JPH04336008A - Rice cooker - Google Patents

Abstract

PURPOSE:To provide a rice cooker by which boiled rice of the same boiled condition can be always obtained by causing a temperature rise conforming to the rice-boiling theory regardless of the amount of rice and the dispersion in water quantity. CONSTITUTION:A rice cooker comprises a deviation judging part 9 for detecting the difference between the target temperature being output from a target temperature producing part 8 at given intervals that have been predetermined by a time-measuring means 7 and the output of a rice-cooking sensor 4, and a change-rate judging part 10 for detecting the difference in the output of the rice-cooking sensor 4 between before a prescribed time and the present; and it is also provided with a control part 6 in which the electric current to be passed to a rice cooking heater 2 is determined by a neural network 11 on the basis of the two inputs sent both from the deviation judging part 9 and from the change-rate judging part 10.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a rice cooker that utilizes neurocontrol to control the amount of heating during rice cooking and is used for the purpose of stably providing delicious rice under any circumstances.

0002

BACKGROUND OF THE INVENTION Conventionally, the heating amount of a rice cooker has been generally controlled by detecting the output of a rice cooking sensor during heating by a rice cooker, and using a microcomputer to determine the amount of heating. As this method, the special public
048, JP-A-59-57616, JP-A-59-155222, etc. utilize the principle that when heating with a constant electric power, the smaller the amount of rice cooked, the more rapidly the temperature rise of the rice cooking sensor increases. The first technology to solve these problems and the Special Publication No. 60-1007 invented to improve these problems.
The second technical idea is to control the amount of heating while comparing the target temperature and the rice cooking sensor output, as disclosed in Japanese Patent Application Laid-Open No. 60-18128. The present invention is a further development of the second technique.

0003

[Problems to be Solved by the Invention] Rice cooking is the process of heating rice, and theory has determined the ideal way to raise the temperature over time. However, in the first technique, the amount of rice cooked is determined by the temperature rise when heating with a constant electric power, so the temperature rises in a shorter time when the amount of rice cooked is small, which contradicts the theory. There are different problems.

[0004] In order to improve this problem, in the second technique, the rice cooking heater is controlled by comparing the generated target temperature with the output of the rice cooking sensor. If the rice cooking sensor accurately detects the temperature of the rice, the smaller the amount of rice cooked, the less power is required, and the temperature rise characteristics as per theory are always achieved regardless of the amount of rice cooked. However, since the rice cooking sensor is in contact with the inner pot and the inner pot is heated by the rice cooking heater, it actually detects a temperature different from the temperature of the rice. When rice is heated with a rice cooker, the temperature of the rice sensor increases, and when heating is stopped, the temperature decreases. Therefore, simply controlling the temperature of the rice cooking sensor to determine whether it is higher or lower than the target temperature inevitably results in a deviation from the actual water temperature, making it difficult to heat the rice according to theory regardless of the amount of rice cooked.

[0005]

[Means for Solving the Problems] The present invention has been made to solve the above-mentioned problems. a deviation determination unit that detects the
It has a change rate judgment section that detects the difference in the output of the rice cooking sensor between a certain time ago and the current time, and uses a neural network to determine the output to the rice cooking heater based on the two inputs from the deviation judgment section and the change rate judgment section. It is equipped with a control section that determines the amount of energization.

[0006] In addition to the inputs from the deviation degree judgment section and the change rate judgment section, it is also effective to use inputs such as room temperature, lid temperature, voltage, menu, hardness preference selection signal, etc. .

Furthermore, it is also effective to provide a side heater and a lid heater in addition to the rice-cooking heater, and to have the amount of current applied to these heaters determined by the output of a neural network.

[0008]

[Operation] According to the configuration of the present invention, it is possible to check not only the degree of deviation between the target temperature and the rice cooking sensor output, but also whether the temperature of the rice cooking sensor is currently rising or falling due to the heating up to this point. The neural network also uses the [rate of change] to determine whether the slope is slow or steep. For example, if the [degree of deviation] is lower than the target temperature on the rice cooking sensor, it can be determined that the amount of heating up to this point is small, and the amount of heating increase. However, if the rate of change of the rice cooking sensor suddenly increases,
Conversely, even if the amount of heating is reduced, the rice cooking sensor predicts that the temperature will be higher than the target temperature after a certain period of time, and the amount of heating will be reduced. In addition, once a neural network has been trained to judge based on the degree of deviation and rate of change,
In addition to "misalignment degree" and "rate of change," the system also uses unlearned inputs such as room temperature, lid temperature, voltage, menu, hardness preference selection signal, etc. to complement the learned pattern appropriately. Determine the amount of electricity that will result in a more appropriate amount of heating.

