Method, device and system for detecting abnormal sound of household appliance

Abstract

The invention relates to the technical field of household appliance manufacturing, and provides a household appliance abnormal sound detection method, device and system, and the method comprises the steps: obtaining at least one to-be-detected vibration signal during the operation of a to-be-detected household appliance; generating a spectrogram to be detected corresponding to each vibration signal to be detected; respectively inputting each spectrogram to be detected into an abnormal sound detection model to obtain an abnormal sound detection result output by the abnormal sound detection model; according to the invention, the abnormal sound detection model is obtained by training based on the sample spectrogram and the abnormal sound type label corresponding to the sample spectrogram, and the sample spectrogram is generated based on the sample vibration signal, so that the abnormal sound detection model carries out image recognition on each spectrogram to be detected; various abnormal sounds of the household appliance to be detected can be accurately detected, the abnormal sound type can be accurately determined, and the detection result has good consistency compared with manual sound listening detection.

Description

Technical Field

[0001] The present invention relates to the technical field of household appliance manufacturing, and in particular to a method, device and system for detecting abnormal sounds of household appliances. Background Art

[0002] Common household appliances include refrigerators, air conditioners, microwave ovens and other equipment. These devices will produce abnormal sounds during use, which may be early warning signs of equipment failure.

[0003] For example, a microwave oven uses microwave radiation to heat water molecules in food, causing them to vibrate rapidly to generate heat, thereby achieving rapid heating. Normally, when using a microwave oven, you simply need to set the heating time and power to start the heating program. A motor-driven turntable or stirrer is installed inside the microwave oven to ensure that the food is heated evenly. However, during use, the microwave oven may have abnormal sound problems such as aging of the magnetron, abnormal motor, loose door seals and / or screws, and electromagnetic failure. These abnormal sounds may be early warning signs of equipment failure and will also affect the user experience.

[0004] At present, abnormal sound detection of household appliances is mostly achieved through manual listening, that is, technicians listen to the sound to determine whether there is abnormal sound when the household appliances are working. However, long-term listening can easily cause fatigue, affect the detection quality, and lead to inaccurate detection results. Moreover, manual listening is highly subjective, and manual experience varies greatly. Different people may get different detection results for the same device, and the consistency of the detection results is poor, and the detection standards cannot be unified. Summary of the invention

[0005] The present invention provides a method, device and system for detecting abnormal sounds of household appliances, which are used to solve the technical problems in the prior art that artificial listening detection leads to inaccurate detection of abnormal sounds of household appliances and poor consistency of detection results.

[0006] The present invention provides a method for detecting abnormal sound of household appliances, comprising the following steps: Acquire at least one vibration signal to be tested when the household appliance to be tested is running; Generating a spectrogram to be tested corresponding to each of the vibration signals to be tested; Inputting each of the spectrograms to be tested into an abnormal sound detection model respectively to obtain an abnormal sound detection result output by the abnormal sound detection model; The abnormal sound detection model is trained based on a sample spectrogram and an abnormal sound type label corresponding to the sample spectrogram, and the sample spectrogram is generated based on a sample vibration signal.

[0007] According to a method for detecting abnormal sounds of household appliances provided by the present invention, a spectrogram to be tested corresponding to each vibration signal to be tested is generated, comprising: Performing short-time Fourier transform on each of the vibration signals to be measured, so as to convert each of the vibration signals to be measured from a time domain signal to a frequency domain signal; Determine the amplitude spectrum of each time window corresponding to the short-time Fourier transform in each of the frequency domain signals; For each frequency domain signal, the amplitude spectra of all time windows are arranged in time order to generate the spectrograms to be tested.

[0008] According to a method for detecting abnormal sounds of household appliances provided by the present invention, short-time Fourier transform is performed on each of the vibration signals to be detected to convert each of the vibration signals to be detected from a time domain signal to a frequency domain signal, including: Each of the vibration signals to be measured is divided into a plurality of time windows, and a windowing process is performed between two adjacent time windows; The windowed vibration signals are subjected to fast Fourier transformation to obtain frequency components in each time window, so as to convert the vibration signals to be measured from time domain signals to frequency domain signals.

[0009] According to a method for detecting abnormal sounds of household appliances provided by the present invention, for each frequency domain signal, after arranging the amplitude spectra of all time windows in time order to generate each of the spectrograms to be tested, the method further includes: The logarithm of the amplitude spectrum of each of the spectrograms to be tested is taken to obtain each spectrogram to be tested after logarithmic coordinate transformation. The spectrogram to be tested input into the abnormal sound detection model is the spectrogram to be tested after logarithmic coordinate transformation. When the abnormal sound detection model is trained, the sample spectrogram is the sample spectrogram after logarithmic coordinate transformation.

[0010] According to a method for detecting abnormal sound of household appliances provided by the present invention, the abnormal sound detection model training process is as follows: Inputting the sample spectrogram into an initial convolutional neural network model to obtain an abnormal sound prediction result output by the initial convolutional neural network model; The abnormal sound prediction result and the abnormal sound type label corresponding to the sample spectrogram are substituted into the loss function. When the loss function converges, the training is completed to obtain the abnormal sound detection model.

[0011] According to a method for detecting abnormal sound of household appliances provided by the present invention, after obtaining at least one vibration signal to be tested when the household appliance to be tested is running, the method further includes: Abnormal noise of the motor is detected based on the vibration characteristics of a vibration signal of the motor to be tested in each rotation cycle of the motor in the household appliance, wherein the vibration signal of the motor to be tested is collected based on a vibration sensor located in a corresponding area of ​​the motor.

[0012] According to a method for detecting abnormal sound of a household appliance provided by the present invention, based on the vibration characteristics of a vibration signal of a motor to be tested in each rotation cycle of the motor in the household appliance, abnormal sound of the motor is detected, comprising: Performing spectrum analysis on the vibration signal of the motor to be tested within each rotation cycle to determine the target frequency band of the sudden change of vibration energy; The first target number of cycles in which the vibration energy of the target frequency band exceeds the energy threshold is counted, and when the ratio of the first target number of cycles to the total number of cycles reaches a first preset ratio threshold, it is determined that the abnormal sound of the motor is the abnormal sound caused by friction between the motor rotor and foreign matter inside the motor, wherein the total number of cycles is determined based on the rotation period of the motor and the total duration of the vibration signal of the motor to be measured.

[0013] According to a method for detecting abnormal sound of a household appliance provided by the present invention, based on the vibration characteristics of a vibration signal of a motor to be tested in each rotation cycle of the motor in the household appliance, abnormal sound of the motor is detected, comprising: Performing kurtosis analysis on the vibration signal of the motor to be measured in each rotation cycle to obtain the kurtosis value in each rotation cycle; The second target number of cycles in which the kurtosis value exceeds the kurtosis threshold is counted, and when the ratio of the second target number of cycles to the total number of cycles reaches a second preset ratio threshold, the abnormal sound of the motor is determined to be the abnormal sound of the motor bearing, wherein the total number of cycles is determined based on the rotation period of the motor and the total duration of the vibration signal of the motor to be measured.

