JP4404494B2 - Concrete compaction judgment method and apparatus - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a technique capable of stopping a vibrator at an appropriate timing when the placed concrete is compacted with a vibrator.
[0002]
[Prior art]
Conventionally, when placing a large amount of concrete, for example, the concrete is transported to a place where the concrete can be transported by a transporting means such as a bucket and is sequentially placed.
At this time, the concrete immediately after placing is not mixed with coarse aggregate and fine aggregate uniformly, so there are many voids and solid concrete cannot be obtained, so insert a vibrator into the placed concrete and vibrate. Has been done to give.
By applying vibration to the concrete, a mortar composed of fine aggregate, cement, and water enters the gap between the coarse aggregate and the coarse aggregate and becomes even, and a dense concrete can be obtained.
[0003]
However, inconvenience occurs even if the vibrator is applied for too long. That is, if the time for applying the vibrator is too long, there is a problem that breathing water is generated on the surface of the concrete.
In this way, the coarse aggregate is settled by the vibrator for a long time, and the breathing water containing ultra fine sand and cement is floated on the surface to generate latency. Since the portion of the latency generated in this way does not generate strength even when it is hardened, when adding concrete to the concrete that has been placed and hardened, the previous concrete and the rear concrete are placed. In order to improve the splicing, the latency is removed in advance.
[0004]
As described above, if the time for applying the vibrator is too long, latency will be generated, so that a lot of labor is required for the removal work, and the amount of the removed concrete is wasted, and new concrete is Only that much extra is needed.
On the other hand, when the time for applying the vibrator is insufficient, a sufficiently solid concrete cannot be obtained, so that the strength is insufficient.
Therefore, the determination of the completion of compaction is very important from the viewpoint of quality control of concrete and the improvement of workability. Conventionally, however, it has relied on the experience and intuition of skilled workers.
[0005]
[Problems to be solved by the invention]
For this reason, it is necessary to always have skilled workers at the concrete placement site, but in the recent construction industry, there is a shortage of this type of skilled worker and an inexperienced person must judge. There were not many situations. For this reason, there were many mistakes made by inexperienced judges.
Furthermore, even a skilled person cannot accurately determine the completion of compaction, and a determination error occurs to some extent, which adversely affects the quality of concrete and the workability of placing work.
[0006]
In view of the above circumstances, the present invention has been made to provide a technique capable of knowing the optimum timing for stopping the vibrator without depending on the judgment of a person.
[0007]
[Means for Solving the Problems]
The concrete compaction judgment method according to claim 1 is:
In the concrete compaction judgment method for judging completion of compaction when compacting the placed concrete with a vibrator,
The input value is the time-dependent data until the sound pressure due to the vibration of the vibrator rises over time from the start of compaction to the end of compaction, exceeds the peak, and falls to the specified sound pressure level. Constructing a neural network with the value as the match rate,
By approximating the appropriate time-dependent data of sound pressure due to vibration of the vibrator from the start of compaction to the end of compaction with an envelope, and setting the joint load coefficient as the teacher data with the maximum match rate, the neural network The process of completing the network;
In actual compaction work,
While detecting the temporal change of sound pressure due to vibration of the vibrator from the start of compaction of concrete, the temporal change of sound pressure due to the detected vibration is sequentially compared with the teacher data using the neural network, and the match rate of both is calculated. The desired process;
And a step of determining completion of compaction when the match rate is equal to or higher than a predetermined threshold value.
[0008]
The concrete compaction judgment method of claim 2 is:
By creating a time-dependent data of sound pressure due to vibration of the vibrator when concrete compaction has not been completed, and using this time-dependent data as teacher data with a minimum matching rate, a neural network can be set. It is characterized by including a process to complete .
[0009]
The concrete compaction judging device according to claim 3 is:
An output means for outputting sound pressure change data over time by vibration of a vibrator for compacting the placed concrete ;
Appropriate time-dependent data until the sound pressure due to vibration of the vibrator rises over time from the start of compaction to the end of compaction, exceeds the peak, and falls to a predetermined sound pressure level. Finding means for calculating a match rate by comparing using predetermined neural network and predetermined teacher data approximated by an envelope,
Comparing the calculated match rate with a predetermined threshold value, and when the match rate is equal to or greater than the threshold value, there is provided a determination means for determining the end of compaction.