[0009] Furthermore, neural networks do not require laws that can be formulated into equations between various combination patterns; if the neural network has learned the situation that occurs when rice is cooked by energizing the rice cooking heater, it can be used to Appropriate control that is more applicable to the actual situation is also carried out when energizing.

0010

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below in detail with reference to the drawings. FIG. 1 is a sectional view of a rice cooker according to an embodiment of the present invention, FIG. 2 is a configuration diagram of a heating control system of the rice cooker, and FIG. 3
These are diagrams showing an example of changes over time in heating control by the rice cooker, where (a) shows a case where the amount of cooked rice is large, and (b) shows a case where the amount of cooked rice is small. FIG. 4 is a diagram showing an example of a neural network,
FIG. 5 is a diagram showing neuron elements forming a neural network. In FIG. 1, a rice cooking heater 2 is attached to the bottom inside a main body 1, an inner pot 3 is detachably inserted into the main body 1, and the inner pot 3 is positioned on top of the rice cooking heater 2. A rice cooking sensor 4 attached to the main body 1 comes into contact with the inner pot 3 and is brought into close contact with the inner pot 3 by the force of a spring 5. A control section 6 is provided inside the main body 1, and a rice cooking sensor 4 and a rice cooking heater 2 are connected to the control section 6.

As shown in FIG. 2, the control section 6 includes a clock means 7 that measures a certain period of time, a target temperature generation section 8 that generates a target temperature that the rice cooking sensor 4 should reach in accordance with the timing of the clock means 7, and a target A discrepancy determining unit 9 calculates the difference (difference) between the target temperature output from the temperature generating unit 8 and the rice cooking sensor 4, stores the output of the rice cooking sensor 4 at regular intervals in the timer 7, and compares the stored value with the current one. A rate of change judgment unit 10 that compares the output values of the rice cooking sensor to determine the rate of change over a certain period of time, a neural network 11 that receives the deviation judgment unit 9 and the rate of change judgment unit 10 as inputs, and a heating amount that is the output of the neural network 11. A heating control unit 12 that receives the heating control unit 12 and an energization control unit 13 that controls energization of the rice cooking heater 2 based on a signal from the heating control unit 12 are provided.

Before explaining its operation step by step, neurocontrol will be briefly explained. Neurocontrol is shown in Figure 4.
By inputting numerical values from the input layer (left side) of the neural network 11 having the structure shown in Figure 1, the numerical values are transmitted between the networks, and some result is output to the output layer (right side). The network consists of a plurality of neuron elements and signal lines connecting them, and one neuron element has the configuration shown in FIG. 5.

The function of the neuron element will be explained with reference to FIG. The neuron element has a plurality of input terminals and one output terminal, and the signal line connected to the output terminal is connected to the output terminal of each neuron element in the upper layer. Each signal line has a transmission coefficient (weighting: W). This means that the signal is amplified and attenuated before it is transmitted, and the signal I is transmitted W times as much. In this way, the value obtained by adding the threshold value Ws to the sum of the values transmitted to each input terminal and processing it using a sigmoid function becomes the output of the neuron element.

[0014] If the above processing is expressed as an arithmetic expression,

0015
]

[Math 1]

Now, the values of the inputs I1 to In are arbitrarily determined. From the equation, it can be seen that when the weighting: W and the threshold value: Ws are changed, the output Z also changes. Z as it should be for input
Learning is the process of manipulating W and Ws so that the value of . In reality, a network is created as shown in Figure 4, and various input patterns (I1
˜In correlation), the computer learns to always output the correct answer at that time. Therefore, the number of input patterns to be learned is large, and the relationship with the correct answer output is nonlinear and has no regularity, so the number of neuron elements constituting the network is required to be large. Once W and Ws are determined, a solution can be found from the equation even for input patterns that have not been trained. Furthermore, as described above, the neural network expresses the learning results by manipulating the values of W and Ws. Since each value of W and Ws has no meaning, the neural network will be treated as a black box below.

Generally speaking, a neurocomputer has the function of manipulating the values of W and Ws through learning, but the computer used to control the rice cooker as in the present invention is equipped with a neural network based on the results of learning. Since the values of W and Ws do not change, it does not mean that the network becomes smarter the more you use it, but if the network is the result of sufficient learning, it will be possible to control the network by taking advantage of the effects of neural control.

