A cyclone detection method and apparatus

Abstract

The application discloses a cyclone detection method and device, and the method comprises the following steps: establishing a sound and vibration monitoring parameter database and a fault attribution database based on preprocessed sound and vibration data; monitoring the cyclone based on the sound and vibration data and the preprocessed sound and vibration data; collecting running process data; establishing a running process parameter database based on the running process data; and optimizing the running process parameters of the cyclone based on the running process parameter database. Based on the collected sound and vibration data, the semi-supervised learning method is used to predict the running faults of the cyclone, so that the predictive maintenance of the equipment running state is realized. The big data analysis based on reinforcement learning is carried out on the running process parameters and the sound and vibration monitoring parameters in the field, the running process parameter optimization of the production process is realized, and finally the intelligent management and control of the cyclone equipment are realized.

Description

Technical Field

[0001] This invention belongs to the field of mineral processing technology, and specifically relates to a hydrocyclone detection method and device. Background Technology

[0002] Hydrocyclones are devices that use centrifugal force to classify mineral slurries and are one of the important process equipment in mineral processing plants. They are typically used to separate suspended solids carried in liquids into multiple discharge streams or "phases" of different particle sizes. In mining engineering, because the particle size of the product from the grinding system directly affects subsequent operations, and mineral particles have strong abrasive properties, hydrocyclones suffer from problems such as rapid wear, short service life, and difficulty in controlling process parameters. Therefore, a hydrocyclone detection method, device, and equipment have been invented to monitor the operating status of hydrocyclone equipment and optimize its operating parameters.

[0003] Chinese patent discloses a method for operating a hydrocyclone (publication number CN102947006B). The hydrocyclone includes a separation chamber configured during use to generate an internal air core for material separation. The method includes measuring vibration parameters of the separation chamber and stability parameters of the internal air core during operation of the hydrocyclone. The measurement results are compared with predetermined relevant parameters indicating stable operation of the hydrocyclone, and the operating parameters of the hydrocyclone are adjusted based on the comparison results.

[0004] The shortcomings of existing technologies are: 1) Vibration signals are used alone for wear detection, or vibration parameters are used alone for operation monitoring, without taking both functions into account; 2) Using vibration signals for wear detection cannot accurately and timely determine whether the hydrocyclone needs maintenance; 3) Using vibration signals for operation status monitoring cannot take into account multiple process parameters of equipment operation, and no operation parameter optimization is performed. Summary of the Invention

[0005] To address the above problems, this invention discloses a hydrocyclone detection method, comprising:

[0006] Establish an acoustic vibration monitoring parameter database and a fault attribution database based on the preprocessed acoustic vibration data;

[0007] The hydrocyclone is monitored based on acoustic and vibration data and pre-processed acoustic and vibration data;

[0008] Collect operational process data;

[0009] Establish an operating process parameter database based on the aforementioned operating process data;

[0010] The operating process parameters of the hydrocyclone are optimized based on the aforementioned operating process parameter database.

[0011] Furthermore, the establishment of the acoustic vibration monitoring parameter database and fault attribution database based on the preprocessed acoustic vibration data includes the following steps:

[0012] Collect acoustic and vibration data;

[0013] Preprocess the acoustic and vibration data;

[0014] The preprocessed acoustic and vibration data were analyzed to determine the spectral characteristic values ​​of the hydrocyclone.

[0015] Calibrate the range of spectral characteristic values ​​of the hydrocyclone under different operating conditions;

[0016] Based on acoustic and vibration data, preprocessed acoustic and vibration data, and spectral characteristic values ​​corresponding to different operating conditions of the hydrocyclone, an acoustic and vibration monitoring parameter database and a fault attribution database are established.

[0017] Furthermore, the spectral characteristic values ​​are one or more of amplitude, frequency, phase, velocity, acceleration, rate of change of amplitude, rate of change of frequency, rate of change of phase, rate of change of velocity, and rate of change of acceleration.

[0018] Furthermore, the specific steps for monitoring the hydrocyclone based on acoustic vibration data and preprocessed acoustic vibration data are as follows:

[0019] Based on acoustic and vibration data and preprocessed acoustic and vibration data, a semi-supervised learning method combining supervised learning based on few samples and unsupervised learning results based on classification samples is used to analyze the failure trend of hydrocyclones.

