Environmental measurement device

Abstract

The environment measurement device is provided with an estimation unit (44) for estimating the temperature distribution or the air flow distribution of the target space (S), and the estimation unit (44) estimates the temperature or the air flow of each region of the first virtual grid (M1) using a predetermined prediction method and constraint conditions added to the prediction method. The constraint condition includes at least one of a value of temperature or airflow in a second imaginary grid (M2) having a smaller number of regions than the first imaginary grid (M1), an inflow / outflow amount of air in at least one region of the first imaginary grid (M1), and at least one temperature measurement of the first imaginary grid (M1).

Description

Technical Field

[0001] The present disclosure relates to an environment measuring device. Background Art

[0002] The following method is known as a method for measuring the temperature distribution or airflow distribution of an object space: a virtual grid is generated by dividing the object space into multiple areas, and the temperature and airflow of the object space are calculated based on the propagation time and propagation distance of the detection sound waves passing through each area of ​​the virtual grid (for example, Patent Document 1).

[0003] Prior art literature

[0004] Patent Literature

[0005] Patent Document 1: International Publication No. 2020 / 110393 Summary of the invention

[0006] -Technical problem to be solved by the invention-

[0007] When the number of unknowns of the temperature or airflow of the imaginary grid is greater than the number of equations based on the propagation time and propagation distance of the detection sound waves passing through each area of ​​the imaginary grid, the temperature or airflow of the imaginary grid can be predicted by a prescribed method, but the accuracy of the predicted value is insufficient. Therefore, the measurement accuracy of the temperature distribution or airflow distribution of the object space is improved by increasing the number of sound wave transceivers to increase the number of equations based on the propagation time and propagation distance of the detection sound waves passing through each area of ​​the imaginary grid. However, if the number of sound wave transceivers is increased, in addition to the increase in cost, the number of sound wave transceivers may sometimes be restricted by the appearance design of the object space and the like, making it impossible to increase the number of sound wave transceivers.

[0008] An object of the present disclosure is to improve the prediction accuracy of temperature distribution or velocity distribution in an object space.

[0009] -Technical solutions to solve technical problems-

[0010] The first aspect is an environment measurement device, which includes a sound wave transceiver 20, a setting unit 42 and an estimation unit 44.

[0011] The sound wave transceiver 20 transmits a detection sound wave toward the target space S and receives the detection sound wave.

[0012] The setting unit 42 sets a first virtual grid M1, wherein the first virtual grid M1 is formed by dividing the object space S into a plurality of regions and the plurality of regions are divided.

[0013] The estimation unit 44 estimates the temperature distribution or the airflow distribution of the target space S based on the propagation time of the detection sound wave through a predetermined area in the first virtual mesh M1.

[0014] The estimation unit 44 estimates the temperature or airflow of each area of ​​the first virtual grid M1 using a predetermined estimation method and constraints added to the estimation method.

[0015] The constraint conditions include at least one of the following three: the temperature or airflow value in the second imaginary grid M2 having fewer regions than the first imaginary grid M1, the inflow and outflow of air in at least one of the regions of the first imaginary grid M1, and at least one temperature measurement value of the first imaginary grid M1.

[0016] In the first aspect, by using the measured value of the temperature or airflow in the second imaginary grid M2, the air inflow and outflow at the boundary of the area of ​​the first imaginary grid M1, or the temperature of at least one area of ​​the first imaginary grid M1 as constraints of the prediction method, the measurement accuracy of the temperature distribution or velocity distribution of the entire object space S can be improved.

[0017] The second aspect is based on the first aspect.

[0018] The inference unit 44 uses a generalized inverse matrix as the prediction method, and the prediction method calculates the temperature or airflow of the air in each area of ​​the first imaginary grid M1 based on the propagation time of the detection sound wave on the propagation path of the object space S, the path length within the first imaginary grid M1 through which the propagation path passes, and the propagation speed within the first imaginary grid M1.

[0019] In the second aspect, the temperature or air flow in each area can be easily obtained by using the generalized inverse matrix.

[0020] The third aspect is based on the first or second aspect.

[0021] The estimation unit 44 generates and outputs a first prediction model of the temperature or airflow of the second virtual grid M2 based on a learning result obtained by machine learning, wherein data that associates the propagation time of the propagation path of the detection sound wave in the object space S with the temperature distribution or the airflow distribution is used as training data,

[0022] The estimation unit 44 uses the output value of the first prediction model as the constraint condition, and based on the learning result after adding the constraint condition to the training data and performing machine learning, produces a second prediction model that outputs the temperature or airflow of the first virtual grid.

