Multi-stage screening equipment and method for green tea processing

By using deep learning-based artificial intelligence technology to monitor and analyze the real-time dynamic changes in filter volume and vibration frequency, the problems of easy clogging and low precision of traditional green tea sieving equipment are solved. This enables intelligent and refined control of the green tea sieving process, improving sieving efficiency and precision.

CN118635112BActive Publication Date: 2024-11-22四川省凤鸣茶业有限公司 +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411095288.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-12
Publication Date
2024-11-22
Estimated Expiration
2044-08-12

AI Technical Summary

Technical Problem

Traditional green tea sieving equipment has a fixed screen aperture, which cannot meet the sieving requirements of different grades of green tea. It is prone to clogging, has low sieving accuracy, and is cumbersome to operate. It also lacks scientific basis and control precision.

Method used

By employing deep learning-based artificial intelligence technology, the temporal dynamic changes of filter volume and vibration frequency are monitored and analyzed in real time. The temporal responsiveness correlation between filter volume and vibration frequency is discovered, enabling intelligent control of vibration frequency.

Benefits of technology

It improves the efficiency and precision of green tea sieving, optimizes the sieving effect of tea leaves, and achieves intelligent and refined control.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118635112B_ABST
    Figure CN118635112B_ABST
Patent Text Reader

Abstract

The present application relates to the field of intelligent screening technology, and specifically discloses a multi-stage screening device and method for green tea processing, which utilizes artificial intelligence technology based on deep learning to monitor and analyze the filtration volume and vibration frequency of the filter in real time, capture the temporal dynamic change pattern of the filtration volume and vibration frequency, and dig out the temporal responsiveness correlation between the filtration volume and vibration frequency, thereby intelligently regulating the vibration frequency of the vibration device based on this temporal responsiveness correlation. In this way, intelligent and refined control of the green tea screening process can be achieved to improve screening efficiency and screening accuracy, thereby optimizing the screening effect of tea leaves.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of intelligent screening technology, and more specifically, to a multi-stage screening device and method for green tea processing. Background Art

[0002] As one of the traditional famous teas in China, green tea has a long history and is deeply loved by people. The processing of green tea is a complex and delicate process, which involves multiple links, including picking, withering, killing green, rolling and drying. Among these links, screening is a key step in the processing of green tea, and its purpose is to classify tea leaves according to different particle sizes and shapes for subsequent processing and packaging.

[0003] Traditional screening equipment has a fixed mesh aperture and limited screening accuracy, which cannot meet the screening requirements of different grades of green tea. It generally has defects such as easy clogging of the mesh, low screening accuracy, and low screening efficiency. In addition, traditional screening equipment sets screening parameters based on manual experience, which is cumbersome to operate and prone to errors. Therefore, a multi-stage screening device and method for green tea processing is desired. Summary of the invention

[0004] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a multi-stage screening device and method for green tea processing, which uses artificial intelligence technology based on deep learning to monitor and analyze the filtration volume and vibration frequency of the filter in real time, capture the temporal dynamic change pattern of the filtration volume and vibration frequency, and dig out the temporal responsiveness correlation between the filtration volume and the vibration frequency, so as to intelligently control the vibration frequency of the vibration device based on this temporal responsiveness correlation. In this way, intelligent and refined control of the green tea screening process can be achieved to improve the screening efficiency and screening accuracy, thereby optimizing the screening effect of tea leaves.

[0005] Accordingly, according to one aspect of the present application, a multi-stage screening device for green tea processing is provided, which includes: a feed port and a discharge port; a screening mechanism arranged between the feed port and the discharge port, wherein the screening mechanism includes a plurality of screening units, each of the screening units includes a screen and a vibration device for driving the screen to vibrate; a weight sensor installed on each of the screens; and a control system, wherein the weight sensor is communicatively connected to the control system, and the control system controls the vibration frequency of each of the vibration devices through the weight data collected by the weight sensor.

[0006] In the above-mentioned multi-stage screening equipment for green tea processing, the control system includes: a filter weight data acquisition module, which is used to obtain the time series of the filter weight within a predetermined time period collected by the weight sensor; a vibration frequency data acquisition module, which is used to obtain the time series of the vibration frequency of the vibration device within the predetermined time period; a data preprocessing module, which is used to perform data preprocessing on the time series of the filter weight and the time series of the vibration frequency to obtain a filtration amount timing input vector and a vibration frequency timing input vector; a timing pattern feature extraction module, which is used to perform timing pattern feature extraction on the filtration amount timing input vector and the vibration frequency timing input vector to obtain a filtration amount timing correlation feature vector and a vibration frequency timing correlation feature vector; a responsiveness analysis module, which is used to perform gamma correction responsiveness analysis on the filtration amount timing correlation feature vector and the vibration frequency timing correlation feature vector to obtain a filtration amount-vibration frequency timing responsiveness correlation feature vector; and a control instruction generation module, which is used to determine a control instruction based on the filtration amount-vibration frequency timing responsiveness correlation feature vector.

