Wind pressure and wind speed intelligent adjusting system and method for winnowing impurity removing machine

Abstract

The invention discloses a wind pressure and wind speed intelligent adjusting system and method for a winnowing impurity removing machine, relates to the technical field of industrial control systems, and can solve the problem of how to realize self-adaptive intelligent adjustment of wind speed and wind pressure in the winnowing impurity removing process so as to improve the separation precision and the equipment operation stability. Comprising the steps that real-time working condition data and historical wind speed data in the winnowing process are obtained; wherein the real-time working condition data comprises material characteristic parameters at the feeding port, the feeding rate and material image data after winnowing; determining an initial wind speed set value of the fan according to the characteristic parameters of the material at the feed port and the feed rate; the real-time impurity content is determined according to the material image data obtained after winnowing, and the current air volume demand degree is determined according to the real-time impurity content; determining a wind speed fluctuation characteristic value according to the historical wind speed data; and according to the wind speed fluctuation characteristic value, the current air volume demand degree and the initial wind speed set value, the target regulation wind speed of the draught fan is determined and executed.

Description

Technical Field

[0001] This invention relates to the field of industrial control system technology, specifically to an intelligent adjustment system and method for air pressure and air speed in an air separator for removing impurities. Background Technology

[0002] In the pretreatment process of Chinese medicinal materials, the air classifier is a key piece of equipment. It generates a controllable airflow through a fan, achieving efficient and environmentally friendly physical separation based on the differences in aerodynamic characteristics between materials and impurities. The core of this technology lies in adjusting and maintaining suitable airflow parameters (mainly air pressure and air velocity) for different types of medicinal materials and impurities with varying densities, ensuring that lighter impurities are effectively blown away while qualified medicinal materials are collected.

[0003] Currently, the air separation control methods commonly used in the industry mostly rely on manual experience settings or simple feedback mechanisms based on fixed parameters. However, in actual production processes, the type of material, moisture content, feed flow rate, and environmental conditions are constantly changing. This relatively static control strategy is difficult to adapt to complex operating condition fluctuations in real time, often leading to incomplete separation or loss of useful materials. The overall control effect and operational economy need to be improved. Summary of the Invention

[0004] To address the current technical challenge of achieving adaptive intelligent adjustment of wind speed and wind pressure during air separation and impurity removal to improve separation accuracy and equipment operational stability, this invention aims to provide an intelligent adjustment system and method for wind pressure and wind speed in an air separation and impurity removal machine. The specific technical solution adopted is as follows: In a first aspect, the present invention provides a method for intelligent adjustment of air pressure and air speed for an air classifier, comprising: acquiring real-time operating condition data and historical air speed data during the air classification process; wherein, the real-time operating condition data includes material characteristic parameters at the feed inlet, feeding rate, and image data of the material after air classification, and the historical air speed data is the air speed data within a preset time period prior to the current moment; determining the initial air speed setpoint of the blower based on the material characteristic parameters at the feed inlet and the feeding rate; determining the real-time impurity content based on the image data of the material after air classification, and determining the current air volume demand based on the real-time impurity content; determining the air speed fluctuation characteristic value based on the historical air speed data; and determining and executing the target adjustment air speed of the blower based on the air speed fluctuation characteristic value, the current air volume demand, and the initial air speed setpoint.

[0005] Secondly, the present invention provides an intelligent adjustment system for air pressure and air speed of an air classifier for removing impurities, comprising: a data acquisition module, an initial air speed calculation module, an air volume demand analysis module, a fluctuation characteristic analysis module, and an adjustment determination and execution module; the data acquisition module is used to acquire real-time operating condition data and historical air speed data during the air classification process; wherein, the real-time operating condition data includes material characteristic parameters at the feed inlet, the feeding rate, and image data of the material after air classification, and the historical air speed data is the air speed data within a preset time period before the current moment; the initial air speed calculation module is used to determine the initial air speed setpoint of the fan based on the material characteristic parameters at the feed inlet and the feeding rate; the air volume demand analysis module is used to determine the real-time impurity content based on the image data of the material after air classification, and determine the current air volume demand based on the real-time impurity content; the fluctuation characteristic analysis module is used to determine the air speed fluctuation characteristic value based on the historical air speed data; and the adjustment determination and execution module is used to determine and execute the target adjustment air speed of the fan based on the air speed fluctuation characteristic value, the current air volume demand, and the initial air speed setpoint.

[0006] Thirdly, the present invention provides an electronic device, comprising: a processor and a memory; wherein the memory is used to store one or more programs, the one or more programs including computer-executable instructions, and when the electronic device is running, the processor executes the computer-executable instructions stored in the memory to cause the electronic device to perform the intelligent adjustment method for wind pressure and wind speed of an air separator as described in the first aspect and any possible implementation thereof.

