Intelligent sunning method, system and device for large-leaf tea leaves

Abstract

The invention discloses an intelligent sunning method, system and device for large-leaf tea leaves, the moisture content of the tea leaves is monitored in real time through a non-contact near-infrared moisture detector, the grid layout is adaptively optimized in combination with a K-nearest neighbor algorithm and a variable-density grid division method, and dynamic path planning is performed by using a DeepPath deep learning algorithm, so that the tea leaf sunning efficiency is improved. The infrared drying module is adopted to compensate a key observation area so as to achieve the aim of homogenizing moisture, the hardware part is composed of a longitudinal movement control mechanism, a vertical lifting control mechanism and a transverse movement control mechanism, and the device integrates moisture content detection and drying heating functions and can accurately move and execute detection and drying tasks. And the uniformity and high efficiency of the tea leaf sunning process are ensured. According to the scheme, intelligent and automatic tea sunning management is realized, and the production quality and efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the field of tea drying technology, mainly to the field of automated tea sun drying technology, and specifically to an intelligent sun drying method, device and system for large-leaf tea. Background Art

[0002] The unique aroma and taste of large-leaf tea stems from its unique germplasm resources and processing techniques. The processing process for large-leaf tea is as follows: fresh leaf picking → spreading and airing → withering → rolling and twisting → sun-drying. Sun-drying is a key step in developing the aroma of large-leaf tea. Sun-drying allows the fresh leaves to lose moisture and react with sunlight, promoting the synthesis of aromatic compounds within the fresh leaves, thereby inducing and enhancing the aroma and flavor of the tea. Sun-drying involves spreading the rolled tea leaves evenly on a bamboo sieve or directly on the floor. Sunlight reduces the moisture content and grassy aroma of the tea leaves, enhances the activity of tea polyphenol oxidase, and promotes changes in the tea leaves' contents and aroma, resulting in a richer aroma and a fresher taste. Currently, tea processing in large-leaf tea-producing areas is largely carried out using traditional manual labor. During the peak spring tea picking season, rain and fog are common, making the sun-drying process particularly vulnerable to weather conditions, which directly impacts the quality of large-leaf tea.

[0003] Traditionally, tea farmers use sunlight to dry their tea leaves. While bamboo screens offer excellent air permeability, promoting the development of tea aroma, they also reduce yields and require more labor. Drying tea leaves on cement floors, on the other hand, lacks air permeability, creates a poor aroma, and can easily cause dead tea leaves, significantly impacting tea quality. Consequently, the actual tea drying process is affected by numerous factors. Rainy days can be challenging, and sunlight intensity may not meet the required drying requirements, leading to high moisture content leaves susceptible to mold. Conversely, overdrying can also compromise tea quality, resulting in uneven quality across tea production. Breaking through the technical bottlenecks of sun drying on rainy and foggy days, and addressing the climatic constraints of large-leaf tea sun drying, is crucial.

[0004] At present, during the sun-drying process of tea production enterprises, tea makers are required to be on site to timely understand the drying conditions of the tea leaves. The manual inspection and natural light drying methods result in low production efficiency, restrict production capacity, make it difficult to meet the batch production needs of tea companies, and there is also the problem of uneven sun-drying of tea leaves.

[0005] In recent years, with the improvement of industrialization and scale and the increase of labor cost, tea gardens and tea enterprises have the trend of intensive development. Unified management of planting end and batch production in processing process are the development trend of tea enterprises. Part of mechanization replaces the original pure manual mode. Although relevant equipment is introduced in the process of tea production, the independent and dispersed processing equipment and the degree of integration with modern manufacturing technology are low, and there is a lack of scientific and effective operation method, which leads to problems such as discontinuous material processing, high labor cost, low production efficiency and low automation level of production line. Tea enterprises need to apply modern scientific and technological innovation, use effective technical means and methods to efficiently and accurately detect and locate the substandard area in the withering of tea, and carry out corresponding drying treatment, so as to improve the withering efficiency and uniformity, on the one hand, improve the production efficiency and increase the product added value, on the other hand, reduce the loss and expenditure, and realize effective cost saving, on the basis of ensuring the stable quality of large leaf tea, realize the batch production and standardization of production. SUMMARY

[0006] It can be seen that the traditional processing technology is low in modernization degree and has a large proportion of labor force. In addition, the main producing area of large leaf tea has more rainy weather in the tea picking season, which restricts the production and processing efficiency of large leaf tea. In order to solve the problems existing in the prior art, the inventors provide an intelligent withering method, system and device for large leaf tea, which can detect the moisture of the fresh leaves placed after fixation and rolling in a targeted area according to the actual conditions, realize the integrated application of moisture detection technology and drying in the withering process of large leaf tea, and well solve the problems in the withering process of tea, such as the delay of controlling tea moisture by manual experience, the large difference in the final quality of the withered tea, the low uniformity quality, and the influence on the withering quality of tea.

[0007] Specifically, the present application is implemented as follows: an intelligent withering method for large leaf tea, after a batch of tea to be withered is laid on the withering field, the method comprises the following steps:

[0008] S1, performing first moisture pre-detection, the online moisture detector performs continuous preliminary pre-detection of the moisture content of the material along the preset path in the material laying area, and obtains the initial moisture content data of the material in the whole area;

[0009] S2, based on the obtained initial moisture content data, obtaining the moisture content distribution characteristic data, and based on the moisture content distribution characteristic data, dividing the area with large moisture content difference and the area with small moisture content difference, and determining the first stage detection grid, and determining the key observation grid unit and the general observation grid unit from the first stage detection grid;

[0010] S3, according to the first stage detection grid, a mobile monitoring path is set, the first stage grid sampling of the moisture content is performed, and the first stage detection data is obtained;

[0011] S4. After a predetermined time interval, perform a second-stage grid sampling, compare the second-stage detection data of each grid with the first-stage detection data, and obtain the material moisture change rate data;

[0012] S5. Based on the material moisture change rate and the second-stage detection data, optimize the key observation grid units and determine the homogenization moisture target, and perform corresponding infrared drying module compensation operations on the key observation grid units;

[0013] S6. After completing the infrared drying module compensation operation in the current stage, the materials on the drying platform are inspected for moisture again, and the key observation grid units are optimized again. If key observation grid units still exist, the corresponding infrared drying module compensation operation is performed again on the key observation grid units until the key observation grid units are eliminated, so that the materials on the entire drying platform are all general observation grid units.

[0014] Furthermore, it also includes step S7: based on a predetermined time interval, a third stage of grid sampling is performed, and then the identification calculation of the key observation grid unit is performed. If a key observation grid unit exists, step S6 is repeated to perform a third infrared drying module compensation operation on the key observation grid unit.

[0015] Furthermore, in step S1, the online moisture detector is a near-infrared moisture detector, which measures the moisture on the surface of the material in a non-contact and high-speed manner to quickly obtain the initial moisture content data of the entire material distribution area; the preset path is: the material laying area is initially gridded to ensure that each grid can cover part of the material area, and the path can continuously pass through each grid row by row. The online moisture detector obtains the initial moisture content data of the tea in the entire sun-drying area along the preset path.

