JP7061893B2 - Shield excavator control system and shield excavator control method - Google Patents

Description

The present invention relates to a shield excavator control system and a shield excavator control method.

Conventionally, when constructing a tunnel or the like by the shield method, the construction environment (the state of the ground such as soil quality and water pressure) at the site excavated by the shield excavator changes from moment to moment depending on the position of the site. Therefore, it is necessary for the operator to control the traveling speed and traveling direction in excavation of the shield excavator in response to changes in the construction environment.
That is, if the operator does not properly control the shield excavator for each of the sites with different construction environments, the accuracy and safety of the excavated tunnel design will decrease.

In addition, the operator develops the experience of controlling the shield excavator in response to changes in the construction environment and improves the skill level by constructing the tunnel at various sites.
When controlling a shield excavator at a site during excavation, a skilled operator applies control knowledge in a similar construction environment in the past to control the current site construction environment. .. However, in the case of an operator with a small number of construction sites, in a construction environment that he has never experienced, it is not possible to control an appropriate shield excavator in the construction environment with poor experience and basic operation knowledge.

For this reason, the accuracy and safety of the design of the tunnel to be excavated vary depending on the operator's skill in controlling each shield excavator.
In order to solve this problem, there is a configuration in which the shield excavator is automatically operated based on the measurement data indicating the rotational state of the cutter of the shield excavator and the propulsion state of the propulsion jack during excavation (see, for example, Patent Document 1).

Japanese Unexamined Patent Publication No. 07-71189

In the above-mentioned Patent Document 1, the operation of the shield excavator by a skilled operator cannot be sufficiently reproduced. That is, since the control is performed by comparing the measured measurement data with the set value, the control is appropriately performed according to the ever-changing construction environment of the site, unlike the control based on the experience of a skilled operator. It is not always the case, and it cannot be said that the accuracy and safety of the excavated tunnel design will be improved.

In addition, using a machine learning model in which human emotion analysis and device operation are learned in advance using teacher data, measurement values similar to teacher data are input to estimate analysis results and operation settings. Is commonly done.
Therefore, it is conceivable to estimate the set value for performing an operation closer to that of a more skilled operator by using a machine learning model in which the operation of the shield excavator is learned by machine learning using the operation of a skilled operator as teacher data. ..

In the machine learning model described above, the set value as the operation amount to be given next to operate the shield excavator is obtained based on a plurality of data including the measured values obtained from the shield excavator.
However, when the measured value is an abnormal value or the measured value is missing, the machine learning model does not require a set value for properly operating the shield excavator. In addition, if an abnormal set value is required, there is a risk of adversely affecting the safety and quality of construction using a shield excavator.

The present invention has been made in consideration of the above circumstances, and when the control of the shield excavator is executed by the machine learning model corresponding to the construction environment of a skilled operator, the shield excavator inputs to the machine learning model. It is an object of the present invention to provide a shield excavator control system and a shield excavator control method for controlling a shield excavator without deteriorating the safety and quality of construction when an abnormal value or a missing value is present in the data to be performed.

In order to solve the above problem, the shield excavator control system of the present invention uses the measured value of the monitoring item data for monitoring the operation status of the shield excavator as input data, and controls the shield excavator output from the machine learning model. The monitoring item data for monitoring the excavation state of the shield excavator, which is input from the detector provided in the shield excavator, is the shield excavator control system that controls the shield excavator by the control data used in. A machine learning model unit that outputs the control data based on the measured value, a rule-based data generation unit that derives the control data based on the measured value and a predetermined rule, and the control based on the measured value. A control unit that performs processing such as outputting data to the machine learning model unit, outputting data to the rule-based data generation unit, or stopping the shield excavator, and measuring the monitoring item data. The data determination unit for determining whether the value is missing or abnormal, and the measured value of the monitoring item data are within the data range of the teacher data used when learning the machine learning model. It is provided with a data range determination unit that determines whether or not, and when the measurement value of the monitoring item data is missing or an abnormal value, the control unit stops the control of the shield excavator and the monitoring item. When the measured value of the data is outside the data range, the rule-based processing unit is made to generate the control data, and when the measured value of the monitoring item data is within the data range, the control data is generated. Is performed by the machine learning model unit .

