JP3160693B2 - Heat load prediction device - Google Patents

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 on a heat supply equipment using a prediction data generating system for generating and outputting predicted data by inputting necessary factor data. The present invention relates to a heat load prediction device using a prediction data creation system for easily and quickly obtaining optimum prediction data.

[0002]

2. Description of the Related Art Generally, a heat load prediction device is a device for predicting, as a heat load, the amount of heat output from a heat supply facility used for air conditioning. Here, the heat supply facility is a facility that supplies heat to an air conditioner installed inside a building such as a building, a house, or a public facility. A heat generating device such as a refrigerator or a cold / hot water generator as a power source is used.

[0003] Further, this type of heat generating device may include a heat storage tank filled with a fluid having a high heat storage effect such as water, and when the heat storage tank is used, the generated heat is stored in the heat storage tank. And releases the accumulated heat when heat consumption occurs. When a heat storage tank is used, heat used during the day can be stored in the heat storage tank at night, and the capacity of the heat generating device can be reduced, so that the initial investment can be reduced.

Further, as described in “April 1993, Journal of Electrical Installations, P347-P350”, inexpensive night-time power is used for heat-generating equipment that uses electricity as a power source and has a heat storage tank. The operation cost can be reduced by using. In addition, there is an advantage for the electric power company that the electric power load can be leveled.

The amount of heat stored in the heat storage tank is determined by predicting the amount of heat to be consumed by the building in the future, and the heat generating device is operated according to the predicted value. In the prediction of the amount of heat used, there are many factors that change the load due to changes in the usage of the building due to changes in the weather, month, day, amount of heat accumulated inside the building, and changes in the day of the week, making it difficult to create an accurate prediction model. For,
This was done based on the experience of the operator based on the fluctuation factors. The prediction based on the operator's experience cannot realize unmanned driving.

For this reason, recently, a thermal load predicting apparatus using this kind of predictive data creating system is disclosed in
As described in Japanese Patent Publication No. 441, a self-organizing model such as a neural network model is used to perform accurate estimation, thereby improving efficiency and unmanning.

The structure of the neural network is such that predictive data, date, measurement data, current time, and other factor data are used as neurons in the input layer, and each predicted data is used as a neuron in the output layer. Further, a three-layer neural network having an intermediate-layer neuron between the input layer and the output layer is configured.

[0010] More specifically, as shown in FIG. 5, the data storage unit 21 stores heat load detection data for the heat supply equipment of the currently controlled and operating heat generating equipment as the actual heat load. The data input unit 22 is manually input with the weather, temperature, day of the week, and the like as the factor data of the heat load fluctuation. The pattern classification unit 23 extracts the actual heat load data stored in the data storage unit 21 and the actual weather data and day of the week data of the factor data that fluctuates the heat load input to the data input unit 22, and based on these data, Then, a feature for specifying the heat load pattern is defined, and the heat load pattern is classified based on the feature.

[0009] The characteristics of the change in the heat load pattern are the characteristics relating to the pattern for one day every hour (daytime and nighttime), the actual weather for three days including the current day such as the maximum temperature, the minimum temperature, the weather, etc., and the day of the week. Defined by the information. For each heat load pattern classified according to this feature, the result in time unit is averaged over the heat load pattern expressed by the ratio of the result in daily amount, and is used as a representative pattern for each classification.

The prediction model creation unit 24 extracts the heat load pattern classified by the pattern classification unit 23 and the patterns related to weather and day of the week, and learns the prediction model 25 based on these patterns. The prediction model 25 is created by inputting past weather data and day of the week, and learning a weight coefficient of a neural network that outputs a heat load pattern. The weighting factor obtained as a result of this learning is
It is stored as the prediction model 25.

The learning of the weight coefficient between each layer and each neuron of the neural network is performed by using a back propagation method in which the adjustment of the network is back-propagated from the output layer to the input layer.

The pattern predicting section 26 includes a data input section 22
Out of the factor data input to the prediction forecast 25, in this example, the weather forecast of the next day is taken out, and the weighting factors stored in the prediction model 25 are used, and the respective neurons (patterns) of the output layer are obtained according to the neural network model. Calculate the value. The pattern indicating the maximum value of each neuron in the output layer is selected as the heat load prediction pattern for the next day.

As a result, the pattern for the next day is accurately obtained in units of 24 hours every day.

[0014]

SUMMARY OF THE INVENTION A thermal load predicting apparatus using the above-described conventional predictive data creating system includes a thermal load detection data for detecting a thermal load performance of a heat supply facility in operation and a past three days of performance data. It has a structure to classify patterns based on data and create a prediction model. In this pattern classification, the next day is collectively formed in units of 24 hours a day, and the predicted load is also predicted as a heat load pattern for one day based on the pattern classification. For this reason, there is a problem that an error between the predicted load and the actual load at each time is not reflected in the calculation of the predicted load amount at another time of the day, and it is not possible to immediately respond to a sudden change in the actual result.

