JP2010134367A - Electric equipment - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide electric equipment capable of specifying a troubled portion, based on a sound. <P>SOLUTION: The electric equipment includes: a plurality of microphones 2 for acquiring the sound; a sound arrival direction specifying part 22 for detecting a direction of a sound source of the sound obtained by the microphone 2; a sound recognition part 23 for recognizing the sound, based on the sound obtained by the microphone 2 and a sound recognition pattern 31 stored preliminarily; and a trouble diagnostic part 24 for detecting a troubled position, based on the direction of the sound source of the sound obtained by the microphone 2, and based on a recognition result by the sound recognition part 23. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a technique for detecting a failure of an electric device such as an air conditioner by voice recognition.

  2. Description of the Related Art Conventionally, a technique for operating an electrical device by voice input using a sound recognition technique has been disclosed. For example, “When the air conditioner accepts an acoustic instruction such as driving, stopping, changing settings, etc. by tapping / speeching, it outputs a notice indicating that the instruction input is to be executed on the air conditioner side. The above-mentioned operation that is displayed in advance is executed only when an acoustic instruction from the instructor that means authorization for the display content of the advance notice is received again (for example, see Patent Document 1). .

  In addition, a technology has been proposed in which air conditioning control is performed by estimating the presence / absence of an air conditioning target person, the amount of activity, the number of people by an environmental sensor including a noise detection sensor, and determining a comfortable temperature / humidity (for example, patent documents). 2).

  In addition, “a refrigeration cycle in which a compressor, an outdoor heat exchanger, a decompressor, and an indoor heat exchanger are sequentially communicated, an outdoor fan that circulates outdoor air to the outdoor heat exchanger, and an indoor heat exchanger An indoor fan that circulates air, a sound sensor provided on the outdoor side, a comparison means that compares a detected sound level of the sound sensor with a set sound level, and the outdoor fan or compression depending on the comparison result of the comparison means An air conditioner having a control means for controlling the number of rotations of the machine has been proposed (see, for example, Patent Document 3).

Japanese Unexamined Patent Publication No. 04-278141 (2nd page, FIG. 1 and FIG. 2) Japanese Patent Laid-Open No. 04-121542 (2nd page, FIG. 1) Japanese Unexamined Patent Publication No. 03-036448 (2nd page, FIG. 6)

In the above-mentioned patent document 1, a technique using voice as a command means to an air conditioner that does not use a transmitter such as a remote controller is disclosed. Then, by allowing the command contents to be approved again by voice, the air conditioner is surely executed in accordance with the intention of the instructor.
However, even if the voice command can be recognized, it cannot be determined from which position the voice command is issued. For this reason, even if the voice does not intend the command, there is a problem that the command is recognized and operated.

  In the above-mentioned Patent Document 2, the amount of activity of an air conditioning subject such as a user is estimated using a noise detection sensor, and the temperature and humidity are determined. However, although it can be inferred that noise is occurring, it cannot be specified in which direction the noise is occurring. For this reason, there is no energy loss associated with air conditioning operation, such as when air conditioning cannot be performed effectively for locations where the air conditioning target is active, or when air conditioning is performed for locations where there is no air conditioning target. It was happening.

  Patent Document 3 discloses a technique in which a sound sensor is provided outside the air conditioner, a detected sound level is compared with a set temperature level, and the rotation speed of the compressor is controlled. However, the detected sound level depends on the installation conditions and installation environment of the air conditioner and the sound sensor, and there are cases where accurate determination cannot be made. In addition, since it is impossible to determine whether the noise detected by the sound sensor is caused by the compressor, even when there is no cause for noise in the compressor, the rotation speed of the compressor is changed and appropriate air conditioning cannot be performed. There was a problem.

  The present invention has been made in order to solve the above-described problems, and provides an electric device capable of detecting a failure based on sound. Further, it is an object of the present invention to obtain an electric device that can be operated with energy saving while improving user comfort based on voice.

  The electrical device according to the present invention includes a plurality of sound acquisition means for acquiring sound, a sound arrival direction detection means for detecting a sound source direction of sound acquired by the sound acquisition means, a sound acquired by the sound acquisition means, Voice recognition means for recognizing voice based on the stored voice pattern, and failure detection means for performing fault detection based on the sound source direction of the voice acquired by the voice acquisition means and the recognition result of the voice recognition means It is.

  The electric device according to the present invention can detect a failure based on the direction of the sound source and the voice recognition result. For this reason, user convenience can be improved.

Embodiment 1 FIG.
In the first embodiment, an air conditioner 100 having a ceiling cassette type indoor unit and an outdoor unit and operable by a remote controller will be described as an example.
FIG. 1 is a diagram showing a hardware configuration of an air conditioner 100 according to Embodiment 1 of the present invention. In FIG. 1, an air conditioner 100 includes a speaker 1, a microphone 2, a control device 3, a display unit 4, and a temperature sensor 5 that measures an indoor temperature. The air conditioner 100 can be connected to a PC 300 that is an external device.

  The speaker 1 corresponds to the sound output means of the present invention, and is controlled by the control device 3 to output a predetermined alarm sound or alarm message. The speaker 1 can be provided in an indoor unit of the air conditioner 100 or a remote controller.

  The microphone 2 corresponds to the sound acquisition means of the present invention, takes sound in the form of a sound signal, and outputs it to the control device 3. A plurality of microphones 2 are provided in a separate device such as an indoor unit, an outdoor unit, or a remote controller. Further, the installation direction and the installation position can be set by the microphone driving device 13 (actuator), and the sound acquisition direction of the microphone 2 can be changed.

  FIG. 2 is a schematic external view of a ceiling cassette type indoor unit and shows an installation example of the microphone 2. In FIG. 2A, three microphones 2 are attached to the lower part of the electrical component box 103 in which the control device 3 is accommodated. As described above, the wiring between the microphone 2 and the control device 3 can be facilitated by installing in the vicinity of the electrical component box 103. In FIG. 2B, one microphone 2 is installed at each of the four corners of the casing 101 of the indoor unit. By installing the microphones 2 apart from each other as described above, the direction of the sound source can be detected more accurately as will be described later.

  The display unit 4 is provided in one or both of the indoor unit and the remote controller, and displays various setting information of the air conditioner 100 and information such as an operation state. The display unit 4 can be configured by appropriately combining a liquid crystal display, a lamp display such as an LED, and the like.

