JP2007064672A - Method of estimating hollow state of tree trunk, its device and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method of estimating a hollow state of a tree trunk not damaging the tree trunk and independent of experience and intuition, its device and a program. <P>SOLUTION: A hammering sound of a tree 1 collected by a microphone 3 is subjected to wavelet analysis, and a hammering sound having characteristic quantities A, B not existing in an analysis result of a sound tree trunk is detected. Existence of a cavity in the tree 1 from which the sound is collected is detected from existence of the characteristic quantities A, B. Hereby, the cavernous state of the tree trunk can be estimated independently of experience and intuition. In addition, since the hammering sound is collected by the microphone 3 to a recording device, the tree trunk 7 is not damaged. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a method for estimating a hollow state of a tree, an apparatus thereof, and a program for estimating a hollow state inside the trunk by collecting and analyzing the sound of a tree trunk.

  Trees planted in public places provide peace of mind to our living environment and fulfill various missions such as disaster prevention and soundproofing. However, they often have some defects, such as cavities and decay. For example, there is a phenomenon of falling down in weather that is not a special strong wind. Many of them are considered to be caused by a decline in trunk support ability caused by decaying fungi caused by some cause into the root stock and causing deterioration of the heartwood.

  Even in trees that have undergone heartwood decay, there is a formation layer even though the support reinforcement in the heartwood part is lost, and green leaves grow up until just before lodging, and it may appear to grow vigorously. For this reason, the conventional diagnostic method that focuses on the living part such as activity determination cannot predict the risk such as lodging.

  Therefore, hitherto, it has been practiced to strike a tree trunk with a wooden hammer, etc., and a person (tree doctor) will hear the sound of the obtained sound and estimate the hollow state inside the tree trunk. However, a tree is a plant that grows and grows in the stem and root, and the cells and tissues constituting the tree are very complicated. In other words, trees are alive and growing repeatedly, and due to various external factors (environmental factors such as soil) during the growth process, the individual shapes of the trees are completely different from one another, and the trunks differ due to the differences in the external shapes. The internal structure is also different. Even if a hammering sound is generated for such a tree, the way the sound is transmitted in the trunk differs from tree to tree due to differences in the external shape of the tree and the retention of moisture in the cavity. Therefore, it is very difficult to estimate the state of the cavity from the percussion sound even if it requires skilled skills based on experience. The work itself was very inefficient.

  For the purpose of improving the efficiency of such work, conventionally, a tree diagnostic device and a trunk interior diagnosis system have been provided. A tree diagnostic device is a diagnostic device that measures the soundness of heartwood by inserting a thin drill into a tree trunk. The resistance value when inserting a tree into a tree trunk is measured, and if the resistance is large, the tree is judged to be healthy, and if the resistance is small, it can be judged that the tree is decayed or hollow (for example, Non-Patent Document 1). ).

  In addition, the trunk interior diagnosis system can make a diagnosis based on the speed at which sound travels inside the trunk. This is a technology that inserts a needle-like terminal into a tree trunk, generates ultrasonic waves therefrom, observes a reflected wave obtained from the tree trunk with a receiver, and displays the result by computer processing. Actually, it has been put to practical use as an apparatus for visualizing a cross section of a trunk by a technique using CT (Computed Tomography) based on the frequency and intensity of a reflected wave (for example, Non-Patent Document 2).

http: // www. toho-leo. co. jp / header. html http: // www. environment. co. jp / picus2. html #Anchor 668348

  In the above-mentioned tree diagnostic device, it is necessary to make a hole in the trunk, resulting in damage to the trunk. Scratching the trunk of the tree has the potential to cause decaying fungi to invade from the wound and newly generate a cavity inside the trunk, which has been a major problem as a method for detecting the cavity inside the trunk. In addition, when trying to apply a conventional tree diagnostic device or trunk interior diagnosis system to a hollow inspection inside an actual trunk, for example, the above-mentioned tree diagnostic device performs an internal hollow inspection from a resistance value when a tree is pierced in a tree trunk. Experience and intuition are necessary for this, and what part of the trunk to inspect must be determined on a trial and error basis based on the measurement results, and the measurement method was not clear. On the other hand, in the above-mentioned trunk interior diagnosis system, it is not necessary to make a hole in the trunk as deep as the above-described tree diagnostic device, but it is necessary to pierce the trunk with a terminal in order to transmit ultrasonic waves to the inside of the trunk. This is also a cavity inspection method that damages the trunk. In addition, there is a large error between the obtained image and the actual cavity state, and there is a problem that experience and intuition for understanding the image are necessary to recognize the cavity.

