Intelligent monitoring system for liupao tea production and method thereof

Abstract

The present application relates to product life cycle management technical field, specifically to a kind of intelligent monitoring system and method for Liubao tea production, system includes disturbance sensing module, hierarchical structure module, abnormality identification module, trend tracking module, closed-loop linkage module.In the present application, by introducing airflow disturbance in fermentation process stage and identifying directional hedging relationship based on the temperature and humidity response difference of nodes before and after disturbance, the disturbance activation position is screened by regional hedging trend, combined with thickness, density and boundary response data to divide three section structure zones, identify response lag area to realize local focus positioning, according to the direction change relationship of collection nodes, structural reverse conduction area is screened out, abnormal response area is identified to realize sudden change positioning, by comparing the difference between disturbance result and first response characteristics, whether the space response mode is mutated is identified, the proactivity of abnormal reaction detection, the pertinence and the effectiveness of intervention effect evaluation are improved.

Description

Technical Field

[0001] This invention relates to the field of product lifecycle management technology, and in particular to an intelligent monitoring system and method for Liubao tea production. Background Technology

[0002] Product lifecycle management (PWM) technology encompasses the information management and control of a product throughout its entire lifecycle, from conceptual design, R&D, pilot production, mass production, use, maintenance, to disposal. Core aspects include product structure management, process management, change management, document management, project management, and quality traceability. This technology aims to improve management efficiency and collaboration throughout the product lifecycle through information integration and data connectivity. Particularly in the manufacturing process, monitoring and recording process execution and control are crucial to ensuring product quality controllability and traceability. Traditional intelligent monitoring systems for Liubao tea production involve collecting and recording data on key process parameters such as temperature, humidity, and fermentation time during processing. This addresses the challenges of large fluctuations in environmental parameters and the difficulty in standardizing processes during Liubao tea processing. Traditional PWM systems use a network of temperature and humidity sensors to collect data on the tea-making environment, then transmit the data remotely to a backend management terminal for visualization and manual analysis. Alarm thresholds are manually set to trigger anomaly alerts.

[0003] Existing technologies rely on fixed networks of temperature and humidity sensors for parameter acquisition and remote data upload for display and recording. However, in the context of frequent environmental disturbances and complex spatial heat and humidity diffusion during the processing of Liubao tea, this approach cannot dynamically identify the evolution trend of abnormal temperature and humidity in local areas. Due to the lack of a local disturbance response mechanism, it is difficult to reveal the starting point and transmission path of abnormal development in the internal structure. Furthermore, relying on manually set alarm thresholds for anomaly alerts can easily lead to misjudgments or omissions in the early stages of temperature and humidity changes. For example, a sudden change in a micro-domain within a short period of time may be ignored if the overall data does not exceed the threshold. As a result, the monitoring effect is limited to surface parameter changes and lacks the ability to perceive the response characteristics of the internal structure, making it difficult to support accurate identification and dynamic control of abnormal states. Summary of the Invention

[0004] To address the shortcomings of existing technologies in handling Liubao tea processing scenarios characterized by frequent environmental disturbances and complex spatial heat and humidity diffusion, which prevent dynamic identification of localized temperature and humidity anomalies, lack of local disturbance response mechanisms, and inability to reveal the starting point and transmission path of anomalies within the internal structure, this invention provides an intelligent monitoring system and method for Liubao tea production. The technical solution is as follows:

[0005] On the one hand, an intelligent monitoring system for Liubao tea production is provided, which includes:

[0006] The disturbance sensing module acquires the distribution nodes of the evaporation layer and core layer regions during the fermentation stage of Liubao tea. A single airflow disturbance is implemented in the evaporation layer. The initial temperature and humidity values ​​of the nodes before the disturbance and the temperature rise and humidity drop within a unit response period after the disturbance are recorded. The direction of temperature and humidity change of adjacent nodes is compared to obtain the micro-domain disturbance response distribution map.

[0007] Based on the micro-domain disturbance response distribution map, the layered structure module calls the pile thickness variation data, unit volume density distribution data and boundary airflow disturbance response value to divide the pile into three segments: evapotranspiration layer, core layer and moisture accumulation layer, and obtains segment response difference block map;

[0008] The anomaly identification module extracts the temperature and humidity rate direction of the acquisition nodes in the multi-level data acquisition nodes during the disturbance period based on the hysteresis region marked in the segmented response difference block diagram, identifies whether there is a case where the direction of one layer is opposite to that of the upper and lower layers at the same time, and obtains the structural disturbance location layer.

[0009] The trend tracking module collects the number of airflow speed direction changes, the length of temperature rise segments, and the number of humidity drop segments within a time period based on the structural disturbance positioning layer. It then divides the continuous change type of the time window into three trend types: rising, oscillating, and stabilizing, and records them to obtain the behavioral trend change trajectory sequence.

[0010] As a further aspect of the present invention, the micro-domain disturbance response distribution map includes the distribution of temperature rise amplitude, humidity fall amplitude, directional opposition node pair distribution, and disturbance activation zone location; the segmented response difference block map includes evapotranspiration layer block, core layer block, moisture accumulation layer block, and disturbance response lag region distribution; the structural disturbance positioning layer includes directional opposition anomaly level, upper and lower opposition region location, and structural disturbance node distribution; and the behavioral trend change trajectory sequence includes temperature rise segment type, humidity drop segment type, airflow change frequency type, and trend switching record type.

[0011] As a further aspect of the present invention, the disturbance sensing module includes:

[0012] The temperature and humidity data acquisition submodule acquires the distribution nodes in the evaporation layer and core layer regions during the fermentation stage of Liubao tea. A single airflow disturbance is implemented at any node in the evaporation layer. The initial temperature and humidity values ​​of the distribution nodes before the disturbance are recorded. The temperature and humidity changes of the distribution nodes within a unit response period after the disturbance are recorded. The temperature rise and humidity fall of the disturbed nodes within a unit period are calculated, and the node disturbance temperature and humidity change is generated.

[0013] The node hedging screening submodule, based on the node disturbance temperature and humidity change, calls the temperature rise and humidity fall of the node within a unit response period, and compares the temperature change direction and humidity change direction of adjacent nodes according to the spatial distribution relationship between nodes. If there is a directional hedging, i.e., one node's temperature rises and its humidity falls, and the adjacent node's temperature falls and its humidity rises, the module filters the node combination pairs that meet the directional hedging relationship and generates a micro-domain directional hedging node pair set.

[0014] The disturbance response map generation submodule calls the micro-domain directional hedging node pair set, marks the region where the directional hedging node pair is located as the disturbance activation region according to the position of the node in the regional distribution, draws the boundary shape and position distribution of the disturbance activation region in spatial coordinates, and generates a micro-domain disturbance response distribution map.

[0015] As a further aspect of the present invention, the layered structure module includes:

[0016] The conduction delay calculation submodule calls the micro-domain disturbance response distribution map, combines the node positions within the segmented blocks, calculates the delay time from the application of boundary airflow disturbance to the start of node response based on the temperature rise and humidity drop values ​​of the nodes within a unit response period, and combines the disturbance timestamp, and calculates the average conduction hysteresis value by block aggregation, generating segmented block airflow conduction delay values;

[0017] The lag region filtering submodule sets the conduction delay benchmark value as the average lag value of the entire block based on the airflow conduction delay value of the segmented block, filters blocks with conduction delay values ​​greater than the benchmark value, extracts the spatial location in the three-segment stack partition identification matrix, and generates a set of disturbance response lag blocks.

[0018] The response difference map generation submodule calls the set of disturbance response hysteresis blocks, overlays the three-segment stack partition identification matrix and the segment block airflow conduction delay value, marks the stack segment type to which the hysteresis block belongs, and attaches the corresponding conduction delay magnitude parameter value to obtain the segment response difference block map.