[0007] This invention offers the following advantages: First, by comprehensively considering real-time feeding conditions and historical optimization knowledge to determine the initial wind speed, a reasonable starting point for control is provided, overcoming the poor adaptability of traditional fixed parameter settings. Second, by introducing real-time impurity content analysis based on machine vision, the sorting effect is directly quantified into an airflow demand signal, achieving effect feedback control with the final quality as the target, thus improving separation accuracy. Third, by analyzing the fluctuation characteristics of historical wind speed data, the amplitude of each adjustment is dynamically constrained, effectively ensuring the stability of equipment operation while pursuing control effects. Finally, by integrating the above three factors to calculate the target adjustment wind speed, the control system can simultaneously respond to material changes, track the sorting effect, and maintain system stability, thereby continuously maintaining efficient, low-consumption, and reliable air-separation impurity removal performance under varying production conditions. Attached Figure Description

[0008] To more clearly illustrate the technical solutions and advantages 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 only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 This is a schematic diagram of the architecture of an intelligent adjustment system for air pressure and air speed in an air separator provided in one embodiment of the present invention; Figure 2 This is a flowchart illustrating an intelligent adjustment method for air pressure and air speed in an air separator for removing impurities, provided as an embodiment of the present invention. Detailed Implementation

[0010] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the specific implementation methods, structures, features, and effects of the present invention will be described in detail below with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0011] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0012] In all division and logarithmic operations involved in this invention, a smoothing mechanism is employed to prevent computer program crashes or invalid values ​​from being generated due to a zero denominator or zero input. Specifically, a correction factor ε, which is a very small positive number, is superimposed on the denominator term of the division operation or the argument term of the logarithmic function, for example, a value of 10 to the power of negative 5, thereby ensuring the robustness and feasibility of the algorithm under extreme conditions.

[0013] Unless otherwise specified, the normalization function Norm() mentioned in this invention uses maximum and minimum value normalization. The maximum and minimum values ​​are preset empirical extreme values ​​derived from a large amount of historical experimental data. If the calculation result exceeds the [0, 1] interval, it is restricted to the [0, 1] range by a truncation function (i.e., if the result is less than 0, it is taken as 0; if it is greater than 1, it is taken as 1) to eliminate the influence of outliers on the evaluation index.

[0014] The following description, in conjunction with the accompanying drawings, details a specific scheme for an intelligent adjustment system and method for air pressure and air speed in an air separator provided by the present invention.

[0015] For example, such as Figure 1 The diagram shown is an architectural schematic of an intelligent adjustment system for wind pressure and wind speed (hereinafter referred to as the intelligent adjustment system) for an air classifier and impurity remover, according to an embodiment of the present invention. The intelligent adjustment system 10 includes: a data acquisition module 11, an initial wind speed calculation module 12, an air volume demand analysis module 13, a fluctuation characteristic analysis module 14, and an adjustment determination and execution module 15. The modules are described below in sequence: (1) Data acquisition module 11.

[0016] The data acquisition module 11 is responsible for synchronously collecting multi-source real-time data and historical operating data related to the air separation process from sensors and monitoring devices deployed at key locations of the air separator, providing a unified input source for subsequent intelligent analysis and decision-making.

[0017] Optionally, the data acquisition module 11 is used to acquire real-time operating condition data and historical wind speed data during the air separation process. The real-time operating condition data includes material characteristic parameters at the feed inlet, the feed rate, and image data of the material after air separation; the historical wind speed data is the wind speed data for a preset period prior to the current moment.

[0018] Specifically, the data acquisition module 11 collects material characteristic parameters (such as average particle size) and feed rate in real time through a feed monitoring unit (e.g., a 3D volume scanner based on laser contour scanning). Simultaneously, the data acquisition module 11 collects image data of the material during its descent after air separation through a result monitoring unit (e.g., a high-speed industrial camera equipped with a high-brightness light source).

[0019] In addition, the data acquisition module 11 continuously reads and caches wind speed data from the wind duct monitoring unit (e.g., differential pressure sensor or anemometer) to provide historical wind speed data for the required time period.

[0020] Then, the data acquisition module 11 aligns and encapsulates the above data according to the timestamp, and then sends it to the initial wind speed calculation module 12, the air volume demand analysis module 13 and the fluctuation characteristic analysis module 14 respectively.

[0021] (2) Initial wind speed calculation module 12.

[0022] The initial wind speed calculation module 12 is responsible for receiving material characteristic parameters and feed rate from the data acquisition module 11, and performing initial wind speed setting calculation. This module determines an initial wind speed setting value that matches the current operating conditions by combining a preset mathematical model and historical optimization knowledge, providing a benchmark for the entire adjustment process.

[0023] Optionally, the initial wind speed calculation module 12 is used to determine the initial wind speed setting value of the blower based on the material characteristic parameters and feeding rate at the feed inlet.

[0024] Specifically, the initial wind speed calculation module 12 first calculates the first wind speed reference value by calling its internally stored first correlation model based on the received material characteristic parameters and feed rate. Then, the initial wind speed calculation module 12 retrieves the target vector representing the operating condition characteristics associated with the optimal separation effect by matching the current operating condition with a pre-generated target operating condition feature vector library. Finally, the initial wind speed calculation module 12 applies the correction rules defined by the target operating condition feature vector to optimize and adjust the first wind speed reference value, outputting the final determined initial wind speed setpoint. This value will be simultaneously sent to the airflow demand analysis module 13 and the adjustment determination execution module 15.

[0025] (3) Air volume demand analysis module 13.

[0026] The airflow demand analysis module 13 is responsible for receiving image data of the material after air separation from the data acquisition module 11, and performing intelligent analysis on the images to evaluate the real-time sorting effect. This module dynamically determines the current required airflow adjustment by calculating the impurity content and combining the adjustment characteristics of the model used by the initial wind speed calculation module 12.

[0027] Optionally, the air volume demand analysis module 13 is used to determine the real-time impurity content based on the material image data after air separation, and to determine the current air volume demand based on the real-time impurity content.

[0028] For example, the air volume demand analysis module 13 can be divided into a visual analysis submodule 131 and a demand calculation submodule 132 to respectively complete impurity identification and demand quantification, which will be described below: (3.1) Visual Analysis Submodule 131.

[0029] Optionally, the visual analysis submodule 131 is used to input material image data into a preset image semantic segmentation model to obtain image classification results; and to calculate the real-time impurity content based on the image classification results.

