JP3390783B2 - Crushing / crushing processing system - Google Patents

Description

Detailed Description of the Invention

[0001]

TECHNICAL FIELD The present invention relates to a crushing / crushing processing system. More specifically, the present invention relates to a crushing / crushing processing system in which a control device has a control means having a learning function.

[0002]

2. Description of the Related Art Conventionally, a crusher or a crusher (hereinafter referred to as a crusher) has been used for treating various kinds of miscellaneous waste such as used refrigerators and chests, so-called oversized waste, or scrap disposal of scrapped automobiles. Crushed and crushed by using a machine).

In the crushing / crushing process, however, various varieties of properties need to be processed as described above, and therefore the crusher is operated manually by a skilled operator. That is, the operator monitors the monitor TV that is imaging the inlet of the crusher, and at the same time, pays attention to the instrument that indicates the current value of the motor driving the crusher and checks the input amount of oversized waste. Making adjustments. For this reason, the operation of this crusher becomes a considerable amount of hard work for the operator, and fatigue accumulates with the lapse of operating time. Therefore, if one operator is allowed to operate for a long time, the work accuracy may decrease. In order to avoid such a situation, at present, a method of performing replacement work by a plurality of operators is adopted.

However, due to the decrease in the number of young workers in recent years, it has become difficult to secure skilled operators. For this reason, the advent of an oversized waste automatic processing system that can be automatically operated has been eagerly desired by the parties concerned.

In order to improve such a situation, Japanese Unexamined Patent Application Publication No. Hei 8
No. 276138, when feeding a crushed object to a crushing / crushing machine by a supply device for crushing / crushing, not only detection of load current of the crushing / crusher but also fluctuation amount of the load current The layer thickness of the object to be crushed in the device and the crushing sound level of the crushing / crushing machine are respectively detected, and based on the detected values, by fuzzy reasoning, the load of the crushing / crushing machine becomes constant, A method for suppressing fluctuation in load of the crusher / crusher has been proposed, which is characterized by controlling the feeding speed of the crushed object from the supply device to the crusher / crusher (see FIG. 7).

However, according to the fluctuation suppressing method proposed above, although the operator's load is expected to be somewhat reduced, the operation of the bulky refuse processing system cannot be automated.

[0007]

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems of the prior art, and is an automatic crushing / crushing processing system for processing objects such as bulky refuse having various miscellaneous properties. It is an object of the present invention to provide a crushing and crushing treatment system which can be operated in the same degree as the manual operation of a skilled operator.

[0008]

[0009]

A first embodiment of the crushing / crushing processing system of the present invention is a crushing / crushing processing system having a crushing function or a crushing function, wherein the crushing / crushing processing system is an object to be processed. To the processing system, an input port into which the object to be processed conveyed by the transfer means is input, and a crushing unit or a crushing unit to crush or crush the object to be processed input into the input port. Control means having a learning function in the control device of the crushing / crushing processing system, comprising: a supply means capable of changing the opening degree to be supplied to the crushing means and a driving means for driving the crushing means or the crushing means ; and a correction means for correcting, learning of the control means, at least the crushing sound generated from the pulverizing means or grinding means during crushing or grinding or milling sound as an input, the A supply unit that uses the conveyance speed of the feeding unit, a captured image of the input port, the opening of the supply unit whose opening can be changed, and the load of the drive unit, and the output of which the opening can be changed Of the opening degree command value and the transfer speed command value of the transfer means,
Accumulates database and know-how of skilled operators
The learning content is corrected by the correction means and supplied.
To deal with seasonal changes in processed materials
And wherein the Rukoto such is.

Second of the crushing / crushing processing system of the present invention
The form is a crushing / crushing system having a crushing function or a crushing function, wherein the crushing / crushing system is
Conveying means for conveying the object to be treated to the same processing system, an input port for introducing the object to be treated conveyed by the conveying means, and crushing means for crushing or crushing the object to be treated introduced into the inlet. Alternatively, a control means having a supply means capable of changing the opening degree to be supplied to the crushing means and a drive means for driving the crushing means or the crushing means, and having a learning function in the control device of the crushing / crushing processing system , content of study
And a correction means for correcting, the learning of the control means,
A crushing sound or a crushing sound generated during crushing or crushing from the crushing means or the crushing means, a conveying speed of the conveying means,
A combination of at least two kinds of the captured image of the charging port, the opening of the supply unit whose opening can be changed, and the load of the driving unit is used as an input, and the opening can be changed as an output. and opening command value of the supply means, made by using the transport speed command value of said conveying means, said modified hand
The database and know-how of skilled operators
It is accumulated and the learning contents are corrected by the correction means.
You can deal with seasonal changes in the supplied materials.
Sea urchin is characterized by Rukoto such by.

