JPH05346281A - Refrigerator - Google Patents

Abstract

PURPOSE:To maintain the temperature in a refrigerator constant by providing a temperature sensor in the refrigerator, an atmospheric temperature sensor, a timer, a memory, a neural network, a target temperature setter and a controller. CONSTITUTION:A temperature sensor 1 in a refrigerator measures a temperature in the refrigerator, and an atmospheric temperature sensor 2 measures the atmospheric temperature at the outer periphery of the refrigerator. A timer 3 counts a temperature measuring interval time. A memory 4 stores past data of several times measured by the sensor 1, 2 at a predetermined time interval according to a signal from the timer 3. A neural network 5 inputs the temperature in the refrigerator, its variation and the atmospheric temperature, its variation of several times in the past from the memory 4, and predicts a temperature in the refrigerator after a predetermined time. A controller 7 so controls the refrigerator that the temperature in the refrigerator after the predetermined time predicted by the network 5 is not out of a set temperature of a target temperature setter 6. Thus, the temperature in the refrigerator is so controlled as not to be largely deviated from a target.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an optimum controller for controlling the temperature inside a refrigerator of a home.

[0002]

2. Description of the Related Art In a conventional refrigerator, a control device controls an equal refrigeration cycle (for example, turns on and off a power source of a compressor) or blows cold air after a temperature inside the refrigerator has a difference from a set temperature. I was trying to bring the temperature inside the refrigerator close to the set temperature by opening and closing the damper on the mouth.

[0003]

However, the thermal inertia and heat capacity of the refrigerating system of the refrigerator and the internal components of the refrigerator, the objects to be cooled, etc. cannot be ignored, and the effects of opening and closing the door and the outside temperature are not constant. No, for this reason the temperature in the refrigerator is
It fluctuates as shown by the curve b in FIG. 3-, and a large difference occurs between the temperature and the set temperature, making it difficult to maintain a constant temperature.

[0004]

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and sets one or more temperature sensors for measuring the temperature in the refrigerator and a target temperature in the refrigerator. Target temperature setting device, a timer for measuring time, a storage device for storing the temperature in the refrigerator in the past obtained by the temperature sensor and its change, and a learning function to predict the state of the refrigerator in the future And a control device for controlling the refrigeration cycle of the refrigerator.

[0005]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 shows a specific example of the control unit of a refrigerator to which the present invention is applied, FIG. 2 shows the control flow thereof, and FIG.

Referring to FIG. 1, the control unit includes an internal temperature sensor 1, an outside air temperature sensor 2, a timer 3, a storage device 4,
It is composed of a neural network 5, a target temperature setting device 6 and a control device 7.

Inside temperature sensor 1, outside air temperature sensor 2
Is composed of, for example, a thermistor, and the inside temperature sensor 1 measures the inside temperature TR of the refrigerator, and the outside air temperature sensor 2 measures the outside air temperature TO around the outside of the refrigerator.

The timer 3 is composed of, for example, a transmitter and a counter, and measures the temperature measurement interval time.

The storage device 4 is composed of, for example, a semiconductor memory, and is measured by the internal temperature sensor 1 and the outside air temperature sensor 2 at a predetermined time interval (here, 5 minutes) by a signal from the timer 3. Data for the past several times (for example, 100 times here) is stored.

The neural network 5 is, for example, a three-layered feedforward connection type neural network composed of dedicated semiconductor hardware,
The learning algorithm uses backpropagation,
From the storage device 4, the refrigerator internal temperature TR (n) and its change ΔTR (n), and the outside air temperature TO (n) outside the refrigerator and its change ΔTO (n) are respectively past several times (for example, 100 here). (Batch) Input as input, predict the refrigerator internal temperature after 5 minutes, and 5 minutes after the prediction, learn by the process described below based on the predicted temperature and the actual temperature.

The target temperature setting device 6 is composed of a slide switch or the like and is used by a user or the like to set the target temperature.

