JPH0688476A - Automatic direction control device of shield machine - Google Patents

Abstract

PURPOSE:To facilitate to set parameters or to grasp influences at changing them by dealing with only a qualitative part, to which direction a shield machine should be faced, in a fuzzy inference part. CONSTITUTION:The locus and the posture angle of a shield machine are detected by means of an offset estimation part 11, the relation of them is set into a model, and dislocation between the advancing direction and the posture angle is estimated as the offset angle. Next, a fuzzy inference part 12 computes dislocation of the position and dislocation of the posture angle of the shield machine from scheduled line information, position detection information, posture angle information, offset angle information, and the like, and outputs a posture control command value based on dislocation of the position and dislocation of the posture angle. Next, a posture control part 13 computes angular moment from the posture control command value and the posture angle detection information, computes the degree of one-sided push based on the angular moment, and selects an optimum jack pattern based on the degree of one-sided push.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a shield machine for forming a passage such as a tunnel in the ground, and more particularly to an automatic directional control device for carrying out a digging operation while minimizing the posture and position deviation from a planned tunnel line.

[0002]

2. Description of the Related Art In a shield machine, a circular excavation part is supported by a plurality of jacks from behind and is propelled, and direction control is required to propel it along a planned tunnel line. . Directional control is performed by applying a rotation moment to the excavation section in the propulsion direction by a plurality of jacks arranged at intervals in the circumferential direction, and this rotation moment turns on any of the plurality of jacks. You can change it. Normally, the number of jacks is 10 or more, and the combination of jacks turned on is called a jack pattern. For example, in the case of 16 jacks, the types of jack patterns reach 2 ¹⁶ .

To briefly explain the conventional direction control device, the current position and direction of the shield machine are compared with the position and direction of a predetermined planned route, and the deviation amount is used as an input to perform fuzzy inference. Using a fuzzy control device that calculates the amount of one-sided push degree change that reduces the amount of deviation, the fuzzy control device is used for each of the position and direction before the excavation in the past and those positions and directions after the excavation. Depending on the amount of movement, the membership function that expresses the amount of one-sided push change is modified to improve the accuracy of direction control, and the circumferential direction is calculated based on the amount of one-sided press change obtained from the corrected membership function. Of the jacks provided in the
There is one in which a jack to be driven by thrust is selected (for example, Japanese Patent Laid-Open No. 2-186097).

As shown in FIG. 10, the fuzzy control device includes a first fuzzy inference unit 51 that defines a rule based on a positional relationship, a second fuzzy inference unit 52 that defines a rule based on an angular relationship, and the above two types. And a third fuzzy inference unit 53 which defines a rule for determining the ratio (weighting coefficient a) of the fuzzy inference unit. The first and second fuzzy inference units 51 and 52 respectively output the amounts of change ΔE ₁ and ΔE ₂ of the _one- sided pushing degree as the inference results. The one-sided pushing degree is a dimensionless value obtained by dividing the rotational moment acting on the excavated portion by the jack thrust, and corresponds to the rotational moment when the jack mounting radius is 1 and the jack total thrust is 1. Also,
In FIG. 10, d is the amount of positional deviation from the planned line, Δd is the amount of positional deviation change, θ is the amount of angular deviation, and Δθ is the amount of change of angular deviation.

[0005] First, the amount of change in single push of the second fuzzy reasoning section 51 and 52 Delta] E _1, by combining Delta] E ₂ obtains pieces pressing degree of change Delta] E of the final output, but, Delta] E ₁ at this time ，
ΔE ₂ is weighted by the multiplier 54 of the weighting coefficient a and the multiplier 55 of the weighting coefficient (1-a), and the value of the weighting coefficient a is determined by the third fuzzy inference unit 53.

[0006]

However, in the fuzzy control device as described above, since the fuzzy inference unit has a three-stage configuration, it is difficult to grasp the influence of the input of each fuzzy inference unit on the output, and the member is difficult to understand. There is a problem that it is difficult to set parameters such as the ship function. In addition, there is no fixed relationship to how much the excavation section bends when the excavation section is pushed with a certain degree of pushing, and if the thrust at that time is different, the rotational moment will be different. On the other hand, even when the same rotational moment is applied, the posture angle varies depending on the soil condition.