[0020] Furthermore, establishing the operating process parameter database based on the operating process data includes the following steps:

[0021] The operating process data of the hydrocyclone under different operating conditions were calibrated.

[0022] A database of operating process parameters is established based on the operating process data and the different operating conditions of the corresponding hydrocyclones.

[0023] Furthermore, the optimization of the hydrocyclone's operating process parameters based on the operating process parameter database includes the following steps:

[0024] Multivariate nonlinear regression analysis was performed on the operating process data and the pre-processed acoustic and vibration data;

[0025] To obtain the functional relationship between production efficiency, energy consumption, and operating process parameters;

[0026] Optimize real-time process parameters based on function correlation.

[0027] Furthermore, after optimizing the operating process parameters of the hydrocyclone based on the operating process parameter database, the process further includes:

[0028] An intelligent management platform for hydrocyclones will be established based on the acoustic and vibration monitoring parameter database, the fault attribution database, and the operating process parameter database.

[0029] The intelligent management platform will be showcased.

[0030] Furthermore, the intelligent management platform includes: a real-time display interface for hydrocyclone operation, underlying data acquisition and storage management, data storage and database management, equipment lifecycle management, production management, safety management, and energy management.

[0031] A hydrocyclone detection device based on the above-mentioned hydrocyclone detection method includes: an acoustic vibration sensor, an acoustic vibration data acquisition edge unit, an acoustic vibration analysis system server, a process data acquisition unit, and a data storage server.

[0032] The acoustic vibration sensor is mounted on the hydrocyclone and connected to the acoustic vibration data acquisition edge unit;

[0033] The acoustic and vibration data acquisition edge unit is connected to the acoustic and vibration analysis system server;

[0034] The acoustic and vibration analysis system server is connected to the data storage server;

[0035] The process data acquisition unit is connected to the data storage server.

[0036] Furthermore, it also includes: deep learning and reinforcement learning analyzers and intelligent platform display interfaces;

[0037] The deep learning and reinforcement learning analyzers and the intelligent platform display interface are respectively connected to the data storage server.

[0038] Compared with the prior art, the embodiments of the present invention have at least the following advantages: they realize the monitoring of the equipment operating status and optimization of the operating process parameters of hydrocyclones in mineral processing plants, and display and manage them through an intelligent management platform;

[0039] (1) The sound and vibration parameters of the hydrocyclone are collected by a contact-type integrated acoustic and vibration sensor, which realizes the early diagnosis and attribution of hydrocyclone faults by acoustic signature and the directional diagnosis and attribution of vibration. Compared with the method of attribution by vibration signal alone, the fault can be predicted earlier.

[0040] (2) Based on the collected acoustic and vibration data, a semi-supervised learning method combining supervised learning based on few samples and unsupervised learning results based on classification samples is used to predict the operation faults of the hydrocyclone, form a fault attribution database, realize predictive maintenance of equipment operation status, and solve the problem of rapid wear and short service life of the hydrocyclone.

[0041] (3) Conduct big data analysis based on reinforcement learning on the on-site operating process parameters and acoustic and vibration monitoring parameters to optimize the operating process parameters of the production process, and finally realize the intelligent management and control of the hydrocyclone equipment, so as to better solve the problem of the difficulty in controlling the operating process parameters and discharge particle size of the hydrocyclone.

[0042] (4) An intelligent management platform for hydrocyclones has been implemented, including underlying data acquisition and storage management, data storage and database management, equipment lifecycle management, production management, safety management, and energy management, achieving comprehensive management of hydrocyclones. This management platform can be extended to the intelligent management of other equipment or processes.

[0043] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention can be realized and obtained by means of the structures pointed out in the description and the drawings. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 A schematic diagram of the structure of a hydrocyclone detection device according to an embodiment of the present invention is shown.

