Flotation Method, Flotation Device, Processor and Flotation System of Coal

Abstract

The present application provides a flotation method, a flotation device, a processor and a flotation system for coal. The flotation method includes: obtaining multiple groups of original data in a flotation cell and the ash content value corresponding to each group of original data; training a preset Bayesian model at least using the multiple groups of original data and the corresponding ash content values to obtain a target Bayesian model; determining a target ash content range corresponding to the flotation data based on the target Bayesian model and the flotation data obtained in real time, and determining the addition amounts of a foaming agent and a collector corresponding to the target ash content range, wherein the flotation data includes real-time image data and real-time audio data. Since the time complexity of the target Bayesian model is within the range of linear complexity, this ensures that the computational amount of the target Bayesian model is small and the computational speed is fast, thereby solving the technical problems of large computational amount and slow computational speed for real-time calculation of ash content values in the prior art.

Description

Technical Field

[0001] The present application relates to the technical field of coal flotation processes, and in particular, to a method for flotation of coal, a flotation device, a computer-readable storage medium, a processor, and a flotation system. Background Technique

[0002] Flotation, as an important process link in coal preparation, is an indispensable part for coking coal preparation plants to maximize the yield of clean coal and fully recover scarce resources. Currently, many countries in the world have fully realized automated coal preparation. For example, the United States, the United Kingdom, Australia, etc. Although the automated control technologies such as on-line ash detection and heavy medium density regulation being promoted in China have been greatly improved, their automation level and popularity are still far from sufficient.

[0003] As a traditional industry, before coal preparation, agents (such as foaming agents and collectors) are added according to the ash value in the washing tank. In the prior art, a detection model can also be obtained through machine learning to on-line detect the ash value in the flotation cell, and determine the addition amounts of the foaming agent and the collector according to the obtained ash value. However, the detection model obtained based on machine learning has problems of large computational amount and slow calculation speed. Summary of the Invention

[0004] The main object of the present application is to provide a method for flotation of coal, a flotation device, a computer-readable storage medium, a processor, and a flotation system, so as to solve the technical problems of large computational amount and slow calculation speed for real-time calculation of ash values in the prior art.

[0005] According to one aspect of an embodiment of the present invention, a method for flotation of coal is provided, including: obtaining multiple groups of original data in a flotation cell, and the ash value corresponding to each group of the original data, where the original data includes original image data and original audio data; training a preset Bayesian model at least using the multiple groups of original data and the corresponding ash values to obtain a target Bayesian model; determining a target ash range corresponding to the flotation data based on the target Bayesian model and the flotation data obtained in real time, and determining the addition amounts of the corresponding foaming agent and collector based on the target ash range, where the flotation data includes real-time image data and real-time audio data.

[0006] Optionally, training a preset Bayesian model at least using the multiple groups of original data and the corresponding ash values to obtain a target Bayesian model includes: preprocessing each group of the original data to obtain multiple groups of target data, where the preprocessing at least includes filtering; training the preset Bayesian model using each of the target data to obtain the target Bayesian model.

[0007] Optionally, preprocess each group of the original data to obtain multiple groups of target data, including: performing wavelet transform on each group of the original data for filtering to obtain multiple groups of filtered data, where each group of the filtered data includes filtered image data and filtered audio data, the filtered image data is used to characterize the glossiness and density of the bubbles in the flotation cell, and the filtered audio data is used to characterize the sound of the bubbles bursting; performing co-frequency combination processing on each group of the filtered image data and the filtered audio data to obtain multiple groups of the target data.

[0008] Optionally, use each group of the target data to train the preset Bayesian model to obtain the target Bayesian model, including: determining multiple target ash value sets according to the ash values corresponding to each group of the original data and a preset ash value range, and each target ash value set includes multiple groups of the original data; constructing the preset Bayesian model where P(c) is the prior probability of the target ash value set, x ij is the jth attribute in the ith group of the original data in the target ash value set, the attribute is one of glossiness, density, or the sound of bursting, w(i) is the weight of the attribute, i starts from 1 and goes up to n, n is the total number of the original data in the target ash value set, j starts from 1 and goes up to m, m is the total number of the attributes in a group of the original data; use multiple target ash value sets to correct the weights of the preset Bayesian model to obtain the target Bayesian model.

[0009] Optionally, use multiple target ash value sets to correct the weights of the preset Bayesian model to obtain the target Bayesian model, including: using the attributes in each target ash value set and the preset Bayesian model to calculate multiple classification correct probabilities; according to correct the weights in the preset Bayesian model to obtain the target Bayesian model, where a j is the classification correct probability of the jth attribute; w(j) is the weight.

[0010] Optionally, based on the target Bayesian model and the flotation data obtained in real time, determining the target ash interval corresponding to the flotation data includes: performing filtering processing on the flotation data by using wavelet transform to obtain filtered flotation data, where the filtered flotation data includes filtered flotation image data and filtered flotation audio data, the filtered flotation image data is used to characterize the glossiness and density of bubbles in the flotation cell, and the filtered flotation audio data is used to characterize the sound of bubble breakage; performing co-frequency combination processing on each group of the filtered flotation image data and the filtered flotation audio data to obtain multiple groups of target flotation data; and using the target Bayesian model and the target flotation data to determine the corresponding target ash interval.

[0011] According to another aspect of the embodiments of the present invention, there is also provided a flotation device for coal, including: an acquisition unit, configured to acquire multiple groups of original data in a flotation cell and the ash value corresponding to each group of the original data, where the original data includes original image data and original audio data; a training unit, configured to train a preset Bayesian model by using at least multiple groups of the original data and the corresponding ash values to obtain a target Bayesian model; and a determination unit, configured to determine the target ash interval corresponding to the flotation data based on the target Bayesian model and the flotation data obtained in real time, and determine the addition amounts of the frother and the collector corresponding to the target ash interval, where the flotation data includes real-time image data and real-time audio data.

[0012] According to still another aspect of the embodiments of the present invention, there is also provided a computer-readable storage medium, where the computer-readable storage medium includes a stored program, and the program executes any one of the flotation methods for coal.

[0013] According to yet another aspect of the embodiments of the present invention, there is also provided a processor, where the processor is used to run a program, and when the program runs, it executes any one of the flotation methods for coal.

[0014] According to one aspect of the embodiments of the present invention, there is also provided a flotation system, including: an image acquisition device, a sound acquisition device, and a flotation device for coal, where the image acquisition device and the sound acquisition device are both communicatively connected to the flotation device, and the flotation device is used to execute any one of the flotation methods for coal.

