JP2897405B2 - Vacuum cleaner - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a vacuum cleaner that detects the amount of dust on the floor and automatically adjusts the suction force.

2. Description of the Related Art Conventionally, there has been a vacuum cleaner of this type, but the suction force set according to the amount of dust or the like is set at about four levels. In some cases, the suction force is set according to the floor condition such as the floor, the tatami mat, and the length of the fur of the carpet. However, the condition of the floor surface can be discriminated only in three stages.

Problems to be Solved by the Invention However, the conventional techniques as described above have the following problems. That is, the amount of dust and the state of the floor surface cannot be set in any of three to four stages, but change continuously, and the suction force of the fan motor set thereby should also be multistage. At the same time, the number of revolutions of the floor nozzle must also be controlled. Further, in setting the suction force and the number of revolutions of the floor nozzle, not only the amount of dust and the material of the floor surface but also the quality of dust should be taken into consideration.

Accordingly, the present invention is to solve such a conventional problem, and uses a neuro-fuzzy inferencer optimized by a learning rule based on a detection value of a dust sensor and a suction force of a fan motor and a rotation speed of a floor nozzle. It is an object of the present invention to provide a vacuum cleaner that determines the details of the vacuum cleaner.

Means for Solving the Problems A vacuum cleaner according to the present invention for achieving the object described above has a fan motor for sucking dust, a floor nozzle for scraping dust on the floor, and detecting sucked dust on the floor. Fuzzy inference based on a dust sensor and a plurality of pieces of information obtained from the output of the dust sensor, determine the number of revolutions of the fan motor and the floor nozzle, and a membership function and a consequent part of the fuzzy inference. function,
It has a neuro-fuzzy inference device that optimizes the inference rules by learning rules such as the steepest descent method.

Function As described above, the membership function of the fuzzy inference device that handles a plurality of pieces of information obtained from the dust sensor is optimized by a learning rule, and the suction force of the fan motor is calculated by the neuro-fuzzy inference device that optimizes the membership function and the like. And the number of revolutions of the floor nozzle is set, and the suction force and the number of revolutions of the floor nozzle are determined finely, and dust can be efficiently removed regardless of the floor surface to be cleaned, and the cleaning is very easy to operate. Machine can provide.

Embodiment An embodiment of the present invention will be described below with reference to the drawings. In FIG. 1, reference numeral 1 denotes a dust sensor provided on a part of the suction hose 9, which is composed of a light emitting unit 11 and a light receiving unit 12 installed so as to face each other. It is supposed to. Reference numeral 2 denotes a dust amount integrating means for calculating an integrated amount of sucked dust for a predetermined time. Reference numeral 3 denotes a dust change rate calculating unit that calculates a change rate of the sucked dust in a predetermined time. Reference numeral 4 denotes a dust quality detection unit that measures the time that the sucked dust passes between the dust sensors 1 to detect dust quality. Numeral 5 is a neuro-fuzzy inference device, and the garbage amount integrating means 2 and the garbage change rate calculating means 3
Then, the suction force of the fan motor 7 and the rotation speed of the floor nozzle 8 are inferred from the output of the dust detection means 4. Numeral 6 denotes a control means for driving the fan motor 7 and the floor nozzle 8 from the inferred values.

The above-described neuro-fuzzy inference device 5 has a configuration as shown in FIG. That is, reference numeral 20 denotes an antecedent part membership function storage means, which stores membership functions relating to the amount of dust, dust quality, and dust change rate. 21, 22,
Reference numeral 23 denotes garbage quality suitability calculating means, garbage quantity suitability calculating means, and garbage change rate suitability calculating means. The garbage quantity, garbage quality,
The degree of conformity between the membership function relating to the dust change rate and the input dust amount, dust quality, and dust change rate is calculated. Reference numeral 24 denotes an antecedent part minimum operation means. The MIN of the three conformances, which are the outputs of the garbage quality adaptation operation means 21, the garbage quantity adaptation operation means 22, and the garbage change rate adaptation operation means 23, are taken. And Numeral 28 is a suction force inference rule storage means for storing inference rules relating to suction force. 26 is a suction force membership function storage means, which stores a membership function relating to the suction force of the consequent part. Reference numeral 25 denotes a consequent part minimum calculating means, which in accordance with the inference rules stored in the suction force inference rule storage means 28, sucks the consequent part conclusion and the suction of the consequent part stored in the suction force membership function storage means 26. Take the MIN of the force membership function as the conclusion of the rule. 27 is a center of gravity calculation means,
After obtaining the conclusions of all the rules, the MAX of all the conclusions is obtained, and the center of gravity is calculated, thereby finally obtaining the suction force. To determine this suction force,
Floor nozzle rotation number inference rule storage means 30 for storing an inference rule for inferring the rotation number of floor nozzle 8, and floor nozzle rotation number storage means for storing a membership function relating to the rotation number of floor nozzle 8. 29 is also involved.

