Slime detection device and slime detection method

Abstract

This invention provides a slime detection device and slime detection method for efficiently determining the slime sedimentation status at the bottom of a borehole. [Solution] The slime detection device A1 includes a measurement information storage unit 22 that records a detection waveform of vibration when a long object is driven into the bottom of a hole, from a sensor device 10 attached to a long object that can reach the bottom of a pile hole of a cast-in-place concrete pile, and a control unit 21 that analyzes the detection waveform. The control unit 21 performs transient spectral analysis on the detection waveform recorded in the measurement information storage unit 22, and generates a judgment model by performing machine learning on the analysis results of the transient spectral analysis according to the presence or absence of slime. The control unit 21 uses the judgment model to output a judgment result on the presence or absence of slime for the analysis results of the transient spectral analysis of the detection waveform in a new pile hole.

Description

Technical Field

[0001] The present disclosure relates to, for example, a slime determination device and a slime determination method for determining the slime sedimentation situation at the bottom of a pile hole in a cast-in-place concrete pile.

Background Art

[0002] When constructing a cast-in-place concrete pile by the earth drilling method using a stabilizing fluid, the excavated soil and fine sand or fine-grained soil floating in the stabilizing fluid settle as slime at the bottom of the pile hole. If slime remains at the bottom of the hole during concrete placement, the slime may be entrained, resulting in a deterioration in the quality of the pile body concrete, or the slime may intervene between the pile and the tip ground, impairing the pile tip support performance. Therefore, in normal construction, bottom hole treatment (slime treatment) is performed to remove the slime before concrete placement.

[0003] The method for determining the slime sedimentation situation is to lower a measuring tape with a weight to the bottom of the hole, measure the depth of the upper surface of the slime from the feeling transmitted to the hand when the weight touches the upper surface of the slime, and determine the slime sedimentation situation before and after the bottom hole treatment.

[0004] In addition, technologies for checking the slime sedimentation situation at the bottom of the hole have also been studied (see, for example, Patent Document 1). In the technology disclosed in this document, the sedimentation situation of slime at the bottom of the pile hole of a cast-in-place concrete pile is checked. A detection waveform when the measuring tape is struck against the bottom of the hole is acquired from a sensor device attached to the measuring tape that can reach the bottom of the hole. Then, the sedimentation situation of the slime is determined by analyzing the detection waveform.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] However, because the detected waveform contains various types of noise, the technology described in Patent Document 1 may not be able to determine the slime sedimentation status. [Means for solving the problem]

[0007] A slime detection device for solving the above problems comprises a measurement information storage unit that records a detection waveform of vibrations when a long object, attached to a sensor device that can reach the bottom of a pile hole in a cast-in-place concrete pile, is driven into the bottom of the hole, and a control unit that analyzes the detection waveform. The control unit then performs transient spectral analysis on the detection waveform recorded in the measurement information storage unit, generates a judgment model by performing machine learning on the analysis results of the transient spectral analysis according to the presence or absence of slime, and outputs a judgment result regarding the presence or absence of slime using the judgment model for the analysis results of the transient spectral analysis of the detection waveform in a new pile hole. [Effects of the Invention]

[0008] According to this disclosure, the slime sedimentation status at the bottom of the borehole can be efficiently determined. [Brief explanation of the drawing]

[0009] [Figure 1] This is an explanatory diagram of the system according to the embodiment. [Figure 2] This is an explanatory diagram of the hardware configuration of the embodiment. [Figure 3] This is an explanatory diagram of the sensor device according to the embodiment. [Figure 4] This is an explanatory diagram of the processing procedure for the slime detection process in the embodiment. [Figure 5] This is an explanatory diagram of the processing procedure for the analysis process of the embodiment. [Figure 6] This diagram illustrates the frequency characteristics of the measurement results of the embodiment, where (a) is an explanatory diagram of the peak range and (b) is an explanatory diagram of the time window. [Figure 7]This diagram illustrates the frequency characteristics of the measurement results of the embodiment, where (a) is the peak range, (b) is the frequency analysis result, (c) is the spectral slice, and (d) is the spectrogram. [Figure 8] This is an explanatory diagram of the confirmation screen of the embodiment. [Figure 9] This diagram illustrates the frequency characteristics of the measurement results of the embodiment, where (a) shows the case with slime and (b) shows the case without slime. [Modes for carrying out the invention]

[0010] Below, an embodiment of the slime detection device and slime detection method will be described using Figures 1 to 9. As shown in Figure 1, in this embodiment, the sedimentation status (presence or absence) of slime SL1 at the bottom of the pile hole H1 of a cast-in-place concrete pile is determined. The pile hole H1 is filled with a stabilizing fluid containing bentonite to stabilize the hole wall after excavation.