[0016] Furthermore, the step of determining the first-stage detection grid in step S2 includes: performing data processing on the initial moisture content data of the material, including calculating the average moisture content, standard deviation, maximum value, and minimum value statistics, determining the pattern and trend of moisture distribution, and identifying possible outliers or moisture content mutation areas, wherein the outliers or moisture content mutation areas include: areas with large moisture content differences and areas where the moisture content exceeds the average value; and marking the outliers or moisture content mutation areas as key observation network units, and marking those with a difference from the average value or standard deviation that does not exceed the threshold as general observation network units.

[0017] Furthermore, the key observation network unit marked in step S2 also includes the following steps: combining the K-nearest neighbor algorithm (KNN) and the variable density-based grid division method to realize adaptive grid algorithm division, creating an initial grid in the material area according to step S1, the K-nearest neighbor algorithm uses the moisture content data of each measurement point as a feature vector, and for each new measurement point, uses the trained KNN model to predict its moisture content category. If the moisture content of adjacent material areas is consistent, they can be divided into a grid. These grids can be large or small and form each other. The area with large differences in adjacent moisture content can be demarcated as a grid and marked as a key area.

[0018] Furthermore, step S2 also includes: simulating the moisture content distribution under the initial grid and estimating the simulation error; adjusting the density of the grid according to the error estimation result and KNN classification, focusing on the key observation grid units of the first-stage detection grid in the material laying area, and the key areas where the moisture content changes dramatically, in order to refine the grid to improve the resolution, further divided into smaller regional grids, which are secondary grids; combined with the minimum working area of ​​the infrared drying equipment, the secondary grid is composed of the drying area of ​​the infrared drying equipment as the minimum grid unit, and the first-stage detection grid contains several grids of different sizes, which are the first-stage detection grids, serving as the basis for subsequent detection.

[0019] Furthermore, the first-stage grid sampling of moisture content in step S3 includes: combining the position of the "first-stage detection grid" in the material laying area, using the DeepPath deep learning algorithm for dynamic path planning, and the online moisture detector executes the optimized path to collect moisture data of the "first-stage detection grid"; in step S4, the step of performing the second-stage grid sampling also includes: using the DeepPath deep learning algorithm for dynamic path planning, and the online moisture detector executes the optimized path to collect the second-stage grid moisture data; in step S4, the predetermined time is 10 minutes, the second-stage grid moisture data is collected, and compared with the data sampled in the first stage to obtain the moisture change rate of the material, and based on the historical data storage in a large-leaf tea intelligent sun-drying system, the deep learning algorithm is used to predict the sun-drying time required for the current material; the target moisture value is pre-set, and learning optimization is performed based on the actual moisture content of the material and the moisture change rate of the material, and the predetermined time interval and the division of the detection stage are adjusted in time.

[0020] Furthermore, in step S5, the optimization of key observation grid units and determination of homogenization moisture targets include: based on the sun-drying rate predicted in the previous step, combined with the moisture situation of the material in the key area with excessive moisture content in the current grid, calculating the quantitative data of the drying compensation operation that should be performed on the material in the key area; for different subdivided grids and under different moisture content conditions, the infrared drying module is combined with the parameters issued by a large-leaf tea intelligent sun-drying system to compensate for the corresponding power and time; after completing the moisture drying task in a key area, the quantitative data of the drying compensation operation for the material in the next key area is recalculated in combination with the current drying time required and the sun-drying rate of the material, the parameters of the infrared drying module are adjusted, and the drying task of the next area is executed; each parameter data and result value is used as a training sample of the model in the large-leaf tea intelligent sun-drying system. As the data set continues to increase, the parameter setting becomes more accurate, stable and efficient.

[0021] Another aspect of the present invention provides an intelligent sun-drying system for large-leaf tea leaves. The moisture detector module has non-contact, high-speed mobile measurement capabilities and is used to pass over the tea leaves to be sun-dried along a pre-planned path. It can detect the moisture content of the tea leaves in real time, store the data, and feed it back to the data processing and analysis module.

[0022] Data processing and analysis module: This module processes initial moisture content data, including calculating statistics such as average moisture content and standard deviation, and identifying moisture distribution patterns, trends, and outliers. It also applies the K-nearest neighbor (KNN) algorithm and variable density gridding to adaptively divide and optimize the grid, distinguishing between key observation and general observation grid cells.

[0023] Path planning module: Integrates the DeepPath deep learning algorithm to dynamically plan the detection path based on the current grid layout to optimize the online moisture detector's travel path and send it to the moisture detector module to control the moisture detector module to move along the travel path;

[0024] Infrared drying module: Equipped with infrared drying equipment with adjustable power, it can perform corresponding drying compensation operations on key observation grid cells according to system instructions to achieve the goal of homogenizing moisture. The infrared drying module can move independently on the travel path, or be integrated with the moisture detector module to share a set of mobile control mechanisms.

[0025] Intelligent Prediction and Control Module: Based on real-time moisture detection data, it monitors the entire sun-drying process in real time, including changes in material moisture and drying efficiency within key or general observation grids. It also instantly adjusts drying strategies and drying parameter controls based on feedback. Based on historical data and real-time detection data, it uses deep learning algorithms to predict the rate of the sun-drying process, optimize drying strategies, and adjust key observation grid areas for drying or moisture detection. It continuously learns and optimizes parameter settings to improve processing efficiency and product quality.

[0026] Data storage and management system: Saves all test data, drying operation records and model training samples, supports data retrieval, analysis and model iterative updates to continuously improve system performance.

[0027] Furthermore, the grid management and optimization module: can manage the detection grids of the first stage and subsequent stages, including division, density adjustment, determination of key points and general observation grids, and dynamically adjust the grid division according to the change of material moisture, and send it to the data processing and analysis module and the path planning module; the human-computer interaction module: is used for the operator to input the requirements of the moisture detection of sun-drying tea leaves, and can set the detection interval time, the detection division stage and the size parameters of the adaptive grid according to the needs, and assist the central operation module to optimize the execution path; each detection task can be adjusted by the operator according to the actual state of sun-drying to achieve personalized, accurate and effective moisture detection; the path planning module: can also combine the current material sun-drying situation and the detection needs of the operator, through deep learning algorithms and path planning algorithms, calculate and match the quantitative data of the detection grid size, the optimal detection path and the drying compensation operation suitable for the execution of this task, and learn and adjust based on historical data values ​​to detect The detection and drying tasks are issued to the intelligent detection and drying integrated device and the tea drying mechanism through the control system; the intelligent detection and drying integrated device transmits the tea moisture value of the detection grid and the position information of the moisture detection module to the control system PLC logic controller and data storage module through the communication module; the information monitoring module: reads the process information and the analysis and calculation results of the path planning module in the control system in real time, and displays the data synchronously, tracks the process parameters in the process, and facilitates the operator to monitor and make the next step of sun-drying execution measures; the information monitoring module is used to read the moisture value of the detection grid and the moisture detection module position information from the control system to establish a definite relationship between each moisture detection module and the detected moisture value, so as to facilitate the operator to monitor and observe the current sun-drying situation; the data storage module: stores the mode information entered by the user, the path information and task information generated by the central operation module, and the process data information generated by the intelligent detection and drying integrated device in real time to realize data storage and interaction.