In the shield excavator control method of the present invention, the measured value of the monitoring item data for monitoring the operation status of the shield excavator is used as input data, and the shield is based on the control data used for controlling the shield excavator output from the machine learning model. It is a shield excavator control method that controls an excavator, and the machine learning model unit measures the monitoring item data that monitors the excavation state of the shield excavator, which is input from a sensor provided in the shield excavator. A machine learning model process that outputs the control data based on the value, a rule-based data generation process in which the rule-based data generation unit derives the control data based on the measured value and a predetermined rule, and a control unit , A control process for causing the machine learning model unit to output the control data, the rule-based data generation unit to output the control data, or the shield excavator to be stopped based on the measured values. And , for the data determination process of determining whether the measured value of the monitored item data is missing or an abnormal value, and when the measured value of the monitored item data is used for learning a machine learning model. The control process includes a data range determination process for determining whether or not the data is within the data range of the teacher data, and the control process includes the shield excavation when the measured value of the monitoring item data is missing or abnormal. When the control of the machine is stopped and the measured value of the monitored item data is out of the data range, the rule-based processing unit is made to generate the control data, and the measured value of the monitored item data is the data range. If it is, it includes having the machine learning model unit generate the control data .

According to the present invention, when the control of the shield excavator is executed by the machine learning model corresponding to the construction environment of a skilled operator, there is an abnormal value or a missing value in the data input from the shield excavator to the machine learning model. In this case, it is possible to provide a shield excavator control system and a shield excavator control method that control a shield excavator without deteriorating the safety and quality of construction.

It is a figure explaining the shield excavator of the object controlled by the shield excavator control system of this embodiment. It is a figure which shows the configuration example of the shield excavator control system by this embodiment. It is a figure which shows the structural example of the monitoring item data table written in the storage part 20. It is a figure explaining the relationship between a training data range and a limit value range. It is a figure of the flowchart which shows an example of the control operation of the shield excavator by the shield excavator control system of this embodiment.

Hereinafter, embodiments of the shield excavator control analysis system according to the present invention will be described with reference to FIG.
FIG. 1 is a diagram illustrating a target shield excavator controlled by the shield excavator control system of the present embodiment.
FIG. 1A shows a conceptual diagram of a shield excavator to be controlled and analyzed by the shield excavator control analysis system of the present embodiment.
The excavation mechanism 50 is a mechanism for excavating the shield excavator 30 while constructing the primary lining S by assembling a segment with an elector (not shown) at the rear portion of the cylindrical skin plate 2 in the arrow D1 direction. .. In the excavation mechanism 50, a mud chamber 7 is provided behind the annular and face plate type cutter 10 having a face 23 in the direction of arrow D1. The mud material 9 is injected into the mud chamber 7 from the mud injection pipe 8 and is strongly kneaded by a kneading blade (not shown) to convert the excavated earth and sand into mud, and the mud pressure is changed to earth pressure and water pressure. By balancing with, the face 23 is stabilized and excavation is performed. Here, the excavated residual soil accumulated in the mud-making chamber 7 of the excavation mechanism 50 is introduced into the screw conveyor 60 and discharged to the outside of the excavated tunnel via the conveyors 62 and 63. The gantry M supports each of the screw conveyor 60 and the conveyors 62 and 63. Further, although not shown, the excavation mechanism 50 is provided with a propulsion jack (described later), and the excavation direction and the propulsion speed are controlled by the propulsion jack.

In the present embodiment, the propulsion direction and propulsion speed of the skin plate 2 are controlled by hydraulic control of the propulsion jack (hereinafter, may be simply referred to as direction control), and the rotation speed of the screw of the screw conveyor 60 (described later). By controlling the screw speed), the mud pressure is controlled (hereinafter, it may be simply referred to as soil pressure control). The shield excavator control system described later acquires each of the monitoring item data indicating the state of these controlled objects, and estimates the control data for controlling the shield excavator by a machine learning model (described later).