Further, even if a sudden change in the actual weather data within 24 hours is accepted, the pattern is classified, and a prediction model is created, the prediction model becomes a special pattern. Many repetitive learnings are required. As a result, there is a problem that it takes a lot of time to create an optimal prediction model, and there is a possibility that the measures for the heat supply facility cannot be made in time.

An object of the present invention is to provide a prediction data creation system capable of creating an optimal prediction model in a short time by using a neural network and responding to a rapid change in results in a short time, and a heat generation system using this system. A load prediction device is provided.

[0017]

According to the present invention, there is provided a predictive data creating system for inputting factor data, creating and outputting predictive data, comprising: a predictive model created in advance; An input unit for temporarily storing, a prediction unit for creating and outputting prediction data from a neural network model by applying the factor data received from the input unit to the prediction model, and a prediction data received from the prediction unit from the input unit. A prediction model creation unit that creates a prediction model by calculating a weighting factor of the neural network by adding the extracted factor and substitutes the prediction model for the prediction model. Further, the prediction model creation unit uses the past actual value and the predicted value as the teacher signal within a certain time.

A heat load prediction device according to the present invention is a heat load prediction device for predicting a heat load on a heat supply facility,
An outside air prediction model, which is created in advance, and a measured outside air temperature and outside air humidity, and a predicted weather, a maximum temperature, and a fluctuation factor data including a minimum temperature for the prediction time, and an outside air input unit for saving, An outside air prediction unit that applies the variation factor data received from the outside air input unit to the outside air prediction model to create and output prediction data of outside air temperature and outside air humidity from a neural network model, and outputs the outside air temperature and the outside air temperature received from the outside air prediction unit. An outside air prediction model creating unit that adds factor data extracted from the outside air input unit to the outside air humidity prediction data, calculates a weight coefficient of the neural network, creates an outside air prediction model, and substitutes the outside air prediction model for the outside air prediction model. , A load prediction model created and received in advance, and the measured load calorie,
A load input unit for inputting and storing the weather to be predicted for the prediction time, and the fluctuation factor data including the outside air temperature and the outside air humidity of the prediction data received from the outside air prediction unit, and the fluctuation factor data received from the load input unit. A load prediction unit that creates and outputs prediction data called load heat from a neural network model by applying the load prediction model, and adds the factor data extracted from the load input unit to the prediction data called load heat received from the load prediction unit. A load prediction model generation unit that calculates a weight coefficient of the neural network to generate a load prediction model, and substitutes the load prediction model for the load prediction model. The outside air prediction model creation unit and the load prediction model creation unit use past actual values and predicted values as teacher signals within a certain period of time.

[0019]

Next, the present invention will be described with reference to the drawings.

FIG. 1 is a functional block diagram showing an embodiment of the present invention.

In FIG. 1, two sets of prediction data creation systems are used. The first forecast data creation system is
The outside air prediction block 9 performs a learning operation based on the factor data 11 to obtain and output a predicted outside air temperature and a predicted outside air humidity 12. The second prediction data creation system is a load prediction block 10 that includes a predicted outside air temperature and a predicted outside air humidity 12.
The learning calculation is performed using both the data and the factor data 13 to obtain and output the predicted load of the final purpose.

Also in the present invention, the prediction data is obtained by using a neural network. For example, FIG.
As shown in FIG. 3A and FIG. 3A, the structure of the neural network is such that factor data including prediction data, date, measurement data, and current time is used as a neuron of an input layer, and each prediction data is used as an output layer. Of neurons. The neural network has a three-layer structure having an intermediate layer neuron between an input layer and an output layer.

First, the first prediction data creation system, that is, the outside air prediction block 9 for calculating the predicted outside air temperature and predicted outside air humidity 12 shown in FIG. 1 will be described.

The outside air input unit 1 for inputting the factor data 11 for model creation stores the received factor data 11 as outside air prediction data. The factor data 11 collects the outside temperature / outside humidity, the expected weather, the expected maximum outside temperature, and the expected minimum outside temperature in one-hour cycles, which are the variation data of the outside temperature / outside humidity which is the prediction data. With the date and time of collection of the collected data,
The past 24 hours are saved. The outside air temperature is measured by a temperature sensor or the like, and the current outside air humidity is measured by a humidity sensor or the like. The current forecast weather, the expected maximum temperature, and the expected minimum temperature are provided by a weather forecast distribution company or the like.