[0046] Reference numerals in the attached figures: 1. Acoustic vibration sensor; 2. Acoustic vibration data acquisition edge unit; 3. Acoustic vibration analysis system server; 4. Process data acquisition unit; 5. Data storage server; 6. Deep learning and reinforcement learning analyzer; 7. Intelligent platform display interface. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] The present invention proposes a hydrocyclone detection method, comprising:

[0049] Establish an acoustic vibration monitoring parameter database and a fault attribution database based on the preprocessed acoustic vibration data;

[0050] The hydrocyclone is monitored based on acoustic and vibration data and pre-processed acoustic and vibration data; in particular, acoustic data is considered at low frequencies and vibration data is considered at high frequencies.

[0051] Collect operational process data;

[0052] The process data acquisition unit 4 develops data acquisition terminal programs for OPC, DB, files and other data types, and uploads sound and vibration signals to the data acquisition edge unit 2 on the data acquisition network for preliminary noise removal and cleaning, and then sends them to the data storage server 5 together with the running process data.

[0053] Establish an operating process parameter database based on the aforementioned operating process data;

[0054] The operating process parameters of the hydrocyclone are optimized based on the aforementioned operating process parameter database.

[0055] In some embodiments, establishing an acoustic vibration monitoring parameter database and a fault attribution database based on preprocessed acoustic vibration data includes the following steps:

[0056] Install the acoustic vibration sensor 1; for example, such as Figure 1 As shown, three acoustic vibration sensors 1 are installed on the outer wall of the hydrocyclone. The first acoustic vibration sensor 1 and the second acoustic vibration sensor 1 are installed on the same horizontal plane, and the angle between the two and the perpendicular line of the hydrocyclone axis is 90°. The third acoustic vibration sensor 1 is located on the plane where the second acoustic vibration sensor 1 and the hydrocyclone axis are located, and is located below the second acoustic vibration sensor 1.

[0057] Collect acoustic and vibration data;

[0058] The specific steps are as follows: connect the acoustic vibration sensor 1 to the acoustic vibration data acquisition edge unit 2 via a network cable, and transmit the sensor data to the acoustic vibration data acquisition edge unit 2;

[0059] Preprocess the acoustic and vibration data;

[0060] The acoustic and vibration data acquisition edge unit 2 can perform preliminary cleaning of the acquired sensor data and transmit the valid data to the acoustic and vibration analysis system server 3 via wired or wireless transmission. The preliminary cleaning specifically involves noise reduction and filtering.

[0061] The preprocessed acoustic and vibration data were analyzed, focusing on sound data at low frequencies and vibration data at high frequencies to determine the spectral characteristics and other acoustic features of the hydrocyclone. Different operating conditions were considered, including normal and fault conditions. Normal operating conditions included different particle size ranges and throughputs, while fault conditions included hydrocyclone cavitation, large fluctuations in mine pressure, screen blockage, and severe wear of the underflow nozzle. Spectral characteristics included one or more of amplitude, frequency, phase, velocity, acceleration, rate of change of amplitude, rate of change of frequency, rate of change of phase, rate of change of velocity, and rate of change of acceleration. Other acoustic features included sound pressure level, sound power, sound intensity, and noise level. Compared to normal operating conditions, fault conditions resulted in changes in parameters such as amplitude, frequency, phase, velocity, acceleration, and their rates of change, manifesting as increased vibration amplitude, spectral curve shift, multiple harmonic shifts, and a large number of harmonics.

[0062] The acoustic and vibration analysis system server 3 stores and processes the pre-processed acoustic and vibration data, analyzing the spectral characteristics of the hydrocyclone. Specifically, the acoustic and vibration data includes time-domain waveform data and a spectrum. The vibration amplitude is obtained from the waveform. Based on the internal air core and fluid flow characteristics of the hydrocyclone, the spectral characteristics are divided into low-frequency acoustic waveform data and high-frequency vibration data. The stability of the hydrocyclone under different operating conditions is determined by observing the changing trends of the acoustic waves and vibration characteristic values. The cause of hydrocyclone failure is determined by observing the changes in the time-frequency domain waveform and the power frequency and other characteristic spectra in the frequency domain, and this serves as reference data for fault diagnosis. Simultaneously, both the pre- and post-processed acoustic and vibration data can communicate in real-time with the data storage server 5 through an open API interface.

