus and method which can estimate the pre-cooling or pre-heating time for performing peak cut, suppressing the wasteful consumption of energy.

In order to achieve the above object, a thermal load predicting apparatus according to the present invention includes:
In the device for predicting the heat load of the target time zone of the air conditioning heat source equipment used for the building,
An actual data storage unit for storing actual data prior to the target time period regarding the heat load of the air-conditioning heat source facility, information on the building external environment, and / or information on the internal environment of the building;
A prediction data acquisition unit for acquiring prediction data of a target time zone for information on the external environment of the building and / or information on the internal environment of the building;
A thermal load prediction model storage unit that stores a thermal load prediction model that is represented by a thermal load, a variable related to the building external environment, and / or a variable related to the internal environment of the building and that is associated with each of a plurality of precooling or preheating time candidates;
With respect to a certain precooling or preheating time candidate, the target time based on the actual data stored in the actual data storage unit, the predicted data acquired by the predicted data acquisition unit, and the thermal load prediction model stored in the thermal load prediction model storage unit A heat load prediction unit for predicting the heat load of the belt,
With
The thermal load predicting unit is configured to perform a second precooling or preheating larger than the first precooling or preheating time candidate if the peak of the heat load predicted by the first precooling or preheating time candidate exceeds a predetermined threshold. The heat load prediction is performed again with a time candidate.

The thermal load prediction method according to the present invention is:
In the method for predicting the heat load of the target time zone of the air conditioning heat source equipment used for the building,
Acquire actual data before the target time period about the heat load of air-conditioning heat source equipment, information about the building external environment, and / or information about the building internal environment,
Get forecast data for the target time zone for information about the external environment of the building and / or information about the internal environment of the building,
Prepare a heat load prediction model that is expressed in terms of heat load, variables related to the building's external environment, and / or variables related to the building's internal environment, and is associated with each of a plurality of precooling or preheating time candidates,
For the first candidate for pre-cooling or pre-heating time, predict the heat load of the target time zone based on the actual data, prediction data, and heat load prediction model,
If the predicted heat load peak for the first precooling or preheating time candidate exceeds a predetermined threshold, the second precooling or preheating time candidate that is larger than the first precooling or preheating time candidate is reheated. Predict,
It is characterized by that.

  According to the present invention, the thermal load is predicted for a certain pre-cooling or pre-heating time, and if the peak exceeds the threshold, the pre-cooling or pre-heating time is further increased to predict the thermal load, so that the peak of the thermal load is determined. The pre-cooling or pre-heating time that is suppressed below the threshold can be obtained. Therefore, wasteful consumption of energy due to pre-cooling or pre-heating operation can be suppressed at the same time as peak cutting is performed.

  Embodiments according to the present invention will be described below with reference to the accompanying drawings.

Embodiment 1 FIG.
FIG. 1 shows a control system for air-conditioning heat source equipment used in a building, which is provided with a first embodiment of a thermal load prediction device according to the present invention. In the control system 2, an air conditioning heat source facility (for example, a cold / hot water generator, an electric refrigerator, etc.) 8 is optimally operated by the facility controller 6 based on the heat load predicted by the heat load prediction device 4. When performing the pre-cooling operation or the pre-heating operation (hereinafter simply referred to as pre-cooling pre-heating operation), the output of the air conditioning heat source facility 8 is constant (for example, 70% of the rated output).

  The thermal load prediction device 4 includes a performance data storage unit 10, a prediction data acquisition unit 12, a thermal load prediction unit 14, a prediction result evaluation unit 16, and a thermal load prediction model storage unit 18.

  The actual data storage unit 10 is for storing the actual heat load before the target time zone of the air conditioning heat source facility 8 as actual data. For this purpose, the equipment controller 6 includes a sensor (not shown) that measures the thermal load of the air conditioning heat source equipment 8.

  The prediction data acquisition unit 12 is for acquiring information related to the building external environment every hour in the target time zone of the thermal load prediction, for example, predicted values of the outside air temperature and the outside air humidity. These predicted values are obtained from predicted values of maximum temperature, minimum temperature, weather (sunny, rain, etc.) obtained directly from a weather information service company via the Internet or obtained from a weather forecast, for example.

  The thermal load prediction unit 14 includes actual data stored in the actual data storage unit 10, predicted data acquired by the predicted data acquisition unit 12, and precooling time or preheating time (preheating time output from the prediction result evaluation unit 16 as described later ( In the following, based on the candidate of the precooling preheating time for simplicity, the heat load prediction is performed with reference to a heat load prediction model (described later) stored in the heat load prediction model storage unit 18. . The thermal load predicted by the thermal load prediction unit 14 is output to the prediction result evaluation unit 16.