[0030] Specifically, the visual analysis submodule 131 deploys a pre-trained, deep learning-based image semantic segmentation model. First, the visual analysis submodule 131 inputs the received high-speed image data into the model. Then, the model performs pixel-level classification of the image, outputting a classification result image indicating whether each pixel belongs to "qualified material" or "impurity." Finally, the visual analysis submodule 131 calculates the ratio of the total number of pixels identified as "impurities" to the total number of pixels identified as "qualified materials" in the classification result image, and sends this ratio as the real-time impurity content to the demand calculation submodule 132.

[0031] (3.2) Demand calculation submodule 132.

[0032] Optionally, the demand calculation submodule 132 is used to obtain the model adjustment parameters of the first associated model; and determine the current air volume demand based on the real-time impurity content and the model adjustment parameters. The model adjustment parameters characterize the adjustment weight of the initial air velocity setpoint as a function of material characteristic parameters and feed rate.

[0033] Specifically, the demand calculation submodule 132 first retrieves the model adjustment parameters of the first associated model from the initial wind speed calculation module 12. Then, the demand calculation submodule 132 calculates the real-time impurity content provided by the visual analysis submodule 131 using these model adjustment parameters to quantify the required airflow adjustment intensity to improve the current sorting effect, i.e., the current airflow demand. Finally, the demand calculation submodule 132 outputs the calculated current airflow demand to the adjustment determination execution module 15.

[0034] (4) Fluctuation characteristic analysis module 14.

[0035] The fluctuation characteristic analysis module 14 is responsible for receiving historical wind speed data sequences from the data acquisition module 11 and analyzing the statistical characteristics of the sequences to assess the stability of the wind turbine's recent operation and provide a constraint reference for the adjustment range.

[0036] Optionally, the fluctuation characteristic analysis module 14 is used to determine the wind speed fluctuation characteristic value based on historical wind speed data.

[0037] Specifically, the fluctuation characteristic analysis module 14 first acquires a series of historical wind speed data within a preset time period (e.g., the first 10 seconds). Then, the module calculates the statistical parameters of this data sequence, such as its variance (reflecting the degree of data dispersion) and range (reflecting the range of data variation). Finally, the fluctuation characteristic analysis module 14 performs comprehensive and normalization processing on these statistical parameters (e.g., calculating the product of variance and range and normalizing it), generating an index to quantify the strength of wind speed fluctuations, namely the wind speed fluctuation characteristic value, and sends this characteristic value to the adjustment determination execution module 15.

[0038] (5) Adjust and determine the execution module 15.

[0039] The adjustment determination execution module 15, as the system's decision and output terminal, is responsible for integrating the initial wind speed setpoint from the initial wind speed calculation module 12, the current air volume demand from the air volume demand analysis module 13, and the wind speed fluctuation characteristic value from the fluctuation characteristic analysis module 14, calculating the final target adjustment wind speed to be executed, and driving the execution mechanism to complete the adjustment.

[0040] Optionally, the adjustment determination execution module 15 is used to determine and execute the target adjustment wind speed of the fan based on the wind speed fluctuation characteristic value, the current air volume demand and the initial wind speed setting value.

[0041] For example, the adjustment determination execution module 15 can be divided into a wind speed fusion calculation submodule 151 and a control output submodule 152, to respectively complete the target value calculation and control command issuance, which will be described below: (5.1) Wind speed fusion calculation submodule 151.

[0042] Optionally, the wind speed fusion calculation submodule 151 is used to determine the target adjustment wind speed based on the wind speed fluctuation characteristic value, the current air volume demand and the initial wind speed setting value.

[0043] Specifically, the wind speed fusion calculation submodule 151 first receives wind speed fluctuation characteristic values ​​from the fluctuation characteristic analysis module 14, the current air volume demand from the air volume demand analysis module 13, and the initial wind speed setpoint from the initial wind speed calculation module 12. Then, the wind speed fusion calculation submodule 151 performs calculations according to a predetermined fusion algorithm. Finally, the wind speed fusion calculation submodule 151 sends the calculated target regulating wind speed value to the control output submodule 152.

[0044] (5.2) Control output submodule 152.

[0045] Optionally, the control output submodule 152 is used to perform target adjustment of the fan speed.

[0046] Specifically, after receiving the target wind speed value from the wind speed fusion calculation submodule 151, the control output submodule 152 first converts the target value into a corresponding control command (e.g., the target frequency signal corresponding to the wind turbine inverter). Then, the control output submodule 152 sends this control command to the wind turbine drive unit in real time via an industrial communication interface (such as an analog output module or fieldbus). Finally, the wind turbine drive unit adjusts its operating state according to the command, causing the actual wind speed to approach and stabilize at the target wind speed value, thus completing one closed-loop control action.

[0047] The above describes the intelligent air pressure and air speed regulation system 10 used in the air separator and its included modules.

[0048] For example, such as Figure 2 The diagram shown is a flowchart illustrating an intelligent adjustment method for air pressure and air speed in an air separator according to an embodiment of the present invention, comprising the following steps: S201. Acquire real-time operating condition data and historical wind speed data during the air separation process. The real-time operating condition data includes material characteristic parameters at the feed inlet, feed rate, and image data of the material after air separation. The historical wind speed data is the wind speed data within a preset time period prior to the current moment.

[0049] For example, this step can be performed by the data acquisition module 11 in the intelligent control system 10 described above.