[0023] The estimation unit 44 estimates the temperature distribution or the airflow distribution of the target space S by using the second prediction model using the propagation time of the propagation path in the target space S as input data.

[0024] In the third aspect, the output of the first prediction model generated by machine learning on the second virtual grid is used as a constraint condition. When machine learning is performed on the basis of the first virtual grid, the second prediction model after adding the constraint condition to the training data and performing machine learning can improve the accuracy of the prediction result compared to the prediction model without adding the constraint condition.

[0025] According to the fourth aspect, based on any one of the first to third aspects,

[0026] The number of regions of the second virtual grid is the largest number of regions for which the temperature or the airflow can be uniquely determined without using the prediction method.

[0027] In the fourth aspect, for example, compared with the second virtual grid M2 having 1 (1×1) region, the second virtual grid M2 having 9 (3×3) regions has more regions and the amount of information is correspondingly larger. Thus, by using the grid with the largest number of regions in the second virtual grid M2 as the second virtual grid, the prediction accuracy can be improved.

[0028] The fifth aspect is based on any one of the first to fourth aspects,

[0029] The measured value of the temperature is a measured value of the temperature sensor 30 arranged in the object space S.

[0030] In the fifth aspect, by adding the temperature value of the area where the temperature sensor is installed as a constraint condition, the prediction accuracy can be improved.

[0031] A sixth aspect is an environment measuring device.

[0032] The environment measurement device includes a sound wave transceiver 20 and an estimation unit 44. The sound wave transceiver 20 transmits a detection sound wave toward the target space S and receives the detection sound wave.

[0033] The estimation unit 44 estimates the temperature distribution or the airflow distribution of the target space S based on the propagation time of the detection sound wave through the first virtual grid M1, wherein the first virtual grid M1 is formed by dividing the target space S into a plurality of regions.

[0034] The estimation unit 44 generates and outputs a prediction model of the temperature or airflow of the first virtual grid M1 based on a learning result obtained by machine learning, wherein data associating the propagation time of the detection sound wave on the propagation path of the object space S with the temperature distribution or the airflow distribution is used as training data.

[0035] The estimation unit 44 estimates the temperature distribution or the airflow distribution of the target space S by using the prediction model using the propagation time of the propagation path in the target space S as input data.

[0036] In the sixth aspect, even without adding constraints, it is possible to predict the temperature distribution or airflow distribution of the target space through machine learning. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 is a block diagram showing the structure of an environment measuring device according to an embodiment;

[0038] Figure 2 is a diagram of the object space in which the first virtual grid is set, viewed from above;

[0039] Figure 3 is a diagram of the object space in which the second virtual grid is set, viewed from above;

[0040] Figure 4 It is a diagram for explaining the speed of airflow in the x-direction and the y-direction within the region;

[0041] Figure 5 is a diagram for explaining the fourth constraint condition;

[0042] Figure 6 is a flowchart showing a method of estimating temperature distribution of an object space;

[0043] Figure 7 is a flowchart showing a method for estimating airflow distribution in a target space;

[0044] Figure 8 is a block diagram showing a configuration of an environment measuring device according to a modification example;

[0045] Fig. 9 It is a diagram for explaining a method of estimating temperature distribution according to a modification example. DETAILED DESCRIPTION

[0046] The following describes the embodiments of the present invention with reference to the accompanying drawings. It should be noted that the following embodiments are essentially only preferred examples and are not intended to limit the present invention, its application objects or its uses. In addition, the various structures of the embodiments, modifications, other examples, etc. described below can be combined or partially replaced within the scope of the present invention.

[0047] (1) Environmental measurement device

[0048] The environmental measuring device 10 of the present embodiment is a device for measuring the temperature distribution and airflow distribution of the object space S. The object space S is, for example, an office or conference room in a building, a guest room or banquet hall in a hotel, a room in a hospital, a storage room in a warehouse, etc., or a room in an apartment or a single-family house. In the following description, the object space S is sometimes referred to as an indoor space S. The indoor space S of the present embodiment is a quadrilateral space having four walls when viewed from above. In the indoor space S, an air conditioning device for air conditioning the indoor space S and a ventilation device for ventilating the indoor space S may also be provided.

