Circuit breaker acoustic diagnosis device and method

Abstract

To make it possible to grasp the state of a circuit breaker in a precise manner in a diagnosis using acoustics.SOLUTION: An acoustic diagnostic device is configured to: extract a first feature amount of an acoustic signal, the acoustic signal being collected by a microphone for each event of a circuit breaker during normal operation of the circuit breaker and corrected for a delay of the acoustic signal caused by a relative distance between the circuit breaker and the microphone; create a plurality of event models that manage data on the corrected first feature amount of the acoustic signal on an event-by-event basis; extract a second feature amount of the acoustic signal, the acoustic signal collected from the microphone at the time of diagnosing the circuit breaker and corrected for a delay in the acoustic signal caused by the relative distance between the circuit breaker and the microphone; calculate the similarity between the second feature amount of the acoustic signal at the time of diagnosis and the first feature amount of the plurality of event models at each time; estimate the time when the similarity is highest as the occurrence timing of each event; and diagnose the circuit breaker on the basis of the plurality of estimated occurrence timings.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an acoustic diagnosis device for a circuit breaker. [Background technology]

[0002] There is a growing need to remotely monitor the status of circuit breakers to perform maintenance without dispatching personnel. To meet this need, methods have been proposed in which various sensors are attached to circuit breakers to grasp their operating status, and the status is judged based on the difference from the normal state. In such diagnostics, attempts have been made to monitor the status by installing current sensors inside the circuit breaker or attaching acceleration sensors to the circuit breaker housing.

[0003] In newly constructed circuit breakers, it is possible to install such sensors for monitoring purposes from the design stage, but retrofitting sensors to existing circuit breakers not only increases the cost of the sensor itself and the processing costs involved in installation, but depending on the type of sensor, it may be difficult to install it in the circuit breaker casing. In particular, if the sensor needs to be installed inside the circuit breaker, installation may not be possible. For this reason, measurement using acoustics, which allows for non-contact measurement, is considered to be effective.

[0004] Patent Document 1 discloses a sound detector constructed to detect the sound emitted by an AC circuit breaker when the AC circuit breaker is switched on and off, and a diagnostic device that diagnoses whether the AC circuit breaker that emitted the sound is abnormal or not based on the sound detected by the sound detector and predetermined criteria information. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Publication No. 2019-184341 Summary of the Invention [Problem to be solved by the invention]

[0006] In Patent Document 1, normality and abnormality are determined by comparing sound information, but the timing of important events in the circuit breaker operation, such as the timing of contact movement due to the on / off operation of the circuit breaker, the timing of contact disengagement, and the timing of contact operation completion, is not taken into consideration.As a result, it is not possible to accurately grasp cases where operation occurs gradually or how the timing of events changes.

[0007] Furthermore, because circuit breaker operation is completed in a few dozen milliseconds, it is necessary to extract the timing of circuit breaker events within a very short time, such as within that few dozen milliseconds. However, because acoustic signals propagate at the speed of sound, the delay due to the distance between the microphone collecting the acoustic signal and the circuit breaker cannot be ignored.

[0008] An object of the present invention is to provide an acoustic diagnostic device and method that enable accurate determination of the state of a circuit breaker through acoustic diagnosis. [Means for solving the problem]

[0009] A diagnostic device according to one aspect of the present invention includes a microphone that converts sound pressure from a circuit breaker into an electrical signal, and an A / D converter that periodically converts the sound pressure converted by the microphone into a digital value to generate an acoustic signal. The acoustic diagnostic device further includes a central processing unit and a storage device. The central processing unit, during normal operation of the circuit breaker, extracts first feature quantities from acoustic signals collected by the microphone for various circuit breaker events, correcting delays in the acoustic signals resulting from the relative distance between the circuit breaker and the microphone, and creates multiple event models that manage data related to the first feature quantities of the corrected acoustic signals for each event. The storage device stores the multiple event models. During circuit breaker diagnosis, the central processing unit extracts second feature quantities from acoustic signals collected by the microphone, correcting delays in the acoustic signals resulting from the relative distance between the circuit breaker and the microphone, calculating similarities between the second feature quantities of the acoustic signals at the time of diagnosis and the first feature quantities of the multiple event models stored in the storage device, estimating the time at which the similarity is highest as the occurrence timing of each event, and diagnosing the circuit breaker based on the estimated occurrence timings. [Effects of the Invention]

