JP2020148368A - Shop air conditioning system - Google Patents

Abstract

To provide a shop air conditioning system capable of improving the accuracy of prediction control for reducing energy consumption while comfortably air-conditioning the inside of a shop by an air conditioner, from the aspect of database.SOLUTION: An air conditioner control part 32 includes a result acquisition mode for controlling the operation of an air conditioner 19 by switching a plurality of air conditioning set temperatures A, and a prediction control mode for controlling the operation of the air conditioner 19 with the optimum value, being the air conditioning set temperature A, which is calculated by machine learning from database by an optimization control part 31. The optimization control part 31 accumulates at least power consumption P and a discomfort index D acquired in the result acquisition mode and the prediction control mode in the database, and updates it.SELECTED DRAWING: Figure 2

Description

The present invention relates to a store air-conditioning system that air-conditions a store in which a refrigerating device such as a showcase is installed.

Conventionally, stores such as convenience stores and supermarkets have been provided with a plurality of refrigerating devices such as refrigerating / freezing showcases. In recent years, in order to save the showcases installed in these stores, a system that predicts the future outside air temperature and controls the showcase, and a database of operation results are stored in a database to predict and operate equipment. Devices have also been developed (see, for example, Patent Document 1 and Patent Document 2).

Japanese Patent No. 6284173 Japanese Unexamined Patent Publication No. 2016-23880

Here, the inside of the store where the refrigerating device including the above-mentioned showcase is installed is air-conditioned by the air-conditioning device based on a predetermined air-conditioning set temperature, but the air-conditioning set temperature is changed and the temperature and humidity in the store are changed. If it changes, it will also affect the operation of the refrigeration equipment installed in the store. However, in the past, the air-conditioning set temperature of the air-conditioning device was set at a fixed value for each season, and the clerk had to make fine adjustments to the air-conditioning set temperature so as to eliminate the discomfort of the customer. It has an adverse effect on operation, and the power consumption of the entire store, including the air conditioner, will increase significantly.

Therefore, it is conceivable to create a database by associating the comfort in the store with the power consumption, etc., and to predict the set temperature of the air conditioner that can achieve both comfort and energy saving, and control the air conditioner. In order to do so, various actual data at the store must be accumulated in the database, and if the store clerk makes appropriate settings or performs predictive control, the number of operations in the same or similar situation will increase, and the actual data There is a problem with diversification. In addition, there is also a problem that it is difficult to confirm the effect only by performing predictive control.

The present invention has been made in order to solve the conventional technical problems, and the accuracy of predictive control for reducing energy consumption while comfortably air-conditioning the inside of a store by an air conditioner can be obtained from the aspect of a database. The purpose is to provide a store air conditioning system that can be improved.

The store air conditioner system of the present invention air-conditions the inside of a store in which a refrigerating device including a showcase is installed, and controls the operation of the air conditioner that air-conditions the inside of the store based on a predetermined air-conditioning set temperature A. It has an air conditioner control unit, and at least a database constructed by accumulating data on indicators showing energy consumption of air conditioners and refrigerating devices and indicators showing comfort in the store, and based on this database. By performing machine learning, an optimization control unit that calculates the optimum value of the air-conditioning set temperature A that minimizes the index showing energy consumption after a predetermined time within the allowable range of the index showing comfort after a predetermined time. The air conditioner control unit controls the operation of the air conditioner by switching between a plurality of air conditioner set temperatures A and controlling the operation of the air conditioner, and using the optimum value calculated by the optimization control unit as the air conditioner set temperature A. The optimization control unit stores and updates at least the data indicating the energy consumption index and the data indicating the comfort index acquired in the performance acquisition mode and the predictive control mode in the database. It is characterized by that.

In the store air conditioner system according to the second aspect of the present invention, in the above invention, the air conditioner control unit has an independent control mode for controlling the operation of the air conditioner according to the air conditioner set temperature A set by using a predetermined input device, which is optimal. The air conditioner control unit is characterized in that at least the data regarding the index indicating the energy consumption and the data regarding the index indicating the comfort acquired in the independent control mode are accumulated and updated in the database.

The store air-conditioning system according to the third aspect of the present invention is characterized in that, in each of the above inventions, the air-conditioning device control unit switches between the actual result acquisition mode and the predictive control mode, or the actual result acquisition mode, the predictive control mode, and the independent control mode. And.

The store air conditioning system according to the fourth aspect of the present invention has an outside air temperature detection unit that detects the outside air temperature T, an in-store temperature detection unit that detects the temperature inside the store, and an in-store humidity detection unit that detects the humidity inside the store. It is equipped with a unit and a power consumption detection unit that detects the power consumption P as an index indicating the energy consumption of the air conditioner and the refrigeration device, and the optimization control unit is an index indicating the comfort in the store from the temperature and humidity in the store. A discomfort index calculation unit that calculates the discomfort index D, and a database construction unit that accumulates at least the outside air temperature T, the air conditioning set temperature A, the power consumption P, and the discomfort index D for a predetermined period to construct a database, and this database construction. Machine learning is performed based on the database constructed by the department, and the power consumption TGT1 and the discomfort index TGT2 after a predetermined time are predicted from at least the latest outside air temperature T, air conditioning set temperature A, power consumption P and discomfort index D. A prediction unit that predicts multiple power consumption TGT1 and discomfort index TGT2 after a predetermined time by applying a plurality of air conditioning set temperatures A to the prediction model creation unit that creates a prediction model and the prediction model created by this prediction model creation unit. From the plurality of power consumption TGT1 and discomfort index TGT2 after a predetermined time predicted by this prediction unit, the air conditioning set temperature at which the power consumption TGT1 after a predetermined time is the smallest within the allowable range of the discomfort index TGT2 after a predetermined time. It is characterized by having an optimum value calculation unit for calculating the optimum value of A.

The store air-conditioning system according to claim 5 is characterized in that, in the above invention, the prediction model creation unit periodically creates and updates a prediction model.

In the store air conditioning system of the invention of claim 6, in the invention of claim 4 or 5, the prediction model creation unit is machine-learned by either linear regression analysis, nonlinear regression analysis, or a combination thereof. It is characterized by performing the above and creating a prediction model.