[0015] In an embodiment of the present invention, in the flotation method of coal, at least a plurality of groups of original data in a flotation cell and the ash content values corresponding to each group of the original data are used to train a preset Bayesian model to obtain a target Bayesian model. Then, the target Bayesian model is used to classify the flotation data obtained in real time to obtain the target ash content interval corresponding to the flotation data, and the addition amounts of the frother and the collector corresponding to the obtained target ash content interval are determined. Compared with the detection algorithm obtained by machine learning training in the prior art for predicting the target ash content interval corresponding to the flotation data obtained in real time, in this solution, at least a plurality of groups of original data and the ash content values corresponding to each group of the original data are used to train a preset Bayesian model to obtain a target Bayesian model, and then the target Bayesian model is used to classify the flotation data to obtain the target ash content interval corresponding to the flotation data. Since the time complexity of the target Bayesian model is within the range of linear complexity, this ensures that the calculation amount of the target Bayesian model is small and the calculation speed is fast. Moreover, by using the flotation data of the flotation cell collected in real time, the corresponding target ash content interval is determined, and the addition amounts of the frother and the collector corresponding to the target ash content interval are determined, realizing the autonomous intelligence of ash flotation, ensuring a high flotation efficiency, and thus solving the technical problems of large calculation amount and slow calculation speed for real-time calculation of ash content values in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The specification drawings forming a part of this application are used to provide a further understanding of this application. The schematic embodiments of this application and their descriptions are used to explain this application and do not constitute an improper limitation to this application. In the drawings:

[0017] Figure 1 Shows a flowchart of a flotation method of coal according to an embodiment of this application;

[0018] Figure 2 Shows a schematic diagram of a data file structure according to an embodiment of this application;

[0019] Figure 3 Shows a schematic diagram of the structure of a flotation device for coal according to an embodiment of this application;

[0020] Figure 4 Shows a schematic diagram of the structure of a flotation system according to an embodiment of this application.

[0021] Among them, the above-mentioned drawings include the following reference numerals:

[0022] 100, Version number; 101, Header length; 102, Data offset; 103, Header checksum; 104, Identifier; 105, Audio format; 106, Image format; 107, Audio size; 108, Image size; 109, Total length; 110, Padding; 112, Data content; 113, Frame header format; 114, Data; 115, Hard disk video recorder; 116, Coal flotation device; 117, POE switch; 118, Power adapter; 119, Image acquisition device; 120, Sound acquisition device; 121, Intrinsically safe mine pick-up microphone; 122, Industrial camera; 123, Chemical dosing control box; 124, Chemical dosing system equipment. Detailed implementation manners

[0023] It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments may be combined with each other. The present application will be described in detail below with reference to the drawings and in combination with the embodiments.

[0024] In order to enable those skilled in the art to better understand the solution of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0025] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily need to be used to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so as to describe the embodiments of the present application here. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units does not necessarily need to be limited to those clearly listed steps or units, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0026] As described in the background art, the computational amount of real-time calculation of ash content in the prior art is large and the calculation speed is slow. In order to solve the above problems, in a typical implementation manner of the present application, a coal flotation method, a flotation device, a computer-readable storage medium, a processor and a flotation system are provided.

[0027] According to an embodiment of the present application, a coal flotation method is provided.

[0028] Figure 1It is a flowchart of a coal flotation method according to an embodiment of the present application. As Figure 1 shown, the flotation method includes the following steps:

[0029] Step S101, obtaining multiple groups of original data in the flotation cell and the ash content value corresponding to each group of the above original data, wherein the above original data includes original image data and original audio data;

[0030] Step S102, training a preset Bayesian model at least using the multiple groups of the above original data and the corresponding above ash content values to obtain a target Bayesian model;

[0031] Step S103, determining a target ash content range corresponding to the above flotation data based on the above target Bayesian model and the flotation data obtained in real time, and determining the addition amounts of the corresponding frother and collector based on the above target ash content range, wherein the above flotation data includes real-time image data and real-time audio data.

[0032] In the above coal flotation method, a preset Bayesian model is trained at least by obtaining multiple groups of original data in the flotation cell and the ash content values corresponding to each group of the above original data to obtain a target Bayesian model. Then, the target Bayesian model is used to classify the flotation data obtained in real time to obtain a target ash content range corresponding to the above flotation data, and the addition amounts of the corresponding frother and collector are determined based on the obtained target ash content range. Compared with the detection algorithm obtained by machine learning training in the prior art for predicting the target ash content range corresponding to the flotation data obtained in real time, in this solution, a preset Bayesian model is trained at least using multiple groups of original data and the ash content values corresponding to each group of original data to obtain a target Bayesian model, and then the target Bayesian model is used to classify the flotation data to obtain a target ash content range corresponding to the flotation data. Since the time complexity of the target Bayesian model is within the range of linear complexity, this ensures that the calculation amount of the target Bayesian model is small and the calculation speed is fast, and the target ash content range is determined based on the flotation data of the flotation cell collected in real time, and the addition amounts of the corresponding frother and collector are determined based on the target ash content range, realizing the autonomous intelligence of ash flotation, ensuring a high flotation efficiency, and thus solving the technical problems of large calculation amount and slow calculation speed for real-time calculation of ash content values in the prior art.

[0033] In a specific embodiment of the present application, a plurality of explosion-proof high-speed cameras and high-fidelity microphones are installed around the flotation cell. Real-time image data of the flotation cell is obtained through the explosion-proof high-speed cameras. Real-time audio data of the flotation cell is obtained through the high-fidelity microphones. Among them, the device side is configured with an FPGA + ARM + VPU embedded controller (i.e., the coal flotation device). The embedded controller is connected to the explosion-proof high-speed cameras and high-fidelity microphones through communication cables, which is convenient for laying and simple, and reduces the maintenance workload caused by cable breakage.

[0034] In the actual application process, the addition amounts of the foaming agent and the collector corresponding to different target ash intervals can be preset in advance. After obtaining the target ash interval corresponding to the real-time obtained flotation data, the addition amounts of the corresponding foaming agent and collector are determined according to the addition amounts of the foaming agent and the collector corresponding to the target ash interval, and automatic dosing is performed.

