Simple and convenient audio control classification system and method for intelligent garbage cans

Abstract

The invention relates to the technical field of garbage classification, in particular to a simple audio control classification system and method for an intelligent garbage can, and the system comprises an audio collection module, an audio file processing module and a motor drive module. The audio acquisition module is used for acquiring collision sound generated when the garbage falls to the baffle and exporting audio; the audio file processing module is used for analyzing audio and judging the subordinate garbage type of an object; and the motor driving module is used for driving various types of garbage storage buckets at the bottom of the baffle plate to throw the garbage into the corresponding garbage storage buckets according to the analyzed garbage types. Collision sounds generated when garbage falls into the garbage can are collected through the audio collection module, the garbage types can be effectively classified according to different audios of objects, and then movement of the motor module at the bottom of the garbage can is controlled; the accuracy of garbage classification can be effectively improved, the effective process of urban garbage classification is accelerated, and the utilization rate of resources is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of garbage classification, and in particular to a simple audio-controlled classification system and method for an intelligent garbage can. Background Art

[0002] In recent years, due to the large amount of garbage generated in the production and life process and the lack of awareness of garbage classification, the environmental pollution has become increasingly severe. Smart trash cans with autonomous garbage classification capabilities have been launched one after another. Among them, smart trash cans with image recognition have a wider audience. By collecting pictures of garbage, they can effectively classify garbage. However, image recognition involves expensive equipment such as cameras, which leads to high production costs of trash cans. Smart trash cans based on audio recognition can effectively solve this problem by using cheap audio acquisition modules, and promote smart trash cans more effectively. Summary of the invention

[0003] The purpose of the present invention is to solve the shortcomings existing in the prior art, and to propose a simple audio-controlled classification system and method for an intelligent trash can. The audio acquisition module collects the impact sound of garbage falling into the bin, and the types of garbage can be effectively classified according to the different audios of the objects, thereby controlling the movement of the motor module at the bottom of the trash can; it can effectively improve the accuracy of garbage classification, speed up the effective process of urban garbage classification, and improve the utilization rate of resources.

[0004] In order to achieve the above object, the present invention adopts the following technical solutions:

[0005] A simple audio-controlled classification system for an intelligent trash can, comprising an audio acquisition module, an audio file processing module and a motor drive module;

[0006] The audio collection module obtains the collision sound emitted when the garbage falls onto the baffle and exports the audio;

[0007] The audio file processing module is used to analyze the audio and determine the type of garbage to which the object belongs;

[0008] The motor driving module drives various types of garbage collection bins at the baffle bottom to be placed into corresponding garbage collection bins according to the garbage types analyzed.

[0009] The present invention also provides a simple audio control classification method for an intelligent trash can. The method is based on the simple audio control classification system for the intelligent trash can and includes the following steps:

[0010] Step 100: Obtain the collision sound made when the garbage falls onto the baffle and export the audio;

[0011] Step 200, analyzing the audio and determining whether the object belongs to the garbage type;

[0012] Step 300: Drive various types of garbage collection bins at the bottom of the baffle according to the garbage types analyzed by the scores, and put the garbage into the corresponding garbage collection bins.

[0013] Preferably, in step 100, the audio is acquired through the wm8978 module and converted into a wav format for storage.

[0014] Preferably, in step 200, the collected wav files are passed into the audio algorithm load_audio_files(data_dir, sample_rate=16000, duration=1) in the main control through the stm32 main control, which accepts a directory data_dir containing audio files and two optional parameters: sampling rate sample_rate (default 16000Hz) and audio duration duration (default 1 second). The function first obtains all the folders of all categories in the directory, and then traverses the .wav files in each category folder to read the audio data. If the audio length is less than 1 second, it is filled with zeros; if it exceeds 1 second, the first 1 second is intercepted. Finally, the processed audio data and the corresponding labels (category indexes) are converted into NumPy arrays, and these arrays and category lists are returned and the generated audio is denoised to obtain more accurate audio data; the acquired audio files are analyzed to obtain the time domain waveform of the wav audio file, and a convolutional neural network model is used. Among them, f is the input signal, g is the convolution kernel (filter), * represents the convolution operation, and n is the index of the output signal. This formula is used to extract the features of the input signal. In this process, the convolution kernel (or filter) g slides on the input signal f, calculates the sum of the point-by-point products of the two, generates a new output signal, analyzes it, converts it into a spectrum diagram, and represents it as a two-dimensional image; the image is processed through this model to obtain the score, embedding vector and spectrogram.

[0015] Preferably, in step 300, the acquired audio file is judged, including using the processed score as a basis. If an ideal score is obtained for a certain type, it can be considered that the garbage disposed belongs to this type, and a corresponding visual spectrogram can be generated. After the judgment is completed, the corresponding type of garbage can is moved to the bottom of the baffle and then the garbage is poured in.