[0007] In the above-mentioned multi-stage screening equipment for green tea processing, the data preprocessing module includes: a timing arrangement unit, which is used to arrange the time series of the filter weight and the time series of the vibration frequency into a filter weight timing input vector and a vibration frequency timing input vector according to the time dimension respectively; a filtration amount calculation unit, which is used to calculate the difference between the filter weights of each two adjacent positions in the filter weight timing input vector to obtain the filtration amount timing input vector.

[0008] In the above-mentioned multi-stage screening equipment for green tea processing, the timing pattern feature extraction module is used to: respectively pass the filtration amount timing input vector and the vibration frequency timing input vector through a timing pattern feature extractor based on a one-dimensional convolutional layer to obtain the filtration amount timing association feature vector and the vibration frequency timing association feature vector.

[0009] In the above-mentioned multi-stage screening equipment for green tea processing, the responsiveness analysis module is used to: use the responsiveness gamma correction fusion module to process the filtration amount time series correlation feature vector and the vibration frequency time series correlation feature vector to obtain the filtration amount-vibration frequency time series responsiveness correlation feature vector.

[0010] In the above-mentioned multi-stage screening device for green tea processing, the responsiveness analysis module is used to: process the filtration amount time series correlation feature vector and the vibration frequency time series correlation feature vector using the following responsiveness gamma correction fusion formula to obtain the filtration amount-vibration frequency time series responsiveness correlation feature vector; wherein the responsiveness gamma correction fusion formula is:

[0011] ;

[0012] in, is the first eigenvector of the time series correlation feature vector of the filtering amount eigenvalues, is the first eigenvector of the vibration frequency time series association eigenvalues, A, B, C and D are adjustment parameters with different values. is the first eigenvector of the filtration amount-vibration frequency time series responsiveness correlation characteristic vector feature values.

[0013] In the above-mentioned multi-stage screening equipment for green tea processing, the control instruction generation module includes: passing the filtration amount-vibration frequency timing responsiveness association feature vector through a classifier-based vibration frequency controller to obtain the control instruction, and the control instruction is used to indicate whether the vibration frequency at the current time point should increase, decrease or remain unchanged.

[0014] In the above-mentioned multi-stage screening equipment for green tea processing, it also includes a training module for training the one-dimensional convolutional layer-based temporal pattern feature extractor, the responsive gamma correction fusion module and the classifier-based vibration frequency controller.

[0015] In the above-mentioned multi-stage screening equipment for green tea processing, the training module includes: a training data acquisition unit, used to acquire training data, the training data including the time series of the training filter weight within a predetermined time period collected by the weight sensor, the time series of the training vibration frequency of the vibration device within the predetermined time period, and the true value of the control instruction of the vibration frequency; a training data arrangement unit, used to arrange the time series of the training filter weight and the time series of the training vibration frequency into a training filter weight timing input vector and a training vibration frequency timing input vector according to the time dimension respectively; a training filtration amount calculation unit, used to calculate the difference between the training filter weights at each two adjacent positions in the training filter weight timing input vector to obtain a training filtration amount timing input vector; a training data timing encoding unit, used to respectively encode the training filtration amount timing input vector and the training vibration frequency timing input vector through the one-dimensional based A convolutional layer temporal pattern feature extractor is used to obtain a training filter quantity temporal correlation feature vector and a training vibration frequency temporal correlation feature vector; a training data response fusion unit is used to use the responsive gamma correction fusion module to process the training filter quantity temporal correlation feature vector and the training vibration frequency temporal correlation feature vector to obtain a training filter quantity-vibration frequency temporal responsiveness correlation feature vector; a classification loss unit is used to pass the training filter quantity-vibration frequency temporal responsiveness correlation feature vector through the classifier-based vibration frequency controller to obtain a classification loss function value; a model training unit is used to train the one-dimensional convolutional layer-based temporal pattern feature extractor, the responsive gamma correction fusion module and the classifier-based vibration frequency controller with the classification loss function value, wherein, in each round of iteration of the training, the training filter quantity-vibration frequency temporal responsiveness correlation feature vector is optimized.

[0016] According to another aspect of the present application, a multi-stage screening method for green tea processing is provided, comprising:

[0017] Acquire a time series of filter screen weight within a predetermined time period collected by a weight sensor;

[0018] Acquire a time series of the vibration frequency of the vibration device within the predetermined time period;

[0019] Preprocessing the filter weight time series and the vibration frequency time series to obtain a filtering amount time series input vector and a vibration frequency time series input vector;

[0020] Performing time series pattern feature extraction on the filtering amount time series input vector and the vibration frequency time series input vector respectively to obtain a filtering amount time series associated feature vector and a vibration frequency time series associated feature vector;

[0021] Performing gamma correction responsiveness analysis on the filtering amount time series correlation feature vector and the vibration frequency time series correlation feature vector to obtain a filtering amount-vibration frequency time series responsiveness correlation feature vector;

[0022] Based on the filtering amount-vibration frequency time series responsiveness correlation characteristic vector, a control instruction is determined.