[0011] To confirm the depth of the pile hole H1, a measuring tape M1 with markings is used. The measuring tape M1 is a long, flat tape measure (long object) of a length that can reach the bottom of the hole H2 (for example, 10m to 100m). A weight W1 is suspended from the tip of this measuring tape M1. Note that the long object is not limited to the measuring tape M1, as long as it can reach the bottom of the hole H2 and vibrates upon contact. For example, a cable or a rod-shaped object may also be used.

[0012] In this embodiment, a slime detection device A1 is further used, which includes a sensor device 10, an amplifier AP1, and a control device 20. The sensor device 10 is attached to the end of the measuring tape M1. Details of the sensor device 10 will be described later. The sensor device 10 is then connected to the management device 20 via a communication line and amplifier AP1.

[0013] (Description of hardware configuration) Using FIG. 2, the hardware configuration of the information processing device H10 that constitutes the management device 20 will be described. The information processing device H10 includes a communication device H11, an input device H12, a display device H13, a storage device H14, and a processor H15. Note that this hardware configuration is an example, and it is also possible to be realized by other hardware.

[0014] The communication device H11 is an interface that establishes a communication path with other devices and executes data transmission and reception. The input device H12 is a device that accepts input of various information, such as a mouse or a keyboard. The display device H13 is a display or the like that displays various information.

[0015] The storage device H14 is a storage device that stores data and various programs for executing various functions of the management device 20. The processor H15 controls each process in the management device 20 using the programs and data stored in the storage device H14. Examples of the processor H15 include a CPU, an MPU, etc. This processor H15 expands the program stored in a ROM or the like to the RAM and executes various processes for each process.

[0016] The processor H15 is not limited to performing software processing for all processes it executes. For example, the processor H15 may include a dedicated hardware circuit (e.g., an application-specific integrated circuit: ASIC) that performs hardware processing for at least a part of the processes it executes.

[0017] (System Configuration) Next, each function of the slime determination device A1 will be described. As shown in Figure 3, the sensor device 10 has accelerometers 12, 13, and 14 fixed to a plate material 11 such as an acrylic plate. The plate material 11 is detachably attached to the surface of the measuring tape M1 using a fixing device ST1 such as a turn clip. In this embodiment, the sensor device 10 measures the acceleration of vibration in three axial directions in three dimensions. Specifically, if the direction perpendicular to the surface of the measuring tape M1 is the X-axis direction, the width direction of the measuring tape M1 is the Y-axis direction, and the length direction of the measuring tape M1 is the Z-axis direction, the sensor device 10 is positioned in the YZ plane.

[0018] Accelerometer 12 measures the acceleration waveform (vibration waveform) in the Z-axis direction. Accelerometer 13 measures the acceleration waveform (vibration waveform) in the Y-axis direction. Accelerometer 14 measures the acceleration waveform (vibration waveform) in the X-axis direction. The measurement signals (time-series data of acceleration) measured by each accelerometer 12 to 14 are transmitted via cable C1. This cable C1 is bundled together with a fixing device ST2 such as a turn clip and fixed to the measuring tape M1.

[0019] As shown in Figure 1, cable C1 is connected to amplifier AP1. Amplifier AP1 amplifies the measurement signals from accelerometers 12-14 and inputs them to the control device 20. The management device 20 is a computer system that performs processing to assist in determining the slime sedimentation status. This management device 20 comprises a control unit 21, a measurement information storage unit 22, and a learning result storage unit 23.