[0014] The present invention also provides a household appliance abnormal sound detection device, comprising the following modules: A vibration signal acquisition module, used to acquire at least one vibration signal to be tested when the home appliance to be tested is running; A spectrogram generating module, used for generating a spectrogram to be tested corresponding to each vibration signal to be tested; A model execution module, used for inputting each of the spectrograms to be tested into an abnormal sound detection model to obtain an abnormal sound detection result output by the abnormal sound detection model; The abnormal sound detection model is trained based on a sample spectrogram and an abnormal sound type label corresponding to the sample spectrogram, and the sample spectrogram is generated based on a sample vibration signal.

[0015] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the program, the method for detecting abnormal sounds of household appliances as described in any one of the above is implemented.

[0016] The present invention also provides a household appliance abnormal sound detection system, comprising: an assembly line body, a lifting mechanism, a vibration detection mechanism and the above-mentioned electronic equipment; The assembly line body is provided with a tooling plate for carrying the household appliance to be tested; The lifting mechanism is located below the assembly line body in a non-contact manner with the assembly line body and is used to lift the tooling plate; The vibration detection mechanism is located above the lifting mechanism, and is used to generate at least one vibration signal to be detected of the household appliance to be detected to the electronic device.

[0017] A household appliance abnormal sound detection system provided according to the present invention also includes a mounting frame, which includes: a column and a transverse support plate, the transverse support plate can be installed on the column axially movably along the column, the column is installed on the lifting mechanism, and the vibration detection mechanism is installed on the transverse support plate with adjustable position.

[0018] According to a household appliance abnormal sound detection system provided by the present invention, the vibration detection mechanism includes: a vibration sensor, a mounting seat, a slider and a cylinder, the mounting seat is provided with a slide rail, the slider is mounted on the slide rail, the vibration sensor is adjustable on the slider, the cylinder is mounted on the mounting seat, the piston rod of the cylinder is connected to the slider, and the mounting seat is adjustable on the transverse support plate.

[0019] According to a household appliance abnormal sound detection system provided by the present invention, the piston rod of the cylinder is connected to the slider via an elastic component.

[0020] According to a household appliance abnormal sound detection system provided by the present invention, the lateral support plate includes: a first support sub-plate, a second support sub-plate and a third support sub-plate, the first support sub-plate can be installed on the column axially movably along the column, the second support sub-plate is movably connected to one end of the first support sub-plate, and the third support sub-plate is movably connected to the other end of the first support sub-plate.

[0021] The present invention also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the method for detecting abnormal sounds of household appliances as described in any one of the above is implemented.

[0022] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method for detecting abnormal sounds of household appliances as described above is implemented.

[0023] The method, device and system for detecting abnormal sounds of household appliances provided by the present invention obtain at least one vibration signal to be tested when the household appliance to be tested is running; generate a spectrogram to be tested corresponding to each of the vibration signals to be tested; and input each of the spectrograms to be tested into an abnormal sound detection model to obtain an abnormal sound detection result output by the abnormal sound detection model. Since the abnormal sound detection model is trained based on sample spectrograms and abnormal sound type labels corresponding to the sample spectrograms, and the sample spectrograms are generated based on sample vibration signals, the abnormal sound detection model can accurately detect various abnormal sounds of the household appliance to be tested and accurately determine the type of abnormal sounds by performing image recognition on each spectrogram to be tested, and the detection results have better consistency than those of manual listening detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0025] Figure 1 It is a flow chart of the method for detecting abnormal sound of household appliances provided by the present invention.

[0026] Figure 2 It is a schematic diagram of a spectrogram in the method for detecting abnormal sounds of household appliances provided by the present invention.

[0027] Figure 3 It is a structural schematic diagram of the abnormal sound detection device for household appliances provided by the present invention.

[0028] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention.

[0029] Figure 5 This is one of the structural schematic diagrams of the household appliance abnormal sound detection system provided by the present invention.

[0030] Figure 6 This is the second structural schematic diagram of the household appliance abnormal sound detection system provided by the present invention.

[0031] Figure 7 This is the third structural schematic diagram of the household appliance abnormal sound detection system provided by the present invention.

[0032] Figure 8 This is the fourth structural schematic diagram of the household appliance abnormal sound detection system provided by the present invention. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0034] It should be noted that in the description of the present invention, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "include one..." do not exclude the existence of other identical elements in the process, method, article or device including the elements. The orientation or position relationship indicated by the terms "upper", "lower" and the like is based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present invention. Unless otherwise clearly specified and limited, the terms "installed", "connected" and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be a connection between the two elements. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0035] The terms "first", "second", etc. in the present invention are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present invention can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are generally of the same type, and the number of objects is not limited. For example, the first object can be one or more.

[0036] In the related art, abnormal sound detection of home appliances mainly relies on manual listening detection. However, long-term listening can easily cause fatigue, which affects the detection quality and leads to inaccurate detection results. In addition, the manual listening detection method is highly subjective and manual experience varies greatly. Different people may get different detection results for the same home appliance product, and the consistency of the detection results is poor, and the detection standards cannot be unified.

[0037] In view of the above technical problems existing in the prior art, an embodiment of the present invention provides a method for detecting abnormal sound of household appliances, such as Figure 1 As shown, the following steps S110 to S130 are included.

[0038] Step S110: obtaining at least one vibration signal to be measured when the home appliance to be measured is in operation, wherein the home appliance to be measured may be a common home appliance such as a refrigerator, a washing machine, an air conditioner and a microwave oven. When the home appliance to be measured is in operation, a vibration sensor (such as an accelerometer) is used to contact the home appliance to be measured to collect the vibration signal of the home appliance to be measured. In this step, by obtaining the vibration signal of the home appliance to be measured collected by the vibration sensor when the home appliance to be measured is in operation, at least one vibration signal to be measured is obtained.

[0039] For any household appliance to be tested, there are many situations that cause abnormal sounds, for example: microwave ovens, magnetron aging, motor abnormalities (or failures), loose door seals and / or screws, and electromagnetic failures, etc., all of which will produce abnormal sounds. Therefore, when at least one vibration signal to be tested is a plurality of vibration signals to be tested, each vibration signal to be tested is acquired based on vibration sensors located in the corresponding areas of each component of the household appliance to be tested that causes abnormal sounds. For example: a vibration sensor is provided on the outer wall of the corresponding area of ​​the magnetron, motor, and door seal and / or screw of the microwave oven. Each vibration sensor simultaneously acquires the vibration signals to be tested of each component, so that all abnormal sounds of the household appliance to be tested can be detected at one time through a subsequent abnormal sound detection model.