  The control device 3 amplifies the audio signal captured by the microphone 2 and separates and captures only a predetermined frequency, and converts the analog signal into a digital signal or converts the digital signal into an analog signal. A D / D / A converter 7, a power amplifier 8 for amplifying an audio signal to be output by the speaker 1, an MPU 9 for executing various operations and processes, a RAM 10 for storing various data being processed or processed, It includes a ROM 11 that stores software executed by the MPU 9 and a voice database 30, and a connection unit 12 that connects to a PC 300 that is an external device.

  FIG. 3 is a diagram showing the configuration of the audio database 30 stored in the functional block of the MPU 9 and the ROM 11. In FIG. 3, the MPU 9 analyzes the digitized voice signal to identify the voice signal that is the target of voice recognition, and applies a predetermined voice recognition algorithm to the voice signal analyzed by the acoustic analyzer 20. The feature amount / supplemental calculation unit 21 for extracting the feature amount, the speech recognition unit 23 for performing speech recognition based on the output result of the feature amount / supplemental calculation unit 21, and the direction of the sound source based on the output result of the acoustic analysis unit 20 The voice arrival direction specifying unit 22 for specifying the fault, the fault diagnosis unit 24 for performing fault diagnosis based on the recognition result of the voice recognition unit 23 and the specification result of the voice arrival direction specifying unit 22, and the voice database 30 stored in the ROM 11 A database update unit 25 for updating is provided.

  The acoustic analysis unit 20 analyzes the voice signal and specifies a voice input section that is a target of voice recognition. Specifically, first, an audio signal is captured every predetermined interval, the amplitude value or power value (hereinafter referred to as audio level) of the audio signal is integrated at a predetermined sampling time, and this integrated amount is calculated for the previous interval. Compared to this, a section where the difference exceeds a threshold value for starting speech recognition is set as a speech recognition start section. Then, a series of a plurality of sections from the voice recognition start section is set as a voice input section, and the voice signal acquired in the voice input section is sent to the feature amount / supplement calculation unit 21 for voice recognition processing. Note that the voice input section may be determined in the frequency domain instead of determining the voice input section based on the sound pressure level.

  The feature amount / supplementary calculation unit 21 performs extraction of the feature amount of the speech signal output by the acoustic analysis unit 20 and other supplementary calculations according to the recognition algorithm of the speech recognition unit 23. The processing contents differ depending on the speech recognition algorithm. For example, when HMM (Hidden Markov Model: Hidden Markov Model) is used as a speech recognition algorithm, a mel cepstrum, a mel cepstrum, and dynamic features of the mel cepstrum in the time domain of the mel cepstrum are obtained from the input speech as acoustic features. Change amount) and a feature vector such as a dynamic feature amount of log power. Then, the obtained signal such as a feature vector is output to the speech recognition unit 23.

  The voice arrival direction specifying unit 22 specifies the direction in which the voice is emitted based on the signal analyzed by the acoustic analysis unit 20. When a certain sound is acquired by a plurality of microphones 2, since the distance from the sound source is different for each microphone 2, a time difference occurs in the timing at which each microphone 2 acquires the sound. The voice arrival direction specifying unit 22 uses this difference to specify the direction of the sound source.

An example of a method for obtaining a plane angle of a sound source with two microphones 2 will be described with reference to FIG. FIG. 4 is a schematic diagram showing an example of how to determine the plane angle of the sound source. In FIG. 4, the distance between the two microphones 2A and 2B is a distance d, and the distance difference between the distance from the sound source to the microphone 2A and the distance to the microphone 2B is a distance difference l.
When the distance between the sound source and the microphones 2A and 2B is sufficiently long, the sound waveform can be approximated as a plane wave, and the following formula can be applied.
τ = 1 / c = dsin (θ) / c
θ = sin−1 (cτ / d)
l: distance difference from sound source to microphone 2, c: speed of sound, d: distance between microphones, τ: delay time

When the sound wave is a stationary wave, the distance between the microphones must be set to be ½ or less of the sound wavelength to be acquired.
d <c / 2fmax = c / fs
c: speed of sound, d: distance between microphones, fmax: maximum frequency, fs: sampling frequency For example, when a voice having a frequency of 5.5 kHz is digitized and acquired by 11 kHz sampling, the distance between microphones 2 is set to 3.09 cm or less. There is a need. Conversely, when the microphone 2 is provided at a distance of 70 cm at both ends of the casing 101 of the indoor unit, the usable frequency is 340 / (0.7 × 2) = 242 Hz or less.

  As shown in FIG. 3B, when a plurality of microphones 2 are installed at a distance from each other, a distance farther from the air conditioner 100 can be measured. If three or more microphones 2 are provided to acquire sound, the solid angle to the sound source can be detected, and the direction of the sound source can be specified more accurately. In the case where a plurality of microphones 2 are provided, the direction of the sound source can be easily specified by installing the microphones 2 so that the distances from one reference microphone 2 to the other microphones 2 are equal. Further, when installing the plurality of microphones 2, not only the installation position in the horizontal direction but also the installation position in the height direction are changed. By doing so, the directivity capable of detecting the voice arrival direction can be changed. In addition, the microphone drive device 13 can change the installation position and the installation direction of the microphone 2 to make it easier to obtain an audio signal.

  The voice database 30 stores a voice recognition pattern 31 used for voice recognition. The speech recognition pattern 31 is a feature amount of speech that is a recognition target when the speech recognition unit 23 performs speech recognition. The normal sound pattern 32 is a driving sound pattern that is generated when the air conditioner 100 performs normal operation. The abnormal sound pattern 33 is a sound pattern of an operation sound when the air conditioner 100 is operated in a state where an abnormality has occurred in a member (for example, a fan) constituting the air conditioner 100. When the driving operation sound varies depending on the operation mode and various settings, the normal sound pattern 32 and the abnormal operation pattern 33 are stored for each operation mode and various settings. The normal voice pattern 32 and the abnormal voice pattern 33 are updated at a predetermined timing by the database update unit 25 described later. The voice pattern initial value 34 is an initial value of the normal voice pattern 32 and the abnormal voice pattern 33.