[0019] As a further aspect of the present invention, the anomaly identification module includes:

[0020] The temperature and humidity rate extraction submodule, based on the segmented response difference block map, obtains the evaporation layer, core layer and moisture accumulation layer segments to which the hysteresis block belongs, extracts the temperature change value and humidity change value recorded by the acquisition node within the disturbance period, calculates the temperature change rate and humidity change rate per unit time, and determines the direction of temperature and humidity change based on the positive and negative values, and generates a multi-layer temperature and humidity rate direction set for the hysteresis block.

[0021] The multi-layer direction comparison submodule calls the multi-layer temperature and humidity rate direction set of the hysteresis block to make directional judgments on the nodes of the three levels in the hysteresis block, compares the temperature and humidity rate directions of the evaporation layer, the core layer and the moisture accumulation layer, identifies whether any layer is simultaneously opposite to the two adjacent layers above and below in both the temperature and humidity rate directions, marks the layer as an abnormal structure layer, and generates an abnormal direction reversal structure set.

[0022] The disturbance location map generation submodule calls the abnormal direction reversal structure set, combines the position index of the abnormal structure layer in the hysteresis block and the coordinate system of the three-segment stack partition identifier matrix, constructs the spatial structure correspondence, plots the spatial location of the abnormal structure layer and the attributes of the segment it is in, and obtains the structural disturbance location layer.

[0023] As a further aspect of the present invention, the trend tracking module includes:

[0024] The disturbance area feature acquisition submodule collects records of airflow velocity direction changes, temperature value sequences, and humidity value sequences of corresponding regional nodes within a specified time period based on the spatial location of the abnormal structural layer marked in the structural disturbance positioning layer. It also counts the number of airflow velocity direction changes, the length of the continuous temperature rise segment, and the number of humidity drop segments within the time period, generating three types of change feature value sets for the disturbance area.

[0025] The continuous change judgment submodule calls the three types of change feature value sets of the disturbance area, arranges the three items of airflow speed direction change number, temperature rise segment length and humidity drop segment number in time order, and judges whether any two items have exceeded the double cycle standard in a continuous manner, that is, the two values ​​exceed the set change threshold value in two consecutive monitoring cycles. If there is a data segment that meets the conditions, it is marked as a continuous abnormal segment and a double cycle continuous abnormal identification segment is generated.

[0026] The monitoring cycle adjustment submodule resets the monitoring cycle time window length within the time period based on the time boundary of the dual-cycle continuous anomaly identification segment, expands the cycle duration forward and backward on both sides of the continuous anomaly segment to construct a new time window, updates the original cycle division method, and re-indexes the position distribution of data segments within the adjusted cycle to generate a dynamic monitoring time window index set.

[0027] The trend trajectory generation submodule calls the dynamic monitoring time window index set, sequentially reads the numerical trends of the three types of change characteristic values ​​in the disturbance area within the time window, and classifies the time window type into three trend types—rising, oscillating, or stabilizing—based on the fluctuation amplitude of the number of changes in airflow speed direction, the gradient of the change in the length of the temperature rise segment, and the periodic repetition rate of the number of humidity drop segments. The module then records the trend type change path in chronological order to obtain the behavioral trend change trajectory sequence.

[0028] As a further aspect of the present invention, the statistical process for the number of airflow velocity direction changes is as follows: in each monitoring cycle, at a time interval of 10 minutes, it is recorded whether the airflow direction change occurs. If there is an angular shift of more than 45 degrees in the direction between two adjacent time intervals, it is considered as one change.

[0029] The statistical process for the length of the continuous temperature rise segment is as follows: the segment with more than 3 consecutive monitoring points in the monotonically increasing sequence is taken as the rise segment, and the cumulative duration in each monitoring cycle is calculated.

[0030] The statistical process for the number of humidity drop segments is as follows: within each monitoring cycle, based on the gradient formed by three adjacent monitoring points in the sequence, identify drop segments that meet the condition of an average drop of more than 5% and a duration of not less than 30 minutes, and use the number of drop segments that meet the condition as the statistical value of the number of humidity drop segments.

[0031] As a further aspect of the present invention, the system also includes a closed-loop linkage module:

[0032] The closed-loop linkage module extracts the continuous switching records of oscillation trend and upward trend based on the behavioral trend change trajectory sequence, locates the time nodes marked as unstable behavior sudden segments, re-performs perturbation actions on the spatial region within the time period, identifies the magnitude of temperature and humidity response changes after the disturbance, compares the characteristics of the first disturbance response, identifies whether there are changes in the spatial response pattern, and obtains the fermentation response mutation distribution area.

[0033] The fermentation response mutation distribution area includes spatial response pattern change points, unstable behavior sudden segment regions, and repeated perturbation response comparison difference areas.

[0034] As a further aspect of the present invention, the closed-loop linkage module includes:

[0035] The sudden segment identification submodule extracts the position index of the continuous switching between oscillation trend and upward trend in the time series based on the behavior trend change trajectory sequence, locates the time nodes where the trends alternate and marks them as unstable behavior sudden segments, identifies the start and end time indexes and corresponding spatial area locations covered by the sudden segment, and generates an unstable trend sudden segment identifier set.

[0036] The disturbance retest execution submodule calls the unstable trend sudden segment identifier set, reapplies a single airflow disturbance action in the spatial region corresponding to the identified time node, records the initial temperature and humidity values ​​of the nodes before and after the disturbance, and the temperature and humidity changes within a unit response period, calculates the temperature and humidity change amplitudes of each node after the disturbance, and generates a re-disturbance temperature and humidity response amplitude set.

[0037] The response mutation identification submodule, based on the re-perturbation temperature and humidity response amplitude set, calls the temperature change amplitude and humidity change amplitude values ​​recorded by the node in the first perturbation operation, calculates the difference between the two perturbation response amplitudes, compares whether the difference exceeds the set spatial response change threshold, identifies the positional relationship in the spatial distribution, records the heap coordinate range corresponding to the node, and obtains the fermentation response mutation distribution area.

[0038] On the other hand, a method for intelligent monitoring of Liubao tea production, which is based on the aforementioned intelligent monitoring system for Liubao tea production, includes the following steps:

[0039] S1: Obtain the distribution nodes of the evaporation layer and core layer regions in the fermentation stage of Liubao tea, record the initial temperature and humidity values ​​of the nodes before airflow disturbance, activate the airflow disturbance actuator once in the node group, call the temperature rise and humidity drop within a unit time period after disturbance, calculate the difference in the direction of temperature and humidity change between adjacent nodes, and generate a micro-domain disturbance response distribution map.

[0040] S2: Based on the micro-domain disturbance response distribution map, obtain the thickness range values, unit volume density values, and boundary airflow disturbance response values ​​of the evaporation layer, core layer, and moisture accumulation layer in the cross-section of the stack body. Call the evaporation layer thickness value and unit density value to calculate the airflow conduction delay time value in the multi-segment region and generate a segmented response difference block map.

[0041] S3: Based on the segmented response difference block diagram, extract the direction of temperature and humidity rate change of the acquisition nodes in the three layers of evaporation layer, core layer and moisture accumulation layer during the disturbance period, and obtain the structural disturbance positioning layer by comparing the consistency of the direction changes of the three layers under the same vertical projection line of the upper and lower layers.

[0042] S4: Based on the structural disturbance positioning layer, obtain the number of times the airflow velocity direction changes, the length of the temperature rise segment, and the number of humidity drop segments of the node within the time period after the disturbance. Determine whether any two items occur consecutively for more than two disturbance cycle thresholds. If the condition is met, adjust the time window length and generate a behavioral trend change trajectory sequence.

[0043] S5: Call the behavior trend change trajectory sequence, combine it with the time segment that is classified as a continuous switching between oscillation trend and upward trend, collect the temperature change amplitude value and humidity change amplitude value within a unit period after the disturbance, filter the area where the deviation ratio exceeds the disturbance response change ratio threshold, and generate the fermentation response mutation distribution area.