[0050] Specifically, the data acquisition module 11 collects material characteristic parameters and feeding rate at the feed inlet in real time through its connected feed monitoring unit. The feed monitoring unit may include a vision sensor, such as a 3D volume scanner, for contour scanning of the material flow. By combining laser scanning with belt speed measurement, it calculates the average particle size of the material (as an example of material characteristic parameters) and the volume or mass per unit time (as the feeding rate) in real time.

[0051] Simultaneously, the data acquisition module 11 collects image data of the material after air separation through its connected result monitoring unit. The result monitoring unit may include a high-speed industrial camera and a high-brightness light source, installed at the observation window of the heavy material outlet or the light impurity outlet, for continuously capturing clear images of the falling material. Furthermore, the data acquisition module 11 continuously reads wind speed readings from the air duct monitoring unit (such as an anemometer) and caches data from the most recent period (e.g., 10 seconds) to form the historical wind speed data sequence.

[0052] Therefore, the data acquisition module 11 provides a complete data foundation for subsequent analysis modules, including real-time operating status and recent operating history.

[0053] S202. Determine the initial wind speed setting value of the blower based on the material characteristic parameters and feeding rate at the feed inlet.

[0054] For example, this step can be performed by the initial wind speed calculation module 12 in the intelligent adjustment system 10 described above. Specifically, the initial wind speed calculation module 12 receives material characteristic parameters and feeding rate from the data acquisition module 11. First, the module calls its internally stored first association model (e.g., a mathematical model that associates wind speed with material characteristics and feeding rate), takes the current material characteristic parameters and feeding rate as input, and calculates a preliminary wind speed value, i.e., a first wind speed reference value. Then, based on the current operating condition information (which may include material characteristics, feeding rate, etc.), the initial wind speed calculation module 12 performs a matching search from a pre-generated target operating condition feature vector library to obtain a feature vector that best represents the optimal separation effect under the current operating condition, i.e., a target operating condition feature vector. Finally, the initial wind speed calculation module 12 applies the correction rules or coefficients contained in the target operating condition feature vector to optimize and adjust the first wind speed reference value, and outputs a wind speed value with better adaptability to the current operating condition as the initial wind speed setpoint for the fan. It should be noted that the specific procedures for the aforementioned sub-steps can be found in S301-S303 below, and will not be repeated here.

[0055] In another possible implementation, the initial wind speed calculation module 12 can pre-establish a lookup table or simple functional relationship between material characteristic parameters, feed rate, and recommended wind speed when determining the initial wind speed setpoint. The initial wind speed setpoint can then be directly obtained by looking up the table or calculating based on real-time collected operating data, without going through the two-stage process of "reference value calculation - feature vector correction". This method is suitable for scenarios with relatively fixed material types and limited range of operating condition variations, and has the advantages of simple implementation and fast response.

[0056] Alternatively, the initial wind speed calculation module 12 can also incorporate data from environmental temperature and humidity sensors for auxiliary correction. This method, after calculating the initial wind speed setpoint, further fine-tunes the setpoint according to a preset correction coefficient based on real-time collected environmental temperature and humidity data to compensate for the impact of air density changes on the wind separation effect.

[0057] Therefore, the initial wind speed calculation module 12, by integrating real-time operating conditions and historical optimization knowledge, sets a reasonable initial operating benchmark for the wind turbine, namely the initial wind speed setpoint. This value is synchronously sent to subsequent modules as the starting point and benchmark for feedback adjustment.

[0058] S203. Determine the real-time impurity content based on the material image data after air separation, and determine the current air volume requirement based on the real-time impurity content.

[0059] For example, this step can be performed by the airflow demand analysis module 13 in the intelligent adjustment system 10 described above. Specifically, the airflow demand analysis module 13 inputs image data into a preset deep learning-based image semantic segmentation model. This model performs pixel-level semantic segmentation on the image, automatically distinguishing between qualified material pixels and impurity pixels in the image, and outputs the classification results. Then, the module counts the number of impurity pixels and qualified material pixels according to the classification results, and calculates their ratio to obtain the real-time impurity content, which directly quantifies the sorting purity of the current batch of materials. Next, the module obtains the key parameter in the first association model used by the initial wind speed calculation module 12—the model adjustment parameter. This parameter characterizes the sensitivity or adjustment weight of the initial wind speed setpoint to changes in material characteristics and feed rate. Finally, the module determines the current airflow demand based on the proportional relationship between the real-time impurity content and this model adjustment parameter (e.g., calculating the ratio of the real-time impurity content to the model adjustment parameter). This demand is a dimensionless value, the magnitude of which directly reflects the intensity and direction of the airflow adjustment required to improve the current sorting effect (i.e., reduce the impurity content). It should be noted that the specific process of the aforementioned sub-steps can be found in S401-S404 below, and will not be repeated here.

[0060] In another possible implementation, when determining the real-time impurity content, the air volume demand analysis module 13 can also perform preprocessing such as grayscale conversion and filtering and noise reduction on the material image, and then use the difference between impurities and materials in grayscale, texture or shape to set a threshold for binarization segmentation, and finally calculate the proportion of impurities by statistically analyzing the area of ​​connected regions.

[0061] In another possible implementation, the airflow demand analysis module 13 can also introduce historical average impurity content as a reference benchmark when determining the current airflow demand, that is, it can also calculate the moving average of impurity content over a recent period. The calculation of airflow demand is adjusted by comparing the deviation between the real-time value and the average value. For example, when the real-time value is significantly higher than the average value, the calculated demand value is increased, and vice versa, thereby making the adjustment smoother and avoiding over-response to short-term fluctuations.

[0062] Therefore, the airflow demand analysis module 13 transforms the visually perceived sorting effect into a quantifiable adjustment demand signal, namely the current airflow demand. This signal accurately indicates the amount of airflow adjustment required to achieve a better sorting effect.