[0049] like Figure 1 and Figure 2 As shown, the environment measuring device 10 of this embodiment includes a sound wave transceiver 20, a temperature detection unit 30, a control unit 40, and a notification unit 50. The environment measuring device 10 estimates the temperature distribution and airflow distribution of the indoor space S based on the propagation time and propagation distance of the detection sound wave sent to the indoor space S.

[0050] (2) Sound wave transceiver

[0051] The sound wave transceiver 20 has a transmitter for transmitting detection sound waves and a receiver for receiving detection sound waves. Four sound wave transceivers 20 are arranged in the indoor space S. Specifically, one sound wave transceiver 20 is provided on each of the four walls of the indoor space S (see FIG. Figure 2 and Figure 3 ).

[0052] The transmitting part and receiving part of the sound wave transceiver 20 of the present embodiment are formed as one body. The position where the sound wave transceiver 20 transmits the sound wave for detection is the same as the position where the sound wave for detection is received. The four sound wave transceivers 20 transmit and receive the sound wave for detection from each other. In other words, the sound wave for detection transmitted by one sound wave transceiver 20 will be received by the other three sound wave transceivers 20.

[0053] The path along which the detection sound wave propagates from one sound wave transceiver 20 to another sound wave transceiver 20 is referred to as the propagation path. In this embodiment, six sound wave propagation paths (L1, L2, ..., L6) are formed in the indoor space S (see Figure 2). The length D (D1, D2, ..., D6) of each propagation path is the straight-line distance between the sound wave transmitting and receiving units 20.

[0054] The time from when one sound wave transceiver 20 transmits a detection sound wave to when another sound wave transceiver 20 receives the detection sound wave is referred to as sound wave propagation time (ToF: Time of Flight). The sound wave propagation time is the time it takes for a detection sound wave to pass through a sound wave propagation path.

[0055] Since the four sound wave transceivers 20 transmit and receive detection sound waves to each other, each sound wave propagation path has two sound wave propagation times (travel time (t + ) and return time (t － )). ToF is sometimes referred to as the travel time (t + ) and return time (t － In this embodiment, six ToFs (ToF1, ToF2, ..., ToF6) correspond to propagation paths (L1, L2, ..., L6), respectively.

[0056] (3) Temperature detection unit

[0057] The temperature detection unit 30 includes two temperature sensors. The two temperature sensors are a first temperature sensor 30a and a second temperature sensor 30b. The first temperature sensor 30a and the second temperature sensor 30b are provided at different positions in the indoor space S. Each temperature sensor 30 transmits a signal indicating a detected temperature value to the control unit.

[0058] (4) Control unit

[0059] The control unit 40 includes a microcomputer mounted on a control substrate and a storage device (specifically, a semiconductor memory) storing software for operating the microcomputer. The control unit 40 is connected to the temperature detection unit 30 and the notification unit 50 via a communication line. The control unit 40 includes a measuring unit 41, a setting unit 42, a storage unit 43, and an estimation unit 44.

[0060] The measuring unit 41 measures the propagation time ToF of each propagation path ( L1 , L2 , . . . , L6 ) from when one sound wave transceiver 20 transmits a detection sound wave to when the detection sound wave is received by another sound wave transceiver 20 .

[0061] like Figure 2 and Figure 3As shown, the setting unit 34 sets the first virtual grid M1 and the second virtual grid M2. The virtual grid M is formed by dividing the indoor space S into a plurality of regions when observed from above and the plurality of regions obtained by the division. Specifically, the virtual grid is obtained by dividing the indoor space S into a plurality of lateral parts and a plurality of longitudinal parts. The first virtual grid M1 of the present embodiment has sixteen regions (4 longitudinal × 4 lateral). The number of regions of the second virtual grid M2 is less than that of the first virtual grid M1. The second virtual grid M2 of the present embodiment has one region (1 longitudinal × 1 lateral). It should be noted that, hereinafter, when observing the indoor space S from above, the lateral direction of the indoor space S is sometimes referred to as the x direction, and the longitudinal direction is sometimes referred to as the y direction.

[0062] The storage unit 33 stores the sound wave propagation path, the length of the sound wave propagation path, the sound wave propagation time, an estimation model described later, and the like.

[0063] The estimation unit 44 estimates the temperature distribution and the airflow distribution of the indoor space S based on the propagation time of the detection sound wave passing through the predetermined area in the first virtual mesh M1. The estimation unit 44 will be described in detail later.

[0064] (5) Notification Department

[0065] The notification unit 50 outputs the estimated results of the temperature distribution and the airflow distribution of the indoor space S. The notification unit 50 is, for example, a display. The user can confirm the estimated results of the temperature distribution and the airflow distribution of the indoor space S displayed on the display.