[0010] According to one aspect of the present invention, it is possible to realize an acoustic diagnostic device that accurately diagnoses the state of a circuit breaker by using sound. [Brief explanation of the drawings]

[0011] [Figure 1] FIG. 1 is a block diagram illustrating an example of a configuration of an acoustic diagnostic apparatus. [Figure 2] FIG. 10 is a diagram illustrating an example of measurements during model generation. [Figure 3] FIG. 10 is a diagram illustrating an example of a model generation process and a diagnosis process. [Figure 4] FIG. 10 is a diagram illustrating an example of feature extraction processing. [Figure 5] FIG. 10 is a diagram illustrating an example of event timing estimation. [Figure 6] FIG. 10 is a diagram illustrating an example of stored data of each event model. [Figure 7] FIG. 10 is a diagram illustrating an example of stored data of an event model using a Gaussian mixture model for calculating similarity. [Figure 8] FIG. 10 is a diagram illustrating an example of diagnostic parameters. DETAILED DESCRIPTION OF THE INVENTION

[0012] In an embodiment, in an acoustic diagnostic device that diagnoses the state of a circuit breaker from an acoustic signal, an event model is created from true values ​​obtained in advance using an electrical sensor and feature quantities obtained by correcting acoustic delays based on the positions of the circuit breaker and a microphone that collects the acoustic signal. By correcting the delay, the feature quantities of the acoustic signal become synchronized with the true values ​​obtained using the electrical sensor. When diagnosing a circuit breaker, various timings during the opening and closing operations of the circuit breaker are accurately estimated based on information about the distance between the microphone that collects the acoustic signal and the circuit breaker, taking into account the delay of the acoustic signal, and the operating state of the circuit breaker is diagnosed based on the estimated timings.

[0013] For example, in one embodiment, an acoustic diagnostic device corrects the delay of an acoustic signal using a microphone whose distance from the circuit breaker is known, and synchronously collects the delay-corrected acoustic signal and measurements from a current sensor and a laser displacement meter that directly measures the stroke of the circuit breaker. The true value used during diagnosis is estimated from a sensor other than the acoustic signal. Meanwhile, the delay of the acoustic signal is calculated and corrected based on the distance between the circuit breaker and the microphone, and feature values ​​of the corrected acoustic signal are determined. Multiple event models are created to estimate the timing of various circuit breaker events. Each event model is data related to the feature values ​​of the acoustic signal that are used to compare the true values ​​of sensors other than the acoustic signal at the same timing when the circuit breaker is operating normally. During diagnosis, the delay in transmission of the acoustic signal is corrected based on position information of the microphone and the circuit breaker, and the similarity between the feature values ​​of the corrected acoustic signal and each event model is calculated. The timing of the feature value with the highest similarity is estimated as the timing of each event. The state of the circuit breaker is diagnosed using diagnostic parameters based on the estimated intervals between each event, etc. A circuit breaker can determine whether it is operating normally by diagnosing whether several important operations are performed at the specified timing, and one of its features is that this timing is used as information for diagnosis. Hereinafter, an embodiment will be described with reference to the drawings. [Example]

[0014] The configuration of the acoustic diagnostic apparatus according to the first embodiment will be described with reference to FIG. As shown in FIG. 1, the acoustic diagnostic device (10) includes a microphone (101) that converts the sound pressure from the circuit breaker into an electrical signal, an A / D converter (102) that converts the electrical signal (sound pressure) from the microphone into a digital data string, and, like a general information processing device, a main memory device (103), an auxiliary memory device (104), a central processing unit (105), an input device (106), and a result output device (107).