[0035] It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. And although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0036] In the actual application process, due to the actual construction site environment of the flotation cell, there will be a lot of interference information in the obtained original data, such as interference from dust, water mist, noise, etc. Therefore, in order to ensure that the obtained target Bayesian model is relatively accurate, and further ensure that the target ash interval obtained by splitting the real-time obtained flotation data according to the target Bayesian model is relatively accurate, in an embodiment of the present application, at least multiple groups of the above-mentioned original data and the corresponding above-mentioned ash values are used to train a preset Bayesian model to obtain a target Bayesian model, including: preprocessing each group of the above-mentioned original data to obtain multiple groups of target data, and the above-mentioned preprocessing at least includes filtering; using each of the above-mentioned target data to train the above-mentioned preset Bayesian model to obtain the above-mentioned target Bayesian model.

[0037] In another embodiment of the present application, preprocessing is performed on each group of the above-mentioned original data to obtain multiple groups of target data, including: using wavelet transform to perform filtering processing on each group of the above-mentioned original data to obtain multiple groups of filtered data, where each group of the above-mentioned filtered data includes filtered image data and filtered audio data. The above-mentioned filtered image data is used to characterize the glossiness and density of bubbles in the flotation cell, and the above-mentioned filtered audio data is used to characterize the breaking sound of the bubbles; performing co-frequency combination processing on each group of the above-mentioned filtered image data and the above-mentioned filtered audio data to obtain multiple groups of the above-mentioned target data. In this embodiment, based on wavelet transform, filtering processing is performed on each group of original data, so that it can be ensured that the multiple groups of filtered data obtained have less interference information. Then, co-frequency combination processing is performed on each group of filtered image data and filtered audio data for data calibration, so that the filtered image data and the filtered audio data can correspond to each other, further ensuring that the subsequent obtained target Bayesian model is relatively accurate and the classification accuracy corresponding to the target Bayesian model is relatively high.

[0038] Specifically, in the actual application process, it is not limited to performing filtering processing on each group of original data through wavelet transform. Other filtering methods can also be used to perform filtering processing on each group of original data, such as low-pass filtering method, band-pass filtering method, clipping filtering method, and so on.

[0039] In a specific embodiment of the present application, the original image data collected by the flameproof camera is processed by a guided filter to remove the useless noise (such as interference such as dust and water mist) in the original image data, and the details such as the original contour of the image are retained to the greatest extent to obtain the filtered image data. The original audio data collected by the flameproof high-fidelity microphone is processed by wavelet transform to filter the noisy signal to obtain the filtered audio data.

[0040] In another specific embodiment of the present application, audio data is collected by a flameproof high-fidelity microphone, and the collection frequency of the flameproof high-fidelity microphone is 50-10 KHz; image data is collected by a flameproof camera, and the collection frequency of the flameproof camera is 20-200 Hz. Then, taking the sampling time of the flameproof high-fidelity microphone as the minimum unit of the update time, and taking 0.5 seconds as a time period, all the filtered audio data within each time period is intercepted; from the image data collected by the flameproof camera, one frame of filtered image data is taken every 0.5 seconds as the image data, and the intercepted filtered audio data and filtered image data are recombined and encapsulated to obtain multiple groups of target data.

[0041] In yet another specific embodiment of the present application, during the process of obtaining target data, an audio-video combined encoding method is further provided. Specifically, the compressed image data and audio data are composed of multiple frames, and a frame is the smallest unit of the data file. Each frame is composed of a frame header and data content. Its length varies with the bit rate. This solution combines the audio data and image data with the same frequency and specifies the data file structure, such as Figure 2As shown, the data file structure includes a frame header format 113 and data 114. Among them, the frame header format 113 includes a version number 100, a header length 101, a data offset 102, a header checksum 103, an identifier 104, an audio format 105, an image format 106, an audio size 107, an image size 108, a total length 109, and padding 110. Specifically, the version number 100 (Version): 4 bits in length. It identifies the currently adopted version number. The header length 101 (Internet Header Length): 4 bits in length. The function of this field is to describe the length of the data packet header. This part occupies 4 bit positions, with a unit of 32 bit (4 bytes), that is, the value of this area = the length of the data frame header (in bit) / (8×4). Therefore, the maximum length of a data frame header is "1111", that is, 15×4 = 60 bytes. The data offset 102 (Data Offset): 8 bits in length. It is used to identify the deviation situation of the sampling time points of audio data and video data. The corresponding value for no deviation is 0. When the deviation value is the largest, the data packet is discarded, and the maximum deviation value is at the 8×256 = 2048 byte position. The header checksum 103 (Header Checksum): 16 bits in length. It is filled by the sending end, and the receiving end uses the CRC algorithm on it to check whether the data packet header is damaged during transmission. The identifier 104 (Identifier): 6 bits in length. It is used to identify the data type contained in the data packet, including audio data, image data, mixed audio and image data, etc. The audio format 105 (Sound Format): 6 bits in length. It refers to the encoding method of digital audio. Different digital audio devices generally correspond to different audio file formats, usually including WAV, MP3, MP3Pro, WMA, MP4, SACD formats, etc. The image format 106 (Image Format): 6 bits in length. It is the format in which the image file is stored on the memory card, usually including JPEG, TIFF, RAW, etc. The audio size 107 (Sound Size): 22 bits in length. The audio size is determined by the bit rate and duration. The bit rate is the amount of data per second of the audio. For audio data of the same format, the higher the bit rate, the better the sound quality. Commonly used high-quality bit rates are: 128 kbps for mp3, 112 kbps for aac (LC), 96 kbps for aac (HE), and 48 kbps for aac (HE-v2). The image size (Image Size): 24 bits in length. The length and width of the image size 108 are in pixels. For example: a picture with a resolution of 640×480 requires approximately 310,000 pixels, and a picture with 2048×1536 requires up to 3.14 million pixels. The higher the resolution of the pixel picture, the more pixels are required. The total length 109 (Total Length): 25 bits in length.The length of the data frame calculated in bytes (including the header and data), so the maximum packet length is 33554431 bytes = 32 MB. Padding 110: Since the header length is not an integer multiple of 32 bits, this field is filled with 0s to adjust it to an integer multiple of 32 bits. Data 114: Stores the specific data content 112, which is a combination of the collected image data and audio data.