[0023] Compared with the prior art, the multi-stage screening equipment and method for green tea processing provided by the present application utilizes artificial intelligence technology based on deep learning to monitor and analyze the filtration volume and vibration frequency of the filter in real time, captures the temporal dynamic change pattern of the filtration volume and vibration frequency, and digs out the temporal responsiveness correlation between the filtration volume and the vibration frequency, thereby intelligently regulating the vibration frequency of the vibration device based on this temporal responsiveness correlation. In this way, intelligent and refined control of the green tea screening process can be achieved to improve screening efficiency and screening accuracy, thereby optimizing the screening effect of tea leaves. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0025] Figure 1 It is a block diagram of a control system in a multi-stage screening device for green tea processing according to an embodiment of the present application.

[0026] Figure 2 It is a schematic diagram of the architecture of a control system in a multi-stage screening device for green tea processing according to an embodiment of the present application.

[0027] Figure 3 It is a block diagram of a data preprocessing module in a multi-stage screening device for green tea processing according to an embodiment of the present application.

[0028] Figure 4 It is a block diagram of a training module in a multi-stage screening device for green tea processing according to an embodiment of the present application.

[0029] Figure 5 The present invention is a flowchart of a multi-stage screening method for green tea processing according to an embodiment of the present application.

[0030] 100 - control system for multi-stage screening equipment for green tea processing, 110 - filter weight data acquisition module, 120 - vibration frequency data acquisition module, 130 - data preprocessing module, 140 - timing pattern feature extraction module, 150 - responsiveness analysis module, 160 - control instruction generation module;

[0031] 131- timing arrangement unit, 132- filtering amount calculation unit;

[0032] 200-training module, 210-training data acquisition unit, 220-training data arrangement unit, 230-training filtering amount calculation unit, 240-training data time series encoding unit, 250-training data response fusion unit, 260-classification loss unit, 270-model training unit. DETAILED DESCRIPTION

[0033] Below, the embodiments of the present application will be described in more detail in conjunction with the accompanying drawings, and the above-mentioned and other purposes, features and advantages of the present application will become more obvious. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the example embodiments described herein. At the same time, the accompanying drawings are used to provide a further understanding of the embodiments of the present application, and constitute a part of the specification, and are used together with the embodiments of the present application to explain the present application, and do not constitute a limitation on the present application. In the drawings, the same reference numerals generally represent the same parts or steps.

[0034] The present application provides a multi-stage screening device for green tea processing, which includes: a feed port and a discharge port; a screening mechanism arranged between the feed port and the discharge port, wherein the screening mechanism includes a plurality of screening units, each of the screening units includes a screen and a vibration device for driving the screen to vibrate; a weight sensor installed on each of the screens; and a control system, wherein the weight sensor is communicatively connected to the control system, and the control system controls the vibration frequency of each of the vibration devices through weight data collected by the weight sensor.

[0035] In a specific implementation, the green tea raw material is fed into the screening mechanism through the feed port, and the control system controls the vibration device to vibrate the screen, so that the green tea raw material slides on the screen and is graded through holes of different sizes, and the graded green tea raw material is discharged from the discharge port.

[0036] In particular, when the vibration device is used to control the vibration of the screen to vibrate and screen the green tea raw material, the vibration frequency has an important influence on the screening effect of green tea. Specifically, the higher the vibration frequency, the smaller the vibration amplitude of the screen. This high-frequency vibration mode can make the green tea leaves jump more frequently on the screen, thereby increasing the contact opportunity between the leaves and the screen, and reducing the possibility of the leaves adhering to the screen, thereby improving the screening accuracy. However, this high-frequency vibration mode will also make the green tea leaves stay on the screen for a shorter time, thereby reducing the screening efficiency. Therefore, in the green tea screening process, in order to maintain the balance between screening efficiency and screening accuracy, it is necessary to select a suitable vibration frequency according to the particle size of the green tea leaves and accurately control the vibration frequency. In traditional green tea screening equipment, fixed vibration parameters are usually set based on manual experience. However, this control method is not only cumbersome to operate and inefficient, but also lacks scientific basis and control accuracy, and cannot ensure the consistency and stability of the screening effect.

[0037] In response to the above technical problems, the technical concept of this application is to use artificial intelligence technology based on deep learning to monitor and analyze the filtration volume and vibration frequency of the filter in real time, capture the time-series dynamic change pattern of the filtration volume and vibration frequency, and dig out the time-series responsiveness correlation between the filtration volume and the vibration frequency, so as to intelligently control the vibration frequency of the vibration device based on this time-series responsiveness correlation. In this way, intelligent and refined control of the green tea screening process can be achieved to improve the screening efficiency and screening accuracy, thereby optimizing the screening effect of tea leaves.

[0038] Figure 1 It is a block diagram of a control system in a multi-stage screening device for green tea processing according to an embodiment of the present application. Figure 2 FIG. 1 is a schematic diagram of the architecture of a control system in a multi-stage screening device for green tea processing according to an embodiment of the present application. Figure 1 and Figure 2As shown, according to the control system 100 of the multi-stage screening equipment for green tea processing according to the embodiment of the present application, it includes: a filter weight data acquisition module 110, which is used to obtain the time series of the filter weight collected by the weight sensor within a predetermined time period; a vibration frequency data acquisition module 120, which is used to obtain the time series of the vibration frequency of the vibration device within the predetermined time period; a data preprocessing module 130, which is used to perform data preprocessing on the time series of the filter weight and the time series of the vibration frequency respectively to obtain a filtration amount time series input vector and a vibration frequency time series input vector; a time series pattern feature extraction module 140, which is used to perform time series pattern feature extraction on the filtration amount time series input vector and the vibration frequency time series input vector respectively to obtain a filtration amount time series correlation feature vector and a vibration frequency time series correlation feature vector; a responsiveness analysis module 150, which is used to perform gamma correction responsiveness analysis on the filtration amount time series correlation feature vector and the vibration frequency time series correlation feature vector to obtain a filtration amount-vibration frequency time series responsiveness correlation feature vector; a control instruction generation module 160, which is used to determine a control instruction based on the filtration amount-vibration frequency time series responsiveness correlation feature vector.