[0044] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0045] By introducing airflow disturbances during the fermentation process and identifying directional counterbalancing relationships based on the temperature and humidity response differences before and after the disturbance, the method filters out disturbance activation locations through regional counterbalancing trends. It then divides the system into three structural zones by combining thickness, density, and boundary response data, identifies lag regions to pinpoint local concerns, and identifies structural reverse transmission zones based on the directional changes of the data collection nodes. Furthermore, it constructs trend trajectory sequences by combining the rhythms of airflow, temperature, and humidity changes over continuous periods, linking trend switching characteristics with spatial behavior feedback to identify anomalous response areas and pinpoint sudden changes. Finally, it identifies whether spatial response patterns have abruptly changed by comparing the differences between the disturbance results and the initial response characteristics. The achieved results include refined perception and trend evolution monitoring of microenvironmental changes during the Liubao tea fermentation process, enabling further intervention and result comparison for abnormal trend states. This enhances the ability to identify and control the fermentation state, improving the foresight, specificity, and effectiveness of abnormal reaction detection and intervention evaluation. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a schematic diagram of an intelligent monitoring system for Liubao tea production provided in an embodiment of the present invention;

[0048] Figure 2 This is a schematic diagram of the system framework of the present invention;

[0049] Figure 3 This is a flowchart of the disturbance sensing module in this invention;

[0050] Figure 4 This is a flowchart of the hierarchical structure module in this invention;

[0051] Figure 5 This is a flowchart of the anomaly identification module in this invention;

[0052] Figure 6 This is a flowchart of the trend tracking module in this invention;

[0053] Figure 7 This is a flowchart of the closed-loop linkage module in this invention;

[0054] Figure 8 This is a flowchart of an intelligent monitoring method for Liubao tea production provided in an embodiment of the present invention. Detailed Implementation

[0055] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0056] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0057] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0058] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0059] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0060] This invention provides an intelligent monitoring system for Liubao tea production, such as... Figure 1-2 The diagram shown illustrates the intelligent monitoring system for Liubao tea production. This system includes:

[0061] The disturbance sensing module acquires the distribution nodes of the evaporation layer and core layer regions during the fermentation stage of Liubao tea. A single airflow disturbance is implemented in the evaporation layer. The initial values ​​of temperature and humidity of the nodes before the disturbance and the temperature rise and humidity drop within a unit response period after the disturbance are recorded. The direction of temperature and humidity change of adjacent nodes is compared, and node pairs with opposing trends are selected as micro-domain anomalies. The regions of adjacent nodes with opposing directions are marked as disturbance activation areas, and the micro-domain disturbance response distribution map is obtained.

[0062] The layered structure module, based on the micro-domain disturbance response distribution map, calls the stack thickness variation data, unit volume density distribution data, and boundary airflow disturbance response value to divide the stack into three segments: evapotranspiration layer, core layer, and moisture accumulation layer. Based on the airflow conduction delay value within the segmented blocks, the disturbance response lag area is selected as the area of ​​interest to obtain the segmented response difference block map.

[0063] The anomaly identification module extracts the temperature and humidity rate directions of the acquisition nodes in the multi-level data collection nodes within the disturbance period based on the hysteresis regions marked in the segmented response difference block diagram, compares the consistency of the multi-level directions, identifies whether there is a situation where the direction of one layer is simultaneously opposite to that of the upper and lower layers, and obtains the structural disturbance location layer.

[0064] The trend tracking module locates the structural disturbance layer, collects the number of changes in airflow velocity direction, the length of temperature rise segments, and the number of humidity drop segments within a time period, and determines whether any two consecutive occurrences exceed the double-cycle standard. If the condition is met, the monitoring cycle time window length is adjusted, and the continuous change type of the time window is divided into three trend types: rising, oscillating, and stabilizing, and recorded to obtain the behavioral trend change trajectory sequence.

[0065] The closed-loop linkage module extracts the continuous switching records of oscillation trend and upward trend based on the behavior trend change trajectory sequence, locates the time nodes marked as unstable behavior sudden segment, re-performs the perturbation action on the spatial area within the time period, identifies the change amplitude of temperature and humidity response after the disturbance, compares the characteristics of the first disturbance response, identifies whether there is a change in spatial response mode, and obtains the fermentation response mutation distribution area.

[0066] The micro-domain disturbance response distribution map includes the distribution of temperature rise amplitude, humidity fall amplitude, directional opposition node pair distribution, and disturbance activation zone location. The segmented response difference block map includes the evapotranspiration layer block, core layer block, moisture accumulation layer block, and disturbance response lag region distribution. The structural disturbance location layer includes the directional opposition anomaly level, upper and lower opposition region location, and structural disturbance node distribution. The behavioral trend change trajectory sequence includes the temperature rise segment type, humidity drop segment type, airflow change frequency type, and trend switching record type. The fermentation response mutation distribution area includes spatial response pattern change points, unstable behavior sudden segment region, and repeated disturbance response comparison difference area.

[0067] Specifically, such as Figure 2 , 3 As shown, the disturbance sensing module includes:

[0068] The temperature and humidity data acquisition submodule acquires the distribution nodes in the evaporation layer and core layer regions during the fermentation stage of Liubao tea. A single airflow disturbance is implemented at any node in the evaporation layer. The initial temperature and humidity values ​​of the distribution nodes before the disturbance are recorded. The temperature and humidity changes of the distribution nodes within a unit response period after the disturbance are recorded. The temperature rise and humidity fall of the disturbed nodes within a unit period are calculated, and the node disturbance temperature and humidity change is generated.

[0069] By deploying a network of temperature and humidity sensors evenly distributed throughout the fermentation environment, the evapotranspiration layer and core layer are divided into multiple micro-regions. Multiple temperature and humidity sensors are deployed in each region, ensuring a node spacing of less than 10cm to obtain sufficiently dense data coverage. A specific node in the evapotranspiration layer is selected as the interference source node, and a single airflow disturbance is implemented. The disturbance is achieved by controlling a directional airflow with a wind speed of 1.5m / s, and the disturbance duration is set to 60 seconds. The control system records the initial temperature T0 and initial humidity H0 of each distributed node before the disturbance. Data changes within a unit response period after the disturbance are collected, with the response period set to 300 seconds. During this period, the temperature and humidity changes T(t) and H(t) of each node are recorded every 10 seconds. The temperature rise ΔT = T(t) of each node is calculated. max -T0, the decrease in humidity ΔH = H0 - H(t) min ), where t max With t min To determine the times when the maximum temperature and minimum humidity occur within the response period, if the temperature at node A rises from 28.5℃ to 32.2℃ and the humidity decreases from 82% to 76% during the disturbance period, then ΔT A =3.7℃, ΔH A =6%, and in this way, ΔT and ΔH of the distributed nodes are obtained to generate the node disturbance temperature and humidity change.

[0070] The node hedging screening submodule, based on the node's temperature and humidity change during the unit response period, calls the node's temperature rise and humidity fall within the unit response period. According to the spatial distribution relationship between nodes, it compares whether there is directional hedging between the temperature and humidity change directions of adjacent nodes. That is, if a node's temperature rises and its humidity falls, and the adjacent node's temperature falls and its humidity rises, it filters out node pairs that meet the directional hedging relationship and generates a micro-domain directional hedging node pair set.