In the present embodiment, the propulsion direction and propulsion speed of the skin plate 2 are controlled (direction control) by hydraulic control or the like as control data of the propulsion jack, and the rotation speed of the screw (screw rotation) is controlled as control data of the screw conveyor 60. By controlling the speed) and the like, control data for controlling the mud pressure (soil pressure control) shall be acquired. Further, in the present embodiment, the monitoring item data includes data for monitoring the excavating construction environment and the operating state of the shield excavator 30 (monitoring item data), for example, cutter torque, cutter speed, propulsion pressure, and propulsion. Speed, propulsion speed instruction sheet deviation value, control earth pressure, face earth pressure average value out of range dummy, agitator torque, screw speed, primary screw pressure, secondary screw pressure, NO. 1 copy stroke, NO. 1 Copy stroke indicated value difference, NO. 1 copy position, NO. 1 Copy position instruction sheet deviation value, pitching, pitching instruction value difference, rolling, rolling instruction value difference, vertical center fold angle, vertical center fold angle instruction value difference, left / right center fold angle, left / right center fold angle instruction value difference, S / M front body direction, S / M front body direction indicated value difference, S / M rear body direction, S / M rear body direction indicated value difference, planned line horizontal deviation (control point), planned line vertical deviation (control point), Direction (control point), direction indication value difference (management point), planned route direction (control point), pitch (management point), planned route pitch (management point), etc.

FIG. 1B shows a conceptual diagram illustrating a propulsion jack that propels the excavation mechanism 50. As shown in FIG. 1 (b), the force point for propelling the surface of the skin plate 2 is set depending on which position the propulsion jack is driven. In FIG. 1 (b), 22 propulsion jacks are shown, but the number is not limited. By distributing the jack pressure of each propulsion jack, the propulsion jack force point position Fx on the x-axis and the propulsion jack force point position Fy on the y-axis are set, and the mark P becomes the position of the force point. The direction in which the excavation mechanism 50 is propelled is set by applying the propelling pressure to the position corresponding to the force point on the surface of the skin plate 2. Control in this direction is performed by a propulsion jack pattern that indicates which of the propulsion jacks to drive. Then, the propulsion speed is set by the amount of oil (oil amount) supplied per unit time to increase the jack pressure.

FIG. 2 is a diagram showing a configuration example of a shield excavator control system according to the present embodiment. In FIG. 2, the shield excavator control system 1 includes a monitoring item data input unit 11, a data determination unit 12, a correction complement unit 13, a data range determination unit 14, an abnormality stop determination unit 15, a control unit 16, and a machine learning model unit 17. , Each of a rule-based data generation unit 18, an output check unit 19, and a storage unit 20 is provided.

The monitoring item data input unit 11 is the data of each of the above-mentioned monitoring items at the timing when a timekeeping signal indicating the passage of a predetermined cycle time (for example, 1 second) from a timekeeping means (timer or the like) (not shown) is supplied. The above-mentioned monitoring item data is acquired from each of the detection means (sensors, measuring instruments, etc.) provided in each part, with the above-mentioned measured value. Further, the monitoring item data input unit 11 sequentially writes and stores the measured values of the acquired monitoring item data in the monitoring item data table in the storage unit 20. At this time, the monitoring item data input unit 11 stores a plurality of measured values for a predetermined period for each type of monitoring item data, and when writing a new measured value, overwrites and writes the oldest measured value. To memorize with.

Further, the monitoring item data input unit 11 supplies the measured values to the machine learning model unit 17 and the rule-based data generation unit 18 as standardized (standardized) numerical values or averaged numerical values based on the monitoring item data. If necessary, perform standardization and averaging calculations. In the case of the measured value such as the controlled earth pressure in the monitoring item data, the averaging process may be supplied as an abnormally high or abnormally low value due to superimposition of noise. As a result, repeated abnormal stops due to noise affect the progress of construction work by the shield excavator. Therefore, the monitoring item data input unit 11 obtains the measured value and the moving average in the past predetermined period for the measured value of the monitoring item data set in advance, and uses this moving average as the measured value of the monitoring item data. Used as.

FIG. 3 is a diagram showing a configuration example of a monitoring item data table written in the storage unit 20. For each record, columns for monitoring item data type, acquired data, loss period, corresponding sensor measurement range, learning data range, and limit value range are provided. The monitoring item data type indicates the type of monitoring item data. The acquired data is the latest measured value of the acquired monitoring item data. The loss period is the elapsed time since the measured value to be measured starts to be supplied as the missing value. As the corresponding sensor measurement range, each of the upper limit value and the lower limit value of the measurement range of the sensor for acquiring the monitoring item data is shown. The learning data range is the range of the measured values of the monitoring item data used for the learning teacher data of the machine learning model (described later) in the machine learning model unit 17 (the range consisting of the upper limit value X2 and the lower limit value X1 of the measured values). Shows. The limit value range is set for each type of monitoring item data, and is a limit value range indicating danger as a measured value (a range consisting of an upper limit value Y2 and a lower limit value Y1 of the limit value).