[0039] In the control system 100 of the multi-stage screening equipment for green tea processing, the filter weight data acquisition module 110 is used to obtain the time series of the filter weight within a predetermined time period collected by the weight sensor. It should be understood that the filter weight is the weight of the green tea raw material on the filter. By collecting the time series of the filter weight in real time, the dynamic change information of the weight of the green tea raw material on the screen during the screening process can be obtained, reflecting the screening efficiency of the green tea raw material. At the same time, by analyzing the time series of the filter weight, it is also possible to determine whether the filter is prone to clogging. For example, if the filter weight remains unchanged for a long time, it means that the filter may be clogged and needs to be cleaned in time.

[0040] In the control system 100 of the multi-stage screening equipment for green tea processing, the vibration frequency data acquisition module 120 is used to obtain the time series of the vibration frequency of the vibration device within the predetermined time period. It should be understood that the vibration frequency is the frequency at which the vibration device drives the screen to vibrate, which determines the beating frequency and screening efficiency of the green tea raw materials on the screen. By collecting the time series of the vibration frequency in real time, it is possible to obtain the vibration control of the screen by the vibration device during the screening process, extract the time series change pattern of the vibration frequency, and combine the weight change information of the filter to dig out the effect mode of the vibration frequency on the screening efficiency of the green tea raw materials, so as to perform adaptive control of the vibration frequency.

[0041] In the control system 100 of the multi-stage screening device for green tea processing, the data preprocessing module 130 is used to perform data preprocessing on the time series of the filter weight and the time series of the vibration frequency to obtain a filtering amount time series input vector and a vibration frequency time series input vector. Specifically, Figure 3 FIG. 1 is a block diagram of a data preprocessing module in a multi-stage screening device for green tea processing according to an embodiment of the present application. Figure 3 As shown, the data preprocessing module 130 includes: a timing arrangement unit 131, which is used to arrange the time series of the filter weight and the time series of the vibration frequency into a filter weight timing input vector and a vibration frequency timing input vector according to the time dimension respectively; a filtration amount calculation unit 132, which is used to calculate the difference between the filter weights of each two adjacent positions in the filter weight timing input vector to obtain the filtration amount timing input vector.

[0042] Specifically, the time series arrangement unit 131 is used to arrange the time series of the filter weight and the time series of the vibration frequency into a filter weight time series input vector and a vibration frequency time series input vector according to the time dimension. It should be understood that considering that the time series of the filter weight and the time series of the vibration frequency have time dependence, that is, there is a correlation between the data at different time points, therefore, in order to maintain the time correlation of the data, the time series of the filter weight and the time series of the vibration frequency are further arranged according to the time dimension respectively, so as to retain the time series characteristics of the data, and provide effective data support for subsequent feature extraction and pattern recognition.

[0043] Specifically, the filtration amount calculation unit 132 is used to calculate the difference between the filter weights at each adjacent two positions in the filter weight time series input vector to obtain the filtration amount time series input vector. It should be understood that the time series of the filter weight represents the weight information of the green tea raw material on the screen at each time point during the screening process, and the filtration amount of the filter refers to the weight of the green tea leaves passing through the filter per unit time. In the technical solution of the present application, it is considered that the filtration amount of the filter has an important characterization significance for the falling speed and screening efficiency of the green tea raw material during the screening process. For example, if the filtration amount is large, it means that the screening efficiency is high; if the filtration amount is small, it means that the screening efficiency is low. Therefore, in order to further understand the screening process of the green tea raw material, by calculating the difference between the filter weights at two adjacent time points, that is, calculating the difference between the filter weights at each adjacent two positions of the filter weight time series input vector, the dynamic change information of the falling speed and screening efficiency of the green tea raw material during the screening process is obtained, thereby obtaining the filtration amount time series input vector.

[0044] In the control system 100 of the multi-stage screening equipment for green tea processing, the timing pattern feature extraction module 140 is used to extract the timing pattern features of the filtering amount timing input vector and the vibration frequency timing input vector respectively to obtain the filtering amount timing correlation feature vector and the vibration frequency timing correlation feature vector. In a specific example of the present application, the encoding method for extracting the timing pattern features of the filtering amount timing input vector and the vibration frequency timing input vector respectively is to pass the filtering amount timing input vector and the vibration frequency timing input vector respectively through a timing pattern feature extractor based on a one-dimensional convolution layer to obtain the filtering amount timing correlation feature vector and the vibration frequency timing correlation feature vector. It should be understood that the one-dimensional convolution layer is a neural network layer specifically used to process one-dimensional data, such as time series data. In the technical solution of the present application, the time series pattern feature extractor based on the one-dimensional convolutional layer can extract the local change characteristics of the filtration amount and vibration frequency of the filter by performing convolution operations on the filtration amount time series input vector and the vibration frequency time series input vector in a sliding window manner in the time dimension, and capture the change pattern of the filtration amount and vibration frequency of the filter in the time dimension, so as to better understand the dynamic process of green tea raw material screening.