[0042] In order to further ensure the classification accuracy of the obtained target Bayesian model, in another embodiment of the present application, the above-mentioned preset Bayesian model is trained using each of the above-mentioned target data to obtain the above-mentioned target Bayesian model, including: determining a plurality of target ash sets according to the ash values corresponding to each group of the above-mentioned original data and the preset ash interval, each of the above-mentioned target ash sets including multiple groups of the above-mentioned original data; constructing the above-mentioned preset Bayesian model where P(c) is the prior probability of the above-mentioned target ash set, and x ij is the jth attribute in the ith group of the above-mentioned original data in the above-mentioned target ash set, the above-mentioned attribute being one of glossiness, density, or breaking sound, w(i) is the weight of the above-mentioned attribute, i starts from 1 and takes values up to n, n being the total number of the above-mentioned original data in the above-mentioned target ash set, and j starts from 1 and takes values up to m, m being the total number of the above-mentioned attributes in a group of the above-mentioned original data; using a plurality of the above-mentioned target ash sets to correct the above-mentioned weight of the above-mentioned preset Bayesian model to obtain the above-mentioned target Bayesian model.

[0043] Specifically, the above-mentioned preset ash interval can be divided according to the actual situation. For example, the above-mentioned preset ash interval can be 8.1 - 8.5, 8.6 - 8.9, etc. Since each group of original data has a corresponding ash value, multiple groups of original data can be divided according to the ash values corresponding to each group of original data and the preset ash interval to obtain a plurality of target ash sets.

[0044] Specifically, the above-mentioned target ash set can be a class of a classification interval, and the glossiness, density, and breaking sound corresponding to each ash value in the above-mentioned target ash set can be the attributes of the above-mentioned target ash set.

[0045] In the actual application process, the Naive Bayes model originates from classical mathematical theory, has stable classification efficiency, and the algorithm itself has a simple logic and is easy to implement. Compared with other classification methods, it has the smallest error rate. When classifying the flotation data in the flotation cell, the number of involved attributes is small and the correlation is not large, and a good classification effect can be obtained. At the same time, considering that in actual applications, the influence of different attributes on the classification result is different, some attributes have a greater impact on the classification while some attributes have a smaller impact on the classification. Therefore, in a specific embodiment of the present application, the above-mentioned preset Bayes model can be constructed based on the weighted Naive Bayes algorithm.

[0046] In another embodiment of the present application, multiple above-mentioned target ash content sets are used to correct the above-mentioned weights of the above-mentioned preset Bayes model to obtain the above-mentioned target Bayes model, including: using the above-mentioned attributes in each of the above-mentioned target ash content sets and the above-mentioned preset Bayes model to calculate multiple classification correct probabilities; according to correct the weights in the above-mentioned preset Bayes model to obtain the above-mentioned target Bayes model, where a j is the classification correct probability of the j-th above-mentioned attribute; w(j) is the above-mentioned weight. Since the attributes in the weighted Naive Bayes algorithm are independent of each other, a classification can be performed separately according to each attribute in each target ash content set to obtain multiple classification correct probabilities, and perform normalization processing on the multiple classification correct probabilities, so as to continuously correct the weights in the preset Bayes model, which ensures that the obtained target Bayes model has a higher classification accuracy and higher precision.

[0047] In order to further ensure that the obtained target ash content interval is more accurate and reduce the influence of other interference information on the classification result, in an embodiment of the present application, based on the above-mentioned target Bayes model and the flotation data obtained in real time, determine the target ash content interval corresponding to the above-mentioned flotation data, including: using wavelet transform to perform filtering processing on the above-mentioned flotation data to obtain filtered flotation data, the above-mentioned filtered flotation data includes filtered flotation image data and filtered flotation audio data, the above-mentioned filtered flotation image data is used to characterize the gloss and density of the bubbles in the above-mentioned flotation cell, and the above-mentioned filtered flotation audio data is used to characterize the breaking sound of the above-mentioned bubbles; perform co-frequency combination processing on each group of the above-mentioned filtered flotation image data and the above-mentioned filtered flotation audio data to obtain multiple groups of target flotation data; use the above-mentioned target Bayes model and the above-mentioned target flotation data to determine the corresponding above-mentioned target ash content interval.

[0048] The embodiment of the present application also provides a coal flotation device. It should be noted that the coal flotation device in the embodiment of the present application can be used to execute the coal flotation method provided by the embodiment of the present application. The following introduces the coal flotation device provided by the embodiment of the present application.

[0049] Figure 3 It is a schematic structural diagram of a coal flotation device according to an embodiment of the present application. As Figure 3 shown, the flotation device includes:

[0050] An acquisition unit 10, configured to acquire multiple groups of original data in the flotation cell, and the ash content value corresponding to each group of the above original data, wherein the above original data includes original image data and original audio data;

[0051] A training unit 20, configured to train a preset Bayesian model at least using multiple groups of the above original data and the corresponding above ash content values to obtain a target Bayesian model;

[0052] A determination unit 30, configured to determine a target ash content interval corresponding to the above flotation data based on the above target Bayesian model and the flotation data acquired in real time, and determine the addition amounts of the corresponding foaming agent and collector based on the above target ash content interval, wherein the above flotation data includes real-time image data and real-time audio data.

[0053] In the above coal flotation device, the acquisition unit is used to acquire multiple groups of original data in the flotation cell, and the ash content value corresponding to each group of the above original data; the training unit is used to train a preset Bayesian model at least through the multiple groups of original data in the flotation cell acquired and the ash content values corresponding to each group of the above original data to obtain a target Bayesian model; the determination unit is used to classify the flotation data acquired in real time through the target Bayesian model to obtain a target ash content interval corresponding to the above flotation data, and determine the addition amounts of the corresponding foaming agent and collector based on the obtained target ash content interval. Compared with the detection algorithm obtained by machine learning training in the prior art for predicting the target ash content interval corresponding to the flotation data acquired in real time, in this solution, at least multiple groups of original data and the ash content values corresponding to each group of original data are used to train a preset Bayesian model to obtain a target Bayesian model, and then the target Bayesian model is used to classify the flotation data to obtain a target ash content interval corresponding to the flotation data. Since the time complexity of the target Bayesian model is within the range of linear complexity, this ensures that the calculation amount of the target Bayesian model is small and the calculation speed is fast, and the target ash content interval is determined through the flotation data of the flotation cell collected in real time, and the addition amounts of the corresponding foaming agent and collector are determined based on the target ash content interval, realizing the autonomous intelligence of ash flotation, ensuring a high flotation efficiency, and thus solving the technical problems of large calculation amount and slow calculation speed for real-time calculation of ash content values in the prior art.