The speech recognition unit 23 performs speech recognition based on the feature amount extracted by the feature amount / supplemental calculation unit 21 and the speech recognition pattern 31 stored in the speech database 30. Speech recognition can be performed by various methods. For example, it can be realized by using the method described in Chapter 5 of “Speech Information Processing” (written by Sadahiro Furui, Morikita Publishing Co., Ltd.). it can.
For example, when an HMM is used as the speech recognition algorithm, the speech to be recognized is represented as an HMM in the speech database 30, and the probability that this HMM generates a feature vector sequence of the actually acquired speech is calculated. The HMM that is output with high probability is used as the recognition result.

  Before performing speech recognition, frequency separation of speech signals is performed, and only speech signals having the same frequency as the speech to be recognized are extracted. By doing in this way, noise can be removed from the speech recognition target and erroneous recognition can be prevented. The frequency separation may be performed by the function of the MPU 9, but can also be performed by a filter. In this case, a digital filter or the like can be provided to separate voice, vibration, noise, etc. for each frequency.

  The failure diagnosis unit 24 performs failure diagnosis based on the voice recognition result of the voice recognition unit 23 and the direction of the sound source specified by the voice arrival direction specifying unit 22. In failure diagnosis, abnormalities associated with secular changes are also estimated. Details of the failure diagnosis will be described later.

The database update unit 25 updates the speech database 30 based on the recognition result of the speech recognition unit 23 and the recognition result of the failure diagnosis unit 24. The database update includes an initial update performed when the air conditioner 100 is installed and a normal update performed at a predetermined timing.
First, the initial update will be described. In the initial state, a predetermined driving operation sound is initially registered as a voice recognition pattern 31 in the voice database 30. In the initial update, the driving operation sound of the test operation performed when the air conditioner 100 is installed is additionally registered as the normal sound pattern 32. At this time, a test operation is performed by combining operation modes such as heating, cooling, and blowing, and various settings such as a strong wind and a weak wind, and an operation sound for each setting is registered. The operation sound differs depending on the installation conditions such as the surrounding material, the pipe length, and the ambient noise in the installation environment of the air conditioner 100, but by updating the voice database 30 in the installation environment of the air conditioner 100, More accurate voice recognition and fault diagnosis can be performed. If the driving operation sound in the trial operation does not match the normal sound pattern 32 initially registered in the sound database 30, it is possible to determine that an abnormality has occurred and perform a notification operation.

In addition, when the installation place of the microphone 2 is close to the wall, there is a possibility that the voice recognition and the voice arrival direction cannot be accurately specified by reflection. When the distance between the sound source and the microphone 2 is short, the sound signal acquired by the microphone 2 is peak-cut, and similarly, there is a possibility that the voice recognition and the voice arrival direction cannot be specified accurately.
In order to cope with such a case, the voice database 30 is initially updated by acquiring voice signals and performing voice recognition over a certain period of time such as 24 hours, one week, and one month after the installation of the air conditioner 100. Also good. Alternatively, sound is generated from a sound generating device such as a buzzer provided in the main body of the air conditioner 100 or the separation type device, and the sound pressure of the sound signal of the buzzer acquired by the microphone 2 is automatically generated by a PGA (programmable gain amplifier). The gain may be adjusted with.
In addition, when the reflection is not superimposed, the microphone 2 acquires a sound having a low spectrum. For this reason, as a countermeasure against reverberation, when a similar voice recognition pattern 31 is continuous, it can be cut, or spectral subtraction or adjustment of the recognition window of the voice recognition unit 23 can be performed.

  Next, normal update of the voice database 30 will be described. The normal update is performed at the timing when voice recognition or failure diagnosis is performed. The database update unit 25 additionally registers the driving operation sound recognized by the voice recognition unit 23 in the voice database 30 as the voice recognition pattern 31 together with information such as the operation mode and various settings of the air conditioner 100. At this time, the operation sound during normal operation is registered as a normal sound pattern 32. In addition, the driving operation sound when the failure is determined by the failure diagnosis is registered as the abnormal sound pattern 33. Thus, the accuracy of voice recognition can be improved by updating the voice database 30 based on the driving operation sound actually acquired during the operation of the air conditioner 100. In the normal update, the sound data of the recognized driving operation sound can be stored as a log. By doing in this way, the failure estimation by the secular change in the failure diagnosis mentioned later can be performed easily.

  Note that the MPU 9 has dedicated programs corresponding to the functions of the acoustic analysis unit 20, the feature / supplementary calculation unit 21, the voice recognition unit 23, the voice arrival direction specifying unit 22, the failure diagnosis unit 24, and the database update unit 25, respectively. It is realized by executing. These programs are stored in the ROM 11, and are read by the RAM 10 when necessary and executed by the MPU 9.

FIG. 5 is an operational flow of failure diagnosis of the air conditioner 100 configured as described above.
When the failure diagnosis is started, the MPU 9 acquires an audio signal via the microphone 2, the filter, the amplifier, and the A / D / D / A converter 7 (S121). Next, an acoustic analysis is performed to identify a voice signal that is a target of voice recognition (S122). Subsequently, feature amount extraction or the like according to the speech recognition algorithm of the speech recognition unit 23 is performed (S123). Next, the voice recognition algorithm of the voice recognition unit 23 recognizes what kind of voice the acquired voice is (S124). Also, the voice arrival direction is specified (S125). In FIG. 5, the voice arrival direction is specified after voice recognition (S124, S125), but both processes can be performed in parallel.

  Then, failure diagnosis is performed based on the voice recognition result and the voice arrival direction (S126). If there is an abnormality (S127), a fault location is specified (S131), and the abnormality is notified by the display unit 4 or the speaker 1 (S132). If there is no abnormality in the failure diagnosis (S127), aged deterioration diagnosis is performed (S128). In the aging deterioration diagnosis, it is determined whether the deterioration is gradually deteriorated due to secular change by comparing the voice recognition result with the voice pattern initial value 34 (S128). If the secular change exceeds the normal range, it is determined that an abnormality has occurred (S129), the failure location is specified (S131), and the abnormality is notified using the speaker 1, the display unit 4, or the like (S132). ). In addition, if the notification method is changed depending on whether there is an abnormality due to a failure or an abnormality due to secular change, the content of the abnormality can be easily communicated to the user. If the secular change is within the normal range (S129), the data is stored in the normal voice pattern 32 of the voice database 30 as the latest normal operation sound (S130).