[0045] In the control system 100 of the multi-stage screening equipment for green tea processing, the responsiveness analysis module 150 is used to perform gamma correction responsiveness analysis on the filtering amount time series correlation feature vector and the vibration frequency time series correlation feature vector to obtain the filtering amount-vibration frequency time series responsiveness correlation feature vector. In a specific example of the present application, the processing method for performing gamma correction responsiveness analysis on the filtering amount time series correlation feature vector and the vibration frequency time series correlation feature vector is to use the responsiveness gamma correction fusion module to process the filtering amount time series correlation feature vector and the vibration frequency time series correlation feature vector to obtain the filtering amount-vibration frequency time series responsiveness correlation feature vector. It should be understood that the vibration frequency has a direct impact on the screening and filtering amount of green tea raw materials. Therefore, it is expected to extract the time series responsiveness correlation relationship between the vibration frequency and the green tea screening and filtering amount, so as to feedback control the vibration frequency based on this time series responsiveness correlation relationship to achieve the purpose of optimizing the tea screening effect. Based on this, in the technical solution of the present application, a responsive gamma correction fusion module is used to perform responsive correlation analysis on the filtering amount time-series correlation feature vector and the vibration frequency time-series correlation feature vector. Specifically, the gamma correction response fusion analysis module introduces a gamma correction function to perform nonlinear transformation on the filtering amount time-series correlation feature vector and the vibration frequency time-series correlation feature vector, and performs nonlinear responsiveness compensation on the time-series correlation relationship between the vibration frequency and the green tea screening filtering amount, so as to strengthen the correlation interactive response relationship between the two, reveal the sensitivity of the green tea screening filtering amount to the vibration frequency, and thus provide a basis for the control of the vibration frequency.

[0046] Specifically, the responsiveness analysis module 150 is used to process the filtering amount time series correlation feature vector and the vibration frequency time series correlation feature vector using the following responsiveness gamma correction fusion formula to obtain the filtering amount-vibration frequency time series responsiveness correlation feature vector; wherein the responsiveness gamma correction fusion formula is:

[0047] ;

[0048] in, is the first eigenvector of the time series correlation feature vector of the filtering amount eigenvalues, is the first eigenvector of the vibration frequency time series association eigenvalues, A, B, C and D are adjustment parameters with different values. is the first eigenvector of the filtration amount-vibration frequency time series responsiveness correlation characteristic vector feature values.

[0049] In the control system 100 of the multi-stage screening equipment for green tea processing, the control instruction generation module 160 is used to determine the control instruction based on the filtration amount-vibration frequency timing responsiveness association characteristic vector. In a specific example of the present application, the implementation method of determining the control instruction based on the filtration amount-vibration frequency timing responsiveness association characteristic vector is to pass the filtration amount-vibration frequency timing responsiveness association characteristic vector through a classifier-based vibration frequency controller to obtain the control instruction, and the control instruction is used to indicate whether the vibration frequency at the current time point should increase, decrease, or remain unchanged. It should be understood that the vibration frequency controller can classify and judge the filtration amount-vibration frequency timing responsiveness association characteristic vector based on a classification algorithm, and predict the optimal adjustment direction of the vibration frequency. Specifically, the vibration frequency controller can learn the classification mapping rules between the timing correlation characteristics of tea filtration amount and vibration frequency and the vibration frequency adjustment direction through a training process, and then use the trained vibration frequency controller to perform feature pattern learning and analysis on the filtration amount-vibration frequency timing responsiveness correlation feature vector. The vibration frequency controller can make classification decisions on the filtration amount-vibration frequency timing responsiveness correlation feature vector based on the classification mapping rules learned during the training process, thereby outputting control instructions indicating that the vibration frequency should increase, decrease or remain unchanged, thereby guiding the adjustment of the vibration frequency to achieve the optimization of the screening effect of the green tea raw materials.

[0050] It should be understood that before using the above neural network model, it is necessary to train the temporal pattern feature extractor based on the one-dimensional convolution layer, the responsive gamma correction fusion module and the vibration frequency controller based on the classifier. That is to say, in the multi-stage screening device for green tea processing of the present application, a training module for training the temporal pattern feature extractor based on the one-dimensional convolution layer, the responsive gamma correction fusion module and the vibration frequency controller based on the classifier is also included.