[0066] (6) Prediction of temperature distribution and airflow distribution

[0067] The estimation unit 44 estimates each of the areas A1 to A2 of the first virtual mesh M1 using a predetermined estimation method and constraints added to the estimation method. 16 The prediction method is based on the sound wave propagation time of each sound wave propagation path in the indoor space S, the path length in each area passed by each sound wave propagation path, and the sound wave propagation speed in the area, to calculate the temperature and airflow of the air in each area. It should be noted that the temperature must be predicted in advance when calculating the airflow. In this embodiment, the generalized inverse matrix is ​​used as the prediction method. The following is a specific description of the inference of the temperature distribution of the indoor space S performed by the environmental measurement device 10 of this embodiment.

[0068] The propagation speed C of the sound wave is expressed by the following relationship using a predetermined coefficient α and temperature T, with v0 being the speed of sound at 0°C (331.5 m / s).

[0069] [Mathematical formula 1]

[0070] v n =v0+α+×t n…Formula (1)

[0071] The round trip sound wave propagation time (ToF) of each detection sound wave is expressed by the following relationship using the length D of the sound wave transmission path and the wind speed v. Figure 4 As shown, v x cosθ represents the velocity of the airflow in the x direction, v y sinθ represents the velocity of the airflow in the y direction.

[0072] [Mathematical formula 2]

[0073]

[0074] [Mathematical formula 3]

[0075]

[0076] (7) Temperature distribution estimation method

[0077] The temperature is calculated based on the sum of the round trip sound wave propagation time. Specifically, the following relationship is established based on equations (1) to (3).

[0078] [Formula 4]

[0079]

[0080] Based on formula (4), we can find the 16 Specifically, for the six detection sound waves passing through the sound wave propagation paths (L1 to L6), in the six ToFs (ToF1 to ToF6), each area A1 to A 16 The sound wave propagation distance D m，n (m=1, 2, ..., 6, n=1, 2, ..., 16) and the speed of sound v n (n=1, 2, ..., 16), the following determinant (5) holds.

[0081] [Formula 5]

[0082]

[0083] (7-1) Constraints

[0084] In this embodiment, when the first condition is satisfied, the following two constraints are added to equation (5), and the generalized inverse matrix is ​​used to find the 16 When the number of regions of the first virtual grid is set to P and the number of sound wave transceivers 20 provided in the indoor space S is set to Q, the first condition is established when the relationship of the following formula (6) is satisfied.

[0085] [Mathematical formula 6]

[0086]

[0087] (7-1-1) The first constraint

[0088] The first constraint condition is the temperature value in the second virtual grid M2. The temperature T of the second virtual grid M2 can be obtained based on equation (4): AVG. Temperature T AVG. represents the average temperature of the indoor space S. In other words, the temperature T AVG. represents the average temperature of the sixteen regions of the first imaginary grid. AVG. The specified expression is added to the determinant (5) as the first constraint.

[0089] (7-1-2) Second constraint

[0090] The second constraint condition is the measurement value of the temperature sensor 30 arranged in the indoor space S. The second constraint condition is the temperature of the grid where the temperature sensor 30 is arranged. In this embodiment, since there are two temperature sensors, the second constraint condition is the temperature of the area where the first temperature sensor 30a is arranged and the temperature of the area where the second temperature sensor 30b is arranged.

[0091] For example, the first temperature sensor 30a is arranged in area A4, and the second temperature sensor 30b is arranged in area A 16 (Refer to Figure 2 ). The temperature detected by the first temperature sensor 30a is 20°C, and the temperature detected by the second temperature sensor 30b is 21°C. At this time, the predetermined calculation formula for making the temperature of area A4 20°C and the predetermined calculation formula for making area A 16 The prescribed operation expression that the temperature is 21°C is added to the determinant (5) as the second constraint condition.

[0092] In this way, the estimation unit 44 adds the first constraint condition and the second constraint condition to the determinant (5), and then uses the generalized inverse matrix to find the region A1 to A2. 16 The estimation unit 44 estimates the temperature distribution of the indoor space S based on this.

[0093] (8) Method for estimating air flow distribution

[0094] The temperature of each area is calculated in advance using equation (5). Based on the difference between the sound wave propagation time of the outgoing path and the sound wave propagation time of the return path, a calculation formula for calculating the airflow can be obtained. Specifically, the following relationship is established based on equations (1) to (3).