  An object of the present invention is to solve the above-described problems, and to provide a method for estimating a hollow state of a trunk, a device therefor, and a program without damaging the trunk and without relying on experience and intuition. Is.

  In order to solve the above problems, a trunk cavity state estimation method according to the present invention is a method for detecting a cavity inside a trunk, and performs a wavelet analysis on the sound of the collected tree and obtains an analysis result of a trunk without a cavity. It is designed to detect the hitting sound of the feature amount.

  Moreover, the hollow state estimation device for a trunk according to the present invention is a recording device that records the sound of a tree trunk, and inputs the recorded sound data, performs wavelet analysis, and is not included in the analysis result of a trunk without a cavity. It is comprised from the computer which performs the hit detection of a feature-value.

  Further, the trunk cavity state estimation apparatus according to the present invention may be configured such that the computer includes an output unit for displaying the presence of a cavity in the trunk.

  Still further, the trunk cavity state estimating device according to the present invention is attached to the trunk in close contact with the impact device for applying a constant impact, and an electrical signal generation for transmitting and controlling the electrical signal to the impact device. It can also be set as the structure which provides an apparatus.

  Furthermore, the trunk cavity state estimating device according to the present invention is attached so as to be in close contact with the trunk, and has a configuration in which at least one vibration converter is provided for converting the vibration inside the trunk as a sound wave. You can also.

  In addition, the trunk cavity state estimating device according to the present invention can employ a configuration in which a sound wave generating device for activating the electrical signal generating device is provided.

  Furthermore, the cavity situation estimation program according to the present invention includes a computer for estimating the trunk cavity situation, data input means for inputting trunk sound data, and data recording means for recording the input sound data. Analyzing means for performing wavelet analysis on the recorded hitting sound data; and discriminating means for discriminating whether or not the hitting sound of the feature quantity does not exist in the analysis result of the trunk without a cavity based on the analysis result by the analyzing means; , An estimation means for estimating the presence / absence of a trunk cavity from the discrimination result by the discrimination means, and an output means for outputting the estimation result by the estimation means.

  Further, the cavity situation estimation program according to the present invention is the above-described cavity situation estimation program, wherein the computer estimates the correctness / incorrectness input means for inputting the correctness / incorrectness of the estimation result by the estimation means, and the correctness / error data input by the correctness / incorrectness input means It is made to function as a storage means for storing as one of the standards.

  The trunk cavity state estimation method, apparatus, and program according to the present invention perform wavelet analysis of a hit sound and investigate a cavity based on the analysis result. Thereby, the presence or absence of a cavity can be determined without damaging the trunk and without relying on experience and intuition. In addition, it is possible to eliminate the decline in trunk support capacity due to deterioration of the heartwood part, and furthermore, it is possible to remove in advance the trees with reduced trunk support capacity and prevent accidents due to lodging. is there.

  The greatest feature of the present invention is that the cavity of the trunk can be found based on a certain rule by performing the wavelet analysis on the sound of the trunk. DESCRIPTION OF EMBODIMENTS Hereinafter, a trunk cavity state estimation method, apparatus, and program according to the present invention will be described with reference to the drawings.

  FIG. 1 is a flowchart showing a trunk cavity state estimation method according to the present invention. The method for estimating the hollow state of the trunk is a method for detecting a hollow inside the trunk, and performing a wavelet analysis on the sound of the collected trunk, and detecting a hammering sound of a feature amount not included in the analysis result of the trunk without the cavity Is done.

  First, the sound of the trunk is collected (step S1). There is no particular limitation on the method of collecting the sound of the hit, and a method of collecting the tree 1 by manually hitting the tree 1 with the mallet 2 or a method of using the impact device 10 mechanically can be considered. The collected sound is input to the computer 5 directly or via the recording device 4 such as an IC recorder in order to perform wavelet analysis (step S2).