The operation will be explained below. When a user fills the inner pot 3 with rice and water, inserts it into the main body 1, and operates a rice cooking switch (not shown), rice cooking begins. At the start of rice cooking, the output of the rice cooking sensor 4 is first input to the target temperature generating section 8 and the rate of change determining section 10. The target temperature generation unit 8 generates a target temperature to be reached after a certain period of time based on the inputted initial output of the rice cooking sensor 4. This target temperature may be read from table data in which an ideal rice cooking curve is stored, or a constant value that increases along the rice cooking curve may be added.

On the other hand, the rate of change determining section 10 stores the input rice cooking sensor output as an initial value. For example, if the output of the rice cooking sensor 4 is 20°C, 20°C is stored in the rate of change determination unit 10, and the target temperature is 24°C, which is 20°C plus 4°C.

Next, the energization control means 13 operates in response to a signal from the heating control section 12, and at the same time as energization of the rice cooking heater 2 is started, the clocking means 7 starts measuring time. In this embodiment, the energization control means 13 is constituted by a relay, and the amount of heating is controlled by changing the ratio of ON time and OFF time during a certain period. For the first certain period of time, heating is performed at a predetermined constant heating amount. For example, if the time measurement by the timer 7 is 1 minute, the initial constant heating is 30 seconds ON and 30 seconds OFF.
FF. When the clocking means 7 finishes counting one minute, the output of the rice cooking sensor 4 is detected by the deviation determining unit 9 and the rate of change determining unit 1.
0, and the [deviation degree] which is the difference between the target temperature and the output of the rice cooking sensor 4 and the [change rate] which is the difference in the output of the rice cooking sensor 4 from a certain time ago and now are calculated.

At the same time, the temperature stored in the rate of change determining section 10 is rewritten as the output of the rice cooking sensor 4. For example 1
If the rice cooking sensor 4 reaches 23°C after heating for 1 minute, the "degree of deviation" is -1 of the difference from the target temperature of 24°C.
℃, "rate of change" is +3℃ difference from the stored 20℃
It is. After determining the rate of change, 23°C is stored in the storage unit of the rate of change determining unit 10 that previously stored 20°C. Neural network 1
1, output will be made according to the learning result. This output may be learned so as to directly output the next heating amount, or may be learned so as to output a correction value for correcting the electric energy based on the current heating result.

If this embodiment outputs a correction value, for example, +5 is output from the input of the network. Upon receiving this output, the heating control unit 12 controls the heating amount 3.
At the same time that 5 seconds is added to 0 seconds and heating control is started for 35 seconds ON and 25 seconds OFF, the timer 7 is reset and starts counting 1 minute again. Further, the target temperature generation unit 8 generates 28° C., which is 4° C. added to the previous target temperature of 24° C., as the target temperature for the next one minute.

[0023] Suppose that one minute later, the temperature of the rice cooking sensor 4 is 30°C. Using the same procedure, the deviation is +2°C from 28°C.
The rate of change is +7°C compared to 23°C. In this case, since the temperature is higher than the target temperature and the rate of change is rapidly increasing, it can be predicted that the deviation from the target will become even larger after the next minute unless the next power is made extremely low. Since this situation is also learned as an input pattern, the output from the neural network 11 will be -25, for example. Heating amount 3
10, which is 5 seconds plus -25, is the next heating amount, which is 10 seconds.
Heating is performed with N and OFF for 50 seconds.

[0024] Thereafter, the amount of heating can be similarly controlled by increasing the target temperature at regular intervals. Although the actual water temperature/rice temperature and the output temperature of the rice cooking sensor 4 are different, there is a correlation depending on the amount of rice cooked. When the amount of rice to be cooked is small, the target temperature can be reached with a small amount of heating. When the amount of rice to be cooked is large, the amount of heat escaping from the inner pot 3 to the water/rice will be larger if the amount of heating is small, so the output of the rice cooking sensor 4 will either not reach the target temperature or will drop even if it has reached the target temperature. The result is that the amount of heating is injected.

Furthermore, if the target temperature is fixed at a constant temperature for a while and then raised again, the amount of heating during the period of fixation becomes the minimum amount of heating necessary to maintain that temperature, making it possible to cook rice efficiently by keeping it lukewarm. It is also possible to achieve a ``preheating'' effect that allows the rice to absorb water well.

