JPH06307185A - Directional control device of shield excavator - Google Patents

Abstract

PURPOSE:To enable the control of the direction of a shield excavator by combining functions of a position detecting section/target course operating section/ operation point reasoning section/operation point correcting section, and setting a shield jack operation point in consideration of working environmental items. CONSTITUTION:Horizontal and vertical slippage between driving directional data inputted from a position detecting section 40 and a planned line recorded in a memory 42 is calculated by a target course operating section 44, and a target course of a shield excavator 10 is set. After that, a jack operation point for leading the shield excavator 10 to the target course is made to fuzzily reason by an operation point reasoning section 46, and an operation point is corrected on the basis of driving environmental data by an operation point correction section 48. A jack pattern is set in the operation point corrected by a jack pattern setting section 50, and a jack control driving section is driven to make directional control of the shield excavator 10 along the planned line.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a direction control device for a shield machine, and more particularly to an improvement of a direction control device for setting a jack pattern based on a deviation amount of a shield machine from a planned line.

[0002]

2. Description of the Related Art The shield work is intended for the ground whose behavior is not easily grasped, the ground is often changed, and the same construction condition is rarely reproduced. For this reason, the automation of construction has not progressed, and in many cases, the directional control of the shield machine has often relied on manual operation by a skilled operator.

In order to solve this problem, shield excavation using fuzzy control or the like has recently been performed. By using such fuzzy control, it is possible to incorporate the experience and know-how of a skilled operator and automatically cause the shield machine to follow the set horizontal and vertical excavation planned lines of the tunnel.

In the conventional direction control using fuzzy reasoning, the current position and propulsion direction of the shield machine are compared with the planned excavation line to obtain the positional deviation amount and the angular deviation amount.
The point of action of the shield jack is fuzzy inferred from these deviation amounts. Then, a jack pattern corresponding to the inferred point of action is set to control the direction of the shield machine.

[0005]

However, the point of action of the shield jack for guiding the shield machine to the target course is as follows.
Not only the shift amount of the shield machine, but also the jack speed, the middle bending angle of the shield machine, the set value of the copy cutter stroke for the horizontal excavation, the tail clearance of the shield machine, and other excavation environments.

With respect to the excavation environment items such as the copy cutter stroke and the center bending angle, when these values become large, the shield excavator itself tends to bend. For example, when the center bending angle becomes large, the shield machine becomes easy to bend in the center bending direction. Also, as the jacking speed increases, the shield machine becomes difficult to bend.

However, in the conventional direction control device using fuzzy inference, no consideration has been given to such excavation environment items. Therefore, there is a problem in that the inferred point of action of the shield jack is not always the optimum point of action.

It is also possible to carry out fuzzy inference that incorporates the above-mentioned excavation environment item, but this has a problem that the fuzzy inference itself becomes extremely complicated.

The present invention has been made in view of such conventional problems, and an object thereof is to obtain a shield jack action point in consideration of the excavation environment item of the shield machine and to control the direction of the shield machine. It is to provide a direction control device for a shield machine that can perform.

[0010]

In order to achieve the above object, the present invention provides a target course calculating means for calculating a target course of a shield machine and a shield machine based on a deviation amount of the shield machine from a planned line. Based on the data of the excavation environment item of the shield machine, and the action point inference means for inferring the shield jack action point for guiding the
It is characterized by including an action point correction means for correcting the inferred shield jack action point and a jack pattern setting means for setting a jack pattern according to the corrected shield jack action point.

Here, the target course calculation means is
Based on the amount of horizontal and vertical deviation with respect to the planned line of the shield machine, calculates the horizontal and vertical target path of the shield machine, the action point inference means, for guiding the shield machine to the target path The shield jack points of action can be configured to infer as horizontal and vertical moment positions.

Further, the action point correcting means, as the excavation environment item, jack speed information of the shield excavator, center bending angle information, copy cutter stroke information,
At least one of the tail clearance information and the soil information can be used to correct the shield jack action point.