  Next, the hitting sound input to the computer 5 is subjected to an editing process for cutting off a silent part and a part including noise as a step before wavelet analysis (step S3), and then a wavelet analysis of the hitting sound is performed (step S3). Step S4). Wavelet analysis is an analysis method proposed by Morlet in the 1980s, and is a method for determining how strongly a target sound wave or the like includes a waveform of a specific frequency called a wavelet. The continuous wavelet transform formula of the time series data f (x) of the wavelet analysis in the present invention is defined by the following equations 1 and 2.

  An expression obtained by substituting the above expression 2 into expression 1 is an expression used in the wavelet analysis in the present invention. The symbols in the above formulas 1 and 2 are described as shown in Table 1 below.

  Then, by using wavelet analysis, the feature amount is detected as a feature amount indicating how strongly a specific frequency is included, and processing corresponding to a predetermined estimation criterion for estimating the cavity state is performed, thereby Is estimated (step S5).

  The cavity state obtained by estimation in step S5 is output and displayed on the display 8 of the computer 5 (step S6). Such output display is performed, for example, in the form of a tree trunk 7 with the presence or absence of a cavity 6 as shown in FIG.

  The result of the wavelet analysis can be displayed by the image of FIG. 3, for example. The process of step S5 will be specifically described based on FIG. FIG. 3 shows the result of wavelet analysis displayed on the display 8 as an output means so that the result can be understood visually.

  The upper side of the display 8 shown in FIG. 3 is a waveform 9 of the hitting sound that is the object of wavelet analysis. The vertical axis represents amplitude and the horizontal axis represents time. The center line has a value of zero amplitude, and the sound wave is displayed as a curve above and below it with zero amplitude as a boundary.

  The lower side of the display 8 shown in FIG. 3 is the result of the wavelet analysis, the vertical axis is the frequency (Hz) value, and the horizontal axis is the time (seconds). In such a display 8, corresponding to the upper hit sound signal, the sound wave represents which frequency component and how strong the sound wave contains at a certain time. In FIG. 3, the strength including the frequency component indicates that the strength including the frequency component corresponding to the order of the regions P, Q, R, S, and T is high. In the above region, the strongest hitting sound is defined as 100, the P region is in the range of 100 to 87.5, the Q region is in the range of 87.5 to 62.5, and the R region is in the range of 62.5 to 37.5. The S region is in the range of 37.5 to 12.5, and the T region is 12.5 or less.

  From the results of the wavelet analysis shown in FIG. 3, the display results in the rectangles in the figure are feature quantities A and B, and compared with the analysis results of the sound of the trunk 7 in which the hollow state inside the trunk 7 is known. The hollow condition of the trunk 7 to be estimated is obtained. According to the result of analysis of the sound of the trunk, where the hollow state inside the trunk 7 is known, the sound of the feature amounts A and B was not found in a healthy tree. Moreover, the existence of the cavity 6 was recognized in the trunk 7 according to the analysis result of FIG. 3 (FIG. 2).

  Next, the cavity condition estimation criteria in step S5 will be described. The hollow state is estimated based on the feature amount. From the above analysis result, the characteristic amount is in a time zone from 1/50 second to 2/50 second from the time when the strongest hitting sound is generated, the strength of the hitting sound is 37.5 or more, and If the sound hits at a frequency of 200 Hz to 1200 Hz, it is estimated that “there is a cavity”. When the two feature quantities exist in different frequency bands in the time period, the determination is made in the uppermost frequency band. In addition to the above, when the characteristic amount is a hitting sound having a frequency of 87.5 or more and 1050 Hz or more when 1/50 second has elapsed since the time when the strongest hitting sound is generated, Presume that there is a cavity. Even if this is not the case, if the feature amount is in a frequency band of 37.5 or more and 680 Hz or more when 2/50 seconds have elapsed, it is estimated that “there is a cavity”. On the other hand, if none of the above applies, it is estimated that there is no cavity.

  Next, an apparatus for carrying out the above trunk cavity state estimation method will be described. FIG. 4 is a schematic diagram showing a configuration of a trunk cavity state estimation apparatus according to the present invention. The hollow state estimation device for a trunk is a recording device 4 that records the sound of a trunk and inputs the recorded sound data, performs wavelet analysis, and hits a feature amount that is not included in the analysis result of a trunk without a cavity. And a computer 5 for detecting sound.