[0063] S204. Determine the characteristic value of wind speed fluctuation based on historical wind speed data.

[0064] For example, this step can be performed by the fluctuation characteristic analysis module 14 in the intelligent regulation system 10 described above, and specifically includes the following steps: (1) Determine the statistical parameters of the historical wind speed data.

[0065] First, the fluctuation characteristic analysis module 14 acquires a sequence of historical wind speed data from the data acquisition module 11 within a preset time period (e.g., 10 seconds) prior to the current moment. Then, the module calculates key statistical parameters of this data sequence to describe its fluctuation characteristics. For example, these statistical parameters include variance and range. Variance measures the dispersion of wind speed data around its average value, reflecting the intensity of wind speed fluctuations; range (the difference between the maximum and minimum values) measures the total range of wind speed variation within that time period.

[0066] (2) Normalize the statistical parameters and use the result as the wind speed fluctuation characteristic value.

[0067] Furthermore, the fluctuation characteristic analysis module 14 performs comprehensive and normalization processing on the calculated statistical parameters. One specific implementation method is to calculate the product of variance and range to comprehensively reflect the intensity and range of wind speed fluctuations. Then, this product value is normalized (e.g., mapped to the 0-1 range), and the result is used as the final wind speed fluctuation characteristic value. A larger characteristic value indicates stronger recent wind speed fluctuations and a less stable system operation; conversely, a smaller value indicates more stable wind speeds.

[0068] S205. Based on the wind speed fluctuation characteristic value, the current air volume demand and the initial wind speed setting value, determine and execute the target wind speed adjustment for the fan.

[0069] For example, this step can be performed by the adjustment determination execution module 15 in the intelligent adjustment system 10 described above. Specifically, the wind speed fusion calculation submodule 151 of the adjustment determination execution module 15 receives the wind speed fluctuation characteristic value from the fluctuation characteristic analysis module 14, the current air volume demand from the air volume demand analysis module 13, and the initial wind speed setpoint from the initial wind speed calculation module 12. The calculation formula is as follows:

[0070] in, This indicates the target wind speed adjustment for the fan. This represents the characteristic value of wind speed fluctuation. This indicates the current air volume demand. This indicates the initial wind speed setting.

[0071] It should be noted that the initial wind speed setting value in the above formula is... It is the basis for regulation. Current air volume demand. As a multiplicative factor for adjusting the intensity: when the sorting effect is poor (high real-time impurity content), >1, which serves to amplify and adjust; when the sorting effect is good, 1. The regulating effect weakens or even reverses, resulting in minor adjustments. Wind speed fluctuation characteristics. Then it serves as a constraint factor for the adjustment range: when recent wind speed fluctuations are large ( When the wind speed is high, the allowable adjustment range is increased accordingly to quickly respond to changes in operating conditions; when the wind speed is stable ( When the value is small, the adjustment range is limited to avoid introducing unnecessary oscillations and ensure stable system operation.

[0072] Finally, the control output submodule 152 of the adjustment and determination execution module 15 will calculate the target adjustment wind speed value. It is converted into specific control commands (such as inverter frequency signals) and sent to the fan drive unit for execution, thereby completing a closed-loop intelligent regulation.

[0073] Based on the above technical solutions, this invention determines the initial wind speed by integrating real-time feeding conditions and historical optimization knowledge, providing a reasonable starting point for control and overcoming the poor adaptability of traditional fixed parameter settings. Secondly, it introduces real-time impurity content analysis based on machine vision, directly quantifying the sorting effect into an airflow demand signal, achieving effect feedback control with the final quality as the target, and improving separation accuracy. Thirdly, by analyzing the fluctuation characteristics of historical wind speed data, it dynamically constrains the amplitude of each adjustment, effectively ensuring the stability of equipment operation while pursuing control effects. Finally, by integrating the above three factors to calculate the target adjustment wind speed, the control system can simultaneously respond to material changes, track the sorting effect, and maintain system stability, thereby continuously maintaining efficient, low-consumption, and reliable air-separation impurity removal performance under varying production conditions.

[0074] For example, in another embodiment of the present invention, a method for intelligent adjustment of air pressure and air speed for an air classifier is provided. The initial air speed setting value of the blower is determined based on the material characteristic parameters at the feed inlet and the feed rate. This specifically includes the following steps: S301. Based on the material characteristic parameters and the feeding rate, determine the first correlation model and calculate the first wind speed reference value.

[0075] In this step, the initial wind speed calculation module 12 calls its internally stored first correlation model. This model defines the mathematical relationship between wind speed and material characteristic parameters and feed rate.

[0076] For example, the first correlation model can be a linear model. Specifically, it uses the material characteristic parameters collected at the current moment. (e.g., average particle size of the material) and feed rate By inputting this model, the first wind speed reference value can be calculated. The calculation formula is:

[0077] in, This represents the calculated first wind speed reference value at time t (unit: m / s). This represents the material characteristic parameters at time t, taking the average particle size as an example; This represents the feed rate at time t; This represents one of the pre-calibrated model adjustment parameters, used to characterize the adjustment weight of the wind speed setpoint as it changes with material characteristic parameters, and... The unit adjustment is with Same, that is, m / s; Similarly, the second pre-calibrated model adjustment parameter is used to characterize the adjustment weight of the wind speed setpoint as the feed rate changes, and... The unit adjustment is with The same, that is, m / s.