FIG. 4 is a diagram illustrating the relationship between the learning data range and the limit value range.
The learning data range of the monitoring item data is the range P1 which is equal to or less than the upper limit value X2 of the control value and is greater than or equal to the lower limit value X1. The permissible range of the monitoring item data is a range P22 that exceeds the upper limit value X2 of the control value and is equal to or less than the upper limit value Y2 of the limit value, and a range that is less than the lower limit value X1 of the control value and is greater than or equal to the lower limit value Y11 of the limit value. It is P11. The abnormal values are a value exceeding the upper limit value Y2 of the limit value and a value less than the lower limit value Y1 of the limit value.

Returning to FIG. 2, the data determination unit 12 determines whether the newly measured measured value is missing (for example, “0” data) or abnormal. At this time, the data determination unit 12 refers to the monitoring item data table in the storage unit 20, and if it exceeds the corresponding sensor measurement range, it is regarded as an abnormality. Then, the data determination unit 12 attaches additional information of the defect or abnormality to the measured value of the monitoring item data determined to be defective or abnormal, and outputs the measured value to the correction complement unit 13.

When the measurement value of the monitoring item data is supplied from the data determination unit 12, the correction complement unit 13 determines whether the additional information is missing or abnormal. Here, when the additional information is missing, the correction complement unit 13 refers to the monitoring item data table in the storage unit 20 and reads out the history of the measured values of the corresponding monitoring item data (measured values acquired in a predetermined period). , A complementary measurement value is generated by averaging the measurement values within a predetermined period, and the complemented measurement value is written and stored in the place of the missing measurement value. At this time, the correction complementing unit 13 counts the period during which the complementation continues from the time when the measurement value is first complemented, and writes the count value in the missing period of the monitoring item data table in the storage unit 20. Remember. In the present embodiment, when the measured value is missing and cannot be obtained, it is shown as a complement that the measured value created by using the measured value in the past history is used as the missing measured value, and the abnormal measured value is within the normal range. Correcting is referred to as correction.

Further, the correction complement unit 13 refers to the monitoring item data table in the storage unit 20 when the additional information is abnormal, and the correction complement unit 13 measures the corresponding sensor of the sensor that measures the measured value of the corresponding monitoring item data. Based on the range, the measured value that is regarded as abnormal is corrected. At this time, the correction complement unit 13 replaces the measured value with this upper limit value when the measured value exceeds the upper limit value of the corresponding sensor measurement range, while the measured value is less than the lower limit value of the corresponding sensor measurement range. , Replace the measured value with this lower limit. Then, the correction complement unit 13 overwrites the corrected measurement value with respect to the original measurement value and stores it in the monitoring item data table in the storage unit 20. In addition, the correction complement unit 13 counts the period during which the correction continues from the time when the correction of the measured value is first performed, and writes the count value in the place of the abnormal period in the monitoring item data table in the storage unit 20. Remember.

In the data range determination unit 14, each measured value of the monitoring item data (including the measured value after complementation and correction) is the maximum value and the minimum value of the measured value in the teacher data used when training the machine learning model. It is determined whether or not it is included in the learning data range (the minimum value (lower limit value of the control value) of the measured value of the teacher data and the maximum value (upper limit value of the control value) or less) having the range of. At this time, the data range determination unit 14 determines whether the measured value of each monitoring item data is included in the learning data range or within the allowable range in the determination result. Further, when the measured value exceeds the upper limit value of the limit value range or is less than the lower limit value of the limit value range, the data range determination unit 14 is a shield excavator (shield excavator) with respect to the abnormal stop determination unit 15. 30 and the same applies thereafter) to transmit an abnormal stop signal for abnormal stop. The above-mentioned training data range is actually measured with the lower limit value obtained by subtracting the first set margin value from the minimum value of the measured value and the upper limit value obtained by adding the second set margin value to the maximum value of the measured value. The range may be configured with a margin for the range of the measured values of the upper limit and the lower limit. The first set margin value and the second set margin value are positive values, and either the same numerical value or different numerical values may be used. Further, each of the first set margin value and the second set margin value may be generated by multiplying the upper limit value and the lower limit value of the measured value by a predetermined coefficient, respectively. However, if each of the upper limit value and the lower limit value is a negative value, it is converted to an absolute value and the training data range is constructed as a positive value by the above processing, or if it is the lower limit value, it is added to the upper limit value. The process of constructing the training data range is performed by the process of subtraction in the case.
Further, when the measured value of the monitoring item data that is neither missing nor abnormal is supplied, the data determination unit 12 resets the missing period and the counting value of the abnormal period of the monitoring item data table in the storage unit 20 to "0". do.