[0028] Working principle of the present invention: The intelligent sun-drying technology solution for large-leaf tea of ​​the present invention is a highly automated and intelligent active sun-drying process auxiliary technology. It uses an integrated hardware and software system to accurately control the moisture management of tea in the sun-drying process. By adjusting the moisture content changes of tea materials during the sun-drying process and actively intervening in drying compensation, it ensures the consistency of tea quality and efficient production. The entire system control is scientific, reasonable, efficient and stable. The main functional modules include: Pre-detection and data pre-processing analysis: First, the system uses an online moisture detector to perform continuous moisture content detection along the material laying area and collects the initial moisture content data of the entire area. Then, based on these data, the system analyzes the distribution characteristics of the moisture content of the material and divides the areas with significant and small differences in moisture content, so as to determine the grid units of different monitoring levels (key observations and general observations), laying the foundation for subsequent refined management; Dynamic monitoring and optimization adjustment: The system performs grid sampling according to the divided grids, obtains moisture data at each stage, and evaluates the moisture change rate by comparing the moisture content data in the same grid in different time periods, and then uses Using the K-nearest neighbor algorithm and adaptive grid division method, the system can dynamically adjust the grid density, merge adjacent grids with similar moisture contents into one grid, and use more refined secondary grid processing to monitor and respond in areas with large moisture changes, so as to achieve centralized processing of efficiency and effectiveness; intelligent drying compensation: for key observation grid units with high moisture content or large changes (moisture content changes slowly), the system will dispatch the infrared drying module to carry out targeted drying operations, and dynamically adjust the drying parameters (such as power and time) according to the preset target moisture value and real-time monitoring data to ensure the consistency of the sun-drying rate, thereby avoiding the adverse effects of harvesting tea leaves at different sun-drying degrees on the quality of tea. Deep learning optimization: During the entire sun-drying and compensatory drying control process of the present invention, the system uses deep learning algorithms to perform dynamic path planning, optimize detection paths, predict sun-drying time, optimize drying strategies, etc. As data accumulates, the system's learning model is continuously optimized, making decisions more accurate and efficient. In particular, with the integrated online moisture detector and drying device design, it can very efficiently realize network disk-like sun-drying field processing; Human-computer interaction and monitoring: Operators can set detection parameters, monitor real-time data, and adjust strategies according to actual conditions through the human-computer interaction module. The information monitoring module displays key information in the sun-drying process in real time to help operators make decisions. The data storage module saves all relevant data and supports fault tracing and analysis; Integrated control and execution: Computer terminals, control systems, and intelligent inspection and drying integrated devices work closely together through a communication network, forming a closed-loop control from receiving instructions, executing detection and drying operations to feedback data, to ensure intelligent management of the entire sun-drying process.

[0029] Beneficial technical effects of the present invention:

[0030] (1) Precise moisture management: An online moisture detector is used to continuously monitor the moisture content of the tea leaves on the drying platform through three-axis control. The preliminary moisture distribution characteristics are obtained through non-contact continuous pre-detection of the moisture content of the tea leaves. The first-stage detection grid, key observation grid unit, and general observation grid unit are determined. Through secondary detection, the key grid unit is re-optimized by comparing the moisture content change rate. The areas with high moisture content and slow change rate are dried to accelerate the moisture drying of the tea leaves in these areas, ensuring the consistency of the moisture management of the entire batch of tea during the withering process. This multi-stage grid sampling method can track the moisture change of each area in detail, timely detect and handle uneven moisture distribution, and actively compensate for uneven areas to achieve precise moisture management during the withering process.

[0031] (2) Intelligent regulation: Data analysis and machine learning algorithms (such as K- nearest neighbor algorithm and DeepPath deep learning algorithm) are used to automatically identify areas with large moisture differences and adaptively adjust the grid density. This method not only improves the targeting of moisture regulation but also dynamically optimizes the detection path and drying strategy, reducing unnecessary energy consumption and improving compensation operation efficiency.

[0032] (3) Homogenization treatment: The infrared drying compensation operation is implemented in the key observation grid unit to effectively solve the problem of local high moisture content, ensuring the uniformity of the moisture content of the entire batch of tea and improving the consistency of the tea quality. At the same time, the drying parameters are continuously optimized based on real-time data feedback to achieve double improvement of drying efficiency and effect. The system can dynamically adjust the drying time interval and detection stage based on historical data and real-time monitoring data. This flexibility considers the time sustainability factors during the drying and waiting drying of the withering process, ensuring that the parameters and expectations of each compensation consider the moisture loss factors of the tea leaves before and after drying and natural withering in the area, thereby optimizing the compensation degree during compensation drying, further ensuring the stability and reliability of the withering process, effectively improving the homogenization efficiency and quality.

[0033] (4) Continuous optimization mechanism: The data of each drying operation is used as a training sample for the intelligent withering system model, which promotes the self-learning and continuous improvement of the system. With the passage of time, the drying strategy will be more accurate and efficient, reducing the need for human intervention. BRIEF DESCRIPTION OF DRAWINGS

[0034] Figure 1 The basic flowchart of the intelligent withering method for large-leaf tea is shown in the figure.

[0035] Figure 2 The basic structure of the intelligent withering system for large-leaf tea is shown in the figure.

[0036] Figure 3 This is a schematic diagram of the basic structure of a grid path in an intelligent sun-drying method for large-leaf tea of ​​the present invention;

[0037] Figure 4 Schematic diagram of the division of key observation grids, general observation grids, and secondary grids in the grid path of the present invention;

[0038] Figure 5 A schematic diagram of the three-dimensional structure of an intelligent sun-drying device for large-leaf tea and its grid division with different moisture contents according to the present invention;

[0039] Figure 6 This is a structural stereogram of the longitudinal sliding mechanism of the device;

[0040] Figure 7 This is a structural stereogram of the transverse movement control mechanism of the device;

[0041] Figure 8 This is a structural stereogram of the vertical lifting mechanism of the device;

[0042] Figure 9 This is a structural stereogram of the drying integrated device for testing this device;

[0043] Figure 10 This is a schematic diagram of the flipping of the drying integrated device for detection of this device;

[0044] Figure 11 This is a schematic diagram of the combined use state of the device;

[0045] Among them: 10 - longitudinal slide rail, 11 - longitudinal drive device, 12 - longitudinal transmission mechanism, 13 - longitudinal slider seat, 14 - transmission belt, 15 - motor output shaft, 16 - driven shaft,

[0046] 20 - lifting mechanism, 21 - lifting drive device, 22 - lifting rail, 23 - lifting screw, 24 - lifting block,

[0047] 30 - transverse slide rail, 31 - transverse drive device, 32 - transverse transmission mechanism, 33 - installation sleeve,

[0048] 40 - integrated detection and drying device, 41 - moisture content detection device, 42 - drying and heating device, 43 - rotation drive motor, 44 - bidirectional mounting plate. DETAILED DESCRIPTION

[0049] To make the objectives, technical solutions, and advantages of the present invention more clearly understood, the present invention will be further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings. It should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present invention. In addition, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present invention.

[0050] Example 1: On a morning during the tea picking season, a tea processing factory used the intelligent sun-drying method, device, and system for large-leaf tea of ​​this embodiment to sun-dry a batch of rolled large-leaf tea leaves. The following are the detailed operating steps and data parameters:

[0051] 1) An intelligent sun-drying system for large-leaf tea leaves divides a 24-square-meter large-leaf tea sun-drying area into 96 initial grids of 0.5 m x 0.5 m.