Here, in order to accurately perform voice recognition and voice arrival direction specification, a plurality of microphones 2 having the same characteristics can be used so that amplitudes and phases can be compared as accurately as possible.
On the other hand, different types of microphones 2 can be used depending on the application. For example, when four microphones 2 having different frequency characteristics are used, one of the four microphones 2 can acquire an accurate voice with a large gain, and the other microphones 2 can reach the voice such as phase difference determination. Used for direction detection. In this way, it is possible to detect the voice arrival direction while acquiring the voice signal with the maximum gain.
Further, in order to obtain directivity, the microphone 2 can be provided with a delay. Alternatively, the microphone 2 having different characteristics can be used depending on the voice to be recognized, such as directivity like a gun microphone.

Next, the failure diagnosis process will be described in detail.
In the failure diagnosis process, whether or not a failure has occurred is diagnosed based on the voice recognition result by the voice recognition unit 23. For example, when the speech recognition unit 23 performs speech recognition using an HMM, the speech to be recognized is represented as an HMM in the speech database 30, and the probability that this HMM generates a feature vector sequence of actually acquired speech The HMM that is output with the highest probability is used as the recognition result. If the recognition result indicates the normal voice pattern 32, it can be determined that no failure has occurred. Further, even when the recognition result indicates the normal voice pattern 32, it is possible to diagnose that a failure has occurred if the probability is equal to or less than a predetermined threshold. The threshold value or the like when performing the failure diagnosis can be arbitrarily set according to the installation environment, the installation years, and the like of the air conditioner 100.
When the obtained driving operation sound is diagnosed as abnormal, the abnormal location is specified based on the direction of the sound source specified by the voice arrival direction specifying unit 22. For example, when the central direction of the indoor unit is a sound source, it is determined that an abnormality has occurred in a component or the like in that direction. Thus, by using the direction of the sound source for failure diagnosis, it is possible to perform failure determination with higher accuracy. In addition, if the identified failure location is notified, the user or the person in charge of repair can easily identify the failure location and the cause of failure.

Even when it is determined that the recognized sound is normal, if the direction of the sound source is a place where the sound hardly sounds, it can be determined that an abnormality has occurred in that place. .
In the failure diagnosis, the diagnosis is performed using the voice recognized by the voice recognition unit 23. The voice is frequency-separated. For this reason, since it can diagnose for every frequency, a more detailed failure diagnosis is possible.

  Next, the failure estimation operation due to secular change will be described. As described above, the normal voice pattern 32 of the voice database 30 is updated when it is determined that the fault diagnosis is normal (see step S130 in FIG. 5). Therefore, since the normal voice pattern 32 of the voice database 30 changes gradually, it is difficult to grasp the change when the deterioration gradually occurs due to secular change. For example, although the dust filter installed in the air conditioner 100 is gradually soiled over time, the normal sound pattern 32 is also periodically updated, so that the change in sound due to the dirt of the dust filter is grasped. Difficult to do. For this reason, in failure estimation due to secular change, the voice pattern initial value 34 is compared with the currently recognized voice pattern, and a difference is detected. If the difference exceeds a predetermined threshold, it is determined that aged deterioration has occurred. By doing so, it is possible to detect deterioration that occurs gradually.

  For example, when the dust filter is contaminated, the sound spectrum of the sound of the air passing through the dust filter is increased, so that it can be determined that the dust filter is clogged with dust. Conventionally, there is a technique for performing filter cleaning by estimating deterioration based on operation time. However, according to the present invention, even when the driver is operating in a poor environment and immediately becomes clogged, the filter is appropriately filtered. Cleaning can be performed.

Here, the failure diagnosis can be performed at an arbitrary timing. For example, it can be performed at a predetermined date and time, or can be performed periodically every predetermined time.
Further, a means for setting the timing for performing the failure diagnosis or an operation means for starting the failure diagnosis may be provided to perform the failure diagnosis at a timing designated by the user. By making it possible for the user to specify the timing for performing failure diagnosis in this way, it is possible to improve user convenience, such as avoiding failure diagnosis while the air conditioner 100 is being used. In addition, a period in which the air conditioner 100 is hardly used can be estimated from the operating state of the air conditioner 100 (for example, spring or autumn), and failure diagnosis can be performed periodically during this period.
Further, a time measuring unit that measures the time after the air conditioner 100 is installed may be provided, and the content of the failure diagnosis may be changed according to the elapsed time after the installation. For example, since it is unlikely that a failure will occur immediately after installation, a simple failure diagnosis can be performed by omitting the diagnosis of aging deterioration, and a diagnosis of aging deterioration can be performed after a predetermined period of time has elapsed since installation. In this way, the load required for failure diagnosis can be reduced.

  Further, even when it is not the failure diagnosis timing, the voice recognition unit 23 performs voice recognition at a predetermined timing, and when a large change (a loud sound) occurs in the standing wave during the operation of the air conditioner 100, the operation at that time is performed. The same control as the control is executed again to determine whether or not the phenomenon is reproduced. When reproducing, an abnormality is notified, and when not reproducing, failure diagnosis is performed. By doing in this way, even when a failure occurs other than a predetermined failure diagnosis timing, it is possible to appropriately perform the failure diagnosis.

In the above description, the audio database 30 is constructed by the MPU 9 of the control device 3. However, depending on the processing capability of the MPU 9, it may be difficult to perform processing with high accuracy. In such a case, the plurality of microphones 2 are connected to the PC 300 connected via the connection unit 12, and the voice database 30 is constructed by performing voice recognition based on the voice signal acquired by the microphone 2. The voice database 30 constructed in this way is stored in the ROM 11 of the control unit. If it is PC300, since the setting for performing voice recognition with high precision, such as designation of a voice recognition area, the position and material of a wall, and a noise filter, can be performed in detail, the voice database 30 with higher precision can be constructed. Can do. In addition, even if it is not PC300, if it is an apparatus which can implement | achieve an equivalent function, such as an application server on the internet, the apparatus to be used is not ask | required.
The communication method between the air conditioner 100 and an external device (such as the PC 300) can be wireless LAN, weak wireless, infrared communication, visible light communication, wireless communication such as RFID, Ethernet (registered trademark), UART (Universal Asynchronos). It can be performed by wired communication such as Receiver Transmitter (I2C), I2C (Inter IC Communication), and PLC (Power Line Communication), and any method can be used as long as communication is possible.
It is also possible to provide the microphone itself at a place different from the air conditioner 100 and the external device, and transfer the audio data acquired by the microphone to the air conditioner 100 and the external device.