[0020] The control unit 21 performs a process to record the slime sedimentation status based on the measurement signal from the sensor device 10. The control unit 21 performs the processes described later (including the analysis stage, learning stage, judgment stage, etc.). By executing the slime judgment program for this purpose, the control unit 21 functions as an analysis unit 211, a learning unit 212, a judgment unit 213, etc.

[0021] The analysis unit 211 acquires measurement signals from accelerometers 12-14 via amplifier AP1 and analyzes the amplified measurement signals. Here, the frequency characteristics of the acceleration of each axis are calculated. The learning unit 212 performs machine learning on frequency characteristics. In this embodiment, deep learning using a CNN (Convolutional Neural Network) is performed. The determination unit 213 determines the presence or absence of slime sedimentation.

[0022] The measurement information storage unit 22 stores measurement management records regarding the slime settlement status of each pile hole. These measurement management records are registered when analysis processing is performed. The measurement management records include information on the site ID, pile number, measurement date and time, measurement location, excavation depth, measurement signals for each axis, vibration frequency characteristics for each axis, analysis results, and judgment results.

[0023] The site ID data area records data related to identifiers used to identify the site where the pile holes were formed. The pile number data area records data related to identifiers (pile numbers) used to identify each pile hole at this site.

[0024] The measurement date and time data area records data related to the year, month, and time the judgment process was performed. The measurement location data area records data related to the location (coordinates) where the measurement was performed on the plan view of the pile hole.

[0025] The drilling depth data area records data related to the depth of the pile hole. The measurement signal data area for each axis records data related to the measurement signals for each axis acquired from accelerometers 12-14. Here, the measurement signal is recorded as acceleration (gal) for each axis against time (sec).

[0026] The frequency characteristic data area for each axis records data related to the results of frequency analysis of the measured signal for each axis. Here, as a frequency characteristic, acceleration (gal) is recorded for each axis in relation to frequency (Hz).

[0027] The analysis results data area records data on the statistical values ​​(mean and standard deviation) of the dominant frequency of acceleration for each axis. The judgment result data area records the result of whether or not slime is present.

[0028] The learning result storage unit 23 stores a judgment model that predicts the slime sedimentation status of each pile hole. This judgment model is registered when a learning process is performed. This judgment model has an input layer of spectra generated by transient spectral analysis, which will be described later, an output layer of the likelihood of slime sedimentation, and a hidden layer connecting the input layer and the output layer.

[0029] (Slime detection process) The slime detection process will be explained using Figure 4. In this case, the measurement information storage unit 22 stores a measurement management record that includes the site ID of the site where the slime sedimentation status is determined, the pile number of the pile hole, and the measurement date and time when the work begins.

[0030] First, a measurement is taken (step S11). Specifically, the worker drops a measuring tape M1 with a weight W1 attached to its tip into the pile hole H1. When the weight W1 reaches the bottom of the hole H2, the worker measures the excavation depth using the scale on the measuring tape M1. Then, using the input device H12 of the management device 20, the excavation depth is recorded in the measurement management record of the measurement information storage unit 22.

[0031] Next, the weight-striking operation is performed (step S12). Specifically, the operator attaches the sensor device 10 to the handle of the measuring tape M1 (directly below the gripping position). Then, the measuring tape M1 is lowered from in front of the hole bottom H2 so as to strike the weight W1 against the hole bottom H2, and then pulled up after striking (striking operation). This causes vibration in the measuring tape.

[0032] In this case, the control device 20 performs the measurement process (step S13). Specifically, when the sensor device 10 measures vibration by acceleration, it outputs a measurement signal for the acceleration during vibration. Here, the accelerometers 12-14 output the measurement signal to the amplifier AP1 via cable C1. The amplifier AP1 amplifies the measurement signals from the accelerometers 12-14. Then, the analysis unit 211 of the control unit 21 of the control device 20 acquires the amplified measurement signals of acceleration for the X, Y, and Z axes.

[0033] Next, the control device 20 performs analysis processing (step S14). Specifically, the analysis unit 211 of the control unit 21 of the control device 20 performs frequency analysis of the acquired amplified measurement signals for the X, Y, and Z axes. Details will be described later.