Further, the jack pattern setting means is
The jack pattern of the shield jack that contributes to propulsion is set according to the corrected point of action of the shield jack, and the extension of shield jacks other than the shield jack that contributes to propulsion is synchronized with and follows the shield jack that contributes to propulsion. In addition, when there is a displacement between the point of action of the shield jack that contributes to the propulsion and the corrected point of action, increase the hydraulic pressure of the shield jack that does not contribute to the propulsion so that they match.
It is desirable to form it so as to generate an opposite moment to the moment due to the shield jack that contributes to propulsion.

Further, the action point inference means can be formed so as to obtain the shield jack action point by fuzzy inference based on the shift amount.

Further, the action point inference means has an input layer, an intermediate layer, and an output layer, and is formed as a neural network circuit for inferring the shield jack action point, and the neural network circuit has its membership. The learning operation may be configured to modify a function based on a backpropagation algorithm, and the learning operation may be controlled to stop during a correction calculation based on the excavation environment item.

[0016]

In the directional control device of the present invention, the target course calculating means calculates and sets the target course of the shield machine based on the deviation amount of the shield machine from the planned line. Then, the action point inference means infers the shield jack action point for guiding the shield machine to the target course. For this reasoning, for example, fuzzy reasoning can be used if necessary.

Then, the action point correcting means corrects the inferred shield jack action point based on the data of the excavation environment item of the shield machine. Such excavation environment items include jack speed information, center bending angle information, copy cutter stroke information, tail clearance information, soil information, and other information. Using at least one of these environment items Correct the point of action.

For example, when the copy cutter is set, or when the shield machine is set in the center-folded state, the shield machine is easily bent. In this case, the correction calculation is performed so that the coordinate value of the inferred point of action becomes small.

Further, as the jacking speed becomes faster, the shield machine itself becomes hard to bend. In this case,
A correction calculation is performed so that the coordinates of the inferred point of action become large.

As described above, instead of using the excavation environment item as a condition for inferring the shield jack action point, it is used as a correction calculation for the inferred shield jack action point.
It is configured to use the excavation environment item. As a result, according to the present invention, it is possible to obtain an appropriate shield jack action point in consideration of the excavation environment item without complicating the inference of the action point.

Therefore, by setting the jack pattern on the basis of the corrected action point and driving it, the shield machine can be accurately controlled along the planned line.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, preferred embodiments of the present invention will be described in detail with reference to the drawings.

First Embodiment FIG. 3 shows a preferred embodiment of the shield machine according to the present invention. In the shield machine 10 of the embodiment, the rotary cutter 12 is provided on the front surface of the cylindrical shield skin plate 22 and is configured to excavate the face 100. The excavated earth and sand is once taken into a chamber 20 formed between the rotary cutter 12 and the partition wall 18, and is further carried out by using the screw conveyor 14.

At the rear end side of the skin plate 22, a tail plate 24 is extended via a center bent portion 23. Between the tail plate 24 and the segment 26 assembled in a ring shape, a water-proof treatment is performed as is well known.

The shield machine 10 includes a plurality of (eight in the embodiment) shield jacks 16 arranged in a ring shape on the inner peripheral surface side of the skin plate 22. Then, extend these shield jacks 16,
By contacting the segment 26, a reaction force for propulsion can be obtained, and thus the shield machine 10 can be moved forward against the earth pressure of the face 100.
Incidentally, 28 is an earth pressure gauge for measuring earth pressure in the chamber 20.

The arrangement of the shield jacks 16 is shown in FIG. In the present embodiment, the shield jacks 16 are arranged symmetrically in the left-right direction and are also arranged vertically symmetrically. Thereby, for example, the shield 1 excavator 1
In order to change the direction of 0 to the left side, the shield jack on the right side is driven to give the shield machine 10 a counterclockwise turning moment. In order to change the direction of the shield machine 10 to the right, an operation reverse to the above may be performed.

Further, in order to change the direction of the shield machine 10 to the upper side, the lower shield jack 16 may be driven to give the shield machine 10 an upward turning moment.

Therefore, a jack pattern is set in any combination from the plurality of shield jacks 16-1, 16-2, ... 16-8 shown in FIG. 4, and the set hydraulic pressure of each jack is controlled to a high pressure. By doing so, the action point P of the shield jack can be set at an arbitrary position. FIG. 4 shows an example of the action points set by the jack pattern. In FIG. 4, it is shown that the action points can be set at the positions A, B, C, D depending on how the jack patterns are set. It represents.