[0052] 2) Starting with the sun-drying of tea leaves, the intelligent drying integrated device moves across the drying platform. The online moisture analyzer emits near-infrared light and detects the amount of reflected light, measuring 10 data points per second with an accuracy of ±0.1%. Its spot diameter is 50mm, ensuring efficient coverage of all grid cells, collecting a total of 96 data points.

[0053] 3) An intelligent sun-drying system for large-leaf tea calculates the average moisture content of all grid nodes and finds that the average moisture content of tea in the sun-drying area is 64%, with the highest moisture content being 77% (grid number #18) and the lowest being 45% (grid number

[0054] #76), the sum of the squares of the differences between the highest and lowest moisture contents and the average yielded 530. This sum of squares was then divided by 98 (the total number of grid cells minus 1), and the square root was taken to obtain the standard deviation. The calculation results show that the standard deviation of the moisture measurement data is approximately 2.32554580, indicating that the moisture distribution of tea leaves in the sun-drying area is uneven.

[0055] 4) The moisture content of grid #18 exceeds the average value by 10%, which is considered an abnormality by an intelligent sun-drying system for large-leaf tea. Based on this, the system identifies areas with extremely high moisture content in the current grid as markers and takes corresponding measures to ensure the uniformity of the overall moisture content of the tea leaves, namely temperature adjustment, wind speed control, and extended sun-drying time.

[0056] Example 2

[0057] An intelligent sun-drying system for large-leaf tea collected moisture values ​​from 96 initial grids in the large-leaf tea sun-drying area. The average moisture content was 64%, the highest was 77% (grid number #18), the lowest was 45% (grid number #76), and the other grids were distributed between 46% and 76%.

[0058] In the intelligent sun-drying system for large-leaf tea, we can use the moisture content of each grid node as a feature vector and use the KNN algorithm to classify the moisture content categories of the tea sun-drying area.

[0059] The calculation steps of the KNN algorithm are:

[0060] 1. Select K value: Determine the number of neighbors, K = 3 (selected based on historical data and experience).

[0061] 2. Distance calculation: Calculate the distance between the point to be classified and all other points.

[0062] 3. Find the nearest K neighbors: Find the K points closest to each other.

[0063] 4. Make a decision: Based on the known categories of the nearest K neighbors, the category of the point to be classified is determined by majority voting or other methods.

[0064] The following grid nodes have moisture content data (expressed as a percentage) and are labeled with their category (high moisture or low moisture):

[0065] Grid Number Moisture content category #01 75％ High moisture #02 70％ High moisture #03 65％ High moisture #04 55％ Low moisture #05 50％ Low moisture #06 45％ Low moisture

[0066] Next, a large-leaf tea intelligent sun-drying system needs to classify a new grid node #07, whose moisture content is 68%.

[0067] Calculation steps:

[0068] 1. Select K value: K=3.

[0069] 2. Distance calculation: Calculate the Euclidean distance between node #07 and all other nodes.

[0070] Distance (#07,#01) = |68-75| = 7

[0071] Distance (#07,#02) = |68-70| = 2

[0072] Distance (#07,#03) = |68-65| = 3

[0073] Distance (#07,#04) = |68-55| = 13

[0074] Distance (#07,#05) = |68-50| = 18

[0075] Distance (#07,#06) = |68-45| = 23

[0076] 3. Find the nearest K neighbors: According to the calculated distance, the 3 nearest neighbors are #02, #03, and #04.

[0077] 4. Decision making: Check the categories of the three nearest neighbors. #02 and #03 are classified as high moisture, and #04 is classified as low moisture. Therefore, node #07 is classified as high moisture.

[0078] The first stage of detection grid division

[0079] Consistency determination: If the moisture content difference between adjacent grids is less than 3%, merge them into one grid.

[0080] Difference determination: If the moisture content difference between adjacent grids is greater than 5%, divide them into different grids and mark the area with large difference as a key area.

[0081] Adaptive grid division: Merge grids with moisture content between 62% and 66%, and separately divide grids #18 and #76

[0082] Simulation and error estimation:

[0083] The simulated moisture content is an estimated value obtained by a large-leaf tea intelligent sun-drying system through analysis of historical data and current environmental conditions. The system assumes that the simulated moisture content is 65%, which is based on the average value in historical data, trend analysis, or a preset standard value.

[0084] The error is determined by calculating the difference between the actual measured value and the simulated value. This can be calculated by the absolute value formula, i.e. Error = |Actual value - Simulated value| Error = |Actual value - Simulated value|.

[0085] The actual moisture content data of the following grid nodes:

[0086] Grid Number Actual moisture content #18 77％ #76 45％ #01 62％ #02 64％ #03 66％

[0087] A large-leaf tea intelligent sun-drying system uses a simulated moisture content of 65% to calculate the error of each grid node:

[0088] Grid #18: Error = |77% - 65%| = 12%

[0089] Grid #76: Error = |45% - 65%| = 20%, the actual moisture content of grid #76 is 20% lower than the simulated value, which is a large error, indicating that the simulated value needs to be adjusted.

[0090] Grid #01: Error = |62% - 65%| = 3%, indicating that the simulated value is very close to the actual value.

[0091] Grid #02: Error = |64% - 65%| = 1%, the error is very small, indicating that the simulated value is accurate.

[0092] Grid #03: Error = |66% - 65%| = 1%

[0093] Error analysis

[0094] By analyzing the error, an intelligent withering system for large-leaf tea identifies the deviation between the simulated value and the actual value. If the error is large, re-evaluate the parameters of the simulation model or consider more influencing factors, such as environmental conditions, tea types, withering time, etc.

[0095] Grid Number Actual moisture content Simulated moisture content error(%) #18 77％ 65％ 12％ #76 45％ 65％ 20％ #01 62％ 65％ 3％ #02 64％ 65％ 1％ #03 66％ 65％ 1％

[0096] Adjusting grid density

[0097] According to the error estimation results, an intelligent withering system for large-leaf tea adjusts the density of the grid, especially in areas with large errors. For grids #18 and #76 with large errors, increase the monitoring frequency from every 5 minutes to every 2 minutes. Redivide the grids around #18 and #76 into smaller grids, 0.25m x 0.25m, shorten the detection distance, and improve the accuracy of monitoring.

[0098] So far, the first stage of grid division is complete, and the withering area is divided into key observation grid units and general observation grid units, with the key observation grid having a smaller granularity and more accurate detection.

[0099] Example 3

[0100] In the withering process of large-leaf tea, intelligent monitoring and adjustment are key to ensuring tea quality. This example describes an intelligent withering method, device, and system for large-leaf tea that combines online moisture detection and deep learning algorithms to optimize the withering process.

[0101] DeepPath algorithm dynamic path planning application

[0102] Initialize the environment: use the moisture detection data of the withering area as the state space, and the moisture content of each grid node as one dimension of the state.

[0103] Define the action space: the action space consists of all possible moisture detection operations, such as moving to adjacent grid nodes for detection.

[0104] State transition: the intelligent detection and drying integrated device (online moisture detector) moves within the withering area, updates the state according to the current position and executed actions.

[0105] Reward function: design a reward function to evaluate the behavior of the intelligent detection and drying integrated device, including:

[0106] Global accuracy (rGLOBAL): if the intelligent detection and drying integrated device reaches the target grid node (area with abnormal moisture content), a positive reward is given; otherwise, a negative reward is given.