[0034] Next, the management device 20 performs recording processing (step S15). Specifically, the determination unit 213 of the control unit 21 of the management device 20 records the measurement signal for each axis and the frequency characteristics of each axis in the measurement management record of the measurement information storage unit 22.

[0035] Next, the control device 20 performs a determination process (step S16). Here, the determination unit 213 outputs the frequency characteristics to the display device H13. In this case, the person in charge checks the frequency characteristics output to the display device H13.

[0036] In the determination process (step S16), if it is determined that the slime has settled, slime treatment is performed (step S17). Specifically, for example, a bottom-cleaning bucket method is used, in which a bucket is used to scoop up the settled slime. Alternatively, an air-lift method, a submersible pump method, or a suction pump method may be used. In the air-lift method, compressed air is blown into the tremie tube to create an upward current in the tube, sucking up the slime along with the muddy water. In the submersible pump method, a submersible pump is used to suck up the slime. In the suction pump method, the tremie tube is connected to a suction pump to suck up and discharge the slime. Then, the determination process is performed again to determine the slime settling status.

[0037] Then, in the determination process (step S16), if it is determined that no slime has settled, the slime determination process is terminated. After that, a concrete pile is constructed in this pile hole H1.

[0038] (Analysis processing) In this embodiment, an analysis is performed to determine the presence or absence of slime. Here, the measuring tape M1 uses acceleration in the direction perpendicular to the surface (X-axis direction), but acceleration in other axial directions may also be used.

[0039] Next, we will explain the analysis process using Figure 5. First, the control unit 21 of the management device 20 performs a process to identify the peak position (step S21). Specifically, the analysis unit 211 of the control unit 21 obtains past measurement management records from the measurement information storage unit 22 that determine the presence or absence of slime. Then, the analysis unit 211 identifies the peak position in the detection waveform of the measurement signal recorded in the measurement management record. Here, the time when the acceleration exceeds a predetermined value is identified as the peak position. This eliminates the influence of the preparation stage before dropping the weight W1 and hand tremors after dropping it, and identifies the time when it hits the bottom of the hole H2.

[0040] As shown in Figure 6(a), the detected waveform 500 is acquired. In this detected waveform 500, the peak is displayed as a negative value. Therefore, in the detected waveform 500, the peak P1 is identified as the minimum value less than or equal to a predetermined acceleration AC1.

[0041] Next, the control unit 21 of the management device 20 performs a process to identify the peak range (step S22). Specifically, the analysis unit 211 of the control unit 21 identifies a predetermined time period as the peak range in the detected waveform, with the peak P1 position (time) in between. Here, the peak range is determined to be the time t1 before the peak P1 position and the time t2 after the peak P1 position. As shown in Figure 6(a), the peak range is determined by specifying the times t1 and t2 for the peak P1 in the detected waveform 500 of the measurement signal.

[0042] Then, the control unit 21 of the management device 20 performs transient spectral analysis processing (step S23). Transient spectral analysis is a method for analyzing the frequency characteristics of a signal that change over time. Here, the short-time Fourier transform is used. In the short-time Fourier transform, the signal is cut into short time windows, and the frequency components that change over time are analyzed by applying the Fourier transform to each time window. Specifically, the analysis unit 211 of the control unit 21 sequentially moves the time windows by a predetermined time difference from the start time of the peak range.

[0043] As shown in Figure 6(b), for example, a time window TW1 of length Lana is sequentially moved by a time difference ΔL. If the peak range [t1+t2] is 0.6 seconds, the length Lana is 0.2 seconds, and the time difference ΔL is 0.02 seconds, 21 time windows TW1 are generated within the peak range. Then, the Fourier transform of the signals of the time windows TW1 is performed.

[0044] Figure 7(a) shows the tenth time window TW1. Figure 7(b) shows the waveform 510 obtained by Fourier transforming the signal within the time window TW1. Figure 7(c) shows a spectral slice 520 created for waveform 510, with frequency on the vertical axis. The width of this spectral slice 520 is the time difference ΔL, and it is color-coded according to amplitude.