[0016] Preferably, in step 300, by analyzing the processed scores, four types of garbage scores are counted:

[0017] If the scores are all below 80, that is, the audio generated by the object does not match any of the garbage types in the model library, a voice announcement will be given to indicate that the object is an unknown type of foreign matter, and it will be judged as a foreign matter. After being judged as a foreign matter, the foreign matter will be discharged from a separate discharge port; for the rest, the type with the highest score will be judged as the type of garbage;

[0018] If the classification of the trash can at the bottom of the baffle meets the classification obtained after the judgment, the baffle is directly opened;

[0019] If the classification of the trash can at the bottom of the baffle does not match the classification obtained after the judgment, the correct type of trash can is moved to the bottom of the baffle by means of bearing rotation or the like, and then placed;

[0020] After the delivery process is completed, the baffle will recover in a short time and wait for the next delivery.

[0021] By adopting the above technical solution, the audio obtained by the collision of the garbage and the baffle is analyzed, the audio is obtained using the wm8978 module, and the convolutional neural network model is used for noise reduction and analysis to obtain a highly accurate judgment result, thereby realizing intelligent garbage classification and refining the audio recognition process, and being able to use a simpler and more economical control method to achieve accurate classification and storage of garbage.

[0022] Compared with the prior art, the present invention has the following beneficial effects:

[0023] 1. Compared with the smart trash cans with traditional complex cameras or advanced voice recognition systems, the smart trash cans based on audio classification are more economical and affordable, because sound sensors are usually cheaper than cameras and the computing resources required to process audio signals are relatively less, thereby reducing costs.

[0024] 2. Since the present invention does not require video surveillance, the smart trash can can achieve effective garbage classification while protecting user privacy, avoiding privacy issues caused by video recording.

[0025] 3. The present invention is not affected by lighting conditions, can work stably in various environments, and has strong adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 is a flow chart of the present invention;

[0027] Figure 2 FIG. 4 is a flowchart of audio processing in an embodiment of the present invention. DETAILED DESCRIPTION

[0028] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings, so that those skilled in the art can better understand the advantages and features of the present invention, thereby making a clearer definition of the protection scope of the present invention. The embodiments described in the present invention are only a part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present invention.

[0029] Reference Figure 1-Figure 2 , a simple audio-controlled sorting system for a smart trash can, including an audio acquisition module, an audio file processing module and a motor drive module;

[0030] The audio collection module obtains the collision sound emitted when the garbage falls onto the baffle and exports the audio;

[0031] The audio file processing module is used to analyze the audio and determine the type of garbage to which the object belongs;

[0032] The motor driving module drives various types of garbage collection bins at the baffle bottom to be placed into corresponding garbage collection bins according to the garbage types analyzed.

[0033] A simple audio-controlled classification method for an intelligent trash can comprises the following steps:

[0034] Step 100: Obtain the collision sound made when the garbage falls onto the baffle and export the audio;

[0035] In step 100, the audio is acquired through the wm8978 module and converted into wav format for storage.

[0036] Among them, the single chip microcomputer module wm8978 is used here. In some specific applications, the infrared sensing module can be used to determine whether there is garbage input to further improve the accuracy. The audio information obtained after sampling is mainly a wav audio file, which contains the time domain waveform data of the audio. In some specific applications, these files are usually stored in the SD card to facilitate subsequent reading operations.

[0037] Step 200, analyzing the audio and determining whether the object belongs to a garbage type; specifically includes:

[0038] S210, obtaining the audio file in the storage unit, and importing it into the trained neural network model;

[0039] S220, denoising audio through a convolutional neural network model to improve accuracy;

[0040] In step S220, the audio signal is gradually enhanced or weakened by using the fade-in and fade-out audio engineering technology in the model preprocessing process.

[0041] S230, calling the model and obtaining the score, embedding vector and spectrogram, and using the score as the main basis to determine the type of garbage;

[0042] In step S230, the pre-processed audio is converted from a time domain signal to a time-frequency domain signal by calculating a short-time Fourier transform (STET) to convert the waveform into a spectrogram to show the change of frequency over time. After the spectrogram conversion is completed, it will be compared with the data sets of different types of garbage in the model and the scores, embedding vectors and spectrograms will be obtained, where the scores will be used as the main results of the judgment. In some cases, the obtained spectrograms can be used for visual judgment.

[0043] S240, if the judgment is successful, proceed to the subsequent step of driving the trash can;

[0044] In step S240, the model will select the highest score among the scores of each data set and use it as a judgment, and then perform subsequent delivery operations.

[0045] S250, if the garbage does not meet any of the categories, it is regarded as foreign matter and discharged from the discharge outlet;

[0046] In step S250, if the scores of each data set are less than the qualified value (<80) after the model analysis is completed, it is determined to be an unknown foreign matter and is discharged from a separate outlet to avoid classification confusion.

[0047] Step 300: Drive various types of garbage collection bins at the bottom of the baffle according to the garbage types analyzed by the scores, and put the garbage into the corresponding garbage collection bins.

[0048] In step 300, the acquired audio file is judged, including using the processed score as a basis. If an ideal score is obtained for a certain type, it can be considered that the garbage disposed belongs to this type, and a corresponding visual spectrogram can be generated. After the judgment is completed, the corresponding type of garbage can is moved to the bottom of the baffle and then the garbage is poured in.