[0107] Path efficiency (rEFFICIENCY): Encourages the intelligent inspection and drying integrated device to choose a shorter path to improve inspection efficiency. The reward value is inversely proportional to the path length.

[0108] Path diversity (rDIVERSITY): By calculating the cosine similarity between the current path and the known paths, the intelligent inspection and drying integrated device is encouraged to explore different paths.

[0109] Policy Network: Use the policy gradient method to train an intelligent path planning model for large-leaf tea, enabling it to select the optimal action based on the current state.

[0110] Path Planning: An intelligent path planning model for large-leaf tea starts from a starting point and gradually selects actions according to a trained policy network until it reaches the target state or reaches the maximum step limit.

[0111] Implementation of testing: The intelligent inspection and drying integrated device moves and collects moisture data of the first-stage detection grid according to the prompts of an intelligent path planning model for large-leaf tea.

[0112] Parameter update: Based on the collected data and reward function, the parameters of the policy network are updated to optimize subsequent path planning.

[0113] Dynamic path planning of the detection grid in the first stage:

[0114] The DeepPath algorithm is used to perform dynamic path planning based on the position of the detection grid in the first stage.

[0115] The online moisture detector collects moisture data according to the optimized path.

[0116] Dynamic path planning for the second phase detection grid:

[0117] After a predetermined time interval (eg, 10 minutes), the second stage gridded moisture is sampled.

[0118] The DeepPath algorithm is used for dynamic path planning to collect moisture data of the second-stage grid.

[0119] Calculation of moisture change rate:

[0120] Calculate the difference between the second-stage detection data and the first-stage detection data of each grid.

[0121] Calculate the moisture change rate: Moisture change rate = (moisture value in the second stage - moisture value in the first stage) / moisture value in the first stage × 100%

[0122] Prediction of drying time:

[0123] Based on historical data storage, deep learning algorithms are used to predict the drying time required for the current material.

[0124] The prediction model may use the following formula: Predicted drying time = f(actual moisture content, moisture change rate, historical data)

[0125] Learn to optimize and tune:

[0126] By presetting the target moisture value, learning optimization is performed based on the actual moisture content and moisture change rate of the material.

[0127] Adjust the scheduled time intervals and the division of the detection phases in time to achieve more precise moisture control.

[0128] Specific implementation steps

[0129] The key observation grids and general observation grids in the sun-drying area are divided into 175 large and small grids. The moisture content of each grid can be measured using an online moisture detector. An intelligent sun-drying system for large-leaf tea uses the DeepPath algorithm to plan the detection path.

[0130] The intelligent inspection and drying integrated device is located in grid #01, and its goal is to detect grid #18 (known to have abnormal moisture content).

[0131] The intelligent inspection and drying integrated device moves to grid #02 or #03 according to the strategy network selection.

[0132] The intelligent drying integrated device executes the action, moves to grid #02, and updates the status.

[0133] Reward calculation:

[0134] The intelligent inspection drying integrated device reaches grid #18, rGLOBAL=+1.

[0135] The path length is 5, rEFFICIENCY=1 / 5.

[0136] The cosine similarity with the known path is 0.1, rDIVERSITY = -0.1.

[0137] Total reward: Rtotal = λ1*rGLOBAL + λ2*rEFFICIENCY + λ3*rDIVERSITY, where λ1, λ2, and λ3 are weight coefficients.

[0138] Parameter update: Update the parameters of the policy network based on the total reward.

[0139] Phase 1 testing:

[0140] Execute the path planned by the DeepPath algorithm to collect the first-stage detection data. The first-stage moisture value of grid #01 is 65%.

[0141] Second stage testing:

[0142] After 10 minutes, the path planned by the DeepPath algorithm was executed to collect moisture data of the second-stage grids. The first-stage moisture value of grid #01 was 60%.

[0143] Calculation of moisture change rate:

[0144] For grid #01: Moisture change rate = (60% - 65%) / 65% × 100% ≈ -7.69%

[0145] Sun-drying time prediction:

[0146] Historical data shows that when the moisture change rate is -7.69%, the deep learning model is used to predict that the required drying time is about 30 minutes.

[0147] Optimization and adjustment:

[0148] If the target moisture content is 50%, the actual moisture content is 60%, and the moisture change rate is -7.69%, the system will increase the drying time or adjust other parameters.

[0149] Updated division of scheduled time intervals and detection phases for better moisture control.

[0150] Example 4

[0151] During the sun-drying process of large-leaf tea, precise control of the tea's moisture content is necessary to ensure uniform drying and high quality. This embodiment describes an intelligent sun-drying method, device, and system for large-leaf tea. This system uses an infrared drying module to compensate for key observation grid cells to optimize the tea drying process.

[0152] This system includes the following key steps:

[0153] Optimization of key observation grid cells:

[0154] Based on the material moisture change rate and the second-stage detection data, determine the key observation grid units that need to be optimized.

[0155] Infrared drying module compensation operation:

[0156] For key observation grid cells, the quantitative data of drying compensation operations are calculated based on the predicted sun-drying rate and current moisture content.

[0157] Duplicate detection and optimization:

[0158] After completing the compensation operation of the infrared drying module, the moisture content of the material on the drying platform is tested again, and the key observation grid units are continued to be optimized.

[0159] The third stage of grid sampling:

[0160] After a predetermined time interval, the third phase of gridded sampling is carried out to identify new key observation grid cells.

[0161] Parameter data as training samples:

[0162] Each parameter data and result value are used as training samples for the system model to improve the accuracy of parameter setting.

[0163] Second stage testing:

[0164] The second stage moisture value of grid #41 is 68%, which is higher than the target moisture content.

[0165] Calculate the drying compensation quantitative data:

[0166] The system predicts that the natural drying rate of the current environment is 2% / 10min. The moisture content of the tea leaves in the target grid is 68%. When the infrared drying power reaches 80%, the moisture loss rate is 16% / 10min.

[0167] Calculate quantitative data for drying compensation work:

[0168] Natural time = (68% - 50%) / (2% / 10 min) = 90 min

[0169] Drying time = (68% - 50%) / (2% + 16% / 10min) = 10min

[0170] The infrared drying module adjusts the power and time according to the calculation results.

[0171] Perform infrared drying module compensation operation:

[0172] Grid #41 was infrared dried with the power adjusted to 80% for 10 minutes.

[0173] Duplicate detection and optimization:

[0174] After drying, the moisture content of grid #41 is tested again. If it drops to 55%, which is still higher than the target moisture content, the infrared drying module compensation operation is continued.

[0175] The third stage of grid sampling:

[0176] After 10 minutes, the third stage of grid sampling is carried out, assuming that the moisture value of grid #41 drops to 50%.

[0177] If there are still key observation grid cells after the third stage detection, repeat step S5.

[0178] Model training and parameter optimization:

[0179] The parameter data of this drying (power 80%, time 10 minutes, moisture value reduced from 68% to 55%) is used as a training sample.

[0180] System model learning optimization to improve the efficiency and accuracy of future drying operations.