[0015] The auxiliary storage device (104) stores an event model (144) that takes into account the delay of the acoustic signal due to the electrical sensor and microphone position, information for determining the relative distance between the microphone and the circuit breaker, such as microphone position information (142), diagnostic parameters (146) for diagnosing based on the timing of each event, diagnostic software (130), and diagnostic history (150).

[0016] Circuit breaker events are the timing of important operations such as when the contacts start to move due to the on / off operation of the circuit breaker, when the contacts separate, and when the contact operation ends. Circuit breaker operations are completed in an extremely short time, typically just over a dozen milliseconds. For this reason, it is necessary to understand and correct the delay in the acoustic signal caused by the distance between the circuit breaker and the microphone.

[0017] Each event model (144) manages data on the feature quantities of the corrected acoustic signal for each event when the circuit breaker is in a normal state. For example, the data may be the average value of the feature quantity vector of the acoustic signal for each event of the circuit breaker, the median of the distribution of the feature quantity vector, or an actually observed feature quantity existing in the vicinity thereof. If the feature quantity vector does not have a simple distribution, a Gaussian mixture model, for example, may be used as the likelihood for the event model.

[0018] The current measurement microphone position information (142) is information indicating the relative distance between the circuit breaker and the microphone. If the circuit breaker installation position is known in advance as a coordinate position, the current measurement microphone position information may be the coordinate position information of the microphone.

[0019] The diagnostic parameters (146) are data for diagnosing the state of the circuit breaker, and may be given as thresholds such as the minimum and maximum values ​​of the interval between circuit breaker events. In this case, if the event timing estimated from the feature amount of the acoustic signal is within the range from the minimum to the maximum value of the event interval, it can be determined that the circuit breaker is operating normally. One of the features of this embodiment is that it focuses on the fact that it is possible to determine whether the circuit breaker is operating normally by diagnosing whether multiple important operations are performed at predetermined timings. The diagnostic parameters will be described later with reference to FIG. 8.

[0020] The diagnostic software (130) that performs overall control processing (111), delay calculation processing (112), feature extraction processing (113), similarity calculation processing (114), and condition diagnosis processing (115) is read from the auxiliary storage device (104) and stored in the main storage device (103). The main storage device (103) has the following work areas required by the diagnostic software (130): an acoustic waveform storage area (121), a delay calculation work area (122), a feature storage area (123), a model storage area (124), a similarity storage area (125), an event timing estimation result (126), a diagnostic parameter storage area (127), and a diagnostic result storage area (128).

[0021] The central processing unit (105) executes the program on the main memory, the input device (106) inputs parameters such as the microphone position, and the result output device (107) outputs the results.

[0022] The delay calculation process (112) calculates the difference in timing between an electric sensor such as a current sensor, a vibration sensor, or a laser displacement meter and an acoustic signal from a microphone as the amount of delay for various events of the circuit breaker.

[0023] The feature extraction process (113) is a process for extracting the average value of the feature of the acoustic signal related to various circuit breaker events and the feature existing near the center of the distribution of the feature, based on the delay-corrected acoustic signal. The feature extraction process will be described later with reference to FIG. 4.

[0024] The similarity calculation process (114) calculates the similarity for each time between the feature of the acoustic signal measured during diagnosis and the feature of the event model (144). Specifically, the occurrence time of each event is calculated from the maximum value of the similarity between the feature of the measured acoustic signal and the feature of the event model (144), and stored as an event timing estimation result (126). The event timing estimation process using the similarity calculation process will be described later with reference to FIG. 5.

[0025] The state diagnosis process (115) diagnoses the state of the circuit breaker based on the diagnostic parameters from the event timings estimated in the similarity calculation process (114).

[0026] The acoustic waveform storage area (121) is an area for storing the acoustic signals collected from the microphone (101).

[0027] The delay calculation work area (122) is a work area for the delay calculation process (112) to calculate the delay of the acoustic signal caused by the microphone (101).

[0028] The feature storage area (123) is an area for storing the feature of the acoustic signal extracted by the feature extraction process (113).