In the store air conditioning system of the invention of claim 7, in the invention of claims 4 to 6, the prediction unit periodically predicts a plurality of power consumption TGT1 and discomfort index TGT2 after a predetermined time, and the optimum value calculation unit predicts. The feature is that the optimum value is calculated every time the unit predicts.

In the invention of claims 4 to 7, the store air conditioner system of the invention of claim 8 has data on the operating factors of the air conditioner and / or the refrigerating device that affect the power consumption P and the discomfort index D in the database. Is additionally provided, the prediction model creation unit creates a prediction model for predicting the power consumption TGT1 and the discomfort index TGT2 after a predetermined time by adding the added operation factor, and the prediction unit creates an air conditioner in the prediction model. By applying a plurality of the added operating factors in addition to the set temperature A, a plurality of power consumption TGT1 and discomfort index TGT2 after a predetermined time are predicted, and the optimum value calculation unit predicts a plurality of predetermined time after the prediction unit. From the power consumption TGT1 and the discomfort index TGT2, the optimum value of the added operating factor that minimizes the power consumption TGT1 after the predetermined time is calculated within the allowable range of the discomfort index TGT2 after the predetermined time, and in the predictive control mode. , The air conditioner control unit and / or the refrigerating device control unit that controls the operation of the refrigerating device of the air conditioner and / or the refrigerating device using the optimum value calculated by the optimum value calculation unit as the added operating factor. It is characterized by controlling the operation.

In the store air-conditioning system of the invention of claim 9, the added operating factors in the above invention are the operation mode of the air-conditioning device, the temperature inside the showcase, the air volume of the circulation fan inside the showcase, and the target low pressure of the refrigerating device. One of the pressure, the high pressure of the refrigerating device, the number of revolutions of the compressor of the refrigerating device, the minimum valve opening of the expansion valve of the refrigerating device, and the operating state of the total heat exchanger that forms part of the air conditioner, or , A combination thereof, or all of them.

According to the present invention, in a store air conditioner system that air-conditions the inside of a store in which a refrigerating device including a showcase is installed, an air conditioner that controls the operation of the air conditioner that air-conditions the inside of the store based on a predetermined air conditioner set temperature A. It has a control unit, and at least a database constructed by accumulating data on indicators showing energy consumption of air conditioners and refrigeration devices and indicators showing comfort in the store, and machine learning based on this database. By performing the above, an optimization control unit for calculating the optimum value of the air conditioner set temperature A in which the index indicating the energy consumption after the predetermined time is the smallest within the allowable range of the index indicating the comfort after the predetermined time is provided. Since the air conditioner control unit is provided with a predictive control mode that controls the operation of the air conditioner with the optimum value calculated by the optimization control unit as the air conditioner set temperature A, the air conditioner and the air conditioner can be used within the range where comfort in the store can be tolerated. The set temperature of the air conditioner can be automatically adjusted to the optimum value that can minimize the energy consumption of the refrigerating device, and the operation of the air conditioner can be controlled. As a result, it becomes possible to reduce the energy consumption of the air conditioner and the freezer while ensuring the comfort in the store.

Further, the air conditioner control unit is provided with a performance acquisition mode for switching a plurality of air conditioner set temperatures A to control the operation of the air conditioner, and the optimization control unit is acquired in the performance acquisition mode and the prediction control mode. At least data on indicators of energy consumption and data on indicators of comfort are stored and updated in the database, so it is possible to store various performance data at the store in the database, and forecast. It becomes possible to perform highly accurate operation in the control mode. In addition, by comparing the actual results of driving in the actual results acquisition mode and the predictive control mode, it is easy to confirm the effect of the predictive control mode.

Further, according to the invention of claim 2, in addition to the above invention, the air conditioner control unit is provided with an independent control mode for controlling the operation of the air conditioner according to the air conditioner set temperature A set by using a predetermined input device. Therefore, for example, the user can control the air conditioner at an arbitrary air conditioner set temperature, and the optimized control unit is comfortable with the data regarding the index indicating at least the energy consumption acquired in the independent control mode. Since the data related to the index showing the sex is also stored and updated in the database, it is possible to further diversify the data stored in the database.

Further, according to the invention of claim 3, the air conditioner control unit switches between the actual result acquisition mode and the predictive control mode, or the actual result acquisition mode, the predictive control mode, and the independent control mode to execute the database. It will be possible to automatically diversify the data of the above, and for example, when using a database built at another store, air conditioning while interweaving the performance acquisition mode and independent control mode with the predictive control mode. It will be possible to operate the equipment, store the actual data in the database, and update it to build the store's own database.

Then, according to the invention of claim 4, at least the outside air temperature T, the air conditioning set temperature A, the power consumption P as an index showing the energy consumption of the air conditioning device and the refrigerating device, and the comfort in the store are determined by the database construction unit. A database is constructed by accumulating the discomfort index D as an index for a predetermined period, and the prediction model creation unit performs machine learning based on this database, and at least the latest outside air temperature T, air conditioning set temperature A, power consumption P, Then, from the discomfort index D, a prediction model for predicting the power consumption TGT1 and the discomfort index TGT2 after a predetermined time is created, and the prediction unit applies a plurality of air conditioning set temperatures A to this prediction model, so that after a predetermined time The power consumption TGT1 and the discomfort index TGT2 of the above are predicted, and the optimum value calculation unit predicts the power consumption TGT1 and the discomfort index TGT2 after the predicted plurality of predetermined times within an allowable range of the discomfort index TGT2 after the predetermined time. Since the optimum value of the air conditioning set temperature A at which the power consumption TGT1 becomes the smallest after a predetermined time is calculated, in the predictive control mode, the discomfort index D in the store is within an acceptable range, and the air conditioner and the refrigerating device are used. The air conditioner set temperature A can be automatically adjusted to the optimum value that can minimize the power consumption P, and the operation of the air conditioner can be controlled. As a result, it becomes possible to accurately reduce the power consumption of the air conditioner and the freezer while ensuring the comfort in the store.