[0054] In a specific embodiment of the present application, a plurality of explosion-proof high-speed cameras and high-fidelity microphones are installed around the flotation cell. The real-time image data of the flotation cell is obtained through the explosion-proof high-speed cameras. The real-time audio data of the flotation cell is obtained through the high-fidelity microphones. Among them, an FPGA+ARM+VPU embedded controller (i.e., the coal flotation device) is configured at the device end. The embedded controller is connected to the explosion-proof high-speed cameras and high-fidelity microphones through communication cables, which is convenient for laying and simple, and reduces the maintenance workload caused by cable breakage.

[0055] In the actual application process, the addition amounts of the foaming agent and the collector corresponding to different target ash intervals can be preset in advance. After obtaining the target ash interval corresponding to the real-time obtained flotation data, the addition amounts of the corresponding foaming agent and collector are determined according to the addition amounts of the foaming agent and the collector corresponding to the target ash interval, and automatic dosing is carried out.

[0056] In the actual application process, due to the actual construction site environment of the flotation cell, there will be a lot of interference information in the obtained original data, such as interference from dust, water mist, noise, etc. Therefore, in order to ensure that the obtained target Bayesian model is relatively accurate, and further ensure that the target ash interval obtained by splitting the real-time obtained flotation data according to the target Bayesian model is relatively accurate, in an embodiment of the present application, the above training unit includes a preprocessing module and a training module. Among them, the above preprocessing module is used to preprocess each group of the above original data to obtain multiple groups of target data, and the above preprocessing at least includes filtering; the above training module is used to use each of the above target data to train the above preset Bayesian model to obtain the above target Bayesian model.

[0057] In another embodiment of the present application, the above-mentioned preprocessing module includes a filtering sub-module and a combining sub-module. Among them, the filtering sub-module is used to perform filtering processing on each group of the above-mentioned original data by using wavelet transform to obtain multiple groups of filtered data. Each group of the above-mentioned filtered data includes filtered image data and filtered audio data. The above-mentioned filtered image data is used to characterize the glossiness and density of the bubbles in the flotation cell, and the above-mentioned filtered audio data is used to characterize the breaking sound of the bubbles. The combining sub-module is used to perform co-frequency combination processing on each group of the above-mentioned filtered image data and the above-mentioned filtered audio data to obtain multiple groups of the above-mentioned target data. In this embodiment, based on wavelet transform, filtering processing is performed on each group of original data, so that it can be ensured that the multiple groups of filtered data obtained have less interference information. Then, co-frequency combination processing is performed on each group of filtered image data and filtered audio data for data calibration, so that the filtered image data and the filtered audio data can correspond to each other, further ensuring that the subsequent obtained target Bayesian model is relatively accurate and the classification accuracy corresponding to the target Bayesian model is relatively high.

[0058] Specifically, in the actual application process, it is not limited to performing filtering processing on each group of original data by using wavelet transform, and other filtering methods can also be used to perform filtering processing on each group of original data, such as low-pass filtering method, band-pass filtering method, clipping filtering method, and so on.

[0059] In a specific embodiment of the present application, the original image data collected by the flameproof camera is processed by a guided filter to remove the useless noise (such as dust, water mist, etc.) in the original image data, and the details such as the original contour of the image are retained to the greatest extent to obtain the filtered image data. The original audio data collected by the flameproof high-fidelity microphone is filtered by wavelet transform to obtain the filtered audio data.

[0060] In another specific embodiment of the present application, audio data is collected by a flameproof high-fidelity microphone, and the sampling frequency of the flameproof high-fidelity microphone is 50 - 10 KHz; image data is collected by a flameproof camera, and the sampling frequency of the flameproof camera is 20 - 200 Hz. Then, taking the sampling time of the flameproof high-fidelity microphone as the minimum unit of the update time, and taking 0.5 seconds as a time period, all the filtered audio data within each time period is intercepted; from the image data collected by the flameproof camera, one frame of filtered image data is taken every 0.5 seconds as the image data, and the intercepted filtered audio data and filtered image data are recombined and encapsulated to obtain multiple groups of target data.

[0061] In another specific embodiment of the present application, during the process of obtaining target data, an audio-video combined coding method is further provided. Specifically, the compressed image data and audio data are composed of multiple frames, and a frame is the smallest unit of the data file. Each frame is composed of a frame header and data content. Its length varies with the bit rate. This solution combines audio data and image data with the same frequency and specifies the data file structure, such as Figure 2As shown, the data file structure includes a frame header format 113 and data 114. Among them, the frame header format 113 includes a version number 100, a header length 101, a data offset 102, a header checksum 103, an identifier 104, an audio format 105, an image format 106, an audio size 107, an image size 108, a total length 109, and padding 110. Specifically, the version number 100 (Version): 4 bits in length. It identifies the version number currently in use. The header length 101 (Internet Header Length): 4 bits in length. The purpose of this field is to describe the length of the data packet header. This part occupies 4 bit positions, with the unit being 32 bit (4 bytes), that is, the value of this area = the length of the data frame header (in bits) / (8×4). Therefore, the maximum length of a data frame header is "1111", that is, 15×4 = 60 bytes. The data offset 102 (Data Offset): 8 bits in length. It is used to identify the deviation situation of the sampling time points of audio data and video data. The corresponding value for no deviation is 0. When the deviation value is the largest, the data packet is discarded, and the maximum deviation value is at the 8×256 = 2048 - byte position. The header checksum 103 (Header Checksum): 16 bits in length. It is filled by the sending end, and the receiving end uses the CRC algorithm on it to check whether the data packet header is damaged during transmission. The identifier 104 (Identifier): 6 bits in length. It is used to identify the data type contained in the data packet, including audio data, image data, mixed audio and image data, etc. The audio format 105 (Sound Format): 6 bits in length. It refers to the encoding method of digital audio. Different digital audio devices generally correspond to different audio file formats, usually including WAV, MP3, MP3Pro, WMA, MP4, SACD formats, etc. The image format 106 (Image Format): 6 bits in length. It is the format in which the image file is stored on the memory card, usually including JPEG, TIFF, RAW, etc. The audio size 107 (Sound Size): 22 bits in length. The audio size is determined by the bit rate and duration. The bit rate is the amount of data per second of the audio. For audio data of the same format, the higher the bit rate, the better the sound quality. Commonly used high - quality bit rates are as follows: mp3 is 128 kbps, aac(LC) is 112 kbps, aac(HE) is 96 kbps, aac(HE - v2) is 48 kbps. The image size (Image Size): 24 bits in length. The length and width of the image size 108 are in pixels. For example, a picture with a resolution of 640×480 requires approximately 310,000 pixels, and a picture with 2048×1536 requires as many as 3.14 million pixels. The higher the resolution of the pixel picture, the more pixels are required. The total length 109 (Total Length): 25 bits in length.The length of the data frame calculated in bytes (including the header and data), so the maximum length of the data packet is 33,554,431 bytes = 32 MB. Padding 110: Since the header length is not an integer multiple of 32 bits, this field is filled with 0s to adjust it to an integer multiple of 32 bits. Data 114: Stores the specific data content 112, which is a combination of the collected image data and audio data.