[0040] It is understandable that: since the vibration of the home appliance to be tested is detected during operation, the home appliance to be tested needs to be placed on a static platform to avoid interference of external vibration on vibration detection, so that each vibration signal to be tested does not contain external vibration components.

[0041] It should be noted that the vibration signal to be tested can be the vibration signal of the household appliance to be tested under different operating conditions. For example, for a microwave oven, at least one vibration signal to be tested can be collected under the three operating conditions of high heat, medium heat and low heat of the microwave oven, thereby achieving comprehensive detection of abnormal sounds of the household appliance to be tested under different operating conditions.

[0042] Step S120: Generate a spectrogram to be tested corresponding to each vibration signal to be tested. Figure 2 As shown, the horizontal axis of the spectrogram is time, the vertical axis is the frequency of the vibration signal, and the value of the coordinate point is the energy of the vibration signal. Since a two-dimensional plane is used to express three-dimensional information, the size of the energy value is represented by the brightness of the color. The greater the brightness of the color, the stronger the vibration energy of the coordinate point. In this step, by generating a spectrogram to be tested corresponding to each vibration signal to be tested, each vibration signal to be tested is converted into an RGB color image, so that the artificial intelligence model based on image detection can be used to recognize the spectrogram to be tested, so as to realize the abnormal sound detection of the household appliance to be tested.

[0043] Step S130: Input each of the spectrograms to be tested into the abnormal sound detection model respectively, and obtain the abnormal sound detection result output by the abnormal sound detection model. The abnormal sound detection model is obtained by training based on the sample spectrogram and the abnormal sound type label corresponding to the sample spectrogram, and the sample spectrogram is generated based on the sample vibration signal. The abnormal sound detection model can be an artificial intelligence model based on a convolutional neural network (CNN), which extracts features from the sample spectrogram and, with the powerful ability of deep learning, can perform deep feature representation of image features converted from these signal features. After training with a large number of sample spectrograms and corresponding abnormal sound type labels, the convolutional neural network gradually learns how to map the input sample spectrogram to the corresponding abnormal sound type label. After completing the training, the obtained abnormal sound detection model can perform inference based on the spectrograms to be tested corresponding to each vibration signal to be tested, so as to detect the abnormal sound type of the household appliance to be tested, and thus determine the cause of the abnormal sound. Since each vibration signal to be tested is collected based on vibration sensors located in the corresponding areas of each component of the home appliance to be tested that causes abnormal sound, after any component fails, the corresponding vibration signal to be tested and the sample vibration signal when any component fails have highly similar signal characteristics, and their corresponding spectrograms have highly similar image features. Therefore, when inferring, the abnormal sound detection model will give the detection result of the abnormal sound caused by any component.

[0044] Specifically, taking a microwave oven as an example, the aging of the magnetron, abnormal motor (or failure), loose door seals and / or screws, and electromagnetic failure of the microwave oven will all produce abnormal sounds, that is, different types of abnormal sounds. Before training the abnormal sound detection model, sample data and corresponding labels are prepared, and five types of microwave oven samples are selected, namely, normal (no abnormal sound) microwave ovens, microwave ovens with aging magnetrons, microwave ovens with abnormal motors, microwave ovens with loose door seals and / or screws, and microwave ovens with electromagnetic failures. There are several microwave ovens of each sample, and it is best if the data of each type of microwave oven sample is equal. For each type of microwave oven sample, a sample vibration signal is collected and a sample spectrogram corresponding to the sample vibration signal is generated. The five types of sample spectrograms are marked with abnormal sound type labels. For example, the abnormal sound type label of the sample spectrogram corresponding to a normal microwave oven is 0, the abnormal sound type label of the sample spectrogram corresponding to a microwave oven with an aging magnetron is 1, the abnormal sound type label of the sample spectrogram corresponding to a microwave oven with an abnormal motor is 2, the abnormal sound type label of the sample spectrogram corresponding to a microwave oven with a loose door seal and / or screws is 3, and the abnormal sound type label of the sample spectrogram corresponding to a microwave oven with an electromagnetic fault is 4. The convolutional neural network is trained using these five types of sample spectrograms and the corresponding abnormal sound type labels. During the training process, different types of sample spectrograms are mapped to corresponding abnormal sound type labels. After the training is completed, an abnormal sound detection model is obtained. During inference, the trained abnormal sound detection model outputs the abnormal sound type to which each spectrogram to be tested belongs as the abnormal sound detection result.

[0045] The abnormal sound detection method of the household appliance of the embodiment of the present invention obtains at least one vibration signal to be tested when the household appliance to be tested is running, generates a spectrogram to be tested corresponding to each of the vibration signals to be tested, and inputs each of the spectrograms to be tested into an abnormal sound detection model to obtain an abnormal sound detection result output by the abnormal sound detection model. Since the abnormal sound detection model is trained based on sample spectrograms and abnormal sound type labels corresponding to the sample spectrograms, and the sample spectrograms are generated based on sample vibration signals, the abnormal sound detection model can accurately detect various abnormal sounds of the household appliance to be tested and accurately determine the type of abnormal sound by performing image recognition on each spectrogram to be tested, and the detection results have good consistency with respect to manual listening detection.

[0046] In some embodiments, step S120 specifically includes step S121 and step S123.

[0047] Step S121: Perform short-time Fourier transform (STFT) on each of the vibration signals to be measured to convert each of the vibration signals to be measured from a time domain signal to a frequency domain signal. Preferably, wavelet analysis is performed on each frequency domain signal to remove clutter and burrs in the waveform corresponding to the frequency domain signal, so as to obtain a more accurate frequency domain signal.

[0048] Step S122: determining the amplitude spectrum of each time window corresponding to the short-time Fourier transform in each of the frequency domain signals, that is, the vibration intensity of the frequency component corresponding to each time window in the frequency domain signal.

[0049] Step S123: for each frequency domain signal, the amplitude spectra of all time windows are arranged in time order to generate each of the spectrograms to be tested. Each spectrogram to be tested can be drawn by a visualization tool, such as the matplotlib library in Python. For example, the drawn spectrogram to be tested is as follows: Figure 2 shown.

[0050] In this embodiment, each vibration signal to be tested is subjected to short-time Fourier transform respectively, so as to quickly and accurately generate a corresponding spectrogram to be tested.

[0051] In some embodiments, the steps of performing short-time Fourier transform on each of the vibration signals to be measured to convert each of the vibration signals to be measured from a time domain signal to a frequency domain signal specifically include: Each of the vibration signals to be measured is divided into a plurality of time windows, and a windowing process is performed between two adjacent time windows. Specifically, first, each vibration signal to be measured is divided into a plurality of time windows (short time windows). Since the vibration signal to be measured is a continuous time domain signal, each vibration signal to be measured can be divided into a plurality of time windows by segmenting the time axis according to the length of the time window. Secondly, a windowing process is performed on the vibration signal between two adjacent time windows, such as a Hanning window.