[0095] [Formula 7]

[0096]

[0097] Based on formula (7), we can find the 16 Specifically, for the six detection sound waves passing through the propagation paths (L1 to L6), in the six ToFs (ToF1 to ToF6), each area A1 to A 16 The propagation distance D m，n cosθ、D m，n sinθ(m=1、2、……、6、n=1、2、……、16) and the speed of sound v xn 、v yn (m=1, 2, ..., 16), the following determinant (8) holds.

[0098] [Mathematical formula 8]

[0099]

[0100] (8-1) Constraints

[0101] In this embodiment, when the second condition is satisfied, the following two constraints are added to equation (8), and the generalized inverse matrix is ​​used to find the 16 When the number of meshes of the first virtual mesh is set to P and the number of sound wave transceivers 20 installed in the indoor space S is set to Q, the second condition is established when the relationship of the following formula (9) is satisfied.

[0102] [Mathematical formula 9]

[0103]

[0104] (8-1-1) The third constraint

[0105] The third constraint condition is the value of the wind speed in the x direction and the value of the wind speed in the y direction in the second virtual grid M2. Based on equation (7), the wind speed v in the x direction of the second virtual grid M2 can be calculated: AVG.x and the wind speed v in the y direction AVG.y . Wind speed v AVG.x and wind speed v AVG.y They represent the average wind speed in the x direction and the average wind speed in the y direction of the indoor space S. In other words, the wind speed v AVG.x The wind speed v represents the average value of the wind speed in the x direction of the sixteen regions of the first imaginary grid M1. AVG.y The wind speed v is the average value of the wind speed in the y direction of the sixteen regions of the first virtual grid M1. AVG.x and wind speed v AVG.yThe specified expression is added to the determinant (8) as the third constraint.

[0106] (8-1-2) The fourth constraint

[0107] The fourth constraint condition is the amount of air flowing in and out of at least one area in the first imaginary grid M1. Specifically, the amount of air flowing into an area in the imaginary grid M is equal to the amount of air flowing out of an area in the imaginary grid M. For example, Figure 5 As shown in the figure, when the wind speed in the x direction of the central area R(i, j) is set to v(i, j) x , the wind speed in the y direction is v(i, j) y When , the air flowing into the central region R(i, j) from the left and right directions and the up and down directions is equal to the air flowing out from the central region R(i, j) to the left and right directions and the up and down directions, so the relationship in equation (10) holds.

[0108] [Formula 10]

[0109] 0 = v (1-1, j) x dy-v(i+1,j) x dy+v(i,j-1) y dx-v(i, j+1) y dx…Formula (10)

[0110] In this embodiment, the regions corresponding to the central region R(i, j) are regions A6, A7, and A 10 , A 11 In this embodiment, at least one of the four regions is selected. For the selected grid, equation (10) is added as the fourth constraint to the determinant (8).

[0111] In this way, the estimation unit 44 adds the third constraint condition and the fourth constraint condition to the determinant (8), and then uses the generalized inverse matrix to find the region A1 to A 16 The estimation unit 44 estimates the airflow distribution in the indoor space S based on this.

[0112] (9) Flow of estimating temperature distribution and airflow distribution in indoor space

[0113] Next, use Figure 6 and Figure 7 The flow of estimating the temperature distribution and airflow distribution of the indoor space S by the control unit 40 will be described. The arrangement of the sound wave transceiver 20 and the temperature sensor 30 is determined in advance and arranged. In order to estimate the airflow distribution, the temperature distribution is estimated in advance.

[0114] (9-1) Flow of the temperature distribution estimation method

[0115] In step S11 , the control unit 40 sets a first virtual mesh M1 for the indoor space S. In the present embodiment, when the indoor space S is viewed from above, the indoor space S is divided into sixteen (4×4) areas.

[0116] In step S12, the control unit 40 generates a 16 The determinant of the first virtual grid M1 of the temperature (see formula (5)).

[0117] In step S13, the control unit 40 determines whether the first condition is satisfied. If it is determined that the first condition is satisfied ("Yes" in step S13), step S14 is executed. If it is determined that the first condition is not satisfied ("No" in step S13), step S16 is executed.

[0118] In step S14, the control unit 40 sets a first constraint condition. Specifically, a second virtual grid M2 is set for the indoor space S. In the present embodiment, when the indoor space is viewed from above, the second virtual grid M2 is a (1×1) area. The temperature of the second virtual grid M2 is used as the first constraint condition. The set first constraint condition is added to the determinant of step S12.