[0063] Calibrate the range of spectral characteristic values ​​of the hydrocyclone under different operating conditions;

[0064] Collect spectral characteristic values ​​of the hydrocyclone under normal operating conditions and various fault conditions, and calibrate them to obtain parameters of the sensor state corresponding to different fault conditions.

[0065] Based on the original acoustic and vibration data, the preprocessed acoustic and vibration data, and the corresponding spectral characteristic value data of the hydrocyclone under different operating conditions, a multi-branch tree node database structure is established by extracting and analyzing the characteristic relationships between the data, and an acoustic and vibration monitoring parameter database and a fault attribution database are established.

[0066] Data storage server 5 will store cyclone parameter data and feature relationship data. This data is mainly in the form of a multi-branch tree node structure, so a low-latency distributed graph database, most suitable for this data structure, is chosen for storage. A virtual containerized deployment approach is adopted, deploying multiple graph database nodes on a single server to maximize the utilization of the server's multi-core processor hardware resources, resulting in optimal query performance. Furthermore, since the graph database supports GraphQL queries, it allows for on-demand and multi-hop queries, enabling efficient operation on complex tree-like node structures. Data storage server 5 needs to store raw acoustic vibration data and preprocessed acoustic vibration data. This data mainly consists of massive binary files, with file sizes ranging from hundreds of KB to tens of MB depending on the sampling frequency and sampling length settings. This data is the main source of storage pressure. Object storage is used to store this file data. The flat design of object storage can be leveraged to minimize the storage and read / write pressure of managing massive amounts of small file metadata, achieving the fastest file read / write performance. Simultaneously, the file lifecycle management function of object storage can easily implement essential file expiration deletion and data dilution functions. Data storage server 5 also needs to store acoustic and vibration spectrum characteristic value data and the corresponding cyclone state data. This data is primarily structured as a time series, and storing it in a time-series database provides optimal read and write performance. Furthermore, time-series databases typically have dedicated compression algorithms for time-series data, significantly saving storage space. Additionally, the query interfaces of time-series databases generally support features designed for time series, such as latest queries, time-period queries, aggregate queries, null value handling, result pagination, and time-series alignment, greatly facilitating the application layer's processing and manipulation of this data.

[0067] In some embodiments, the specific steps for monitoring the hydrocyclone based on acoustic vibration data and preprocessed acoustic vibration data are as follows:

[0068] Acoustic and vibration data mining: A deep learning and reinforcement learning analyzer 6 reads acoustic and vibration data and preprocessed acoustic and vibration data from the data storage server 5. Through deep learning algorithms, especially a semi-supervised learning method combining supervised learning based on few samples and unsupervised learning based on classification samples, fault trend analysis is performed on the hydrocyclone to achieve full-cycle management of the hydrocyclone. Using big data statistical analysis methods, the distribution patterns (Berle distribution, normal distribution, etc.) of each parameter (amplitude, frequency, phase, velocity, acceleration) in the acoustic and vibration spectrum data are obtained. Upper and lower thresholds are set according to a certain confidence level as a basis for fault diagnosis. Simultaneously, data augmentation is performed through component obfuscation and noise obfuscation. Negative samples and pseudo-labels are generated from the original samples (original samples formed by preprocessed acoustic and vibration data and operational process data) to solve the problem of scarce fault state label data and provide a guarantee for data representation extraction. A neural network structure and loss function (exemplary, support vector machine or deep residual network) are designed to perform a self-supervised agent task based on the supervision signal provided by the pseudo-labels, extracting data representations related to the health status, reducing the impact of data variability, and lowering the false alarm rate. Negative samples generated through data augmentation can constrain the distribution characteristics of observed data under different health conditions, providing a guarantee for the data separability assumption required for condition monitoring and reducing the false negative rate. For example, upper and lower thresholds for acoustic and vibration data and operational process data are set by combining the calibrated range of sensor parameters under normal conditions with a conventional 70% confidence level, serving as a basis for fault diagnosis. Then, reinforcement learning is used to continuously improve the confidence level, ultimately raising it to over 85%.