The outside air prediction unit 2 performs an experiment or the like in advance based on the factor data 11 taken out from the outside air input unit 1 to obtain an optimum predicted value. From the data of 3,
By the forward propagation method in which the adjustment of the neural network propagates from the input layer to the output layer every hour, the predicted outside air temperature and the predicted outside air humidity during that time are obtained,
Output. The obtained predicted outside air temperature and predicted outside air humidity are stored as prediction data for 24 hours.

The outside air prediction model 3 is composed of the outside air prediction model creator 4, which stores variation factor data such as the outside air temperature and the outside air humidity stored in the outside air input section 1 and the predicted outside air temperature and predicted outside air output from the outside air prediction section 2. From the humidity, the neural network adjustment is created by a back propagation method in which the output layer propagates from the output layer to the input layer.

The outside air prediction model creating unit 4 calculates the predicted outside air temperature / predicted outside air humidity from 23:00, which is, for example, 23:00 at the present time, 22:00 passed, or 2 hours predicted one hour before to 22:00 the next day. The predicted weather, predicted maximum temperature, predicted minimum temperature, month and day, and the current outdoor temperature /
The outside air prediction model is changed by the back propagation method every hour so that the data corresponding to the current outside air humidity and the current time in the input layer match.

FIG. 2A shows a structural diagram of a neural network in the outside air prediction unit 2.
(B) shows a prediction result in the outside air prediction unit 2.

As shown in the figure, the neurons in one input layer are provided with fluctuation factor data as "variable factor data on air conditioning and sanitary equipment, Shoichi Makino, et al., Co-authored by Shokokusha P9-P26".
As described in the above, factors affecting the change in weather are adopted. The illustrated factor data includes the expected weather on the due date to be forecast, the predicted maximum and minimum temperatures that affect the weather, the date and time that affects the weather,
It includes the outside air temperature and the outside air humidity currently being measured, which are reference values of the predicted outside air temperature and the predicted outside air humidity output by the forward propagation method, and the current time.

The neurons in the other output layer list the predicted outside air temperature and the predicted outside air humidity every hour.

For example, as shown in the drawing, the predicted data shows a curve with a predicted outside air temperature peak between 23:00 and 22:00 the next day, and a curve with a predicted outside air humidity having a valley. I'm drawing.

Next, returning to FIG. 1, the second prediction data creation system, that is, the load prediction block 10 for calculating the predicted load 14 shown in FIG. 1 will be described.
The operation function of each block is the same as that of the outside air prediction block 9 of the first prediction data creation system described above.

The load input unit 5 receives the factor data 13 and the predicted outside air temperature and predicted outside air humidity 1 output from the outside air prediction unit 2.
2 is input and stored as load prediction data and load prediction model creation data. The factor data 13 is variation factor data for the predicted load including the predicted weather, the date, the day of the week, and the secondary load up to the present.

The load predicting unit 6 calculates a secondary-side predicted load for 24 hours from the fluctuation factor data from the load input unit 5 and the load prediction model 7 created in advance by the forward propagation method. .

The fluctuation factor data input to the load prediction unit 6 is described in “Air Conditioning Equipment Takeshi Yoshimura et al.
As described in “30”, it is said that factors due to fluctuations in weather conditions such as weather, outside air temperature and outside air humidity, and factors related to heat generated by the human body and lighting and OA equipment are large. For this reason, the factor data includes the predicted weather on the day to be predicted, the predicted outside air temperature / outside humidity at the predicted time, and the usage status of the room which is a factor of the fluctuation of the heat generated by the human body and the lighting / OA equipment. Is indirectly expressed, and the current load calorific value, which is a reference value of the predicted load output by the forward propagation method, is adopted.

Depending on the air-conditioning system, cooling and heating may need to be realized at the same time, and the same factors of variation affect the load as the cold water load requiring cooling and the hot water load requiring heating. Therefore, independent prediction models are created and used.

The load prediction model 7 is obtained by the load prediction model creator 8 based on the secondary factor fluctuation factor data extracted from the load input unit 5 and the secondary predicted load output from the load prediction unit 6. It is created by the back propagation method.

FIG. 3A shows a structural diagram of the neural network in the load predicting section 6.
(B) shows a prediction result in the load prediction unit 6.

As shown in the figure, the expected weather, the predicted outside air temperature, the predicted outside air humidity, the month, the day, the day of the week, the measured load, and the current time are listed as the fluctuation factor data in the input layer neurons. In the neurons of the layer, the predicted loads for each hour are listed.

The prediction results show that, in the example shown, the predicted cold water load increases during the day and the predicted hot water load decreases during the day.