In this case, if the prediction model creation unit periodically creates and updates the prediction model as in the invention of claim 5, the optimum value of the air conditioning set temperature A can be accurately set according to the change in the environment surrounding the store. Will be able to be calculated.

For machine learning in the prediction model creation unit, either one of linear regression analysis as in the invention of claim 6 and non-linear regression analysis, or a combination thereof can be considered.

Further, as in the invention of claim 7, the prediction unit periodically predicts a plurality of power consumption TGT1 and discomfort index TGT2 after a predetermined time, and the optimum value calculation unit calculates the optimum value each time the prediction unit predicts. If this is done, the optimum value of the air conditioning set temperature A can be calculated accurately in response to changes in the situation in the store.

Further, as in the invention of claim 8, data on the operating factors of the air conditioner and / or the refrigerating device that affect the power consumption P and the discomfort index D are additionally provided in the database, and the prediction model creation unit is added. A prediction model for predicting the power consumption TGT1 and the discomfort index TGT2 after a predetermined time is created by adding the specified operation factors, and the prediction unit applies a plurality of operation factors added in addition to the air conditioner set temperature A to the prediction model. As a result, a plurality of power consumption TGT1 and discomfort index TGT2 after a predetermined time are predicted, and the optimum value calculation unit predicts the discomfort index after a predetermined time from the plurality of power consumption TGT1 and discomfort index TGT2 after a predetermined time predicted by the prediction unit. Within the range that TGT2 can tolerate, the power consumption TGT1 after a predetermined time is the smallest. The optimum value of the added operating factor is calculated, and the air conditioner control unit and / or the refrigeration device control unit that controls the operation of the refrigeration device. However, if the operation of the air conditioner and / or the refrigeration device is controlled by using the optimum value calculated by the optimum value calculation unit as an added operating factor, the power consumption P other than the outside air temperature T and the air conditioner set temperature A And the operating factor that affects the discomfort index D is also controlled by the optimum value, and it is possible to further improve the comfort and reduce the power consumption.

In this case, the added operating factors include the operation mode of the air conditioner, the temperature inside the showcase, the air volume of the circulation fan inside the showcase, and the target low pressure of the refrigerating device, as in the invention of claim 9. , The high pressure of the refrigerating device, the number of revolutions of the compressor of the refrigerating device, the minimum valve opening of the expansion valve of the refrigerating device, and the operating state of the total heat exchanger constituting a part of the air conditioner can be considered.

It is a schematic plan view of the store of one Example to which the store air-conditioning system of this invention was applied. It is a block diagram of one Example of the store air-conditioning system of this invention. It is a figure explaining the control mode of the air conditioner control unit in FIG. 2 and its switching. It is a block diagram of the optimization control part in FIG. It is a flowchart explaining the operation of the optimization control unit of FIG. It is another flowchart explaining the operation of the optimization control part of FIG. It is a figure which shows the structure of the database which the optimization control part of FIG. 3 constructs. It is a figure explaining the data used for the machine learning by the optimization control part of FIG. It is the figure which showed the value to be predicted by the optimization control part of FIG. 3 and the data of FIG. 8 for the past day side by side. It is a figure explaining the determination of the optimum value of the air-conditioning set temperature. It is a figure explaining the prediction model of the power consumption by the decision tree analysis. It is a figure explaining the prediction model of the discomfort index by the decision tree analysis. It is a block diagram of another Example of the store air-conditioning system of this invention when the prediction model is created on the server side of a management company.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
(1) Store 1
FIG. 1 shows a schematic plan view of a general convenience store as an embodiment of the store 1. In the sales floor 2 of the store 1, an open type showcase 4 (refrigerated and frozen. There are multiple units) and product shelves 6 to 8 are installed from the back when viewed from the front entrance 3, and in the machine room 9 on the left side. A reach-in type showcase 11 (refrigerated and frozen. There are multiple units) is installed, a toilet 12 is arranged in front of the showcase 11, and a cash register space 13, a cooking room 14, and an office 16 are arranged on the right side. An indoor unit 17 of an air conditioner 19 (FIG. 2) for air-conditioning the inside of the store 1 is provided on the ceilings of the sales floor 2, the cooking room 14, and the office 16.

In the figure, 18 is a fryer (fried food cooker) provided in the cooking room 14. Further, in this application, the inside of the thick line frame shown in FIG. 1 is the inside of the store, and the outside of the frame is the outside of the store. Then, a refrigerator 15 for circulating the refrigerant in the showcases 4 and 11 and an outdoor unit 20 of the air conditioner 19 are installed outside the store, and the showcases 4 and 11 and the refrigerator 15 (compressors constituting the refrigerant circuit) are installed. The refrigerator 21 (FIG. 2) is configured by the showcases 4 and 11 (including the internal circulation fan of the showcases 4 and 11), and the air conditioner 19 is configured by the indoor unit 17 and the outdoor unit 20. Further, the store 1 is provided with a total heat exchanger 22 for exchanging heat between the outside air supplied to the store and the air discharged to the outside of the store, which also constitutes a part of the air conditioner 19 in the present application. It shall be.

(2) Store air conditioning system 24
Next, FIG. 2 shows an example of the configuration of the store air-conditioning system 24 of the present invention provided in the store 1 of FIG. The store air conditioner system 24 of this embodiment includes the above-mentioned air conditioner 19 and refrigeration device 21, and a power consumption detection unit 26 including a power meter that detects the total power consumption P of the air conditioner 19 and the refrigeration device 21. , An outside air temperature detection unit 27 composed of a temperature sensor for detecting the outside air temperature T of the store 1, an in-store temperature detection unit 28 including a temperature sensor for detecting the temperature inside the store 1, and a humidity sensor for detecting the humidity in the store 1. The store interior humidity detection unit 29 is composed of an optimization control unit 31 composed of a microcomputer equipped with a microprocessor, an air conditioner control unit 32, and a refrigeration device control unit 33.

In the embodiment, the power consumption P is an index indicating the energy consumption of the air conditioner 19 and the refrigerating device 21. Further, the discomfort index D in the store 1, which will be described later, is an index indicating the comfort in the store 1.