  Also, active noise cancellation can be performed. Specifically, when the speech recognized by the speech recognition unit 23 has a certain or higher speech spectrum that is a standing wave, any component that constitutes the air conditioner 100 is a noise sound due to deterioration over time or the like. Is output from the speaker 1 facing the sound source direction of the recognized sound. In this way, noise noise can be canceled out, so that even if a sound trouble occurs due to deterioration over time or failure after installation of the air conditioner 100, it is possible to detect abnormal driving operation sound associated with the failure. . Moreover, since the noise sound accompanying driving | operation operation | movement can be negated, a user's comfort can be improved. Furthermore, materials such as a sound deadening material to be provided in the air conditioner 100 can be reduced. Note that when performing active noise cancellation, the speech recognition unit 23 periodically determines whether there is a change in the noise source and whether noise in another direction has increased.

  As described above, according to the air conditioner 100 according to the first embodiment, the failure detection is performed based on the voice recognition result and the sound source direction, so that the failure position and the failure content can be appropriately grasped. . Also, the voice recognition pattern used for voice recognition can be updated. For this reason, even when the driving operation sound varies depending on the installation environment of the air conditioner 100, more accurate voice recognition and failure diagnosis can be performed.

  In the first embodiment, it has been described that the initial update or the normal update of the voice database 30 is performed. However, the voice database 30 may not be updated, and voice recognition may be performed using the voice database 30 constructed in advance. In this case, various normal sound patterns 32 and abnormal sound patterns 33 corresponding to the operation mode and various settings are registered by a test operation or the like. Even in this way, failure diagnosis can be performed by voice recognition, and the burden on the air conditioner 100 required for updating the voice database 30 can be eliminated.

  In the first embodiment, a case where a plurality of microphones 2 are provided and the sound source direction is specified using a difference in timing at which each microphone 2 obtains sound has been described as an example. However, only one microphone 2 is provided. There may be. In this case, the microphone head of a highly directional microphone can be modified so as to face in various directions, and sound can be acquired from a plurality of directions, and the direction with the highest sound pressure level can be set as the sound source direction.

Embodiment 2. FIG.
In the second embodiment, an example will be described in which vibration is detected by a microphone 2 serving as a voice acquisition unit, and failure diagnosis is performed by vibration. Here, it is assumed that the microphone 2 includes a diaphragm (not shown) inside, and obtains sound by a change in capacitance caused by sound pressure. The air conditioner according to the second embodiment is different from the first embodiment described above only in the function of the MPU, and therefore the description will focus on the differences.

  FIG. 6 is a diagram illustrating a functional block of the MPU 9 according to the second embodiment and a configuration of the vibration database 150 stored in the ROM 11. In FIG. 6, the MPU 9 a includes a vibration analysis unit 40, a vibration recognition unit 41, a vibration arrival direction specifying unit 42, and a failure diagnosis unit 43.

  The vibration database 50 stores a vibration pattern 51. The vibration pattern 51 is a vibration pattern of vibration generated by operation when the air conditioner is operating normally. In addition, when the vibration by the operation of the air conditioner varies depending on the operation mode and various settings, the normal vibration pattern 51 is stored for each operation mode and various settings.

  The vibration analysis unit 40 performs frequency analysis of vibration acquired by the microphone 2 and then performs frequency analysis to detect a vibration spectrum. The vibration recognition unit 41 compares the detection result of the vibration analysis unit 40 with the normal vibration pattern 51 stored in the vibration database 50. Then, it is recognized whether or not the detection result of the vibration analysis unit 40 is the same as the normal vibration pattern 51, or is coincident with either the normal vibration pattern 51 multiplication or the normal vibration pattern 51 frequency division. The vibration arrival direction specifying unit 42 specifies the direction of the vibration source based on the detection result of the vibration analysis unit 40. The vibration source is specified based on the arrival delay times of vibrations acquired by the plurality of microphones 2, similarly to the voice arrival direction specifying unit 22 described in the first embodiment. Then, the failure diagnosis unit 43 diagnoses the presence / absence of the failure and the direction of the failure based on the recognition result of the vibration recognition unit 41 and the detection result of the vibration arrival direction specifying unit 42. For example, when the microphones 2 are installed at the four corners of the casing of the indoor unit as shown in FIG. 2B, vibrations reach the four microphones 2 at the same time (from the four microphones 2 to the vibration source). If the vibration is different from the vibration pattern during normal operation, it is diagnosed that some failure has occurred near the center of the indoor unit (for example, a fan (not shown)).

  When the microphone 2 using the capacitor is used as the sound acquisition means, the diaphragm provided in the microphone 2 is changed by the sound pressure, so the sound signal is acquired by the change in the capacitance. As described above, the microphone 2 that acquires sound by vibration is designed so that a gain is obtained by matching the frequency characteristic to the sound region (audible region), so that sound can be easily acquired. However, if the microphone 2 is attached to a device such as an air conditioner that vibrates due to a driving operation, the vibration in the microphone 2 is affected by the vibration of the driving operation. In order to remove such influence, filtering is performed to obtain an audio signal.

  In the second embodiment, frequency separation of vibrations acquired by the microphone 2 and frequency analysis are performed, and the presence or absence of a failure is diagnosed based on the vibration frequency. As described above, it is possible to perform failure diagnosis and specification of a failure position also by vibration sensed by the microphone 2 that is a voice acquisition unit. Further, a more detailed failure diagnosis can be performed by using in combination with the sound failure diagnosis described in the first embodiment.

Embodiment 3 FIG.
In the third embodiment, an air conditioner 200 that performs air conditioning control based on a voice recognition result and can be operated by voice input will be described as an example. In the third embodiment, the same reference numerals are assigned to the same components as those in the first embodiment.

  FIG. 7 is a diagram illustrating a hardware configuration of the air conditioner 200 according to the third embodiment. The third embodiment is different from the first embodiment in that an ultrasonic transmitter 14 that can generate ultrasonic waves in an arbitrary direction is provided.

  FIG. 8 is a diagram showing the functional blocks of the MPU 9 according to the third embodiment and the structure of the voice database 60 and the activity database 70 stored in the ROM 11. In FIG. 8, the MPU 9 uses the acoustic analysis unit 20, the feature amount / supplement calculation unit 21, the speech recognition unit 23, the speech arrival direction specifying unit 22, the recognition result of the speech recognition unit 23, and the specification result of the speech arrival direction specifying unit 22. Based on a human body detection unit 82 that detects a human body activity region and activity amount, a database update unit 83, an air conditioning control unit 85 that controls the operation of the air conditioner 200, and an operation instruction from a remote controller (not shown). An operation I / F (interface) 84 that converts the operation signal is provided.

