JP2001179122A - Waste pulverizing device with steam explosionproof specification - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable an operator to drive a waste pulverizing device of steam explosionproof specifications even when an installed ITV camera fogs and the stagnation state before roll feeder is invisible. SOLUTION: This waste pulverizing device has an ITV camera 34 which photographs the state before the roll feeder 13, an image processor which calculates the stagnation quantity before the roll feeder in accordance with the output signal therefrom and a loading material sensor 33 which measures the waste quantity on a supply conveyor 11 and operates automatically to pulverize the wastes in accordance with the state of the supply conveyor, the state of the roll feeder, the state of the pulverizing machine 16 and the stagnation quantity before the roll feeder. The device has a stagnation quantity estimating device 4, such as a neural network, for estimating the stagnation quantity before the roll feeder. The device is automatically operated by using the estimated value output of the stagnation quantity estimating device 4 when the acquisition of the reasonable stagnation quantity information from the ITV cameras is not possible on account of ejected steam.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a crusher having a function of preventing explosion due to steam spray, and more particularly to an automatic operation device of the crusher.

[0002]

2. Description of the Related Art In a facility for separating and recycling industrial waste and bulky waste from homes, a skilled operator is required to operate a device for crushing and processing waste, but the operation is automatically controlled. Technical developments to be performed by the apparatus have been performed, and as described in Japanese Patent Application Nos. 10-365881 and 11-189934, predetermined results are being obtained.

[0003] By the way, such a crushing apparatus includes:
May contain flammable materials and sometimes explosive waste, such as household gas cylinders and gas tanks. In order to prevent an explosion accident even if such combustible materials are mixed, steam may be blown into an inlet and a crushing chamber in front of a roll feeder to form an explosion-proof atmosphere, followed by crushing. With this device, the oxygen concentration in the atmosphere is reduced by the ejection of steam, so even when flammable materials are crushed, combustion and ignition can be prevented and explosion can be avoided,
When the atmosphere becomes hot and humid, and especially when the temperature is not high such as at the time of startup, steam becomes mist-like and visibility becomes poor, so that measurement is difficult.

In the automatic operation of the crusher, it is considered to be efficient to keep the amount of waste accumulated before the crusher inlet, that is, the roll feeder and the load of the crusher constant, and the roll feeder obtained from the image of the ITV camera is used. Previous waste retention information is required. However, the ITV camera cannot be used when performing steam ejection such as at the time of startup. For this reason, automatic operation was not possible, and the experienced operator had no choice but to operate manually and rely on intuition without image information.

[0005]

The problem to be solved by the present invention is to solve the problem in a steam explosion-proof waste crusher even if the installed ITV camera becomes cloudy in fog and the state in front of the roll feeder becomes invisible. An object of the present invention is to provide a waste crushing apparatus provided with a method for supplying information on the amount of waste accumulated before a roll feeder when an ITV camera cannot be used.

[0006]

Means for Solving the Problems To solve the above-mentioned problems, an ITV equipped with a supply conveyor, a roll feeder, and a crusher for supplying steam to at least a crushing chamber to prevent explosion, and for photographing a state before the roll feeder. Equipped with an image processing device that calculates the amount of stay before the roll feeder based on the camera and its output signal, and a load sensor that measures the amount of waste on the supply conveyor, the state of the supply conveyor, the state of the roll feeder, and the state of the crusher The steam explosion-proof specification waste crushing apparatus of the present invention, which automatically operates based on the accumulated amount before the roll feeder to crush the waste, includes a accumulated amount estimation device for estimating the accumulated amount before the roll feeder. Prepare, IT
When proper stay amount information cannot be obtained from the image signal of the V camera, automatic operation is performed using the estimated value output of the stay amount estimating device. The staying amount estimation device is a neural network that inputs the waste amount measured by the load sensor, the state of the supply conveyor, the state of the roll feeder, and the state of the crusher, and estimates and outputs the staying amount before the roll feeder. Is preferred.