  The recording device 4 is connected to a microphone 3 that collects a hitting sound. By hitting the tree 1 with the mallet 2, the sound of the trunk 7 is generated, and the sound is collected by the microphone 3 and recorded in the recording device 4. In the microphone 3 and the recording device 4, at least one set of sound recording data recorded on one tree 1 is input to the computer 5, wavelet analysis is performed, and there is no analysis result of a trunk without a cavity. The hitting detection of the feature amount is performed. The computer 5 incorporates a program for wavelet analysis of the hitting data and finally estimating the hollow state of the trunk 7, as well as hitting conversion software, hitting transfer software, and the like. Necessary application software is installed.

  Note that at least one microphone 3 and recording device 4 may be provided for one tree 1, and a plurality of microphones 3 and recording devices 4 may be provided for one tree 1. By providing a plurality, the same effects as those obtained when a plurality of vibration converters 12 described later are provided can be obtained.

  Further, the computer 5 may be configured to include an output unit for displaying the presence / absence of a cavity in the trunk based on the estimation result of the cavity state of the trunk 7. As an example of display by such an output means, it is performed in the form of a diagram of a trunk 7 marked with or without a cavity 6 as shown in FIG.

  When collecting the sound of a tree trunk, as shown in FIG. 5, an impact device 10 may be used instead of the mallet 2. The impact device 10 is attached in close contact with the surface of the trunk 7, and the impact device 10 is fixed to the surface of the trunk 7 by sending a constant electrical signal from the electrical signal generator 11. It gives an impact. By adopting such a configuration, there is no variation in impact when a person hits the trunk 7 with the tree 2, and it becomes possible to always give a constant magnitude impact to the trunk 7, thereby improving the accuracy of cavity condition estimation. It becomes possible.

  In addition, as shown in FIG. 6, a mode in which the vibration conversion device 12 is connected to the recording device 4 instead of the microphone 3 can be considered. The vibration conversion device 12 is attached in close contact with the surface of the trunk 7, and converts vibrations inside the trunk as sound waves when recording the hitting sound of the trunk 7. At least one vibration conversion device 12 is provided on the surface of the trunk 7, and the sound waves converted by the vibration conversion device 12 are recorded on one or more recording devices 4 corresponding to the vibration conversion devices 12.

  FIG. 7 shows a cross-sectional view of the trunk 7 in the case where a plurality of the vibration converting devices 12 are provided on the surface of the trunk 7. The impact applied using the impact device 10 propagates inside the tree trunk 7 as vibrations and reaches a plurality of vibration conversion devices 12 that convert each vibration as sound waves. At that time, since the distance from the impact device 10 to each of the plurality of vibration converters 12 is different, the time until the vibration reaches each of them is different. Furthermore, since the speed of vibration propagation is different between the cavity 6 inside the trunk and the non-cavity portion 13, this difference in vibration propagation speed is added to the previous time difference. The computer 5 can estimate the number, size, and position of the cavities using the time difference in which the vibration is propagated.

  Moreover, as shown in FIG. 8, the structure which provided the sound wave generator 14 for starting the said electric signal generator 11 can also be considered. The sound wave generator 14 gives an impact to the trunk 7 by an impact device 10 that gives an impact to the surface of the trunk 7 and an electrical signal generator 11 that gives an electrical signal corresponding to the impact, and causes one or more vibrations caused thereby. This is effective when the impact device 10 that gives an impact to the surface of the trunk 7 and the one or more recording devices 4 are operated in synchronization. That is, by providing the sound wave generating device 14, the sound wave generating device 14 first generates a predetermined sound wave, and subsequently receives the generated sound wave, and then gives an electric signal corresponding to the impact after a certain period of time. Is activated, the impact device 10 that actually impacts the surface of the trunk 7 is used to impact the trunk 7, and after a certain time, one or more recording devices 4 convert the vibration of the trunk 7 into sound waves. A series of processing such as recording is performed.

  Note that the sound wave generated by the sound wave generator 14, the time from when the sound wave is received until the electric signal generator 11 is started, and the time until one or more recording devices 4 are started are arbitrary values. Can be set.

  Next, a hollow state estimation method for the trunk 7 and a program for executing the apparatus will be described with reference to the drawings.