[0029] The model storage area 124 is a storage area for storing the feature quantities of the acoustic signals in various events of the circuit breaker as event models. Examples of the event models will be described later with reference to FIGS.

[0030] The similarity storage area (125) is an area for storing the calculation results of the similarity calculation process (114).

[0031] The event timing estimation result (126) is an area for storing the event timing estimation result obtained by the similarity calculation process (114). The diagnostic parameter storage area (127) is an area for storing diagnostic parameters used in the condition diagnosis process (115).

[0032] The diagnosis result storage area (128) is an area for storing the diagnosis results of the circuit breaker by the state diagnosis processing (115).

[0033] FIG. 2 is a diagram showing an example of measurements during model generation in the first embodiment.

[0034] First, the acoustic diagnostic device synchronizes the true values ​​of sensors (211, 212, 213) other than acoustic signals, such as electrical sensors including current sensors, vibration sensors, and laser displacement meters, with the acoustic signal of the microphone (101) to correct delays in the acoustic signal. Feature quantities of the corrected acoustic signal are used to generate event models (144) for various events of the circuit breaker during normal operation.

[0035] In order to accurately determine the timing of the on and off operations of the circuit breaker (201), electrical sensors (211, 212, 213) such as a current sensor, a vibration sensor, and a laser displacement meter are installed on the circuit breaker (201). The logger (220) has a function of A / D converting the acoustic signal collected by the microphone (101) when the circuit breaker (201) is operating and the signals from the sensors (211, 212, 213), and synchronizes the signals from the electrical sensors (211, 212, 213) and the microphone (101) to convert them into digital signals. A model generation computer (230) is connected to the circuit breaker (201) to generate each event model (144) using the distance (2142) between the circuit breaker (201) and the microphone (101). The model generation computer (230) performs the model generation process (3001) shown in FIG. 3 to create each event model (144). The model generation computer (230) mainly handles the delay calculation process (112) and feature extraction process (113) shown in FIG. 1, corrects the delay of the acoustic signal collected by the microphone (101), and creates the feature of the acoustic signal during normal operation of various events of the circuit breaker (201) as an event model.

[0036] FIG. 3 is a diagram for explaining the model generation process (3001) executed by the model generation computer (230) and the diagnostic process (3002) executed by the diagnostic software (130).

[0037] First, we will explain the model generation process (3001) that generates, as an event model, feature quantities of various circuit breaker acoustic signals at important event timings such as when the contacts start to move due to the on / off operation of the circuit breaker, when the contacts separate and engage, and when the contact operation ends. The event model generation process (3001) is mainly performed by the delay calculation process (112) and feature quantity extraction process (113) in Figure 1.

[0038] Sensor signal 1 (3101), sensor signal 2 (3102), and sensor signal 3 (3103) correspond to the electrical sensors (211, 212, 213) shown in Figure 2, and detect the vibrations and movements of the circuit breaker caused by various circuit breaker events without delay. On the other hand, the acoustic signal collected from the microphone (101) is delayed depending on the distance between the circuit breaker (201) and the microphone. This delay is calculated by the delay calculation process (3112) using the speed of sound to calculate the amount of delay between the electrical sensors (3101, 3102, 3103), and the timing of the acoustic signal and the electrical sensor signal is synchronized.

[0039] The acoustic signal collected by the microphone (101) is stored as acoustic sensor information (acoustic waveform (121)) (3121). Features are extracted from the stored acoustic signal by a feature extraction process (3113). Taking into account the delay of the acoustic signal, an event model can be generated from the features of the acoustic signal under normal conditions at timing synchronized with the electrical sensors (3101, 3102, 3103). For example, under normal conditions, the features of the acoustic signal collected from the microphone at the timing of important events such as the start of contact movement, the timing of contact disengagement, and the timing of contact operation completion due to the on / off operation of the circuit breaker are stored as an event model. In FIG. 3, features are extracted from the acoustic signal collected by the microphone, and the delay of the acoustic signal is input to model generation (3114) in delay calculation processing (3112), and a model (144) of each event is generated from the features of the acoustic signal synchronized with the electrical sensors (3101, 3102, 3103). However, after performing delay calculation processing 3112, the delay may be input to feature extraction processing (3113), and features may be obtained from the acoustic signal with the delay corrected.