[0069] In some embodiments, establishing an operating process parameter database based on the operating process data includes the following steps:

[0070] The operating process data of the hydrocyclone under different operating conditions were calibrated.

[0071] The system collects and calibrates operational data for the hydrocyclone under various operating conditions to obtain the corresponding operational parameters for each sensor. Specific operational data includes the slurry concentration and pressure entering the hydrocyclone, processing capacity, and different particle sizes at the hydrocyclone outlet. Normal operating conditions include different particle size ranges and processing volumes; fault conditions include hydrocyclone cavitation, large pressure fluctuations, screen blockage, and severe wear of the underflow nozzle.

[0072] Data storage server 5 establishes an operating process parameter database based on operating process data and the different operating conditions of the corresponding hydrocyclones.

[0073] In some embodiments, the specific content of optimizing the operating process parameters of the hydrocyclone based on the operating process parameter database is as follows:

[0074] A deep learning and reinforcement learning analyzer 6 reads the operational process data and preprocessed acoustic and vibration data from the data storage server 5. A big data analysis method based on reinforcement learning is used to optimize the operational process parameters, enabling production and energy management of the hydrocyclone. The calculation results are then returned to the data storage server 5. Utilizing prior knowledge (including mechanistic knowledge obtained from multi-energy domain dynamic models and theoretical logic based on empirical predictions), data-level fusion can be performed on corresponding multi-parameter parameters to construct new state parameters, providing richer and more refined data and feature inputs. Multivariate nonlinear regression analysis is performed on the operational process data and preprocessed acoustic and vibration data. Combined with long-term historical data, the correlations between various parameters can be mined, especially the functional correlations between production efficiency parameters, energy consumption parameters, and the collected operational process parameters. These functional correlations are then used to optimize real-time operational process parameters, providing suggestions for adjusting the operational process data. Specifically, the operational process parameters are adjusted to optimize the output particle size, throughput, and hydrocyclone lifespan, including input concentration, flow rate, pressure, overflow pipe insertion depth, and underflow diameter.

[0075] In some embodiments, after optimizing the operating process parameters of the hydrocyclone based on the operating process parameter database, the method further includes:

[0076] An intelligent management platform for hydrocyclones is established based on the acoustic and vibration monitoring parameter database, fault attribution database, and operating process parameter database stored on data storage server 5. The platform includes a real-time display interface for hydrocyclone operation, underlying data acquisition and storage management, data storage and database management, equipment lifecycle management, production management, safety management, and energy management, enabling comprehensive management of hydrocyclones.

[0077] The intelligent management platform will be showcased. The intelligent management platform for the hydrocyclone will be displayed on the intelligent platform display interface 7 using a smart touchscreen or LCD screen.

[0078] Based on the above-mentioned hydrocyclone detection method, this embodiment proposes a hydrocyclone detection device, including: an acoustic vibration sensor 1, an acoustic vibration data acquisition edge unit 2, an acoustic vibration analysis system server 3, a process data acquisition unit 4, and a data storage server 5;

[0079] The acoustic vibration sensor 1 is mounted on the hydrocyclone and connected to the acoustic vibration data acquisition edge unit 2;

[0080] The acoustic and vibration data acquisition edge unit 2 is connected to the acoustic and vibration analysis system server 3;

[0081] The acoustic and vibration analysis system server 3 is connected to the data storage server 5;

[0082] The process data acquisition unit 4 is connected to the data storage server 5.

[0083] The acoustic and vibration sensor 1 is an integrated acoustic and vibration sensor that can collect acoustic and vibration data from the hydrocyclone. While the acoustic data is non-directional, it allows for early fault prediction, while the vibration data is directional, enabling fault direction diagnosis. The integrated acoustic and vibration sensor collects sound and vibration signals from the same point, leading to more accurate fault diagnosis. For example, the installation locations on the hydrocyclone are primarily the conical and cylindrical sections. Each hydrocyclone typically has three acoustic and vibration sensors 1 installed: two at a 90° angle to the same cylinder and one along the cylinder's axis.