The abnormal stop determination unit 15 refers to the monitoring item data table in the storage unit 20 for each cycle in which the set value of the control data is estimated, and each of the missing period and the abnormal period of the monitoring item data is set in advance. It is determined whether or not the set limit period has been exceeded. At this time, when either the missing period or the abnormal period of each of the monitoring item data exceeds the respective limit period, the abnormal stop determination unit 15 abnormally stops the shield excavator. Here, when the defect period exceeds the limit period, it is conceivable that the sensor provided in the shield excavator or provided around the shield excavator is out of order. If the shield excavator continues to operate without knowing the operating condition of the shield excavator and the excavation condition corresponding to the excavation including the surrounding environmental condition during excavation, the safety and quality of the construction may be deteriorated. Therefore, the shield excavator control system 1 in the present embodiment abnormally stops the shield excavator when the loss of the measured value continues.

In addition, if the abnormal period exceeds the limit period, if the above sensor is out of order, the operating state of the shield excavator is abnormal (the state where the low propulsion speed continues), or the surroundings where the excavator is being excavated. It is conceivable that the environmental condition of the vehicle is abnormal (such as the condition where high controlled earth pressure continues).
If the sensor is out of order, or if the shield excavator continues to operate while the operating condition of the shield excavator and the surrounding environmental conditions during excavation are abnormal, the safety and quality of the construction may deteriorate. Therefore, the shield excavator control system 1 in the present embodiment abnormally stops the shield excavator when the abnormality of the measured value continues. Further, the abnormal stop determination unit 15 also abnormally stops the shield excavator even when the abnormal stop signal is supplied from the data range determination unit 14.

The control unit 16 estimates the set value of the control data for the operation of the shield excavator based on the determination result of whether or not the measured value of the monitoring item data from the data range determination unit 14 is included in the learning data range. It controls which of the machine learning model unit 17 and the rule-based data generation unit 18 is to perform the operation. At this time, when the above determination result includes the measured values of all the monitoring item data within the learning data range, the control unit 16 estimates the set value of the control data for the operation of the shield excavator by the machine learning model unit. Let 17 do it. On the other hand, when the measurement value of any of the monitoring item data is outside the learning data range and is included in the permissible range, the control unit 16 rules to estimate the set value of the control data for the operation of the shield excavator. Let the base data generation unit 18 do this.

The machine learning model unit 17 generates a machine learning model from the teacher data for estimating the set value of the control data for the operation of the shield excavator according to the input measured value by using the machine learning algorithm. , Estimate the set value of the control data of the operation of the shield excavator corresponding to the measured value using the generated machine learning model. For example, in the machine learning model in the present embodiment, the set value of the monitoring item data (feature amount vector) is input to the machine learning algorithm as input data, and the control data (hydraulic control data of the propulsion jack, screw rotation of the screw conveyor 60) is input. Estimates and outputs the set value of (speed, etc.).

Therefore, the machine learning model unit 17 first inputs the measured value of the monitoring item data, which is a feature quantity vector for which the correct control data is known, into the machine learning algorithm as teacher data, and estimates the set value of the control data. Generate a machine learning model. Then, the machine learning model unit 17 inputs the measured value of the monitoring item data supplied from the sensor or the like into the trained machine learning model, and sets the control data presumed to match the input measured value. Estimate the value.