[0052] The vibration signals to be measured after the windowing process are subjected to a fast Fourier transform (Fast Fourier Transform, FFT) to obtain the frequency components in each time window, so as to convert the vibration signals to be measured from time domain signals to frequency domain signals. Specifically, the vibration signals in each time window after the windowing process are subjected to a fast Fourier transform to obtain the frequency components in each time window, thereby obtaining the frequency domain signals of the vibration signals to be measured.

[0053] In this embodiment, during the short-time Fourier transform process, the vibration signal in each time window is windowed to reduce the boundary effect between frames and avoid the gap between two adjacent time windows that causes the omission of some key signal features, so that the finally generated spectrograms to be tested are more accurate.

[0054] In some embodiments, for each frequency domain signal, after the amplitude spectra of all time windows are arranged in chronological order to generate each of the spectrograms to be tested, the method further includes: taking the logarithm of the amplitude spectrum of each of the spectrograms to be tested to obtain each spectrogram to be tested after logarithmic coordinate transformation, wherein the logarithmic transformation can be performed with a base of 10 or a natural constant e as the base. The spectrogram to be tested input to the abnormal sound detection model is the spectrogram to be tested after logarithmic coordinate transformation, and when the abnormal sound detection model is trained, the sample spectrogram is the sample spectrogram after logarithmic coordinate transformation.

[0055] In this embodiment, since the logarithmic coordinate transformation can compress the dynamic range of the data, the data details of the spectrogram to be tested and various types of sample spectrograms are clearer after logarithmic coordinate transformation, that is, the characteristics of the vibration signal can be better displayed. In the training stage of the abnormal sound detection model, features can be extracted faster and more accurately from the sample spectrogram after logarithmic coordinate transformation, and training can be performed, so that the trained abnormal sound detection model has more accurate abnormal sound detection results for the tested home appliances.

[0056] In some embodiments, the abnormal sound detection model training process is as follows: the sample spectrogram is input into an initial convolutional neural network model to obtain an abnormal sound prediction result output by the initial convolutional neural network model; the abnormal sound prediction result and the abnormal sound type label corresponding to the sample spectrogram are substituted into a loss function, for example: a cross-entropy loss function (Cross-Entropy Loss). When the loss function converges, the training is completed and the abnormal sound detection model is obtained.

[0057] The above training process is an iterative process. Specifically, the training process includes iteratively executing the following steps 1 and 2.

[0058] Step 1: Input the current sample spectrogram into the initial convolutional neural network model, the initial convolutional neural network model outputs the abnormal sound prediction result of the current sample spectrogram, and substitutes the abnormal sound prediction result and the abnormal sound type label corresponding to the sample spectrogram into the loss function.

[0059] Step 2: When the loss function converges, the training is completed and the abnormal sound detection model is obtained. Otherwise, back propagation is performed to update the model parameters of the initial convolutional neural network model, and the next sample spectrogram is determined as the current sample spectrogram, and jump to step 1.

[0060] For some household appliances that contain motors, such as washing machines and microwave ovens, there are many cases of motor abnormalities, which can produce a variety of abnormal sounds. For example, if there is a foreign object in the motor, the rotor will contact the foreign object in each rotation cycle to produce abnormal sounds. Another example is that due to improper matching between the ball bearings and the inner and outer rings of the bearings in the motor, the collision between the ball bearings and the inner and outer rings during the operation of the motor will also produce abnormal sounds. The various abnormal sounds produced by the motor may not have obvious features in the spectrogram, resulting in the inability to detect some abnormal sounds of the motor through the spectrogram to be tested.

[0061] For example, to more accurately detect abnormal sounds caused by motors, in some embodiments, after step S110, it further includes: detecting abnormal sounds of the motors based on the vibration characteristics of the vibration signal of the motor to be tested in each rotation cycle of the motor in the household appliance, wherein the vibration signal of the motor to be tested is acquired based on a vibration sensor located in a corresponding area of ​​the motor. Specifically, each vibration sensor has a unique device ID, and when the vibration signal to be tested is obtained from the vibration sensor in step S110, it is determined whether the vibration signal to be tested is a vibration signal of the motor to be tested according to the ID of the vibration sensor.

[0062] In this embodiment, the abnormal sound detection result of the home appliance to be tested is determined based on the abnormal sound detection result output by the abnormal sound detection model and the abnormal sound detection result of the motor. Taking the microwave oven as an example, the abnormal sound detection model outputs the detection results of abnormal sound caused by aging of the magnetron and abnormal sound caused by electromagnetic failure, and the abnormal sound of the motor is obtained by performing feature analysis on the vibration signal of the motor to be tested. Then, what is detected this time is: abnormal sound caused by aging of the magnetron, abnormal sound caused by electromagnetic failure and abnormal sound of the motor, thereby making the abnormal sound detection of the home appliance to be tested more comprehensive.

[0063] In some embodiments, based on the vibration characteristics of the vibration signal of the motor to be tested in each rotation cycle of the motor in the household appliance, the step of detecting abnormal sound of the motor includes: Perform spectrum analysis on the vibration signal of the motor to be tested in each rotation period to determine the target frequency band of the vibration energy mutation, that is, the vibration feature is the vibration energy. Specifically, the frequency domain signal of the vibration signal of the motor to be tested can be obtained by performing fast Fourier transform on the vibration signal of the motor to be tested, and the target frequency band of the vibration energy mutation can be determined based on the frequency domain signal. Among them, the rotation period T can be calculated by the rated speed of the motor, and the calculation formula is T = 60 / RPM, RPM is the rated speed of the motor, and the unit is: revolutions per minute.

[0064] The first target number of cycles in which the vibration energy of the target frequency band exceeds the energy threshold is counted, and when the ratio of the first target number of cycles to the total number of cycles reaches a first preset ratio threshold, it is determined that the abnormal sound of the motor is the abnormal sound caused by friction between the motor rotor and foreign matter inside the motor, wherein the total number of cycles is determined based on the rotation period of the motor and the total duration of the vibration signal of the motor to be measured.

[0065] It can be understood that the total duration of the vibration signal of the motor to be tested is the duration of an abnormal sound detection. If there is a foreign object inside the motor, each rotation of the rotor will cause different degrees of friction with the foreign object. When the friction force is large, the vibration energy of the vibration signal of the motor to be tested is large, and abnormal sound will be generated. Therefore, in this embodiment, the first target number of cycles in which the vibration energy of the target frequency band exceeds the energy threshold is counted, wherein the energy threshold can be set according to actual conditions, for example: 0.3G-0.6G, G is the unit of gravity acceleration, and when the ratio of the first target number of cycles to the total number of cycles reaches the first preset ratio threshold, it is determined that the abnormal sound of the motor is the abnormal sound caused by friction between the motor rotor and the foreign object inside the motor. Among them, the first preset ratio threshold can be set according to actual conditions, for example: 15%~20%.