[0119] In step S15, the control unit 40 sets a second constraint condition. Specifically, the temperature of the region where the first temperature sensor 30a and the second temperature sensor 30b are arranged is used as the second constraint condition. The set second constraint condition is added to the determinant of step S12.

[0120] In step S16 , the control unit 40 calculates the temperature of each region by calculating the determinant using the generalized inverse matrix.

[0121] In step S17 , the control unit 40 outputs the temperature distribution of the indoor space S to the display based on the temperature of each mesh calculated in step S16 .

[0122] (9-2) Flow of the method for estimating airflow distribution

[0123] Step S21 is the same as the above-mentioned step S11, and therefore the description thereof is omitted.

[0124] In step S22, the control unit 40 generates a 16 The determinant of the hypothetical grid of wind speed (see formula (8)).

[0125] In step S23, the control unit 40 determines whether the second condition is satisfied. If it is determined that the second condition is satisfied ("Yes" in step S23), step S24 is executed. If it is determined that the second condition is not satisfied ("No" in step S23), step S26 is executed.

[0126] In step S24, the control unit 40 sets a third constraint condition. Specifically, a 1×1 second virtual grid M2 is set for the indoor space S. The wind speed in the x direction and the wind speed in the y direction of the second virtual grid M2 are used as the third constraint condition. The set third constraint condition is added to the determinant of step S22.

[0127] In step S25, the control unit 40 sets a fourth constraint condition. Specifically, the fourth constraint condition is that the amount of air flowing into the area selected in the first virtual grid M1 is equal to the amount of air flowing out of the area selected in the first virtual grid M1 (see equation (10)). The set fourth constraint condition is added to the determinant of step S22.

[0128] In step S26 , the control unit 40 calculates the wind speed in the x direction and the wind speed in the y direction of each mesh by calculating the determinant using the generalized inverse matrix.

[0129] In step S27 , the control unit 40 outputs the airflow distribution of the indoor space S to the display based on the airflow of each mesh calculated in step S26 .

[0130] (10) Features

[0131] (10-1) Feature 1

[0132] The estimation unit 44 of the environment measurement device 10 of the present embodiment estimates the temperature or airflow of each area of ​​the first virtual grid M1 using a predetermined estimation method and a constraint condition added to the estimation method. The constraint condition includes at least one of the following three: the value of the temperature or airflow in the second virtual grid M2 having fewer areas than the first virtual grid M1, the inflow and outflow of air in at least one area of ​​the first virtual grid M1, and at least one temperature measurement value of the first virtual grid M1.

[0133] In the present embodiment, by adding constraint conditions to a predetermined method, the measurement accuracy of the temperature distribution or the airflow distribution in the entire target space S can be improved.

[0134] (10-2) Feature 2

[0135] In the environment measuring device 10 of the present embodiment, a generalized inverse matrix is ​​used as a predetermined prediction method. By using a generalized inverse matrix, even when the number of unknowns of the temperature or airflow of the virtual grid is greater than the number of equations based on the propagation time and propagation distance of the detection sound wave passing through each area of ​​the virtual grid, the temperature and airflow of each area can be obtained.

[0136] (10-3) Feature 3

[0137] In the environment measuring device 10 of this embodiment, the temperature measurement value that is the constraint condition is the measurement value of the temperature sensor 30 arranged in the indoor space S. The actual measurement value of the temperature of at least one area of ​​the first virtual mesh M1 can be used as the constraint condition.

[0138] (11) Modification

[0139] Next, a description will be given of a configuration of an environment measurement device 10 according to a modified example that is different from the environment measurement device 10 of the above-described embodiment.

[0140] like Figure 8 As shown, the environment measurement device 10 includes a learning unit 45, a first prediction model U1, and a second prediction model U2. The learning unit 45 performs learning by machine learning using data that associates the sound wave propagation time of each sound wave propagation path in the indoor space S with the temperature distribution and the airflow distribution as training data.

[0141] use Fig. 9 The learning of temperature distribution is described in detail. The learning unit 45 performs machine learning by simulating the propagation time ToF and the speed of sound v corresponding to multiple temperature distribution patterns (pattern 1 to pattern N) of the indoor space S as the first training data. The speed of sound v here is the speed of sound in the second imaginary grid (1×1). For one temperature distribution pattern, six ToFs (ToF1 to ToF6) and the speed of sound v corresponding to one temperature distribution pattern are associated and learned N times. By using such learning data to enable the learning unit 45 to perform "supervised learning", a first prediction model U1 is generated as a learning result.