  The voice database 60 stores a voice recognition pattern 61 for performing voice recognition. There are two types of voice recognition patterns 61, a life voice pattern 62 and an operation voice pattern 63. The life sound pattern 62 is a feature amount of typical life sound of the user in the installation environment of the air conditioner 200. For example, in the case of the air conditioner 200 installed in an office or the like, a mouse click sound, a keyboard keying sound, a telephone ringing sound, and the like are stored as the life sound pattern 62. The operation sound pattern 63 stores a sound pattern for performing an operation by sound. For example, in a case where a predetermined operation is performed with “twice applause”, an audio pattern when clap is performed twice is stored.

  The activity database 70 stores, as an activity pattern 71, the date and time, the user activity area and the activity amount at the date and time in the environment where the air conditioner 200 is installed. The user can set the activity database 70 in advance. Further, the activity database 70 can be updated by learning according to the user activity status. It can be said that the activity database 70 is information indicating when, where, and what activity the user performs within a predetermined period such as 24 hours, one week, and one month.

  The voice recognition unit 23 specifies the voice by referring to the voice database 60 based on the feature quantity extracted by the feature quantity / supplementary calculation unit 21. The result of the voice recognition is output to the human body detection unit 82, and is output to the operation I / F 84 when the recognized voice is the operation voice. Note that the voice recognition method is the same as that in the first embodiment.

  The space recognition unit 81 performs space recognition based on the feature amount extracted by the feature amount / supplemental calculation unit 21 and the direction of the sound source specified by the voice arrival direction specifying unit 22. The operation of space recognition will be described later.

  The human body detection unit 82 detects the presence / absence of the user, the user's activity area, the amount of activity of the user, and the like based on the voice recognition result of the voice recognition unit 23 and the direction of the sound source specified by the voice arrival direction. The human body detection operation will be described later.

The database update unit 83 performs initial update when the air conditioner 200 is installed, and updates the voice database 60 and the activity database 70 based on the detection results of the voice recognition unit 23 and the human body detection unit 82 at a predetermined cycle.
First, the update of the voice database 60 will be described. The voice database 60 stores in advance the voice recognition pattern 61 of the user's life sound, but the life sound varies depending on the user. For this reason, for example, when the voice recognition unit 23 repeatedly detects specific voice data within a predetermined period (24 hours, one week, one month, etc.), it is determined that the voice is the user's life sound, A voice recognition pattern 61 is written in the voice database 60. By doing in this way, the precision of the speech recognition by the speech recognition part 23 can be improved.
Next, the update of the activity database 70 will be described. In the human body detection within a predetermined period (24 hours, one week, one month, etc.), the human body detection unit 82 learns the detection contents and the time and place thereof and stores them in the activity database 70. The information to be stored is, for example, when and where the user was present. Further, when the user performs an operation of the air conditioner 200, the time and contents may be learned and stored in the activity database 70.

  The operation I / F 84 receives an input signal sent from an operation means such as a remote controller (not shown) and transmits it to the air conditioning control unit 85 as an operation signal. When the voice recognized by the voice recognition unit 23 is an operation command, the recognized voice is transmitted to the air conditioning control unit 85 as an operation signal.

  The air conditioning control unit 85 performs air conditioning control such as cooling and heating based on an input signal from the operation I / F 84. At this time, the user's activity area is grasped by referring to the activity database 70, and the user's activity area is selectively air-conditioned. For example, control is performed so as to send a wind with a swing so as to give a cool feeling to the user's activity area. Alternatively, control is performed so that air conditioning is not performed outside the user activity area. The air conditioning operation will be described later.

Next, the space recognition operation will be described.
At the time of initial setting such as when the air conditioner 200 is installed, the sound signal is acquired by the microphone 2 for a certain period (for example, 24 hours, 1 week, 1 month, etc.), and the direction of the sound source is acquired by the sound arrival direction specifying unit 22. Is identified. Then, the sound spectrum of the sound signal is compared for each sound source arrival direction, and a place where the difference of the sound spectrum is large is defined as the indoor / outdoor boundary.

Spatial recognition is performed at the time of initial setting, and is also repeatedly performed at a predetermined timing during normal operation of the air conditioner 200. In this way, it is possible to grasp the change in the space due to the opening / closing of the door and the movement of the fixture.
Further, space recognition by a camera, a photodiode, a pyroelectric sensor, a thermopile, a distance measuring sensor, and a radar may be performed together. In this case, when the voice recognition unit 23 recognizes a predetermined voice, the detection target can be detected more accurately by performing recognition using a camera or the like. Furthermore, space recognition may be performed by an external device such as the PC 300.

  Spatial recognition can also be performed by ultrasonic waves. In this case, the ultrasonic transmitter 14 emits an ultrasonic wave in a predetermined direction, receives the reflected wave with the microphone 2, and measures the distance from a non-detected object such as a wall to perform space recognition. That is, the microphone 2 as a voice acquisition unit is used as an ultrasonic sensor. The microphone 2 to be used may be a piezoelectric element type or a capacitor type as long as it can receive sound waves and ultrasonic waves. Moreover, if the microphone 2 and the speaker are integrally configured, the number of parts can be reduced. When the ultrasonic signal is received by the microphone 2, the distance to the detection target is measured based on the received signal. The principle of measurement is the same as that of a general ultrasonic sensor. At this time, detection accuracy can be improved by repeatedly transmitting and receiving ultrasonic waves using a plurality of microphones 2. When ultrasonic waves are transmitted over a wide area, the distance to an obstacle such as a wall is determined from the time difference at which the microphone 2 receives the reflected ultrasonic waves.

Moreover, although the propagation speed of an ultrasonic wave changes with air temperature, distance detection with high precision can be performed by using the detection result of the temperature sensor 5 (not shown) with which the air conditioner 200 is provided. Note that the velocity v of the sound wave may be easily obtained by the following equation.
v≈331.5 + 0.61 * t (m / s)
t: Air temperature (° C)

  Further, for example, when the distance to the detection target is far and the distance cannot be detected by the ultrasonic wave, sensing may be performed with the wavelength of the sound wave region. In this case, the presence / absence determination of a person is performed by voice recognition or an infrared sensor, and a sound wave having a long wavelength is generated in the absence area. By doing in this way, space recognition can be performed without giving a user discomfort.