[0010]
The concrete compaction judging device according to claim 4 is:
Furthermore, learning means is provided for setting a coupling load coefficient of the neural network of the obtaining means based on the temporal change data of the sound pressure due to the vibration of the vibrator from the start to the end of the compaction as a reference.
The concrete compaction judging device according to claim 5 is:
Furthermore, a standby unit is provided for outputting a vibrator stop command after a predetermined standby time has elapsed after the determination unit determines that compaction has been completed.
In the concrete compaction judging device according to claim 6 ,
The output means is configured to detect a change in sound pressure due to vibration of the vibrator with a microphone and output data with time change of the signal.
[0011]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, a concrete compaction determination method and apparatus according to the present invention will be described in detail with reference to the drawings showing embodiments thereof.
[0012]
In FIG. 1, 10 is a concrete compaction apparatus used in the present invention. Reference numeral 1 denotes a compacting work vehicle having a vibrator 2 at the tip of the boom 11. The vibrator 2 vibrates the vibrating body 21 using an electric or hydraulic motor as a power source. The vibrator 2 can be inserted into a desired place by operating the boom 11 to give vibration.
Reference numeral 3 denotes a microphone that detects sound caused by vibration of the vibrator 2, and is installed in a location facing the vibrator 2 side in front of the operator's driver's seat. It is attached via a buffer so that the vibration of the compacting work vehicle 1 is not transmitted directly.
4 is a control device installed at the rear of the driver's seat. As shown in FIG. 2, a sound level meter 41 that outputs a sound pressure level due to vibration of the vibrator 2 detected by the microphone 3 and a sound level meter 41 The A / D converter 42 that converts the output sound pressure level signal into a digital signal, the determination computer 43 that determines the timing to stop the vibrator from the change in sound pressure, and the determination computer 43 determines that compaction is complete. And a buzzer 5 that rings when it is turned on.
In the above-described configuration, the microphone 3 and the control device 4 constitute the concrete compaction determination device of the present invention, the microphone 3 constitutes a part of the output means, and the control device 4 determines the learning means and the seeking means. Means and standby means.
[0013]
In FIG.
A is concrete that has already been cast and has been hardened, and B is uncured concrete that has been additionally cast by a bucket or the like (not shown). C is a temporary frame for holding the concrete to be added.
The concrete B has, for example, almost no slump and is slightly raised immediately after placement as shown in the figure.
The vibrator 2 is inserted into the concrete B in such a state, and vibrations are generated to start the compacting operation. From this point, the microphone 3 and the control device 4 are activated.
The control device 4 shown in FIG. 2 also includes a learning computer 44 described later. As will be described later, the learning computer 44 and the determination computer 43 provided in the control device 4 are programmed with a learning function by a neural network and a compaction end determination function.
As shown in FIG. 3, the neural network programmed in the learning computer 44 and the determination computer 43 constitutes a hierarchical network including an input layer, an intermediate layer, and an output layer.
[0014]
Next, a detailed configuration of the learning computer 44 in FIG. 2 will be described.
A learning computer 44 serving as a learning means creates a first accumulation unit 441 that accumulates sound pressure level data based on the sound pressure level detected by the microphone 3, and creates teacher data from the accumulated sound pressure level data. Unit 442, a second storage unit 443 that sequentially stores the created teacher data, and a learning unit 444 that sequentially inputs the teacher data and obtains a connection load coefficient of the neural network. The sound pressure level refers to the amplitude of sound waves.
[0015]
A detailed configuration of the determination computer 43 will be described.
The determination computer 43 serving as the seeking means, determination means, and standby means includes a third storage unit 431 that stores sound pressure level data based on the sound pressure level detected by the microphone 3, and the stored sound pressure level data. A creation unit 432 that creates input data from the input unit, a search unit 433 that sequentially inputs the created input data and obtains a match rate between the actual data and the execution data by a neural network, and the obtained match rate is preset. And a determination unit 434 that determines the end of compaction in comparison with the threshold value. The solicitation unit 433 corresponds to a solicitation unit, and the determination unit 434 corresponds to a determination unit.