In the crushing / crushing processing system of the present invention, it is preferable that the crushing / crushing processing system comprises a correction means for correcting the learning content of the control means having the learning function. Further, in the crushing / crushing processing system of the present invention, the learning may be performed off-line.

Here, the carrying means is, for example, a supply conveyor, and the carrying speed is the number of rotations of the supply conveyor. The control means having the learning function is composed of, for example, a neural network or a fuzzy neural network.

[0013]

Since the crushing / crushing processing system of the present invention is configured as described above, the automatic operation of the crushing / crushing processing system can be performed by the skilled operator by making the control means of the control device learn the operating procedure of the skilled operator. To the same extent as

According to a preferred embodiment of the present invention, which comprises a correction means for correcting the content of learning, the control means performs learning corresponding to the change in the property of the object being processed even after the crushing / crushing processing system is installed. Since it can be easily done, it is easy to deal with changes in the properties of the object to be treated.

[0015]

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, the present invention will be described based on the embodiments with reference to the accompanying drawings, but the present invention is not limited to the embodiments.

FIG. 1 is a schematic view showing a conventional oversized waste processing system S'to which the present invention is applied. In this oversized waste processing system S ′, as shown in FIG.
The bulky refuse input port 1a of the crusher 1 is imaged by the video camera 10 and projected on the monitor TV (not clearly shown) of the control room R, and the crushing sound generated from the crusher 1 is generated by the crusher 1. The sound is collected by the sound collecting microphone 12 provided on the side and flowed from the speaker (not clearly shown) in the control room R, and the transfer speed of the transfer means, for example, the rotation speed and the opening of the supply conveyor 2 are changed. The possible supply means, for example, the opening of the feeder 3, the rotation speed of the crushing means, for example, the rotation speed of the crusher, the load of the driving means of the crushing means, for example, the motor load, etc. are provided in the control room R (clearly shown. It has not been). Then, the operator comprehensively judges the monitor image, the displayed opening of the feeder 3, the rotation speed, the load of the motor, etc.
The opening degree and the number of revolutions of the supply conveyor 2 are adjusted so that the crusher 1 is operated in an optimum state.

For example, as for the treatment of large-sized waste from households, the processing time in the crusher 1 is different between the refrigerator and the wooden chest. That is, when the refrigerator is put in, the processing takes a long time, whereas when the wooden chest is put in, the processing does not take much time. Therefore, when the refrigerator is thrown in, the transport speed of the supply conveyor 2 is slowed down, that is, the number of rotations of the supply conveyor 2 is reduced, while when the wooden chest is thrown in, the transport speed of the supply conveyor 2 is increased, that is, Increase the number of rotations of the supply conveyor 2. Further, even in the same refrigerator, if the size is different, the processing time is different and the opening degree of the feeder 3 needs to be changed. That is, when a large refrigerator is loaded, the opening of the feeder 3 is increased and the conveying speed of the supply conveyor 2 is slowed, while when a small refrigerator is loaded, the opening of the feeder 3 is decreased. At the same time, the feeding speed of the supply conveyor 2 is increased. In addition, since the processing speed is different between the metal product and the plastic product, it takes a long time to process the metal product when a metal sound is heard from the speaker, so the conveying speed of the supply conveyor 2 is slowed. . Furthermore, when it is confirmed that the amount of staying in the input port 1a is larger than the image on the monitor TV, the conveying speed of the supply conveyor 2 is slowed down or the feeder is fed in order to prevent oversized dust from overflowing from the input port 1a. Increase the opening of 3.

On the other hand, in the scrap processing including the pre-crushing processing, the conveying speed of the supply conveyor 2 and the feeder 3 depend on the amount of staying in the charging port 1a and the material to be scraped.
The opening degree of is adjusted. For example, when the amount of staying in the input port 1a is increased, the conveyance speed of the supply conveyor 2 is slowed or the opening degree of the feeder 3 is increased. Further, when a soft material is supplied, the conveying speed of the supply conveyor 2 is increased or the opening degree of the feeder 3 is increased.

FIG. 2 is a bulky refuse processing system S shown in FIG.
It shows the state of the preparatory stage for automating '. That is, a neural network (hereinafter abbreviated as NN) or a fuzzy neural network (hereinafter FN)
The control device 20 ′ is provided with the control means 22 having a learning function, which is configured by N.
The measured value input to the instrument provided in the control room R is input to 0'in parallel. In this case, the image from the video camera 10 is input to the control device 20 ′ via the image processing device 24, and the crushing sound from the sound collecting microphone 12 is input to the control device 20 ′ via the audio processing device 26. .