As described above, in this embodiment, the rice cooking sensor 4 is controlled to rise in accordance with the target temperature generated by the target temperature generating section 8. The neural network 11 is made to learn not only the binary comparison but also the degree of increase in the rice cooking sensor 4 due to overheating and the degree of decrease in the rice cooking sensor 4 due to insufficient heating, so that rice is always cooked regardless of the amount of rice cooked. This makes it possible to ideally raise the temperature in accordance with theory. As a result, you can always get the same finished rice. Since the output of the neural network 11 is interpolated from input patterns other than those learned, even small differences in the amount of water can cause the amount of heating to change minutely, ensuring that the same finished rice is always obtained.

Further, the neural network 11 is provided with all or one of the following information: [room temperature], [lid temperature], [voltage], [menu], and [hardness preference selection signal], which the customer selects with a switch or the like according to his/her preference. By also entering the
By learning the optimal output based on these input conditions, it becomes possible to control the amount of heating according to the input. Even more stable,
We can provide a finished product that meets the customer's preferences.

Further, in this embodiment, only the heating amount of the rice cooking heater 2 is output from the neural network 11, but as another embodiment, as shown in FIG.
If a lid heater 2b and a lid heater 2b are provided and the amount of heating thereof is controlled by the output of the network, it is possible to heat the inner pot evenly from the entire circumference, and the balance of these heatings can also be controlled by the learning of the network.

[0029]

[Effects of the Invention] The present invention uses a neural network to determine the amount of electricity to be applied to the rice cooking heater based not only on the degree of deviation between the target temperature and the rice cooking sensor output, but also on the rate of change in the output of the rice cooking sensor from a certain time ago to the current time. Because we decided to let you decide, regardless of the amount of rice being cooked, the ideal temperature rise that matches the rice cooking theory can be achieved, and the rice can always be cooked to the same degree.Furthermore, even slight differences in the amount of water can cause the amount of heating to change minutely, making it always possible to This has the effect of allowing you to cook rice with the same result.

[0030] A similar effect can be obtained using a neural network that determines the amount of electricity to be applied to the rice cooking heater by inputting the room temperature, lid temperature, voltage, menu, hardness preference selection signal, etc. in addition to the degree of deviation and rate of change. A similar effect can also be obtained by installing a side heater on the side of the main body and a lid heater on the lid side, and controlling the amount of heating by a neural network.

[Brief explanation of drawings]

FIG. 1 is a sectional view of a rice cooker showing an embodiment of the present invention.

FIG. 2 is a configuration diagram of the heating control system of the rice cooker.

FIG. 3 is a diagram showing an example of heating control of the rice cooker. (a) When the amount of rice cooked is large. (b) When the amount of cooked rice is small.

FIG. 4 is a diagram showing an example of a neural network.

FIG. 5 is a diagram showing neuron elements forming a neural network.

FIG. 6 is a sectional view of a rice cooker according to another embodiment.

[Explanation of symbols]

2　Rice cooking heater 2a Side heater 2b Lid heater 4 Rice cooking sensor 6 Control section 7 Timekeeping means 8 Target temperature generation section 9 Misalignment judgment section 10 Change rate judgment section 11 Neural network

Claims (3)

[Claims] [Claim 1] A main body, an inner pot housed in the main body, a rice cooking sensor provided in the main body and in close contact with the inner pot to detect the temperature of the inner pot, and a rice cooking heater provided in the main body to heat the inner pot. In a rice cooker that has a timer that measures a certain period of time and a target temperature generation section that generates ideal temperature characteristics of the rice cooking sensor as rice cooking progresses, the timer (7) measures the time at a predetermined time interval. a deviation determining unit (9) that detects the difference between the target temperature output from the target temperature generating unit (8) and the output of the rice cooking sensor (4), and a rice cooking sensor (4) between a certain time ago and the current time. and a change rate determining unit (10) that detects the output difference between
Based on the two inputs from 9) and the change rate judgment unit (10), the neural network (11) controls the rice cooking heater (
2) A rice cooker comprising a control section (6) that determines the amount of electricity supplied to the rice cooker. [Claim 2] The neural network (11) receives inputs from the misalignment determining unit (9) and the rate of change determining unit (10) as well as inputs such as [room temperature], [lid temperature], [voltage], [menu], and [hardness]. Enter the rice cooking heater (
2. The rice cooker according to claim 1, wherein the amount of electricity applied to the rice cooker is determined. [Claim 3] Apart from the rice cooking heater (2), there is also a side heater (2a) on the side of the main body, and a lid heater (2b) on the lid of the main body.
), and the amount of energization to the side heater (2a) and the amount of energization to the lid heater (2b) are determined by the output of the neural network (11).
Or the rice cooker according to claim 2.