[0078] It should be pointed out that, , The model adjustment parameters, pre-calibrated using linear regression, are determined by fitting and analyzing the material characteristic parameter sequences, feed rate sequences, and corresponding historical wind speed setpoint sequences that have been verified as suitable in practice from historical operating data. The least squares method is then used to find the parameters that minimize the overall error between the model output and the historical setpoints. , .parameter The magnitude of the value directly determines the sensitivity of the initial wind speed setpoint to changes in material characteristic parameters. The larger the value, the greater the adjustment range of the wind speed setpoint as the material characteristics change, that is, the higher the adjustment weight of the model in response to changes in material characteristics. Similarly, I will not elaborate further.

[0079] It can be understood that the wind speed reference value in the above formula consists of two parts, one of which is linearly proportional to the particle size of the material (from...). (Adjustment), the other part is linearly proportional to the feed rate (by...) The adjustment reflects the basic air volume required to process different material quantities and characteristics.

[0080] S302. Determine the characteristic vector of the target working condition.

[0081] For example, the initial wind speed calculation module 12 determines the target operating condition feature vector, specifically including the following steps:

[0082] (1) Obtain historical operating condition data, corresponding historical wind speed data, and historical separation effect data under various historical operating conditions.

[0083] Specifically, the initial wind speed calculation module 12 retrieves pre-recorded data from multiple wind separation processes from the historical database. This data includes: historical operating condition data, such as characteristic parameters (particle size, moisture content, etc.) and feed rate of different batches of materials; corresponding historical wind speed data, i.e., the actual wind speed value under that operating condition; and historical separation effect data, i.e., the final separation result achieved under that operating condition and wind speed, such as the impurity content rate at the heavy material outlet and / or the material loss rate at the light impurity outlet.

[0084] (2) Extract features from the historical operating condition data and the corresponding historical wind speed data to obtain a feature vector set that reflects the relationship between operating conditions and wind speed.

[0085] Specifically, the initial wind speed calculation module 12 performs dimensionality reduction and feature extraction on the collected multidimensional historical data to capture the essential relationship between operating conditions and wind speed. One specific implementation method is to use principal component analysis (PCA) algorithm. This algorithm analyzes historical operating condition data (such as material characteristics and feed rates under different moisture contents and impurities) and their corresponding historical wind speed data to calculate the principal component directions that can best explain the data variance. Each principal component direction vector contains a magnitude. With the direction angle θ. By extracting the directions of the first few principal components, the original high-dimensional working condition-wind speed data pairs can be mapped to a low-dimensional feature space, forming a series of feature vectors that reflect their core correlation. The set of these vectors constitutes the feature vector set.

[0086] (3) Determine the target working condition feature vector based on the historical separation effect data and the feature vector set.

[0087] Specifically, the goal of the initial wind speed calculation module 12 is to find a feature vector that leads to the optimal separation effect. To this end, each feature vector obtained in the previous step (corresponding to a historical operating condition mode) is first associated with its corresponding historical separation effect data (such as the cleanup rate). For a vector in the feature vector set, it represents the wind speed achieved under the operating condition mode. There is a certain relationship between this and the separation effect. To quantify this relationship, an intermediate index is introduced. For example, the coefficient of variation under this operating condition is calculated. Its formula is:

[0088] in, This represents the variation coefficient under the a-th working condition; This indicates the historical wind speed data used under this operating condition; θ represents the principal component orientation angle (in radians) corresponding to this operating condition data in the principal component analysis. It should be noted that the wind speed values... It is related to the principal component orientation angle θ that characterizes the working condition. It is a normalization process for the direction angle, making It can comprehensively reflect the wind speed used in the characteristic direction of a specific working condition.

[0089] Furthermore, the degree of impurity removal under this operating condition is calculated:

[0090] in, This indicates the degree of impurity removal under the a-th working condition; This represents the actual impurity removal rate achieved under this operating condition, and is a known parameter. It can be understood that this is used to evaluate characteristics under specific operating conditions (through...). Under the manifestation of this, the separation effect achieved (by...) Efficiency (reflected) The value can be understood as the wind speed associated cost required per unit of impurity removal rate. The smaller the value, the more efficient the wind speed selection is usually under this operating condition.

[0091] Finally, the feature vectors under various historical operating conditions and their corresponding noise reduction levels were compared. The target matrix is ​​formed as follows:

[0092] In the matrix above, each row represents a historical process, for example... This represents the feature vector components after feature extraction for the a-th historical working condition. The corresponding level of impurity removal is determined. The initial wind speed calculation module 12 applies a preset multi-objective optimization algorithm, such as multi-objective particle swarm optimization, to this target matrix, aiming to simultaneously minimize the level of impurity removal and optimize other relevant indicators, searching within the feature vector set. This algorithm ultimately selects one or a set of Pareto optimal solutions, and the representative feature vector selected from these is determined as the target operating condition feature vector. This vector encapsulates the core operating condition feature information learned from historical data, which can guide efficient separation.

[0093] Therefore, the initial wind speed calculation module 12 analyzes historical data, extracts the correlation features between operating conditions and wind speed, and optimizes based on the separation effect, ultimately obtaining a knowledge-based target operating condition feature vector to guide real-time wind speed correction.

[0094] S303. Based on the target operating condition feature vector, the first wind speed reference value is corrected to obtain the initial wind speed setting value. The target operating condition feature vector represents the operating condition characteristics associated with the optimal separation effect.

[0095] It should be noted that before performing this step, the initial wind speed calculation module 12 needs to construct or have a target operating condition feature vector library built into it. The construction of this library is based on the method described in S302: by analyzing and optimizing various historical operating condition data, multiple sets of representative operating condition feature vectors that can lead to efficient separation are selected from the historical data. These selected feature vectors together constitute the target operating condition feature vector library, where each vector encapsulates a historically proven efficient operating condition mode and wind speed adjustment experience.