[0066] For example: the motor speed is 1000RPM, and the total duration of the motor vibration signal to be measured is 2 seconds. Then within this 2-second duration, the total number of cycles of the motor rotation is approximately 33, that is, the motor rotates 33 times. If the first preset proportional threshold is set to 18%, when the vibration energy exceeds the target number of cycles of the energy threshold by 6, it is determined that the abnormal sound of the motor is the abnormal sound caused by friction between the motor rotor and foreign matter inside the motor.

[0067] In this embodiment, the vibration energy of the target frequency band in each rotation cycle of the motor is used as the vibration feature, and the vibration feature is analyzed to determine the abnormal sound generated by the friction between the motor rotor and the foreign matter inside the motor.

[0068] In some embodiments, detecting abnormal noise of a motor based on vibration characteristics of a vibration signal of a motor to be tested in each rotation cycle of a motor in a household appliance includes: The kurtosis analysis is performed on the vibration signal of the motor to be tested in each of the rotation cycles to obtain the kurtosis value in each of the rotation cycles, that is, the vibration characteristic is the kurtosis. In this embodiment, due to improper matching between the balls and the inner and outer rings of the bearings in the motor, when the motor is running, the collision between the balls and the inner and outer rings will generate instantaneous large energy, and the instantaneous large energy is reflected in the vibration signal of the motor to be tested in the form of kurtosis.

[0069] The second target number of cycles in which the kurtosis value exceeds the kurtosis threshold is counted, and when the ratio of the second target number of cycles to the total number of cycles reaches a second preset ratio threshold, the abnormal sound of the motor is determined to be the abnormal sound of the motor bearing, wherein the total number of cycles is determined based on the rotation period of the motor and the total duration of the vibration signal of the motor to be measured.

[0070] The kurtosis threshold and the second preset ratio threshold can be set according to the actual situation. The kurtosis threshold can be 8, and the second preset ratio threshold can be 10%~15%. For example: the motor speed is 1000RPM, and the total duration of the motor vibration signal to be measured is 2 seconds. Then within this 2-second duration, the total number of cycles of the motor rotation is about 33, that is, the motor rotates 33 times. If the second preset ratio threshold is set to 10%, the second target cycle number with a kurtosis value exceeding 8 is 4, that is, the ratio of the second target cycle number to the total number of cycles is 12%, which is greater than 10%, so it is determined that the motor bearing has abnormal sound.

[0071] In this embodiment, the kurtosis of the vibration signal in each rotation cycle of the motor is used as the vibration feature, and the kurtosis is analyzed to determine the abnormal sound of the motor bearing, that is, the abnormal sound caused by improper matching of the ball and the inner and outer rings in the motor bearing.

[0072] The abnormal sound detection device for household appliances provided by the present invention is described below. The abnormal sound detection device for household appliances described below and the abnormal sound detection method for household appliances described above can be referred to each other.

[0073] The abnormal sound detection device of household appliances according to the embodiment of the present invention is as follows: Figure 3 As shown, the following functional modules 310 to 330 are included.

[0074] The vibration signal acquisition module 310 is used to acquire at least one vibration signal to be tested when the household appliance to be tested is running.

[0075] The spectrogram generating module 320 is used to generate a spectrogram to be tested corresponding to each vibration signal to be tested.

[0076] The model execution module 330 is used to input each of the spectrograms to be tested into the abnormal sound detection model to obtain the abnormal sound detection result output by the abnormal sound detection model.

[0077] The abnormal sound detection model is trained based on a sample spectrogram and an abnormal sound type label corresponding to the sample spectrogram, and the sample spectrogram is generated based on a sample vibration signal.

[0078] In the abnormal sound detection device for household appliances of this embodiment, at least one vibration signal to be tested is obtained when the household appliance to be tested is running; a spectrogram to be tested corresponding to each of the vibration signals to be tested is generated; each of the spectrograms to be tested is input into the abnormal sound detection model to obtain the abnormal sound detection result output by the abnormal sound detection model. Since the abnormal sound detection model is trained based on the sample spectrogram and the abnormal sound type label corresponding to the sample spectrogram, and the sample spectrogram is generated based on the sample vibration signal, the abnormal sound detection model can accurately detect various abnormal sounds of the household appliance to be tested and accurately determine the abnormal sound type by performing image recognition on each spectrogram to be tested, and the detection result has good consistency with respect to manual listening detection.

[0079] In some embodiments, the spectrogram generation module 320 includes the following modules.

[0080] The signal conversion module is used to perform short-time Fourier transform on each of the vibration signals to be measured, so as to convert each of the vibration signals to be measured from a time domain signal to a frequency domain signal.

[0081] The amplitude spectrum determination module is used to determine the amplitude spectrum of each time window corresponding to the short-time Fourier transform in each of the frequency domain signals.

[0082] The amplitude spectrum arrangement module is used to arrange the amplitude spectra of all time windows in time order for each frequency domain signal to generate each of the spectrograms to be tested.

[0083] In some embodiments, the signal conversion module is specifically used to divide each of the vibration signals to be measured into multiple time windows, and perform windowing processing between two adjacent time windows; perform fast Fourier transform on each of the vibration signals to be measured after the windowing processing to obtain the frequency components within each of the time windows, so as to convert each of the vibration signals to be measured from time domain signals to frequency domain signals.

[0084] In some embodiments, the household appliance abnormal sound detection device also includes: a logarithmic coordinate conversion module, which is used to arrange the amplitude spectra of all time windows in chronological order for each frequency domain signal to generate each of the spectrograms to be tested, and then take the logarithm of the amplitude spectrum of each of the spectrograms to be tested to obtain each spectrogram to be tested after logarithmic coordinate transformation. The spectrogram to be tested input into the abnormal sound detection model is the spectrogram to be tested after logarithmic coordinate transformation. When the abnormal sound detection model is trained, the sample spectrogram is the sample spectrogram after logarithmic coordinate transformation.

[0085] In some embodiments, the abnormal sound detection model training process is as follows: the sample spectrogram is input into the initial convolutional neural network model to obtain the abnormal sound prediction result output by the initial convolutional neural network model; the abnormal sound prediction result and the abnormal sound type label corresponding to the sample spectrogram are substituted into the loss function, and when the loss function converges, the training is completed to obtain the abnormal sound detection model.

[0086] In some embodiments, the household appliance abnormal sound detection device also includes: a motor abnormal sound detection module, which is used to detect the abnormal sound of the motor based on the vibration characteristics of the motor vibration signal to be tested in each rotation cycle of the motor in the household appliance after obtaining at least one vibration signal to be tested when the household appliance to be tested is running, and the vibration signal of the motor to be tested is collected based on a vibration sensor located in the corresponding area of ​​the motor.