[0016]
Then, the connection weight coefficients between the layers of the hierarchical neural network obtained by learning in the learning unit 444 of the learning computer 44 are sent to the determination computer 43 via a medium 45 such as a LAN, and the determination computer 43 So as to make appropriate determination by the soliciting unit 433 as a coupling load coefficient between the layers of the neural network of the soliciting unit 433.
[0017]
Therefore, the determination computer 43 calculates the match rate by processing the temporal change of the sound pressure level that is sequentially input by the compaction determination function by the neural network, and the calculated match rate is a predetermined threshold value. If it is (for example, 0.99) or more, it is determined that compaction has been completed, and after a predetermined delay time (for example, 4 seconds) has elapsed, the buzzer 5 is sounded to notify that compaction has been completed.
[0018]
Next, an example of creating teacher data will be described.
First, in a plurality of cases at a certain concrete placement site, the change over time in the sound pressure level from the start of compaction to the end of compaction when an expert determines the end of compaction is measured. The actually measured broken line and its envelope are shown in FIGS. 4A, 4B, 4C, and 4D.
Observing the envelope in FIG. 3 reveals that there is a certain regularity common to each case.
[0019]
Specifically, the sound pressure level (absolute value) detected by the microphone 3 immediately after the start of the compacting operation is 0.1 as shown in FIGS. 4 (a), (b), (c), and (d). Low level around Pa.
Thereafter, as the compaction operation proceeds, the sound pressure level rises (S1) and rises to a high level peak.
Then, as the compaction operation further proceeds, the sound pressure level decreases (S2), and thereafter, the sound pressure level is slightly increased (S3). After 4 to 5 seconds, it is determined that the compaction is finished. This tends to be common in each case although there is a difference in sound pressure level and elapsed time.
[0020]
In this way, the reason why the sound pressure changes with the passage of time since the start of applying the vibrator will be examined with reference to FIG.
In the first stage of the initial compaction as shown in FIG. 5 (A), since the placed concrete has a lot of voids and the vibration region of the vibrator is narrow, the sound pressure is the sound pressure in the air. Close to the level (about 0.05Pa). That is, the sound pressure level is low.
Next, in the second stage of compaction as shown in FIG. 5B, as the compaction progresses, the voids of the aggregate are reduced by vibration, and the vibration region is expanded. As a result, the vibration is propagated over a wide area and the wide area vibrates, so the sound pressure level seems to increase.
[0021]
Furthermore, in the third stage of compaction as shown in FIG. 5 (C), the aggregate voids become smaller, and the mortar fills the voids, and the fluidization of the concrete begins. Since the mechanical impedance of concrete becomes small, the change of vibration to individual propagation sound becomes small, and it seems that the sound pressure level starts to decrease.
Then, in the fourth stage of compaction as shown in FIG. 5D, after the gap of the aggregate is filled with mortar and the concrete is almost fluidized as a whole, the compaction area is expanded. It seems that the sound pressure slightly increases.
Breathing may occur simultaneously with the expansion of the compaction area, but from the viewpoint of ensuring the strength, which is the most necessary condition for concrete, it is determined that compaction is completed after the fourth stage has elapsed. It seems to be a thing.
[0022]
First, teacher data created based on data over time of the sound pressure level indicated by the envelopes shown in FIGS. 4A, 4B, 4C, and 4D is used as the learning computer 44. To the learning unit 444.
One teacher data created at this time is set every 2 seconds in a period from the start of compaction to 60 seconds later, as shown in FIG. 6, and is composed of a total of 31 pieces of data. Note that the period and setting interval of the teacher data are not limited to this.
FIG. 6A shows an example of teacher data having a pattern in which the sound pressure level gradually increases from the start of compaction, then decreases after passing the peak, and the sound pressure level becomes 0 after 22 seconds.