[0096] Furthermore, the initial wind speed calculation module 12 will use the first wind speed reference value calculated by S301. The correction is performed by combining the target operating condition feature vector determined in S302. Specifically, the module finds the best-matching target vector in the target operating condition feature vector library based on the currently collected real-time operating condition data (material characteristics, feed rate, etc.). The matching process can be based on metrics such as Euclidean distance and cosine similarity. After matching the target vector, the module parses its preset correction coefficient from it. The correction factor This target vector was calculated through optimization by comparing the optimal wind speed under historically efficient operating conditions it represents with the reference wind speed given by the first correlation model during its construction. Subsequently, the module calculates the initial wind speed setpoint using the following formula:

[0097] in, This indicates the final determined initial wind speed setting value; This represents the correction coefficient parsed from the matched target vector, and its value is greater than 0; This represents the first wind speed reference value. For example, if the target vector indicates that the current operating condition is similar to a certain efficient operating condition in the past, and the required wind speed under that condition is higher than the first reference value... If it is about 10% higher, then the corresponding correction factor is... It might be 1.1. Finally, the wind speed value after correction using this eigenvector. This means that the initial wind speed setting is more in line with the best historical experience, providing a better starting point for subsequent dynamic adjustments.

[0098] Based on the above technical solution, this embodiment of the invention establishes a first correlation model to quickly respond to real-time operating conditions and provide a wind speed benchmark. It then introduces a target operating condition feature vector learned from historical data to intelligently correct this benchmark. This method not only considers the current material state but also incorporates operating condition knowledge corresponding to historically optimal separation effects. This ensures that the determined initial wind speed setpoint possesses both real-time adaptability and the superiority of historical experience, laying a solid foundation for accurate and efficient control of the entire wind separation process.

[0099] For example, in another embodiment of the present invention, a method for intelligent adjustment of air pressure and air speed for an air classifier is provided, wherein the real-time impurity content is determined based on the image data of the material after air classification, and the current air volume requirement is determined based on the real-time impurity content, specifically including the following steps:

[0100] S401. Input the material image data into the preset image semantic segmentation model to obtain the image classification result.

[0101] Optionally, this step can be performed by the visual analysis submodule 131 within the airflow demand analysis module 13. Specifically, the visual analysis submodule 131 internally deploys a pre-trained image semantic segmentation model. This model receives real-time image data of the air-separated material from the result monitoring unit (high-speed industrial camera). For example, the image semantic segmentation model could be a deep learning-based image segmentation model.

[0102] Before inputting image data into the model, the visual analysis submodule 131 can perform necessary preprocessing, such as size standardization and brightness normalization, to ensure input consistency. Subsequently, the preprocessed image is fed into the model for forward inference. One specific implementation of the preset image semantic segmentation model is a deep learning model based on the U-Net architecture, which can efficiently perform pixel-level semantic segmentation. The model outputs a segmentation result image of the same size as the input image, where each pixel is classified into a specific category, such as "qualified medicinal material," "impurity," or "background," which is the image classification result.

[0103] S402. Calculate the real-time impurity content based on the image classification results.

[0104] Furthermore, after receiving the image classification results output by the model, the visual analysis submodule 131 analyzes the results to quantify the sorting effect. Specifically, the module counts the total number of pixels classified as "impurities" in the segmentation result image, and simultaneously counts the total number of pixels classified as "qualified medicinal materials." Then, the real-time impurity content is calculated using the following formula:

[0105] in, This represents the real-time impurity content at time j, and is a dimensionless ratio. This represents the total number of pixels identified as impurities in the image at time j. This represents the total number of pixels in the image at time j that are identified as qualified medicinal materials. It should be noted that this formula approximates the mass or volume ratio of impurities to materials in physical space by statistically analyzing the pixel ratio of impurities to qualified materials in the image, thereby directly and objectively quantifying the sorting purity of the current batch of materials. The higher the value, the more impurities are mixed in the material after sorting, and the worse the sorting effect.

[0106] S403. Obtain the model adjustment parameters of the first correlation model. The model adjustment parameters characterize the adjustment weight of the initial wind speed setpoint as the material characteristic parameters and feed rate change.

[0107] Optionally, this step can be performed by the demand calculation submodule 132 in the air volume demand analysis module 13. Specifically, the demand calculation submodule 132 obtains the model adjustment parameters from the initial wind speed calculation module 12 in the first associated model used to calculate the initial wind speed setpoint. The demand calculation submodule 132 reads this parameter value directly through the internal communication interface.

[0108] S404. Determine the current air volume demand based on the real-time impurity content and model adjustment parameters.

[0109] Furthermore, the demand calculation submodule 132 calculates the current air volume demand using the following formula:

[0110] in, The value represents the current air volume demand at time j, which is a dimensionless adjustment factor; k represents the model adjustment parameter for the first correlation model. The dimensionless model adjustment parameters are obtained from S403.

[0111] It should be noted that this formula represents the real-time impurity content, which indicates the degree of poor sorting performance. With parameters characterizing the adjustment characteristics of the current wind speed model Related. Impurity content The higher the value, the less ideal the current sorting effect, and theoretically, a larger airflow adjustment is needed to improve it. Therefore, the numerator term drives this process. Increase. The denominator is the model adjustment parameter, whose physical meaning lies in "normalization" or "calibration": if the current model's adjustment weights are already large (i.e., the model is very sensitive to changes in materials), then for the same impurity content, the required additional adjustment degree will increase. It should be relatively reduced, and vice versa. Therefore, It comprehensively reflects the intensity of airflow adjustment demand based on real-time sorting effect feedback and after calibration with the characteristics of the current model. A value greater than 1 usually indicates that the airflow needs to be increased. A value less than 1 indicates that the airflow may be reduced, while A value of 1 indicates that the current airflow is appropriate.