(3) Air conditioner control unit 32 and refrigeration device control unit 33
The air conditioner control unit 32 is provided in the air conditioner 19, and is an air conditioner including the outdoor unit 20 and the indoor unit 17 described above so that the temperature inside the store 1 becomes the air conditioner set temperature A based on the air conditioner set temperature A described later. Controls the operation of 19. In the figure, 35 is an input device for setting the air conditioner set temperature A and the like in the air conditioner control unit 32. In the embodiment, it is assumed that the input device 35 is configured by the switch provided in the air conditioner control unit 32, but the air conditioner set temperature A and the like can be set by remote control from the management company via the Internet. May be good. In that case, the network constitutes the input device 35. Further, the refrigerating device control unit 33 is provided in the refrigerating device 21, and the refrigerating device 21 including the refrigerating machine 15 and the showcases 4 and 11 described above so that the internal temperature of each of the showcases 4 and 11 becomes a predetermined set temperature. Control the operation of.

(4) Control mode of the air conditioner control unit 32 As described above, the air conditioner control unit 32 sets the outdoor unit 20 and the indoor unit 17 so that the temperature inside the store 1 becomes the air conditioner set temperature A based on the air conditioner set temperature A. The operation of the air conditioner 19 including the above is controlled, and in the embodiment, it has three types of control modes: a performance acquisition mode, a predictive control mode, and an independent control mode.

(4-1) Achievement acquisition mode Of these, the achievement acquisition mode is a control mode that automatically switches between a plurality of air conditioning set temperatures A. In the embodiment, the air-conditioning set temperature A is automatically (forcibly) switched between 25 ° C. and 26 ° C. at predetermined time (for example, 1 hour) to control the operation of the air-conditioning device 19.

(4-2) Predictive control mode Next, in the predictive control mode, the operation of the air conditioner 19 is controlled with the optimum value Auto calculated by the optimization control unit 31 as described later as the air conditioning set temperature A.

(4-3) Independent control mode In the independent control mode, the operation of the air conditioner 19 is controlled by setting the input value set by the input device 35 described above as the air conditioning set temperature A.

(4-4) Switching of Control Mode In the embodiment, the air conditioner control unit 32 sets the above-mentioned actual result acquisition mode, predictive control mode, and independent control mode every predetermined period (for example, every day) as shown in FIG. ) Is repeated.

(5) Configuration of Optimization Control Unit 31 Next, in the optimization control unit 31, the total power consumption P (index indicating energy consumption) of the air conditioner 19 and the refrigeration device 21 output by the power consumption detection unit 26 is combined. ), The outside air temperature T detected by the outside air temperature detection unit 27, the temperature inside the store 1 detected by the store temperature detection unit 28, and the humidity inside the store 1 detected by the store humidity detection unit 29 are input. Then, the optimization control unit 31 transmits and receives data between the air conditioner control unit 32 of the air conditioner 19 and the refrigerating device control unit 33 of the refrigerating device 21, and optimizes these operations as described later. Perform optimization control.

Next, the detailed configuration of the optimization control unit 31 will be described with reference to FIG. The optimization control unit 31 of the embodiment includes a data reception unit 34, a discomfort index calculation unit 36, a database construction unit 37, a prediction model creation unit 38, a prediction calculation unit 41, and a data transmission unit 46. The prediction calculation unit 41 has a prediction unit 42 and an optimum value calculation unit 44. Further, the database construction unit 37 has a database 37A constructed in the storage device.

The data receiving unit 34 of the optimization control unit 31 performs an operation of receiving data from the air conditioner control unit 32 and the refrigerating device control unit 33, as well as the power consumption P from the power consumption detection unit 26 described above and the outside air temperature. It receives data on the outside air temperature T from the detection unit 27, the temperature inside the store 1 from the store temperature detection unit 28, and the humidity inside the store 1 from the store humidity detection unit 29. Further, the data transmission unit 46 performs an operation of transmitting data to the air conditioner control unit 32 and the refrigeration device control unit 33.

(6) Operation of Optimization Control Unit 31 With the above configuration, the construction of the database by the optimization control unit 31 and the optimization control will be described next with reference to FIGS. 5 to 12.
(6-1) Construction of database 37A by database construction unit 37 The data reception unit 34 of the optimization control unit 31 in FIG. 4 has the power consumption P from the power consumption detection unit 26 and the outside air temperature in step S1 of FIG. Receives data on the outside air temperature T from the detection unit 27, the temperature inside the store 1 from the store temperature detection unit 28, and the humidity inside the store 1 from the store humidity detection unit 29. Further, the data regarding the air conditioning set temperature A is received from the air conditioner control unit 32.

At this time, the data on the temperature and humidity in the store 1 are sent to the discomfort index calculation unit 36. The discomfort index calculation unit 36 calculates the discomfort index D (an index indicating the comfort in the store 1) in the store 1 from the temperature and humidity in the store 1 received from the data reception unit 34, and the data reception unit 34. Return to. Then, the data receiving unit 34 transmits the data of the outside air temperature T, the air conditioner set temperature A, the power consumption P, and the discomfort index D to the database construction unit 37.

The database construction unit 37 of the optimization control unit 31 accumulates the received data of the outside air temperature T, the air conditioning set temperature A, the power consumption P, and the discomfort index D for a predetermined period (one year in the example), and FIG. A database 37A having the contents shown in the above is constructed. In FIG. 7, the air-conditioning set temperature is described as the air-conditioning setting, and the power consumption is described as electric power.

In this case, in the embodiment, the database construction unit 37 periodically (in the embodiment, every 30 minutes as shown in FIG. 7) data of the outside air temperature T, the air conditioning set temperature A, the power consumption P, and the discomfort index D are stored in the database. Recorded in 37A, first accumulated for one year, and after accumulating for one year, every time new data is recorded, the oldest data that has passed one year is deleted from the oldest data for the past one year. The data will be updated sequentially. In FIG. 7, the value of each outside air temperature T for every 30 minutes for one year is shown by t1 to t17520, the value of each air conditioning set temperature A is shown by a1 to a17520, and the value of each power consumption P is shown by p1 to p17520. , The value of each discomfort index D is shown by d1 to d17520.