[0087] In some embodiments, the motor abnormal sound detection module is specifically used to perform spectral analysis on the vibration signal of the motor to be tested within each of the rotation cycles to determine a target frequency band in which the vibration energy changes suddenly; count the first target cycle number in which the vibration energy of the target frequency band exceeds an energy threshold, and when the ratio of the first target cycle number to the total number of cycles reaches a first preset ratio threshold, determine that the motor abnormal sound is an abnormal sound caused by friction between the motor rotor and foreign matter inside the motor, wherein the total number of cycles is determined based on the rotation cycle of the motor and the total duration of the vibration signal of the motor to be tested.

[0088] In some embodiments, the motor abnormal sound detection module is specifically used to perform kurtosis analysis on the vibration signal of the motor to be tested within each of the rotation cycles to obtain the kurtosis value within each of the rotation cycles; count the second target number of cycles whose kurtosis value exceeds the kurtosis threshold, and when the ratio of the second target number of cycles to the total number of cycles reaches a second preset ratio threshold, determine that the motor abnormal sound is the motor bearing abnormal sound, wherein the total number of cycles is determined based on the rotation cycle of the motor and the total duration of the vibration signal of the motor to be tested.

[0089] Figure 4 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4 As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430 and a communication bus 440, wherein the processor 410, the communication interface 420 and the memory 430 communicate with each other through the communication bus 440. The processor 410 may call the logic instructions in the memory 430 to execute the abnormal sound detection method of the household appliance, and the method includes the following steps.

[0090] At least one vibration signal to be measured is obtained when the household appliance to be measured is running.

[0091] Generate a spectrogram to be tested corresponding to each of the vibration signals to be tested.

[0092] Each of the spectrograms to be tested is input into an abnormal sound detection model to obtain an abnormal sound detection result output by the abnormal sound detection model.

[0093] The abnormal sound detection model is trained based on a sample spectrogram and an abnormal sound type label corresponding to the sample spectrogram, and the sample spectrogram is generated based on a sample vibration signal.

[0094] In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0095] The embodiment of the present invention also provides a household appliance abnormal sound detection system, such as Figure 5 , Figure 6 and Figure 7 As shown, it includes: an assembly line body 1, a lifting mechanism 2, a vibration detection mechanism 3 and the above-mentioned electronic equipment 4.

[0096] The assembly line body 1 is provided with a tooling plate 5 for carrying the household appliance to be tested (for example, a microwave oven 6). During testing, the household appliance to be tested is placed on the tooling plate 5, and the assembly line body 1 is used to convey the tooling plate 5. Specifically, the assembly line body 1 can be (but not limited to) the following structure: the assembly line body 1 includes a first bracket 24 and a second bracket 25, the two brackets are arranged opposite to each other, and a conveyor belt or conveyor roller (not shown in the figure) with the same conveying speed is arranged on the top of each of the first bracket 24 and the second bracket 25, and both ends of the tooling plate 5 are placed on the conveyor belts or conveyor rollers of the first bracket 24 and the second bracket 25, respectively, to realize the conveyance of the tooling plate 5. In addition, there is a gap between the two brackets so that the lifting mechanism 2 can lift the tooling plate 5 from the middle of the two brackets.

[0097] The lifting mechanism 2 is located below the assembly line body 1 in a non-contact manner, and is used to lift the tooling plate 5, and lift the tooling plate 5 with the home appliance to be tested to be separated from the assembly line body 1. Specifically, the lifting mechanism 2 can be (but not limited to) the following structure: the lifting mechanism 2 includes a support frame 7 and a lifting component installed on the support frame 7, and the lifting component is used to lift the tooling plate 5. The lifting component can be a cylinder lifting structure or a motor lifting structure. For the cylinder lifting structure, it includes a cylinder 8 and a lifting plate 9 connected to the end of the piston rod of the cylinder 8. The tooling plate 5 is lifted by the lifting plate 9 through the extension and contraction of the piston rod of the cylinder 8. Furthermore, in order to prevent the vibration of the assembly line body 1 from being transmitted from the ground to the support frame 7, and then transmitted upward from the support frame 7 to the home appliance to be tested, an anti-vibration foot cup 10 is provided at the bottom of the support frame 7 to further prevent the external vibration from affecting the vibration signal to be tested.

[0098] The vibration detection mechanism 3 is located above the lifting mechanism 2, and is used to generate at least one vibration signal to be detected of the home appliance to be tested to the electronic device 4. After receiving the at least one vibration signal to be tested, the electronic device 4 executes the home appliance abnormal sound detection method of the above embodiment to detect abnormal sound of the home appliance to be tested.

[0099] It should be noted that: in this embodiment, the electronic device 4 can be an industrial computer, which can not only execute the abnormal sound detection method of household appliances in the above-mentioned embodiment, but also integrate the control programs of the assembly line body 1, the lifting mechanism 2 and the vibration detection mechanism 3 into the electronic device 4. For example, the electronic device 4 can control the start and stop of the assembly line body 1. When the tooling plate 5 is transmitted to the top of the lifting mechanism 2, the lifting mechanism 2 is controlled to lift the tooling plate 5 upward. After the lifting is in place, the vibration detection mechanism 3 is controlled to collect at least one vibration signal to be measured of the household appliance to be tested on the tooling plate 5, and obtain at least one vibration signal to be measured collected by the vibration detection mechanism 3. After the vibration signal is collected, the lifting mechanism 2 is controlled to fall back, and the tooling plate 5 returns to the assembly line body 1 and is transmitted to the next process.

[0100] Of course, a socket 13 is also provided on the assembly line body 1 to provide power for the home appliance to be tested. The home appliance to be tested is connected to the socket 13 via a soft wire, and the vibration of the assembly line body 1 will hardly be transmitted to the home appliance to be tested through the soft wire. Preferably, the socket 13 can be set on the tooling board 5.

[0101] In the household appliance abnormal sound detection system of the present embodiment, since the lifting mechanism 2 itself does not contact the assembly line body 1, the tooling plate 5 is also separated from the assembly line body 1 after being lifted, forming a static bearing platform. Therefore, the vibration of the assembly line body 1 will not be transmitted to the household appliance to be tested, thereby preventing the vibration of the assembly line body 1 from interfering with the vibration signal to be tested, making the collected at least one vibration signal to be tested more accurate and the detection of abnormal sound more accurate.

[0102] The abnormal sound detection system for home appliances of this embodiment can be used in home appliance production scenarios and maintenance scenarios. For home appliance production scenarios, the assembly line body 1 can be docked at the end of the production line, and the tooling board 5 can be shared by the assembly line body 1 and the production line. Each home appliance product from the production line is directly moved to the assembly line body 1 through the tooling board 5, so that all newly produced home appliances can be tested for abnormal sounds to prevent unqualified products from entering the market. For maintenance scenarios, the home appliances to be repaired are placed on the tooling board 5 of the assembly line body 1 in turn, and they can be tested in turn. Compared with traditional manual listening detection, the efficiency of maintenance detection is improved.