[0112] Based on the above technical solution, this invention utilizes machine vision technology to quantify the sorting effect (real-time impurity content) in real time and objectively, and combines this with the inherent adjustment characteristics (model adjustment parameters) of the wind speed control model to calculate the precise current airflow demand. This method directly transforms the final quality indicator (impurity content) into process control parameters (airflow demand), realizing effect-oriented closed-loop feedback regulation. This allows airflow regulation to not only respond to operating condition inputs but also directly track the sorting target, thereby significantly improving the accuracy and adaptability of control.

[0113] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0114] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for intelligent adjustment of air pressure and air velocity in an air separator for removing impurities, characterized in that, The method includes: Acquire real-time operating condition data and historical wind speed data during the air separation process; wherein, the real-time operating condition data includes material characteristic parameters at the feed inlet, feed rate, and material image data after air separation, and the historical wind speed data is the wind speed data within a preset time period before the current moment; The initial wind speed setting value of the blower is determined based on the material characteristic parameters at the feed inlet and the feed rate. The real-time impurity content is determined based on the image data of the material after air separation, and the current air volume requirement is determined based on the real-time impurity content. Based on the historical wind speed data, determine the wind speed fluctuation characteristic value; Based on the wind speed fluctuation characteristic value, the current air volume demand, and the initial wind speed setting value, the target wind speed adjustment for the fan is determined and executed.

2. The intelligent adjustment method for air pressure and air velocity in an air separator for removing impurities according to claim 1, characterized in that, Based on the material characteristic parameters at the feed inlet and the feed rate, the initial wind speed setpoint of the blower is determined, specifically including: Based on the material characteristic parameters and the feeding rate, a first correlation model is determined and a first wind speed reference value is calculated; Based on the target operating condition feature vector, the first wind speed reference value is corrected to obtain the initial wind speed setting value; wherein, the target operating condition feature vector represents the operating condition features associated with the optimal separation effect.

3. The intelligent adjustment method for air pressure and air velocity in an air separator for removing impurities according to claim 2, characterized in that, The method further includes: Acquire historical operating condition data, corresponding historical wind speed data, and historical separation effect data under various historical operating conditions; Feature extraction is performed on the historical operating condition data and the corresponding historical wind speed data to obtain a feature vector set reflecting the correlation between operating conditions and wind speed. Based on the historical separation effect data and the feature vector set, the target working condition feature vector is determined.

4. The intelligent adjustment method for air pressure and air velocity in an air separator for removing impurities according to claim 1, characterized in that, Determining the real-time impurity content based on the image data of the material after air separation specifically includes: The material image data is input into a preset image semantic segmentation model to obtain the image classification result; The real-time impurity content is calculated based on the image classification results.

5. The intelligent adjustment method for air pressure and air velocity in an air separator for removing impurities according to claim 3, characterized in that, The current air volume demand is determined based on the real-time impurity content, specifically including: Obtain the model adjustment parameters of the first associated model; wherein, the model adjustment parameters are used to characterize the adjustment weight of the initial wind speed setpoint as the material characteristic parameters and the feed rate change; The current air volume demand is determined based on the real-time impurity content and the model adjustment parameters.

6. The intelligent adjustment method for air pressure and air velocity in an air separator for removing impurities according to claim 1, characterized in that, Based on the historical wind speed data, wind speed fluctuation characteristic values ​​are determined, specifically including: Determine the statistical parameters of the historical wind speed data; The statistical parameters are normalized, and the result is used as the wind speed fluctuation characteristic value.

7. The intelligent adjustment method for air pressure and air velocity in an air separator for removing impurities according to claim 2, characterized in that, The first association model includes linear models.

8. The intelligent adjustment method for air pressure and air velocity in an air separator for removing impurities according to claim 4, characterized in that, The types of preset image semantic segmentation models include deep learning-based image segmentation models.

9. The intelligent adjustment method for air pressure and air velocity of an air separator for removing impurities according to any one of claims 1-8, characterized in that, Acquire real-time operating data during the air separation process, specifically including: The material characteristic parameters and the feed rate are collected by the feed monitoring unit; The result monitoring unit collects image data of the material after air separation.

10. An intelligent adjustment system for air pressure and air velocity in an air separator, characterized in that, The system includes: a data acquisition module, an initial wind speed calculation module, an air volume demand analysis module, a fluctuation characteristic analysis module, and an adjustment determination and execution module; The data acquisition module is used to acquire real-time operating condition data and historical wind speed data during the air separation process; wherein, the real-time operating condition data includes material characteristic parameters at the feed inlet, feed rate and material image data after air separation, and the historical wind speed data is wind speed data within a preset time period before the current moment; The initial wind speed calculation module is used to determine the initial wind speed setting value of the blower based on the material characteristic parameters at the feed inlet and the feed rate. The air volume demand analysis module is used to determine the real-time impurity content based on the material image data after air separation, and to determine the current air volume demand based on the real-time impurity content. The fluctuation characteristic analysis module is used to determine the wind speed fluctuation characteristic value based on the historical wind speed data. The adjustment determination and execution module is used to determine and execute the target adjustment wind speed of the fan based on the wind speed fluctuation characteristic value, the current air volume demand and the initial wind speed setting value.

Images