JPH11125448A - Thermal load-predicting device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a thermal load predicting device wherein, by using only an actual load data, a load is predicted with high precision while relatively simple process is executed. SOLUTION: A data employment-deciding part 12 which decides whether an inputted actual thermal load data is employed as a load prediction data or not, and calculation parts 14 and 18 wherein a weight coefficient of actual thermal load for latest few days which is decided to be employed at the data employment deciding part is calculated and weighted movement average process is executed to predict thermal load pattern for the next day, are provided.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a heat load estimating apparatus for estimating a heat load in, for example, a heat source apparatus system for operating and controlling a heat source apparatus of a building air conditioner.

[0002]

2. Description of the Related Art In air conditioning equipment for buildings, a system for estimating a heat load required on the next day, storing heat by utilizing a low electricity rate at night in accordance with the prediction, and saving energy is becoming widespread. In addition, it is important to predict in advance the heat load necessary for setting pool water and facility temperature, for example, at events performed in indoor pools and various leisure facilities.

[0003] In order to improve the accuracy of heat load prediction, this type of heat load prediction device uses weather data such as temperature, humidity, rainfall rate, and the like, which are one of the factors greatly affecting the predicted value, and outside air data. There are known ones that are used, those that are obtained from complicated calculation formulas, and those that are statistically obtained from the past accumulated load data.

[0004]

However, in the conventional method as described above, a large amount of input data is required, such as weather data, outside air data, load data, day of the week data, or the amount of data stored in a database. In the case of data input by the operator, an input error or an input error may occur, which may cause a serious defect in the load prediction process.

On the other hand, in the case of using a complicated calculation formula, an enormous program for performing this calculation and equipment having performance suitable for the calculation are required, and the apparatus becomes considerably complicated. In addition, in the case of statistically obtaining from the past accumulated load actual data, there is a problem that it is not only necessary to accumulate data for a considerable period of time, but also that it cannot cope with a change in the surrounding environment. .

In view of the above problems, an object of the present invention is to provide a thermal load predicting apparatus capable of performing a highly accurate load prediction while performing relatively simple processing using only actual load data.

[0007]

According to a first aspect of the present invention, there is provided a data adoption determining section for determining whether or not to use input thermal load actual data as load prediction data, and a data adoption determining section. Calculating a weight coefficient of the heat load results for the latest several days determined to be adopted in the above, performing a weighted moving average process, and predicting a heat load pattern of the next day. It is a prediction device.

[0008] Thus, the weighted moving average processing is performed based on the detected actual load data and using the latest actual heat load for several days excluding abnormal values, thereby achieving highly accurate heat load prediction. Can do it.

According to a second aspect of the present invention, the data adoption deciding section has a calendar section for deciding whether or not to adopt data based on a calendar, and an abnormal value cutting section for excluding data far from recent data. The heat load prediction device according to claim 1, wherein:

According to a third aspect of the present invention, in the thermal load predicting apparatus according to the first aspect, the calculation of the weighting factor is determined based on a difference between an actual heat load and a predicted value thereof. . The invention according to claim 4 is the thermal load prediction device according to claim 1, wherein the weight coefficient increases as newer data. According to a fifth aspect of the present invention, there is provided the heat load estimating apparatus according to the second aspect, wherein a weight of each day can be changed in a calendar. The means described in claims 3 to 5 may be used alone or in combination.

[0011]

Embodiments of the present invention will be described below with reference to the drawings. Examples of the heat source device targeted by the heat load prediction device A include an air conditioner for cooling and heating a building, an air conditioner for a relatively large vehicle, and an air conditioner for a hot water pool.

The thermal load estimating apparatus A of this embodiment includes:
An input unit 10 for inputting actual load data detected by a detector attached to a heat source device, and data for determining whether to use the actual load data input to the input unit 10 as the latest load prediction data The adoption determining unit 12 compares the actual thermal load value and the predicted value in the actual load data determined to be adopted by the data adoption determining unit 12, and weights the actual thermal load, and the data adoption determining unit 14. The data storage unit 16, the data storage unit 16, which stores the latest load result data determined to be adopted by the unit 12 while updating the data daily while erasing the oldest data instead.
A load prediction calculation unit 18 that receives data from the server and performs a heat load prediction on the next day, and an output unit 20 that outputs the heat load prediction data obtained by the load prediction calculation unit 18 to the outside.