(6-2) Creation of prediction model (1)
Next, in step S2 of FIG. 5, the prediction model creation unit 38 of the optimization control unit 31 performs the latest (in the example, the latest one day's worth) outside air temperature T to T48 and the air conditioning set temperature A in the database 37A. Based on the data of ~ A48, power consumption P to P48, and discomfort indexes D to D48, a prediction model for predicting power consumption TGT1 after a predetermined time (1 hour in the example) by machine learning, and the same as a predetermined time A prediction model for predicting the discomfort index TGT2 (also in the example after 1 hour) is created.

FIG. 8 shows the data used for this machine learning, and FIG. 9 shows the actual value TGT1real of the power consumption TGT1 after 1 hour (after a predetermined time) and the discomfort after 1 hour (after a predetermined time) to be predicted by the prediction model. The actual value TGT2real of the index TGT2 is shown in each row alongside the data of FIG. 8 for one day. In addition, also in FIGS. 8 and 9, the air-conditioning set temperature is described as the air-conditioning setting and the power consumption is described as electric power.

The prediction model of this example is a calculation formula of multiple regression analysis, which is one of the linear regression analyzes represented by the following formulas (I) and (II).
TGT1 = b0 + b1 x T + b2 x T1 + ... + b196 x D48 ... (I)
TGT2 = c0 + c1 x T + c2 x T1 + ... + c196 x D48 ... (II)
The formula (I) is a prediction model for predicting the power consumption TGT1 after 1 hour, and the formula (II) is a prediction model for predicting the discomfort index TGT2 after 1 hour.

In the machine learning of step S2 of FIG. 5, when the prediction model creation unit 38 applies the data of each row of FIG. 9 to the equations (I) and (II), the consumption after 1 hour calculated in all the rows. The common coefficients b0 to b196 and c0 to c196 such that the difference between the power TGT1 and its actual value TGT1real and the difference between the discomfort index TGT2 after 1 hour and its actual value TGT2real are minimized are determined.

Then, the prediction model creation unit 38 predicts the power consumption TGT1 after 1 hour (Equation (I)) and the discomfort index TGT2 after 1 hour (Equation (Equation (Equation)). II)) is output in step S3 of FIG. 5 (indicated by reference numeral 39 in FIG. 4). It should be noted that the prediction model creation unit 38 regularly creates and updates the prediction model (once a day in the embodiment).

(6-3) Optimization Control Next, in step S4 of FIG. 6, the prediction unit 42 of the prediction calculation unit 41 of the optimization control unit 31 has the latest (in the embodiment, the latest one day's worth) outside air temperature. Data of T, air conditioning setting A, power consumption P, and discomfort index D are acquired from the database 37A, applied to the prediction models represented by the above-mentioned equations (I) and (II) in step S5, and in steps S6 and S7. The power consumption TGT1 and the discomfort index TGT2 after 1 hour are predicted and output (the prediction result 43 is shown in FIG. 4). It should be noted that the prediction unit 42 periodically (in the example, once an hour) predicts and outputs the power consumption TGT1 and the discomfort index TGT2 after one hour.

Here, in step S5 of FIG. 6, the prediction unit 42 changes the current air-conditioning set temperature A in FIG. 8 to a plurality of values, for example, 21 ° C., 22 ° C., 23 ° C., and 24 ° C. The power consumption TGT1 and the discomfort index TGT2 after hours are predicted. Now, when the predicted result is as shown in FIG. 10, that is, when the air conditioning set temperature A is 21 ° C., the power consumption TGT1 after 1 hour is "5 kW", and the discomfort index TGT2 after 1 hour is. "Unpleasant", the power consumption TGT1 after 1 hour when the air conditioner set temperature A is 22 ° C. is "4 kW", the discomfort index TGT2 after 1 hour is "comfortable", and when the air conditioner set temperature A is 23 ° C. The power consumption TGT1 after 1 hour is "3 kW", the discomfort index TGT2 after 1 hour is "comfortable", and the power consumption TGT1 after 1 hour is "4 kW" when the air conditioning set temperature A is set to 24 ° C. Assuming that the discomfort index TGT2 is "comfortable", the optimum value calculation unit 44 selects 23 ° C. in step S8 of FIG. 6 and calculates it as the optimum value Auto of the air conditioning set temperature A (“Result” in FIG. 10). : Indicated by "Selection").

The optimum value of 23 ° C. is the air-conditioning set temperature A at which the power consumption TGT1 after 1 hour is the smallest within the allowable range (comfort) of the discomfort index TGT2 after 1 hour. Next, the data transmission unit 46 transmits the optimum value Auto of the air conditioning set temperature A to the air conditioner control unit 32 in step S9 of FIG.

When the air conditioner control unit 32 is in the predictive control mode described above, the air conditioner control unit 32 changes the air conditioning set temperature A to the received optimum value Auto. Then, the operation of the air conditioner 19 is controlled by setting this optimum value Auto as the air conditioning set temperature A. The optimum value calculation unit 44 calculates the optimum value Outlook every time the prediction unit 42 predicts, that is, periodically.

(6-4) Creation of prediction model (Part 2)
Next, using FIGS. 11 and 12, the prediction model for predicting the power consumption TGT1 after 1 hour and the prediction for predicting the discomfort index TGT2 after 1 hour by the decision tree analysis which is one of the nonlinear regression analyzes. An example in which a model is created and optimization control is performed will be described.

In this case, the prediction model creation unit 38 of the optimization control unit 31 of FIG. 4 has the latest (also for the latest one day) outside air temperature T to T48, air conditioning set temperatures A to A48, and power consumption P in the database 37A. Based on the data of ~ P48 and discomfort indexes D to D48, the prediction model of the decision tree analysis that predicts the power consumption TGT1 after a predetermined time (also 1 hour) by machine learning, and the same after a predetermined time (1 hour later). ) Create a prediction model for decision tree analysis that predicts the discomfort index TGT2.