[0051] Figure 4 FIG. 1 is a block diagram of a training module in a multi-stage screening device for green tea processing according to an embodiment of the present application. Figure 4As shown, the training module 200 includes: a training data acquisition unit 210, which is used to acquire training data, wherein the training data includes a time series of the training filter weight within a predetermined time period collected by the weight sensor, a time series of the training vibration frequency of the vibration device within the predetermined time period, and a true value of the control instruction of the vibration frequency; a training data arrangement unit 220, which is used to arrange the time series of the training filter weight and the time series of the training vibration frequency into a training filter weight time series input vector and a training vibration frequency time series input vector according to the time dimension; a training filter amount calculation unit 230, which is used to calculate the difference between the training filter weights at each two adjacent positions in the training filter weight time series input vector to obtain a training filter amount time series input vector; a training data time series encoding unit 240, which is used to respectively pass the training filter amount time series input vector and the training vibration frequency time series input vector through the one-dimensional convolutional layer-based A timing pattern feature extractor is used to obtain a training filter quantity timing correlation feature vector and a training vibration frequency timing correlation feature vector; a training data response fusion unit 250 is used to use the responsive gamma correction fusion module to process the training filter quantity timing correlation feature vector and the training vibration frequency timing correlation feature vector to obtain a training filter quantity-vibration frequency timing responsiveness correlation feature vector; a classification loss unit 260 is used to pass the training filter quantity-vibration frequency timing responsiveness correlation feature vector through the classifier-based vibration frequency controller to obtain a classification loss function value; a model training unit 270 is used to train the one-dimensional convolutional layer-based timing pattern feature extractor, the responsive gamma correction fusion module and the classifier-based vibration frequency controller with the classification loss function value, wherein, in each round of iteration of the training, the training filter quantity-vibration frequency timing responsiveness correlation feature vector is optimized.

[0052] In the technical solution described above, the training filter amount time series association feature vector and the training vibration frequency time series association feature vector respectively express the local time series association features of the training filter amount and the training vibration frequency. However, considering the difference in source time series distribution of the training filter amount and the training vibration frequency, the training filter amount time series association feature vector and the training vibration frequency time series association feature vector will also have feature distribution misalignment, so that when the training filter amount time series association feature vector and the training vibration frequency time series association feature vector are processed using the responsive gamma correction fusion module, the obtained training filter amount-vibration frequency time series responsiveness association feature vector will have a more significant local distribution inconsistency of feature distribution, that is, the training filter amount-vibration frequency time series responsiveness association feature vector has a more significant prior-posteriori class probability causal association missing relative to the source domain time series distribution of the training filter weight and the training vibration frequency, thereby affecting the accuracy of the classification result.

[0053] Based on this, in a preferred embodiment, when the training filter amount-vibration frequency timing responsiveness associated feature vector is input into a vibration frequency controller based on a classifier to obtain a control instruction, the present application optimizes the training filter amount-vibration frequency timing responsiveness associated feature vector, specifically comprising the following steps: obtaining an increase probability value, a reduction probability value, and a retention probability value corresponding to whether the vibration frequency at the current time point should increase, decrease, or remain unchanged, obtained by inputting the training filter amount-vibration frequency timing responsiveness associated feature vector into the vibration frequency controller based on the classifier;

[0054] After deducting one third of the increase probability value, the reduction probability value, and the retention probability value respectively, calculate the square root of the sum of their squares, and add it to one third to obtain the strategy probability value;

[0055] Probabilistically converting each eigenvalue of the training filtration amount-vibration frequency time series responsiveness associated eigenvector based on a probability activation function, such as a sigmoid function, to obtain a probabilistic filtration amount-vibration frequency time series responsiveness associated eigenvector;

[0056] Determine a filter amount-vibration frequency timing responsiveness association cognitive symbol value based on a comparison between each characteristic value of the probabilistic filter amount-vibration frequency timing responsiveness association characteristic vector and the strategy probability value, wherein the filter amount-vibration frequency timing responsiveness association cognitive symbol value is equal to one, zero and negative one in response to the characteristic value of the probabilistic filter amount-vibration frequency timing responsiveness association characteristic vector being greater than, equal to and less than the strategy probability value, respectively;

[0057] Calculating the mean of all eigenvalues ​​of the probabilistic filtration amount-vibration frequency timing responsiveness correlation eigenvector to obtain the overall phase shift value of the filtration amount-vibration frequency timing responsiveness correlation;

[0058] Each eigenvalue of the probabilistic filtering amount-vibration frequency timing responsiveness association eigenvector is multiplied by the filtering amount-vibration frequency timing responsiveness association cognitive symbol value and the filtering amount-vibration frequency timing responsiveness association overall phase shift value to obtain a probabilistic filtering amount-vibration frequency timing responsiveness association cognitive eigenvector and a probabilistic filtering amount-vibration frequency timing responsiveness association phase shift eigenvector;

[0059] The probabilistic filtering amount-vibration frequency timing responsiveness associated cognitive feature vector and the probabilistic filtering amount-vibration frequency timing responsiveness associated phase shift feature vector are respectively multiplied by a first weight and a second weight as hyperparameters, and then the difference is calculated to obtain a probabilistic filtering amount-vibration frequency timing responsiveness associated differential feature vector; and

[0060] The absolute value of each eigenvalue of the probabilistic filtration amount-vibration frequency timing responsiveness association differential eigenvector is taken to obtain an optimized training filtration amount-vibration frequency timing responsiveness association eigenvector.