Further, the feeder 3 by this control means 22
The opening command value and the rotation speed command value of the supply conveyor 2 are
A comparator 28 provided in the controller 20 'compares the command values by the operator, and the deviation thereof is fed back to the NN or FNN. Then, the control means 22 sets NN or F so that this deviation becomes zero.
Modify each coefficient of the NN network. This correction work is repeated, for example, about 5000 times. That is,
The NN or FNN constituting the control means 22 is repeatedly learned. Further, as inputs during this learning, for example, a monitor image, crushing sound, feeder opening, crusher rotation speed,
The motor load and supply conveyor speed are used. In this case, learning may be performed by using a combination of two or more of the inputs as inputs. For example, a combination of monitor image and crushing sound, a combination of feeder opening and crusher rotation speed, a combination of monitor image and crushing sound and feeder opening, a monitor image and crushing sound, feeder opening and motor load. Combination of monitor image, crushing sound, feeder opening, motor load and supply conveyor rotation speed, monitor image, crushing sound, feeder opening, crusher rotation speed, motor load and supply conveyor rotation speed, etc. You may use.

The learning may be performed online or offline. Here, online learning means learning NN or FNN while driving the crusher 1, and offline learning means learning NN or FNN separately from the operation of the crusher 1. Say.

Next, an example of each of the NN and FNN will be described.

FIG. 3 shows a schematic diagram of the NN. This NN is a hierarchical network, and a monitor image is input from the input side.
The crushing sound, the feeder opening, the crusher rotation speed, the motor load, and the supply conveyor rotation speed are input, and the feeder opening command value and the supply conveyor rotation speed command value are output from the output side.

FIG. 4 shows a schematic diagram of the FNN. This network has a fuzzy rule-like structure, and when a certain input pattern is given, an output pattern corresponding to it is obtained through a structure called a fuzzy rule. is there. For example, as shown in FIG. 4, from the input side, a monitor image, crushing sound, feeder opening, crusher rotation speed,
The motor load and the supply conveyor rotation speed are input, and the feeder opening command value and the supply conveyor rotation speed command value are output from the output side.

As shown in FIG. 5, membership functions in this FNN are triangular (see (a)), differentiable function types (see (b)), and Gaussian basis types. (See the same (c)), a sigmoid function (see the same (d)), etc. are used. Then, the membership function is modified by iterative learning by the backpropagation method or the like as in the NN.

FIG. 6 is a bulky refuse automatic processing system S including the NN or FNN for which learning has been completed by the above procedure.
FIG. This oversized garbage automatic processing system S
As is apparent from FIG. 6, basically, the part related to the manual operation by the operator in FIG. 2 is excluded. Further, the control device 20 is provided with a correction means 30 in which a database and know-how of a skilled operator are accumulated in order to cope with seasonal changes in the supplied bulky refuse. The NN and FNN can be appropriately corrected by the data supplied from the correction means 30 or the know-how of a skilled operator.

As described above, according to this embodiment, the control device 20 is provided with the control means 22 composed of the NN or FNN having the learning function, and the NN or FNN is operated by a skilled operator to operate the oversized waste processing system S. Since the automatic operation of the bulky waste processing system S is performed after learning the point, the automatic operation of the bulky waste processing system S can be performed to the same extent as the manual operation by a skilled operator.
That is, the bulky waste processing system S can be automated. Further, since the learning of the NN or FNN is automatically performed, troublesome parameter adjustment becomes unnecessary. Furthermore, since the data is supplied from the database of the correction means 30 to the control means 22, it is possible to cope with the variation in the type of the bulky refuse simply by changing the supplied data. In addition, even for a bulky waste processing system S in a business establishment where no skilled operator is present, automatic operation is possible only by supplying data from a database from a similar bulky waste processing system S.

Although the present invention has been described above based on the embodiments, the present invention is not limited to the embodiments and various modifications can be made. For example, the present invention can be applied not only to a bulky waste treatment system but also to various systems using a crusher or a crusher.

[0029]

As described in detail above, according to the present invention,
The control device is provided with a control means having a learning function composed of NN or FNN, and a trained operator learns the operating procedure of the crushing / crushing processing system before the crushing / crushing processing system is automatically operated. Therefore, it is possible to obtain an excellent effect that the crushing / crushing processing system can be automatically operated while ensuring the same operation accuracy as a manual operation by a skilled operator. That is, the excellent effect that the crushing / crushing system can be automated can be obtained. Moreover, since the control means composed of this NN or FNN is automatically learned,
The effect that no troublesome parameter adjustment is required can be obtained.