FIG. 4 is a transition diagram showing the relationship between the calculated load prediction used for creating the prediction model and the teacher signal. The first hour (A), the several hours (B), and the 24th hour (C). Is shown. As shown in the figure, when the prediction model is created every hour, a model can be created for a time at which there is already an actual value. For example, the actual value of several hours is multiplied by the actual value of the first hour. At the time of abnormal weather such as when away, an hour later, prediction according to the abnormal weather can be performed.

As described above, in this embodiment, the fluctuation factor data of the prediction data to be output is input and stored as the factor data, and the prediction data is referred to by referring to the factor data and a previously created prediction model. The forecast data creation system is used to create an outside air prediction block that forecasts the outside temperature and outside humidity that are greatly affected by weather conditions, and adds the predicted outside temperature and predicted outside air humidity to the factor data to determine the building equipment and human body density. A load prediction block for predicting the load to be affected is separately provided, and data having different fluctuation patterns are separately processed to simplify the calculation.

Further, since each prediction data creation system inputs the prediction result immediately as the factor data of the prediction model creation section, the prediction data can be immediately output to the supply equipment with the prediction cycle shortened. .

Although the thermal load predicting apparatus using the predictive data creating system according to the present embodiment has been described with reference to functional blocks, the functional blocks may have another configuration. Also, the quantity of stored data and the like may be set to another quantity.

The present invention is not limited by the above description in terms of block configuration and quantity as long as the above functions are satisfied.

[0046]

As described above, according to the present invention, the prediction model has a configuration in which the prediction model is created using the prediction data which is the output of the system and the factor data which fluctuates the prediction data. The prediction data can be output in a short time with respect to the cycle. As a result, an error between the prediction data at each time and the actual data can be reflected in the calculation of the prediction data at another time of the day, and a rapid change in the actual performance can be dealt with in a short time.

Further, the system predicts the outside air temperature and the outside air humidity which are greatly affected by the weather conditions, and adds the predicted outside air temperature and the predicted outside air humidity to the factor data to load the building equipment and the human body density. And a part for estimating the data, and has a configuration in which data having different fluctuation patterns are separately processed. For this reason, creation of an individual optimal prediction model is simple and can be performed in a short time.

[Brief description of the drawings]

FIG. 1 is a functional block diagram showing one embodiment of the present invention.

FIG. 2 is a structural diagram (A) showing an example of a neural network in an outside air prediction unit of FIG. 1, and a prediction result diagram (B).

3 is a structural diagram (A) showing an example of a neural network in a load prediction unit in FIG. 1, and a prediction result diagram (B).

FIG. 4 is a transition diagram showing a relationship between a calculated load prediction used for creating the prediction model in FIG. 1 and a teacher signal.

FIG. 5 is a functional block diagram showing an example of the related art.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 outside air input unit 2 outside air prediction unit 3 outside air prediction model 4 outside air prediction model creation unit 5 load input unit 6 load prediction unit 7 load prediction model 8 load prediction model creation unit 9 outside air prediction block 10 load prediction block 11, 13 factor data 12 Predicted outside air temperature and predicted outside air humidity 14 Predicted load

Continuation of front page (56) References JP-A-6-102908 (JP, A) JP-A-4-15441 (JP, A) JP-A-4-58188 (JP, A) JP-A-8-100940 (JP) (A) JP-A-6-313605 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) F24F 11/02 G05B 13/02 L G06F 15/18 550

Claims (2)

(57) [Claims] 1. A heat load predicting apparatus for predicting a heat load on a heat supply facility, comprising: an outside air prediction model prepared in advance; a measured outside air temperature and outside air humidity; Temperature, inputting and storing fluctuation factor data including the minimum temperature, and predicting data such as outside temperature and outside humidity from a neural network model by applying the fluctuation factor data received from the outside air input unit to the outside air prediction model. The outside air prediction unit that creates and outputs the outside air prediction model by adding the factor data extracted from the outside air input unit to the prediction data of the outside air temperature and the outside air humidity received from the outside air prediction unit and calculating the weight coefficient of the neural network. An outside air prediction model creation unit that creates and substitutes the outside air prediction model with the outside air prediction model; A load prediction model, measured heat load, and, weather to predict the prediction time,
A load input unit for inputting and storing fluctuation factor data including the outside air temperature and the outside air humidity of the prediction data received from the outside air prediction unit, and applying the fluctuation factor data received from the load input unit to the load prediction model, A load prediction unit that creates and outputs prediction data of load heat from the model, and calculates the weighting factor of the neural network by adding the factor data extracted from the load input unit to the prediction data of load heat received from the load prediction unit. A heat load prediction device, comprising: a load prediction model generation unit that generates a load prediction model and replaces the load prediction model with the load prediction model. 2. The thermal load predicting device according to claim 1, wherein the outside air predictive model generating unit and the load predictive model generating unit use past actual values and predicted values as teacher signals within a predetermined time.