The neuro-fuzzy inference unit 5 includes an antecedent membership function storage unit 20, a suction force inference rule storage unit 28, and a suction force membership function storage unit 26 included in the neuro-fuzzy inference unit 5 which can be easily realized by a microcomputer. The membership function and the inference rule stored in the floor nozzle rotation number inference rule storage means 30 and the floor nozzle rotation number membership function storage means 29 are used for cleaning the amount of dust, the rate of change of dust and the data of dust quality and cleaning. From the data of the suction force of the fan motor 7 and the rotation speed of the floor nozzle 8 to be set in consideration of the feeling, the steepest descent method (one of the learning rules used for the neural network, in which the error function is minimized) ) Is optimally set according to the learning rule. The control means 6 calculates and controls the phase control amounts of the fan motor 7 and the floor nozzle 8 based on the determined suction force and the rotation speed of the floor nozzle.

Next, the operation of the vacuum cleaner having the above configuration will be described. The light emitted from the light emitting unit 11 of the dust sensor 1 can be received by the light receiving unit 12 when there is no dust, but cannot be received by the light receiving unit 12 because it is blocked when the dust passes. Therefore, the presence or absence of dust can be determined from the output of the light receiving unit 12. The dust amount integrating means 2 integrates the dust detected by the dust sensor 1 for a certain period of time (for example, 0.1 second). By integrating, the amount of dust on the floor at that time can be determined. FIG. 3 shows the degree of change in the integrated value of the amount of dust when cleaning is continuously performed. After starting cleaning in the same figure
Until T1, garbage is reduced at once, indicating that garbage on the floor surface has been removed. In addition, after T1, as shown in the same figure, it is large depending on the type of garbage that is removed (A)
(B) and (C). In the case of (C), the integrated value of dust is almost 0, indicating that almost all of the dust has been removed before T1. This is the case where the floor to be cleaned is a wooden floor, a cushion floor, a tatami mat, or the like. In addition, when the floor surface is a carpet, dust is buried between the hairs, and the amount of dust is generally relatively small compared to wooden floors and tatami mats. That is, the characteristic shows that the integrated value of the amount of dust gradually decreases as shown in FIGS. When the change rate of the dust amount is calculated by the dust change rate calculating means 3 in this way, it is possible to estimate what the characteristics of the floor surface currently being cleaned are. A small change in the amount of garbage indicates that the garbage is difficult to collect, and a large change in the garbage amount indicates that the garbage is easy to collect. I have. 4th
The figure shows a pulse waveform of dust detected by the dust sensor 1. FIG. 7A shows a pulse waveform when cotton dust is sucked, and FIG. 7B shows a pulse waveform when sand dust is sucked. By detecting this pulse waveform, the dust quality detecting means 4 can determine the quality of the dust as to whether the sucked dust is large and light like cotton dust or small and heavy like sand dust. . The optimum suction force for cleaning is determined by the amount and quality of dust on the floor, the characteristics of the floor, and the like. The neuro-fuzzy inference unit 5 infers from the output value of the detection means 4.