[0142] The learning unit 45 performs machine learning on the learning data obtained by adding the output value of the first prediction model U1 to the above-mentioned first training data as the second training data. The learning unit 45 performs machine learning on the output value of the first prediction model U1, the propagation time ToF, and the speed of sound v corresponding to the multiple temperature distribution patterns (pattern 1 to pattern N) of the indoor space S as training data. The speed of sound v here is the speed of sound in the first imaginary grid (4×4). For one temperature distribution pattern, six ToFs (ToF1~ToF6) and the output value of the first prediction model added as the seventh data and the speed of sound v (v1~v16 ) are associated and N learnings are performed. By using such learning data, the learning unit 45 performs "supervised learning" to generate a second prediction model U2 as a learning result.

[0143] The estimation unit 44 estimates the temperature distribution of the indoor space S by inputting the actual six ToFs as input data into the second prediction model. The constraint condition of Modification 1 is the output value of the first prediction model U1. By adding the output value as a constraint condition and as training data, the prediction accuracy of the temperature distribution of the indoor space S can be improved. Since the estimation method of the airflow distribution of Modification 1 is the same as the estimation method of the temperature distribution described above, the description is omitted.

[0144] (12) Other Implementation Methods

[0145] The above-mentioned embodiment may also adopt the following structure.

[0146] In the above-mentioned embodiments and variations, the second virtual grid M2 may not be a virtual grid having one region (1×1). The number of regions of the second virtual grid M2 can be the largest number of regions that can uniquely determine the temperature or airflow without using a predetermined prediction method. For example, the second virtual grid M2 can be the largest number of regions in the entire region through which the propagation path of the detection sound wave passes.

[0147] The above-mentioned modification may not use constraint conditions. That is, the estimation unit 44 generates and outputs a prediction model of the temperature or airflow of the first virtual grid based on the learning result obtained by machine learning, wherein in the machine learning, data that associates the propagation time of the detection sound wave on the propagation path of the indoor space S with the temperature distribution and the airflow distribution is used as training data. The estimation unit 44 can estimate the temperature distribution and the airflow distribution of the indoor space S by using the propagation time on the propagation path of the indoor space S as input data and using the prediction model.

[0148] Specifically, the learning unit 45 uses the propagation time ToF and the sound speed v corresponding to the multiple temperature distribution patterns (pattern 1 to pattern N) of the indoor space S as the first training data for machine learning. The sound speed v here is the sound speed in the first virtual grid (4×4). For one temperature distribution pattern, six ToFs (ToF1 to ToF6) and the sound speed v (v1 to v 16 ) are associated and N learnings are performed. By using such learning data, the learning unit 45 performs "supervised learning", and a prediction model is generated as a learning result.

[0149] In the above embodiment, there may be one constraint condition. When estimating the temperature distribution, the constraint condition may be the first constraint condition or the second constraint condition. When estimating the airflow distribution, the constraint condition may be the third constraint condition or the fourth constraint condition.

[0150] As a constraint condition when estimating the airflow distribution, it is also possible to add it when the following three-dimensional airflow distribution conditions are met.

[0151] [Mathematical formula 11]

[0152]

[0153] In the above-mentioned variation, the constraint condition is not limited to the output value of the 1×1 virtual grid. For example, the learning unit 45 performs machine learning on the 2×2 virtual grid by adding the output value of the 1×1 virtual grid as training data. The output value of the inference model of the 2×2 virtual grid generated based on the learning result can also be used as a constraint condition. In this way, in addition to the output value of the 1×1 virtual grid, the output value of the 2×2 virtual grid is also used as a constraint condition. In this way, for multiple low-order grids with fewer regions than the first virtual grid, the inference accuracy of the temperature distribution and airflow distribution in higher-order virtual grids is improved by performing machine learning in stages in the order of 1×1 virtual grid, 2×2 virtual grid, ... and generating an inference model.

[0154] The measuring unit 41 may not be disposed in the control unit 40 . The measuring unit 41 may be disposed in the sound wave transceiver 20 . The measuring unit 41 may also be a device independent of other devices.

[0155] The environment measurement device 10 may be a device that estimates the temperature distribution or airflow distribution in the indoor space S.

[0156] The control unit 40 of the above-mentioned modification example may have the first prediction model U1 and the second prediction model U2 in advance, or may generate the first prediction model U1 and the second prediction model U2 each time through learning.