[0084] The acoustic vibration data acquisition edge unit 2 can perform preliminary cleaning of the acquired sensor data and transmit the valid data to the acoustic vibration analysis system server 3 via wired or wireless transmission. In some embodiments, the acoustic vibration data acquisition edge unit 2 may only have data transmission function, and the preprocessing function (preliminary cleaning) is performed in the acoustic vibration analysis system server 3.

[0085] The acoustic and vibration analysis system server 3 stores and processes the collected acoustic and vibration data, analyzes the spectral characteristics of the hydrocyclone as reference data for fault diagnosis, and allows both pre- and post-processed acoustic and vibration data to communicate in real time with the data storage server 5 via an open API interface. The acoustic and vibration analysis system server 3 needs to perform actual calibration on the collected acoustic and vibration data and the hydrocyclone's status, especially calibrating data under specific fault conditions of the hydrocyclone, including different degrees of slag outlet blockage and different particle sizes of overflow.

[0086] The process data acquisition unit 4 can collect on-site process data through the DCS system's open OPC protocol, or collect sensor data through various sensors and edge terminals, and send the collected process data to the data storage server 5.

[0087] The main function of data storage server 5 is to store acoustic and vibration data, pre-processed acoustic and vibration data, and operational process data, and to display the monitored acoustic and vibration data and operational process data in real time on the intelligent platform display interface 7. Simultaneously, by establishing an intelligent platform for the hydrocyclone, encompassing underlying data acquisition, storage, and management; data storage and database management; equipment lifecycle management; production management; safety management; and energy management, comprehensive management of the hydrocyclone is achieved. Ultimately, this results in the formation of a database of acoustic and vibration monitoring parameters, a database of fault attribution, and a database of operational process parameters for the hydrocyclone.

[0088] The hydrocyclone detection device also includes: a deep learning and reinforcement learning analyzer 6 and an intelligent platform display interface 7;

[0089] The deep learning and reinforcement learning analyzer 6 and the intelligent platform display interface 7 are respectively connected to the data storage server 5.

[0090] The deep learning and reinforcement learning analyzer 6 can read preprocessed acoustic and vibration data and operational process data from the data storage server 5 in real time. Through deep learning algorithms, especially a semi-supervised learning method combining supervised learning based on few samples and unsupervised learning based on classification samples, it performs fault trend analysis on the hydrocyclone, achieving full-cycle management of the hydrocyclone. Based on big data analysis using reinforcement learning, it optimizes operational process parameters, realizing production and energy management of the hydrocyclone, and returns the calculation results to the data storage server 5.

[0091] The intelligent platform display interface 7 can display the pre-processed acoustic and vibration data and operating process data in real time in the intelligent hydrocyclone platform, and can also display the overall functions of the intelligent platform.

[0092] The hydrocyclone detection method and device proposed in this invention realize the monitoring of the equipment operation status and optimization of the operating process parameters of hydrocyclones in mineral processing plants, and are displayed and managed through an intelligent management platform;

[0093] (1) The sound and vibration parameters of the hydrocyclone are collected by a contact-type integrated acoustic and vibration sensor, which realizes the early diagnosis and attribution of hydrocyclone faults by acoustic signature and the directional diagnosis and attribution of vibration. Compared with the method of attribution by vibration signal alone, the fault can be predicted earlier.

[0094] (2) Based on the collected acoustic and vibration data, a semi-supervised learning method combining supervised learning based on few samples and unsupervised learning results based on classification samples is used to predict the operation faults of the hydrocyclone, form a fault attribution database, realize predictive maintenance of equipment operation status, and solve the problem of rapid wear and short service life of the hydrocyclone.

[0095] (3) Conduct big data analysis based on reinforcement learning on the on-site operating process parameters and acoustic and vibration monitoring parameters to optimize the operating process parameters of the production process, and finally realize the intelligent management and control of the hydrocyclone equipment, so as to better solve the problem of the difficulty in controlling the operating process parameters and discharge particle size of the hydrocyclone.

[0096] (4) An intelligent management platform for hydrocyclones has been implemented, including underlying data acquisition and storage management, data storage and database management, equipment lifecycle management, production management, safety management, and energy management, achieving comprehensive management of hydrocyclones. This management platform can be extended to the intelligent management of other equipment or processes.