[0062] In order to further ensure the classification accuracy of the obtained target Bayesian model, in another embodiment of the present application, the above training module includes a determination sub-module, a construction sub-module, and a correction sub-module. Among them, the determination sub-module is used to determine a plurality of target ash value sets according to the ash values corresponding to each group of the above original data and a preset ash value interval, and each of the above target ash value sets includes multiple groups of the above original data; the construction sub-module is used to construct the above preset Bayesian model Among them, P(c) is the prior probability of the above target ash value set, and x ij is the j-th attribute in the i-th group of the above original data in the above target ash value set, and the above attribute is one of glossiness, density, or breaking sound. w(i) is the weight of the above attribute. i starts from 1 and takes values until n, where n is the total number of the above original data in the above target ash value set. j starts from 1 and takes values until m, where m is the total number of the above attributes in a group of the above original data; the correction sub-module is used to use multiple above target ash value sets to correct the above weights of the above preset Bayesian model to obtain the above target Bayesian model.

[0063] Specifically, the above preset ash value interval can be divided according to the actual situation. For example, the above preset ash value interval can be 8.1 - 8.5, 8.6 - 8.9, etc. Since each group of original data has a corresponding ash value, multiple groups of original data can be divided according to the ash values corresponding to each group of original data and the preset ash value interval to obtain a plurality of target ash value sets.

[0064] Specifically, the above target ash value set can be a class of a classification interval, and the glossiness, density, and breaking sound corresponding to each ash value in the above target ash value set can be the attributes of the above target ash value set.

[0065] In the actual application process, the Naive Bayes model originates from classical mathematical theory, has stable classification efficiency, and the logic of the algorithm itself is simple and easy to implement. Compared with other classification methods, it has the smallest error rate. When classifying the flotation data in the flotation cell, the number of attributes involved is small and the correlation is not large, and a good classification effect can be obtained. At the same time, considering that in actual application, the influence of different attributes on the classification result is different, some attributes have a greater impact on the classification while some other attributes have a smaller impact on the classification. Therefore, in a specific embodiment of the present application, the above-mentioned preset Bayes model can be constructed based on the weighted Naive Bayes algorithm.

[0066] In another embodiment of the present application, the above-mentioned correction sub-module includes a calculation sub-module and a weight correction sub-module. Among them, the above-mentioned calculation sub-module is used to calculate multiple classification correct probabilities by using the above-mentioned attributes in each of the above-mentioned target ash content sets and the above-mentioned preset Bayes model; the above-mentioned weight correction sub-module is used to correct the weights in the above-mentioned preset Bayes model to obtain the above-mentioned target Bayes model, where a j is the classification correct probability of the j-th above-mentioned attribute; w(j) is the above-mentioned weight. Since the attributes in the weighted Naive Bayes algorithm are independent of each other, multiple classification correct probabilities can be obtained by performing classification separately according to the attributes in each target ash content set, and normalize the multiple classification correct probabilities, so as to continuously correct the weights in the preset Bayes model, which ensures that the obtained target Bayes model has a higher classification accuracy and a higher precision.

[0067] In order to further ensure that the obtained target ash content interval is more accurate and reduce the influence of other interference information on the classification result, in an embodiment of the present application, the above-mentioned determination unit includes a filtering processing module, a same-frequency combination module and a determination module. Among them, the above-mentioned filtering processing module is used to perform filtering processing on the above-mentioned flotation data by using wavelet transform to obtain filtered flotation data. The above-mentioned filtered flotation data includes filtered flotation image data and filtered flotation audio data. The above-mentioned filtered flotation image data is used to characterize the glossiness and density of the bubbles in the above-mentioned flotation cell, and the above-mentioned filtered flotation audio data is used to characterize the breaking sound of the above-mentioned bubbles; the above-mentioned same-frequency combination module is used to perform same-frequency combination processing on each group of the above-mentioned filtered flotation image data and the above-mentioned filtered flotation audio data to obtain multiple groups of target flotation data; the above-mentioned determination module is used to determine the corresponding above-mentioned target ash content interval by using the above-mentioned target Bayes model and the above-mentioned target flotation data.

[0068] The above-mentioned coal flotation device includes a processor and a memory. The above-mentioned acquisition unit, training unit, determination unit, etc. are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to implement corresponding functions.

[0069] The processor contains a kernel, and the kernel retrieves the corresponding program units from the memory. One or more kernels can be set, and by adjusting the kernel parameters, the technical problems of large computational complexity and slow calculation speed in real-time calculation of ash content values in the prior art can be solved.

[0070] The memory may include non-permanent memory in a computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. The memory includes at least one storage chip.

[0071] An embodiment of the present invention provides a computer-readable storage medium, on which a program is stored, and when the program is executed by a processor, the above-mentioned coal flotation method is implemented.

[0072] An embodiment of the present invention provides a processor, which is used to run a program, and when the program runs, the above-mentioned coal flotation method is executed.

[0073] In a typical embodiment of the present application, a flotation system is further provided, as Figure 4 shown. The flotation system includes an image acquisition device 119, a sound acquisition device 120, and a coal flotation device 116. Among them, the above-mentioned image acquisition device 119 and the above-mentioned sound acquisition device 120 are both communicatively connected to the above-mentioned coal flotation device 116, and the above-mentioned coal flotation device 116 is used to execute any one of the above-mentioned coal flotation methods.