As described above, it is possible to select and set arbitrary operating points A, B, C and D by setting the pattern of the shield jack which contributes to the propulsion to which the high hydraulic pressure is applied.
However, only by setting the jack pattern that contributes to such propulsion, it is possible to set only a very rough point of action,
For example, it is impossible to set the action point as shown by P in FIG. In order to solve this problem, the extension of the shield jack that does not contribute to the propulsion is made to follow the shield jack that contributes to the propulsion in synchronization, and the action points of the shield jack that contributes to the propulsion (for example, A, B, and
C and D) and the desired point of action P are displaced, the hydraulic pressure of the shield jack that does not contribute to the propulsion is increased so that they coincide with each other, and the moment due to the shield jack that contributes to the propulsion is increased. It is configured to generate a reverse moment.

As a method for selecting and setting such a jack pattern, for example, the technique of Japanese Patent Application No. 4-325149 proposed by the applicant of the present patent application can be adopted.

In addition to this, the following jack pattern selection and setting method may be adopted. That is, while supplying a high hydraulic pressure to the shield jack that contributes to the promotion of the shield machine, the low pressure oil is supplied as a low-pressure synchronizing pressure to the shield jack that does not contribute to the promotion of the shield machine to extend the shield jack. Follow the tuned shield jack that contributes to. At this time, if there is a displacement between the point of action (for example, A, B, C, D) of the shield jack that contributes to the propulsion and the desired point of action P, the two points should be matched so that they coincide with each other. The high hydraulic pressure supplied to the shield jack that contributes to propulsion is reduced to a pressure higher than the low pressure synchronizing pressure, and this is supplied as high pressure synchronizing pressure to a part or all of the shield jacks that do not contribute to the propulsion and is propelled to the shield jack. The force is exerted to change the center of the moment of the shield machine, and the direction of the shield machine is controlled.

As a method of selecting and setting such a jack pattern, for example, the technique disclosed in Japanese Patent Laid-Open No. 4-325150 proposed by the applicant of the present patent application can be adopted.

FIG. 5 shows an example of controlling the direction of the shield machine 10 so as to move along the planned line 200 by setting the jack pattern. In order to advance the shield machine 10 along the planned line 200 of the tunnel, the reference point of the shield machine 10 is on the planned line 200, and the direction of the excavation route 300 is the direction of the planned line 200. Need to match.

Therefore, the current reference point position of the shield machine 10b is detected, and the planned line 20 of the shield machine 10 is detected.
A horizontal and vertical position deviation amount Δδ from 0 and a horizontal and vertical deviation angle Δθ in the excavation direction with respect to the planned line 200 are obtained as the deviation amount.

Based on the deviation amount thus obtained, the target course of the shield machine 10 is calculated, and the shield jack action point for guiding the shield machine 10 to the target course is set.

Then, a combination pattern of the shield jacks 16 is set and driven according to the set shield jack action point. This enables shield machine 1
The shield machine 10 can be advanced along the planned line 200 by controlling the direction of 0 toward the target course.

FIG. 1 shows a shield machine 10 of this embodiment.
A block diagram of a directional control device used in the is shown.

The direction control device of the embodiment includes a position data detection unit 40 for detecting the current position and the direction of the excavation of the shield machine 10, a memory 42 in which data of the planned line 200 of the shield machine 10 is stored, and A target course calculation unit 44 that calculates a target course of the shield machine 10 based on the position data and the data of the planned line 200 is included.

As the position data detecting section 40, a gyro compass or a progress state detecting device proposed in Japanese Patent Application No. 4-326137 of the present applicant can be used. Further, instead of the position data detection unit 40, the shield machine 10
The position data of may be obtained by surveying or the like.

The target course calculation unit 44 includes the data of the planned line 200 stored in the memory 42 and the position data detection unit 4
Based on the position data input from 0, the amount of deviation of the shield machine 10 with respect to the planned line 200 in the horizontal and vertical directions (positional deviation amount, angular deviation amount) is calculated. Then, the target course of the shield machine 10 is calculated based on the calculated deviation amount.