[0049] In step 300, by analyzing the processed scores, four types of garbage scores are counted:

[0050] If the scores are all below 80, that is, the audio generated by the object does not match any of the garbage types in the model library, a voice announcement will be given to indicate that the object is an unknown type of foreign matter, and it will be judged as a foreign matter. After being judged as a foreign matter, the foreign matter will be discharged from a separate discharge port; for the rest, the type with the highest score will be judged as the type of garbage;

[0051] If the classification of the trash can at the bottom of the baffle meets the classification obtained after the judgment, the baffle is directly opened;

[0052] If the classification of the trash can at the bottom of the baffle does not match the classification obtained after the judgment, the correct type of trash can is moved to the bottom of the baffle by means of bearing rotation or the like, and then placed;

[0053] After the delivery process is completed, the baffle will recover in a short time and wait for the next delivery.

[0054] To sum up, the present invention collects the impact sound of garbage falling into the bin through the audio collection module. According to the different audios of the objects, the types of garbage can be effectively classified, and then the movement of the motor module at the bottom of the trash can is controlled; it can effectively improve the accuracy of garbage classification, speed up the effective process of urban garbage classification, and improve the utilization rate of resources.

[0055] The description and practice disclosed in the present invention are easy to think and understand for ordinary technicians in the technical field, and several improvements and modifications can be made without departing from the principles of the present invention. Therefore, modifications or improvements made without departing from the spirit of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A simple audio-controlled classification system for smart trash cans, characterized in that: It includes an audio acquisition module, an audio file processing module and a motor drive module; The audio collection module obtains the collision sound emitted when the garbage falls onto the baffle and exports the audio; The audio file processing module is used to analyze the audio and determine the type of garbage to which the object belongs; The motor driving module drives various types of garbage collection bins at the baffle bottom to be placed into corresponding garbage collection bins according to the garbage types analyzed.

2. A simple audio-controlled classification method for smart trash cans, characterized in that: The method is based on the simple audio-controlled classification system for the smart trash can according to claim 1, and comprises the following steps: Step 100: Obtain the collision sound made when the garbage falls onto the baffle and export the audio; Step 200, analyzing the audio and determining whether the object belongs to the garbage type; Step 300: Drive various types of garbage collection bins at the bottom of the baffle according to the garbage types analyzed by the scores, and put the garbage into the corresponding garbage collection bins.

3. The simple audio-controlled classification method for a smart trash can according to claim 1, characterized in that: In step 100, the audio is acquired through the wm8978 module and converted into wav format for storage.

4. The simple audio-controlled classification method for a smart trash can according to claim 1, characterized in that: In step 200, the collected wav files are passed to the audio algorithm load_audio_files in the main control through the stm32 main control. The algorithm is used to load audio files and preprocess them. It accepts a directory data_dir containing audio files and two optional parameters: the sampling rate sample_rate and the audio duration duration. The function is used to first obtain all the folders of all categories in the directory, and then traverse the .wav files in each category folder to read the audio data. If the audio length is less than 1 second, it is filled with zeros. If it exceeds 1 second, the first 1 second is captured; finally, the processed audio data and the corresponding labels are converted into NumPy arrays, and these arrays and category lists are returned, and the generated audio is subjected to noise reduction to obtain more accurate audio data; the acquired audio file is analyzed to obtain the time domain waveform of the wav audio file, and a convolutional neural network model is used Among them, f is the input signal, g is the convolution kernel, * indicates the convolution operation, and n is the index of the output signal. This formula is used to extract the features of the input signal. In this process, the convolution kernel g slides on the input signal f, calculates the sum of the point-by-point products of the two, generates a new output signal, analyzes it, converts it into a spectrum diagram, and represents it as a two-dimensional image; the image is processed through this model to obtain the score, embedding vector and spectrogram.

5. The simple audio-controlled classification method for intelligent trash can according to claim 4, characterized in that: In step 300, the acquired audio file is judged, including using the processed score as a basis. If an ideal score is obtained for a certain type, it can be considered that the garbage disposed belongs to this type, and a corresponding visual spectrogram can be generated. After the judgment is completed, the corresponding type of garbage can is moved to the bottom of the baffle and then the garbage is poured in.

6. The simple audio-controlled classification method for intelligent trash can according to claim 5, characterized in that: In step 300, by analyzing the processed scores, four types of garbage scores are counted: If the scores are all below 80, that is, the audio generated by the object does not match any of the garbage types in the model library, a voice announcement will be given to indicate that the object is an unknown type of foreign matter, and it will be judged as a foreign matter. After being judged as a foreign matter, the foreign matter will be discharged from a separate discharge port; for the rest, the type with the highest score will be judged as the type of garbage; If the classification of the trash can at the bottom of the baffle meets the classification obtained after the judgment, the baffle is directly opened; If the classification of the trash can at the bottom of the baffle does not match the classification obtained after the judgment, the correct type of trash can is moved to the bottom of the baffle by the bearing rotation method and then placed; After the delivery process is completed, the baffle will recover in a short time and wait for the next delivery.

Images