  When the thermal load predicted by the thermal load prediction unit 14 is input, the prediction result evaluation unit 16 receives the peak value of the thermal load based on the precooling preheating time candidates (including 0 hours) used for the thermal load prediction. Is for determining whether or not exceeds a preset threshold value. The prediction result evaluation unit 16 outputs a new precooling preheating time candidate to the thermal load prediction unit 14 when the threshold is exceeded, and the thermal load prediction including the precooling preheating time when the threshold is not exceeded. Is output to the equipment controller 6. The prediction result evaluation unit 16 prepares values for a plurality of stages in advance in addition to 0 hours, such as candidates for precooling preheating time, for example, three stages of 1 hour, 2 hours, and 3 hours. The candidate precooling preheating time and its time interval (1 hour in the above example) do not limit the present invention.

  The thermal load prediction model storage unit 18 stores a different thermal load prediction model for each candidate of the precooling preheating time selected by the prediction result evaluation unit 16.

と表される。ここで、
ｔ：予測対象時刻
ｋ：予冷予熱時間
ｙ_ｋ，ｔ：時刻ｔの熱負荷
Ｎ：熱負荷に影響を与える要因の数
ｘ_{ｋ，ｔ，ｉ}：熱負荷に影響を与える要因の時刻ｔの値
ａ：パラメータ
である。式（１）は、候補となる予冷予熱時間毎に与えられる。パラメータａ_{ｋ，ｔ，０}は、時刻（ｔ−１）の熱負荷が時刻ｔの熱負荷に影響を与える影響の度合いを表し、パラメータａ_{ｋ，ｔ，ｉ}は、時刻ｔの各要因の値が時刻ｔの熱負荷に与える影響の度合いを表す。パラメータａは、時刻により異なる値をとり、例えば別の建物に本実施形態に係る熱負荷予測装置４を適用するなどして得た実績データに基づいて予め同定しておく。
熱負荷に影響を与える要因として外気温度と外気湿度を選択した場合、式（１）は、

となる。ここで、ｘ_{ｋ，ｔ，１}は時刻ｔの外気温度、ｘ_{ｋ，ｔ，２}は時刻ｔの外気湿度である。 When the PLMS (Periodic Least Mean Square) method, which is one of the statistical methods of air conditioning heat load prediction, is used as an example of the prediction method of the heat load prediction unit 14, the heat load prediction model is

It is expressed. here,
t: Prediction target time k: Precooling preheating time y _{k, t} : Thermal load at time t N: Number of factors affecting the thermal load x _{k, t, i} : Value of time t of factors affecting the thermal load a: a parameter. Equation (1) is given for each candidate precooling preheating time. Parameter a _{k, t, 0} represents the degree of influence that the thermal load at time (t-1) affects the thermal load at time t, and parameter a _{k, t, i} is the value of each factor at time t. Represents the degree of influence on the thermal load at time t. The parameter a takes a different value depending on the time and is identified in advance based on actual data obtained by applying the thermal load prediction device 4 according to the present embodiment to another building, for example.
When outside air temperature and outside air humidity are selected as factors that affect the heat load, equation (1) becomes

It becomes. Here, x _{k, t, 1} is the outside air temperature at time t, and x _{k, t, 2} is the outside air humidity at time t.

  (N-1) The case where the heat load prediction from 22:00 on day 2 to 21:00 on day n is performed is taken as an example. First, the predicted data of the thermal load at 22:00 on (n-1) day is stored in the actual data storage unit 10 and the actual data of the thermal load at 21:00 on (n-1) day and the predicted data acquisition unit 12 are stored. (2) is calculated from the prediction data of the outside air temperature and the outside air humidity at 22:00 on the (n-1) day acquired in step (2). Thereafter, by applying the calculated predicted value of the heat load one hour before and the predicted values of the outside air temperature and the outside humidity at the prediction target time to the formula (2), the heat load up to 21:00 on the nth day Are sequentially calculated.

  Instead of or in addition to the outside temperature and outside humidity, factors relating to the external environment of the building may be used as factors affecting the heat load, or factors relating to the internal environment of the building (for example, indoor temperature) may be used. Good. Thus, the heat load prediction model is expressed by the heat load of the air conditioning heat source facility, the variable related to the building external environment, and / or the variable related to the building internal environment.

  As another heat load prediction model, as shown in FIG. 2, a neural network model considering the precooling preheating time may be used. The model shown in the figure performs a heat load prediction from 22:00 on (n-1) day to 21:00 on n day. Specifically, for each neuron in the input layer, the same time of the previous day, that is, actual data of heat load from 22:00 on the (n-2) day to 21:00 on the (n-1) day, and n days By inputting the prediction data of the maximum temperature and the minimum temperature and the candidate for the precooling preheating time, the thermal load of each neuron in the output layer from 22:00 on (n−1) day to 21:00 on n day Get the predicted value. The weight between the neurons is preliminarily learned by performing learning using the error back propagation method or the like using actual data obtained by applying the thermal load prediction device 4 according to the present embodiment to another building, for example. Identify it. The weight varies depending on the time.

  In FIG. 2, the predicted maximum temperature and the predicted minimum temperature are input to the input layer in addition to the heat load at the same time on the previous day, but in addition to or instead of these, the target time zone (one Predicted data of outside air temperature and outside air humidity every hour) may be used. In addition to the outside air temperature and the outside air humidity, information related to the building external environment such as the amount of solar radiation and information related to the building internal environment such as the indoor temperature may be used as the prediction data.