[0181] Example 4: A device for detecting and drying the moisture content of tea leaves during sun-drying, which is installed on a sun-drying table for use. The device comprises two sets of longitudinal sliding mechanisms, each of which is arranged on opposite sides of the sun-drying table. Each set comprises a longitudinal slide rail 10, a longitudinal drive device 11 and a longitudinal transmission mechanism 12. The longitudinal slider seat 13 is installed on the longitudinal slide rail 10; the longitudinal drive device 11 can drive the longitudinal slider seat 13 to move synchronously in the longitudinal direction; the device comprises two sets of vertical lifting mechanisms, each of which is arranged on a longitudinal slider seat 13 and can move with the longitudinal slider seat 13. Each set comprises a vertically arranged lifting stroke mechanism 20, a lifting drive device 21 and a lifting block 24. The transverse control mechanism is installed between the two lifting blocks 24. The space spans above the tea drying table, and the two lifting drive devices 21 can operate synchronously to drive the transverse control rod to rise and fall horizontally; the transverse control mechanism includes a transverse slide rail 30, a transverse drive device 31 and a transverse transmission mechanism 32. The detection and drying integrated device 40 is installed on the transverse slide rail 30 and is connected to the transverse transmission mechanism 32. The transverse drive device 31 can drive the detection and drying integrated device 40 to move laterally along the transverse slider; the detection and drying integrated device 40 includes a moisture content detection device 41 and a drying and heating device 42. The moisture content detection device 41 is used to detect the moisture content of the tea material, and the drying and heating device 42 is used to dry the tea material to accelerate the reduction of its moisture content.

[0182] Preferably, the longitudinal transmission mechanism 12 is a transmission belt 14, one end of which is connected to the motor output shaft 15 of the longitudinal driving device 11, and the other end is connected to the driven rotating shaft 16. The longitudinal sliding block seat 13 can be slidably installed on the longitudinal sliding rail 10 and is fixedly connected to the transmission belt 14. The longitudinal driving device 11 can drive the transmission belt 14 to move by rotating forward and backward, thereby driving the longitudinal sliding block seat 13 to move longitudinally along the longitudinal sliding rail 10 synchronously.

[0183] Preferably, the lifting stroke mechanism 20 includes a lifting slide rail 22 and a lifting screw 23. The lifting screw 23 is connected to the motor of the lifting drive device 21. The lifting block 24 is slidably mounted on the lifting slide rail 22 and is threadedly fitted with the lifting screw 23. The lifting block 24 can be lifted and lowered along the lifting slide rail 22 as the lifting screw 23 rotates forward and backward.

[0184] Preferably, both ends of the transverse slide rail 30 are fixedly connected to the inner sides of the two lifting blocks 24 and maintained parallel to the drying table. The transverse transmission mechanism 32 is a transverse screw, which is connected to the motor shaft of the transverse drive device 31. The detection and drying integrated device 40 includes a mounting sleeve 33, which is mounted on the transverse slide rail 30 and is installed in conjunction with the transverse screw thread. The mounting sleeve 33 can move laterally along the transverse slide rail 30 with the forward and reverse rotation of the transverse screw; the detection and drying integrated device 40 is installed on the mounting sleeve 33.

[0185] Preferably, the detection and drying integrated device 40 includes: a rotary drive motor 43, a bidirectional mounting plate 44, a moisture content detection device 41 and a drying and heating device 42; the rotary drive motor 43 is installed on the mounting sleeve 33, and its rotating shaft is horizontally arranged along the longitudinal direction and extends horizontally along the longitudinal direction. The bidirectional mounting plate 44 is connected to the rotating shaft of the rotary drive motor 43, and the rotary drive motor 43 can drive the bidirectional mounting plate 44 to rotate with the rotating shaft; the moisture content detection device 41 and the drying and heating device 42 are respectively installed on both sides of the bidirectional mounting plate 44.

[0186] Preferably, the rotary drive motor 43 can rotate 180° in reverse, or only rotate 180° each time, so as to always keep the working surfaces of the moisture content detection device 41 and the drying and heating device 42 parallel to the drying table.

[0187] Preferably, the moisture content detection device 41 is a near-infrared moisture meter; and the drying and heating device 42 is a microwave dryer or a far-infrared dryer.

[0188] Preferably, the two sets of longitudinal drive devices 11 and lifting drive devices 21 of the longitudinal sliding mechanism and the vertical lifting mechanism use servo motors and are equipped with corresponding servo drivers, which can provide synchronous, same speed and same torque control, and can set the same speed command value to the two servo drivers through programming to ensure that the two motors maintain synchronous operation; or a rotary encoder is installed on each motor, which can synchronously control the two motors to maintain synchronous operation through the encoder.

[0189] The moisture content detection and drying integrated device for tea leaf sunning is arranged on the side of the sunning table for installation, and the size specification is adapted to the size of the sunning table. Main moving parts include a longitudinal sliding mechanism, a vertical lifting mechanism and a transverse control mechanism, which are connected with the control end and uniformly controlled by the control end program. The longitudinal sliding mechanism and the transverse control mechanism control the planar position displacement of the detection and drying integrated device, and the vertical lifting mechanism controls the height position of the detection and drying integrated device relative to the sunning table surface. The detection and drying integrated device can be moved to and stopped at any point or region on the sunning table surface according to the control of the control end. The detection and drying integrated device combines the dual-function hardware structure of the moisture content detector and the drying equipment, so as to realize the online detection of the moisture content of the tea leaf material on the sunning table surface, or the compensatory drying of the tea leaf material on the sunning table surface, and accelerate the rapid reduction of the moisture content of the material in the local region.

[0190] Example 5: The purpose of this example is to compare the effect difference of the sunning time of the conventional sunning method and the intelligent sunning method of large-leaf tea proposed in the present application. By collecting multiple batches of tea sunning test data, using mathematical formulas and statistical analysis knowledge, the sunning time under the two methods is compared and analyzed to verify the effectiveness and superiority of the present application.

[0191] 1. Test data

[0192] In this test, the time period with similar sunshine environment was selected, and the conventional sunning method and the intelligent sunning method proposed in the present application were used to sun large-leaf tea. The weight of each batch of material was 200Kg, and the initial moisture content was slightly different. The test data is shown in the following table:

[0193] Traditional sunning time

[0194]

[0195] Sunning time using the present application

[0196]

[0197] 2. Analysis method

[0198] The statistical analysis methods such as mean, standard deviation and t-test are used to compare and analyze the traditional sunning time and the sunning time after using the present application.

[0199] 3. Results and analysis

[0200] Mean and standard deviation calculation

[0201] Firstly, the mean and standard deviation of the traditional sunning time and the sunning time after using the present application are calculated to evaluate the overall level and dispersion degree of the two methods.

[0202] Traditional sun green time mean value:

[0203]

[0204] Traditional sun green time standard deviation:

[0205]

[0206] Sun green time mean value after adopting the application:

[0207]

[0208] Sun green time standard deviation after adopting the application:

[0209]

[0210] t-test analysis

[0211] In order to verify whether there is a significant difference between the two sun green methods in sun green time, t-test is used for analysis.

[0212] T value calculation formula:

[0213]

[0214] Wherein: n CT and n GJ are the batch quantities of traditional sun green method and sun green method after adopting the application respectively, both are 8.

[0215] The calculated mean value and standard deviation are substituted into the t value calculation formula to obtain the t value.

[0216] Through t-test analysis, the following conclusions are obtained:

[0217] (1) Compared with the traditional sun green method, the sun green time of the large leaf tea intelligent sun green method of the application is significantly reduced.