[0007] The steam explosion-proof waste crushing apparatus of the present invention calculates the amount of waste accumulated before the roll feeder based on the image of the ITV camera when the ITV camera can be used normally, and performs automatic operation based on the calculated amount. However, if the fog rises in the steam atmosphere and the visibility is not good and the image of the ITV camera cannot be relied on, the automatic operation can be performed based on the staying amount estimated value estimated by a neural network or the like. Therefore, even when steam explosion protection is being performed, there is no need to perform manual operation by a veteran operator, and automatic operation can be continued.

[0008] The amount of stay before the roll feeder is not a variable that can be easily calculated based on a simple functional relationship using a single or a small number of observation values, but the relationship between the amount of supply to the inlet and the amount of crushing treatment. Cannot be estimated unless it is known over time. Since the amount of accumulation is the amount of accumulation due to the difference between the inlet flow rate and the outlet flow rate, it can be easily estimated if the flow rates at both the inlet and the outlet can be measured. However, it is almost impossible to measure the flow rate at the outlet of the roll feeder with the waste crushing apparatus. However, the outlet flow rate can be estimated with a certain degree of accuracy from the crusher current value corresponding to the crushing load, the crushing sound, and the like. However, it is very difficult to quantitatively link those measured values to the outlet flow rate. Therefore, if a neural network having a learning ability and a generalization ability is used, a complicated function relationship can be expressed as a relatively simple function to obtain a reasonable estimated value.

The neural network is preferably configured to learn based on the relationship between the amount of stay obtained by the image processing device and the values of various inputs when the image signal of the ITV camera can obtain appropriate stay amount information. . By learning by comparing the relationship between the estimated value of the staying amount and the actual staying amount during the period when correct information can be obtained, and determining a practically appropriate coefficient for the related variable, even if the actual staying amount can not be determined. An accurate retention amount can be estimated. Especially I
If the TV camera is trained while it is operating properly, it is possible to obtain an estimated value suitable for the latest situation whenever an estimation is required.

Further, the neutral network inputs the variation value of the estimated staying amount, sets the staying amount output of the image processing device as a bias, and sets the output of the image processing device as an initial value when switching to operation based on the estimated value. It is preferable to include an output integrator that outputs the estimated stay amount. By using such an integrator, the output of the integrator keeps track of the output value while the image processing device is operating properly, and when the camera cannot be used, the output can be switched to the dampress. it can.

Further, there is provided an arithmetic circuit for estimating the input amount from the supply conveyor by inputting the output of the load sensor, the speed signal of the supply conveyor, and the bulk density information of the object, and the input amount estimated by the arithmetic circuit. Information can be entered into a neural network. When such an arithmetic circuit is used, the number of input variables of the neural network is reduced, so that difficulty in learning is greatly reduced.

Further, the second steam explosion-proof waste crushing apparatus of the present invention is configured to input a waste amount measured by a load sensor, a state of a supply conveyor, a state of a roll feeder, and a state of a crusher to input a waste. An arithmetic unit for estimating the staying amount from the input amount of waste and the estimated value of the amount of waste discharged is provided. When it is not possible to obtain appropriate information of the staying amount from the image signal of the ITV camera, the estimated value output of the arithmetic unit is used. It is characterized by automatic driving.

According to a second aspect of the present invention, the amount of stagnation is estimated from the material balance of the waste in the processing apparatus instead of the neural network. The value obtained by integrating the difference between the amount charged into the inlet and the amount discharged from the crusher corresponds to the sum of the amount of stay before the roll feeder and the amount of stay of the crusher. The amount of stagnation during crushing in the crusher is determined almost in proportion to the load current of the crusher and the increment of the crushing sound. Therefore, the accumulated amount before the roll feeder can be estimated from this integrated value.

The arithmetic unit according to the second aspect of the present invention also receives a variation value of the estimated staying amount, sets the staying amount output of the image processing unit as a bias, and switches the output of the image processing unit to an initial value when switching to operation based on the estimated value. It is preferable to output the estimated staying amount. According to the second aspect of the present invention as well, the automatic operation can be performed based on the staying amount estimated value calculated by the arithmetic unit when the image of the ITV camera cannot be relied on due to fog or the like. Automatic operation can be continued.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A steam explosion-proof waste crushing apparatus according to the present invention will be described below in detail with reference to the drawings based on embodiments. FIG. 1 is a partially cutaway perspective view showing a steam explosion-proof waste crushing apparatus according to a first embodiment of the present invention, FIG. 2 is an input / output signal diagram of the crusher operating apparatus, and FIG. 3 is a first embodiment. FIG. 4 is a block diagram showing a state of the neural network of the first embodiment during learning, FIG. 5 is a block diagram showing a state of the neural network of the first embodiment during automatic operation, FIG. 6 is a block diagram showing a stagnant amount estimating operation unit according to the second embodiment.