  An example of the prediction process of the precooling preheating time of the thermal load prediction apparatus 4 having such a configuration will be described with reference to the flowchart of FIG. 3 together with FIG. First, in step S1, the prediction result evaluation unit 16 selects 0 hour as a candidate for the precooling preheating time (in other words, sets the precooling preheating time to 0 hour). In step S2, the prediction result evaluation unit 16 outputs this set value to the heat load prediction unit 14, and the heat load prediction unit 14 stores the heat stored in the heat load prediction model storage unit 18 under the condition of the precooling preheating time 0. Thermal load prediction is performed using a load prediction model. In step S <b> 3, the thermal load prediction unit 14 outputs the obtained predicted value of the thermal load (FIG. 4) to the prediction result evaluation unit 16, and the prediction result evaluation unit 16 performs the heat predicted by the thermal load prediction unit 14. When the load is input, it is determined whether or not the peak value of the thermal load exceeds a preset threshold value.

  When the predicted peak value of the heat load exceeds the threshold value in step S3 [see FIG. 4 (a)], the process proceeds to step S4, and the prediction result evaluation unit 16 determines whether the precooling preheating time exceeds the upper limit. Determine whether. Since the precooling preheating operation is performed in a time zone other than the normal operation, it is necessary to set the maximum possible precooling preheating time due to the point that the start time of the precooling preheating operation is after the normal operation stop time and the restrictions on facility operation Considering this point, the user (for example, a building manager) predetermines the upper limit of the precooling preheating time. When the precooling preheating time does not exceed the upper limit, the process proceeds to step S5, and the prediction result evaluating unit 16 increases the precooling preheating time set in step S1 by one step (for example, 1 hour). This increase is not limited to one hour. Then, it returns to step S2, the prediction result evaluation part 16 outputs the newly set pre-cooling preheating time to the heat load prediction part 14, and the heat load prediction part 14 performs heat load prediction. If the peak value of the heat load exceeds the threshold and the precooling preheating time has not reached the upper limit, the heat load prediction is performed while increasing the precooling preheating time (steps S2 to S5). In step S3, when the predicted peak value of the heat load becomes equal to or less than the threshold value (see FIG. 4B), the flow ends. Thereafter, the prediction result evaluation unit 16 outputs a heat load prediction including a precooling preheating time to the equipment controller 6. Alternatively or in combination with this, a heat load prediction including a precooling preheating time may be provided to a building manager or the like via display means such as a monitor. The flow is also terminated when the precooling preheating time exceeds the upper limit in step S4. In this case, for example, the fact that the pre-cooling preheating time has reached the upper limit is provided to the building manager or the like via the display means such as a monitor together with the upper limit.

  As described above, according to the present embodiment, the shortest precooling preheating time can be predicted while keeping the peak of the air conditioning heat load below the threshold value. Therefore, the power consumption at the peak time and the energy required for precooling and preheating can be predicted. Can be reduced.

と表される。ここで、
ｔ：予測対象時刻
Ｌ：方位
ｑ_AC,t：時刻ｔの空調熱負荷
θ_0,L,t：方位Ｌ時刻ｔの等価外気温
［＝外気温度＋（日射吸収率×日射量）／表面熱伝達率］
θ_R,t：時刻ｔの室内温度
Ｊ_L,t：方位Ｌ、時刻ｔの日射量
ζ_R,t：時刻ｔの室内発生熱
ζ_o,t：時刻ｔの外気導入量
Δｈ_t：時刻ｔの室内外エンタルピー差
ａ：パラメータ
Ｎp、Ｎq、Ｎr：次数（過去何時間までの影響を考慮するかを表す次数）
である。パラメータａおよび次数Ｎp、Ｎq、Ｎrは、例えば別の建物に本実施形態に係る熱負荷予測装置４Ａを適用するなどして得た実績データに基づいて予め同定しておく。 Embodiment 2. FIG.
FIG. 5 shows a control system provided with Embodiment 2 of the thermal load prediction apparatus according to the present invention. Hereinafter, the same or similar components as those of the first embodiment are represented by the same reference numerals or the same reference numerals with appropriate subscripts added thereto. In the thermal load prediction device 4A, a physical model that takes into account the heat storage load of the building frame is stored in the thermal load prediction model storage unit 18A. In this embodiment, as a physical model, it is described in “Air-conditioning load prediction for rational operation management of a heat storage tank” (Hirori Yoshida, Architectural Institute of Japan Planning Series, No. 495, pages 77-83, 1997). Use what you have. The heat load prediction model is

It is expressed. here,
t: prediction target time L: direction q _{AC, t} : air-conditioning heat load θ _{0, L, t at} time t: equivalent outside temperature at direction L time t [= outside air temperature + (solar absorption rate × insolation amount) / surface heat Transmission rate]
θ _{R, t} : Indoor temperature at time t J _{L, t} : Direction L, Solar radiation amount at time t ζ _{R, t} : Indoor heat generated at time t ζ _{o, t} : Outside air introduction amount at time t Δh _t : Time t Indoor / outdoor enthalpy difference a: Parameters Np, Nq, Nr: Orders (orders representing how many hours in the past are considered)
It is. The parameter a and the orders Np, Nq, and Nr are identified in advance based on actual data obtained by applying the thermal load prediction device 4A according to the present embodiment to another building, for example.