[0040] An event model is created for each type of event of the circuit breaker (201). Therefore, each type of event is assigned an event ID to identify the event, and an event model is created for each event ID. The event models are managed as an event model correspondence table as shown in Figure 6, which will be described later.

[0041] FIG. 4 is a diagram for explaining the feature extraction process (3113) shown in FIG.

[0042] An acoustic waveform (401) is frequency analyzed (410) to generate a spectrogram (402) as a time series of short-time spectra. At this time, the window stride used to calculate the short-time spectra is set to a time shorter than 1 ms. A required frequency range is extracted from the obtained short-time spectra, and multiple frames are stacked to extract features as a feature vector (403). When creating features from a delay-corrected acoustic signal, the time of the spectrogram 402 is shifted by the delay calculated by the delay calculation process (112), thereby obtaining features from the corrected acoustic signal. Alternatively, when creating an event model from the feature vector (403), the event model may be created by shifting the delay calculated by the delay calculation process (112).

[0043] The diagnosis process (3002) in Fig. 3 will now be described. The diagnosis process (3002) is mainly performed by the delay calculation process (112), feature extraction process (113), similarity calculation process (114), and condition diagnosis process (115) in Fig. 1. The acoustic signal collected by the microphone (101) is stored as acoustic sensor information (acoustic waveform (121)), and features are extracted by the feature extraction process (113). The feature extraction procedure is the same as that of the model generation process (3001).

[0044] The delay of the acoustic signal is calculated by a delay calculation process (112) based on the microphone position (142) at the time of diagnosis. In a similarity calculation process (114), a model stored in each event model (144) is used to compare each event model with the feature of the acoustic signal at the time of diagnosis to calculate the similarity. The acoustic signal collected by the microphone (101) at the time of diagnosis is subjected to a delay calculation process (112) that calculates the delay of the acoustic signal based on the relative position of the microphone (101) and the circuit breaker (201), and is input to a similarity calculation process (114). In the similarity calculation process (114), the feature extracted from the acoustic signal at the time of diagnosis is shifted by the delay amount calculated in the delay calculation process (112) and compared with the feature of each event model (144). This comparison is performed for each time period, as shown in FIG. 4. The occurrence time of each event is calculated from the maximum similarity for each event, and the delay is corrected to obtain the event timing estimation result.

[0045] A circuit breaker can determine whether it is operating normally by diagnosing whether important operations are performed at predetermined timings.

[0046] FIG. 5 is a diagram illustrating the estimation of event timing.

[0047] In Figure 5, the horizontal axis represents time, and the vertical axis represents the similarity calculated by the similarity calculation process 114 in Figure 3. The similarity is calculated by comparing an event model with the acoustic signal of the circuit breaker being diagnosed, and the value increases if there is a high degree of similarity. By comparing the event model for each event, Event 1, Event 2, and Event 3 of the circuit breaker, with the feature quantities of the acoustic signal of the diagnostic device being diagnosed, it is possible to determine the occurrence time of each event at which the similarity is greatest.

[0048] For example, the occurrence time of event 1 can be estimated as time 511, the occurrence time of event 2 as time 512, and the occurrence time of event 3 as time 513.

[0049] In this way, the occurrence timing of each event can be accurately estimated by taking into account the delay of the acoustic signal, synchronizing with the electrical sensors (3101, 3102, 3103), comparing the event model under normal conditions with the feature of the acoustic signal during diagnosis, and calculating the similarity. A circuit breaker can determine whether it is operating normally by diagnosing whether important operations are occurring at the specified timing. Therefore, in the diagnosis process (115), as shown in Figure 8, thresholds (804) for the range of maximum and minimum values ​​of the interval between specific events of the circuit breaker are given as diagnostic parameters for the interval between specific events and the time interval between the trigger signal and each event. The diagnosis is made based on whether the interval falls within this range. If there is a limit on only one side of the maximum or minimum value, for example, the maximum value can be set to +∞ and the minimum value to -∞ to represent no constraint.