[0061] It is expressed as:

[0062]

[0063]

[0064]

[0065] in, is the probability-based filtering amount-vibration frequency time series responsiveness correlation feature vector The eigenvalues ​​at the positions, is the first element of the optimized training filter-vibration frequency temporal responsiveness correlation feature vector The eigenvalues ​​at the positions, is the strategy probability value, is the first of the increase probability value, the reduction probability value and the retention probability probability value, Indicates the cognitive symbol value associated with the filtering amount and the vibration frequency timing response, is the overall phase shift value associated with the filtering amount and the vibration frequency timing response, and denote the first weight and the second weight as hyperparameters respectively.

[0066] That is, in the above preferred embodiment, the strategy cognitive phase change response of the training filter amount-vibration frequency timing responsiveness associated feature vector is obtained by comparing the probabilistic amplitude of the eigenvalue of the training filter amount-vibration frequency timing responsiveness associated feature vector with the strategy probability representation, and the phase shift response of the eigenvalue of the training filter amount-vibration frequency timing responsiveness associated feature vector relative to the strategy probability representation of the feature as a whole is subjected to a feature distribution sequence invariance transformation based on the strategy probability differential distribution expansion, so as to realize the causal constraint of the posterior class probability of the training filter amount-vibration frequency timing responsiveness associated feature vector on its prior feature distribution representation, so as to improve the accuracy of the control instructions obtained by inputting the training filter amount-vibration frequency timing responsiveness associated feature vector into the vibration frequency controller based on the classifier.

[0067] In summary, the multi-stage screening device for green tea processing according to the embodiment of the present application is explained, which uses artificial intelligence technology based on deep learning to monitor and analyze the filtration volume and vibration frequency of the filter in real time, captures the time-series dynamic change pattern of the filtration volume and vibration frequency, and digs out the time-series responsiveness correlation between the filtration volume and the vibration frequency, thereby intelligently regulating the vibration frequency of the vibration device based on this time-series responsiveness correlation. In this way, intelligent and refined control of the green tea screening process can be achieved to improve screening efficiency and screening accuracy, thereby optimizing the screening effect of tea leaves.

[0068] Figure 5 Flow chart of a multi-stage screening method for green tea processing according to an embodiment of the present application. Figure 5 As shown, the multi-stage screening method for green tea processing according to the embodiment of the present application includes the following steps: S1, obtaining the time series of the filter weight within a predetermined time period collected by the weight sensor; S2, obtaining the time series of the vibration frequency of the vibration device within the predetermined time period; S3, performing data preprocessing on the time series of the filter weight and the time series of the vibration frequency to obtain a filtration amount timing input vector and a vibration frequency timing input vector; S4, performing timing pattern feature extraction on the filtration amount timing input vector and the vibration frequency timing input vector to obtain a filtration amount timing correlation feature vector and a vibration frequency timing correlation feature vector; S5, performing gamma correction responsiveness analysis on the filtration amount timing correlation feature vector and the vibration frequency timing correlation feature vector to obtain a filtration amount-vibration frequency timing responsiveness correlation feature vector; S6, determining a control instruction based on the filtration amount-vibration frequency timing responsiveness correlation feature vector.

[0069] Here, those skilled in the art will appreciate that the specific operations of each step in the multi-stage screening method for green tea processing have been described in detail above. Figures 1 to 4 The multi-stage screening equipment for green tea processing has been described in detail, and therefore, its repeated description will be omitted.

[0070] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to being implemented by adopting the above specific details.

[0071] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the module division is only a logical function division, and there may be other division methods in actual implementation. The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0072] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0073] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference to a figure in a claim should not be considered as limiting the claim to which it relates.

[0074] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units stated in the device claim can also be implemented by one unit through software or hardware.

[0075] Finally, it should be noted that the above description has been given for the purpose of illustration and description. In addition, the above embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A multi-stage screening device for green tea processing, characterized in that: include: Feed inlet and discharge outlet; A screening mechanism disposed between the feed port and the discharge port, wherein the screening mechanism comprises a plurality of screening units, each of the screening units comprises a screen and a vibration device for driving the screen to vibrate; A weight sensor installed on each of the screens; A control system, wherein the weight sensor is communicatively connected to the control system, and the control system controls the vibration frequency of each of the vibration devices through weight data collected by the weight sensor; The control system comprises: A filter weight data acquisition module is used to acquire a time series of filter weight within a predetermined time period collected by a weight sensor; A vibration frequency data acquisition module, used to acquire a time series of the vibration frequency of the vibration device within the predetermined time period; A data preprocessing module, used for performing data preprocessing on the time series of the filter weight and the time series of the vibration frequency to obtain a filtering amount time series input vector and a vibration frequency time series input vector; A time series pattern feature extraction module, used for respectively performing time series pattern feature extraction on the filtering amount time series input vector and the vibration frequency time series input vector to obtain a filtering amount time series associated feature vector and a vibration frequency time series associated feature vector; A responsiveness analysis module, used for performing a gamma correction responsiveness analysis on the filtering amount time series correlation characteristic vector and the vibration frequency time series correlation characteristic vector to obtain a filtering amount-vibration frequency time series responsiveness correlation characteristic vector; A control instruction generation module, used for determining a control instruction based on the filtration amount-vibration frequency time series responsiveness correlation characteristic vector; The control instruction generating module comprises: The filtering amount-vibration frequency time series responsiveness associated feature vector is passed through a classifier-based vibration frequency controller to obtain the control instruction, wherein the control instruction is used to indicate whether the vibration frequency at the current time point should increase, decrease or remain unchanged.