[0071] A node topology graph is constructed through spatial modeling, mapping nodes to spatial coordinates (x, y, z) according to their actual positions. The condition for determining adjacent nodes is that the Euclidean distance d ≤ 15cm. All node pairs are traversed. For each pair of adjacent nodes (i, j), the functions sgn(ΔTi) and sgn(ΔTj) for temperature change direction and sgn(ΔHi) and sgn(ΔHj) for humidity change direction are called. The sgn function is defined as follows: if the variable is positive, it returns +1; if negative, -1; and if zero, 0. The function checks if the directional opposition condition is met, i.e., ΔTi > 0 ∧ ΔHi < 0 ∧ ΔTj < 0 ∧ ΔHj > 0, or if the mirror condition is met. Taking nodes A and B as an example, if ΔTA = +3.2℃, ΔHA = -5%, ΔTB = -1.1℃, and ΔHB = +3%, then the node pair meets the directional opposition condition. After filtering, the node pair (A, B) is recorded. To avoid misjudgment and introduce interference thresholds, a temperature change threshold ΔT is set. th =1.0℃, humidity change threshold ΔH th =2.0%, and is only included in the judgment range when the absolute values ​​of ΔT and ΔH both exceed the threshold; generate a set of micro-domain directional hedging nodes.

[0072] The disturbance response map generation submodule calls the micro-domain directional hedging node pair set. Based on the position of the node in the regional distribution, the region where the directional hedging node pair is located is marked as the disturbance activation region. The boundary shape and position distribution of the disturbance activation region are drawn in spatial coordinates to generate the micro-domain disturbance response distribution map.

[0073] The spatial location of the node pair is analyzed and mapped to the actual spatial model inside the fermentation chamber based on the 3D coordinates. The fermentation chamber is divided into 30×30×10 grid units using a grid method, with each grid having a side length of 10cm. The grid region containing the node pair is marked as the interference activation area. The boundary is delineated using the α-shape algorithm. The outline of the interference area is formed by connecting grids that meet the threshold, and the boundary point set of the activation area is obtained. Then, closed boundary curves are generated using Bezier interpolation or linear interpolation. Taking the node pair (A, B) as an example, A is located at... (10cm, 20cm, 10cm), B is located at (20cm, 20cm, 10cm), and the regions where they are located are merged to form a long strip of interference activation band; the activation regions of the opposite direction node pairs are superimposed in sequence, and the activation intensity is displayed by color mapping. The color depth is normalized according to the product of the node pair disturbance intensity ΔT and ΔH. For example, if the disturbance pair (A, B) has ΔT×ΔH=3.2×6=19.2, then a color value of 0.82 is assigned (normalized to the 0~1 interval), and a micro-domain disturbance response distribution map is obtained.

[0074] Specifically, such as Figure 2 , 4 As shown, the hierarchical structure module includes:

[0075] The conduction delay calculation submodule calls the micro-domain disturbance response distribution map, combines the node positions within the segmented blocks, calculates the delay time from the application of the boundary airflow disturbance to the start of the node response based on the temperature rise and humidity drop values ​​of the nodes within a unit response period, and combines the disturbance timestamp, and calculates the average conduction hysteresis value by block aggregation, generating the segmented block airflow conduction delay value;

[0076] Based on the perturbation activation regions marked in the diagram, the system identifies the sensor nodes included in each segment block. The system matches the actual spatial coordinates of the nodes with their regional locations and assigns them to the corresponding segment block number. After obtaining the specific location of each node, the system begins to analyze the temperature and humidity response behavior exhibited by the nodes after the perturbation begins, focusing on the dynamic process of temperature rise and humidity decrease within a unit response period. For each node, the system extracts the initial temperature and humidity values ​​and the time-varying sequence data, comparing when the node data begins to show a temperature rise after the perturbation is applied. The response start time can be determined by observing the decreasing humidity trend, and the time when the disturbing airflow is first applied can be recorded. By combining the two time points, the delay time between the disturbance and the response of each node can be obtained. After determining the conduction delay of each node, the nodes are classified and aggregated into segments. The average conduction lag value of the block is calculated by summarizing the delay data of the nodes within the block. For example, if a block contains five nodes and the recorded delay values ​​are 45 seconds, 48 ​​seconds, 51 seconds, 49 seconds and 47 seconds respectively, then the average delay of the block is 48 seconds, and the segment block airflow conduction delay value is generated.

[0077] The lag region filtering submodule sets the conduction delay benchmark value as the average lag value of the entire block based on the airflow conduction delay value of the segmented blocks, filters blocks with conduction delay values ​​greater than the benchmark value, extracts the spatial location in the three-segment stack partition identification matrix, and generates a set of disturbance response lag blocks.

[0078] The system summarizes and statistically analyzes the delay values ​​of each block. By traversing the average delay time recorded in each block, it calculates the overall average lag level of airflow disturbance transmission in the entire fermentation area. This value serves as a benchmark for subsequent screening processes. After the benchmark value is established, the system compares the lag value of each block with the benchmark to determine whether it exhibits a delay higher than the overall average level. Blocks with lag values ​​exceeding the benchmark are considered slow-responding areas, screened, and added to the lag block set. To eliminate interference caused by boundary errors or abnormal fluctuations, the system also sets a lag judgment buffer. Only when the lag time exceeds the benchmark value by a certain range will the system determine it as a valid lag block, which can improve the accuracy of data screening. After screening out the lag blocks, the system also calls the three-segment stack partition identification matrix to mark and map the spatial location of each lag block in the matrix, clarifying the specific stack segment location to which each block belongs. For example, if a block is located in the upper middle section of the fermentation stack, its location will be marked as "middle section - upper layer" in the lag block set, generating a disturbance response lag block set.

[0079] The response difference map generation submodule calls the set of disturbance response lag blocks, overlays the three-segment stack partition identification matrix and the segment block airflow conduction delay value, marks the stack segment type to which the lag block belongs, and attaches the corresponding conduction delay magnitude parameter value to obtain the segment response difference block map.

[0080] The system performs graphic composition, loading a three-segment stack partition identifier matrix and aligning it with the lag block set. Through index matching, the position of each lag block is associated with its stack type. The system finds its stack segment based on the block's index value in the stack matrix; for example, a block located in the first 1 / 3 of the matrix is ​​classified as the front segment, the middle 1 / 3 as the middle segment, and so on. After classifying the lag blocks, the system outputs the airflow conduction delay values ​​recorded for each block along with its segment type, forming a preliminary data annotation map. In visualization, the system maps the three-dimensional stack structure onto a two-dimensional graphic coordinate system, using color blocks. The system displays the areas where lagging blocks are located, distinguishing the degree of delay by the intensity of color. For example, the longer the delay, the darker the color. The system also adds text labels to each color block, displaying the block number, the segment it belongs to, and the delay value, such as "Block K21, front segment, delay 62 seconds". The overall image uses a grid structure to overlay the stack cross-section diagram, ensuring that the location of each lagging block is referential in actual space. The image also includes a legend area to explain the delay level range represented by each color and to indicate the boundary division of the three stack segments, determining which areas should optimize airflow application strategies or adjust the stack structure layout, and obtaining segmented response difference block maps.

[0081] Specifically, such as Figure 2 , 5 As shown, the anomaly detection module includes:

[0082] The temperature and humidity rate extraction submodule is based on the segmented response difference block map to obtain the evaporation layer, core layer and moisture accumulation layer segments to which the hysteresis block belongs. It extracts the temperature change value and humidity change value recorded by the acquisition node within the disturbance period, calculates the temperature change rate and humidity change rate per unit time, and determines the direction of temperature and humidity change based on the positive and negative values ​​to generate a multi-layer temperature and humidity rate direction set of the hysteresis block.