[0045] Then, the spectral slice 520 is placed in the order of the time window TW1. For example, it is placed at the end of the time window TW1. Note that the spectral slices only need to be arranged in the order of the time window TW1. Figure 7(d) shows the spectrogram 530, in which each spectral slice 520 is arranged according to the time window TW1.

[0046] Next, the control unit 21 of the management device 20 performs the teacher information registration process (step S24). Specifically, the analysis unit 211 of the control unit 21 records the generated spectrogram in the measurement information storage unit 22.

[0047] Next, the control unit 21 of the management device 20 performs machine learning processing (step S25). Specifically, when a predetermined number of measurement management records are recorded in the measurement information storage unit 22, the learning unit 212 of the control unit 21 obtains a dataset from the measurement information storage unit 22 that combines the spectrogram and the judgment result. The learning unit 212 then uses deep learning to generate a judgment model that predicts the presence or absence of slime, with the spectrograms in the X, Y, and Z directions as input layers and the judgment result as the output layer. Here, a judgment model that predicts the judgment result from the spectrogram is generated and recorded in the learning result storage unit 23.

[0048] Next, we will explain the case of carrying out new excavations. Here, the control unit 21 of the management device 20 executes the process of acquiring a new detected waveform (step S31). Specifically, the analysis unit 211 of the control unit 21 acquires a new detected waveform from the measurement information storage unit 22 for the new excavation. At this stage, the presence or absence of slime has not been determined.

[0049] Next, the control unit 21 of the management device 20 performs the following processes in the same manner as in steps S21 to S23: peak position identification process (step S32), peak range identification process (step S33), and transient spectral analysis process (step S34).

[0050] Next, the control unit 21 of the management device 20 performs a determination process (step S35). Specifically, the determination unit 213 of the control unit 21 inputs newly generated spectrograms in the X, Y, and Z directions into the determination model recorded in the learning result storage unit 23, and obtains the determination result of whether or not slime is present. The control unit 21 then outputs this result to the display device H13.

[0051] As shown in Figure 8, the display device H13 outputs a confirmation screen 600. The confirmation screen 600 includes a stake number field 601, a measurement date and time field 602, a measurement position field 603, and a judgment result field 604. Furthermore, the confirmation screen 600 includes analysis result fields 60X, 60Y, and 60Z for the X, Y, and Z axes, respectively. The analysis result fields 60X, 60Y, and 60Z each contain graphs and spectrograms showing amplitude, respectively. The judgment result field 604 outputs the judgment result determined by the judgment model.

[0052] (Operation of this embodiment) If slime has accumulated, before slime removal, the accumulated slime acts as a cushion. If the slime is removed, the hard bottom of the hole is exposed, which increases the vibration frequency during hammering.

[0053] (Effects of this embodiment) (1) In this embodiment, the sensor device 10 is attached to the handle of the measuring tape M1 (directly below the gripping position). Since the sensor device 10 is attached to an existing measuring tape M1, the workload for the worker can be reduced.

[0054] (2) In this embodiment, the control device 20 performs a measurement process (step S13). This allows the vibration caused by the impact of the weight W1 at the tip of the measuring tape M1 to be detected by acceleration.

[0055] (3) In this embodiment, the control device 20 performs an analysis process (step S14). The impact of the weight W1 changes depending on whether or not slime settles at the bottom of the hole H2. The presence or absence of slime settlement can be determined according to this impact condition.

[0056] (4) In this embodiment, the acceleration of the measuring tape M1 in the horizontal direction (perpendicular to the surface, width direction) is detected. When the weight W1 is struck against the bottom of the hole H2, slack occurs in the measuring tape M1. The impact situation can be understood from this slack. In particular, the measuring tape M1 sags significantly in the direction perpendicular to the surface (X-axis direction), and the impact situation can be efficiently understood in the X-axis direction.

[0057] (5) In this embodiment, the management device 20 performs recording processing (step S15). This allows for the recording of evidence regarding the slime sedimentation status. (6) In this embodiment, the control unit 21 of the management device 20 performs a process to identify the peak position (steps S21, S32). This makes it possible to identify the timing when the weight W1 is struck against the bottom of the hole H2.