Next, a human body detection operation for detecting a user's activity area will be described.
At the time of initial setting such as when the air conditioner 200 is installed, a voice signal is acquired by the microphone 2 for a certain period (for example, 24 hours, one week, one month, etc.), and the voice recognition unit 23 performs voice recognition. In the voice database 60, a voice recognition pattern 61 of typical user life sounds (user voice, telephone ringing sound, mouse click sound, keyboard keying sound, etc.) is registered in advance, and a predetermined voice recognition algorithm is registered. Voice recognition is performed. In addition, the direction of the sound source is specified by the voice arrival direction specifying unit 22. When the user's life sound is recognized, the human body detection unit 82 determines that the direction of the sound source is the user's activity area. If the user's life sound cannot be recognized, it is determined that the user is outside the activity area. When determining whether or not it is the user's activity area, it is possible to determine by setting a criterion such as whether or not the living sound has been repeatedly recognized exceeding a predetermined threshold.

The user's activity area can be detected in conjunction with the time, lighting ON / OFF state, and the like. In this case, a time measuring means (not shown) is provided in the air conditioner 200, and the presence / absence of the user is detected according to the time. Alternatively, the lighting ON / OFF state is acquired, and when it is OFF, it is determined that the user is absent. Further, the user activity area may be detected by an infrared sensor or a camera. An external device such as the PC 300 may be used.
Moreover, this information is utilized when the installation position of the indoor unit of the air conditioner 200 is preset. Furthermore, the user's activity area may be detected by emitting ultrasonic waves in a predetermined direction by the ultrasonic transmitter 14 and determining the presence / absence of the user.

Next, the air conditioning control operation will be described.
When the air-conditioning operation is started, the air-conditioning operation is started at the temperature and the operation mode set by operation means such as a remote controller. At this time, the activity database 70 is referred to selectively air-condition the region recognized as “indoor” by space recognition or the user activity area. For example, control is performed such as sending a wind with a swing to give a cool feeling toward the user's activity area. When the activity database 70 can predict the user's activity schedule in advance, a predictive operation such as starting / stopping cooling may be performed in advance. By doing in this way, useless air conditioning can be omitted and energy consumption can be reduced.

In addition, voice recognition and human body detection are performed during air conditioning operation, and air conditioning control is performed based on the detection result. For example, if the user's daily sound such as the user's voice or keyboard tapping sound cannot be detected for a predetermined time, it is determined that there is no user and the air conditioning is stopped for that area, or the system is operated in the energy saving mode. Control such as to do.
Also, when the user's voice is detected and the direction of arrival of the voice changes frequently, it is determined that the user has repeatedly moved, the amount of activity is large, and the human body generates a large amount of heat. Air conditioning control such as setting or setting the heating to be weak is performed. By doing in this way, a user's comfort can be improved. Even when the air conditioning control is performed based on the voice recognition result, it is possible to operate by a user's manual setting.

Next, an operation method using voice will be described.
When the acquired voice is recognized as a voice operation signal, the voice operation signal is transmitted to the air conditioning control unit 85 via the operation I / F 84. The air conditioning control unit 85 refers to the sound source direction of the voice operation signal and the activity database 70 to determine whether or not this voice operation signal is issued from the user's activity area, and accepts only voice operation from the activity area. Then, the voice operation is confirmed by the voice output from the speaker 1 or the display by the display unit 4, and the voice operation is executed when there is a predetermined reaction from the user. In addition, an area where voice operation is possible is set in advance, and voice operation from outside this area may not be accepted. For example, in the case of setting to activate the voice operation function with two applauses and then raise the set temperature with one applause, it is determined whether or not to accept the voice operation based on the result of voice recognition and the sound source direction. In this way, erroneous recognition of voice operation can be prevented.

  Moreover, it can connect with a network and can interlock | cooperate with the other air conditioner 200, a circulator, a ventilation fan, a humidifier, a dehumidifier, an air cleaner, etc. For example, when interlocking with another air conditioner 200, human body detection is performed on both sides, and if a user exists only in the air-conditioning target area of one air conditioner 200, the air conditioned escapes from that area. The other blows to form an air curtain so that there is no air. By doing in this way, air-conditioning efficiency can be improved. In addition, when air conditioning is performed with a plurality of air conditioners 200, when a certain air conditioner 200 is operating at full power, air conditioned air is sent from another air conditioner 200 to assist air conditioning. To do. By doing in this way, each air conditioner 200 can air-condition efficiently. Further, the operation history, setting information, electricity bill, etc. of the air conditioner 200 can be confirmed from an external device via a network, or remote operation can be performed by the external device.

  Moreover, since the air conditioner 200 of the present invention can determine the presence / absence of the user by the human body detection unit 82, an alarm operation can be performed when a human body is detected outside the working hours. In this case, for example, when an alarm operation is unnecessary, for example, when a security guard enters the room, a voice that emits a specific voice from the authentication system is linked with an authentication system such as an RFID connected to the air conditioner 200 via a network. When the harmony machine 200 recognizes it, the human body detection may be stopped for a predetermined time. Conventionally, there is a technique for detecting a human body using a camera, but the air conditioner 200 according to Embodiment 3 can detect a human body at a lower cost. In addition, there is a technology that detects human bodies using a radiation sensor or infrared pyroelectric sensor, but when detecting human bodies by acquiring temperature using a radiation sensor, it takes time to acquire the temperature. Can not do. In addition, according to the infrared pyroelectric sensor, it is possible to determine whether there is a change in infrared rays, but it is not possible to determine a detailed situation such as the size of a detection target. According to the air conditioner 200 according to the third embodiment, since human body detection is performed by voice recognition, it can be determined in real time. Also, the position and size of the detection target can be detected based on the voice arrival direction. Also, an alarm operation can be performed through the Internet or a telephone line.

  Thus, according to the air conditioner 200 according to Embodiment 3, space recognition and human body detection are performed based on the voice recognition result and the sound source direction, and air conditioning control is performed based on the space recognition result and the human body detection result. I did it. For this reason, it is possible to drive with energy saving while improving user comfort.