[0083] The system identifies the physical stack segment locations corresponding to the lagging blocks marked in the diagram. Combining this with the layer classification information of each block in three-dimensional space, it categorizes them into three regions: evapotranspiration, core, or moisture accumulation. For lagging blocks, the system locates all data acquisition nodes under the block and retrieves temperature and humidity records during the airflow disturbance cycle. Particular attention is paid to the continuous data sequence recorded by nodes from the start of the disturbance to the end of the response. The system calculates the rate of temperature change per unit time, i.e., the speed at which temperature rises or falls over time, based on the data. Simultaneously, the humidity change process is extracted, and the humidity change is calculated. The rate of change is calculated, for example, if the temperature of a node rises from 30.1℃ to 31.5℃ within 20 seconds after a disturbance, the average temperature change rate is 0.07℃ / s, and the humidity drops from 85% to 80%, with a humidity change rate of 0.25% / s. By analyzing the positive and negative signs of the temperature and humidity rates, the direction of temperature and humidity change is determined. If the temperature rate is positive and the humidity rate is negative, the direction of change for the node is the direction of temperature increase and humidity decrease. The system marks the direction of change rate for each node and clusters and classifies them according to the layer segment in which the node is located, generating a multi-layer temperature and humidity rate direction set for the lag block.

[0084] The multi-layer direction comparison submodule calls the multi-layer temperature and humidity rate direction set of the hysteresis block to make directional judgments on the nodes of the three levels in the hysteresis block, compares the temperature and humidity rate directions of the evaporation layer, the core layer and the moisture accumulation layer, identifies whether any layer is simultaneously opposite to the two adjacent layers above and below in both the temperature and humidity rate directions, marks the layer as an abnormal structure layer, and generates an abnormal direction reversal structure set.

[0085] A systematic analysis was conducted on node data belonging to the evapotranspiration layer, core layer, and moisture accumulation layer within the lag block. Nodes at each of the three levels in each block were extracted and their directions were summarized. The dominant trends in temperature and humidity rates within each layer were calculated, i.e., the direction of temperature and humidity change for the majority of nodes was statistically analyzed. If more than 50% of the nodes exhibited the same trend, this trend was taken as the dominant direction. For example, in the core layer of a certain block, if 5 out of 7 nodes showed a decreasing temperature and increasing humidity trend, then the dominant direction of temperature and humidity rates was marked as "decreasing temperature and increasing humidity". The system compares the main directional relationships between the evaporation layer and the core layer, and between the core layer and the moisture accumulation layer. If the core layer is opposite to the upper and lower layers in both dimensions, that is, the direction of temperature change and the direction of humidity change are completely reversed, the core layer is determined to be an abnormal structure layer. The comparison logic also applies to the case where the evaporation layer or moisture accumulation layer is located in the middle layer. The system will mark the middle layer segment that satisfies the two-dimensional direction reversal relationship as the structure reversal point, and record the layer segment position and associated node in the hysteresis block. The abnormal structure layers that meet the conditions are summarized to generate an abnormal direction reversal structure set.

[0086] The disturbance location map generation submodule calls the abnormal direction reverse structure set, combines the position index of the abnormal structure layer in the hysteresis block and the coordinate system of the three-segment stack partition identifier matrix, constructs the spatial structure correspondence, plots the spatial location and layer attributes of the abnormal structure layer, and obtains the structural disturbance location layer.

[0087] The system reconstructs the location of each recorded anomalous structural layer. Based on the number of the lag block and the coordinate system of the three-segment stack partition identification matrix, the system can accurately locate the spatial position of the layer in the stack structure and confirm which specific segment and level it belongs to. For example, if an anomalous layer is recorded in block K041 and is located in the core layer of the middle section of the stack, the system will locate the corresponding position of the core layer of the middle section of the stack structure in the figure. The system constructs a layered structural model of the stack space and, based on the position coordinates of the anomalous structural layers, superimposes them one by one onto the stack structure profile. Each anomalous layer is highlighted with a separate color and accompanied by text labels showing the block number, stack segment, and inversion feature type. At the same time, the legend area explains the correspondence between the multi-segment attributes and color markings, so that each highlighted block in the figure can be clearly identified. For example, the inversion area of ​​the front wet layer can be marked in red, the inversion area of ​​the core layer can be marked in blue, and the evapotranspiration layer can be marked in green, thus obtaining the structural disturbance location layer.

[0088] Specifically, such as Figure 2 , 6 As shown, the trend tracking module includes:

[0089] The disturbance area feature acquisition submodule collects records of airflow velocity direction changes, temperature value sequences, and humidity value sequences of corresponding regional nodes within a specified time period based on the spatial location of the abnormal structure layer marked in the structural disturbance location layer. It also counts the number of airflow velocity direction changes, the length of the continuous temperature rise segment, and the number of the humidity drop segment within the time period, generating three types of change feature value sets for the disturbance area.

[0090] By matching the corresponding spatial node set within the stack body using layer coordinate information, the specific sensor node number and layout position under each abnormal layer segment are identified. A specified analysis time period is selected in the system time series, such as a 30-minute data window after airflow disturbance. Raw data on changes in airflow velocity direction, temperature, and humidity recorded by the aforementioned nodes within this time range are extracted and organized into data sequence sets by node. The records of changes in airflow velocity direction are analyzed, and the number of times the direction changes within the time period is counted, i.e., the number of sudden changes from one direction to another. For example, if the wind direction alternates multiple times between southeast and northwest, the number of changes is the cumulative number of switching events. For temperature change data... The system scans the entire data sequence, identifies and counts the length of intervals where the temperature rises continuously. Specifically, it detects the length of data segments that rise continuously from a low point to a local high point, reflecting the duration of the temperature increase. Regarding humidity changes, the system identifies the number of data segments where humidity shows a significant downward trend within the same time period. The criterion for this is that the humidity drops by more than 3% in a short period of time, which constitutes a sudden drop segment. Once this condition is met, the segment is numbered and counted. Each abnormal structural layer forms a set of three types of change characteristic values ​​within a specified time period: the number of airflow direction changes, the total length of the continuous temperature rise segment, and the number of sudden humidity drop segments, generating a set of three types of change characteristic values ​​for the disturbance region.

[0091] The continuous change judgment submodule calls the three types of change feature value sets of the disturbance area, arranges the three items of airflow speed direction change number, temperature rise segment length and humidity drop segment number in time order, and judges whether any two items have exceeded the double cycle standard in a continuous manner, that is, the two values ​​exceed the set change threshold value in two consecutive monitoring cycles. If there is a data segment that meets the conditions, it is marked as a continuous abnormal segment and a double cycle continuous abnormal identification segment is generated.

[0092] Each set of feature values ​​is arranged in chronological order according to the time series and organized into multiple monitoring period units of equal length. Each period represents the characteristic behavior within a fixed time window, for example, every 10 minutes is a period. The number of airflow changes, the length of the temperature rise segment, and the number of humidity drop segments in each period are listed in a comparison table. A comparative analysis is performed between every two adjacent periods. The system checks whether any two feature values ​​exceed their respective set critical change standards in two consecutive periods. For example, the number of airflow direction changes is greater than 5 times in two consecutive periods, the length of the temperature rise segment is greater than 120 seconds in two consecutive periods, and the number of humidity drop segments is greater than 2 in two consecutive periods. Once any two of the above conditions are met, the system marks the time segment as a continuous abnormal segment. During the marking process, the system will scroll the period sliding window to ensure that no short-term but continuous change behavior is missed. The start and end times of the continuous abnormal segment, as well as the specific abnormal structural layer segment position and period number, will be recorded to generate a dual-period continuous abnormal identification segment.

[0093] The monitoring cycle adjustment submodule resets the monitoring cycle time window length within the time period based on the time boundary of the dual-cycle continuous anomaly identification segment, expands the cycle duration forward and backward on both sides of the continuous anomaly segment to construct a new time window, updates the original cycle division method, and re-indexes the position distribution of data segments within the adjusted cycle, generating a dynamic monitoring time window index set.