[0103] In some embodiments, the household appliance abnormal sound detection system further includes a mounting frame, the mounting frame including: a column 11 and a transverse support plate 12, the transverse support plate 12 can be installed on the column 11 axially movable (i.e., moving up and down) along the column 11, the column 11 is installed on the lifting mechanism 2, and the vibration detection mechanism 3 is installed on the transverse support plate 12 in an adjustable position, so that the vibration detection mechanism 3 contacts the corresponding area of ​​the component that causes the abnormal sound in the household appliance to be tested. The transverse support plate 12 moves up and down along the column 11, so as to adjust the height of the vibration detection mechanism 3 according to the height of the household appliance to be tested after lifting. The vibration detection mechanism 3 is installed on the transverse support plate 12 in an adjustable position, and the installation position of the vibration detection mechanism 3 can be adjusted according to the size of the household appliance to be tested, and the installation position of the vibration detection mechanism 3 can also be adjusted according to the corresponding area of ​​each component that causes abnormal sound in different household appliances to be tested. Specifically, a plurality of mounting holes 23 are provided on the transverse support plate 12, and the mounting holes 23 can be randomly distributed so that the distance between each mounting hole 23 and the home appliance to be tested and the projection position on the home appliance to be tested are different. The vibration detection mechanism 3 is installed in different mounting holes 23 to adjust the installation position of the vibration detection mechanism 3.

[0104] The column 11 is installed on the jacking mechanism 2, and the jacking mechanism 2 does not contact the assembly line body 1. Therefore, the column 11, the transverse support plate 12 and the vibration detection mechanism 3 do not contact the assembly line body 1, avoiding the interference of the vibration of the assembly line body 1 on the vibration detection mechanism 3, so that the collected vibration signal to be measured is more accurate.

[0105] In some embodiments, Figure 8As shown, the vibration detection mechanism 3 includes: a vibration sensor 14, a mounting seat 15, a slider 16 and a cylinder 17. The mounting seat 15 is provided with a slide rail 18, and the slider 16 is installed on the slide rail 18, that is, the slider 16 can slide on the slide rail 18, the vibration sensor 14 is installed on the slider 16 with an adjustable position, the cylinder 17 is installed on the mounting seat 15, the piston rod of the cylinder 17 is connected to the slider 16, and the mounting seat 15 is installed on the transverse support plate 12 with an adjustable position. Specifically, a plurality of countersunk holes that cooperate with the mounting holes 23 are also provided at the bottom of the mounting seat 15. The vibration sensor 14 is installed on the slider 16 with an adjustable position, and the position of the vibration sensor 14 relative to the home appliance to be tested is further adjusted, as shown in FIG. Figure 8 In the embodiment, the installation height of the vibration sensor 14 on the slider 16 can be adjusted, the longitudinal position of the vibration sensor 14 can be adjusted, and the lateral or horizontal position of the vibration detection mechanism 3 on the lateral support plate 12 can be adjusted, so that the vibration sensor 14 can be more accurately located at the corresponding area of ​​each component that causes abnormal sound in the home appliance to be tested. The cylinder 17 pushes the slider 16 to move, so as to fine-tune the distance between the vibration sensor 14 and the home appliance to be tested, so that the two keep in contact.

[0106] In some embodiments, the piston rod of the cylinder 17 is connected to the slider 16 through an elastic component, such as a spring 19, to prevent the vibration sensor 14 from pressing against the appliance to be tested, causing damage to the outer surface of the appliance to be tested. In addition, due to the pressing too tightly, when encountering a large vibration, the outer surface of the appliance to be tested may be deformed, and the stress of the deformation will also act on the vibration sensor 14, so that the stress is also mistakenly collected as part of the vibration signal to be tested. Moreover, after the deformation, the vibration sensor 14 and the outer surface of the appliance to be tested may not be in contact, and the vibration signal cannot be collected. Therefore, in this embodiment, the piston rod of the cylinder 17 is connected to the slider 16 through an elastic component, and the elastic component has a buffering effect on the vibration, avoiding the above-mentioned situation of incorrectly collecting the vibration signal or failing to collect the vibration signal.

[0107] In some embodiments, the transverse support plate 12 includes: a first support sub-plate 20, a second support sub-plate 21 and a third support sub-plate 22, the first support sub-plate 20 can be installed on the column 11 movably along the axial direction of the column, the second support sub-plate 21 is movably connected to one end of the first support sub-plate 20, and the third support sub-plate 22 is movably connected to the other end of the first support sub-plate 20. Wherein, the movable connection includes a sliding connection or a rotating connection, which can be adjusted according to the shape and size of the home appliance to be tested. For example, the movable connection is a rotating connection. For a home appliance to be tested in the shape of a rectangular parallelepiped, the second support sub-plate 21 and the third support sub-plate 22 can be adjusted to be parallel and perpendicular to the first support sub-plate 20, respectively, and the vibration signal to be tested can be collected from at least three sides. For some cylindrical home appliances to be tested with smaller sizes, such as a wall breaking machine, the second support sub-plate 21 and the third support sub-plate 22 can be adjusted to have an angle less than 90 degrees with the first support sub-plate 20, respectively, to form an enclosed structure for the wall breaking machine, so as to collect the vibration signals to be tested at multiple positions on the side.

[0108] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the household appliance abnormal sound detection method provided by the above-mentioned methods, which includes the following steps.

[0109] At least one vibration signal to be measured is obtained when the household appliance to be measured is running.

[0110] Generate a spectrogram to be tested corresponding to each of the vibration signals to be tested.

[0111] Each of the spectrograms to be tested is input into an abnormal sound detection model to obtain an abnormal sound detection result output by the abnormal sound detection model.

[0112] The abnormal sound detection model is trained based on a sample spectrogram and an abnormal sound type label corresponding to the sample spectrogram, and the sample spectrogram is generated based on a sample vibration signal.

[0113] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for detecting abnormal sounds of household appliances provided by the above methods is implemented, and the method includes the following steps.

[0114] At least one vibration signal to be measured is obtained when the household appliance to be measured is running.

[0115] Generate a spectrogram to be tested corresponding to each of the vibration signals to be tested.

[0116] Each of the spectrograms to be tested is input into an abnormal sound detection model to obtain an abnormal sound detection result output by the abnormal sound detection model.

[0117] The abnormal sound detection model is trained based on a sample spectrogram and an abnormal sound type label corresponding to the sample spectrogram, and the sample spectrogram is generated based on a sample vibration signal.

[0118] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0119] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0120] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting abnormal sound of household appliances, characterized in that: include: Acquire at least one vibration signal to be tested when the household appliance to be tested is running; Generating a spectrogram to be tested corresponding to each of the vibration signals to be tested; Inputting each of the spectrograms to be tested into an abnormal sound detection model respectively to obtain an abnormal sound detection result output by the abnormal sound detection model; The abnormal sound detection model is trained based on a sample spectrogram and an abnormal sound type label corresponding to the sample spectrogram, and the sample spectrogram is generated based on a sample vibration signal.