The control device 7 is composed of a semiconductor memory, a compressor control circuit, and the like, and the temperature in the refrigerator after 5 minutes predicted by the neural network 5 does not deviate from the set temperature obtained from the target temperature setting device 6. Control the refrigerator.

The operation of the control device configured as described above will be described with reference to FIG.

First, the number of measurements is started from 0 (step 201), and the internal temperature TR (0) and the outside temperature TO are measured.
(0) is measured (step 202) and n = 1 is set (step 203).

Next, the inside temperature TR (n) and the outside temperature TO
(N) is measured (step 204) and the difference ΔTR (n) ΔTO (n) from the previous measured value is calculated (step 20).
5).

If 1 is added to n and n is less than 100 times, the process returns to step 204 after waiting for 5 minutes. 10 in the above process
The measured value for 0 times is stored in the storage device 4.

When 100 measured values are stored in the storage device 4, the internal temperature TR of the refrigerator is stored in the neural network 5.
(N) and change in temperature inside the chamber ΔTR (n) and outside temperature To
(N) and the change in outside temperature ΔTo (n) are input 100 times in the past and the temperature TF in the refrigerator after 5 minutes is predicted (step 21).
0), the expected temperature TF is sent to the controller 7 (step 21
1) Based on this, the cooling device is controlled.

After waiting for a predetermined interval of 5 minutes (step 212), the internal temperature TR (n) and the outside air temperature TO (n) are measured (step 213), and the difference ΔTR from the previous measured value is measured.
(N) ΔTO (n) is calculated (step 214), and the internal temperature TR (n) and the internal temperature change ΔTR (n) and the outside temperature To (n) and the outside temperature change in the neural network 5 are calculated. ΔTo (n) is input and the actual internal temperature with respect to the predicted temperature TF is learned using the back propagation method (step 215).

1 is added to n, and thereafter (step 210)
To (step 216) are repeated.

Here, the temperature inside the refrigerator of the future refrigerator is predicted by using the neural network, but the temperature inside the refrigerator is not limited to the temperature inside the refrigerator, and the operating state of the compressor and the rotation of the fan are predicted. You can also do it.

[0021]

As described above, since the neural network 6 continues to learn the past temperature and its change, it becomes possible to accurately predict the future temperature over time. Therefore, as shown by the curve a in FIG. 3, it is possible to control the temperature inside the refrigerator of the future refrigerator so as not to deviate largely from the target.

[Brief description of drawings]

FIG. 1 is a block diagram of a control unit of an embodiment of a refrigerator to which the present invention is applied.

FIG. 2 is a control flow of an embodiment of a refrigerator.

FIG. 3 is a curve a-internal temperature change of the embodiment curve b-internal temperature change of the conventional example.

[Explanation of symbols]

 1 Internal temperature sensor 2 Outdoor temperature sensor 3 Timer 4 Storage device 5 Neural network 6 Target temperature setting device 7 Control device

Claims (2)

[Claims] 1. A one or more temperature sensor for measuring a temperature inside a refrigerator, a target temperature setting device for setting a target temperature in the refrigerator, a timer for measuring time, and a past obtained by the temperature sensor. Is characterized by having a storage device for storing the temperature in the refrigerator and its change, a neural network having a learning function for predicting the future state of the refrigerator, and a control device for controlling the refrigeration cycle of the refrigerator. And a refrigerator. 2. One or more temperature sensors for measuring the temperature inside the refrigerator, one or more temperature sensors for measuring the outside temperature of the refrigerator, and a target temperature setting device for setting a target temperature inside the refrigerator. A timer that measures time, a storage device that stores the temperature in the refrigerator and past changes obtained by the internal temperature sensor and the external temperature sensor, and a learning function that predicts the future state of the refrigerator. A refrigerator having a neural network for controlling a refrigerator and a control device for controlling a refrigeration cycle of the refrigerator.