JPH04256718A - Cleaner - Google Patents

Abstract

PURPOSE:To delicately determine the suction force of a fan motor by using a neuro-fuzzy inference device which has plural operation modes, and is optimized by a learning rule from a detected value of a refuse sensor in the respective operation modes. CONSTITUTION:The cleaner is constituted of a fan motor for sucking refuse, a refuse sensor 1 for detecting the refuse on the floor surface, an operation mode setting means 8 which can designate a control range of suction force of the fan motor 9, and a neuro-fuzzy inference device 10 for selecting the operation mode by the operation mode setting means 8, determining the suction force of the fan motor 9 by an output of the refuse sensor 1, and also, optimizing various parameters of a fuzzy inference by a learning rule of the utmost dive method, etc.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a vacuum cleaner which has a plurality of operating modes and which detects the amount of dirt on a floor surface and automatically adjusts its suction power.

0002

2. Description of the Related Art In recent years, vacuum cleaners have been required to be able to detect the amount of dirt on the floor, the type of the floor, etc., and finely adjust the suction force of the fan motor.

Conventionally, this type of vacuum cleaner has a plurality of operation modes, and the suction power is set according to the amount of dirt, etc.
It was set in stages. Additionally, there are devices that set the suction force depending on the condition of the floor, such as the length of the pile of the floor, tatami mats, or carpet, but these too can only differentiate the condition of the floor into three levels.

0004

[Problem to be solved by the invention] In such conventional vacuum cleaners, the amount of dirt and the condition of the floor surface cannot be set in three to four levels, but change continuously. Since the quality of the garbage is not taken into consideration, there is a problem in that the optimal suction force cannot be obtained for the amount, quality, floor condition, etc. of the garbage in each operation mode.

[0005] The present invention solves the above problems by finely setting the suction force of the fan motor in each operation mode using a neuro-fuzzy inference machine optimized by a learning rule from the detected value of the dust sensor. It is an object.

[0006]

[Means for Solving the Problems] In order to achieve the above object, the present invention provides a fan motor for sucking in dust, a dust sensor for detecting dust on the floor, and a control range of the suction force of the fan motor. An operation mode setting means that can be specified, and an operation mode is selected by the operation mode setting means, and in each operation mode, the suction force of the fan motor is determined based on the output of the garbage sensor, and according to a learning rule such as the steepest descent method. The solution to the problem is that it is equipped with a neuro-fuzzy inference machine that optimizes various parameters of fuzzy inference.

[0007]

[Operation] The present invention uses the above-mentioned problem-solving means to generate a neurotransmitter that is optimized by a learning rule from the output of the garbage sensor.
The suction force of the fan motor can be set using a fuzzy reasoner, and the suction force can be determined in detail, allowing dirt to be removed efficiently regardless of the floor surface to be cleaned.Moreover, it is very easy to operate, and has a silent operation mode. If you select
By setting the maximum suction power to an appropriate value, it is possible to clean quietly.

[0008]

[Embodiment] An embodiment of the present invention will be described below with reference to FIG.

As shown in the figure, the dust sensor 1 is installed on a part of the suction hose 2, and consists of a light emitting part 3 and a light receiving part 4, which are placed opposite each other. I'm trying to pass. Garbage amount calculation means 5
calculates the cumulative amount of garbage sucked in over a predetermined period of time. The dust change rate calculating means 6 calculates the change rate of the sucked-in dust over a predetermined period of time. The garbage quality detection means 7 measures the time taken for the sucked garbage to pass between the garbage sensors 1, and detects the garbage quality. The operation mode setting means 8 selects the operation mode using a switch or the like (not shown) and sets the maximum suction force of the fan motor 9. Neuro-fuzzy reasoner 1
0 infers the suction force of the fan motor 9 from the outputs of the waste amount integrating means 5, the waste change rate calculating means 6, and the waste quality detecting means 7 according to the operating mode selected by the operating mode setting means 8. The control means 11 drives the suction force of the fan motor 9 with the suction force inferred by the neuro-fuzzy inference device 10.