[0058] (7) In this embodiment, the control unit 21 of the management device 20 performs a process to identify the peak range (steps S22, S33). This makes it possible to extract the detection waveform that is affected by slime in response to the impact of the weight W1.

[0059] (8) In this embodiment, the control unit 21 of the management device 20 performs transient spectral analysis processing (steps S23, S34). This allows the properties of the signal (stationary / transient) to be evaluated by evaluating the main frequency components and their temporal changes, as well as the stability / instability of the frequency components. For example, in transient conditions, the temporal changes in the statistical properties of time-series data can be grasped.

[0060] Figure 9 shows the results of the transient spectral analysis. Figure 9(a) is spectrogram 700 with slime present, and Figure 9(b) is spectrogram 701 without slime. Spectrogram 701 has larger amplitudes in all three axes compared to spectrogram 700. Therefore, the presence or absence of slime can be visually identified by the color distribution and color area of ​​the spectrogram.

[0061] (9) In this embodiment, the control unit 21 of the management device 20 performs teacher information registration processing (step S24) and machine learning processing (step S25). This makes it possible to generate a determination model for determining the presence or absence of slime.

[0062] (10) In this embodiment, the control unit 21 of the management device 20 performs a determination process (step S35). This makes it possible to evaluate the likelihood of the presence or absence of slime using the determination model.

[0063] This embodiment can be implemented with the following modifications. This embodiment and the following modifications can be combined with each other to the extent that they do not contradict each other technically. In the above embodiment, the slime detection device A1 comprises a sensor device 10, an amplifier AP1, and a management device 20. The hardware configuration is not limited to the above configuration.

[0064] In the above embodiment, the sensor device 10 is attached to the measuring tape M1. The mounting location for the sensor device 10 may be a rod-shaped material, as long as it is a flexible member that can reach the bottom of the hole H2. In the above embodiment, an accelerometer 12 measures acceleration in the Z-axis direction, an accelerometer 13 measures acceleration in the Y-axis direction, and an accelerometer 14 measures acceleration in the X-axis direction. The number of sensors is not limited to this. For example, acceleration in only one of the axes may be used.

[0065] In the above embodiment, the process of driving in the weight is performed (step S12). Here, the process of driving in the weight W1 may be performed by a machine instead of by an operator. In the above embodiment, the control unit 21 of the management device 20 performs a process to identify the peak range (steps S22, S33). Here, in the detected waveform, a predetermined time period spanning the peak position is identified as the peak range. The method for identifying the peak range is not limited to this. For example, the range around the peak position until the acceleration becomes "almost zero" (a reference value) may be identified as the peak range. Alternatively, the peak range may be identified based on a region where the change in acceleration becomes flat and is below a predetermined value. Alternatively, small peaks may be counted before and after the peak, and a peak range consisting of times t1 and t2 may be identified within a predetermined range of small peaks.

[0066] In the above embodiment, the control unit 21 of the management device 20 performs transient spectral analysis processing (steps S23, S34). In this embodiment, a short-time Fourier transform is performed as transient spectral analysis, but it is not limited to the short-time Fourier transform. For example, wavelet transform, Wignerville distribution, and Hilbert-fan transform can be used. The wavelet transform analyzes both time and frequency information simultaneously by calculating the correlation between wavelet functions of various scales and positions and the signal. The Wignerville distribution uses a function that represents the simultaneous time and frequency distribution of the signal. The Hilbert-fan transform extracts the instantaneous amplitude and frequency of the signal.

[0067] In the above embodiment, the peak range [t1+t2] was set to 0.6 seconds, the length Lana to 0.2 seconds, and the time difference ΔL to 0.02 seconds. The length Lana and time difference ΔL should be lengths that allow a predetermined number of time windows TW1 to be generated within the peak range. In addition, the time difference ΔL was set to one-tenth of the length Lana, but any predetermined ratio is acceptable.