[0016]

[Embodiment 1] A steam explosion-proof waste crushing apparatus of the present embodiment has a supply conveyor 11, a charging chute 12, and a roll feeder 1 as shown in a partially cutaway perspective view in FIG.
3. The waste crushing apparatus 1 including the crusher 16, in which combustible materials are blown by blowing steam from a steam nozzle 19 provided in a crushing chamber having the crusher 16 and a steam nozzle 20 provided in a portion of the roll feeder 13. This is a steam explosion-proof crushing treatment device that prevents ignition by frictional heat or ignition of sparks generated by crushing. This device has a central control panel 3
Thus, the supply of the wastes 21 and 22 to the crusher 16 can be controlled automatically or by remote control, and the optimum crushing treatment can be performed.

The supply conveyor 11 has a receiving hopper at the receiving end for receiving waste from a waste carrier or the like, and transports the received waste 21 and 22 to the entrance of the input chute 12. The charging chute 12 guides the wastes 21 and 22 dropped from the supply conveyor to the crusher 16 via the roll feeder 13. The roll feeder 13 includes a feed roller 14 and a biting roller 15, and adjusts the size of the wastes 21 and 22 in the input chute to adjust the feed speed. The feed roller 14 and the biting roller 15 swing in conjunction with each other, change the distance between the roller surface and the floor of the input chute 12 according to the size of the waste, adjust the opening of the roll feeder 13, and remove the waste. It bites well and sends it to the crusher 16.

The crusher 16 thins the waste and sends it out to the next sorting step by the action of a cutter bar 17 fixed to the main body and a hammer 18 embedded on the outer periphery of the rotating roller. The crushing machine 16 is relatively small, and small and medium-sized waste can be crushed as it is. However, large-sized waste cannot be crushed unless it is sent to the roll feeder 13 in an appropriate size and gradually fed. In order to prevent accidents and failures when the load is too large, a mechanism is provided that constantly measures the load current and trips and stops operation when the current exceeds a predetermined threshold.

The central control unit 3 switches the control method based on the result of measuring the size of the wastes 21 and 22 on the supply conveyor 11 and discriminating the large waste 21 and the small and medium waste 22 from each other. drive. The central control unit 3 has an I / O interface function for inputting various measured values obtained from sensors provided at various parts of the waste crushing and processing apparatus 1 and outputting an operation signal. It has a man-machine interface function that enables operations related to components. In addition, starting and stopping of equipment that does not require an advanced judgment function is sequence-controlled by a programmable controller (PLC).
In addition, the automatic supply of the waste to the crusher 16 is performed.

As shown in the schematic diagram of the control system for waste crushing treatment shown in FIG. 2, the central controller 3 has an ammeter for measuring the load current of the drive motor of the crusher 16 and an opening indicating the opening of the roll feeder 13. Total, supply conveyor 11 and roll feeder 1
In addition to the output of the speedometer 3, the ITV camera 34 for photographing the entrance of the roll feeder 13 from above the input chute 12, the size of the waste measured by the laser sensor 33, which is detected by the ultrasonic sensor 35 Waste height,
Information on the crushing sound caught by the microphone 36 is input.

The laser sensor 33 is provided above the end portion of the supply conveyor 11 and scans in a direction perpendicular to the conveyor surface to measure the cross section of the waste. Therefore, the laser sensor 33 moves downward by taking into account the moving speed of the conveyor. Passing waste 21
, Width, and height, that is, the size can be estimated. An appropriate number of ultrasonic sensors 35 are provided in front of the roll feeder 13 to detect waste having an appropriate height, and the large waste 21 detected by the laser sensor 33 is actually injected into the injection chute 12. Confirm that it was done.
The microphone 36 picks up the sound at the crushing position of the crusher 16 and compares it with a predetermined sound level to determine whether the waste has been crushed.

Further, the image signal acquired by the ITV camera 34 is used by an image processing device to detect the amount of waste accumulated before the roll feeder. Since the image of the portion in front of the roll feeder 13 acquired by the ITV camera 34 is binarized, a plurality of the binarized images are stored at predetermined time intervals, and a stagnant state can be known by taking a logical product of them. Is output as a measurement result of the staying amount. The central controller 3 performs a control logic operation based on these input signals and calculates a speed command signal of the supply conveyor 11, a roll feeder opening command signal, a roll feeder rotation speed command signal, and a roll feeder normal rotation / reverse rotation / stop command signal. It occurs and controls the supply to the crusher.

However, in order to prevent explosion due to mixing of combustibles in the waste, steam is blown out, especially after starting,
Since the surroundings do not become sufficiently hot for about 2 hours from the time, the vapor becomes mist-like and obstructs the field of view, so that it is impossible to measure the amount of stay correctly with an ITV camera. Therefore, in the present invention, the staying amount estimating device 4 is attached to the central control device 3 so that when the fog rises in the steam atmosphere and the visibility becomes ineffective and the image of the ITV camera cannot be used, the staying amount estimating device 4 is used. Automatic operation is performed based on the estimated residence amount estimation value.

In this embodiment, the amount of stay is estimated by a neural network. The neural network learns the relationship between the measured value of the stay amount output from the image processing device and the state of various input signals while the ITV camera can be used, and estimates the stay amount using the neural network that has an appropriate structure by learning. And use it for control. The neural network is a hierarchical neural network having an internal structure as shown in FIG. 3, which includes an input layer, one intermediate layer, and an output layer, and uses a back propagation method for learning.

The input unit of the input layer inputs the input amount from the conveyor, the roll feeder opening, the crusher load current, and the crushing sound, and also inputs the change in the roll feeder opening, the crusher load current, and the crushing sound. It is input, transformed into a coupling load based on a nonlinear function as an internal function, and distributed to units in the intermediate layer. The amount of change such as the roll feeder opening may be calculated based on the signal output of each instrument. The unit of the middle layer calculates the product of each grade, multiplies it by the coupling weight, normalizes it, and supplies it to the output layer. The output layer outputs the amount of change in the amount of stay as the sum of inputs. In the present embodiment, the structure is simplified by using the intermediate layer as one layer, but it goes without saying that more precise calculation is possible with a multilayer structure.

The neural network of the present embodiment has a sigmoid function within the function, to realize a non-linear function can be adjusted right and left asymmetrically C ^∞ class. The sigmoid function is f
Represented by (x) = 1 / (1 + exp (-x)), the output _{o j = 1 / (1 +} exp (-w g (x j + w c)))) become. Here, w _g is a center position of the sigmoid function, w _c is a gradient, and a desired nonlinear function can be realized by setting these to appropriate values. Note that, instead of the sigmoid function, an arc tangent function (tan
h) and the like may be used.

The input amount from the supply conveyor can be quantitatively estimated based on the accumulation amount on the supply conveyor measured by the laser sensor 33 and the speed of the supply conveyor in consideration of the bulk density of the object. , A value separately calculated by an arithmetic unit is used. The smaller the number of variables used in the neural network, the easier the function becomes.
It is preferable to directly calculate the variables whose relations are quantitatively grasped by the arithmetic unit because the number of variables handled by the neural network is reduced.

The roll feeder opening defines the amount of waste input to the crusher. A change in the injection state appears as a change in the opening, but the injection amount is more directly related to the opening itself. The crusher load current represents the load associated with the processing of waste and is a variable for estimating the charging state.
However, the change amount of the load current has a stronger relationship. The crushing sound is a sound that is generated when the waste is crushed, and is a basis for estimating the input state, and is considered to be more related to the crushing sound change amount.

These values are input and processed by a neural network to estimate the variation in the amount of stay. In a normal state, actual measurement by an ITV camera is performed.
When the V-camera cannot be used and the operation is switched to the operation based on the estimated value, the measured value immediately before can be used, and the inflow state and the outflow state of the waste can be easily estimated from the various input variables that can be obtained from the absolute value of the accumulated amount. Here, the variation of the staying amount is calculated. In addition, since it is variational, it does not need to be treated in time series, and the functional relationship is simplified.

Before using the neural network, learning is performed to determine an appropriate connection weight and the like. FIG. 4 is a block diagram showing a learning state of the neural network. The neural network 41 performs learning by the back propagation method. Learning is performed by inputting the accumulated amount of debris on the conveyor measured by the laser sensor, supply conveyor speed, roll feeder opening, crusher load current, crushing sound, etc. as input signals, and learning and inputting the change in the amount of accumulation before the roll feeder. Do as. The change amount of the staying amount before the roll feeder is given as a result of differentiating the staying amount obtained by performing image processing on the image acquired by the ITV camera by the image processing device 37 with the differentiator 38. In addition, since the supply amount from the supply conveyor is obtained from the supply conveyor speed, the accumulation amount on the conveyor, and the bulk density according to a predetermined relational expression, an input amount estimation arithmetic unit that inputs these variables and outputs an input amount estimation value. 42 may be provided so that the calculation result is taken into the neural network 41.

The learning is performed by comparing the staying amount estimation output when each input signal is given to the neural network with the learning input, and when the error is larger than a certain set value, adjusting the connection weight of the neural network. It is. If the maximum value and the average error of all the learning data become smaller than a certain appropriate value, the learning is terminated. If the end condition is not satisfied, a series of learning work is repeated until the standard is satisfied. It should be noted that the condition may not be satisfied even after a considerable number of repetitions, but it may be successful by changing the initial value of the weight of the neural network and re-executing. If that does not work, reconstruct the neural network structure, such as changing the number of hidden layers, and train again. The data used for learning was obtained by performing automatic driving when the ITV camera was usable. The learning can be either online learning performed at the same time as driving, or offline learning using stored data.

A steam explosion-proof waste crusher is operated using a learned neural network in which the connection weight and the like are fixed to appropriate values. FIG. 5 is a block diagram of the neural network during automatic driving. The neural network 41 constantly inputs and calculates an input signal, estimates a variation value of the staying amount, and outputs the estimated value. The output of the neural network 41 is integrated by an output integrator 43 provided externally, and supplied to the automatic control system in the central control unit 3 via the changeover switch 44 as an estimated value of the amount of stay.
The image processing device 37 is connected to the bias terminal of the output integrator 43.
Are connected via an on / off switch 45. Also, an image availability determination circuit 39 is provided, and the changeover switch 44 and the on / off switch 45 are operated based on the result of determining whether or not an image is available from the output of the ITV camera. The image processing device 37 processes the output signal of the ITV camera 34 and outputs the amount of stay before the roll feeder as a measured value.

While the ITV camera operates normally and the amount of stay before the roll feeder can be accurately measured from the image acquired, the changeover switch 44 is turned on to the measurement side and the measured value given by the image processing device 37 is automatically controlled by the automatic control system. Send to Further, the on / off switch 45 is turned on, and supplies the measured value to the offset terminal of the output integrator 43 sequentially.

When the steam is blown out, fog is generated in the roll feeder portion and visibility becomes poor, and the camera lens becomes cloudy, so that the staying state before the roll feeder cannot be seen from the image of the ITV camera. Then, the image availability determination circuit 39 is activated and the on / off switch 45 is turned on.
Is turned off, and the changeover switch 44 is switched to the estimation side.
The output integrator 43 integrates and outputs the variation value of the estimated stay amount of the neural network 41 using the offset value at this time as an initial value. The output of the output integrator 43 is sent to the automatic control system by the changeover switch 44 as an estimated value instead of the staying amount measured value based on the output of the ITV camera. The accumulated amount is a ratio occupied by sediments in the space before the roll feeder, and a resolution of a value between 0 and 10 satisfies the operational necessity.

The offset amount of the output integrator 43 always tracks the staying measurement value. When switching to the estimated value, the integration is started based on the staying value measurement value at that time. The input of the estimated value is switched by a bumpless press. In the steam explosion-proof specification waste crusher of this embodiment, even if the image of the ITV camera cannot be seen, the operation does not depend on the intuition of the operator,
Automatic operation can be continued by accurately estimating the amount of residence.

In the automatic operation of the waste crusher, it was assumed that images from the ITV camera were used.
In the technology used in the waste crusher of this embodiment, the IT
Since the automatic operation can be performed with appropriate accuracy without always using the V camera, it is possible to always perform an operation that does not rely on the camera image as required, not only when the steam is blown out. Needless to say, the apparatus of this embodiment can be effectively used even when the apparatus is not of the steam explosion-proof type. In the present embodiment, a method using a hierarchical neural network is employed. Alternatively, a recurrent neural network capable of expressing dynamics or a fuzzy neural network that determines and uses an internal structure may be used. it can.

[0037]

Embodiment 2 FIG. 6 is a block diagram showing a second embodiment of the present invention. In this embodiment, a neural network is used in the first embodiment, and a device for calculating mass balance is provided. The difference between the amount of waste on the supply conveyor and the amount of crushed and discharged is integrated. To estimate the amount of residence.

A feeding amount estimating device 46 for inputting a feeding conveyor speed, a result of measuring a deposition amount using a laser sensor provided on the feeding conveyor, and bulk density information of an object, and estimating a feeding amount from the feeding conveyor. And the discharge conveyor speed,
It is provided with a discharge amount estimating device 47 that inputs the measurement result of the deposition amount by the laser sensor on the discharge conveyor and the bulk density information of the object, and estimates the discharge amount from the discharge conveyor, and outputs the difference between the two. It is input to the integrator 43. The output integrator 43 has the same configuration as the first embodiment and performs the same operation. Further, a crusher staying amount estimating device 48 for inputting the crusher load current and the crushing sound and estimating the staying amount of the crushing chamber is provided,
The output is subtracted from the estimated output of the crusher staying amount of the output integrator 43, and then supplied to the automatic control system via the changeover switch 44. The changeover switch 44 has the same configuration as in the first embodiment and performs the same operation.

The accumulated amount in the apparatus can be obtained by integrating the difference between the amount of waste input from the supply conveyor and the amount of waste discharged after processing from the discharge conveyor. The amount of residence in the device is
This is the sum of the accumulated amount before the roll feeder and the accumulated amount during crushing by the crusher. On the other hand, the amount of residence in the crusher can be estimated because it is substantially proportional to the load current of the crusher and the increment of the crushing noise with no load. Therefore, the remaining amount before the roll feeder can be calculated by subtracting the estimated value in the crusher from the estimated value in the device. In addition, the accumulation amount on the conveyor may be measured using a weighing scale. When measuring with a weighing scale, there is no need to use information on the bulk density.

The arithmetic unit in this embodiment also inputs the variation value of the estimated staying amount, sets the staying amount output of the image processing device as a bias, and switches the output of the image processing device to the initial value when switching to operation based on the estimated value. It is preferable to output the estimated staying amount to the bumpless. Also in the second embodiment, when explosion-proofing is possible, automatic operation can be performed based on the estimated amount of stay calculated by the arithmetic unit when the image of the ITV camera cannot be relied on due to fog or the like. Even in such a case, the automatic operation can be continued. In addition,
As in the first embodiment, it is possible to perform an operation that does not rely on an image from an ITV camera at normal times, and the same technical idea can be applied to a crushing apparatus that is not a steam explosion-proof type.

[0041]

The steam explosion-proof waste crushing apparatus of the present invention can automatically perform the crushing work in the early stage of steam ejection, which was conventionally performed by a skilled operator, thereby reducing operating costs. Was completed. In addition, since appropriate operation can always be performed, it is not necessary to expect a high safety factor, and it is possible to avoid an increase in the size of the equipment. Furthermore, the operation can be performed under the same conditions even during the steam ejection, and the crushing efficiency can be maintained. In addition, since control can be performed only by switching the input signal, there is no need to replace a control device used for automatic operation with a device used for a waste crusher to which steam explosion protection is not applied.

[Brief description of the drawings]

FIG. 1 is a partially cutaway perspective view showing a steam explosion-proof waste crushing apparatus of the present invention.

FIG. 2 is a diagram showing input / output signals of a crusher operating device according to the present invention.

FIG. 3 is a configuration diagram of a neural network used in the first embodiment of the present invention.

FIG. 4 is a block diagram showing a state at the time of learning of the neural network of the first embodiment.

FIG. 5 is a block diagram showing a state of the neural network according to the first embodiment during automatic operation.

FIG. 6 is a block diagram showing a staying amount estimating operation unit according to a second embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Waste crushing processing apparatus 11 Supply conveyor 12 Injection chute 13 Roll feeder 14 Feeding roller 15 Biting roller 16 Crusher 17 Cutter bar 18 Hammer 19,20 Steam nozzle 21,22 Waste 3 Central control panel 33 Laser sensor 34 ITV camera Reference Signs List 35 ultrasonic sensor 36 microphone 37 image processing device 38 differentiator 39 image availability determining circuit 4 staying amount estimating device 41 neural network 42 input amount estimating operation device 43 output integrator 44 changeover switch 45 on / off switch 46 input amount estimating device 47 discharge Quantity estimation device 48 Crusher retention amount estimation device

 ──────────────────────────────────────────────────の Continuing from the front page (72) Inventor Takao Kanamaru 118 Notsuka, Noda-shi, Chiba Kawasaki Heavy Industries Noda Plant (72) Inventor Junichi Okumura 1780 Kamikono, Yachiyo-shi, Chiba Kawasaki Heavy Industries Yachiyo Plant F term (reference) 4D004 AA46 CA04 CA50 CB03 CB13 CB43 CB46 DA01 DA02 DA16 DA20 4D067 EE01 EE39 EE45 FF01 FF11 GA18 GA20 GB03

Claims (7)

[Claims] 1. A waste crushing treatment apparatus comprising a supply conveyor, a roll feeder, and a crusher, which supplies steam to a crushing chamber to prevent explosion, and captures an image of a state before the roll feeder.
Equipped with an image processing device that calculates the amount of stay before the roll feeder based on the TV camera and its output signal, and a load sensor that measures the amount of waste on the supply conveyor, the state of the supply conveyor, the state of the roll feeder, and the crusher In a waste crushing device for automatically crushing and processing waste based on a state and a stay amount before a roll feeder, a waste amount estimating device for estimating a stay amount before the roll feeder is provided, and an image of the ITV camera is provided. A steam explosion-proof waste crushing apparatus, wherein automatic operation is performed by using an estimated value output of the staying amount estimating device when information on a proper staying amount cannot be obtained from the signal. 2. The apparatus according to claim 1, wherein the accumulated amount estimating device is a neural network, and the neural network inputs a waste amount measured by the load sensor, a state of a supply conveyor, a state of a roll feeder, and a state of a crusher to form a roll. The steam explosion-proof specification waste crushing apparatus according to claim 1, wherein the amount of stay before the feeder is estimated and output. 3. The neural network includes an ITV
The steam explosion-proof specification according to claim 2, wherein learning is performed based on a relationship between the amount of stay obtained by the image processing apparatus and the value of each of the inputs when the image signal of the camera can acquire appropriate stay amount information. Waste crushing equipment. 4. The operation based on the estimated value, wherein the neutral network includes an output integrator, and the output integrator inputs a variation value of the estimated staying amount and biases the staying amount output of the image processing apparatus. 4. The steam explosion-proof waste crushing apparatus according to claim 2, wherein when switching, the estimated staying amount is output with the output of the image processing apparatus as an initial value. 5. An operation circuit for inputting an output of the load sensor, a speed signal of a supply conveyor and bulk density information of an object to estimate an amount of supply from the supply conveyor, and the amount of input estimated by the operation circuit. The steam explosion-proof waste crushing apparatus according to any one of claims 2 to 4, wherein information is input to the neural network. 6. The input amount of waste and the discharge amount of waste by inputting the amount of waste, the state of a supply conveyor, the state of a roll feeder, and the state of a crusher measured by the accumulation amount estimation device by the load sensor. 2. A steam explosion-proof waste crushing device according to claim 1, wherein the calculating device estimates the amount of residence from the estimated value of the above. 7. The operation device includes an output integrator, the output integrator inputs a variation value of the estimated staying amount, and biases the staying amount output of the image processing apparatus. 7. The steam explosion-proof waste crushing apparatus according to claim 6, wherein when switching, the estimated retention amount is output with the output of the image processing apparatus as an initial value.