[0074] The above flotation system includes an image acquisition device, a sound acquisition device, and a coal flotation device. The flotation device is communicatively connected to the image acquisition device and the sound acquisition device, and the flotation device is configured to perform any one of the above coal flotation methods. In the above flotation method, at least by using multiple groups of original data in the flotation cell and the ash content values corresponding to each group of the original data, a preset Bayesian model is trained to obtain a target Bayesian model. Then, the target Bayesian model is used to classify the real-time acquired flotation data to obtain the target ash content range corresponding to the flotation data, and the addition amounts of the foaming agent and the collector corresponding to the obtained target ash content range are determined. Compared with the detection algorithm obtained by machine learning training in the prior art for predicting the target ash content range corresponding to the real-time acquired flotation data, in this solution, at least multiple groups of original data and the ash content values corresponding to each group of the original data are used to train a preset Bayesian model to obtain a target Bayesian model, and then the target Bayesian model is used to classify the flotation data to obtain the target ash content range corresponding to the flotation data. Since the time complexity of the target Bayesian model is within the range of linear complexity, this ensures that the computational amount of the target Bayesian model is small and the computational speed is fast. Moreover, by using the flotation data of the real-time acquired flotation cell, the corresponding target ash content range is determined, and the addition amounts of the foaming agent and the collector corresponding to the target ash content range are determined, realizing the autonomous intelligence of ash flotation, ensuring a high flotation efficiency, and thus solving the technical problems of large computational amount and slow computational speed for real-time calculation of ash content values in the prior art.

[0075] In a specific embodiment of the present application, as Figure 4 shown, the flotation system further includes a mine intrinsically safe sound pick-up machine 121, an industrial camera 122, a dosing control box 123, a dosing system device 124, a POE switch 117 (Power Over Ethernet, abbreviated as POE), a power adapter 118, and a hard disk video recorder 115.

[0076] An embodiment of the present invention provides a device, which includes a processor, a memory, and a program stored on the memory and executable on the processor. When the processor executes the program, it realizes at least the following steps:

[0077] Step S101, obtaining multiple groups of original data in the flotation cell, and the ash content value corresponding to each group of the original data, wherein the original data includes original image data and original audio data;

[0078] Step S102, training a preset Bayesian model at least using multiple groups of the original data and the corresponding ash content values to obtain a target Bayesian model;

[0079] Step S103: Based on the above target Bayesian model and the flotation data obtained in real time, determine the target ash interval corresponding to the above flotation data, and determine the addition amounts of the frother and collector corresponding to the above target ash interval, wherein the above flotation data includes real-time image data and real-time audio data.

[0080] The devices in this article can be servers, PCs, PADs, mobile phones, etc.

[0081] The present application also provides a computer program product, which, when executed on a data processing device, is adapted to execute a program initialized with at least the following method steps:

[0082] Step S101: Obtain multiple groups of original data in the flotation cell and the ash value corresponding to each group of the above original data, wherein the above original data includes original image data and original audio data;

[0083] Step S102: At least use multiple groups of the above original data and the corresponding above ash values to train a preset Bayesian model to obtain a target Bayesian model;

[0084] Step S103: Based on the above target Bayesian model and the flotation data obtained in real time, determine the target ash interval corresponding to the above flotation data, and determine the addition amounts of the frother and collector corresponding to the above target ash interval, wherein the above flotation data includes real-time image data and real-time audio data.

[0085] In the above embodiments of the present invention, the descriptions of the respective embodiments have their own emphases. For parts not detailed in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0086] In several embodiments provided by the present application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only illustrative. For example, the above unit division can be a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of units or modules can be in electrical or other forms.

[0087] The units described above as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0088] In addition, in each embodiment of the present invention, each functional unit can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0089] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the above-mentioned methods in each embodiment of the present invention. The foregoing storage medium includes: various media such as USB flash drives, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), mobile hard disks, magnetic disks, or optical discs that can store program codes.

[0090] From the above description, it can be seen that the above embodiments of the present application achieve the following technical effects:

[0091] 1) In the coal flotation method of the present application, at least by obtaining multiple sets of original data in the flotation cell and the ash content values corresponding to each set of the above original data, a preset Bayesian model is trained to obtain a target Bayesian model. Then, the target Bayesian model is used to classify the flotation data obtained in real time to obtain the target ash content interval corresponding to the above flotation data, and the addition amounts of the corresponding foaming agent and collector are determined based on the obtained target ash content interval. Compared with the detection algorithm obtained by machine learning training in the prior art for predicting the target ash content interval corresponding to the flotation data obtained in real time, in this solution, at least multiple sets of original data and the ash content values corresponding to each set of original data are used to train a preset Bayesian model to obtain a target Bayesian model, and then the target Bayesian model is used to classify the flotation data to obtain the target ash content interval corresponding to the flotation data. Since the time complexity of the target Bayesian model is within the range of linear complexity, this ensures that the calculation amount of the target Bayesian model is small and the calculation speed is fast, and by using the flotation data of the flotation cell collected in real time, the corresponding target ash content interval is determined and the addition amounts of the corresponding foaming agent and collector are determined based on the target ash content interval, realizing the autonomous intelligence of ash flotation and ensuring a high flotation efficiency, thus solving the technical problems of large calculation amount and slow calculation speed for real-time calculation of ash content values in the prior art.

[0092] 2) In the flotation device for coal of the present application, the acquisition unit is used to acquire multiple groups of original data in the flotation cell and the ash content value corresponding to each group of the above-mentioned original data; the training unit is used to train a preset Bayesian model at least through the multiple groups of original data in the flotation cell obtained and the ash content values corresponding to each group of the above-mentioned original data to obtain a target Bayesian model; the determination unit is used to classify the flotation data obtained in real time through the target Bayesian model to obtain the target ash content interval corresponding to the above-mentioned flotation data, and determine the addition amounts of the corresponding foaming agent and collector based on the obtained target ash content interval. Compared with the detection algorithm obtained by machine learning training in the prior art for predicting the target ash content interval corresponding to the flotation data obtained in real time, in this solution, at least multiple groups of original data and the ash content values corresponding to each group of original data are used to train a preset Bayesian model to obtain a target Bayesian model, and then the target Bayesian model is used to classify the flotation data to obtain the target ash content interval corresponding to the flotation data. Since the time complexity of the target Bayesian model is within the range of linear complexity, this ensures that the computational amount of the target Bayesian model is small and the computational speed is fast, and by using the flotation data of the flotation cell collected in real time, the corresponding target ash content interval is determined and the addition amounts of the corresponding foaming agent and collector are determined based on the target ash content interval, realizing the autonomous intelligence of ash flotation, ensuring a high flotation efficiency, and thus solving the technical problems of large computational amount and slow computational speed for real-time calculation of ash content values in the prior art.