As with the envelopes in FIG. 4, the sound pressure level of the teacher data increases as the compaction operation proceeds (S1), and increases to a high level peak. Further, as the compaction operation further proceeds, the sound pressure level decreases (S2), and thereafter, the sound pressure level starts to increase slightly (S3). The feature is common to each case of FIG. 4 in that compaction is completed after 4 to 5 seconds. It has. Then, the output from the neural network at this time, that is, the teacher signal is set to the maximum value (for example, 0.99).
FIG. 6B shows an example of teacher data having a pattern in which the sound pressure level gradually increases from the start of compaction, decreases after passing the peak, and becomes 0 within 20 seconds. This teacher data indicates a pattern that cannot be determined as the end of compaction. Then, the output from the neural network at this time, that is, the teacher signal is set to a minimum value (for example, 0.01).
In this way, two pieces of teacher data having the maximum value and the minimum value of the teacher signal are created from one record data. These teacher signals express the matching rate between the teacher data and the actual data.
[0023]
A case where teacher data as shown in FIG. 6A is input to the neural network of the learning unit 444 will be described with reference to FIG.
The teacher data indicated by the envelope in FIG. 6A is composed of 31 sound pressure level data (0 to 1) from D (0) to D (30). These teacher data are input to the input units N (0) to N (i) of the input layer of the neural network of FIG. That is, the teacher data D (0) at the start of compaction is input to the input unit N (0), the teacher data D (1) after 2 seconds is input to the input unit N (1), and sequentially input.
[0024]
In the intermediate layer units M (0) to M (j) of the neural network of FIG. 3, output data I (0) to I (i) from the input unit of the input layer and the connection from the input layer to the intermediate layer Using the load factor W (ji) and the offset values of the units M (0) to M (j) of the intermediate layer, the input data U (0) to the units M (0) to M (j) of each intermediate layer ) To U (j).
The output data H (0) to H (j) from the units M (0) to M (j) of each intermediate layer are obtained by the sigmoid function shown in FIG. As shown in FIG. 3C, the input / output characteristics of the sigmoid function are as follows. When the input data U (uj) is 0, the output data becomes 0.5, and when the input data exceeds 0, the input / output characteristics gradually increase. The output data approaches to.
That is, according to the sigmoid function having such characteristics, output data of 0 to 1 can be obtained instead of 0 or 1 as in a general threshold function.
The output data of each intermediate layer unit is calculated as input data of the output layer unit P. That is, using the output data H (0) to H (j) from each intermediate layer unit, the coupling load coefficient V (kj) from the intermediate layer to the output layer, and the offset value of the output layer unit, Input data U (k) to the output layer unit P is obtained.
In the output layer unit P, the output data O (k) is obtained by using the input data S (0) to U (k) and the sigmoid function.
[0025]
From the difference between the input teacher signal and the output data O (k) of the output layer, an error δk with respect to the coupling load coefficient connected to the output layer unit and the offset value of the intermediate layer unit is obtained.
Further, from the error δk, the coupling load coefficient Vkj from the intermediate layer to the output layer, and the output data of the intermediate layer, the coupling load coefficient connected to the unit of the intermediate layer and the error δj with respect to the offset of the unit of the intermediate layer are obtained.
[0026]
By adding the error δk, the output data of the intermediate layer unit, and the constant α, the coupling load coefficient Vkj connected from the intermediate layer unit to the output layer unit is corrected. Also, the offset of the output layer unit is corrected by adding the product of the error δk and the constant β.
Next, by adding the error δj in the intermediate layer unit and the product of the output data of the input layer unit and the constant α, the coupling load coefficient Wji connected from the input layer unit to the intermediate layer unit is corrected. . Also, the offset of the unit of the intermediate layer is corrected by adding the product of the error δj and the constant β.
In this way, a neural network is constructed in which output data suitable for the teacher data is obtained by sequentially correcting the combined weight coefficients.
[0027]
Next, the coupling load coefficient is corrected in the same manner using another pattern of teacher data. By applying such back-propagation algorithm to multiple patterns of teacher data, it is possible to determine a neural network connection weighting factor that can make an appropriate decision with respect to changes in sound pressure levels of various patterns over time. You can get it.
Further, by learning a pattern in which the end of compaction cannot be determined as shown in FIG. 6B, for example, when the sound pressure changes as in FIG. 6 during actual compaction, FIGS. The accuracy of determining whether the compaction has been completed or not completed in the time period between is increased.