Further, when the machine learning model unit 17 obtains correct / incorrect feedback on the set value of the control data estimated using the generated machine learning model, the machine learning model unit 17 includes the feedback in the teacher data to reconstruct the machine learning model. I do. Then, after the reconstruction, the machine learning model unit 17 estimates the set value of the control data that matches the measured value of the input monitoring item data by using the reconstructed machine learning model.
Here, machine learning creates a model in which the measured value of the monitoring item data is used as an explanatory variable and the set value of the control data is used as an explained variable. That is, by creating a model between the input explanatory variables and the output explained variables by machine learning and using this model, the explained variables that are highly relevant corresponding to the combination of the explanatory variables. To be obtained. By using this model, it is possible to clearly obtain which of the control data, which is the explained variable, needs to be controlled in response to the change in the monitoring item data, which is the explanatory variable. Here, as the machine learning technique for creating a model, any of commonly used techniques such as decision tree learning, neural network, genetic programming, support vector machine, and deep learning may be used.

The rule-based data generation unit 18 has, for example, the function of an expert system, stores a set of production rules consisting of a condition unit and a conclusion unit, and stores the measured values of the supplied monitoring item data and the production of the condition unit. Determine the conformity with the rules.
Then, the rule-based data generation unit 18 uses the content described in the conclusion unit corresponding to the condition unit having the highest conformity as the conclusion of the inference, that is, the set value of the control data shown in the conclusion unit as an estimated value. Is output as.

In the output check unit 19, each of the set values of the control data as the estimated values supplied from each of the machine learning model unit 17 and the rule-based data generation unit 18 is a set value range preset for each control data. It is determined whether or not it is included in.
At this time, when the set value of the control data exceeds the upper limit value of the above setting range, the output check unit 19 corrects this set value to the upper limit value and outputs it to the shield excavator. Further, when the set value of the control data is less than the lower limit value of the set range, the output check unit 19 corrects this set value to the lower limit value and outputs the set value to the shield excavator.

FIG. 5 is a diagram showing an example of a flowchart of a control operation of a shield excavator by the shield excavator control system of the present embodiment.
Step S1: The monitoring item data input unit 11 inputs the measured value of each monitoring item data from each of the sensors for measuring the measured value of the monitoring item data provided in the shield excavator, and stores the measured value. It is written and stored in the monitoring item data table of 20. Then, the monitoring item data input unit 11 stores the measured value of the monitoring item data in which the moving average is set as the measured value in advance, that is, the past measured value in a predetermined range for calculating the moving average. Read from the monitoring item data table of, and calculate the moving average of the measured values. Then, the monitoring item data input unit 11 writes the calculated result of the obtained moving average as the measurement value of the monitoring item data and stores the measurement value in the monitoring item data table of the storage unit 20.

Step S2: The data determination unit 12 refers to the monitoring item data table of the storage unit 20, extracts the latest measured value, and determines whether or not the measured value is abnormal or missing. At this time, if the measured value is abnormal or missing, the data determination unit 12 advances the process to step S3, while if the measured value is not abnormal or missing, the process proceeds to step S4.

Step S3: The correction complement unit 13 confirms the additional information added to the measured value of the monitoring item data supplied from the data determination unit 12. Then, when the measured value is indicated as abnormal in the additional information, the correction complementing unit 13 corrects the measured value by using the upper limit value or the lower limit value of the corresponding sensor measurement range corresponding to the numerical value of the measured value. conduct. The correction complement unit 13 sets the corrected measurement value as a new measurement value in the monitoring item data table of the storage unit 20, and overwrites the measurement value before the correction. At this time, the correction complement unit 13 performs a process of advancing the counting of the abnormal period of the corrected monitoring item data in the monitoring item data table of the storage unit 20.

Further, when the measured value is indicated to be missing in the additional information, the correction complementing unit 13 refers to the monitoring item data table of the storage unit 20 and refers to the average value (or intermediate value) of the past measured values in a predetermined range. Etc. are obtained to supplement the missing measured values. In the monitoring item data table of the storage unit 20, the correction complement unit 13 sets the complemented measurement value as a new measurement value, writes it to the location of the missing measurement value before complementation, and stores it. At this time, the correction complement unit 13 performs a process of advancing the counting of the missing period of the complemented monitoring item data in the monitoring item data table of the storage unit 20.

Step S4: The data range determination unit 14 refers to the measured value of each monitoring item data in the monitoring item data table of the storage unit 20, and determines whether or not the limit value range is exceeded. At this time, when the measured value exceeds the limit value range, the data range determination unit 14 transmits an abnormal stop signal to the abnormal stop determination unit 15 and proceeds to the process in step S8, while the measured value is in the limit value range. If it is included in, the process proceeds to step S5.