[0094] The system initiates a process to reset the time window length. For each identified anomalous segment, the system automatically calculates the start and end times, extending outwards by a fixed period before and after each segment. For example, it extends by one period unit on each side. If the original period unit is 10 minutes, then it extends by 10 minutes before and after each segment, enclosing the continuous anomalous segments within a larger new time window. This new time window is used to capture the transitional changes before and after the anomalous segment. The system arranges the newly generated time windows in chronological order, updates the original period numbers and segmentation methods, replaces the old time series distribution, rescans the original recorded data of the entire disturbance area, and reassigns each data point to the newly defined monitoring time window, forming a new data index structure within the time window. That is, the node values, start and end times, behavioral characteristics, and period numbers corresponding to each new period are all reclassified and updated. This process ensures that the background changes before and after the anomalous segment are synchronously included in the analysis, generating a dynamic monitoring time window index set.

[0095] The trend trajectory generation submodule calls the dynamic monitoring time window index set, reads the numerical trend of the three types of change characteristic values ​​in the disturbance area within the time window in turn, and classifies the time window type into three trend types of rising, oscillating or stabilizing based on the fluctuation amplitude of the number of changes in airflow speed direction, the gradient of the change in the length of the temperature rise segment and the periodic repetition rate of the number of humidity drop segments. The trend type change path is recorded in chronological order to obtain the behavioral trend change trajectory sequence.

[0096] The system reads the dynamic trends of three types of characteristic values ​​in the disturbance region within each time window. It evaluates the fluctuation of the number of changes in airflow velocity and direction within each time window. If the fluctuation range continues to increase, the trend is judged to be "rising". If there are frequent fluctuations but the overall range does not expand significantly, it is characterized as "oscillation". If the fluctuation amplitude gradually weakens and tends to stabilize, it is judged as "stabilizing". The same method is applied to the trend judgment of the length of the temperature rise segment and the number of humidity drop segments. The system analyzes the slope and repetition characteristics of the indicators in multiple consecutive time windows by setting comparison standards, forming a trend label for each time window, marked as "rising", "oscillation" or "stabilizing". The trend type of the time window is recorded in sequence and sorted by time. For example, if an abnormal structural layer shows "rising" in the first cycle after the disturbance, "oscillation" in the second and third cycles, and "stabilizing" in the fourth cycle, then the trend trajectory is "rising → oscillation → stabilizing". The system will attach the trajectory to the corresponding structural layer segment and time index to obtain the behavioral trend change trajectory sequence.

[0097] Specifically, such as Figure 2 , 7 As shown, the closed-loop linkage module includes:

[0098] The sudden segment identification submodule extracts the location index of the continuous switching between oscillating trend and upward trend in the time series based on the behavioral trend change trajectory sequence, locates the time nodes where the trends alternate and marks them as unstable behavioral sudden segments, identifies the start and end time indexes and corresponding spatial area locations covered by the sudden segment, and generates a set of unstable trend sudden segment identifiers.

[0099] By analyzing the trend type corresponding to each time window in the trend sequence, the system identifies key points where "oscillation" and "upward" trends alternate. Specifically, it scans from the beginning of the sequence, reading the trend markers window by window. Once a time window is found to be "oscillating" while the adjacent windows are "upward," or if the window itself is "upward" while the windows before and after it are "oscillating," it is considered a trend switching point. The system records the index of the time window as a trend change node and locates the corresponding actual time point. If multiple trend alternation points occur densely in the same time period, the system merges the consecutive alternation nodes into a sudden segment, recording the start time of the first switching time window as the starting point and the end time of the last switching window as the ending point, thus determining the complete time range of the unstable behavior sudden segment. The system combines the spatial coordinate information of each time window in the trend trajectory to extract the spatial location of the nodes involved in the sudden segment, forming a three-dimensional stacked index set of the covered area, and generating a set of unstable trend sudden segment identifiers.

[0100] The disturbance retest execution submodule calls the unstable trend sudden segment identifier set, re-applies a single airflow disturbance action in the spatial region corresponding to the identified time node, records the initial temperature and humidity values ​​of the nodes before and after the disturbance, and the temperature and humidity changes within a unit response period, calculates the temperature and humidity change amplitudes of each node after the disturbance, and generates a re-disturbance temperature and humidity response amplitude set.

[0101] The system retrieves the time index and spatial coordinates corresponding to the sudden segment. Within the region, it reactivates the airflow disturbance device and applies a single disturbance operation to the corresponding node of the sudden segment. This ensures that the disturbance wind speed, direction, and duration are consistent with the first disturbance conditions to maintain comparability. Before the disturbance is applied, the system records the initial temperature and humidity values ​​of the node and continuously samples within a unit response cycle to collect the complete temperature and humidity change process. The response cycle is set to 300 seconds to capture the entire node response process, with a sampling frequency of once every 5 seconds. After the disturbance, the system analyzes the temperature and humidity data sequences of each node, extracting the maximum temperature change and humidity change, which are represented as the temperature and humidity response amplitudes under the re-disturbance conditions. The system also calculates the total temperature increase and humidity decrease for each node. For example, if the initial temperature of a node is 29.0℃ and rises to a maximum of 33.5℃ after the disturbance, the temperature change amplitude is 4.5℃, and the humidity decreases from 83% to 77%, with a humidity change amplitude of 6%. The re-disturbance data of the involved nodes are then combined into a complete dataset to generate a re-disturbance temperature and humidity response amplitude set.

[0102] The response mutation identification submodule, based on the re-perturbation temperature and humidity response amplitude set, calls the temperature change amplitude and humidity change amplitude values ​​recorded by the node in the first perturbation operation, calculates the difference between the two perturbation response amplitudes, compares whether the difference exceeds the set spatial response change threshold, identifies the positional relationship in the spatial distribution, records the heap coordinate range corresponding to the node, and obtains the fermentation response mutation distribution area.

[0103] The system calls the nodes to record temperature and humidity changes during the initial perturbation test and matches them one by one with the new values ​​obtained during the subsequent perturbation. The difference between the temperature and humidity changes between the two perturbations is obtained by subtraction. For example, if the initial temperature change is 3.2℃ and the subsequent perturbation is 5.0℃, the difference is 1.8℃. The system sets a spatial response change threshold as a judgment standard. If the temperature difference exceeds 1.5℃ or the humidity difference exceeds 3.0%, it is judged as a response mutation point. The system screens the nodes one by one, marks the nodes whose response change amplitude increases or decreases significantly after the perturbation, and statistically analyzes the spatial distribution pattern of the nodes. The positions of the nodes in the three-dimensional coordinate system of the stack are summarized. For example, some response mutation points are concentrated in the core layer in the middle of the stack, while others are distributed in the front evaporation layer. The system organizes the positions of the nodes judged as response mutations into a coordinate set, aggregates and outputs the boundaries of the mutation blocks by region, and forms a fermentation response mutation distribution area layer.

[0104] Please see Figure 8 The intelligent monitoring method for Liubao tea production is based on the aforementioned intelligent monitoring system for Liubao tea production and includes the following steps:

[0105] S1: Obtain the distribution nodes of the evaporation layer and core layer regions in the fermentation stage of Liubao tea, record the initial temperature and humidity values ​​of the nodes before airflow disturbance, activate the airflow disturbance actuator once in the node group, call the temperature rise and humidity drop within a unit time period after disturbance, calculate the difference in the direction of temperature and humidity change between adjacent nodes, and generate a micro-domain disturbance response distribution map.

[0106] S2: Based on the micro-domain disturbance response distribution map, obtain the thickness range values, unit volume density values, and boundary airflow disturbance response values ​​of the evaporation layer, core layer, and moisture accumulation layer in the cross-section of the stack. Call the evaporation layer thickness value and unit density value to calculate the airflow conduction delay time value in the multi-segment region and generate a segmented response difference block map.

[0107] S3: Based on the segmented response difference block diagram, extract the direction of temperature and humidity rate change of the acquisition nodes in the three layers of evaporation layer, core layer and moisture accumulation layer during the disturbance period. By comparing the consistency of the direction change of the three layers under the same vertical projection line of the upper and lower layers, obtain the structural disturbance positioning layer.