[0218] (2) From the mean value, the traditional sun green time mean value is 5.4625h, while the sun green time mean value after adopting the application is 4.3125h, which is reduced by about 21%.

[0219] (3) The t-test result shows that there is a significant difference between the two sun green methods in sun green time, and the difference has statistical significance.

[0220] It should be understood that the above-described specific embodiments of the present invention are merely illustrative or illustrative of the principles of the present invention and do not constitute limitations of the present invention. Therefore, any modifications, equivalent substitutions, improvements, etc. made without departing from the spirit and scope of the present invention should be included within the scope of protection of the present invention. In addition, the appended claims are intended to cover all variations and modifications that fall within the scope and metes and bounds of the appended claims, or equivalents thereof.

Claims

1. An intelligent sun-drying method for large-leaf tea leaves, wherein a batch of tea leaves to be sun-dried is spread flat on the drying ground and loaded with materials, characterized in that The following steps are also included: S1. Perform the first moisture pre-detection. The online moisture detector performs a continuous preliminary pre-detection of the material moisture content along the material laying area according to the preset path to obtain the initial moisture content data of the material in the entire area. S2. Based on the acquired initial moisture content data, obtain material moisture content distribution characteristic data, divide the material moisture content distribution characteristic data into areas with large moisture content differences and areas with small moisture content differences, and determine a first-stage detection grid. From the first-stage detection grid, determine key observation grid units and general observation grid units; S3. Develop a mobile monitoring route based on the first-stage detection grid, perform the first-stage grid sampling of moisture content, and obtain the first-stage detection data; S4. After a predetermined time interval, perform a second-stage grid sampling, compare the second-stage detection data of each grid with the first-stage detection data, and obtain material moisture change rate data; S5. Based on the material moisture change rate and the second-stage detection data, optimize the key observation grid units and determine the homogenization moisture target, and perform corresponding infrared drying module compensation operations on the key observation grid units; S6. After completing the infrared drying module compensation operation in the current stage, the materials on the drying platform are inspected for moisture again, and the key observation grid units are optimized again. If key observation grid units still exist, the corresponding infrared drying module compensation operation is performed again on the key observation grid units until the key observation grid units are eliminated, so that the materials on the entire drying platform are all general observation grid units. Or, it also includes step S7: after a predetermined time interval, perform a third stage of grid sampling, and then perform an identification calculation of the key observation grid unit. If a key observation grid unit exists, continue to repeat step S6 and perform a third infrared drying module compensation operation on the key observation grid unit.

2. The intelligent sun-drying method for large-leaf tea according to claim 1, characterized in that: In step S1, the online moisture detector is a near-infrared moisture detector, which measures the moisture on the surface of the material in a non-contact and high-speed manner to quickly obtain the initial moisture content data of the entire material distribution area; the preset path is: the material laying area is initially gridded to ensure that each grid can cover part of the material area, and the path can continuously pass through each grid row by row. The online moisture detector obtains the initial moisture content data of the tea in the entire sun-drying area along the preset path.

3. The intelligent sun-drying method for large-leaf tea according to claim 1, characterized in that: The step of determining the first-stage detection grid described in step S2 includes: performing data processing on the initial moisture content data of the material, including calculating the average moisture content, standard deviation, maximum value, and minimum value statistics, determining the pattern and trend of moisture distribution, and identifying possible outliers or moisture content mutation areas, wherein the outliers or moisture content mutation areas include: areas with large moisture content differences and areas where the moisture content exceeds the average value; and marking the outliers or moisture content mutation areas as key observation network units, and marking the areas where the difference from the average value or standard deviation does not exceed the threshold as general observation network units.

4. The intelligent sun-drying method for large-leaf tea according to claim 3, characterized in that: The steps of marking the key observation network unit in step S2 include: combining the K-nearest neighbor algorithm (KNN) and the variable density-based grid division method to realize adaptive grid algorithm division, creating an initial grid in the material area according to step S1, and the K-nearest neighbor algorithm uses the moisture content data of each measurement point as a feature vector. For each new measurement point, the trained KNN model is used to predict its moisture content category. If the moisture content of adjacent material areas is consistent, they can be divided into a grid. These grids can be large or small and form each other. The area with large differences in adjacent moisture content can be demarcated as a grid and marked as a key area. Alternatively, step S2 further includes: simulating the moisture content distribution under the initial grid and estimating the simulation error; adjusting the density of the grid based on the error estimation result and the KNN classification, and further dividing the key observation grid units of the first-stage detection grid in the material laying area, and the key areas where the moisture content changes dramatically, into smaller regional grids to improve the resolution by refining the grid to improve the resolution, which are the secondary grids; Combined with the minimum working area of ​​the infrared drying equipment, the secondary grid is composed of the drying area of ​​the infrared drying equipment as the minimum grid unit. The first-stage detection grid contains several grids of different sizes. This grid is the first-stage detection grid, which serves as the basis for subsequent detection.

5. The intelligent sun-drying method for large-leaf tea according to claim 1, characterized in that: The first-stage grid sampling of moisture content in step S3 includes: using the DeepPath deep learning algorithm to perform dynamic path planning based on the location of the "first-stage detection grid" in the material laying area, and the online moisture detector executes the optimized path to collect moisture data of the "first-stage detection grid"; The step S4 of performing the second-stage grid sampling further includes: using the DeepPath deep learning algorithm to perform dynamic path planning, and the online moisture detector executes the optimized path to collect the second-stage grid moisture data; The predetermined time in step S4 is 10 minutes, and the grid moisture data of the second stage is collected and compared with the data sampled in the first stage to obtain the material moisture change rate. Based on the historical data storage in the intelligent sun-drying system for large-leaf tea, a deep learning algorithm is used to predict the sun-drying time required for the current material; By presetting the target moisture value, learning optimization is carried out based on the actual moisture content of the material and the moisture change rate of the material, and the predetermined time intervals and the division of the detection stages are adjusted in a timely manner.

6. The intelligent sun-drying method for large-leaf tea according to any one of claims 2 to 5, characterized in that: In step S5, optimizing the key observation grid units and determining the homogenization moisture target includes: based on the sun-drying rate predicted in the previous step and in combination with the moisture content of the material in the key area with excessive moisture content in the current grid, calculating quantitative data for the drying compensation operation that should be performed on the material in the key area; for different subdivided grids and different moisture content conditions, the infrared drying module compensates for the corresponding power and time in combination with parameters issued by a large-leaf tea intelligent sun-drying system; After completing the moisture drying task in a key area, the quantitative data of the drying compensation operation for the materials in the next key area is recalculated based on the current drying time and the material's drying rate. The parameters of the infrared drying module are adjusted to carry out the drying task in the next area. Each parameter data and result value is used as a training sample for the model in the intelligent sun-drying system for large-leaf tea. As the data set continues to increase, the parameter setting becomes more accurate, stable and efficient.