[0050] FIG. 6 shows an example of an event model representation (event model correspondence table) of the feature values ​​for each event. As shown in FIG. 6, the event ID that identifies each event is managed in correspondence with each event model. The event model shown in FIG. 6 can be implemented when the feature values ​​corresponding to an event are stable, and is represented by a single representative value of a simple feature value. This can be calculated using, for example, the median or mean value of the distribution when the model was created. Similarity can be calculated using cosine distance or Euclidean distance. [Example]

[0051] Figure 7 shows an example of an event model where the feature quantities for each event are expressed using a Gaussian mixture model. A feature quantity model (602) is managed corresponding to each event ID (601) as an event model. The feature quantities have the mean value, covariance matrix, and weight ω of n_component features, and are considered similar by calculating the likelihood for each time.

[0052] The relationship between the feature values ​​extracted in advance from the measurement data and the state of the device at the time of measurement may be represented by reference points appropriately placed in the feature value space, and model parameters that quantify the state corresponding to the data near the feature values ​​relative to the reference points. [Explanation of symbols]

[0053] 101 Microphone 102 A / D conversion 103 Main storage 104 Auxiliary storage 105 Central Processing Unit 106 Input Device 107 Result output device 111 Overall control processing 112 Delay calculation processing 113 Feature Extraction Processing 114 Similarity calculation processing 121 Acoustic Waveform 122 Delay Calculation Work 123 Feature storage area 124 model storage area 125 Similarity storage area 126 Event timing estimation results 127 Diagnostic parameter storage area 128 Diagnostic result storage area 130 Diagnostic Software 142 Microphone Position 144 Event Models 146 Diagnostic Parameters 150 Diagnostic History 201 Circuit Breaker 211 Electrical Sensor 1 212 Electrical Sensor 2 213 Electrical Sensors 3 2142 Microphone position 220 Logger 230 Model Generation Calculator 3001 Model generation process 3002 Diagnostic Processing 3101 Sensor signal 1 3102 Sensor signal 2 3103 Sensor signal 3 3112 Delay calculation processing 3121 Acoustic Waveform 3113 Feature Extraction Processing 3114 Model Creation 401 Acoustic Waveform 402 Spectrogram 403 Feature Vector 410 Frequency Analysis 420 Frame Stacking 501 Event 1 Similarity 502 Event 2 Similarity 503 Event 3 Similarity 511 Event 1 timing estimate 512 Event 2 Timing Estimates 513 Event 3 Timing Estimate 601 Event ID 602 Feature Model 702 Feature Model (Gaussian Mixture Model) 801 Diagnostic parameter number 802 Diagnostic Event ID 1 803 Diagnostic Event ID 2 804 Diagnostic threshold information

Claims

1. A microphone that converts the sound pressure from the circuit breaker into an electrical signal, an A / D converter that converts the sound pressure converted into an electric signal by the microphone into a digital value at a constant period to generate an acoustic signal, The acoustic diagnostic device further includes a central processing unit and a storage device, The central processing unit extracting a first feature amount of an acoustic signal collected by the microphone for various events of the circuit breaker during normal operation of the circuit breaker, the acoustic signal being corrected for a delay in the acoustic signal caused by a relative distance between the circuit breaker and the microphone; creating a plurality of event models that manage data relating to the first feature amount of the corrected acoustic signal for each event; the storage device stores the plurality of event models; the plurality of event models stored in the storage device, An event model obtained by extracting a first feature of an acoustic signal collected by the microphone and correcting a delay of the acoustic signal resulting from a relative distance between the circuit breaker and the microphone is managed in association with an event ID that identifies various events of the circuit breaker, and a feature represented by a Gaussian mixture model is managed in association with each event; The central processing unit extracting a second feature amount from the acoustic signal collected from the microphone during diagnosis of the circuit breaker and correcting a delay in the acoustic signal caused by a relative distance between the circuit breaker and the microphone; calculating a similarity for each time between a second feature amount of the acoustic signal at the time of diagnosis and a first feature amount of the plurality of event models stored in the storage device; The time when the similarity is highest is estimated as the occurrence timing of each event, Diagnose the circuit breaker based on the estimated occurrence timings. An acoustic diagnostic device characterized by:

2. A microphone that converts the sound pressure from the circuit breaker into an electrical signal, an A / D converter that converts the sound pressure converted into an electric signal by the microphone into a digital value at a constant period to generate an acoustic signal, The acoustic diagnostic device further includes a central processing unit and a storage device, The central processing unit extracting a first feature which is an average value of a feature vector of an acoustic signal collected by the microphone for various events of the circuit breaker during normal operation of the circuit breaker and corrected for a delay in the acoustic signal caused by a relative distance between the circuit breaker and the microphone, or a median value of a distribution of the feature vector; creating a plurality of event models that manage data relating to the first feature amount of the corrected acoustic signal for each event; the storage device stores the plurality of event models; The central processing unit During diagnosis of the circuit breaker, a second feature is extracted, which is an average value of a feature vector of the acoustic signal collected from the microphone and corrected for a delay of the acoustic signal caused by a relative distance between the circuit breaker and the microphone, or a median value of a distribution of the feature vector; calculating a similarity for each time between a second feature amount of the acoustic signal at the time of diagnosis and a first feature amount of the plurality of event models stored in the storage device; The time when the similarity is highest is estimated as the occurrence timing of each event, Diagnose the circuit breaker based on the estimated occurrence timings. An acoustic diagnostic device characterized by:

3. The acoustic diagnostic apparatus according to claim 1 or 2, The central processing unit a signal input from an electrical sensor provided in the circuit breaker and a digital value of the microphone are synchronized, and a delay of the acoustic signal collected by the microphone relative to the signal input from the electrical sensor is calculated; An acoustic diagnostic device characterized by:

4. The acoustic diagnostic apparatus according to claim 1 or 2, the storage device stores diagnostic parameters for managing maximum and minimum values ​​of intervals between specific events of the circuit breaker as thresholds; The central processing unit diagnosing the circuit breaker based on the estimated occurrence timings and the diagnostic parameters; An acoustic diagnostic device characterized by:

5. A microphone that converts the sound pressure from the circuit breaker into an electrical signal, an A / D converter that converts the sound pressure converted into an electric signal by the microphone into a digital value at a constant period to generate an acoustic signal, a first feature extraction processing unit that extracts feature values ​​of acoustic signals collected by the microphone for various events of the circuit breaker during normal operation of the circuit breaker, and that corrects delays in the acoustic signals resulting from a relative distance between the circuit breaker and the microphone; a model creation unit that creates a plurality of event models that manage data relating to feature quantities of the corrected acoustic signal for each event; an event model storage unit that stores the plurality of event models; The plurality of event models stored in the event model storage unit For each event ID identifying various events of the circuit breaker, an event model obtained by extracting a first feature amount of the acoustic signal collected by the microphone and correcting a delay of the acoustic signal caused by the relative distance between the circuit breaker and the microphone is associated and managed, and for each event, a feature amount indicated by a Gaussian mixture model is associated and managed, a second feature extraction processing unit that extracts, during diagnosis of the circuit breaker, feature amounts of the acoustic signal collected from the microphone and corrected for a delay in the acoustic signal caused by a relative distance between the circuit breaker and the microphone; a similarity calculation processing unit that calculates a similarity for each time between a feature amount of the acoustic signal at the time of diagnosis and the feature amounts of the plurality of event models stored in the event model storage unit, and estimates the time at which the similarity is highest as the occurrence timing of each event; a diagnosis processing unit that diagnoses the circuit breaker based on the estimated occurrence timings, An acoustic diagnostic device characterized by:

6. A microphone that converts the sound pressure from the circuit breaker into an electrical signal, an A / D converter that converts the sound pressure converted into an electric signal by the microphone into a digital value at a constant period to generate an acoustic signal, a first feature extraction processing unit that extracts an average value of feature vectors of acoustic signals collected by the microphone for various events of the circuit breaker during normal operation of the circuit breaker, and that corrects for delays in the acoustic signals resulting from a relative distance between the circuit breaker and the microphone, or a median value of a distribution of the feature vectors; a model creation unit that creates a plurality of event models that manage data relating to feature quantities of the corrected acoustic signal for each event; an event model storage unit that stores the plurality of event models; a second feature extraction processing unit that extracts, during diagnosis of the circuit breaker, an average value of feature vectors of the acoustic signal collected from the microphone and corrected for a delay in the acoustic signal caused by a relative distance between the circuit breaker and the microphone, or a median value of a distribution of the feature vectors; a similarity calculation processing unit that calculates a similarity for each time between a feature amount of the acoustic signal at the time of diagnosis and the feature amounts of the plurality of event models stored in the event model storage unit, and estimates the time at which the similarity is highest as the occurrence timing of each event; a diagnosis processing unit that diagnoses the circuit breaker based on the estimated occurrence timings, An acoustic diagnostic device characterized by:

7. An acoustic diagnostic method for an acoustic diagnostic device having a microphone that converts sound pressure from a circuit breaker into an electrical signal, and an A / D converter that periodically converts the sound pressure converted into an electrical signal by the microphone into a digital value to generate an acoustic signal, a central processing unit of the acoustic diagnostic device extracts feature quantities of acoustic signals collected by the microphone for various events of the circuit breaker during normal operation of the circuit breaker, and corrects delays in the acoustic signals resulting from the relative distance between the circuit breaker and the microphone; creating a plurality of event models that manage data relating to the feature quantities of the corrected acoustic signals for each event; a storage device of the acoustic diagnostic apparatus stores the plurality of event models created by the central processing unit; the plurality of event models stored in the storage device, For each event ID identifying various events of the circuit breaker, an event model obtained by extracting a first feature amount of the acoustic signal collected by the microphone and correcting a delay of the acoustic signal caused by the relative distance between the circuit breaker and the microphone is associated and managed, and for each event, a feature amount indicated by a Gaussian mixture model is associated and managed, the central processing unit, during diagnosis of the circuit breaker, extracts a feature amount of the acoustic signal collected from the microphone and corrected for a delay in the acoustic signal caused by a relative distance between the circuit breaker and the microphone; calculating a similarity for each time between a feature of the acoustic signal at the time of diagnosis and a feature of the plurality of event models stored in the storage device, and estimating the time at which the similarity is highest as the occurrence timing of each event; and diagnosing the circuit breaker based on the estimated occurrence timings. An acoustic diagnostic method comprising:

8. An acoustic diagnostic method for an acoustic diagnostic device having a microphone that converts sound pressure from a circuit breaker into an electrical signal, and an A / D converter that periodically converts the sound pressure converted into an electrical signal by the microphone into a digital value to generate an acoustic signal, a central processing unit of the acoustic diagnostic device extracts an average value of feature vectors of acoustic signals collected by the microphone for various events of the circuit breaker during normal operation of the circuit breaker, and corrects delays of the acoustic signals resulting from the relative distance between the circuit breaker and the microphone, or a median value of a distribution of feature vectors; creating a plurality of event models that manage data relating to the feature quantities of the corrected acoustic signals for each event; a storage device of the acoustic diagnostic apparatus stores the plurality of event models created by the central processing unit; the central processing unit, during diagnosis of the circuit breaker, extracts a feature amount of the acoustic signal collected from the microphone and corrected for a delay in the acoustic signal caused by a relative distance between the circuit breaker and the microphone; calculating a similarity for each time between a feature of the acoustic signal at the time of diagnosis and a feature of the plurality of event models stored in the storage device, and estimating the time at which the similarity is highest as the occurrence timing of each event; and diagnosing the circuit breaker based on the estimated occurrence timings. An acoustic diagnostic method comprising:

Images