In order to deal with the above problems, the outputs ΔE ₁ and Δ of the _first and second fuzzy inference units 51 and 52 are
A learning function is needed to adjust the membership function of E ₂ . This learning function is for increasing or discounting the output when the positional deviation amount d exceeds a specified value. In other words, this learning function is considered to correspond to gain adjustment in proportional control. However, what is changed and how to adjust it (for example, whether the relationship between the rotation moment and the posture angle has changed, or whether the jack thrust has changed, etc.) is treated as ambiguous, and the parameters are determined. The process is obscured. This is because the fuzzy inference was used to directly obtain the one-sided push degree, and thus another element was included in the input / output correspondence.

[0008] That is, elements such as the relationship between the rotational moment and the amount of change in the posture angle, which are unknown depending on the soil condition and must be actually excavated, and the trajectory of the shield machine, the planned line. It is because it is treated at the same time with the element that is determined only by the geometrical aspect, that is, whether or not it is placed on. For this reason, the input / output relation of the fuzzy inference unit becomes poor, and which parameter is changed
It is difficult to determine where and what kind of effect will occur.

In view of the above problems, an object of the present invention is to clarify the input / output relationship of a fuzzy inference unit by separating elements that change from moment to moment in relation to soil quality from fuzzy reasoning. An object of the present invention is to provide an automatic direction control device that makes it easy to grasp the influence of parameter setting and change.

[0010]

An automatic direction control device according to the present invention detects a trajectory and an attitude angle of a shield machine and models the relationship between them, and uses a deviation between the traveling direction and the attitude angle as an offset angle. An offset estimation unit for estimating and planning line information, position detection information, attitude angle detection information, and attitude control command values by fuzzy inference based on the position deviation and attitude angle deviation of the shield machine from the information such as the offset angle. A fuzzy inference unit that outputs the rotational moment is calculated from the posture control command value from the fuzzy inference unit and the posture angle detection information, and the degree of one-sided depression is calculated based on the rotational moment. An attitude control unit for selecting an optimum jack pattern based on the degree of one-sided pressing.

[0011]

In the present invention, the relationship between the rotational moment and the amount of change in the actual attitude angle and the offset angle are obtained from the actual excavation data, and the portion for learning for these soil conditions is separated from the fuzzy inference unit to make fuzzy inference. The inference unit handles only the trajectory control, that is, the qualitative part of which direction the shield machine should be directed.

[0012]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram of an automatic direction control device according to the present invention, which includes an offset estimation unit 11, a fuzzy inference unit 12, and an attitude control unit 13.

Now, the relationship among the offset angle, the amount of angular deviation, and the amount of positional deviation will be described with reference to FIGS. In FIG. 2, when the shield machine main body 14 slides sideways with respect to the direction of the trajectory (travel direction), the angle between the direction of the trajectory and the direction the shield machine body 14 is facing is the offset angle θ. It is _HOF .
On the other hand, in FIG. 3, if the shield machine main body 14 advances from a certain unit distance (for example, 40 mm) position i to the next unit distance position (i + 1) and becomes a state shown by a solid line, the shield machine main body 14 The distance from the point P determined in advance to the planned line S is the amount of positional deviation d.
It becomes _H. Further, at the position (i + 1), the shield machine body 14 is oriented in the direction indicated by the line D1, and the line D2 considering the offset angle θ _HOF with respect to the direction of the locus.
When an attempt is made to correct the target azimuth angle indicated by, the angle θ _H between the direction of the target azimuth angle and the direction the shield machine body 14 is currently facing is the amount of angular deviation. Also in the vertical direction,
Similarly, the offset angle θ _VOF , the angular deviation amount θ _V , and the positional deviation amount d _V are defined respectively.

The offset estimation unit 11 uses the gyro, inclinometer, and other measuring equipment installed in the shield machine main body 14 to detect the shield based on the trajectory information based on the position detection information and the attitude angle detection information. Excavator body 1
The relationship between the trajectory and the posture angle of 4 is modeled, and the deviation between the trajectory direction (travel direction) and the posture angle is offset angle θ _HOF ,
Estimate as θ _VOF . Then, in the planned line information sent from the setter (not shown), the offset angles θ _HOF and θ are included in the information about the angles in the horizontal direction and the vertical direction, respectively.
_{The VOFs} are added by the adder 16 and the horizontal and vertical target azimuth angles are output.