2. The multi-stage screening equipment for green tea processing according to claim 1, characterized in that: The data preprocessing module comprises: A time series arrangement unit, used for arranging the time series of the filter weight and the time series of the vibration frequency into a filter weight time series input vector and a vibration frequency time series input vector according to the time dimension respectively; The filtering amount calculation unit is used to calculate the difference between the filter weights of every two adjacent positions in the filter weight time series input vector to obtain the filtering amount time series input vector.

3. The multi-stage screening equipment for green tea processing according to claim 2, characterized in that: The temporal pattern feature extraction module is used to: The filtering amount time series input vector and the vibration frequency time series input vector are respectively passed through a time series pattern feature extractor based on a one-dimensional convolutional layer to obtain the filtering amount time series associated feature vector and the vibration frequency time series associated feature vector.

4. The multi-stage screening device for green tea processing according to claim 3, characterized in that: The responsiveness analysis module is used to: The responsive gamma correction fusion module is used to process the filtering amount time series correlation feature vector and the vibration frequency time series correlation feature vector to obtain the filtering amount-vibration frequency time series responsiveness correlation feature vector.

5. The multi-stage screening equipment for green tea processing according to claim 4, characterized in that: The responsiveness analysis module is used to: The filtering amount time series correlation feature vector and the vibration frequency time series correlation feature vector are processed using the following responsiveness gamma correction fusion formula to obtain the filtering amount-vibration frequency time series responsiveness correlation feature vector; wherein the responsiveness gamma correction fusion formula is: in, is the first eigenvector of the time series correlation feature vector of the filtering amount eigenvalues, is the first eigenvector of the vibration frequency time series association eigenvalues, A, B, C and D are adjustment parameters with different values. is the first eigenvector of the filtration amount-vibration frequency time series responsiveness correlation characteristic vector feature values.

6. The multi-stage screening equipment for green tea processing according to claim 5, characterized in that: It also includes a training module for training the one-dimensional convolutional layer-based temporal pattern feature extractor, the responsive gamma correction fusion module and the classifier-based vibration frequency controller.

7. The multi-stage screening equipment for green tea processing according to claim 6, characterized in that: The training module comprises: A training data acquisition unit, used to acquire training data, the training data including a time series of a training filter weight collected by a weight sensor within a predetermined time period, a time series of a training vibration frequency of the vibration device within the predetermined time period, and a true value of a control instruction of the vibration frequency; A training data arrangement unit, used for arranging the time series of the training filter weight and the time series of the training vibration frequency into a training filter weight time series input vector and a training vibration frequency time series input vector according to the time dimension respectively; A training filter amount calculation unit, used for calculating the difference between the training filter weights of every two adjacent positions in the training filter weight time series input vector to obtain a training filter amount time series input vector; A training data time series encoding unit, used for respectively passing the training filter amount time series input vector and the training vibration frequency time series input vector through the time series pattern feature extractor based on the one-dimensional convolution layer to obtain a training filter amount time series associated feature vector and a training vibration frequency time series associated feature vector; A training data response fusion unit, used for processing the training filter amount time series correlation feature vector and the training vibration frequency time series correlation feature vector using the responsiveness gamma correction fusion module to obtain a training filter amount-vibration frequency time series responsiveness correlation feature vector; A classification loss unit, used for passing the training filter quantity-vibration frequency time series responsiveness associated feature vector through the classifier-based vibration frequency controller to obtain a classification loss function value; A model training unit is used to train the one-dimensional convolutional layer-based temporal pattern feature extractor, the responsive gamma correction fusion module and the classifier-based vibration frequency controller with the classification loss function value, wherein in each round of iteration of the training, the training filter amount-vibration frequency timing responsiveness association feature vector is optimized.

8. A multi-stage screening method for green tea processing, characterized in that: include: Acquire a time series of filter screen weight within a predetermined time period collected by a weight sensor; Acquire a time series of the vibration frequency of the vibration device within the predetermined time period; Preprocessing the filter weight time series and the vibration frequency time series to obtain a filtering amount time series input vector and a vibration frequency time series input vector; Performing time series pattern feature extraction on the filtering amount time series input vector and the vibration frequency time series input vector respectively to obtain a filtering amount time series associated feature vector and a vibration frequency time series associated feature vector; Performing gamma correction responsiveness analysis on the filtering amount time series correlation feature vector and the vibration frequency time series correlation feature vector to obtain a filtering amount-vibration frequency time series responsiveness correlation feature vector; Determining a control instruction based on the filtering amount-vibration frequency time series responsiveness correlation characteristic vector; The method of determining a control instruction based on the filtering amount-vibration frequency timing responsiveness association characteristic vector comprises: passing the filtering amount-vibration frequency timing responsiveness association characteristic vector through a classifier-based vibration frequency controller to obtain the control instruction, wherein the control instruction is used to indicate whether the vibration frequency at the current time point should increase, decrease, or remain unchanged.