[0097] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting hydrocyclones, characterized in that, include: Establish an acoustic vibration monitoring parameter database and a fault attribution database based on the preprocessed acoustic vibration data; The hydrocyclone is monitored based on the raw acoustic and vibration data and the preprocessed acoustic and vibration data; Collect operational process data; Establish an operating process parameter database based on the aforementioned operating process data; The operating process parameters of the hydrocyclone are optimized based on the aforementioned operating process parameter database. The optimization of the hydrocyclone's operating process parameters based on the operating process parameter database includes the following steps: Multivariate nonlinear regression analysis was performed on the operating process data and the pre-processed acoustic and vibration data; To obtain the functional relationship between production efficiency, energy consumption, and operating process parameters; Optimize real-time process parameters based on function correlation.

2. The hydrocyclone detection method according to claim 1, characterized in that, The establishment of the acoustic vibration monitoring parameter database and fault attribution database based on the preprocessed acoustic vibration data includes the following steps: Collect raw acoustic and vibration data; Preprocess the raw acoustic and vibration data; The preprocessed acoustic and vibration data were analyzed to determine the spectral characteristic values ​​of the hydrocyclone. Calibrate the range of spectral characteristic values ​​of the hydrocyclone under different operating conditions; Based on the original acoustic and vibration data, the preprocessed acoustic and vibration data, and the corresponding spectral characteristic values ​​of the hydrocyclone under different operating conditions, an acoustic and vibration monitoring parameter database and a fault attribution database are established.

3. The hydrocyclone detection method according to claim 2, characterized in that, The spectral characteristic values ​​are one or more of amplitude, frequency, phase, velocity, acceleration, rate of change of amplitude, rate of change of frequency, rate of change of phase, rate of change of velocity, and rate of change of acceleration.

4. The hydrocyclone detection method according to claim 1, characterized in that, The specific steps for monitoring the hydrocyclone based on the raw acoustic and vibration data and the preprocessed acoustic and vibration data are as follows: Based on the original acoustic and vibration data and the preprocessed acoustic and vibration data, a semi-supervised learning method combining supervised learning based on few samples and unsupervised learning results based on classification samples is used to analyze the failure trend of hydrocyclones.

5. The hydrocyclone detection method according to claim 1, characterized in that, The process parameter database established based on the process data includes the following steps: The operating process data of the hydrocyclone under different operating conditions were calibrated. A database of operating process parameters is established based on the operating process data and the different operating conditions of the corresponding hydrocyclones.

6. The hydrocyclone detection method according to claim 1, characterized in that, After optimizing the operating process parameters of the hydrocyclone based on the operating process parameter database, the process further includes: An intelligent management platform for hydrocyclones will be established based on the acoustic and vibration monitoring parameter database, the fault attribution database, and the operating process parameter database. The intelligent management platform will be showcased.

7. The hydrocyclone detection method according to claim 6, characterized in that, The intelligent management platform includes: a real-time display interface for hydrocyclone operation, underlying data acquisition and storage management, data storage and database management, equipment lifecycle management, production management, safety management, and energy management.

8. A hydrocyclone detection device based on the hydrocyclone detection method according to any one of claims 1-7, characterized in that, include: Acoustic vibration sensor (1), acoustic vibration data acquisition edge unit (2), acoustic vibration analysis system server (3), process data acquisition unit (4), and data storage server (5); The acoustic vibration sensor (1) is mounted on the hydrocyclone and connected to the acoustic vibration data acquisition edge unit (2); The acoustic vibration data acquisition edge unit (2) is connected to the acoustic vibration analysis system server (3); The acoustic vibration analysis system server (3) is connected to the data storage server (5); The process data acquisition unit (4) is connected to the data storage server (5).

9. The hydrocyclone detection device according to claim 8, characterized in that, It also includes: a deep learning and reinforcement learning analyzer (6) and an intelligent platform display interface (7); The deep learning and reinforcement learning analyzer (6) and the intelligent platform display interface (7) are respectively connected to the data storage server (5).

Images