The obtained joint load coefficient is sent to the determination computer 43 through a medium 45 such as a LAN, and the joint load coefficient between the layers of the neural network constituting the obtaining unit 433 is set.
[0028]
Next, referring to the flowchart of FIG. 8, a procedure for determining the timing of completion of compaction in actual placing work using the determination computer 43 in which the combined load coefficient obtained by the above processing is set. explain.
It is set so that the output of the A / D converter 42 is taken into the determination computer 43.
Then, the operation of the vibrator is started in step S1, the vibration sound detected by the microphone in step S2 is filtered in step S3, and an envelope of the sound pressure level change with time is created in step S4.
In step S5, the created envelope is approximated by a cubic curve.
In step S6, the sound pressure level data is stored in the third storage unit 431 every predetermined time, for example, every 2 seconds.
For example, as shown in FIG. 7, four sound pressure level data E (0) to E (3) are accumulated after 6 seconds. In step S7, the creation unit 432 sets all the remaining sound pressure level data E (4) to E (30) to 0, and 31 sound pressure level data E (0) to E (30). Create
[0029]
In step S8, the 31 pieces of sound pressure level data E (0) to E (30) created by the creation unit 432 are input to the obtaining unit 433 in which the joint load coefficient sent from the learning computer 44 is set. That is, the sound pressure level data E (0) is input to the input unit N (0) of the input layer, and the other sound pressure level data E (1) to E (30) are sequentially input to the input units N (1) to N ( Enter in 30).
In step S 9, the match rate output from the unit P of the output layer is input to the determination unit 434.
In step S10, the determination unit 434 compares the match rate with a predetermined threshold value, for example, 0.95. If the match rate is equal to or greater than the threshold value, it is determined that compaction is complete, and the process proceeds to step S11. If it is less than the threshold value, the process returns to step S2, and further waits until data E (4) after 2 seconds is accumulated, and thereafter, similarly, the five sound pressure level data E detected by the microphone are used. The same processing is performed on (0) to E (4) and the sound pressure level data E (5) to E (30) created by the creation unit to determine and determine the match rate.
If the match rate based on the five sound pressure level data from the start of compaction to 8 seconds later is also less than the threshold value, the sound pressure level data for every 2 seconds is sequentially accumulated, and the match rate is similarly determined and determined. . Further, as shown in FIG. 7 (b), nine sound pressure level data up to 16 seconds later are inputted, and the coincidence rate is obtained and determined. Also in this case, it is not determined that the compaction is finished.
[0030]
After the determination unit 434 determines that the compaction is finished, in step S11, after waiting for a predetermined time, for example, after 4 seconds, the buzzer 5 is sounded and notified in step S12. Step S11 corresponds to standby means.
The worker who has heard the ringing of the buzzer 5 stops the compacting operation by stopping the vibrator 2 in step S13.
[0031]
Thus, when the compaction work by the vibrator is stopped by the judgment of the judgment computer by the neural network, a sufficiently solid concrete is produced before a large amount of breathing water is generated, regardless of human judgment. Therefore, there is no generation of latency and sufficient strength is obtained.
[0032]
In the above embodiment, the envelope data is approximated by a curve. However, the present invention is not limited to this, and the envelope data may be directly used as an input value. In the above embodiment, the sigmoid function is used as an output function to each unit. However, the present invention is not limited to this, and a threshold function may be used. Moreover, although the intermediate layer is only one layer, the present invention is not limited to this, and a plurality of layers may be provided.
Further, the data recognition section (the section from the start of compaction to 4 seconds before the end of compaction) in the input value can be extended or contracted as appropriate without being limited to this embodiment. Accordingly, the standby time can be changed as appropriate including 0 seconds.
It should be noted that the determination load computer may also be used as a learning computer to obtain the coupling load coefficient.
Further, the microphone is not limited to the location shown in FIG. 1, and may be provided at a position closer to the vibrator as long as the vibration is not transmitted directly.
Further, instead of the buzzer, an alarm lamp may be turned on or blinked, or the vibrator may be automatically stopped.