2. The method for detecting abnormal sound of household appliances according to claim 1, characterized in that: Generating a spectrogram to be tested corresponding to each vibration signal to be tested, comprising: Performing short-time Fourier transform on each of the vibration signals to be measured, so as to convert each of the vibration signals to be measured from a time domain signal to a frequency domain signal; Determine the amplitude spectrum of each time window corresponding to the short-time Fourier transform in each of the frequency domain signals; For each frequency domain signal, the amplitude spectra of all time windows are arranged in time order to generate the spectrograms to be tested.

3. The method for detecting abnormal sound of household appliances according to claim 2, characterized in that: Performing short-time Fourier transform on each of the vibration signals to be measured to convert each of the vibration signals to be measured from a time domain signal to a frequency domain signal, comprising: Each of the vibration signals to be measured is divided into a plurality of time windows, and a windowing process is performed between two adjacent time windows; The windowed vibration signals are subjected to fast Fourier transformation to obtain frequency components in each time window, so as to convert the vibration signals to be measured from time domain signals to frequency domain signals.

4. The method for detecting abnormal sound of household appliances according to claim 2, characterized in that: For each frequency domain signal, after arranging the amplitude spectra of all time windows in time order to generate each of the spectrograms to be tested, the method further includes: The logarithm of the amplitude spectrum of each of the spectrograms to be tested is taken to obtain each spectrogram to be tested after logarithmic coordinate transformation. The spectrogram to be tested input into the abnormal sound detection model is the spectrogram to be tested after logarithmic coordinate transformation. When the abnormal sound detection model is trained, the sample spectrogram is the sample spectrogram after logarithmic coordinate transformation.

5. The method for detecting abnormal sound of household appliances according to claim 1, characterized in that: The abnormal sound detection model training process is as follows: Inputting the sample spectrogram into an initial convolutional neural network model to obtain an abnormal sound prediction result output by the initial convolutional neural network model; The abnormal sound prediction result and the abnormal sound type label corresponding to the sample spectrogram are substituted into the loss function. When the loss function converges, the training is completed to obtain the abnormal sound detection model.

6. The method for detecting abnormal sound of household appliances according to any one of claims 1 to 5, characterized in that: After obtaining at least one vibration signal to be tested when the household appliance to be tested is running, the method further includes: Abnormal noise of the motor is detected based on the vibration characteristics of a vibration signal of the motor to be tested in each rotation cycle of the motor in the household appliance, wherein the vibration signal of the motor to be tested is collected based on a vibration sensor located in a corresponding area of ​​the motor.

7. The method for detecting abnormal sound of household appliances according to claim 6, characterized in that: Based on the vibration characteristics of the motor vibration signal to be tested in each rotation cycle of the motor in the home appliance, abnormal motor noise is detected, including: Performing spectrum analysis on the vibration signal of the motor to be tested within each rotation cycle to determine the target frequency band of the sudden change of vibration energy; The first target number of cycles in which the vibration energy of the target frequency band exceeds the energy threshold is counted, and when the ratio of the first target number of cycles to the total number of cycles reaches a first preset ratio threshold, it is determined that the abnormal sound of the motor is the abnormal sound caused by friction between the motor rotor and foreign matter inside the motor, wherein the total number of cycles is determined based on the rotation period of the motor and the total duration of the vibration signal of the motor to be measured.

8. The method for detecting abnormal sound of household appliances according to claim 6, characterized in that: Based on the vibration characteristics of the motor vibration signal to be tested in each rotation cycle of the motor in the home appliance, abnormal motor noise is detected, including: Performing kurtosis analysis on the vibration signal of the motor to be measured in each rotation cycle to obtain the kurtosis value in each rotation cycle; The second target number of cycles in which the kurtosis value exceeds the kurtosis threshold is counted, and when the ratio of the second target number of cycles to the total number of cycles reaches a second preset ratio threshold, the abnormal sound of the motor is determined to be the abnormal sound of the motor bearing, wherein the total number of cycles is determined based on the rotation period of the motor and the total duration of the vibration signal of the motor to be measured.

9. A household appliance abnormal sound detection device, characterized in that: include: A vibration signal acquisition module, used to acquire at least one vibration signal to be tested when the home appliance to be tested is running; A spectrogram generating module, used for generating a spectrogram to be tested corresponding to each vibration signal to be tested; A model execution module, used for inputting each of the spectrograms to be tested into an abnormal sound detection model to obtain an abnormal sound detection result output by the abnormal sound detection model; The abnormal sound detection model is trained based on a sample spectrogram and an abnormal sound type label corresponding to the sample spectrogram, and the sample spectrogram is generated based on a sample vibration signal.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the method for detecting abnormal sounds of household appliances according to any one of claims 1 to 8 is implemented.

11. A household appliance abnormal sound detection system, characterized in that: include: An assembly line body, a lifting mechanism, a vibration detection mechanism, and the electronic device according to claim 10; The assembly line body is provided with a tooling plate for carrying the household appliance to be tested; The lifting mechanism is located below the assembly line body in a non-contact manner with the assembly line body and is used to lift the tooling plate; The vibration detection mechanism is located above the lifting mechanism, and is used to generate at least one vibration signal to be detected of the household appliance to be detected to the electronic device.

12. The household appliance abnormal sound detection system according to claim 11, characterized in that: It also includes a mounting frame, which includes: a column and a transverse support plate, the transverse support plate is installed on the column axially movably along the column, the column is installed on the jacking mechanism, and the vibration detection mechanism is installed on the transverse support plate with adjustable position.

13. The household appliance abnormal sound detection system according to claim 12, characterized in that: The vibration detection mechanism includes: a vibration sensor, a mounting seat, a slider and a cylinder, the mounting seat is provided with a slide rail, the slider is mounted on the slide rail, the vibration sensor is mounted on the slider in an adjustable position, the cylinder is mounted on the mounting seat, the piston rod of the cylinder is connected to the slider, and the mounting seat is mounted on the transverse support plate in an adjustable position.

14. The household appliance abnormal sound detection system according to claim 13, characterized in that: The piston rod of the cylinder is connected to the slider via an elastic component.

15. The household appliance abnormal sound detection system according to claim 13, characterized in that: The transverse support plate includes: a first support sub-plate, a second support sub-plate and a third support sub-plate. The first support sub-plate can be installed on the column axially movably along the column, the second support sub-plate is movably connected to one end of the first support sub-plate, and the third support sub-plate is movably connected to the other end of the first support sub-plate.

16. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for detecting abnormal sounds of household appliances according to any one of claims 1 to 8 is implemented.

17. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method for detecting abnormal sounds of household appliances according to any one of claims 1 to 8 is implemented.

Images