[0108] S4: Based on the structural disturbance location layer, obtain the number of times the airflow velocity direction changes, the length of the temperature rise segment, and the number of humidity drop segments of the nodes within the time period after the disturbance. Determine whether any two items occur consecutively for more than two disturbance cycle thresholds. If the condition is met, adjust the time window length and generate a sequence of behavioral trend change trajectories.

[0109] S5: Call the behavioral trend change trajectory sequence, combine it with the time segment that is continuously switching between oscillation trend and upward trend, collect the temperature change amplitude value and humidity change amplitude value within a unit period after the disturbance, filter the area where the deviation ratio exceeds the disturbance response change ratio threshold, and generate the fermentation response mutation distribution area.

[0110] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A Liupu tea production intelligent monitoring system, characterized in that, The system includes: The disturbance sensing module acquires the distribution nodes of the evaporation layer and core layer regions during the fermentation stage of Liubao tea. A single airflow disturbance is implemented in the evaporation layer. The initial temperature and humidity values ​​of the nodes before the disturbance and the temperature rise and humidity drop within a unit response period after the disturbance are recorded. The direction of temperature and humidity change of adjacent nodes is compared to obtain the micro-domain disturbance response distribution map. The disturbance sensing module includes: The temperature and humidity data acquisition submodule acquires the distribution nodes in the evaporation layer and core layer regions during the fermentation stage of Liubao tea. A single airflow disturbance is implemented at any node in the evaporation layer. The initial temperature and humidity values ​​of the distribution nodes before the disturbance are recorded. The temperature and humidity changes of the distribution nodes within a unit response period after the disturbance are recorded. The temperature rise and humidity fall of the disturbed nodes within a unit period are calculated, and the node disturbance temperature and humidity change is generated. The node hedging screening submodule, based on the node disturbance temperature and humidity change, calls the temperature rise and humidity fall of the node within a unit response period, and compares the temperature change direction and humidity change direction of adjacent nodes according to the spatial distribution relationship between nodes. If there is a directional hedging, i.e., one node's temperature rises and its humidity falls, and the adjacent node's temperature falls and its humidity rises, the module filters the node combination pairs that meet the directional hedging relationship and generates a micro-domain directional hedging node pair set. The disturbance response map generation submodule calls the micro-domain directional hedging node pair set, marks the region where the directional hedging node pair is located as the disturbance activation region according to the position of the node in the regional distribution, draws the boundary shape and position distribution of the disturbance activation region in spatial coordinates, and generates a micro-domain disturbance response distribution map. Based on the micro-domain disturbance response distribution map, the layered structure module calls the pile thickness variation data, unit volume density distribution data and boundary airflow disturbance response value to divide the pile into three segments: evapotranspiration layer, core layer and moisture accumulation layer, and obtains segment response difference block map; The hierarchical structure module includes: The conduction delay calculation submodule calls the micro-domain disturbance response distribution map, combines the node positions within the segmented blocks, calculates the delay time from the application of boundary airflow disturbance to the start of node response based on the temperature rise and humidity drop values ​​of the nodes within a unit response period, and combines the disturbance timestamp, and calculates the average conduction hysteresis value by block aggregation, generating segmented block airflow conduction delay values; The lag region filtering submodule sets the conduction delay benchmark value as the average lag value of the entire block based on the airflow conduction delay value of the segmented block, filters blocks with conduction delay values ​​greater than the benchmark value, extracts the spatial location in the three-segment stack partition identification matrix, and generates a set of disturbance response lag blocks. The response difference map generation submodule calls the set of disturbance response hysteresis blocks, superimposes the three-segment stack partition identification matrix and the segment block airflow conduction delay value, marks the stack segment type to which the hysteresis block belongs, and attaches the corresponding conduction delay magnitude parameter value to obtain the segment response difference block map. The anomaly identification module extracts the temperature and humidity rate direction of the acquisition nodes in the multi-level data acquisition nodes during the disturbance period based on the hysteresis region marked in the segmented response difference block diagram, identifies whether there is a case where the direction of one layer is opposite to that of the upper and lower layers at the same time, and obtains the structural disturbance location layer. The anomaly detection module includes: The temperature and humidity rate extraction submodule, based on the segmented response difference block map, obtains the evaporation layer, core layer and moisture accumulation layer segments to which the hysteresis block belongs, extracts the temperature change value and humidity change value recorded by the acquisition node within the disturbance period, calculates the temperature change rate and humidity change rate per unit time, and determines the direction of temperature and humidity change based on the positive and negative values, and generates a multi-layer temperature and humidity rate direction set for the hysteresis block. The multi-layer direction comparison submodule calls the multi-layer temperature and humidity rate direction set of the hysteresis block to make directional judgments on the nodes of the three levels in the hysteresis block, compares the temperature and humidity rate directions of the evaporation layer, the core layer and the moisture accumulation layer, identifies whether any layer is simultaneously opposite to the two adjacent layers above and below in both the temperature and humidity rate directions, marks the layer as an abnormal structure layer, and generates an abnormal direction reversal structure set. The disturbance location map generation submodule calls the abnormal direction reversal structure set, combines the position index of the abnormal structure layer in the hysteresis block and the coordinate system of the three-segment stack partition identifier matrix, constructs the spatial structure correspondence, plots the spatial location and layer attributes of the abnormal structure layer, and obtains the structural disturbance location layer. The trend tracking module collects the number of airflow speed direction changes, the length of temperature rise segments, and the number of humidity drop segments within a time period based on the structural disturbance positioning layer. It then divides the continuous change type of the time window into three trend types: rising, oscillating, and stabilizing, and records them to obtain the behavioral trend change trajectory sequence.

2. The intelligent monitoring system for Liubao tea production according to claim 1, characterized in that: The trend tracking module includes: The disturbance area feature acquisition submodule collects records of airflow velocity direction changes, temperature value sequences, and humidity value sequences of corresponding regional nodes within a specified time period based on the spatial location of the abnormal structural layer marked in the structural disturbance positioning layer. It also counts the number of airflow velocity direction changes, the length of the continuous temperature rise segment, and the number of humidity drop segments within the time period, generating three types of change feature value sets for the disturbance area. The continuous change judgment submodule calls the three types of change feature value sets of the disturbance area, arranges the three items of airflow speed direction change number, temperature rise segment length and humidity drop segment number in time order, and judges whether any two items have exceeded the double cycle standard in a continuous manner, that is, the two values ​​exceed the set change threshold value in two consecutive monitoring cycles. If there is a data segment that meets the conditions, it is marked as a continuous abnormal segment and a double cycle continuous abnormal identification segment is generated. The monitoring cycle adjustment submodule resets the monitoring cycle time window length within the time period based on the time boundary of the dual-cycle continuous anomaly identification segment, expands the cycle duration forward and backward on both sides of the continuous anomaly segment to construct a new time window, updates the original cycle division method, and re-indexes the position distribution of data segments within the adjusted cycle to generate a dynamic monitoring time window index set. The trend trajectory generation submodule calls the dynamic monitoring time window index set, sequentially reads the numerical trends of the three types of change characteristic values ​​in the disturbance area within the time window, and classifies the time window type into three trend types—rising, oscillating, or stabilizing—based on the fluctuation amplitude of the number of changes in airflow speed direction, the gradient of the change in the length of the temperature rise segment, and the periodic repetition rate of the number of humidity drop segments. The module then records the trend type change path in chronological order to obtain the behavioral trend change trajectory sequence.