7. An intelligent sun-drying system for large-leaf tea leaves based on the intelligent sun-drying method for large-leaf tea leaves according to any one of claims 1 to 6, characterized in that include: Moisture detector module: This module has non-contact, high-speed mobile measurement capabilities. It is used to pass over the tea leaves to be sun-dried along a pre-planned path. It can detect the moisture content of the tea leaves in real time, store the data, and feed it back to the data processing and analysis module. Data processing and analysis module: This module processes initial moisture content data, including calculating statistics such as average moisture content and standard deviation, and identifying moisture distribution patterns, trends, and outliers. It also applies the K-nearest neighbor (KNN) algorithm and variable density gridding to adaptively divide and optimize the grid, distinguishing between key observation and general observation grid cells. Path planning module: Integrates the DeepPath deep learning algorithm to dynamically plan the detection path based on the current grid layout to optimize the online moisture detector's travel path and send it to the moisture detector module to control the moisture detector module to move along the travel path; Infrared drying module: Equipped with infrared drying equipment with adjustable power, it can perform corresponding drying compensation operations on key observation grid cells according to system instructions to achieve the goal of homogenizing moisture. The infrared drying module can move independently on the travel path, or be integrated with the moisture detector module to share a set of mobile control mechanisms. Intelligent prediction and control module: Based on real-time moisture detection data, it can monitor the entire sun-drying process in real time, including material moisture changes and drying efficiency within key observation grids or general observation grids. It can also adjust drying strategies and drying parameters based on feedback information. Based on historical data and real-time detection data, it can use deep learning algorithms to predict the rate of the sun-drying process, optimize drying strategies, and adjust key observation grid areas for drying or moisture detection. Ability to continuously learn and optimize parameter settings to improve processing efficiency and product quality; Data storage and management system: Saves all test data, drying operation records and model training samples, supports data retrieval, analysis and model iterative updates to continuously improve system performance.

8. The intelligent sun-drying system for the large-leaf tea intelligent sun-drying method according to claim 7 is characterized in that Also included are: Grid management and optimization module: It can manage the detection grids of the first stage and subsequent stages, including division, density adjustment, determination of key and general observation grids, and dynamically adjust the grid division according to the changes in material moisture, and send the data to the data processing and analysis module and the path planning module; Human-computer interaction module: This module allows operators to input requirements for moisture testing of sun-dried tea leaves. The module can set testing intervals, test phases, and adaptive grid size parameters as needed, and assists the central computing module in optimizing the execution path. Each test task can be adjusted by the operator based on the actual state of the sun-dried tea leaves, achieving personalized, accurate, and effective moisture testing. The path planning module can also combine the current material drying situation and the operator's inspection requirements, and use deep learning algorithms and path planning algorithms to calculate and match quantitative data on the inspection grid size, optimal inspection path, and drying compensation operations suitable for the execution of this task. It also performs learning and optimization based on historical data values, and issues the inspection and drying tasks to the intelligent inspection and drying integrated device and tea drying mechanism through the control system. The intelligent detection and drying integrated device transmits the tea moisture value of the detection grid and the position information of the moisture detection module to the control system PLC logic controller and data storage module through the communication module; Information monitoring module: reads the process information in the control system and the analysis and calculation results of the path planning module in real time, and displays the data synchronously to track the process parameters, making it easier for operators to monitor and make the next step of sun-drying execution; The information monitoring module is used to read the moisture value of the detection grid and the position information of the moisture detection module from the control system to establish a definite relationship between each moisture detection module and the detected moisture value, so as to facilitate the operator to monitor and observe the current sun-drying situation; Data storage module: real-time storage of mode information entered by the user, path information and task information generated by the central operation module, and process data information generated by the intelligent inspection and drying integrated device, to realize data storage and interaction.

9. An intelligent sun-drying device based on the intelligent sun-drying method for large-leaf tea according to any one of claims 1 to 6, which is installed on a sun-drying table and is characterized in that include: There are two sets of longitudinal sliding mechanisms, each set is arranged at the opposite sides of the drying platform, each set includes a longitudinal sliding rail (10), a longitudinal driving device (11) and a longitudinal transmission mechanism (12), and the longitudinal sliding block seat (13) is installed on the longitudinal sliding rail (10); the longitudinal driving device (11) can drive the longitudinal sliding block seat (13) to move synchronously along the longitudinal direction; There are two sets of vertical lifting mechanisms, each of which is arranged on a longitudinal sliding block seat (13) and can move with the longitudinal sliding block seat (13). Each set includes a vertically arranged lifting stroke mechanism (20), a lifting drive device (21) and a lifting block (24). The transverse control mechanism is installed between the two lifting blocks (24) and spans above the drying table. The two lifting drive devices (21) can operate synchronously to drive the transverse control rod to move horizontally. The transverse movement control mechanism includes a transverse movement slide rail (30), a transverse movement driving device (31) and a transverse movement transmission mechanism (32); the detection and drying integrated device (40) is installed on the transverse movement slide rail (30) and connected to the transverse movement transmission mechanism (32); the transverse movement driving device (31) can drive the detection and drying integrated device (40) to move transversely along the transverse movement slide rail; The detection and drying integrated device (40) comprises a moisture content detection device (41) and a drying and heating device (42). The moisture content detection device (41) is used to detect the moisture content of the tea material, and the drying and heating device (42) is used to dry the tea material to accelerate the reduction of its moisture content.

10. The drying device according to claim 9, characterized in that the longitudinal transmission mechanism (12) is a transmission belt (14), one end of which is connected to the motor output shaft (15) of the longitudinal drive device (11), and the other end is connected to the driven shaft (16); the longitudinal slider seat (13) is slidably installed on the longitudinal slide rail (10) and is fixedly connected to the transmission belt (14); the longitudinal drive device (11) can drive the transmission belt (14) to move by forward and reverse rotation, thereby driving the longitudinal slider seat (13) to move longitudinally along the longitudinal slide rail (10) synchronously; The lifting stroke mechanism (20) includes a lifting slide rail (22) and a lifting screw (23). The lifting screw (23) is connected to the motor of the lifting drive device (21). The lifting block (24) is sleeve-mounted on the lifting slide rail (22) and is threadedly mounted on the lifting screw (23). The lifting block (24) can be lifted and lowered along the lifting slide rail (22) as the lifting screw (23) rotates forward and backward. The two ends of the transverse slide rail (30) are fixedly connected to the inner sides of the two lifting blocks (24) and are kept parallel to the drying table. The transverse transmission mechanism (32) is a transverse screw rod, which is connected to the motor shaft of the transverse drive device (31). The detection and drying integrated device (40) includes a mounting sleeve (33). The mounting sleeve (33) is mounted on the transverse slide rail (30) and is fitted with the transverse screw thread. The mounting sleeve (33) can move transversely along the transverse slide rail (30) with the forward and reverse rotation of the transverse screw rod. The detection and drying integrated device (40) is mounted on the mounting sleeve (33). The detection and drying integrated device (40) comprises: A rotary drive motor (43), a bidirectional mounting plate (44), a moisture content detection device (41) and a drying and heating device (42); The rotary drive motor (43) is mounted on the mounting sleeve (33), and its rotary shaft is arranged horizontally along the longitudinal direction and extends horizontally along the longitudinal direction. The bidirectional mounting plate (44) is connected to the rotary shaft of the rotary drive motor (43), and the rotary drive motor (43) can drive the bidirectional mounting plate (44) to rotate along with the rotary shaft. The moisture content detection device (41) and the drying and heating device (42) are respectively mounted on both sides of the bidirectional mounting plate (44). The rotary drive motor (43) can rotate 180 degrees in reverse, or only rotate 180 degrees each time, so as to always keep the working surfaces of the moisture content detection device (41) and the drying and heating device (42) parallel to the drying table.

Images