The prediction model creation unit 38 determines the difference between the power consumption TGT1 after 1 hour and its actual value TGT1real predicted in all the rows of FIG. 9, and the difference between the discomfort index TGT2 after 1 hour and its actual value TGT2real. A decision tree having the smallest value is created as shown in FIGS. 11 and 12. FIG. 11 is a prediction model of the power consumption P that predicts the power consumption TGT 11 after 1 hour by the decision tree analysis, and FIG. 12 is a prediction model of the discomfort index D that predicts the discomfort index TGT2 after 1 hour by the decision tree analysis.

(6-5) Optimization Control in the Cases of FIGS. 11 and 12 Next, the prediction unit 42 of the prediction calculation unit 41 of the optimization control unit 31 applies a plurality of air conditioner set temperatures A to the prediction model of FIG. Then, a plurality of power consumption TGT1s after 1 hour are predicted, and a plurality of air conditioner set temperatures A are also applied to the prediction model of FIG. 12, and a plurality of discomfort indexes TGT2 after 1 hour are predicted.

In this example, assuming that 21 ° C, 22 ° C, 23 ° C, and 24 ° C are applied as the air conditioning set temperature A, in the prediction model of FIG. 12, when the air conditioning set temperature A is 22 ° C, 23 ° C, and 24 ° C. The discomfort index TGT2 after 1 hour becomes comfortable, and in the prediction model of FIG. 11, when the air conditioning set temperature is set to 23 ° C., the power consumption TGT1 after 1 hour becomes the smallest 3 kW. Therefore, also in the case of this embodiment, the optimum value calculation unit 44 calculates 23 ° C. as the optimum value Auto of the air conditioning set temperature A, and the data transmission unit 46 transmits the data to the air conditioner control unit 32.

As described in detail above, the store air conditioner system 24 of the present invention includes an air conditioner control unit 32 that controls the operation of the air conditioner 19 that air-conditions the inside of the store 1 based on a predetermined air conditioner set temperature A, and at least an air conditioner. It has a database 37A constructed by accumulating data on an index (power consumption P) indicating energy consumption of 19 and the refrigerating device 21 and data on an index (discomfort index D) indicating comfort in the store 1. By performing machine learning based on the database 37A, the air conditioner set temperature A at which the index (TGT1) indicating the energy consumption after the predetermined time is the smallest within the allowable range of the index (TGT2) indicating the comfort after the predetermined time. A predictive control mode in which an optimization control unit 31 for calculating the optimum value opt of the above is provided, and the operation of the air conditioner 19 is controlled by setting the optimum value opt calculated by the optimization control unit 31 as the air conditioner set temperature A in the air conditioner control unit 32. The air conditioner set temperature A is automatically set to the optimum value Opt that can minimize the energy consumption (power consumption P) of the air conditioner 19 and the refrigerating device 21 within the range where the comfort in the store 1 is acceptable. It becomes possible to control the operation of the air conditioner 19 by adjusting to. As a result, the energy consumption (power consumption P) of the air conditioner 19 and the refrigerating device 21 can be reduced while ensuring the comfort in the store 1.

Further, the air conditioner control unit 32 is provided with a performance acquisition mode for switching a plurality of air conditioner set temperatures A to control the operation of the air conditioner 19, and the optimization control unit 31 is in the performance acquisition mode and the prediction control mode. Since the acquired data on the index indicating energy consumption (power consumption P) and the data on the index indicating comfort (discomfort index D) are accumulated and updated in the database 37A, various data in the store 1 are accumulated. The actual data can be accumulated in the database 37A, and the operation with high accuracy can be performed in the predictive control mode. In addition, by comparing the actual results of driving in the actual results acquisition mode and the predictive control mode, it becomes easier to confirm the effect of the predictive control mode.

Further, in the embodiment, the air conditioner control unit 32 is provided with an independent control mode for controlling the operation of the air conditioner 19 according to the air conditioner set temperature A set by using the predetermined input device 35, so that the user can perform arbitrary air conditioning. The air conditioner 19 can be controlled at the set temperature A, and the optimization control unit 31 obtains data and comfort regarding at least an index (power consumption P) indicating energy consumption acquired in the independent control mode. Since the data related to the indicated index (discomfort index D) is also accumulated and updated in the database 37A, it is possible to further diversify the data accumulated in the database 37A.

Further, if the air conditioner control unit 32 switches and executes the actual result acquisition mode, the predictive control mode, and the independent control mode as in the embodiment, the data in the database 37A can be automatically diversified. At the same time, for example, when using the database 37A constructed in another store, the air conditioner 19 is operated while interweaving the predictive control mode with the actual result acquisition mode and the independent control mode, and the actual data is stored in the database 37A. It will be possible to build a database 37A unique to the store by accumulating and updating it.

Then, as in the embodiment, the database construction unit 37 shows at least the outside air temperature T, the air conditioner set temperature A, the power consumption P as an index showing the energy consumption of the air conditioner 19 and the refrigerating device 21, and the comfort in the store 1. A database 37A is constructed by accumulating the discomfort index D as an index for a predetermined period, and the prediction model creation unit 28 performs machine learning based on this database 37A, and at least the latest outside air temperature T, air conditioning set temperature A, and power consumption. A prediction model for predicting the power consumption TGT1 and the discomfort index TGT2 after a predetermined time is created from P and the discomfort index D, and the prediction unit 42 applies a plurality of air conditioning set temperatures A to this prediction model. A plurality of power consumption TGT1 and discomfort index TGT2 after a predetermined time are predicted, and the optimum value calculation unit 44 allows the discomfort index TGT2 after a predetermined time from the predicted power consumption TGT1 and discomfort index TGT2 after a plurality of predetermined time. If the optimum value Aopt of the air conditioning set temperature A at which the power consumption TGT1 becomes the smallest after a predetermined time is calculated within the range that can be achieved, in the predictive control mode, the discomfort index D in the store 1 is within an acceptable range. The air conditioner set temperature A can be automatically adjusted to the optimum value Auto that can minimize the power consumption P of the air conditioner 19 and the refrigerating device 21, and the operation of the air conditioner 19 can be controlled. As a result, the power consumption of the air conditioner 19 and the refrigerating device 21 can be accurately reduced while ensuring the comfort in the store 1.