[0093] 3) The flotation system of the present application includes an image acquisition device, a sound acquisition device, and a coal flotation device. The flotation device is communicatively connected to the image acquisition device and the sound acquisition device, and the flotation device is configured to perform any of the above-mentioned coal flotation methods. In the above-mentioned flotation method, at least by obtaining multiple sets of original data in the flotation cell and the ash content values corresponding to each set of the original data, a preset Bayesian model is trained to obtain a target Bayesian model. Then, the target Bayesian model is used to classify the real-time obtained flotation data to obtain the target ash content range corresponding to the flotation data, and the addition amounts of the corresponding foaming agent and collector are determined based on the obtained target ash content range. Compared with the detection algorithm obtained by machine learning training in the prior art for predicting the target ash content range corresponding to the real-time obtained flotation data, in this solution, at least multiple sets of original data and the ash content values corresponding to each set of the original data are used to train a preset Bayesian model to obtain a target Bayesian model, and then the target Bayesian model is used to classify the flotation data to obtain the target ash content range corresponding to the flotation data. Since the time complexity of the target Bayesian model is within the range of linear complexity, this ensures that the computational amount of the target Bayesian model is small and the computational speed is fast. Moreover, by using the real-time collected flotation data of the flotation cell, the corresponding target ash content range is determined and the addition amounts of the corresponding foaming agent and collector are determined based on the target ash content range, realizing the autonomous intelligence of ash flotation, ensuring a high flotation efficiency, and thus solving the technical problems of large computational amount and slow computational speed for real-time calculating the ash content value in the prior art.

[0094] The foregoing are only the preferred embodiments of the present application and are not intended to limit the present application. For those skilled in the art, various modifications and changes can be made to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

Claims

1. A flotation method for coal, characterized in that, Including: Obtaining multiple groups of original data in the flotation cell and the ash content value corresponding to each group of the original data, where the original data includes original image data and original audio data; Training a preset Bayesian model at least using multiple groups of the original data and the corresponding ash content values to obtain a target Bayesian model; Based on the target Bayesian model and the flotation data obtained in real time, determining the target ash content interval corresponding to the flotation data, and determining the addition amounts of the corresponding frother and collector based on the target ash content interval, where the flotation data includes real-time image data and real-time audio data; Training a preset Bayesian model at least using multiple groups of the original data and the corresponding ash content values to obtain a target Bayesian model, including: Performing preprocessing on each group of the original data to obtain multiple groups of target data, where the preprocessing at least includes filtering; Training the preset Bayesian model using each of the target data to obtain the target Bayesian model; Performing preprocessing on each group of the original data to obtain multiple groups of target data, including: Using wavelet transform to perform filtering on each group of the original data to obtain multiple groups of filtered data, where each group of the filtered data includes filtered image data and filtered audio data, the filtered image data is used to characterize the glossiness and density of the bubbles in the flotation cell, and the filtered audio data is used to characterize the sound of the bubbles breaking; Performing co-frequency combination processing on each group of the filtered image data and the filtered audio data to obtain multiple groups of the target data.

2. The flotation method according to claim 1, wherein Training the preset Bayesian model using each of the target data to obtain the target Bayesian model, including: Determining multiple target ash content sets according to the ash content values corresponding to each group of the original data and a preset ash content interval, and each of the target ash content sets includes multiple groups of the original data; Construct the preset Bayesian model where P(c) is the prior probability of the target ash set, and x ij is the j-th attribute in the i-th group of the original data in the target ash set, the attribute is one of glossiness, density, or breaking sound, w(i) is the weight of the attribute, i starts from 1 and goes up to n, n is the total number of the original data in the target ash set, j starts from 1 and goes up to m, and m is the total number of the attributes in a group of the original data; Using the multiple target ash content sets to correct the weights of the preset Bayesian model to obtain the target Bayesian model.

3. The flotation method according to claim 2, characterized in that, Using the multiple target ash content sets to correct the weights of the preset Bayesian model to obtain the target Bayesian model, including: Calculating multiple classification correct probabilities using the attributes in each of the target ash content sets and the preset Bayesian model; According to correct the weights in the preset Bayesian model to obtain the target Bayesian model, where a j is the correct classification probability of the j-th attribute; w(j) is the weight.

4. The flotation method according to any one of claims 1 to 3, characterized in that Based on the target Bayesian model and the flotation data obtained in real time, determining the target ash content interval corresponding to the flotation data, including: Using wavelet transform to perform filtering on the flotation data to obtain filtered flotation data, the filtered flotation data includes filtered flotation image data and filtered flotation audio data, the filtered flotation image data is used to characterize the glossiness and density of the bubbles in the flotation cell, and the filtered flotation audio data is used to characterize the sound of the bubbles breaking; Performing co-frequency combination processing on each group of the filtered flotation image data and the filtered flotation audio data to obtain multiple groups of target flotation data; Using the target Bayesian model and the target flotation data to determine the corresponding target ash content interval.

5. A flotation device for coal, characterized in that, Including: An acquisition unit, configured to acquire multiple groups of original data in a flotation cell, and the ash content value corresponding to each group of the original data, wherein the original data includes original image data and original audio data; A training unit, configured to train a preset Bayesian model at least using the multiple groups of the original data and the corresponding ash content values to obtain a target Bayesian model; A determination unit, configured to determine a target ash content range corresponding to the flotation data based on the target Bayesian model and the flotation data acquired in real time, and determine the addition amounts of the corresponding frother and collector based on the target ash content range, wherein the flotation data includes real-time image data and real-time audio data; The training unit includes: A preprocessing module, configured to preprocess each group of the original data to obtain multiple groups of target data, and the preprocessing at least includes filtering; A training module, configured to train the preset Bayesian model using each of the target data to obtain the target Bayesian model; The preprocessing module includes: A filtering sub-module, configured to perform filtering on each group of the original data using wavelet transform to obtain multiple groups of filtered data, wherein each group of the filtered data includes filtered image data and filtered audio data, the filtered image data is used to characterize the glossiness and density of the bubbles in the flotation cell, and the filtered audio data is used to characterize the sound of the bubbles breaking; A combination sub-module, configured to perform co-frequency combination processing on each group of the filtered image data and the filtered audio data to obtain multiple groups of the target data.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein the program executes the flotation method of coal according to any one of claims 1 to 4.

7. A processor, characterized in that, The processor is configured to run a program, wherein when the program runs, it executes the flotation method of coal according to any one of claims 1 to 4.

8. A flotation system, characterized in that, Including: An image acquisition device, a sound acquisition device, and a coal flotation device, wherein the image acquisition device and the sound acquisition device are both communicatively connected to the flotation device, and the flotation device is configured to execute the flotation method of coal according to any one of claims 1 to 4.

Images