Further, a microphone, a control device, and a buzzer may be incorporated in a unit that is separated and independent from the compacting work vehicle 1.
Furthermore, instead of detecting the change in sound pressure due to the microphone, a change in the electric current of the electric motor of the vibrator drive source and a change in the load of the hydraulic motor are detected electrically, and the compaction state is based on this. May be determined to stop the vibrator.
[0033]
【The invention's effect】
As described above, according to the present invention, since it is possible to know the appropriate timing for stopping the vibrator regardless of human judgment, it is not possible to obtain sufficient strength because the vibrator is applied for too short time. It is possible to solve the problem of breathing water being generated due to the excessive time,
It is possible to reduce the burden of monitoring the timing of compaction judgment by the operator, and it is also possible to cope with the current situation where skilled workers are insufficient.
[Brief description of the drawings]
FIG. 1 is a configuration diagram showing a configuration of an embodiment of a concrete compaction apparatus according to the present invention.
FIG. 2 is a block diagram of a main part.
FIG. 3 is a configuration diagram of a neural network.
FIG. 4 is a diagram showing changes in the surface height of concrete.
FIG. 5 is a schematic diagram for explaining the state of concrete compaction work.
FIG. 6 is an example of teacher data.
FIG. 7 is an example of actual input data.
FIG. 8 is a flowchart illustrating a procedure for determining completion of compaction.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 10 Concrete compaction apparatus 1 Compaction work vehicle 2 Vibrator 3 Microphone 4 Control apparatus 5 Buzzer 43 Judgment computer 431 Third accumulation part 432 Creation part 433 Finding part 434 Judgment part 44 Learning computer 441 First accumulation part 442 Creation part 443 Second storage unit 444 Learning unit 45 Media such as LAN

Claims (6)

In the concrete compaction judgment method for judging completion of compaction when compacting the placed concrete with a vibrator,
The input value is the time-dependent data until the sound pressure due to the vibration of the vibrator rises over time from the start of compaction to the end of compaction, exceeds the peak, and falls to the specified sound pressure level. Constructing a neural network with the value as the match rate,
By approximating the appropriate time-dependent data of sound pressure due to vibration of the vibrator from the start of compaction to the end of compaction with an envelope, and setting the joint load coefficient as the teacher data with the maximum match rate, the neural network The process of completing the network;
In actual compaction work,
While detecting the temporal change of sound pressure due to vibration of the vibrator from the start of compaction of concrete, the temporal change of sound pressure due to the detected vibration is sequentially compared with the teacher data using the neural network, and the match rate of both is calculated. The desired process;
A concrete compaction judging method, comprising: a step of judging that compaction is finished when the match rate is equal to or greater than a predetermined threshold value. By creating a time-dependent data of sound pressure due to vibration of the vibrator when concrete compaction has not been completed, and using this time-dependent data as teacher data with a minimum matching rate, a neural network can be set. The process to complete
The concrete compaction judging method according to claim 1, wherein An output means for outputting sound pressure change data over time by vibration of a vibrator for compacting the placed concrete;
Appropriate time-dependent data until the sound pressure due to vibration of the vibrator rises over time from the start of compaction to the end of compaction, exceeds the peak, and falls to a predetermined sound pressure level. Finding means for calculating a match rate by comparing using predetermined neural network and predetermined teacher data approximated by an envelope,
A determination means for comparing the calculated match rate with a predetermined threshold value, and determining that the compaction is completed when the match rate is equal to or greater than the threshold value;
A concrete compaction judging device characterized by comprising: 4. A learning means for setting a coupling load coefficient of a neural network of a search means based on a temporal change data of sound pressure due to vibration of a vibrator from a start to an end of a compaction as a reference. Concrete compaction judgment device described in 1. Waiting means for outputting a vibrator stop command after a predetermined waiting time has elapsed since the determination means determines that compaction has ended.
The concrete compaction determination device according to claim 3, wherein the concrete compaction determination device is provided. The concrete compaction determination according to any one of claims 3, 4, and 5, wherein the output means detects a change in sound pressure due to vibration of the vibrator with a microphone and outputs data with a lapse of time of the signal. apparatus.

Images