Step S5: The abnormal stop determination unit 15 refers to the monitoring item data table of the storage unit 20 and determines whether or not there is monitoring item data having a missing period or an abnormal period exceeding the limit period. Then, the abnormal stop determination unit 15 advances the process to step S8 when there is a monitoring item data having a missing period or an abnormal period exceeding the limit period, while the process proceeds to step S6 when there is no monitoring item data. Proceed.

Step S6: The data range determination unit 14 determines whether or not each of all the measured values of the monitoring item data is included in the respective learning data range. At this time, the data range determination unit 14 proceeds to the process in step S7 when all the measured values are included in the learning data range, and on the other hand, when any of the measured values is not included in the learning data range, the data range determination unit 14 proceeds to the process. The process proceeds to step S9.

Step S7: When it is determined that all the measured values are included in the learning data range, the control unit 16 causes the machine learning model unit 17 to estimate the set value of the control data. As a result, the machine learning model unit 17 refers to the monitoring item data table of the storage unit 20, reads out the latest measured value, inputs it to the machine learning model, and outputs the estimated value as the set value of the control data. It is output to the check unit 19 and the process proceeds to step S1.

Step S8: The abnormal stop determination unit 15 forcibly excavates the shield when an abnormal stop signal is supplied from the data range determination unit 14, or when there is a monitoring item data having a defect period or an abnormal period exceeding the limit period. Stop the machine abnormally.

Step S9: In the control unit 16, each of the measured values in all the monitored item data is not included in the training data range corresponding to each monitored item data, and the measured value not included in this learning data range is the limit. When it is not included in the value range, that is, when it is determined that the measured value not included in the training data range of the monitoring item data is within the permissible range (beyond the training data range and within the range below the limit value). , The rule-based data generation unit 18 is made to estimate the set value of the control data. As a result, the rule-based data generation unit 18 refers to the monitoring item data table of the storage unit 20, reads out the latest measured value, inputs it to the condition unit, and sets the estimated value output from the conclusion unit as control data. It is output as a value to the output check unit 19, and the process proceeds to step S1.

According to the present embodiment, according to the above-described configuration, the measured values of all the monitoring item data supplied from the sensors provided in the shield excavator are included in the training data range, or the training data range and the limit value range are included. Depending on whether it is within the permissible range between and, or if it exceeds the limit value range, whether to estimate the setting value of the control data by the machine learning model or rule base, or to stop the shield excavator abnormally. Therefore, if there is an abnormal value or a missing value in the measured value, the shield excavator can be controlled without deteriorating the safety and quality of the construction.

Further, according to the present embodiment, when the measured value of the monitoring item data does not exceed the limit value range, even if the measured value is missing or an abnormal measured value exists, each of the missing period and the abnormal period is provided. Therefore, even if the measured value such as noise is supplied from the sensor, the measured value will be observed in a predetermined period, so that the shield excavator will not be transiently stopped abnormally (emergency stop), and the shield excavator will be excavated. The operation of the machine can be continued and the progress of the construction will not be affected.

Further, according to the present embodiment, when the measured value of the monitoring item data is prone to noise, a moving average for a predetermined period is calculated and this moving average is used as the measured value, so that the measured value is within the limit value range. Even if it exceeds momentarily, that is, even if the superimposed measured value such as noise is supplied from the sensor, the abnormality of the measured value will be observed not only at that time but also in a predetermined period, so due to noise of the sensor etc. The shield excavator can be continued to operate without abnormally stopping the shield excavator in a transient manner, and the progress of the construction is not affected.

A program for realizing the function of the shield excavator control system of FIG. 1 in the present invention is recorded on a computer-readable recording medium, and the program recorded on the recording medium is read into the computer system and executed. Thereby, the shield excavator may be controlled. The term "computer system" as used herein includes hardware such as an OS (Operating System) and peripheral devices. Further, the "computer system" shall also include a WWW (World Wide Web) system provided with a homepage providing environment (or display environment). The "computer-readable recording medium" is a portable medium such as a flexible disk, a magneto-optical disk, a ROM (Read Only Memory), a CD-ROM (Compact Disc --Read Only Memory), or a built-in computer system. A storage device such as a hard disk. Furthermore, a "computer-readable recording medium" is a volatile memory (RAM (Random Access)) inside a computer system that serves as a server or client when a program is transmitted via a network such as the Internet or a communication line such as a telephone line. It also includes those that hold the program for a certain period of time, such as Memory)).