  FIG. 9 is a flowchart showing the first embodiment of the cavity status estimation program according to the present invention. The hollow state estimation program uses the computer 5 to estimate the hollow state of the trunk 7, data input means (step 1) for inputting sound data of the trunk 7, and data recording for recording the input sound data. Means (step 2), analysis means (step 3) for performing wavelet analysis on the recorded hitting sound data, and feature quantities that are not present in the analysis result of the trunk 7 having no cavity based on the analysis result by the analysis means Discriminating means for discriminating the presence / absence of the hitting sound (step 4), estimating means for estimating the presence / absence of the cavity of the trunk 7 from the discrimination result by the discriminating means, and output for outputting the estimation result by the estimating means It is made to function as a means (step 6).

  With respect to the computer 5 that executes the cavity situation estimation program according to the first embodiment, first, the sound data of the collected tree trunk is taken into the computer 5 by the data input means (step 1) and taken by the data input means. The hit sound data is recorded on the hardware in the computer 5 by the recording means (step 2).

  Next, wavelet analysis will be performed by the analyzing means on the hitting data recorded by the recording means, but as a previous step, an editing process for cutting off silent portions and noise-containing portions in the hitting data After that, wavelet analysis of the hit sound data is performed (step 3).

  After that, based on the analysis result by the analysis means, the determination means determines the presence / absence of a hitting sound of a feature amount that does not exist in the analysis result of the trunk without a cavity (step 4), and estimates from the determination result by the determination means The presence / absence of the cavity of the trunk 7 is estimated by means (step 5). At this time, if it is determined in step 4 that there is no feature value, it is immediately estimated as “no cavity” in step 5; conversely, if it is determined that the feature value is present, in step 5, a predetermined amount is determined. Based on the estimation criterion, the existence of the cavity of the trunk 7 such as “with a cavity” or “without a cavity” is estimated.

  Finally, the estimation result of “with cavities” or “without cavities” by the estimation means is output and displayed on the display 8 of the computer 5 by the output means (step 6), and the cavity condition estimation program according to the present invention is executed. A series of processing of the computer 5 ends.

  FIG. 10 is a flowchart showing a second embodiment of the cavity status estimation program according to the present invention. The program is the hollow state estimation program according to the first embodiment, wherein the computer 5 further inputs a correct / incorrect input means (step 7) for inputting the correctness / incorrectness of the estimation result by the estimating means (step 5), and the correct / incorrect input means. It is made to function as a storage means (step 8) for storing the correct / incorrect data input as described above as one of the estimation criteria.

  Regarding the cavity situation estimation program according to the second embodiment, a series of processing from step 1 to step 6 executed by the computer 5 is the same as that of the first embodiment. In the cavity condition estimation program according to the second embodiment, after the process of step 6 is completed for the computer 5, the correctness / incorrectness of the estimation result in step 5 is further input to the computer 5 by the correct / incorrect input means (step 7). ). Such an input method is performed by simply inputting “correct” or “incorrect” to the computer 5 based on whether or not the existence of the cavity of the trunk 7 actually matches the estimation result. Is called. The correct / incorrect data input by the correct / incorrect input means is stored in the hardware in the computer 5 by the storage means to be used as one of the estimation criteria in the estimating means in Step 5 (Step 8).

  The cavity situation estimation program according to the second embodiment utilizes one of learning algorithms generally called a neural network, and the computer 5 learns the estimation result of the cavity situation of the trunk 7 so far. Subsequent estimation is performed based on learning. That is, when the hit data is subjected to wavelet analysis (step 3) and the presence of a feature amount is determined (step 4), the existence of the cavity of the trunk 7 is estimated (step 5) based on a predetermined estimation criterion. However, such estimation criteria include correct / incorrect data for previous estimation results that the computer 5 has learned and recorded so far. As a result of the estimation of the existence / non-existence of the hollow of the trunk 7, the correctness / incorrectness of the actual existence / non-existence of the hollow of the trunk 7 is input to the computer 5 and recorded. By causing the computer 5 to repeatedly learn whether the estimation result is correct or incorrect, it is possible to dramatically improve the probability that the estimation result matches the actual existence of the hollow of the trunk 7.