In the above description, the air conditioner has been described as an example, but the electric device according to the present invention can be applied to various devices. For example, the present invention can be applied to AV equipment such as an air cleaner, a humidifier, a dehumidifier, a ventilation fan, and a television set. In addition, it can be applied to refrigerators to control the number of times the refrigerator door is opened and closed by voice recognition, and when it is frequently opened and closed, the set temperature can be lowered, the air curtain at the door entrance is operated, and a warning can be issued. is there.
In addition, it is possible to perform operation control such as grasping the user's activity time zone by voice recognition, reducing the driving ability outside the activity time zone, and putting the display device such as a liquid crystal in a standby state or a stopped state.
In addition, since the failure location can be detected by voice recognition, for example, repair consultation by telephone can be smoothly performed and the satisfaction of the user can be improved, and the work efficiency of the repair staff can be improved.

  In addition, the direction of the microphone can be changed manually / automatically during installation or during operation of the air conditioner, and can be disposed at an optimal position for acquiring sound.

It is a figure which shows the hardware constitutions of the air conditioner which concerns on Embodiment 1 of this invention. It is an external appearance schematic diagram of a ceiling cassette type indoor unit, and shows an installation example of a microphone. It is a figure which shows the structure of the functional block and audio | voice database of MPU which concern on Embodiment 1 of this invention. It is a figure explaining how to obtain the plane angle of a sound source. It is an operation | movement flow of the fault diagnosis which concerns on Embodiment 1 of this invention. It is a figure which shows the structure of the functional block of MPU which concerns on Embodiment 2 of this invention, and a vibration database. It is a figure which shows the hardware constitutions of the air conditioner which concerns on Embodiment 3 of this invention. It is a figure which shows the structure of the functional block of MPU which concerns on Embodiment 3 of this invention, an audio | voice database, and an activity database.

Explanation of symbols

  1 speaker, 2 microphone, 2A microphone, 2B microphone, 3 control device, 4 display unit, 5 temperature sensor, 6 filter / amplifier, 7 A / D / D / A converter, 8 power amplifier, 9 MPU, 10 RAM, 11 ROM, 12 connection unit, 13 microphone driving device, 14 ultrasonic transmitter, 20 acoustic analysis unit, 21 feature quantity / supplement calculation unit, 22 voice arrival direction specifying unit, 23 voice recognition unit, 24 fault diagnosis unit, 25 database update Part, 30 voice database, 31 voice recognition pattern, 32 normal voice pattern, 33 abnormal voice pattern, 34 voice pattern initial value, 40 vibration analysis part, 41 vibration recognition part, 42 vibration arrival direction specifying part, 43 fault diagnosis part, 50 Vibration database, 51 vibration pattern, 52 normal vibration pattern, 60 voice database 61, voice recognition pattern, 62 life voice pattern, 63 operation voice pattern, 70 activity database, 71 activity pattern, 81 space recognition unit, 82 human body detection unit, 83 database update unit, 85 air conditioning control unit, 100 air conditioner, 101 housing, 102 flap, 103 electrical box, 200 air conditioner.

Claims (15)

Audio acquisition means for acquiring audio from each of a plurality of directions;
Voice arrival direction detection means for detecting the sound source direction of the voice acquired by the voice acquisition means;
Voice recognition means for recognizing voice based on the voice acquired by the voice acquisition means and a voice pattern stored in advance;
An electrical apparatus comprising: a failure detection unit that detects a failure based on a sound source direction of a voice detected by a voice arrival direction detection unit and a recognition result of the voice recognition unit. The electrical device according to claim 1, wherein the failure detection unit periodically detects a failure. The electrical apparatus according to claim 1, further comprising an updating unit configured to update the previously stored voice pattern at a predetermined timing based on a voice recognition result by the voice recognition unit. With audio output means,
The voice output unit outputs a voice obtained by inverting the phase of the frequency of the voice when it is determined that the failure detection unit is normal based on a voice recognition result by the voice recognition unit. The electrical apparatus in any one of Claims 1-3. Among the output signals from the sound acquisition means, vibration analysis means for detecting a vibration signal from a vibration source excluding a sound source;
Vibration arrival direction detection means for detecting the vibration source direction of the vibration signal detected by the vibration analysis means;
A vibration recognition means for recognizing vibration based on a vibration signal detected by the vibration analysis means and a vibration pattern stored in advance;
The failure detection unit performs failure detection based on a vibration source direction detected by the vibration arrival direction detection unit and a recognition result of the vibration recognition unit. Electrical equipment as described in. 2. A space recognition unit that learns a sound source direction of the sound acquired by the sound acquisition unit and a recognition result of the sound recognition unit for a predetermined period and performs space recognition based on the learned result. The electrical apparatus according to any one of claims 5 to 6. When the voice acquired by the voice acquisition unit is recognized as a control command by the voice recognition unit, and the sound source direction of the acquired voice is a predetermined space recognized by the space recognition unit The electric device according to claim 6, further comprising operation control means for performing operation control according to a voice recognition result by the voice recognition means. Human body detecting means for detecting the presence or absence of a user based on a voice recognition result by the voice recognition means;
The electric device according to any one of claims 1 to 7, wherein the operation control unit performs operation control according to a detection result of the human body detection unit. When the voice recognition means recognizes a voice generated by a user,
The human body detection means repeatedly obtains the sound source direction of the voice recognized by the voice recognition means at a predetermined timing, determines the amount of activity of the user based on the amount of change in the sound source direction,
The electric device according to claim 8, wherein the operation control unit performs operation control according to an activity amount of the user. The electric device according to any one of claims 1 to 9, further comprising a voice acquisition direction setting unit that sets a voice acquisition direction of the voice acquisition unit. The electric device according to any one of claims 1 to 10, further comprising notification means for notifying information on a failure detected by the failure detection means. A plurality of the voice acquisition means;
The electrical equipment according to any one of claims 1 to 11, wherein the plurality of voice acquisition units are provided such that a distance from a reference voice acquisition unit is constant. It has a connection that can be connected to external devices,
The electric apparatus according to claim 1, wherein various kinds of information including at least the sound pattern are acquired by the external apparatus.   The electrical device according to any one of claims 1 to 13, which is an air conditioner. The electrical device according to any one of claims 1 to 13 is an air conditioner,
The sound acquisition means is mounted on the lower end, both ends, or inside the casing of the indoor unit.

Images