The neuro-fuzzy inference device 10 is constructed as shown in FIG. 2, and the antecedent membership function storage means 12 stores membership functions relating to the amount of garbage, the quality of garbage, and the rate of change of garbage. Waste quality suitability calculation means 1
3. The garbage amount compatibility calculation means 14 and the garbage change rate compatibility calculation means 15 input membership functions related to garbage quality, garbage amount, and garbage change rate stored in the antecedent membership function storage means 12, respectively. The degree of compatibility with the garbage quality, garbage amount, and garbage change rate is calculated. The antecedent part minimum calculation means 16 takes the MIN of the three compatibility values that are the outputs of the waste quality compatibility calculation means 13, the waste amount compatibility calculation means 14, and the waste change rate compatibility calculation means 15, and calculates the MIN of the three compatibility values, which are the outputs of the waste quality compatibility calculation means 13, the waste amount compatibility calculation means 14, and the waste change rate compatibility calculation means 15, and calculates the MIN of the three compatibility degrees, which is the output of the waste quality compatibility calculation means 13, the waste amount compatibility calculation means 14, and the waste change rate compatibility calculation means 15. do. The suction force inference rule storage means 17 stores inference rules regarding suction force. The suction force membership function storage means 18 stores membership functions related to the suction force of the consequent part. The consequent part minimum calculation means 19 calculates the antecedent part conclusion and the suction force members of the consequent part stored in the suction force membership function storage means 18 according to the inference rules stored in the suction force inference rule storage means 17. The MIN of the ship function is taken as the conclusion of the rule. The center of gravity means 20 obtains each conclusion for all the rules, takes the MAX of all the conclusions, calculates the center of gravity, and finally obtains the suction force. This neuro-fuzzy inference device 10 can be easily realized by a microcomputer. Further, the membership functions and inference rules stored in the antecedent membership function storage means 12, the attraction force inference rule storage means 17, and the attraction force membership function storage means 18 included in the neuro-fuzzy inference device 10 are The steepest descent method (one of the learning rules used in neural networks) is calculated in advance from the data on the amount of trash, the rate of change in trash, the quality of trash, and the suction force of the fan motor 9 that should be set in consideration of the operating feel when cleaning. , which is a method that minimizes the error function). The control means 11 calculates and controls the phase control amount of the fan motor 9 based on the set suction force.

Next, to explain the operation of the above configuration, the light emitted from the light emitting part 3 of the dust sensor 1 can be received by the light receiving part 4 when there is no dust, but if dust passes through it, it is blocked and the light cannot be received. Part 4 cannot receive light. Therefore, the presence or absence of dust can be determined from the output of the light receiving section 4. The trash amount integrating means 5 integrates the amount of trash detected by the trash sensor 1 for a certain period of time (for example, 0.1 seconds). By adding up the amount, you can find out the amount of garbage on the floor at that time.

FIG. 3 shows the degree of change in the integrated value of the amount of dust when cleaning is continued. In the same figure, from the start of cleaning until T1, the amount of garbage decreases at once, but
This indicates that the dirt on the floor surface has been removed. Also T
As shown in FIG. 3, from 1 onwards, the steps are roughly divided into FIGS. 3(A), (B), and (C) depending on how the garbage is removed thereafter. In the case of FIG. 3C, the accumulated value of dust is almost 0, indicating that most of the dust has been removed by T1. This is the case when the floor surface to be cleaned is a wooden floor, cushion floor, tatami mat, etc. Additionally, when the floor is a carpet, dirt gets buried between the piles, and generally the amount of dirt is relatively large compared to wooden floors or tatami mats, making it difficult to remove. In other words, as shown in FIGS. 3A and 3B, the integrated value of the amount of garbage gradually decreases. When the rate of change in the amount of garbage is calculated by the rate of change calculation means 6 in this way, it is possible to estimate the characteristics of the floor surface that is currently being cleaned. A small rate of change in the amount of trash indicates that the floor surface is difficult to remove trash from, and a large rate of change in the amount of trash indicates that the floor surface is easy to remove trash from. ing. Figure 4 shows the pulse waveform of the garbage detected by the garbage sensor 1. Figure 4 (A) shows the pulse waveform when cotton garbage is sucked in, and Figure 4 (B) shows the pulse waveform when dust is sucked in. The pulse waveform for the case is shown. Therefore, by detecting this pulse waveform with the garbage quality detection means 7, the quality of the garbage can be determined, such as whether the sucked-in garbage is large and light like cotton garbage or small and heavy like sand garbage. I understand. The optimal suction force for cleaning is determined by the amount of dirt on the floor, the quality of the dirt, the characteristics of the floor, etc., and this is determined by the amount of dirt accumulation means 5, the dirt change rate calculation means 6, and the quality of the dirt. From the output value of the detection means 7, neuro-
Inference is made using the fuzzy inference device 10.