Experimental example 1

  Next, in order to confirm the effect of the trunk cavity state estimation method and apparatus and program having the above-described configuration, the trunk cavity state estimation apparatus according to the present invention was configured and experimented. About 100 percussion sounds, that is, a total of 2142 percussion sounds, were collected from each of 23 tree trunks for which a diagnosis result of “with / without a cavity” was actually obtained by a tree doctor. The collected sound was analyzed by wavelet, and the judgment result of “with / without cavity” was compared with that of a tree doctor. As the trunk cavity state estimation device, the “pack electret condenser microphone; ECM-MS957” is used as the microphone 3 and “Sony Corporation; IC RECORDER ICD-ST25” is used as the IC recorder 4, and the personal computer 5 uses Windows as the OS. The one equipped with (registered trademark) was used.

  As a result, of the 2142 hitting sounds, the judgment result of “with / without a cavity” for the 1787 hitting sounds coincided with that of the tree doctor. Assuming that the judgment result of the tree doctor is correct, the hit rate corresponding to the correct answer was 83%. This is a very high effect. In addition, the hit ratio for about 100 hits per 23 tree trunks was 53% minimum and 99% maximum. In the present embodiment, there is a hit rate of 53%, in this example, there is a variation in impact caused by a person hitting the trunk with a wooden fence, noise during recording, trunk type, size, etc. This is because there are many inadequate points in the example in which part of the invention is implemented in terms of estimation accuracy because the physical conditions are not considered.

It is a flowchart which shows the hollow condition estimation method of the trunk concerning this invention. It is sectional drawing which shows the trunk of an estimation result. It is explanatory drawing which shows the example of a display display of the analysis result of a wavelet analysis. It is a schematic explanatory drawing which shows the structure of the hollow condition estimation apparatus of the trunk of this invention. It is a schematic explanatory drawing which shows the structure of the hollow condition estimation apparatus of the trunk of this invention. It is a schematic explanatory drawing which shows the structure of the hollow condition estimation apparatus of the trunk of this invention. It is a schematic explanatory drawing which shows the structure of the hollow condition estimation apparatus of the trunk of this invention. It is a schematic explanatory drawing which shows the structure of the hollow condition estimation apparatus of the trunk of this invention. It is a flowchart which shows 1st embodiment of the cavity condition estimation program concerning this invention. It is a flowchart which shows 2nd embodiment of the cavity condition estimation program concerning this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Tree 2 Kiso 3 Microphone 4 Sound recording device 5 Computer 6 Cavity 7 Trunk 8 Display 9 Waveform 10 Impact device 11 Electric signal generator 12 Vibration converter 13 Non-cavity part 14 Sound wave generator A, B Characteristic quantity

Claims (8)

  A method for detecting a cavity inside a trunk, characterized by wavelet analysis of the sound of the collected trunk and detecting a sound of a characteristic amount not included in the analysis result of a trunk without a cavity Estimation method.   It consists of a recording device that records the sound of the sound of the trunk, and a computer that inputs the recorded sound data, performs wavelet analysis, and detects the sound of the feature that is not in the analysis result of the trunk without a cavity. Characteristic trunk cavity estimation device.   3. The trunk cavity state estimation apparatus according to claim 2, wherein the computer includes output means for displaying the presence / absence of a cavity inside the trunk.   3. The apparatus according to claim 2, further comprising: an impact device that attaches the tree trunk in close contact, and that applies a constant impact; and an electrical signal generator that transmits and controls an electrical signal to the impact device. The trunk cavity state estimation device according to claim 3.   5. The apparatus according to claim 2, further comprising at least one vibration conversion device that is attached in close contact with the tree trunk and converts vibrations inside the tree trunk as sound waves. Trunk cavity state estimation device.   6. The trunk cavity state estimation device according to claim 4 or 5, further comprising a sound wave generation device for activating the electrical signal generation device.   In order to estimate the hollow state of the trunk, a computer is used. Data input means for inputting the sound data of the trunk, data recording means for recording the input sound data, and wavelet analysis for the recorded sound data. Analysis means to perform, discrimination means for discriminating the presence or absence of hitting of a feature quantity that does not exist in the analysis result of a trunk without a cavity based on the analysis result by the analysis means, and determination of the cavity of the trunk from the discrimination result by the discrimination means A cavity condition estimation program that functions as an estimation unit that estimates presence / absence and an output unit that outputs an estimation result by the estimation unit. In the cavity situation estimation program, the computer functions as correct / incorrect input means for inputting the correctness / incorrectness of the estimation result by the estimating means and as storage means for storing the correct / incorrect data input by the correct / incorrect input means as one of the estimation criteria. The hollow state estimation program according to claim 7, wherein: