Living noise impact detection system and method

Abstract

To provide a living noise impact detection system and method capable of estimating a noise level generated by residents' daily activities in an apartment building that affects lower-floor units, and estimating noise-generating activities and notifying residents.SOLUTION: A living noise impact detection system 100 includes an acceleration sensor 10 fixed to a floor slab 2 of an upper-floor unit, a control unit 20, and a noise notification device 30. The acceleration sensor 10 detects vibration acceleration 4 generated in the floor slab 2. The control unit 20 estimates a noise level B transmitted to a lower-floor unit and a noise behavior Y of living noise from acceleration data 5 indicating time variation of the vibration acceleration. The noise notification device 30 notifies residents of the upper-floor unit of the noise level B and the noise behavior Y of the living noise when the estimated noise level B exceeds a predetermined threshold value.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a system and method for detecting the influence of sounds in an apartment building. [Background technology]

[0002] In apartment complexes and other collective housing, noise from daily life, such as walking noise, transmitted from other residences can be a problem. However, it is usually difficult to completely block such noise using architectural techniques alone. Therefore, in order to maintain a good living environment, consideration must be given to the way residents live, and technologies to support this consideration have been proposed (for example, Patent Documents 1 and 2).

[0003] The "noise detection system" in Patent Document 1 includes a vibration detection sensor that detects vibrations on the floor above and outputs a voltage signal corresponding to the magnitude of the vibration, a noise conversion unit, a noise conversion condition setting unit, a vibration and noise conversion table, and a control unit. The noise conversion unit converts the voltage signal from the vibration detection sensor into the value of noise on the floor below and outputs the converted noise, and the control unit outputs a warning output command to the warning unit if the converted noise value is greater than a threshold value.

[0004] In the "noise monitoring system" of Patent Document 2, a management server placed on the Internet has a noise data collection unit, a noise level registration unit, and a noise level notification unit. The noise data collection unit sequentially stores the noise levels detected in each room of an apartment building or the like in a noise database. The noise level registration unit registers occupants of rooms whose noise levels are below the noise threshold on a whitelist, and registers occupants of rooms whose noise levels exceed the noise threshold on a blacklist. The noise level notification unit notifies occupants of rooms registered on the whitelist that useful services will be provided, and notifies occupants of rooms registered on the blacklist that they should be careful. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Patent No. 3751628 [Patent Document 2] Patent Publication No. 2021-76383 Summary of the Invention [Problem to be solved by the invention]

[0006] The system in Patent Document 1 uses a vibration detection sensor installed in the slab of one's own apartment to estimate the noise generated in the apartment on the floor below, and if the estimated value exceeds a threshold, it issues a warning to the apartment where the noise is generated.

[0007] The system of Patent Document 1 has the following problems. (1) A vibration detection sensor detects vibrations on the floor above and outputs a voltage signal corresponding to the magnitude of the vibration, and a noise conversion unit converts the voltage signal from the vibration detection sensor into a noise value on the floor below and outputs the converted value. The noise generated in the apartment on the floor below is estimated based on the obtained noise value. However, while this system can estimate the level of noise generated in the apartment below (hereinafter referred to as the estimated noise level), it cannot identify the noise-generating behavior. (2) When the estimated noise level exceeds the threshold, a warning is issued in real time at that point in time. Therefore, it is not possible to review the warning situation later, and it is not possible to evaluate the impact if the warning occurs repeatedly.

[0008] Furthermore, since the system of Patent Document 2 directly uses the noise levels detected in each room of an apartment building or the like, this system cannot estimate the volume of noise occurring in the dwelling units on the floors below.

[0009] The present invention has been invented to solve the above-mentioned problems. That is, the first object of the present invention is to provide a living sound influence detection system and method that can estimate the noise level caused by the living activities of residents of an apartment building in the lower floors, and can estimate the noise-generating activities and notify the residents. A second object of the present invention is to provide a system and method for detecting the impact of noise on residents that allows them to look back at any time on noise-generating behavior that they have been notified of and to evaluate the impact of such behavior. [Means for solving the problem]

[0010] According to the present invention, in an apartment building having an upper floor dwelling and a lower floor dwelling, a life sound influence detection system is provided for detecting the influence of life sounds of the upper floor dwelling on the lower floor dwelling, an acceleration sensor fixed to a floor slab of the upper floor dwelling unit and detecting a vibration acceleration occurring in the floor slab; a control unit that estimates the noise level propagating to the lower floor dwelling unit and the noise behavior of the living sound from acceleration data indicating a time change of the vibration acceleration; A living sound influence detection system is provided which includes a noise notification device which, when the estimated noise level exceeds a predetermined judgment value, notifies the resident of the upstairs residence of the noise level as well as the noisy behavior of the living sounds.

[0011] In addition, according to the present invention, the above-mentioned life sound influence detection system is used, Detecting the vibration acceleration occurring in the floor slab; estimating the noise level propagating to the lower floor dwelling unit from the acceleration data; estimating the noise behavior of the living sound from the acceleration data; A living sound influence detection method is provided which, when the estimated noise level exceeds the judgment value, notifies the resident of the upper floor dwelling unit of the noise level and the noisy behavior of the living sound. Effect of the Invention

[0012] The floor slab of the upper floor apartment is less susceptible to the noises of the apartment below, but is susceptible to vibrations caused by noises from the apartment above. In order to utilize this vibration characteristic, in the present invention, an acceleration sensor is fixed to the floor slab of the upper floor dwelling unit, so that the noise level propagating to the lower floor dwelling unit can be estimated from the acceleration data generated on the floor slab.

[0013] In addition, the control unit estimates the noise level propagating to the dwelling on the floor below and the noise-generating behavior of daily life sounds from acceleration data that indicates the time change of vibration acceleration, so that the resident can be notified of the noise-generating behavior (noisy behavior of daily life sounds) together with the estimated noise level. [Brief description of the drawings]

[0014] [Figure 1] 1 is an overall configuration diagram of a life sound influence detection system according to the present invention; [Diagram 2] FIG. 2 is an overall configuration diagram of a control unit. [Diagram 3] 1A to 1C are diagrams illustrating examples of changes in vibration acceleration over time in an impact system Y1, a drop system Y2, and a walking system Y3. [Figure 4] FIG. 13 is a diagram showing an example of a change over time in vibration acceleration 4 of the equipment system Y4 and the drag system Y5. [Diagram 5] FIG. 2 is a functional explanatory diagram showing a specific example of an action inference device. [Figure 6] 1 is an overall configuration diagram of a noise analysis device. [Figure 7] FIG. 1 is an explanatory diagram of a noise analysis device. [Figure 8] FIG. 13 is a diagram showing noise occurrence frequencies C of noise behaviors (behavior-specific noise occurrence frequencies C2) calculated by an occurrence frequency calculation device. [Figure 9] FIG. 13 is an explanatory diagram of contribution degree E3 by action. [Figure 10] This is a layout diagram of the experimental equipment simulating an upper floor dwelling unit. [Figure 11] FIG. 1 is a diagram showing the relationship between sampling frequency and F1-score. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0015] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. In addition, common parts in each drawing are given the same reference numerals, and duplicated explanations will be omitted.

[0016] FIG. 1 is a diagram showing the overall configuration of a life sound effect detection system 100 according to the present invention. The life sound influence detection system 100 is a system for detecting the influence of life sounds from an upper floor dwelling on a lower floor dwelling in an apartment building 1 having an upper floor dwelling and a lower floor dwelling below the upper floor dwelling.

[0017] In FIG. 1, the life sound effect detection system 100 includes an acceleration sensor 10, a control unit 20, and a noise notification device 30.

[0018] The acceleration sensor 10 is fixed to the floor slab 2 of the upper floor dwelling unit and detects the vibration acceleration 4 occurring in the floor slab 2. The acceleration sensor 10 is preferably a piezoelectric acceleration sensor. The piezoelectric acceleration sensor can detect accelerations of 5 μG to 10 kG in a frequency range of 0.5 to 20 kHz, for example. The acceleration sensor 10 is not limited to a piezoelectric acceleration sensor, but may be a strain gauge type, a capacitance type, a servo type, or the like.

[0019] The control unit 20 is preferably a computer (PC) and includes an input / output device, a storage device, and an arithmetic unit. The storage device is a RAM (Random Access Memory) and a ROM (Read Only Memory). The control unit 20 estimates the noise level B propagating to the lower floor dwelling unit and the noise behavior Y of the living sounds from the acceleration data 5 indicating the change over time of the vibration acceleration 4. Noise level B is an estimate of the noise level transmitted to the dwelling on the floor below. Note that the control unit 20 may be connected to an internet line, and some of the functions of the control unit 20 may be performed by an external computer (cloud processing).

[0020] Living noises in apartment buildings can be broadly categorized into (1) noise from household appliances, (2) noise from housing facilities and structures, (3) noise from audio equipment, and (4) other noises. "Noise from household appliances" refers to sounds from washing machines, refrigerators, air conditioners, outdoor units, vacuum cleaners, household heat pump water heaters, etc. "Noise from housing facilities and structures" refers to sounds such as doors opening and closing, furniture being moved, and water supply and drainage sounds from the bath. "Noise from audio equipment" includes sounds from televisions, audio equipment, alarm clocks, and musical instruments such as pianos, drums, and guitars. "Other noises" include sounds such as footsteps indoors or on the stairs, talking, pets barking, etc. In addition, the living sounds that are the object of detection in this invention are limited to "sounds that are caused by the vibration of the dwelling floor due to living and that are then transmitted to the dwelling on the floor below," and do not include the sound of a television, the cries of pets, etc. There are various kinds of living sounds, but the one that is particularly likely to be a problem in an apartment building is the sound that is caused by the vibration of the dwelling floor due to living and that is then transmitted to the dwelling on the floor below. The living sounds that are the subject of the present invention are such floor vibration sounds.

[0021] In the present invention, the noise behaviors Y of daily life sounds are classified into impact-related Y1, falling-related Y2, walking-related Y3, equipment-related Y4, and dragging-related Y5. Each noise behavior will be described in detail later.

[0022] When the estimated noise level B exceeds a predetermined judgment value, the noise notification device 30 notifies the resident of the upper floor dwelling unit of the noise behavior Y of the living sound together with the estimated noise level B. The noise notification device 30 is preferably a display device capable of outputting audio. A smartphone may also be used as a part of the noise notification device 30.

[0023] FIG. 2 is a diagram showing the overall configuration of the control unit 20. As shown in this figure, the control unit 20 comprises a data storage device 22, a noise estimator 24, and an activity estimator .

[0024] The data storage device 22 stores acceleration data 5 indicating the change in vibration acceleration 4 over time. The data storage device 22 is preferably a magnetic disk (eg, HDD) or flash memory (eg, SSD, USB memory, SD card) connectable to a computer (PC). The data storage device 22 also stores the estimated value of the noise level propagating to the lower floor dwelling unit (noise level B) and the estimated noise behavior Y of the living sound together with the time of occurrence.

[0025] The noise estimation device 24 estimates the noise level B propagating to the lower floor dwelling unit from the acceleration data 5 indicating the change over time of the vibration acceleration 4.

[0026] FIG. 3 is a diagram showing an example of change over time in vibration acceleration 4 of the impact system Y1, the drop system Y2, and the walking system Y3, and FIG. 4 is a diagram showing an example of change over time in vibration acceleration 4 of the equipment system Y4 and the drag system Y5. In each of Figures 3 and 4, the horizontal axis represents time (seconds) and the vertical axis represents the magnitude of vibration acceleration 4 (m / s 2 ). In FIG. 3, (A) is an example of a jump in the impact system Y1, (B) is an example of a fall in a bathroom bucket in the drop system Y2, and (C) and (D) are examples of walking in slippers and running in slippers in the walking system Y3. In addition, in FIG. 4, (A) and (B) are examples of the use of a kitchen food processor and a drum washing machine spin-drying cycle, which are equipment system Y4, and (C) and (D) are examples of dragging a chair and a bathroom washbasin, which are dragging systems Y5. In the examples of Figures 3 and 4, the time span on the horizontal axis is one second.

[0027] According to the performance standards for buildings and rooms by use set by the Architectural Institute of Japan, the preferable performance level (Class 1) recommended by the Architectural Institute for apartment buildings is a noise level of 35 dB(A) or less. According to another source, internal noise from shared facilities in apartment buildings is considered to be "low noise" at a noise level of 30 dB(A), "somewhat loud" at 40 dB(A), and "quite loud" at 50 dB(A). Therefore, the actual feeling of noise in an apartment building needs to be judged based on the noise level B, not the maximum value X of the vibration acceleration 4 described above.

[0028] FIG. 5 is a diagram showing the configuration of the noise estimation device 24. The noise estimation device 24 includes a frequency analyzer 24a, a radiated sound calculator 24b, a band synthesis value calculator 24c, and a maximum level calculator 24d.

[0029] The frequency analyzer 24a obtains a time waveform 6 for each octave band by passing the acceleration data 5 through an octave band filter and then through an effective value detection circuit with a time weighting characteristic F. This frequency analysis is similar to the frequency analysis function of a general noise system. The time weighting characteristic F is a process used in general sound level meters, and is usually used when evaluating impulsive sounds such as floor impact sounds. The frequency range should be the octave band with center frequencies of 31.5 Hz, 63 Hz, 125 Hz, 250 Hz, and 500 Hz, which is the target frequency range for heavy floor impact sound insulation performance measurement according to JIS A 1418-2. Furthermore, when obtaining an octave band waveform up to 500 Hz, the sampling frequency needs to cover the upper limit frequency of the 500 Hz band (= 500 × √2 ≒ 707 Hz).

[0030] The indoor sound pressure level Lp due to radiation from a vibrating floor slab can theoretically be calculated using equation (1). Here, Lp is the indoor sound pressure level, La is the vibration acceleration level of the floor slab, ΔLa is the vibration acceleration level amplification amount (assuming ceiling effect), f is the frequency, k is the radiation coefficient, S is the radiation area, and A is the indoor equivalent sound absorption area (room surface area x indoor average sound absorption coefficient).

[0031]

number

[0032] The radiated sound calculator 24b calculates the room average sound pressure level 7 due to radiation of each octave band from the time waveform 6 of each octave band. In this calculation, the radiation efficiency k is a value determined by the physical properties of the floor and the frequency. In the example described below, a value was adopted assuming a concrete slab thickness of 200 mm, which is typical for floors in reinforced concrete apartment buildings. The radiation area S and the room equivalent sound absorption area A are set to approximate values. In the example described below, a room the size of a bedroom is assumed, with a floor area of ​​10 m 2 , room surface area 50m 2 The average indoor sound absorption coefficient was set to be the typical measurement value for a residential room.

[0033] The band synthesis value calculator 24c synthesizes the energy of the room average sound pressure levels 7 of the respective octave bands to calculate the sum of the amounts of energy 8. That is, the synthesized level is calculated by taking the sum of the amounts of energy 8 of the time waveforms of the respective octave bands. By adding an A-weighting correction value to each band during energy synthesis, a correction can be made that corresponds to human hearing, and a value equivalent to noise level B, which is an index of sound loudness, can be calculated.

[0034] The maximum level calculator 24d estimates a noise level B, which is the maximum level within the target time period, for the combined total energy amount 9.

[0035] The behavior estimation device 26 uses, for example, a discrimination program that uses a plurality of acceleration data 5 indicating the time change of vibration acceleration 4 that divides the noise behavior Y of daily sounds into a plurality of categories, and performs machine learning on some of the data as learning data and the rest as evaluation data.

[0036] In FIG. 2, the control unit 20 further includes a noise analysis device 28 for analyzing the noise level B.

[0037] FIG. 6 is a diagram showing the overall configuration of the noise analysis device 28. In this figure, a noise analysis device 28 includes an occurrence frequency calculation device 28a, an influence degree calculation device 28b, and a level-based contribution degree calculation device 28c. The occurrence frequency calculation device 28a calculates a noise occurrence frequency C, which is the relationship between a noise level B and the number of occurrences within a predetermined period of time. The influence degree calculation device 28b calculates an influence degree evaluation value D of the noise level B from the noise occurrence frequency C. The level-specific contribution calculation device 28c calculates the level-specific contribution E of the noise level B from the noise occurrence frequency C.

[0038] FIG. 7 is an explanatory diagram of the noise analysis device 28. As shown in FIG. In this example, the noise analysis device 28 targets all noises detected within a predetermined period (hereinafter, all noises). The predetermined period is, for example, one month, but may be another period (several days or several weeks).

[0039] 7A is a diagram showing the noise occurrence frequency C of all noises (hereinafter, total noise occurrence frequency C1) calculated by occurrence frequency calculation device 28a. In this figure, the horizontal axis represents the index value (dB) of noise level B, and the vertical axis represents the number of occurrences within a given period. The index value of the noise level B for one time shown in Fig. 7(A) is determined by the above-mentioned noise estimation device 24. Therefore, this figure is created by determining the noise level B from a large number of vibration acceleration waveforms, more than 3000 times.

[0040] In FIG. 7(A), the range of the index value of the noise level B that exceeds the judgment value is divided into a first level, a second level, and a third level. As mentioned above, in an apartment building, a noise level of 30 dB(A) is considered "low," 40 dB(A) is "slightly loud," and 50 dB(A) is "quite loud." In this example, the judgment value is set to 30 dB(A), and noises below the judgment value are ignored here as they are not noises that will cause dissatisfaction to the residents. In this example, the first level is in the range of 30 dB(A) or more and less than 40 dB(A), the second level is in the range of 40 dB(A) or more and less than 50 dB(A), and the third level is in the range of 50 dB(A) or more. The above levels are merely examples, and other ranges may be set.

[0041] FIG. 7B is an example of a total noise influence evaluation value D (hereinafter, total noise influence evaluation value D1) calculated by the influence calculation device 28b. The influence degree calculation device 28b calculates a total noise influence degree evaluation value D1 by adding up each occurring event that is equal to or greater than the judgment value in FIG. 7(A) after weighting it for each level. Therefore, the total noise influence evaluation value D1 indicates the degree of influence on the occupants of the noise level B of all noises exceeding the judgment value that occurred during a predetermined period.

[0042] FIG. 7C is an example of the level-specific contribution degree E of the total noise (hereinafter, total noise level-specific contribution degree E1) calculated by the level-specific contribution degree calculation device 18c. The contribution degree E1 by all noise levels is obtained by comparing the magnitude of each level obtained by weighting and adding up each event that occurs equal to or exceeds the judgment value in FIG. 7(A) for each level. Therefore, the contribution degree E1 by total noise level indicates the contribution degree of the impact on occupants of noise level B of all noise exceeding the judgment value that occurred during a specified period, divided into first level, second level, and third level.

[0043] The noise analysis device 28 can also target each noise behavior Y of the living sounds estimated by the above-mentioned behavior estimation device 26.

[0044] FIG. 8 is a diagram showing the noise occurrence frequency C of a noise behavior Y (hereinafter, behavior-specific noise occurrence frequency C2) calculated by the occurrence frequency calculation device 28a. In this figure, (A) is an example of jumping in the impact system Y1, (B) is an example of walking in slippers in the walking system Y3, (C) is an example of a vacuum cleaner in the equipment system Y4, and (D) is an example of dragging a chair in the dragging system Y5. 8, the horizontal axis represents the index value (dB) of noise level B, and the vertical axis represents the number of occurrences of noisy behavior Y within a predetermined period. The predetermined period is, for example, one month, but may be other periods (several days or several weeks). Each graph in Figure 8 was created by calculating the noise level B from the vibration acceleration waveforms of noise behavior Y occurring at least 90 times.

[0045] 8(A) to 8(D), the range of index values ​​for noise level B is divided into a first level, a second level, and a third level. This division is the same as in the case of FIG. 7(A).

[0046] Similarly, the impact evaluation value D of a noise behavior (hereinafter, behavior-specific impact evaluation value D2) can be calculated by the impact calculation device 28b, and the level-specific contribution calculation device 18c can calculate the level-specific contribution E of the behavior (hereinafter, behavior-specific level contribution E2).

[0047] In FIG. 6, the noise analysis device 28 further includes an activity-specific contribution degree calculation device 28d. The activity-specific contribution calculation device 28d calculates an activity-specific contribution E3 from the activity-specific noise generation frequency C2 and the activity-specific level-specific contribution E2, which are the noise generation frequency C and level-specific contribution E of each noise activity Y of daily sounds.

[0048] FIG. 9 is an explanatory diagram of the activity-specific contribution degree E3. The contribution level by behavior E3 shows the contribution level of each noise behavior exceeding the judgment value that occurred during a specified period to the impact on residents of noise level B, categorized by noise behavior.

[0049] The above-mentioned data storage device 22 further stores the noise level B, the total noise occurrence frequency C1, the total noise impact evaluation value D1, the contribution degree E1 by total noise level, the noise occurrence frequency C2 by activity, the contribution degree E2 by activity level, or the contribution degree E3 by activity.

[0050] The noise notification device 30 also notifies the resident of the upper floor dwelling unit 2F of the noise-generating behavior analyzed by the noise estimation device 24, and displays some or all of the data stored in the data storage device 22. The noise-generating behavior is a noise-generating behavior Y of daily life sounds, and in the above example, it is classified into impact-related Y1, falling-related Y2, walking-related Y3, equipment-related Y4, and dragging-related Y5. In addition, this notification is not limited to when noise is generated, but can be made at any time at the resident's request. In addition, residents can check the following data at any time on the noise notification device 30 (display device, smartphone, etc.). That is, it is possible to confirm the above-mentioned total noise occurrence frequency C1, activity-specific noise occurrence frequency C2, total noise impact evaluation value D1, activity-specific impact evaluation value D2, total noise level-specific contribution E1, activity level-specific contribution E2, and activity-specific contribution E3.

[0051] The life sound influence detection method according to the present invention uses the life sound influence detection system 100 described above to detect vibration acceleration 4 occurring in the floor slab 2 of the upper floor dwelling unit. In addition, the noise level B propagating to the lower floor dwelling unit is estimated from acceleration data 5 indicating the time change of vibration acceleration 4. Furthermore, noise behavior Y of everyday sounds is estimated from acceleration data 5. Furthermore, if the estimated noise level B exceeds a predetermined judgment value, the noise level B as well as the noise behavior Y of daily life sounds are notified to the resident of the upper floor residence. EXAMPLES

[0052] An example in which a time-series data analysis tool is used as a machine learning discrimination program will be described below.

[0053] (Validation data) Figure 10 shows the layout of the experimental equipment simulating the upper floor dwelling units. (A) Dwelling unit 1 is constructed using a large slab method without small beams, and (B) Dwelling unit 2 is constructed using a small beam frame structure, both of which are typical slab construction methods for reinforced concrete apartment buildings. The floor slabs of the upper floor units are 200mm thick prestressed reinforced concrete (PRC) for unit 1 and 120mm to 180mm thick reinforced concrete (RC) for unit 2. In the test, the acceleration sensors 10 were fixed at two and four locations indicated by ▼ marks in the figure. The acceleration sensors 10 were piezoelectric acceleration sensors, and the data sampling frequency was 12.8 kHz for dwelling unit 1 and 2.56 kHz for dwelling unit 2.

[0054] (Recorded acceleration data 5) Noisy behaviors Y of daily life were recorded and classified into four patterns: impact type Y1, falling type Y2, walking type Y3, and dragging type Y5. Equipment type Y4 was excluded from the study due to the small amount of data. The impact patterns of Y1 include jumping, opening and closing a door, dropping a washbasin, dropping a bucket, etc. The drop pattern Y2 includes a washbowl drop, a bucket drop, etc. The walking pattern for Y3 is slipper walking, walking, and slipper jogging. The dragging patterns of Y5 include dragging a chair, dragging a bath stool, opening and closing a sliding door, opening and closing a sliding door, etc. The length of each data is 1 second. The number of recorded cases is approximately 1,300 for impact type Y1, approximately 900 for fall type Y2, approximately 3,000 for walking type Y3, and approximately 1,500 for dragging type Y5.

[0055] (Analysis method) The analysis was carried out according to the following procedure. (1) Classify noise behavior Y of daily life sounds and record all acceleration data 5. (2) All acceleration data 5 for dwelling unit 1 and dwelling unit 2 are randomly divided into training and evaluation data in a ratio of 7:3. (3) Features are extracted from the learning acceleration data 5 according to feature extraction logic. The feature extraction logic is, for example, the logic shown in FIG. (4) The learning acceleration data 5 is input into a time series data analysis tool for machine learning, and a discrimination program that estimates the behavior of daily sounds from the acceleration data 5 is created. (5) The acceleration data 5 for evaluation is input into the discrimination program, an estimation result is obtained, and the estimation performance is evaluated. During the analysis, the data sampling frequency was unified to 2.56 kHz by downsampling.

[0056] (estimated performance) Table 1 shows the estimation results of four patterns (Y1, Y2, Y3, Y5) of living sounds based on the acceleration sensor 10 near the center of the slab.

[0057] [Table 1]

[0058] In Table 1, Precision (hereinafter, Pr) means TP / (TP+FP), where TP is the case where the predicted result is positive and the true result is also positive, and FP is the case where the predicted result is positive and the true result is negative. In addition, Recall (hereinafter, Re) means TP / (TP+FN), where FN is the case where the predicted result is negative and the true result is positive. Furthermore, F1-score means 2(Re×Pr) / (Re+Pr) and corresponds to the accuracy rate of estimation.

[0059] From the above-mentioned examples, an average F1-score of 0.74 was obtained, and it was confirmed that a general-purpose discrimination program can be constructed by learning using data from various floor construction methods. EXAMPLES

[0060] FIG. 11 is a graph showing the relationship between the sampling frequency and the F1-score. In this figure, the horizontal axis is the sampling frequency, and the vertical axis is the average F1-score. Note that this figure examines the analysis accuracy due to changes in sampling frequency under the condition that only the data of dwelling unit 1 was used for learning and evaluation.

[0061] From FIG. 11, we can see that even when the data sampling frequency is reduced to about 300 Hz by downsampling, classification can be performed with an accuracy of about 80%. For example, there is no particular problem even if the sampling frequency is lowered to 2.56 kHz. On the other hand, in the frequency analysis described above, when determining octave band waveforms up to a frequency of 500 Hz, the sampling frequency must cover at least twice the upper limit frequency of 500 Hz (=500×√2≒707 Hz). Therefore, by setting the sampling frequency of the acceleration data 5 to at least twice the upper limit frequency of 500 Hz and as low as possible, the load of data analysis can be reduced as much as possible.

[0062] As described above, the floor slab 2 of the upper floor dwelling unit is less susceptible to the effects of living noise from the dwelling unit on the floor below, but is more likely to vibrate due to living noise from the dwelling unit on the floor above. In order to utilize this vibration characteristic, in the present invention, an acceleration sensor 10 is fixed to the floor slab 2 of the upper floor dwelling unit, so that the noise level B propagating to the lower floor dwelling unit can be estimated from the acceleration data 5 generated on the floor slab 2.

[0063] In addition, the system is equipped with a control unit 20 that estimates the noise level B propagating to the dwelling unit on the floor below and the noise-generating behavior Y of daily sounds from acceleration data 5 that indicates the change over time in vibration acceleration 4, so that the resident can be notified of the noise-generating behavior (noisy behavior Y of daily sounds) along with the estimated noise level B.

[0064] Furthermore, the noise notification device 30 can issue a notification not only when noise occurs but also at any time desired by the resident. In addition, residents can check the above-mentioned total noise occurrence frequency C1, activity-specific noise occurrence frequency C2, total noise impact evaluation value D1, activity-specific impact evaluation value D2, total noise level-specific contribution E1, activity level-specific contribution E2, and activity-specific contribution E3 at any time using the noise notification device 30. Therefore, the occupants can look back at any time on the noise-generating behaviors that they have been notified of, and evaluate the extent of their impact.

[0065] The scope of the present invention is not limited to the above-described embodiment, but is indicated by the claims, and further includes all modifications within the meaning and scope equivalent to the claims. [Explanation of symbols]

[0066] a,b threshold, B noise level, C noise occurrence frequency, C1 total noise occurrence frequency, C2 noise occurrence frequency by behavior, D impact evaluation value, D1 total noise impact evaluation value, D2 impact evaluation value by behavior, E contribution by level, E1 contribution by total noise level, E2 contribution by behavior level, E3 contribution by behavior, X maximum value, Y noise behavior, Y1 impact / fall system, Y2 walking system, Y3 equipment system, Y4 dragging system, 1 apartment building, 2 floor slab, 4 vibration acceleration, 5 acceleration data, 6 octave band time waveform, 7 indoor average sound pressure level, 8 sum of energy amount, 10 acceleration sensor, 22 data storage device, 24 noise estimation device, 26 behavior estimation device, 28 noise analysis device, 28a occurrence frequency calculation device, 28b impact calculation device, 28c contribution by level calculation device, 28d Activity-specific contribution calculation device, 30 Noise notification device, 100 Life sound impact detection system