In this case, if the prediction model creation unit 38 periodically creates and updates the prediction model as in the embodiment, the optimum value Opt of the air conditioning set temperature A can be accurately adjusted according to the change in the environment surrounding the store 1. Will be able to be calculated.

For machine learning in the prediction model creation unit, either linear regression analysis as in the examples, nonlinear regression analysis, or a combination thereof can be considered.

Further, as in the embodiment, the prediction unit 42 periodically predicts a plurality of power consumption TGT1 and discomfort index TGT2 after a predetermined time, and the optimum value calculation unit 44 calculates the optimum value Auto every time the prediction unit 42 predicts. By doing so, it becomes possible to quickly respond to changes in the situation in the store 1 and accurately calculate the optimum value opt of the air conditioning set temperature A.

(7) Other Examples of the Optimization Control Unit Next, FIG. 13 shows another configuration example of the optimization control unit 31 constituting the store air-conditioning system 24 of the present invention. In the embodiment of FIG. 4, all the functions of the optimization control unit 31 are provided on the store 1 side, but in this embodiment, the prediction model creation unit 38 of FIG. 4 is the server of the management company that manages the store 1. It is provided on the 47 side.

That is, in the store air conditioning system 24 in this case, the optimization control unit 31A on the store 1 side does not create a prediction model, but transmits the database 37A constructed by the database construction unit 37 on the optimization control unit 31A side to the server 47. To do. Then, the prediction model creation unit 38 of the server 47 creates a prediction model (reference numeral 39) and transmits it to the prediction calculation unit 41 of the optimization control unit 31A.

The prediction unit 42 of the optimization control unit 31A uses this prediction model to predict the power consumption TGT1 and the discomfort index TGT2 after one hour, calculate the optimum value opt of the air conditioner set temperature A, and use the data transmission unit 46. It will be transmitted to the air conditioner control unit 32. If the prediction model is created on the server 47 side in this way, there is an advantage that the control on the store 1 side is simplified.

(8) Other configurations In the above-described embodiment, multiple regression analysis is adopted as the linear regression analysis performed by machine learning, but the generalized linear regression analysis and the like can also be used. Further, as for the non-linear regression analysis described above, not only the decision tree analysis shown in other examples but also random forest analysis, neural network analysis, SVM analysis and the like can be used.

Further, in the embodiment, the power consumption TGT1 and the discomfort index TGT2 after 1 hour are predicted by the multiple regression analysis, or the power consumption TGT1 and the discomfort index TGT2 after 1 hour are predicted by the determination tree analysis. Not limited to this, the power consumption TGT1 may be predicted by linear regression analysis, the discomfort index TGT2 may be predicted by nonlinear regression analysis, or vice versa, and so on.

Further, in the above-described embodiment, the database 37A is constructed by accumulating the outside air temperature T, the air conditioner set temperature A, the power consumption P, and the discomfort index D for a predetermined period (1 year). In addition, data regarding the operating factors of the air conditioner 19 and the refrigerating device 21 that affect the power consumption P and the discomfort index D may be additionally provided in the database 37A.

The operating factors to be added in this case include the operation mode of the air conditioner 19 (cooling, heating, dry, etc.), the internal temperature of the showcases 4 and 11, the air volume of the internal circulation fan of the showcases 4 and 11, and the refrigerating device. The target low pressure of 21, the high pressure of the refrigerating device 21, the number of revolutions of the compressor of the refrigerating device 21 (inverter control), the minimum valve opening degree of the expansion valve of the refrigerating device 21, and the operating state of the total heat exchanger 22 can be considered.

Then, in that case, the prediction model creation unit 38 creates a prediction model that predicts the power consumption TGT1 and the discomfort index TGT2 after a predetermined time (1 hour) by adding the added operation factor, and the prediction unit 42 creates a prediction model. By applying a plurality of operating factors added in addition to the air conditioning set temperature A to the prediction model, a plurality of power consumption TGT1 and discomfort index TGT2 after a predetermined time (1 hour) are predicted.

The optimum value calculation unit 44 determines the power consumption TGT1 and the discomfort index TGT2 after a plurality of predetermined times (1 hour) predicted by the prediction unit 42 within a range in which the discomfort index TGT2 after a predetermined time (1 hour) can be tolerated. The optimum value of the added operating factor that minimizes the power consumption TGT1 after an hour (1 hour) is calculated.

If the air conditioner control unit 32 and the refrigerating device control unit 33 control the operation of the air conditioner 19 and the refrigerating device 21 with the optimum value calculated by the optimum value calculation unit 44 as an added operating factor, the outside air temperature T The operating factors that affect the power consumption P and the discomfort index D other than the air-conditioning set temperature A are also controlled by the optimum values, and it is possible to further improve the comfort and reduce the power consumption.

Further, assuming that a showcase is installed in addition to the showcases 4 and 11 in the store 1 of the embodiment, for example, as the above-mentioned operating factor regarding the showcase and the refrigerating apparatus, the data of similar devices are averaged. It is conceivable to combine the values, maximum values, minimum values, and standard deviations into one. By doing so, even if the configuration of the showcase of the store 1 is changed thereafter, the database 37A can be constructed without any trouble, and the power consumption TGT1 and the discomfort index TGT2 can be predicted and the optimum values can be calculated. It will be possible.

Further, in the embodiment, the power consumption P is used as an index indicating the energy consumption of the air conditioner 19 and the refrigerating device 21, but the operating state of the air conditioner 19 and the refrigerating device 21 (for example, the number of revolutions of the compressor) is used as the energy consumption. It may be used as an indicator. However, even in that case, the power consumption detection unit 26 will calculate the power consumption P from the operating state. Further, in the embodiment, the discomfort index D is adopted as an index indicating the comfort in the store 1, but the present invention is not limited to this, and for example, the temperature and humidity itself in the store 1 can be adopted.