In this embodiment, in order to avoid sudden direction control of the shield machine 10, a planned line 200 as shown in FIG.
When the amount of positional deviation with respect to is small, the target point T is set so as to coincide with the planned line 200 200 mm ahead. Further, as shown in FIG. 7, when the displacement amount is large,
It is formed so that the target course 400 is set so as to coincide with the planned line 200 one ring ahead (1000 mm).

That is, in the case shown in FIG. 6, 200 m
Target line T0, T1, T2, ... at the pitch interval of m
00 Set above.

Then, the control displacements Δδh and Δδv in the horizontal and vertical directions are calculated and output as the amounts of deviation with respect to the planning line 200. Further, the control angles Δθh and Δθv in the horizontal and vertical directions are calculated and output as the angle difference between the planned line 200 and the excavation direction of the shield machine.

In the case shown in FIG. 7, the target course 400 is set so that the shield machine will ride on the planned line 200 at the end of the segment for one ring. At this time, 100
When the 0 mm wide segment is used, the target points T1 ... T4 are set on the target course 400 at intervals of 200 mm.

The control displacement is calculated and output as the amount of deviation from each target point, and the control angle is calculated and output as the angle difference between the target path 400 and the excavation direction of the shield machine.

Then, the control amounts (Δδh, Δθ) in the horizontal and vertical directions calculated by the target course calculating unit 44.
h) and (Δδv, Δθv) are input to the action point inference unit 46.

The point-of-action inferring unit 46, based on the control amount thus input, that is, the control amount in the horizontal and vertical directions for guiding the shield machine 10 to the target course,
It is configured to infer the shield jack point of action. In the embodiment, the shield jack action point is inferred using fuzzy inference.

FIG. 8 shows a fuzzy inference method performed by the example of the action point inference unit 46, and FIG. 9 shows an example of a control rule used for this fuzzy inference.

In the embodiment, as shown in FIG. 8A, the input membership function corresponding to the angular deviation amount and the positional deviation amount and the output membership function shown in FIG. 8B are preset. Has been done. Here, NB is negative big, NS is negative small, ZE is zero, PS is positive small, and PB is positive big.

For example, when the horizontal angular displacement amount and positional displacement amount are a0 and b0, respectively, the shield jack action point c0 (x0) in the horizontal direction is inferred as follows.

First, the input membership functions corresponding to the angular displacement amount a0 are PS and ZE. Then, from FIG. 8A, the fitness of the membership functions of PS and ZE is 0.6 and 0.4.

Further, the input membership functions corresponding to the positional deviation amount b0 are NS and ZE, and their matching degrees are 0.7 and 0.3, respectively.

Therefore, the angle deviation amount a0 and the positional deviation amount b
The combination of membership functions corresponding to 0 is shown in FIG.
There are four patterns (1) to (4) shown in (A).

The four input patterns are collated with the control rules shown in FIG. 9 to obtain output patterns corresponding to the respective input patterns as shown in FIG. 8 (B).

For example, in the input pattern (1), the amount of angular displacement and the amount of positional displacement are PS and NS.
The output pattern becomes ZE when collated with the control rule shown in FIG. Further, in the input pattern 2, since the angular displacement amount and the positional displacement amount are PS and ZE, respectively, the output pattern thereof is PS.

In this way, four output patterns can be obtained. At this time, the fitness of the output membership function takes a smaller value of the fitness of the two input membership functions. For example, in pattern 1, the smaller value of the fitness values 0.6 and 0.7 of the two input membership functions is 0.6. In pattern 2, 0.6, 0.3
The smaller value is 0.3.

In this way, when four output patterns are obtained, as shown in FIG. 8C, these patterns are integrated and the center of gravity c of the output membership function is determined.
Calculate the value of 0. At this time, a large value is selected for the matching degree of the overlapping portion of the two patterns. For example, the overlapping portion of NS and ZE is larger in ZE, so this value is adopted.

When a combination of output patterns as shown in FIG. 8C is obtained in this way, the center of gravity c0 of this combination pattern becomes the horizontal shield jack action point.