The data adoption deciding section 12 includes a calendar section 12a and an abnormal value cutting section 12b which perform processing in parallel. As shown in FIG. 2, the calendar section 12a indicates whether a specific day is a weekday, a holiday, or a peculiar day, and can select whether to adopt data on these days. It is configured as follows. That is, for example, when only the actual load data on weekdays is adopted, “adopt” is selected on weekdays, and “don't adopt” is selected otherwise.

On the other hand, in this embodiment, the abnormal value cut unit 12b compares the actual load data of the current day with the actual load data adopted by the data adoption deciding unit 12 two times before and last, and compares the actual load data of the current day. Only when the integrated load value is within a predetermined range, the actual load data of the day is adopted as the load prediction data. That is, only the actual load data of the day satisfying the following three equations is adopted as the load prediction data, and the other data is not adopted. In _{S n-2 × αL <S} n <S n-2 × αH (1) S n-1 × αL <S n <S n-1 × αH (2) αLL <S n (3) where, S _{n -2:} the second previous data load actual integrated value S _n-1: previous data load actual integrated value S _n: today's load ERA accumulated value .alpha.H: allowable upper limit (relative value) .alpha.L: allowable lower limit value (relative value) αLL: lower limit value.

In this manner, the actual data of the current day and the latest several days are compared, and the actual heat load of the present day is greatly different, and when the difference exceeds a set limit, or on a holiday or the like. Therefore, if the actual value of the day is less than the set limit, the data is not adopted, and the actual data used for the load prediction of the next day is not updated. On the other hand, when the data of the current day is adopted, the data is stored as the latest data for load prediction, the data before that is moved down one day at a time, and the oldest data is deleted, so that the latest data is always stored. The next day's load prediction is performed using the minute result data.

The weight coefficient calculating unit 14 compares the actual heat load value and the predicted value of the actual load data determined to be adopted by the data adoption determining unit 12 and weights the actual heat load. The weight coefficient K is given by the following equation according to the error between the actual data integrated value and the integrated value of the data predicted last time. 1 / K = 1 + (| Accumulated load actual-Accumulated load forecast |) / Accumulated load forecast

The data storage section 16 stores actual heat load data for the past several days (three days in this embodiment). The latest load result data determined is added, and the data is updated daily while deleting the oldest data instead. At the same time, the weight coefficient K is stored while being sequentially updated.

As shown in FIG. 4, the load prediction calculation unit 18 calculates the latest three days of load actual data stored in the data storage unit 16 and the respective weighting factors K ₁ , K ₂ , and K ₃ . Load prediction at each time on the next day is performed.
That is, the predicted value at each time of the next day = (D _n−2 × K ₁ + D _n−1 × K ₂ + D _n
× K ₃ ) / (K ₁ + K ₂ + K ₃ ) Here, D _n-2 : the actual load value at the current time in the data before last D _n-1 : the actual load value at the current time in the previous data D _n : today This is the actual load value at that time in the data. The predicted value at each time obtained by this equation is stored in the data storage unit 16 and further output from the output unit 20.

As described above, according to the error between the predicted value and the actual value for each day, the weight coefficient is given so that the larger the error is, the smaller the weight coefficient becomes. Determines that the day is a peculiar day, such as a sudden change in temperature or a change in driving state, and reduces the weighting coefficient for that day to reduce the effect of the prediction. When the error is large for three days or more, the predicted value becomes small for the first time.

Next, the operation of the heat load predicting / calculating device A configured as described above will be described with reference to the flowchart shown in FIG. First, the actual heat load data detected from the heat source device is totaled at regular intervals, for example, every hour, input from the input unit 10 and temporarily stored in the data storage unit 18, and the arithmetic processing of the day is performed. Time, for example, 22
It is set at time (step 1).

Then, when the set time has been reached, the data adoption determining section 12 determines whether or not to use the current load actual data as the load prediction data (step 2). That is, for example, when the calendar unit 12a selects to use the load result data only on weekdays, the abnormal value cut unit does not use the load result data other than weekdays, and even if the load result data is on weekdays. If it is determined to be an abnormal value in 12b, the load prediction of the current day is output as it is as the load prediction of the next day without adopting the data, and the data is deleted from the data storage unit 18 at the same time.

With respect to the actual load data on the day determined to be adopted by the data adoption decision unit 12, the weight coefficient calculation unit 1
In step 4, the weight coefficient K ₃ of the load data of the present day is calculated (step 3), and the weight coefficient K ₃ and the load data of the present day are stored in the data storage unit 16, and the oldest data is deleted. Update coefficients.

Then, the actual load data and the weight coefficient K _{3 for} today, and also the previous and previous load actual data and the weight coefficients K ₂ and K ₁ are sent to the load prediction calculation unit 18, where the load is predicted ( Step 4) The prediction result is stored in the data storage unit 16 and output from the output unit 20.

On the basis of the load prediction, the heat source control device performs an operation schedule creation process, and controls the nighttime heat storage operation and the air conditioning operation of the next day based on the process. In this way, using only the load actual data detected from the heat source device as input data, predicting the next day's heat load pattern,
A highly accurate heat load prediction can be performed with a simple configuration.

In the embodiment, an example is shown in which the heat load of the next day is predicted based on the latest heat load results for the past three days. However, the present invention is not limited to this. In addition, since the above-mentioned heat load prediction device has a relatively simple configuration, it does not require a high-performance and high-precision function such as a commercially available general-purpose computer, so-called personal computer or EWS. It can also be constructed with a widely used sequencer or the like.

[0026]

As described above, according to the present invention,
Based on the detected actual load data, the weighted moving average processing is performed using the latest thermal load results for the last few days excluding abnormal values, so that relatively simple processing can be performed using only actual load data. It is possible to perform highly accurate load prediction while performing. Since it can be constructed with only a minimum amount of data and simple calculations, it does not require high-function and configuration functions as hardware, and requires a small amount of work, so that the cost is low.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a heat load prediction calculation device according to an embodiment of the present invention.

FIG. 2 is a diagram attached to a description of a calendar section of the embodiment shown in FIG. 1;

FIG. 3 is a diagram attached to description of an abnormal value cut section of the embodiment shown in FIG. 1;

FIG. 4 is a diagram attached to description of a load prediction calculation unit of the embodiment shown in FIG. 1;

FIG. 5 is a flowchart illustrating the operation of the embodiment shown in FIG. 1;

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Input part 12 Data adoption determination part 12a Calendar part 12b Abnormal value cut part 14 Weight coefficient calculation part 16 Data storage part 18 Load prediction calculation part 20 Output part

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Chiharu Sakashita 3-4-1 Shibaura, Minato-ku, Tokyo Inside NTT Facilities Inc. (72) Inventor Noboru Makita 11th Asahimachi Haneda-cho, Ota-ku, Tokyo No. 1 Inside EBARA CORPORATION (72) Inventor Hideaki Yanagishita 11-1 Haneda Asahimachi, Ota-ku, Tokyo Inside EBARA CORPORATION

Claims (5)

[Claims] 1. A data adoption deciding section for deciding whether or not to use input thermal load actual data as load prediction data, and a heat application for the latest several days decided to be adopted by the data adoption deciding section. A heat load predicting apparatus, comprising: a calculating unit that calculates a weight coefficient of a load result, performs a weighted moving average process, and predicts a heat load pattern on the next day. 2. The apparatus according to claim 1, wherein the data adoption determining unit includes a calendar unit that determines whether or not to adopt data based on a calendar, and an abnormal value cut unit that excludes data that is far from recent data. 2. The heat load prediction device according to 1. 3. The heat load predicting apparatus according to claim 1, wherein the calculation of the weight coefficient is determined based on a difference between the actual heat load and a predicted value thereof. 4. The thermal load predicting apparatus according to claim 1, wherein the weight coefficient increases as newer data. 5. The heat load predicting device according to claim 2, wherein the weight of each day can be changed in a calendar.