Further, the program may be transmitted from a computer system in which this program is stored in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the "transmission medium" for transmitting a program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line. Further, the above program may be for realizing a part of the above-mentioned functions. Further, a so-called difference file (difference program) may be used, which can realize the above-mentioned function in combination with a program already recorded in the computer system.

1 ... Shield excavator control system 11 ... Monitoring item data input unit 12 ... Data judgment unit 13 ... Correction complement unit 14 ... Data range judgment unit 15 ... Abnormal stop judgment unit 16 ... Control unit 17 ... Machine learning model unit 18 ... Rule base Data generation unit 19 ... Output check unit 20 ... Storage unit

Claims (5)

Shield excavator control that controls the shield excavator by using the measured value of the monitoring item data that monitors the operation status of the shield excavator as input data and the control data used for controlling the shield excavator output from the machine learning model. It ’s a system,
A machine learning model unit that outputs the control data based on the measured value of the monitoring item data that monitors the excavation state of the shield excavator, which is input from the detector provided in the shield excavator.
A rule-based data generation unit that derives the control data based on the measured value and a predetermined rule,
Based on the measured value, the control unit causes the machine learning model unit to output the control data, the rule-based data generation unit to output the control data, or the shield excavator to be stopped. ,
A data determination unit that determines whether the measured value of the monitoring item data is missing or an abnormal value, and
A data range determination unit that determines whether or not the measured value of the monitoring item data is within the data range of the teacher data used when learning the machine learning model.
Equipped with
When the measurement value of the monitoring item data is missing or abnormal, the control unit stops the control of the shield excavator, and when the measurement value of the monitoring item data is out of the data range, the control unit stops control of the shield excavator. Have the rule-based processing unit generate control data, and have the machine learning model unit generate the control data when the measured value of the monitoring item data is within the data range.
It features a shield excavator control system. If the measured value of the monitoring item data is missing, it is complemented, and if it is an abnormal value, it is further provided with a complementing correction unit.
The shield excavator control system according to claim 1 , wherein the data range determination unit determines whether or not the measured value of the monitored item data supplemented and corrected is within the data range. Further, an abnormality determination unit for determining whether or not the measured value of the monitoring item data is missing or the abnormal value continues for a predetermined period is provided.
The control unit
If the abnormal value of the monitoring item data is outside the data range and within a predetermined allowable value, the monitoring item data is supplied to the rule-based data generation unit.
The shield according to claim 1 or 2 , wherein the shield excavator is stopped when the defect continues for a predetermined period or is an abnormal value exceeding the allowable value continuously for a predetermined period. Excavator control system. From claim 1 , the monitoring item data input unit moves and averages the measured values of a predetermined type of monitoring item data in the monitoring item data within a predetermined range, and uses the result of the moving average as the measured value. The shield excavator control system according to any one of claims 3 . Shield excavator control that controls the shield excavator by using the measured value of the monitoring item data that monitors the operation status of the shield excavator as input data and the control data used for controlling the shield excavator output from the machine learning model. It ’s a method,
A machine learning model process in which the machine learning model unit outputs the control data based on the measured value of the monitoring item data for monitoring the excavation state of the shield excavator, which is input from the sensor provided in the shield excavator. ,
A rule-based data generation process in which the rule-based data generation unit derives the control data based on the measured value and a predetermined rule, and
Based on the measured value, the control unit performs a process of outputting the control data to the machine learning model unit, outputting the control data to the rule-based data generation unit, or stopping the shield excavator. The control process to make
A data determination process for determining whether the measured value of the monitoring item data is missing or an abnormal value, and
A data range determination process for determining whether or not the measured value of the monitoring item data is within the data range of the teacher data used when learning the machine learning model.
Including
In the control process, when the measured value of the monitoring item data is missing or abnormal, the control of the shield excavator is stopped, and when the measured value of the monitoring item data is out of the data range, the control process is described. This includes having the rule-based processing unit generate control data and having the machine learning model unit generate the control data when the measured value of the monitoring item data is within the data range.
Shield excavator control method.

Images