The fuzzy inference unit 12 includes first and second subtraction units 17 and 18, and the first subtraction unit 17 detects the detected horizontal and vertical posture angles and horizontal and vertical target orientations. The difference from the angle is calculated and output as the angle deviation amounts θ _H and θ _V. On the other hand, the second subtraction unit 18 takes the difference between the detected current position (horizontal direction, vertical direction) of the shield machine 14 and the position information (horizontal direction, vertical direction) of the planned line information, The positional deviation amounts d _H and d _V are output. When the inference unit 19 receives the angular displacement amounts θ _H and θ _V and the positional displacement amounts d _H and d _V from the first and second subtraction units 17 and 18, they perform horizontal and vertical excavation by fuzzy inference. The direction in which the machine should travel is calculated, and the addition of the offset angle is output as the attitude control command value.

FIG. 4 shows an example of a rule in the inference unit 19. For example, ++ (−−) is the upper (lower) in the vertical direction.
Direction, it means to output a command value that makes a large displacement in the left (right) direction of the horizontal direction, and + (-) means that it is slightly displaced in the up (down) direction and the left (right) direction. It means to output the command value.

The attitude control unit 13 includes a shield machine main body 1
One-sided pushing degree F in the horizontal and vertical directions for posture control of No. 4
_H and F _V are output, and a rotation moment calculation for calculating horizontal and vertical rotation moments M _H and M _V based on the posture control command value from the fuzzy inference unit 12 and the detected posture angle value. The device 13A, the one-sided push degree calculation device 13B for calculating the one-sided push degrees F _H and F _V based on the calculated rotational moment and the detected value of the jack thrust, and the jack pattern of the jack pattern based on the calculated one-side push degree. The jack pattern selection device 13C for selection. Since the calculation methods for the horizontal direction and the vertical direction are the same, only the horizontal direction will be described below.

First, the rotation moment calculation device 13A determines an angle (1 mm, for example, 10 mm) that should be corrected in one attitude control pitch (for example, 10 mm) from the attitude control command value in the horizontal direction and the detected attitude angle in the horizontal direction from the fuzzy inference unit 12. Calculate the correction angle). Next, the relationship between the rotational moment M _H and the amount of change Δθ _{H in the} amount of angular deviation is modeled, and the rotational moment corresponding to the corrected angle is calculated as the rotational moment M _H to be applied to the excavated portion using this model.

Assuming that the positional deviation amount at a unit distance position i of the shield machine is d _Hi , the angular deviation amount is θ _Hi , and the rotational moment applied at that time is M _Hi , the objective of the rotational moment calculation device 13A is , Unit distance L (eg 40 m
The purpose is to calculate the rotational moment M _{H, i + 1} at the next unit distance position (i + 1) which is advanced by m). For this purpose, first, it is necessary to obtain the angle change amount Δθ _Hi that changes from the position i to the next position (i + 1), and this is performed as follows.

When the horizontal angle changes from θ ₄ to θ ₁ while advancing one pitch (for example, 40 mm), the angle change amount Δθ _H is represented by (θ ₄ −θ ₀ ). If the rotational moment M _H acting during this period is constant, it is considered that there is a constant relationship between the angle change amount Δθ _H and the rotational moment M _H. The quantity Δθ _H is expressed by the following mathematical formula 1.

[0021]

[Equation 1] Here, K ₁ is a proportional constant, and M _Hof is a moment necessary to maintain a constant posture angle. Based on such an idea, the relationship between the rotation moment and the amount of change in angle is modeled by approximation by a linear expression. The procedure will be described below.

FIG. 5 shows an example of the calculated values of the angle θ and the rotation moment. Here, the section 6 is the latest section, the end point of the latest section 6, the average rotation moment M _H6 of the section 6 and the amount of angle change Δθ. _H6 is _calculated by the following equations 2 and 3.