Next, the process of inferring the suction force will be explained. The inference rule of the fuzzy inference in this example is ``If the amount of garbage is large, the garbage is heavy and small, such as sand, and the floor surface is difficult to remove (the rate of change in the amount of garbage is small), the suction force will be increased. It is formed based on general judgments such as "I want more." The amount of garbage is ``large'', the rate of change in the amount of garbage is ``small'', the quality of garbage is ``heavy'', etc.
The qualitative concept of suction force being “very large” is shown in Figures 5 (A), (B), (C) and Figures 6 (A), (B).
, (C). The garbage amount compatibility computing means 14 calculates the compatibility between the input from the garbage amount accumulating means 5 and the membership function regarding the amount of garbage stored in the antecedent membership function storage means 12 by taking the MAX of both. . The garbage change rate compatibility calculation means 15 similarly calculates the degree of compatibility between the input from the garbage change rate calculation means 6 and the membership function of the garbage change rate stored in the antecedent membership function storage means 12. The waste quality compatibility calculation means 13 similarly calculates the compatibility with respect to the input from the waste quality detection means 7 and the membership function of the waste quality stored in the antecedent membership function storage means 12. The antecedent minimum calculation means 16 takes the MIN of the three degrees of fitness and uses it as the conclusion of the antecedent. The consequent part minimum calculation means 19 calculates the antecedent part conclusion and the consequent part stored in the suction force membership function storage means 18 according to any of the rules stored in the suction force inference rule storage means 17. M of the suction force membership function
Take IN as the conclusion of the rule. At this time, the operation mode setting means 8 selects one of the rules (A), (B), and (C) from the suction force inference rule storage means 17, and in each rule, the maximum suction force is Preselected. Note that (A), (B), and (C) are stored in the suction force inference rule storage means 17 as, for example, a strong mode, a normal mode, a quiet mode, etc., respectively.

After determining the respective conclusions for all the rules, the center of gravity calculation means 20 takes the MAX of all the conclusions and calculates the center of gravity to finally determine the suction force. The control means 11 calculates and controls the phase control amount of the fan motor 9 based on the determined suction force. In this example, the MAX-MIN combination method and the center of gravity method are used as inference methods, but other methods are also possible, and the suction force, which is the consequent, is expressed by a membership function, but in reality It goes without saying that it can also be expressed numerically or in linear form.

[0015]

As is clear from the above embodiments, the present invention includes a fan motor for sucking in dust, a dust sensor for detecting dust on the floor, and a control range of the suction force of the fan motor. an operation mode setting means that can be specified; and selecting an operation mode using the operation mode setting means;
Equipped with a neuro-fuzzy inference device that determines the suction force of the fan motor based on the output of the dust sensor in each operation mode and optimizes various parameters for fuzzy inference using learning rules such as the steepest descent method, Since we have optimized the configuration of the neuro-fuzzy inference machine using learning rules such as laws, even if the number of inputs in fuzzy inference increases, it is easy to optimize the inference rules between them and their configuration. Using fuzzy inference based on the amount of trash, quality of trash, and rate of change in trash, the optimal suction force can be determined in detail based on the desired operating mode, allowing efficient removal of trash regardless of the floor surface to be cleaned or the type of trash. It is very easy to operate, and if you select the silent operation mode, the suction force is kept low for quiet cleaning.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a vacuum cleaner according to an embodiment of the present invention.

[Figure 2
] Block diagram of the vacuum cleaner's neuro-fuzzy inference device

[Figure 3] Characteristic diagram showing changes in the amount of dust of the same vacuum cleaner

[Figure 4]
(A), (B) Diagram showing the waveform of the garbage quality of the same vacuum cleaner

[Figure 5] (A), (B), (C) Diagram showing the membership function of the same vacuum cleaner

[Figure 6] (A), (B), (C) Diagram showing the membership function of the same vacuum cleaner

[Explanation of symbols]

1 Garbage sensor 8 Operation mode setting means 9 Fan motor 10 Neuro-fuzzy inference machine

Claims (3)

[Claims] 1. A fan motor for suctioning dust, a dust sensor for detecting dust on the floor, an operation mode setting means for specifying a control range of the suction force of the fan motor, and an operation mode set by the operation mode setting means. A neuro-fuzzy inference machine that selects a mode, determines the suction force of the fan motor based on the output of the dust sensor in each operation mode, and optimizes various parameters for fuzzy inference using learning rules such as the steepest descent method. Vacuum cleaner with. 2. The cleaning device according to claim 1, wherein the neuro-fuzzy reasoner recognizes the amount of garbage on the floor, the quality of the garbage, and the type of the floor from the output of the garbage sensor, and determines the suction force of the fan motor. Machine. 3. The vacuum cleaner according to claim 1, wherein the neuro-fuzzy inference device optimizes the shapes of the antecedent membership function and the consequent membership function and the number of inference rules as various parameters.