[0157] The above embodiments and variations are described, but it should be understood that various changes can be made to their forms and specific circumstances without departing from the subject matter and scope of the claims. In addition, as long as the functions of the objects of the present disclosure are not destroyed, the above embodiments and variations can also be appropriately combined and replaced. The above-mentioned "first", "second" ... these words are only used to distinguish the sentences containing the above words, and are not intended to limit the number and order of the sentences.

[0158] - Industrial Applicability -

[0159] In summary, the present disclosure is useful for an environment measuring device.

[0160] - Explanation of symbols -

[0161] 10 Environmental measurement device

[0162] 20 Sound wave transceiver

[0163] 42 Setting section

[0164] 44 Speculation Department

[0165] M1 First imaginary grid

[0166] M2 Second imaginary grid

[0167] S Object space (indoor space)

[0168] U1 First prediction model

[0169] U2 Second prediction model

Claims

1. An environmental measurement device, characterized in that: The environment measurement device includes a sound wave transceiver (20), a setting unit (42) and an estimation unit (44). The sound wave transceiver (20) transmits a detection sound wave toward the target space (S) and receives the detection sound wave. The setting unit (42) sets a first virtual grid (M1), wherein the first virtual grid (M1) is formed by dividing the object space (S) into a plurality of regions and the plurality of regions are divided. The estimation unit (44) estimates the temperature distribution or airflow distribution of the target space (S) based on the propagation time of the detection sound wave through a predetermined area in the first virtual grid (M1). The estimation unit (44) estimates the temperature or airflow of each area of ​​the first virtual grid (M1) using a predetermined estimation method and constraints added to the estimation method. The constraint condition includes at least one of the following three: the value of temperature or airflow in a second imaginary grid (M2) having fewer regions than the first imaginary grid (M1), the inflow and outflow of air in at least one of the regions of the first imaginary grid (M1), and at least one temperature measurement value of the first imaginary grid (M1).

2. The environmental measurement device according to claim 1, characterized in that: The inference unit (44) uses a generalized inverse matrix as the prediction method, and the prediction method calculates the temperature or airflow of the air in each area of ​​the first imaginary grid (M1) based on the propagation time of the detection sound wave on the propagation path of the object space (S), the path length within the first imaginary grid (M1) through which the propagation path passes, and the propagation speed within the first imaginary grid (M1).

3. The environmental measurement device according to claim 1 or 2, characterized in that: The inference unit (44) produces and outputs a first prediction model of the temperature or airflow of the second virtual grid based on a learning result obtained by machine learning, wherein data that associates the propagation time of the propagation path of the detection sound wave in the object space (S) with the temperature distribution or the airflow distribution is used as training data, The estimation unit (44) uses the output value of the first prediction model as the constraint condition, and based on the learning result after adding the constraint condition to the training data and performing machine learning, produces a second prediction model that outputs the temperature or airflow of the first virtual grid (M1), The estimation unit (44) estimates the temperature distribution or the airflow distribution of the target space (S) by using the second prediction model using the propagation time of the propagation path of the target space (S) as input data.

4. The environmental measuring device according to any one of claims 1 to 3, characterized in that: The number of regions of the second virtual grid (M2) is the largest number of regions for which the temperature or the airflow can be uniquely determined without using the prediction method.

5. The environmental measuring device according to any one of claims 1 to 4, characterized in that: The measured value of the temperature is a measured value of a temperature sensor (30) arranged in the object space (S).

6. An environmental measurement device, characterized in that: The environment measurement device comprises a sound wave transceiver (20) and an estimation unit (44). The sound wave transceiver (20) transmits a detection sound wave toward the target space (S) and receives the detection sound wave. The estimation unit (44) estimates the temperature distribution or airflow distribution of the target space (S) based on the propagation time of the detection sound wave through the first virtual grid (M1), wherein the first virtual grid (M1) is formed by dividing the target space (S) into a plurality of regions and the plurality of regions are divided. The inference unit (44) produces and outputs a prediction model of the temperature or airflow of the first virtual grid (M1) based on a learning result obtained by machine learning, wherein data that associates the propagation time of the detection sound wave on the propagation path of the object space (S) with the temperature distribution or the airflow distribution is used as training data, The estimation unit (44) estimates the temperature distribution or the airflow distribution of the target space (S) by using the prediction model using the propagation time of the propagation path of the target space (S) as input data.

Images