In the present embodiment, the actual data storage unit 10 </ b> A stores actual solar radiation amount, outdoor temperature, outdoor air humidity, and actual indoor temperature before the target time period as actual building temperature information as building external environment information. Remember as. The predicted data acquisition unit 12A acquires predicted data of the amount of solar radiation, the outside air temperature, and the outside air humidity in the target time zone as building external environment information. The predicted data acquisition unit 12A is also configured to acquire the set temperature of air conditioning during normal operation from the schedule storage unit (not shown) of the facility as the building internal environment information. This schedule is determined by the building manager in consideration of the time zone, day of the week, season, and the like. The thermal load prediction device 4A also includes a fixed data storage unit 20 for storing fixed data required when the thermal load prediction unit 14 predicts the thermal load. As fixed data, for example,
・ Direction, ground surface reflectance, solar radiation absorption rate, surface heat transfer coefficient, window area, wall area other than windows for each external wall of the building ・ Heat generated indoors at each time ・ Amount of outside air introduced into the building at each time (During air conditioning operation)
Is mentioned. Various data on the outer wall of the building is prepared based on the building design data. The time variation pattern of the heat generated in the room is data that cannot be measured normally, so prepare it with reference to the settings at the time of building design. As the time variation pattern of the outside air introduction amount, a pattern determined as a schedule during the air conditioning operation is prepared.

  For example, the case where the heat load prediction from 22:00 on the (n-1) day to 21:00 on the n day is performed with the pre-cooling preheating operation as one hour will be described.

と予測される。この予測室内温度は、２３：００以降の熱負荷を予測するのに必要となる。等価外気温θ_0,L,ｔは、予測データ取得部１２Ａで取得した日射量、外気温度、固定データ記憶部２０に記憶された建物の外壁に関するデータを得ることで求められる。Ｎｐが１以上の場合には、（ｎ−１）日の２１：００以前の等価外気温θ_{0,L,ｔ−ｐ}（ｐ＝１，．．．，Ｎｐ）が必要となるが、これらは、固定データ記憶部２０に記憶された建物の外壁に関するデータと、実績データ記憶部１０Ａに記憶された日射量、外気温度とから求まる。Ｎｑが１以上の場合には、（ｎ−１）日の２１：００以前の室内温度θ_R,t−ｑ（ｑ＝１，．．．，Ｎｑ）が必要となるが、実績データ記憶部１０Ａから得られる。日射量Ｊ_L,tは、予測データ取得部１２Ａから求まる。Ｎｒが１以上の場合には、（ｎ−１）日の２１：００以前の日射量Ｊ_Ｌ,t−ｒ（ｒ＝１，．．．，Ｎｒ）が必要となるが、実績データ記憶部１０Ａから得られる。（ｎ−１）日の２２：００の室内発生熱ζ_R,tは、固定データ記憶部２０から得られる。したがって、（パラメータａは予め同定されているので）（ｎ−１）日の２２：００の室内温度θ_R,tが予測値として求められる。 (N-1) When the air conditioning is stopped at 22:00 on the day, the thermal load q _{AC, t on} the left side of Equation (3) is zero. Furthermore, since there is no outside air intake, the fifth term on the right side is also zero. Therefore, when air conditioning is stopped, the room temperature is

It is predicted. This predicted indoor temperature is necessary to predict the heat load after 23:00. The equivalent outside air temperature θ _{0, L, t} is obtained by obtaining data relating to the amount of solar radiation, the outside air temperature acquired by the prediction data acquisition unit 12A, and the outer wall of the building stored in the fixed data storage unit 20. When Np is 1 or more, the equivalent outside air temperature θ _{0, L, tp} (p = 1,..., Np) before 21:00 on (n−1) day is required. Is obtained from the data relating to the outer wall of the building stored in the fixed data storage unit 20, the amount of solar radiation and the outside air temperature stored in the result data storage unit 10A. When Nq is 1 or more, the room temperature θ _{R, t-q} (q = 1,..., Nq) before 21:00 on (n−1) day is required. From 10A. The amount of solar radiation J _{L, t} is obtained from the prediction data acquisition unit 12A. When Nr is 1 or more, the solar radiation amount J _{L, tr} (r = 1,..., Nr) before 21:00 on the (n−1) day is required, but the actual data storage unit From 10A. (N-1) The indoor generated heat ζ _{R, t} at 22:00 on the day is obtained from the fixed data storage unit 20. Therefore, (because the parameter a is identified in advance), the room temperature θ _{R, t} at 22:00 on (n−1) day is obtained as a predicted value.

(N-1) In the case of normal operation at 22:00 on the day, the thermal load is calculated using Equation (3). In Equation (3), the equivalent outside air temperature (first term on the right side), the amount of solar radiation (third term), and the heat generated in the room (fourth term) are obtained in the same manner as when air conditioning is stopped. (N-1) The room temperature θ _{R, t} at 22:00 on the day is the set temperature of the air conditioning acquired by the prediction data acquisition 12A. When Nq is 1 or more, the room temperature θ _{R, t-q} (q = 1,..., Nq) before 21:00 on (n−1) day is required. From 10A. The outside air introduction amount ζ _{o, t} is obtained from the fixed data storage unit 20. Indoor and outdoor enthalpy difference Delta] h _t is determined using the set temperature of the predicted value and the room temperature 22:00 outside air temperature and the ambient humidity of the (n-1) day. Therefore, the predicted value of the heat load of 22:00 on (n-1) days is obtained using these values.

  After 23:00, by repeating the same calculation, the heat load and the room temperature at each time before the precooling preheating operation can be obtained. However, when the orders Np, Nq, and Nr are 1 or more, after 22:00 acquired by the prediction data acquisition unit 12A together with the actual data before 21:00 stored in the actual data storage unit 10A or alone. It is necessary to use the forecast data.

と予測される。例えば、朝の空調の通常運転開始時刻を８：００とすると、式（５）において、７：００の室内温度は、右辺の第１項から第４項については式（４）を用いて室内温度を予測する場合と同じようにして得られる。 During the precooling preheating operation, the thermal load q _{AC, t} is constant at a preset air conditioning output value q ₀ (for example, 70% of the rated output). In formula (3), if there is no outside air intake, the room temperature is

It is predicted. For example, assuming that the normal operation start time of morning air conditioning is 8:00, in the equation (5), the indoor temperature at 7:00 is calculated using the equation (4) for the first to fourth terms on the right side. It is obtained in the same way as when the temperature is predicted.

  In this manner, the heat load is predicted when the candidate for the precooling preheating time is 1 hour. The heat load is predicted in the same manner for other candidates for the precooling preheating time.

Embodiment 3 FIG.
In the first embodiment, the air conditioning output during the precooling preheating operation is set to a constant value (for example, 70% of the rated output) when predicting the precooling preheating time during which the peak of the heat load is lowered below the threshold value. The air conditioning output during the precooling preheating operation can also be changed, and the precooling preheating time and the air conditioning output suitable for lowering the peak of the heat load below the threshold are obtained. The thermal load prediction apparatus according to the present invention has substantially the same configuration as the thermal load prediction apparatus 4 of FIG. 1 and will be described below with reference to FIG.

  In the thermal load prediction device 4 according to the present embodiment, the prediction result evaluation unit 16 includes a plurality of pre-cooling preheating time candidates and air conditioning output candidates, for example, three stages of 60%, 70%, and 80% of the rated output. The value of the stage is prepared.

と表される。ここで、ｐは空調出力であり、その他の記号は式（２）と同様である。式（６）は、予冷予熱時間ｔと空調出力ｐの組み合わせ候補毎に与えられる。なお、熱負荷に影響を与える要因として外気温度や外気湿度の代わりにまたはこれらに加えて他の建物外部環境に関する要因を用いてもよいし、建物内部環境に関する要因（例えば、室内温度）を用いてもよい。 When the PLMS method is used as an example of the prediction method of the thermal load prediction unit 14, the thermal load prediction model stored in the thermal load prediction model storage unit 18 is:

It is expressed. Here, p is an air-conditioning output, and other symbols are the same as in equation (2). Equation (6) is given for each combination candidate of the precooling preheating time t and the air conditioning output p. In addition to or in addition to the outside temperature and outside humidity, factors relating to the external environment of the building may be used as factors affecting the heat load, or factors relating to the internal environment of the building (for example, indoor temperature) may be used. May be.

を用いてもよい。ここで、ｘ_{ｋ，ｔ，３}は、予冷予熱時間ｋのときの時刻ｔの空調出力である。空調出力ｘ_{ｋ，ｔ，３}は、空調停止時の時刻ｔには０で、空調運転時（通常運転の場合も含む）の時刻ｔには常に一定値（予冷予熱運転時の空調出力値）とする。その他の記号は式（２）と同じである。式（７）は、予冷予熱運転時の空調出力を、熱負荷に影響を与える要因として考慮することを意味する。 Instead of equation (6)

May be used. Here, x _{k, t, 3} is the air conditioning output at time t when the pre-cooling preheating time k. The air conditioning output x _{k, t, 3} is 0 at the time t when the air conditioning is stopped, and is always a constant value (the air conditioning output value during the precooling preheating operation) at the time t during the air conditioning operation (including the normal operation). And Other symbols are the same as those in formula (2). Equation (7) means that the air conditioning output during the precooling preheating operation is considered as a factor affecting the heat load.

  Alternatively, a neural network model may be used instead. In this case, the neural network is the same as that described in the first embodiment, except that a neural network in which a neuron that inputs an air conditioning output at the time of precooling preheating operation is added to the input layer of the neural network in FIG.

You may use the physical model which considered the heat storage load of the building frame demonstrated in Embodiment 2. FIG. In this case, in applying equation (5) to determine the room temperature during precooling preheating operation, except that candidate values predetermined stepwise to the air conditioning output q ₀ is assigned it is in the second embodiment It is the same as explained.

  Hereinafter, an example of the prediction process of the thermal load prediction device 4A will be described with reference to the flowcharts of FIGS. 1 and 6. First, in step S61, the prediction result evaluation unit 16 selects 0 hour as a precooling preheating time candidate and the maximum value as an air conditioning output candidate during precooling preheating operation (in other words, the precooling preheating time is 0 hour and the air conditioning output is selected). Set to maximum). In step S62, the prediction result evaluation unit 16 outputs these set values to the heat load prediction unit 14, and the heat load prediction unit 14 stores the set values in the heat load prediction model storage unit 18 under the conditions of the precooling preheating time 0 and the maximum air conditioning output. The thermal load prediction is performed using the stored thermal load prediction model (note that the air conditioning output does not affect the calculation because the precooling preheating time is zero at this time). In step S63, the thermal load prediction unit 14 outputs the obtained predicted value of the thermal load to the prediction result evaluation unit 16, and the prediction result evaluation unit 16 receives the thermal load predicted by the thermal load prediction unit 14. Then, it is determined whether or not the peak value of the thermal load exceeds a preset threshold value.

  If the peak value of the heat load exceeds the threshold value in step S63, the process proceeds to step S64, and the prediction result evaluation unit 16 determines whether the air conditioning output has reached the upper limit. If the air conditioning output has reached the upper limit, the process proceeds to step S65, and the prediction result evaluation unit 16 determines whether or not the precooling preheating time exceeds the upper limit. When the precooling preheating time does not exceed the upper limit, the process proceeds to step S66, and the prediction result evaluating unit 16 increases the precooling preheating time by one step (for example, one hour) and sets the air conditioning output to the minimum value. Then, it returns to step S62, the prediction result evaluation part 16 outputs the newly set pre-cooling preheating time and air-conditioning output to the heat load prediction part 14, and the heat load prediction part 14 performs heat load prediction.

  If the air conditioning output does not reach the upper limit in step S64, the process proceeds to step S67, and the prediction result evaluation unit 16 increases the air conditioning output by one step. Thereafter, the flow returns to step S62.

  In step S63, if the peak value of the heat load exceeds the threshold and the precooling preheating time has not reached the upper limit or the air conditioning output during the precooling preheating operation has not reached the upper limit, the precooling preheating time or the air conditioning output is set. The heat load is predicted while increasing (steps S62 to S67). In step S63, when the predicted peak value of the thermal load is equal to or less than the threshold value, the flow ends. Thereafter, the prediction result evaluation unit 16 outputs a heat load prediction including a precooling preheating time and an air conditioning output to the equipment controller 6. Alternatively or in combination with this, a heat load prediction including a precooling preheating time and an air conditioning output may be provided to a building manager or the like via display means such as a monitor. In step S65, the flow is also terminated when the precooling preheating time exceeds the upper limit.

  Thus, according to this embodiment, in addition to the precooling preheating time, the air conditioning output during the precooling preheating operation can be determined. Therefore, the air conditioning output is as small as possible to enable peak cut even in the same precooling preheating time. As a result, energy consumption due to the precooling preheating operation can be further suppressed.

Embodiment 4 FIG.
FIG. 7 shows a control system including a fourth embodiment of the thermal load prediction apparatus according to the present invention. The thermal load prediction device 4B according to the present embodiment includes a thermal load prediction model learning unit 30 in addition to the configuration of the first embodiment, and the results accumulated in the result data storage unit 10B by operating the control system 2B. Based on the data, the heat load prediction model stored in the heat load prediction model storage unit 18B is learned. This is to improve the prediction accuracy of the thermal load prediction model by correcting the parameters and weights of the thermal load prediction model to match the actual results. In the record data storage unit 10B, the actual heat load, outside air temperature, and outside air humidity of the air conditioning heat source facility 8 are accumulated every precooling preheating time.

を学習モデルとして用いる。ここで、ｅ_ｋ，ｔは予測誤差である。式（８）は予冷予熱時間の候補毎に与えられる。 When the above-described PLMS method is used as the prediction method of the thermal load prediction unit 14, the thermal load prediction model learning unit 30

Is used as a learning model. Here, e _{k, t} is a prediction error. Equation (8) is given for each candidate for the precooling preheating time.

During learning, x _{k, t, 1} and x _{k, t, 2} include the outside air temperature and the outside air humidity at each time t when the precooling preheating operation is performed with the precooling preheating time k stored in the record data storage unit 10B. The actual values y _{k, t} , y _{k, t−1} are the thermal loads at times t and t−1 when the precooling preheating operation is performed with the precooling preheating time k stored in the result data storage unit 10B. Substitute in 8). Parameters a _{k, t, i} (i = 0, 1, 2) that minimize the sum of squares of the prediction errors ek _{, t} obtained in this way are obtained for each time t. Thus, the parameters at each time are obtained for each precooling preheating time.

を学習モデルとして用い、実績データ記憶部１０Ｂに蓄積されたデータを利用して、予冷予熱時間と空調出力の組み合わせ毎に各時刻のパラメータを求める。 When considering both the precooling preheating time and the air conditioning output during the precooling preheating operation, the actual data storage unit 10B stores the heat load, the outside air temperature, and the outside air humidity for each combination of the precooling preheating time and the air conditioning output during the precooling preheating operation. Accumulate performance data. The heat load prediction model learning unit 30 corresponds to the equation (6).

Is used as a learning model, and parameters at each time are obtained for each combination of precooling preheating time and air conditioning output using data accumulated in the result data storage unit 10B.

を用いてもよい。但し、式（１０）のｘ_{ｋ，ｔ，３}には、空調停止時の時刻ｔでは０を代入し、空調運転時（通常運転の場合も含む）の時刻ｔでは常に予冷予熱運転時の空調出力（一定値）を、実績データ記憶部１０Ｂから得て代入する。 Instead, it corresponds to equation (7)

May be used. However, 0 is substituted for xk _{, t, 3} in the expression (10) at the time t when the air conditioning is stopped, and the air conditioning during the precooling preheating operation is always performed at the time t during the air conditioning operation (including the normal operation). An output (a constant value) is obtained from the record data storage unit 10B and substituted.

  Alternatively, a neural network model may be used instead. In this case, first, the actual heat load, the highest temperature, and the lowest temperature are stored in the result data storage unit 10B in association with each precooling preheating time. Next, in the input layer of the neural network of FIG. 2, the actual data of the heat load from 22:00 on (n-2) day to 21:00 on (n-1) day, the highest temperature and the lowest temperature on day n And the actual data of the pre-cooling preheating time for n days are input, and the predicted value of the thermal load from 22:00 on the (n-1) day to 21:00 on the n day is output from the output layer. Then, the difference between these predicted values and the actual data of the heat load from 22:00 on the (n-1) day to 21:00 on the n day is obtained as a prediction error, and the sum of squares of the prediction error is minimized. Next, the weights between neurons are corrected using an error back-propagation method or the like.

  According to the present embodiment, the accuracy of the thermal load prediction is improved by reflecting the results in the thermal load prediction model, and as a result, a more accurate precooling preheating time and an air conditioning output value during the precooling preheating operation can be obtained. it can.

Embodiment 5. FIG.
FIG. 8 shows a control system including a fifth embodiment of the thermal load predicting apparatus according to the present invention. The thermal load prediction device 4C according to the present embodiment includes a fixed data storage unit 20C and a thermal load prediction model storage unit 18C that stores a physical model that takes into account the heat storage load of the building frame, as in the second embodiment. As in the fourth embodiment, the thermal load prediction model learning unit 30C is provided.

を学習モデルとして用いる。ここで、ｅ_ｔは予測誤差である。 The thermal load prediction model learning unit 30C

Is used as a learning model. Here, _et is a prediction error.

実績データのベクトルに関する式

を用いて、

と表すことができる。 Expression (11) is an expression related to a parameter vector.

Formula for vector of actual data

Using,

It can be expressed as.

Each component of the actual data vector phi _t of formula (13) is equivalent outside temperature, room temperature, amount of sunlight, indoor heat generated, the outside air introduction amount is indoor and outdoor enthalpy difference. Among these, those that can be directly measured are stored in advance in the result data storage unit 10C. For data that cannot be directly measured (calculated by calculation), data that is the basis of calculation is stored in the actual data storage unit 10C. The thermal load prediction model learning unit 30C performs necessary calculations based on the calculation data and the fixed data stored in the fixed data storage unit 20C.

ここで、

である。 At the time of learning, the thermal load prediction model learning unit 30C estimates the parameter vector A by sequentially applying the following formulas based on the formula (14).

here,

It is.

Incidentally, the order Np, Nq, with respect to the Nr, may be used those already identified, may be calculated as a value the sum of squares of prediction errors e _t is minimized.

The block diagram which shows the control system of air-conditioning heat-source equipment provided with Embodiment 1 of the thermal load prediction apparatus which concerns on this invention. The figure which shows the neural network which is an example of the thermal load prediction model memorize | stored in the thermal load prediction model memory | storage part of FIG. The flowchart which shows an example of the prediction process by the thermal load prediction apparatus of FIG. When performing a cooling operation, (a) is a graph showing an example of a heat load whose peak exceeds a threshold value, and (b) is a graph showing an example of a heat load where pre-cooling is performed and the peak is set to a threshold value or less. The block diagram which shows the control system of the air-conditioning heat-source equipment provided with Embodiment 2 of the thermal load prediction apparatus which concerns on this invention. The flowchart which shows an example of the prediction process by Embodiment 3 of the thermal load prediction apparatus which concerns on this invention. The block diagram which shows the control system of the air-conditioning heat-source equipment provided with Embodiment 4 of the thermal load prediction apparatus which concerns on this invention. The block diagram which shows the control system of the air-conditioning heat-source equipment provided with Embodiment 5 of the thermal load prediction apparatus which concerns on this invention.

Explanation of symbols

4,4A, 4B, 4C Thermal load prediction device

Claims (5)

In the device for predicting the heat load of the target time zone of the air conditioning heat source equipment used for the building,
An actual data storage unit for storing actual data prior to the target time period regarding the heat load of the air-conditioning heat source facility, information on the building external environment, and / or information on the internal environment of the building;
A prediction data acquisition unit for acquiring prediction data of a target time zone for information on the external environment of the building and / or information on the internal environment of the building;
A thermal load prediction model storage unit that stores a thermal load prediction model that is represented by a thermal load, a variable related to the building external environment, and / or a variable related to the internal environment of the building and that is associated with each of a plurality of precooling or preheating time candidates;
With respect to a certain precooling or preheating time candidate, the target time based on the actual data stored in the actual data storage unit, the predicted data acquired by the predicted data acquisition unit, and the thermal load prediction model stored in the thermal load prediction model storage unit A heat load prediction unit for predicting the heat load of the belt,
With
The thermal load predicting unit is configured to perform a second precooling or preheating larger than the first precooling or preheating time candidate if the peak of the heat load predicted by the first precooling or preheating time candidate exceeds a predetermined threshold. A thermal load prediction apparatus, wherein thermal load prediction is performed again with a time candidate. In the device for predicting the heat load of the target time zone of the air conditioning heat source equipment used for the building,
An actual data storage unit for storing actual data prior to the target time period regarding the heat load of the air-conditioning heat source facility, information on the building external environment, and / or information on the internal environment of the building;
A prediction data acquisition unit for acquiring prediction data of a target time zone for information on the external environment of the building and / or information on the internal environment of the building;
Expressed with variables related to thermal load, building external environment, and / or building internal environment, and supports each combination of one of multiple precooling or preheating time candidates and one of multiple air conditioning output candidates during precooling or preheating operation A thermal load prediction model storage unit for storing the attached thermal load prediction model,
With respect to a certain pre-cooling or pre-heating time candidate and a certain air conditioning output candidate, the actual data stored in the actual data storage unit, the predicted data acquired by the predicted data acquisition unit, and the thermal load prediction model stored in the thermal load prediction model storage unit Based on the thermal load prediction unit that predicts the thermal load of the target time zone,
With
The thermal load prediction unit relates to the first precooling or preheating time candidate and the first air conditioning output candidate, the actual data stored in the actual data storage unit, the prediction data acquired by the prediction data acquisition unit, and the thermal load prediction model storage Based on the thermal load prediction model stored in the unit, the thermal load in the target time zone is predicted, and if the predicted thermal load peak exceeds a predetermined threshold, the first precooling or preheating time candidate and The thermal load prediction is performed again with a second air conditioning output candidate larger than the first air conditioning output candidate, or the second precooling or preheating time candidate larger than the first precooling or preheating time candidate and the first air conditioning output. A thermal load prediction device, wherein the thermal load prediction is performed again with a third air conditioning output candidate smaller than the candidate. In the actual data storage unit, actual data before the target time zone is stored for each pre-cooling or pre-heating time,
The apparatus further comprises a thermal load prediction model learning unit that learns the thermal load prediction model stored in the thermal load prediction model storage unit using the actual data before the target time period stored in the actual data storage unit. The thermal load prediction apparatus according to claim 1. The actual data storage unit stores actual data before the target time period for each precooling or preheating time and air conditioning output during precooling or preheating operation.
The apparatus further comprises a thermal load prediction model learning unit that learns the thermal load prediction model stored in the thermal load prediction model storage unit using the actual data before the target time period stored in the actual data storage unit. The thermal load prediction apparatus according to claim 2. In the method for predicting the heat load of the target time zone of the air conditioning heat source equipment used for the building,
Acquire actual data before the target time period about the heat load of air-conditioning heat source equipment, information about the building external environment, and / or information about the building internal environment,
Get forecast data for the target time zone for information on the external environment of the building and / or information on the internal environment of the building,
Prepare a heat load prediction model that is expressed in terms of heat load, variables related to the building's external environment, and / or variables related to the building's internal environment, and is associated with each of a plurality of precooling or preheating time candidates,
For the first candidate for pre-cooling or pre-heating time, predict the heat load of the target time zone based on the actual data, prediction data, and heat load prediction model,
If the predicted heat load peak for the first precooling or preheating time candidate exceeds a predetermined threshold, the second precooling or preheating time candidate that is larger than the first precooling or preheating time candidate is reheated. Predict,
The thermal load prediction method characterized by the above-mentioned.

Images