In a preferred embodiment of the present invention,
Since the NN or the FNN is corrected by supplying the data from the database of the correction means to the control means, it is possible to cope with the fluctuation of the type of dust only by changing the supplied data. Further, according to this preferable mode, even for a crushing / crushing treatment system in a business establishment without a skilled operator, a similar crushing / crushing system can be used.
It is also possible to obtain an excellent effect that the automatic operation becomes possible only by supplying the data from the database from the grinding processing system.

[Brief description of drawings]

FIG. 1 is a block diagram of a conventional configuration of a bulky refuse processing system to which the present invention is applied.

FIG. 2 is a block diagram showing a state in which the control means is learning the operating procedure of a skilled operator in the bulky waste processing system of the conventional configuration.

FIG. 3 is a schematic diagram of an example of a neural network that constitutes the control means of the present invention.

FIG. 4 is a schematic diagram of an example of a fuzzy neural network that constitutes the control means of the present invention.

FIG. 5 is a graph of a membership function used in the same fuzzy neural network, where (a) shows a triangular function, (b) shows a differentiable function type, and (c) the same. Indicates a Gaussian basis type, and (d) indicates a sigmoid function.

FIG. 6 is a block diagram showing a state in which the present invention is applied to the oversized waste processing system having the conventional configuration shown in FIG.

FIG. 7 is a conceptual diagram related to a control method disclosed in Japanese Patent Laid-Open No. 8-276138.

[Explanation of symbols]

1 crusher 1a Input port 2 supply conveyor 3 feeders 10 video cameras 12 collecting microphone 20 Control device 22 Control means having learning function 24 Image processing device 26 Speech processing device 28 Comparator 30 Corrective measures S crushing / crushing processing system

─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Shiro Nakabayashi 2-4-1 Hamamatsucho, Minato-ku, Tokyo World Trade Center Building Kawasaki Heavy Industries, Ltd. Tokyo Head Office (72) Inventor Takao Kanemaru 118 Futatsuka, Noda, Chiba Prefecture Address Kawasaki Heavy Industries Ltd. in Noda Factory (72) Inventor Yusho Kato 1780, Yachiyo Uetakano, Chiba Prefecture Kawasaki Heavy Industries Ltd. in Yachiyo Factory (56) Reference JP 5-192604 (JP, A) JP Flat 8-276138 (JP, A) JP-A-3-12244 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) B02C 1/00-25/00 G05B 13/02 G06F 15 / 18 540 G06F 15/18 550

Claims (5)

(57) [Claims] 1. A crusher having a crushing function or a crushing function.
A crushing system, The crushing and crushing processing system uses
Transporting means for transporting the
The input port where the object to be processed is input and the input port
Crushing means for crushing or crushing the input object
Is a supply means that can change the opening to be supplied to the crushing means,
Drive means for driving the crushing means or the crushing means.
e, The crushing / crushing system controller has a learning function.
Control means to, Correction means for correcting the learning contentTo
Prepare, Learning of the control means includes at least the crushing as input
Crushing or crushing from means or crushing means
Crushing sound or crushing sound, the conveying speed of the conveying means,
A captured image of the inlet and a supply means whose opening can be changed.
The degree of opening and the load of the driving means are used, and as an output
The opening command value of the supply means whose opening can be changed,
It is made by using the conveying speed command value of the feeding means., The correction means includes a database and a skilled operator.
Know-how is accumulated, and the learning content is
For seasonal changes in the processed materials supplied after modification,
Being able to do it Crushing and crushing characterized by
Processing system. 2. A crusher having a crushing function or a crushing function.
A crushing system, The crushing and crushing processing system uses
Transporting means for transporting the
The input port where the object to be processed is input and the input port
Crushing means for crushing or crushing the input object
Is a supply means that can change the opening to be supplied to the crushing means,
Drive means for driving the crushing means or the crushing means.
e, The crushing / crushing system controller has a learning function.
Control means to, Correction means for correcting the learning contentTo
Prepare, Whether the learning of the control means is the crushing means or the crushing means
Sound or crushing sound generated during crushing or crushing
The transport speed of the transport means, the captured image of the inlet, the opening
The opening of the supply means whose degree can be changed,And of the drive means
Input at least two or more combinations from the load
Used as a power supply and the output of which can be changed by the supplier
The step opening command value and the transfer speed command value of the transfer means
Made with, The correction means includes a database and a skilled operator.
Know-how is accumulated, and the learning content is
For seasonal changes in the processed materials supplied after modification,
Being able to do it Crushing and crushing characterized by
Processing system. 3. The crushing / grinding processing system according to claim 1, wherein the conveying means is a supply conveyor, and the conveying speed is a supply conveyor rotation speed. 4. The crushing-grinding processing system according to claim 1 or 2, wherein said learning is performed by off-line. 5. The crushing / crushing processing system according to claim 1, wherein the control means having the learning function comprises a neural network or a fuzzy neural network.