Further, in the embodiment, the air conditioner control unit 32 is provided with a performance acquisition mode, a prediction control mode, and an independent control mode, but the invention of claim 1 is not limited to this, and a performance acquisition mode and a prediction control mode are provided. The present invention is also effective in switching between them.

1 Store 4, 11 Showcase 15 Refrigerator 17 Indoor unit 19 Air conditioner 20 Outdoor unit 21 Refrigerator 22 Total heat exchanger 24 Store air conditioning system 26 Power consumption detector 27 Outside air temperature detector 28 In-store temperature detector 29 In-store humidity detector 31 Optimization control unit 32 Air conditioner control unit 33 Refrigeration equipment control unit 34 Data reception unit 35 Input device 36 Discomfort index calculation unit 37 Database construction unit 37A Database 38 Prediction model creation unit 41 Prediction calculation unit 42 Prediction unit 44 Optimum value calculation unit

Claims (9)

In a store air conditioning system that air-conditions the inside of a store where a freezing device including a showcase is installed
An air conditioner control unit that controls the operation of the air conditioner that air-conditions the inside of the store based on a predetermined air-conditioning set temperature A.
At least, it has a database constructed by accumulating data on an index showing the energy consumption of the air conditioner and the refrigerating device and data on an index showing the comfort in the store, and machine learning is performed based on the database. By doing so, an optimization control unit that calculates the optimum value of the air-conditioning set temperature A in which the index indicating the energy consumption after a predetermined time becomes the smallest within an acceptable range of the index indicating the comfort after a predetermined time is provided. Prepare,
The air conditioner control unit
A track record acquisition mode that controls the operation of the air conditioner by switching a plurality of the air conditioner set temperatures A,
It has a predictive control mode for controlling the operation of the air conditioner with the optimum value calculated by the optimization control unit as the air conditioning set temperature A.
The optimization control unit
A store air-conditioning system characterized in that data on at least the index indicating energy consumption and data on the index indicating comfort acquired in the performance acquisition mode and the prediction control mode are stored and updated in the database. The air conditioner control unit
It has an independent control mode that controls the operation of the air conditioner according to the air conditioner set temperature A set by using a predetermined input device.
The optimization control unit
The store air-conditioning system according to claim 1, wherein at least data on an index indicating energy consumption and data on an index indicating comfort acquired in the independent control mode are stored and updated in the database. .. The first or second aspect of the air conditioner control unit is characterized in that the performance acquisition mode and the prediction control mode, or the performance acquisition mode, the prediction control mode, and the independent control mode are switched and executed. Store air conditioning system described in. An outside air temperature detector that detects the outside air temperature T,
The in-store temperature detection unit that detects the temperature inside the store,
The in-store humidity detection unit that detects the humidity in the store,
A power consumption detection unit for detecting power consumption P as an index indicating the energy consumption of the air conditioner and the refrigeration device is provided.
The optimization control unit
A discomfort index calculation unit that calculates a discomfort index D as an index indicating comfort in the store from the temperature and humidity in the store, and
A database construction unit that builds the database by accumulating at least the outside air temperature T, the air conditioner set temperature A, the power consumption P, and the discomfort index D for a predetermined period.
Machine learning is performed based on the database constructed by the database construction unit, and the power consumption TGT1 after a predetermined time is obtained from at least the latest outside air temperature T, air conditioning set temperature A, power consumption P, and discomfort index D. And a prediction model creation unit that creates a prediction model for predicting the discomfort index TGT2,
By applying a plurality of the air conditioner set temperatures A to the prediction model created by the prediction model creation unit, the prediction unit predicts the power consumption TGT1 and the discomfort index TGT2 after a predetermined time.
From the power consumption TGT1 and the discomfort index TGT2 after a plurality of predetermined times predicted by the prediction unit, the air conditioning in which the power consumption TGT1 after a predetermined time is the smallest within an allowable range of the discomfort index TGT2 after a predetermined time. The store air-conditioning system according to any one of claims 1 to 3, further comprising an optimum value calculation unit for calculating the optimum value of the set temperature A. The store air-conditioning system according to claim 4, wherein the prediction model creation unit periodically creates and updates the prediction model. Claim 4 or claim 5 characterized in that the prediction model creation unit performs machine learning by using either linear regression analysis or nonlinear regression analysis, or a combination thereof, and creates the prediction model. Store air conditioning system described in. The prediction unit periodically predicts a plurality of the power consumption TGT1 and the discomfort index TGT2 after a predetermined time, and the optimum value calculation unit calculates the optimum value each time the prediction unit predicts. The store air conditioning system according to any one of claims 4 to 6. The database is additionally provided with data on the operating factors of the air conditioner and / or the freezer that affect the power consumption P and the discomfort index D.
The prediction model creation unit creates a prediction model that predicts the power consumption TGT1 and the discomfort index TGT2 after a predetermined time by adding the added operation factor, and also creates a prediction model.
The prediction unit predicts the power consumption TGT1 and the discomfort index TGT2 after a predetermined time by applying a plurality of the added operating factors in addition to the air conditioning set temperature A to the prediction model.
The optimum value calculation unit receives the power consumption after a predetermined time within a range in which the discomfort index TGT2 after a predetermined time can be tolerated from the power consumption TGT1 and the discomfort index TGT2 after a plurality of predetermined times predicted by the prediction unit. Calculate the optimum value of the added operating factor that makes TGT1 the smallest,
In the predictive control mode, the air conditioner control unit and / or the refrigeration device control unit that controls the operation of the refrigeration device uses the optimum value calculated by the optimum value calculation unit as the added operating factor. The store air-conditioning system according to any one of claims 4 to 7, wherein the operation of the air-conditioning device and / or the refrigeration device is controlled. The added operating factors include the operation mode of the air conditioner, the temperature inside the showcase, the air volume of the circulation fan inside the showcase, the target low pressure of the refrigerating device, the high pressure of the refrigerating device, and the like. The number of revolutions of the compressor of the refrigerating device, the minimum valve opening of the expansion valve of the refrigerating device, the operating state of the total heat exchanger forming a part of the air conditioner, or a combination thereof, or The store air-conditioning system according to claim 8, wherein the store air-conditioning system is all of them.

Images