[0023]

[Equation 2]

[Equation 3] Here, M _{11 to} M ₁₄ are rotational moments calculated during excavation of the section 6, θ ₀ is a detected value of the angle at the end point of the section 5, and θ ₄ is a detected value of the angle at the end point of the section 6.

FIG. 6a is a plot (P1) of the relationship between the average rotational moment and the angle change amount obtained by the above-mentioned method in the sections 1 to 5 where the excavation has already been completed before the section 6 (P1). To P5), and FIG.
The relation between the average rotational moment M _H6 obtained in the section 6 and the angle change amount Δθ _H6 is added to a as P6.
It should be noted that the past data used for creating the model in the section 6 are the four sections 2, 3, 4, and 5 before the latest section 6 here, and therefore the data of the section 1 is deleted in FIG. 6b. ing.

In this way, the model creation for section 6
A linear approximation model represented by Equation 1 is created from the data of 5 points in 5 sections, which is the combination of the data of the latest section 6 and the data of each of the sections 2 to 5 before that, and the proportional constant K _{1 is set} to the linear expression. The inclination of the approximate straight line and the moment M _Hof are used as offset values and are _calculated by the methods described below.

Referring to FIG. 7, in step SS1, the degree of data dispersion at 5 points is determined. This is to determine whether the data pattern of 5 points is a pattern to which the direct calculation method described later can be applied. The direct calculation method is a method of estimating the linear approximation line from the data of 5 points, and cannot be applied when the data of 5 points concentrate on 1 point. The data points need to be scattered in a relationship. 5
If the point data are appropriately scattered, the process proceeds to step SS2.

In step SS2, the slope K _{1 of the} linear approximation line is directly calculated from the data of 5 points as shown in FIG. 6b.
And the offset value M _Hof are calculated. A typical example of the calculation method is the least squares method, but other methods may be used. Inclination K ₁ and offset value M _Hof
Is calculated, it is determined whether or not the slope K ₁ is within the predetermined ranges K _{1 max} (upper limit value) and K _{1 min} (lower limit value) shown in FIG. 8 (step SS3).
SS4). Then, the inclination K ₁ is K _{1 min} <K ₁ <K
If the condition of _{1 max} is satisfied, the linear approximation model Δθ
= K ₁ (M−M _Hof ) is determined (step SS5).

On the other hand, if it is determined in step SS1 that the degree of data dispersion is not appropriate, or if it is determined in steps SS3 and SS4 that the slope K ₁ exceeds the predetermined range, the slope is determined in section 5. The obtained one is used (step SS6).

Next, at step SS7, the angle change amount Δθ _{H6 at} the end point of the section 6 is as shown by the broken line in FIG.
It is determined whether or not it is within the allowable range, and if it is within the allowable range, the offset value obtained in the section 5 is used (step SS8). This is because, if all the data are within a certain allowable range with respect to the previously created model, the previous model sufficiently exhibits the behavioral characteristics and the model need not be changed.

On the other hand, if the angle change amount Δθ _H6 is out of the predetermined range as indicated by point P6 in FIG. 9 in step SS7, the process proceeds to step SS9 to move the linear approximation straight line in parallel. And adjust the offset value. In this case, since the slope K ₁ is known as the previous value, the offset value M _{Hof6 of the} straight line passing through the point P6 in FIG. 9 can be obtained by the following formula 4.

[0031]

[Equation 4]

The offset value M _{Hof of the} latest model is obtained by averaging the offset values of several sections including the latest section 6 thus obtained. For example, if the number of sections is three, the offset value M _Hof of the latest model can be calculated by the following expression 5.

[0033]

[Equation 5] However, M _H4 and M _H5 are rotational moments obtained by model creation in sections 4 and 5, respectively, and Δ
θ _H4 and Δθ _H5 are the angular variations at the end points of the sections 4 and 5, respectively.

The angle change amount Δθ _Hi can be obtained by the linear approximation model created as described above. From this angle change amount Δθ _Hi , the rotational moment M in the horizontal direction can be obtained.
_{H, i + 1} can be obtained. The rotational moment in the vertical direction can be similarly obtained.

In the one-sided push degree calculation device 13B, the jack thrust force measured by the shield machine main body 14 and the rotational moment command values M _H and M _V are used for the horizontal and vertical directions expressed by the following formulas 6 and 7. One-sided push degree (power point coordinates)
Calculate F _H and F _V.