3. The intelligent monitoring system for Liubao tea production according to claim 2, characterized in that: The statistical process for the number of airflow velocity and direction changes is as follows: in each monitoring cycle, at a time interval of 10 minutes, it is recorded whether the airflow direction change occurs. If there is an angular shift of more than 45 degrees in the direction between two adjacent time intervals, it is considered as one change. The statistical process for the length of the continuous temperature rise segment is as follows: the segment with more than 3 consecutive monitoring points in the monotonically increasing sequence is taken as the rise segment, and the cumulative duration in each monitoring cycle is calculated. The statistical process for the number of humidity drop segments is as follows: within each monitoring cycle, based on the gradient formed by three adjacent monitoring points in the sequence, identify drop segments that meet the condition of an average drop of more than 5% and a duration of not less than 30 minutes, and use the number of drop segments that meet the condition as the statistical value of the number of humidity drop segments.

4. The intelligent monitoring system for Liubao tea production according to claim 1, characterized in that: The system also includes a closed-loop linkage module: The closed-loop linkage module extracts the continuous switching records of oscillation trend and upward trend based on the behavioral trend change trajectory sequence, locates the time nodes marked as unstable behavior sudden segments, re-performs perturbation actions on the spatial region within the time period, identifies the magnitude of temperature and humidity response changes after the disturbance, compares the characteristics of the first disturbance response, identifies whether there are changes in the spatial response pattern, and obtains the fermentation response mutation distribution area. The fermentation response mutation distribution area includes spatial response pattern change points, unstable behavior sudden segment regions, and repeated perturbation response comparison difference areas.

5. The intelligent monitoring system for Liubao tea production according to claim 4, characterized in that: The closed-loop linkage module includes: The sudden segment identification submodule extracts the position index of the continuous switching between oscillation trend and upward trend in the time series based on the behavior trend change trajectory sequence, locates the time nodes where the trends alternate and marks them as unstable behavior sudden segments, identifies the start and end time indexes and corresponding spatial area locations covered by the sudden segment, and generates an unstable trend sudden segment identifier set. The disturbance retest execution submodule calls the unstable trend sudden segment identifier set, reapplies a single airflow disturbance action in the spatial region corresponding to the identified time node, records the initial temperature and humidity values ​​of the nodes before and after the disturbance, and the temperature and humidity changes within a unit response period, calculates the temperature and humidity change amplitudes of each node after the disturbance, and generates a re-disturbance temperature and humidity response amplitude set. The response mutation identification submodule, based on the re-perturbation temperature and humidity response amplitude set, calls the temperature change amplitude and humidity change amplitude values ​​recorded by the node in the first perturbation operation, calculates the difference between the two perturbation response amplitudes, compares whether the difference exceeds the set spatial response change threshold, identifies the positional relationship in the spatial distribution, records the heap coordinate range corresponding to the node, and obtains the fermentation response mutation distribution area.

6. A method for intelligent monitoring of Liubao tea production, characterized in that, The intelligent monitoring system for Liubao tea production according to any one of claims 1-5 includes the following steps: S1: Obtain the distribution nodes of the evaporation layer and core layer regions during the fermentation stage of Liubao tea, record the initial temperature and humidity values ​​of the nodes before airflow disturbance, activate the airflow disturbance actuator once within the node group, call the temperature rise and humidity fall within a unit time period after disturbance, calculate the difference in the direction of temperature and humidity change between adjacent nodes, and generate a micro-domain disturbance response distribution map. The specific steps are as follows: The distribution nodes of the evaporation layer and core layer regions in the fermentation stage of Liubao tea are obtained. A single airflow disturbance is implemented at any node in the evaporation layer. The initial temperature and humidity values ​​of the distribution nodes before the disturbance are recorded. The temperature and humidity changes of the distribution nodes within a unit response period after the disturbance are recorded. The temperature rise and humidity fall of the disturbed nodes within a unit period are calculated to generate the node disturbance temperature and humidity change. Based on the node's temperature and humidity change, the temperature rise and humidity fall of the node within a unit response period are called. According to the spatial distribution relationship between the nodes, the temperature change direction and humidity change direction of adjacent nodes are compared to see if there is directional opposition. That is, if a node's temperature rises and its humidity falls, and the adjacent node's temperature falls and its humidity rises, node pairs that meet the directional opposition relationship are selected to generate a micro-domain directional opposition node pair set. The set of micro-domain directional hedging nodes is invoked. Based on the position of the nodes in the regional distribution, the region where the directional hedging node pairs are located is marked as the disturbance activation region. The boundary shape and position distribution of the disturbance activation region are drawn in spatial coordinates to generate a micro-domain disturbance response distribution map. S2: Based on the micro-domain disturbance response distribution map, obtain the thickness range values, unit volume density values, and boundary airflow disturbance response values ​​of the evapotranspiration layer, core layer, and moisture accumulation layer in the cross-section of the stack. Using the evapotranspiration layer thickness and unit density values, calculate the airflow conduction delay time values ​​within the multi-segment region, and generate a segmented response difference block map. The specific steps are as follows: The micro-domain disturbance response distribution map is called, and combined with the node positions in the segmented blocks, the temperature rise and humidity fall of the nodes within a unit response period are calculated based on the disturbance timestamp, and the delay time from the application of the boundary airflow disturbance to the start of the node response is calculated. The average conduction hysteresis value is obtained by aggregation by block, and the segmented block airflow conduction delay value is generated. Based on the airflow conduction delay value of the segmented blocks, the conduction delay benchmark value is set as the average lag value of the entire block. Blocks with conduction delay values ​​greater than the benchmark value are selected, and their spatial positions in the three-segment stack partition identification matrix are extracted to generate a set of disturbance response lag blocks. The set of hysteresis blocks in the disturbance response is called, and the three-segment stack partition identification matrix and the airflow conduction delay value of the segment block are superimposed. The stack segment type to which the hysteresis block belongs is marked, and the corresponding conduction delay magnitude parameter value is attached to obtain the segment response difference block map. S3: Based on the segmented response difference block diagram, extract the direction of temperature and humidity rate change of the acquisition nodes in the three layers of evapotranspiration layer, core layer, and moisture accumulation layer during the disturbance period. By comparing the consistency of the directional changes of the three layers under the same vertical projection line between the upper and lower layers, obtain the structural disturbance positioning layer. The specific steps are as follows: Based on the segmented response difference block map, the evapotranspiration layer, core layer and moisture accumulation layer segments to which the hysteresis block belongs are obtained. The temperature change value and humidity change value recorded by the acquisition node during the disturbance period are extracted. The temperature change rate and humidity change rate per unit time are calculated. The direction of temperature and humidity change is determined according to the positive and negative values, and a multi-layer temperature and humidity rate direction set of the hysteresis block is generated. The hysteresis block multi-layer temperature and humidity rate direction set is called to determine the directionality of the nodes in the three levels of the hysteresis block. The temperature and humidity rate directions of the evaporation layer, the core layer and the moisture accumulation layer are compared. It is identified whether any layer is simultaneously opposite to the two adjacent layers above and below in both the temperature and humidity rate directions. The layer segment is marked as an abnormal structure layer and an abnormal direction reversal structure set is generated. The abnormal direction reversal structure set is invoked, and combined with the position index of the abnormal structure layer in the lag block and the coordinate system of the three-segment heap partition identifier matrix, the spatial structure correspondence is constructed, the spatial location and layer attributes of the abnormal structure layer are plotted, and the structural disturbance positioning layer is obtained. S4: Based on the structural disturbance positioning layer, obtain the number of times the airflow velocity direction changes, the length of the temperature rise segment, and the number of humidity drop segments of the node within the time period after the disturbance. Determine whether any two items occur consecutively for more than two disturbance cycle thresholds. If the condition is met, adjust the time window length and generate a behavioral trend change trajectory sequence. S5: Call the behavior trend change trajectory sequence, combine it with the time segment that is classified as a continuous switching between oscillation trend and upward trend, collect the temperature change amplitude value and humidity change amplitude value within a unit period after the disturbance, filter the area where the deviation ratio exceeds the disturbance response change ratio threshold, and generate the fermentation response mutation distribution area.

Images