Claims

1. A living sound influence detection system for detecting an influence of living sounds of the upper floor dwelling unit on the lower floor dwelling unit in an apartment building having an upper floor dwelling unit and a lower floor dwelling unit, an acceleration sensor fixed to a floor slab of the upper floor dwelling unit and detecting a vibration acceleration occurring in the floor slab; a control unit that estimates the noise level propagating to the lower floor dwelling unit and the noise behavior of the living sound from acceleration data indicating a time change of the vibration acceleration; A noise notification device that notifies the resident of the upper floor of the noise level and the noisy behavior of the living sounds when the estimated noise level exceeds a predetermined judgment value.

2. The control unit a noise estimation device that estimates the noise level from the acceleration data; The life sound influence detection system according to claim 1 , further comprising: an activity estimation device that estimates the noisy activity from the acceleration data.

3. The noise estimation device comprises: a frequency analyzer for passing the acceleration data through octave band filters having center frequencies of 31.5 Hz, 63 Hz, 125 Hz, 250 Hz, and 500 Hz to obtain a time waveform of each octave band; a sound radiation calculator for calculating an indoor average sound pressure level due to radiation of each of the octave bands from the time waveform; a band synthesis value calculator for synthesizing the energy of the room average sound pressure levels of the respective octave bands to calculate a total energy amount; 3. The life sound influence detection system according to claim 2, further comprising: a maximum level calculator that estimates the noise level, which is the maximum level within a target time period, by taking into account a time weighting characteristic of the sum of the synthesized energy amounts.

4. The noise behavior of the living sounds is an impact type, a drop type, a walking type, a machine type, and a drag type, The living sound influence detection system of claim 2, wherein the activity estimation device uses a discrimination program that is machine-learned using a portion of the multiple acceleration data classified as the noisy activities as learning data and the rest as evaluation data.

5. The control unit includes a noise analysis device that analyzes the noise level, the noise analysis device comprising: an occurrence frequency calculation device for calculating a noise occurrence frequency, which is a relationship between the noise level and the number of occurrences within a predetermined period of time; an influence calculation device for calculating an influence evaluation value of the noise level from the noise occurrence frequency; The life sound influence detection system according to claim 1 , further comprising: a level-based contribution degree calculation device that calculates a level-based contribution degree of the noise level from the noise occurrence frequency.

6. The noise analysis device is provided with an activity-specific contribution calculation device that calculates an activity-specific contribution from the noise occurrence frequency and the level-specific contribution of each of the noise activities of daily sounds, which are the activity-specific noise occurrence frequency and the activity-specific contribution. The daily sound influence detection system of claim 5.

7. the control unit comprises a data storage device for storing the acceleration data; 2. The life sound influence detection system according to claim 1, wherein the data storage device stores the estimated noise level and the noise behavior of the life sound together with the time of occurrence thereof.

8. the data storage device stores the noise level, the noise occurrence frequency, the impact evaluation value, and the level-specific contribution degree of all noises, and the noise occurrence frequency, the level-specific contribution degree, or the activity-specific contribution degree of each of the noisy activities; The life sound influence detection system according to claim 7 , wherein the noise notification device displays a part or all of the data stored in the data storage device.

9. Using the life sound influence detection system according to claim 1, Detecting the vibration acceleration occurring in the floor slab; estimating the noise level propagating to the lower floor dwelling unit from the acceleration data; estimating the noise behavior of the living sound from the acceleration data; A method for detecting the influence of living sounds, which notifies the resident of the upper floor dwelling unit of the noise level and the noisy behavior of the living sounds when the estimated noise level exceeds the judgment value.

10. The noise behavior of the living sounds is an impact type, a drop type, a walking type, a machine type, and a drag type, The living sound influence detection method according to claim 9, further comprising using a discrimination program that is machine-learned using a portion of the multiple acceleration data classified as the noise behavior as learning data and the remainder as evaluation data.

11. 10. The living sound influence detection method according to claim 9, wherein a sampling frequency of the acceleration data is set in a range higher than twice an upper limit frequency of an octave band of 500 Hz and lower than a target frequency of the acceleration sensor.

Images