As described above, the action point inference unit 46 of the embodiment first moves in the horizontal direction on the basis of the horizontal control amount, that is, the angular deviation amount and the positional deviation amount input from the target course calculating unit 44. Estimate the shield jack action point x0.

Next, using the same method, the action point inference unit 4
Reference numeral 6 estimates the shield jack action point y0 in the vertical direction based on the control amount in the vertical direction input from the target course calculation unit 44.

FIG. 10 shows the shield jack action point obtained by such fuzzy inference, that is, the moment position P in the horizontal and vertical directions of the shield jack.
It is shown. Here, X and Y represent the center-of-gravity position in the horizontal direction and the center-of-gravity position in the vertical direction, which are obtained by the method shown in FIG.

By the way, when the excavation environment of the shield machine 10 is constant, the value of the action point P does not change,
When the shield excavation environment changes, the value of the action point P also changes. Incorporating such changes in the excavation environment into fuzzy inference makes the fuzzy inference itself complicated and is not practical.

In this embodiment, the position of the shield jack action point P0 obtained by fuzzy inference is changed to the action point correction unit 4
I am entering 8.

Then, the action point correction unit 48 is formed so as to perform the correction calculation of the input shield jack action point P based on the data of the excavation environment item of the shield machine 10. Examples of excavation environment items used for such correction calculation include copy cutter stroke of shield excavator, center bending angle, jack speed, tail clearance, cutter, etc., and soil conditions such as water content ratio, particle size, earth pressure, etc. To be

In the embodiment, the copy cutter stroke,
The center bending angle, jack speed, tail clearance, etc. are adopted as the excavation environment items to perform the correction calculation.

FIG. 2 shows a concrete structure of the working point inference unit 46 of the embodiment. The working point inference unit 46 calculates the correction of the working point P0 based on the copy cutter stroke, the center bending angle, and the jacking speed. And a second correction unit 4 that performs correction calculation based on the tail clearance.
8b and.

FIG. 11 shows the copy cutter stroke,
Adjustment parameters n1, n2, and n3 corresponding to the center bending angle and the jack speed are shown. For example, if the stroke of the copy cutter is increased, the shield machine 10 is likely to bend by the amount of excess excavation around the rotary cutter 12. Further, if the center bending angle of the shield machine 10 is set, that is, the middle bending angle is set between the skin plate 12 and the tail plate 24, the shield machine itself is easily bent by that amount. Therefore, as shown in FIGS. 11 (A) and 11 (B),
As the copy cutter stroke and the center bending angle increase, the values of the adjustment parameters n1 and n2 are set smaller. On the contrary, when the jacking speed becomes faster, the shield machine 10 itself becomes difficult to bend. Therefore, FIG.
As shown in (C), the adjustment parameter n3 is set to increase as the jack speed increases.

The first correction section 48 adjusts each of the adjustment parameters n1, n2 shown in FIG. 11 based on the input copy cutter information, center bending angle information, and jack speed information.
Obtain n3 and input it to the shield jack action point P
An operation for multiplying the x0 and y0 coordinates of 0 is performed. Then, the values x ′ and y ′ obtained by this multiplication are the second correction unit 48.
Input to b.

The tail plate 2 is provided at a plurality of locations on the inner peripheral surface of the tail plate 24 of the shield machine 10.
A proximity switch 30 for detecting the outer peripheral surface of the segment 26 to be assembled inside 4 is provided. The proximity switch 30 includes a plurality of switch parts having different sensitivities, and measures the distance between the tail plate 24 and the outer peripheral surface of the segment 26, that is, the tail clearance. Then, the measured value is input to the second correction unit 48b.

In the second correction section 48b, the correspondence between the tail clearance and its adjustment parameter n4 is stored in advance as data, and the adjustment parameter n4 corresponding to the input tail clearance value is obtained. Then, this adjustment parameter n4 is multiplied by the x ', y'coordinates of the action point input from the first correction section 48a, and this is set as the corrected X, Y coordinates of the shield jack action point P. Output to the unit 50.

In this way, the action point correction unit 4 of the embodiment
In 8, the x0 and y0 coordinates of the shield jack action point P0 fuzzy inferred by the action point inference unit 46 are corrected and calculated based on the input excavation environment items such as the copy cutter stroke, the center bending angle, the jack speed, and the tail clearance. Then, the new shield jack action point P corresponding to the excavation environment item is output to the jack pattern setting unit 50.

The jack pattern setting section 50 selects and sets an optimum jack pattern for obtaining the point of action based on the corrected X and Y coordinates of the shield jack point of action P input in this way.

The jack control drive unit 52 is shown in FIG.
Seal jacks 16-1, 16-2, ... 16 shown in
-8 is drive-controlled according to the input jack pattern, and the shield machine 10 is directed to the target point T for direction control.

The direction control device of this embodiment has the above-mentioned structure, and its operation will be described below.

FIG. 12 shows a flow chart of a procedure for obtaining the shield jack action point by the direction control device of the embodiment.

The direction control device for the shield machine of the embodiment sets the target points T at a pitch interval of 200 mm, as shown in FIGS. Then, the horizontal moment X and the vertical moment Y of the action point P of the shield jack are calculated so as to move the shield machine 10 toward the set target point T, and the jack pattern matching the action point P is set. . The direction of the shield machine 10 is controlled by controlling the shield jack 16 based on the set jack pattern.

At this time, first, the target course calculation unit 44 and the action point inference unit 46, as shown in steps S10 and S20,
Fuzzy inference for determining the horizontal and vertical coordinates (x0, y0) of the point of action P0 of the shield jack 16 is performed.

In this fuzzy inference, first, the target course calculating unit 44 sets a target point T 200 mm ahead based on the positional deviation amount and the angular deviation amount of the shield machine 10 as shown in FIGS. The control amounts in the horizontal and vertical directions for guiding the shield machine 10 to T are calculated and output to the action point inference unit 46.

The action point inference unit 46 fuzzy infers the horizontal and vertical moment amounts x0 and y0 of the shield jack action point P0 based on the horizontal and vertical control amounts thus input. Here, x0 of the shield jack action point P0 obtained by fuzzy inference
, Y0 coordinates are P0 = (-0.2, 0.4).

Next, as shown in steps S11 and S21, the first correction section 48a, for the x0 and y0 coordinates of the operating point P0 obtained by the inference, copy cutter stroke data, center bending amount data, and jack speed. The adjustment parameters n1, n2, and n3 obtained from the data and the like are multiplied to perform the operation point correction calculation. Here, the value obtained by multiplying the adjustment parameters of n1, n2, and n3 is 110%. Thereby, the corrected value of the action point is (−0.
22, 0.44).

Next, as shown in steps S12 and S22, the second correction section 48b finds the adjustment parameter n4 based on the input tail clearance data, and uses this as the action point data output from the first correction section 48a. Perform multiplication operation. For example, when the adjustment parameter n4 is 90%, the XY coordinate of the action point is (-0.198,0.
396).

X of the action point P thus obtained,
For the Y coordinate, a correction calculation is performed based on the values of each of the excavation environment items such as copy cutter stroke, center bending angle, jack speed, and tail clearance with respect to the fuzzy inferred values of x0 and y0 of the shield jack action point P0. It becomes a thing.

Accordingly, in step S30, the corrected and calculated moment values X and Y in the horizontal and vertical directions are combined, and the position (X, Y) of the shield jack action point P is combined.
Ask for.

As described above, according to the present invention, the shield jack action point P in consideration of the excavation environment item of the shield machine 10 can be obtained without complicating the fuzzy inference, and as a result, various excavations are performed. Even under the environment, the direction control of the shield machine 10 can be performed well.

Second Embodiment Next, a preferred second embodiment of the present invention will be described. The same reference numerals are given to the members corresponding to the first embodiment, and the description thereof will be omitted.

The feature of this embodiment is that the action point inference unit 46 shown in FIG. 1 is formed as a neural network circuit shown in FIG. 13 (C). Here, in FIG.
(A) is the membership function (input membership function) of the antecedent part of the neural network circuit,
FIG. 13B shows the membership function (output membership function) of the consequent part. Further, the control rule between the antecedent part and the consequent part in this neural network circuit is determined as shown in FIG.

With this configuration, also in the direction control device of this embodiment, the x0 and y0 coordinates of the shield jack action point P0 are calculated based on the control amounts in the horizontal and vertical directions input from the target course calculation unit 44. Can be inferred. Note that FIG. 13C shows a case where the horizontal position x0 of the shield jack action point P0 is obtained based on data such as the amount of deviation and the angle of deviation with respect to the horizontal direction.

Here, for example, the output of the consequent part corresponding to the input control amount has a PS conformity of 0.45,
Assuming that the matching degree of ZE is 0.7, the center-of-gravity position calculation unit integrates these PS and ZE, and FIG.
The data as shown in (1) is created and the position of the center of gravity is obtained as the value of the horizontal acting point x0.

The process up to this point is the same as in the first embodiment, but in this embodiment, the value of the action point P0 thus obtained is compared with the result of the actual direction control, and the back propagation is performed. The method is characterized by adding a function of correcting the membership function by learning.

For example, as shown in FIG. 15B, as a result of direction control of the shield machine 10 based on the shield jack action point P0 obtained by fuzzy inference, the target point T
On the other hand, assume that the position actually moved is Q. At this time, the insufficient amount of position and angle is D2, θ2
Is. Therefore, based on the positional deviation amount D1 and the angular deviation amount θ1 between the original position where the control is started and the target point T, and the deviation amount D2 and the angular deviation amount θ2 between the actual movement position Q and the target point T,
The deviation amount insufficiency rate and the deviation angle insufficiency rate are calculated to create a teacher signal.

Then, the integrated data output from the center-of-gravity position calculation unit becomes as shown in FIG. 15C (that is, the center-of-gravity position of this combination data becomes the teacher signal lt). , Adjust the PS value. That is, the values of ZE and PS are, for example, 0.55 and 0.50.
Learn so that In this learning, 0.50 and 0.55 are input as teacher signals from PS and ZE in the consequent part of the neural network circuit. In this way, by using the method of backpropagation and adjusting the value of the membership function by learning, the entire circuit can be configured so that optimum output data can be obtained based on the input data.

FIG. 16 shows an example of the antecedent part membership function (input membership function) after learning.

Such a backpropagation method is generally performed in a neural network circuit, but the characteristic point here is that the action point correction section 48 makes a correction calculation based on the excavation environment item. In this case, the neural network circuit is configured not to perform learning.

FIG. 17 shows an example thereof.

For example, when the values of the copy cutter stroke, the center bending angle, the jack speed, etc. exceed the predetermined set values, the neural network circuit forming the action point inference unit 46 is formed so as not to perform learning. There is.
As a result, the neural network circuit can always perform learning under a constant condition and set an optimum membership function.

The present invention is not limited to the above embodiments, and various modifications can be made within the scope of the present invention.

For example, in the above-mentioned embodiment, the case where the shield jack action point is inferred by fuzzy inference has been described as an example, but other inference methods can be adopted if necessary.

Further, in the above-mentioned embodiment, the case where the direction control of the hydraulic shield machine is performed is described as an example, but the present invention is not limited to this, and the direction control of various types of shield machine other than this is performed. It can also be applied in cases.

[0098]

As described above, according to the present invention,
It is possible to obtain an accurate point of action of the shield jack in consideration of the excavation environment item of the shield machine and to control the direction so that the shield machine moves along the planned line.

Further, according to the present invention, the direction control of the shield machine can be automated by using a technique such as fuzzy inference. Especially, unlike the conventional direction control device using fuzzy inference, the shield machine Not only the amount of deviation from the planned line of the machine, but also jack speed of shield machine, center bending angle, copy cutter, tail clearance,
Corrective calculation based on various excavation environment items such as soil quality can be performed to find the optimum shield jack action point that matches the excavation environment of the shield excavator.Therefore, direction control to move the shield excavator along the planned line , It is possible to perform better.

[Brief description of drawings]

FIG. 1 is a block diagram showing a preferred example of a direction control device for a shield machine according to the present invention.

FIG. 2 is a detailed block diagram of an action point inference unit of the direction control device shown in FIG.

FIG. 3 is a schematic explanatory diagram of a shield machine used in the present embodiment.

4 is an explanatory diagram showing an arrangement of shield jacks of the shield machine shown in FIG. 3. FIG.

FIG. 5 is an explanatory diagram showing a state of a shield machine that moves in a state of being displaced from a planned line.

FIG. 6 is an explanatory diagram for setting a target point when the positional deviation amount of the shield machine with respect to the planned line is equal to or less than a predetermined threshold value.

FIG. 7 is an explanatory diagram for setting a target point when the amount of positional deviation with respect to the planned line is equal to or greater than a predetermined threshold value.

FIG. 8 is an explanatory diagram showing an example in which a shield jack action point is calculated using a fuzzy inference method.

9 is an explanatory diagram of a control rule used in the fuzzy inference shown in FIG.

10 is an explanatory diagram of a case where the shield jack action points for X and Y seats are obtained using the fuzzy inference shown in FIG. 8. FIG.

FIG. 11 is an explanatory diagram of adjustment parameters for a point of application obtained by fuzzy inference.

FIG. 12 is a flowchart showing the operation of the direction control device according to the embodiment.

13 is an explanatory diagram in the case where the action point inference unit of the direction control device shown in FIG. 1 is configured by a neural network circuit.

14 is an explanatory diagram of control rules of the neural network circuit shown in FIG.

15 is an explanatory diagram showing an example of a learning operation of the neural network circuit shown in FIG.

16 is an explanatory diagram of a membership function after learning of the neural network circuit shown in FIG.

17 is an explanatory diagram showing a learning timing of the neural network circuit shown in FIG.

[Explanation of symbols]

 10 shield machine 44 target course calculation unit 46 action point inference unit 48 jack pattern setting unit 50 jack control drive unit

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Koichi Ito 1-7-1 Kyobashi, Chuo-ku, Tokyo Toda Construction Co., Ltd. (72) Inventor Tadashi Higuchi 1-1-7 Kyobashi, Chuo-ku, Tokyo Toda Incorporated Co., Ltd. (72) Inventor Takeshi Yanagura 1-7-1 Kyobashi, Chuo-ku, Tokyo Toda Construction Co., Ltd.

Claims (6)

[Claims] 1. A target course calculating means for calculating a target course of the shield machine and a shield jack action point for guiding the shield machine to the target course based on a deviation amount of the shield machine from a planned line. Action point inference means, action point correction means for correcting the inferred shield jack action point based on the data of the excavation environment item of the shield machine, and a jack pattern corresponding to the corrected shield jack action point is set. A direction control device for a shield machine, comprising: a jack pattern setting means. 2. The target course calculating means according to claim 1, wherein the target course calculating means calculates a horizontal and vertical target course of the shield machine based on horizontal and vertical deviations of the shield machine from a planned line. A direction control device for a shield machine, wherein the point-of-action reasoning means infers a shield jack point of action for guiding the shield machine to the target course as a moment position in horizontal and vertical directions. 3. The operating point correction means according to claim 1, wherein the excavation environment item includes jack speed information of the shield machine, middle bending angle information, copy cutter stroke information, and tail clearance information. , A direction control device for a shield machine, which corrects a shield jack action point by using at least one of soil information. 4. The jack pattern setting means according to claim 1, wherein the jack pattern setting means sets a jack pattern of the shield jack that contributes to propulsion according to the corrected shield jack action point, and contributes to propulsion. The extension of shield jacks other than the shield jack is made to follow the shield jack that contributes to the propulsion in synchronization, and there is a displacement between the point of action of the shield jack that contributes to the propulsion and the corrected point of action. In this case, increase the hydraulic pressure of the shield jack that does not contribute to the propulsion so that they match.
A direction control device for a shield machine, which generates an inverse moment to a moment by a shield jack that contributes to propulsion. 5. The direction control device for a shield machine according to claim 1, wherein the action point inference means obtains the shield jack action point by fuzzy inference based on the deviation amount. . 6. The neural network circuit according to claim 1, wherein the action point inference means has an input layer, an intermediate layer, and an output layer, and is a neural network circuit that obtains the shield jack action point by inference. The neural network circuit is formed to perform a learning operation for modifying its membership function based on a backpropagation algorithm, and the learning operation is stopped and controlled during a correction operation based on the excavation environment item. Direction control device for shield machine.