[0068] In the above embodiment, the control unit 21 of the management device 20 performs machine learning processing (step S25). Here, deep learning using a CNN (Convolutional Neural Network) is performed. The learning method is not limited to deep learning, as long as the presence or absence of slime can be determined according to the frequency characteristics. For example, support vector machines (SVM), decision trees / random forests, k-nearest neighbors (k-NN), Gaussian process regression (GPR), principal component analysis (PCA), etc., may be used. Support vector machines are a method that separates data in a high-dimensional feature space. Decision trees / random forests are a method that divides data into a tree structure and predicts the state. Random forests are a method that improves prediction accuracy by combining multiple decision trees. k-nearest neighbors are a method that makes predictions based on the state of the k closest existing data for new data. Gaussian process regression is a method that enables prediction considering the uncertainty of the data. Principal component analysis is a method that reduces the dimensionality of the data and extracts important features.

[0069] In the above embodiment, the control unit 21 of the management device 20 performs machine learning processing (step S25). Here, a judgment model is generated that predicts the presence or absence of slime, using the spectrograms in the X, Y, and Z directions as input layers and the judgment results as output layers. Alternatively, separate judgment models may be generated for each of the X, Y, and Z directions. The presence or absence of slime may then be determined by statistical processing (e.g., weighted majority voting) of the judgment results output from each judgment model.

[0070] Next, the technical concepts that can be understood from the above embodiments and alternative examples are described below. (a) The slime determination device according to any one of claims 1 to 4, characterized in that the measurement information storage unit records the detection waveform of the acceleration of three-dimensional vibration in three axes acquired from the sensor device.

[0071] (b) The slime detection device according to any one of items 1 to 4 or (a), characterized in that deep learning is performed in the machine learning. (c) The slime detection device according to any one of claims 1 to 4, (a) or (b), characterized in that a measuring tape is used as the long object. [Explanation of Symbols]

[0072] H1...Pile hole, H2...Bottom of hole, SL1...Slime, M1...Measurement tape, W1...Weight, A1...Slime detection device, 10...Sensor device, 11...Plate material, 12,13,14...Accelerometer, ST1,ST2...Fixing device, AP1...Amplifier, 20...Management device, 21...Control unit, 211...Acquisition unit, 212...Analysis unit, 213...Determination unit, 22...Measurement information storage unit, 23...Learning result storage unit.

Claims

1. A slime determination device for determining the state of slime settlement at the bottom of a pile hole, comprising: a measurement information storage unit that records a detected waveform of vibration when a long object, attached to a sensor device that can reach the bottom of a pile hole of a cast-in-place concrete pile, is driven into the bottom of the hole; and a control unit that analyzes the detected waveform, the device for determining the state of slime settlement at the bottom of the pile hole, The control unit, In the detection waveform recorded in the measurement information storage unit, transient spectral analysis is performed. The results of the transient spectral analysis are used to generate a judgment model by performing machine learning based on the presence or absence of slime. A slime detection device characterized by outputting a determination result regarding the presence or absence of slime using the determination model described above, based on the analysis results of transient spectral analysis of the detected waveform in a new pile hole.

2. The slime detection device according to claim 1, characterized in that the control unit performs a short-time Fourier transform as the transient spectral analysis.

3. The slime detection device according to claim 2, characterized in that the control unit extracts the detected waveform in a plurality of time windows within a peak range that includes the peak position of the detected waveform in the short-time Fourier transform.

4. The control unit, For each of the aforementioned time windows, spectral slices are generated with amplitudes color-coded according to frequency. The slime detection device according to claim 3, characterized in that it generates a spectrogram in which the spectral slices are arranged in the order of the time window.

5. A slime determination method for determining the state of slime settling at the bottom of a pile hole, using a slime determination device comprising: a measurement information storage unit that records a detected waveform of vibration when a long object, attached to a sensor device that can reach the bottom of a pile hole of a cast-in-place concrete pile, is driven into the bottom of the hole; and a control unit that analyzes the detected waveform, the device being used to determine the state of slime settling at the bottom of the pile hole, The control unit, In the detection waveform recorded in the measurement information storage unit, transient spectral analysis is performed. The results of the transient spectral analysis are used to generate a judgment model by performing machine learning based on the presence or absence of slime. A slime detection method characterized by outputting a determination result regarding the presence or absence of slime using the determination model described above, based on the analysis results of transient spectral analysis of the detected waveform in a new pile hole.

Images