[0036]

[Equation 6]

[Equation 7] Here, n is the number of jacks used for excavation, r is the attachment radius of the jack, and F _J is the jack thrust force per jack (supply hydraulic pressure to the jack × jack cylinder area).

The jack pattern selection device 13c selects the optimum jack pattern from the jack pattern table prepared in advance corresponding to the combination of the one-side pushing degrees F _H and F _V.

As described above, according to the present invention, the pushing degree F _H , F _V is calculated every time the shield machine advances one short distance (for example, 40 mm), and the optimum pushing degree is obtained. Select the jack pattern and proceed to the next section. In the embodiment, the model is created every 40 mm, but by reducing this to 20 mm and 10 mm, the model becomes closer to continuous control and a more accurate model is created. The tunnel construction accuracy can be improved.

[0039]

As described above, the present invention separates from the fuzzy inference unit a numerical value that changes every moment due to the soil condition, such as the offset angle and the relationship between the rotational moment and the angle change amount, and the fuzzy inference unit is separated. It is a direction control device that limits the relations handled by the department to only the parts that are easy to understand qualitatively called rules. The fuzzy inference unit should proceed from the current state (position, direction, etc.) of the shield machine. I try to determine the direction. This is a problem of how to take the locus, and can be determined without being influenced by soil quality or the like. Therefore, it is not necessary to change the parameters during the actual construction. As a result, the relationship between the inputs and outputs of the fuzzy inference unit is clarified, and it is easy to understand the effect of parameter settings and changes. In addition, the parameters can be tuned in advance by a desktop simulation.

On the other hand, since the numerical values that change every moment due to the relationship with the soil such as the relationship between the offset angle and the rotational moment and the angle change amount are estimated from the actual data by the offset estimation unit and the attitude control unit, You can directly see what has changed and how. Therefore, it is easy to judge whether the estimation method is good or bad, and the reliability of the estimation is improved.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an exemplary embodiment of the present invention.

FIG. 2 is a diagram for explaining an offset angle generated in the shield machine main body.

FIG. 3 is a diagram showing a relationship among an angular deviation amount, a positional deviation amount, and an offset angle.

FIG. 4 is a diagram showing an example of rules of a fuzzy inference unit shown in FIG. 1.

FIG. 5 is a diagram showing an example of a detected value of a posture angle and a calculated value of a rotation moment.

FIG. 6 is a diagram showing an example of a model based on a relationship between an angle change amount and a rotation moment.

FIG. 7 is a flowchart for explaining the operation of creating the model shown in FIG.

FIG. 8 is a characteristic diagram for explaining the determination condition in FIG.

9 is a characteristic diagram for explaining another determination condition in FIG. 7. FIG.

FIG. 10 is a diagram for explaining the operation of the fuzzy control device in the conventional automatic direction control device.

[Explanation of symbols]

 11 Offset Estimating Unit 12 Fuzzy Inference Unit 13 Attitude Control Unit 13A Rotational Moment Calculator 13B One-sided Push Degree Calculator 13C Jack Pattern Selector 14 Shield Machine

Claims (1)

[Claims] 1. A tunnel is excavated while propelling the excavation section with a plurality of jacks, and the excavation section is set as a planned line while appropriately selecting a jack pattern that defines which one of the plurality of jacks is turned on. In the automatic direction control device of the shield machine for propelling along, the trajectory and the attitude angle of the shield machine are detected and the relationship between them is modeled, and the deviation between the traveling direction and the attitude angle is estimated as an offset angle. Outputs a posture control command value by fuzzy inference based on the positional deviation of the shield machine and the deviation of the attitude angle from the offset estimator, planning line information, position detection information, attitude angle detection information, and information such as the offset angle. A rotation moment is calculated from the fuzzy inference unit, the attitude control command value from the fuzzy inference unit and the attitude angle detection information, and the rotation moment is calculated. An automatic direction control device for a shield machine, comprising: an attitude control unit that calculates a one-sided pushing degree based on a turning moment, and selects an optimum jack pattern based on the calculated one-sided pushing degree. .