Next, the process of inferring the suction force will be described. The inference rule of the fuzzy inference according to the present embodiment is that “a large amount of trash, heavy and small trash such as sand, and a floor surface where trash cannot be easily removed (the rate of change in the amount of trash is small). It is formed based on a general judgment such as "more." The qualitative concepts such as "large amount of garbage", "small change rate of garbage amount", "heavy garbage quality", and "very large" suction force are shown in Figs. 5 (A) and 5 (B). , (C) and FIG. 6 (A), (B). The garbage amount suitability calculating means 21 obtains the degree of conformity with respect to the input from the garbage amount integrating means 2 and the membership function relating to the garbage amount stored in the antecedent membership function storage means 20 by taking MAX of both. . The dust change rate matching degree calculating means 23 similarly obtains the matching degree with respect to the input from the dust change rate calculating means 3 and the membership function of the dust change rate stored in the antecedent membership function storage means 20. The garbage quality adaptation calculating means 21 similarly obtains the degree of conformity with respect to the input from the garbage quality detecting means 4 and the garbage quality membership function stored in the antecedent membership function storage means 20. Antecedent part minimum calculation means
In 24, the MIN of the three conformances is taken to be the conclusion of the antecedent part. In the consequent part minimum calculating means 25, the consequent part conclusion and the suction force membership of the consequent part stored in the suction force membership function storage means 26 are determined in accordance with the rules stored in the suction force inference rule storage means. Function
Take MIN as the conclusion of the rule.

After obtaining the respective conclusions for all the rules, the center-of-gravity calculating means 27 takes the MAX of all the conclusions and calculates the center of gravity to finally obtain the suction force.
The control means 6 calculates and controls the phase control amount of the fan motor 7 based on the determined suction force. Floor nozzle 8
In the same manner as in the process of determining the suction force, the conclusion of the antecedent is calculated, and the floor nozzle rotation number inference rule storage means 30 and the floor nozzle rotation number membership function storage means 29 are determined.
From this, the rotation speed of the floor nozzle 8 is determined. In this embodiment, the MAX-MIN synthesis method and the center of gravity method are used in the inference method. However, other methods are also possible, and the consequent suction force is expressed by a membership function. Needless to say, it can also be represented by a numerical value or a line format.

Effects of the Invention As described above, the vacuum cleaner of the present invention has the following effects. As the number of inputs and outputs in fuzzy inference increases, it becomes more difficult for humans to optimize the inference rules and their configurations between them. In the case of the present invention, the membership function and the inference rule are optimized using a learning rule such as the steepest descent method, so that even when a plurality of information is input to the fuzzy inference device, the optimal suction force and the floor nozzle are obtained. Can be determined. In addition, since the amount of dust, the dust quality, and the material of the floor surface are used to determine fine and optimal suction power and the number of rotations of the floor nozzle by fuzzy inference, dust can be efficiently removed regardless of the floor surface and dust quality to be cleaned. Moreover, it is possible to provide a vacuum cleaner having a very good operation feeling.

[Brief description of the drawings]

FIG. 1 is a block diagram showing one embodiment of the vacuum cleaner of the present invention,
FIG. 2 is a block diagram showing the configuration of a neuro-fuzzy inference device, FIG. 3 is a characteristic diagram showing a change in the amount of dust, FIG. 4 is a diagram showing a waveform of dust, FIG. 5 and FIG. It is a figure showing a ship function. 1 garbage sensor 2 garbage amount integrating means 3 garbage change rate calculating means 4 garbage quality detecting means 5 neuro fuzzy inference device

──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Haruo Terai 1006 Kadoma Kadoma, Osaka Prefecture Matsushita Electric Industrial Co., Ltd. (56) References JP-A-2-131374 (JP, A) JP-A-2-152427 (JP, A) (58) Fields investigated (Int. Cl. ⁶ , DB name) A47L 9/28

Claims (2)

(57) [Claims] 1. A fan motor for sucking dust, a floor nozzle for scraping dust on the floor, a dust sensor for detecting sucked dust on the floor, and a plurality of pieces of information obtained from an output of the dust sensor. Fuzzy reasoning, the number of revolutions of the fan motor and the floor nozzle were determined, and the antecedent and consequent part membership functions of the fuzzy inference and the inference rules were optimized by learning rules such as the steepest descent method. Vacuum cleaner with neuro-fuzzy inference device. 2. The method according to claim 1, wherein the neuro-fuzzy inference unit recognizes the amount of dust on the floor surface, the quality of the dust, and the material of the floor surface from the output of the dust sensor, and determines the rotation speed of the fan motor and the floor nozzle. Vacuum cleaner as described.