Fertilizer production self-adaptive control method and system based on strategy selection

Abstract

The invention discloses a strategy selection-based fertilizer production adaptive control method and system. The method comprises the following steps of: acquiring a continuous sample time sequence image and production process data; constructing a fault-tolerant envelope; judging whether the sample is in a heat and mass transfer state or not according to the continuous sample time sequence image; sample fertilizer particles are placed in a bypass detection cavity, and time sequence temperature and humidity data and time sequence air pressure data are obtained; determining a micro temperature and humidity transition signal according to the time sequence temperature and humidity data and the time sequence air pressure data; calculating a temperature and humidity dynamic index according to the micro temperature and humidity transition signal; calculating an interval stability indicating quantity based on the temperature and humidity dynamic indexes, and determining the current process stage of the sample fertilizer particles; determining to enter a strategy switching area through the interval stability indication quantity, and selecting a target control strategy from a preset strategy library according to the fault-tolerant envelope corresponding to the process stage; and issuing the target control strategy to an execution mechanism in the production line. The strategy selection precision in the fertilizer production process can be effectively improved.

Description

Technical Field

[0001] This application relates to the field of fertilizer production control strategies, and in particular to an adaptive control method and system for fertilizer production based on strategy selection. Background Technology

[0002] As the modern fertilizer industry continues to develop towards high capacity, low energy consumption, and stable product quality, the dynamic controllability and process stability of fertilizer production processes have become increasingly critical. Especially in the production processes of highly hygroscopic fertilizers such as urea, ammonium phosphate, and ammonium chloride, the granulation, drying, sizing, and coating processes are interdependent and highly sensitive, and their operational stability has become an important factor affecting the continuity of the production line and product performance.

[0003] Existing technologies generally employ traditional automatic control methods based on macroscopic parameters such as temperature, humidity, airflow, material layer thickness, and particle moisture content. These methods control granulation uniformity, drying degree, and coating quality through feedback adjustments to heating intensity, ventilation volume, liquid spraying volume, and coating conditions. This control method primarily relies on adjusting the production environment and process parameters to indirectly maintain particle stability. However, in actual production, the hygroscopic and desiccant properties of fertilizer granules cause significant moisture responses to even minor fluctuations in environmental conditions. For example, in the granulation stage, a slight increase in ambient humidity or uneven local airflow can lead to increased moisture absorption by small particles, surface liquefaction, and increased viscosity, potentially causing material return or sticking to the drying trays. In the drying stage, particles are highly sensitive to changes in drying intensity and humidity, potentially resulting in under-dried or over-dried granules, increasing the risk of moisture absorption or re-moistening in subsequent production stages, affecting particle quality and production efficiency. Therefore, existing control methods, while adjusting macroscopic environmental parameters, struggle to provide precise, real-time responses to rapid changes in particle appearance. This could lead to inaccurate selection of production process strategies, affecting production stability and increasing energy consumption and production costs.

[0004] There is currently no good solution to the above problems. Summary of the Invention

[0005] This application provides an adaptive control method and system for fertilizer production based on strategy selection, which is used to improve the accuracy of strategy selection during fertilizer production.

[0006] To achieve the above objectives, the embodiments of this application adopt the following technical solutions: Firstly, a strategy-selection-based adaptive control method for fertilizer production is provided, applied to a fertilizer production line, wherein the fertilizer production line includes a bypass detection chamber, and the method includes: Collect continuous time-series images of sample fertilizer granules and production process data corresponding to each process stage in the fertilizer production line; Construct a fault tolerance envelope for each process type based on the production process data corresponding to each process stage; Determine whether the sample fertilizer particles are in a state of heat and mass exchange based on the time series images of continuous samples; If the sample fertilizer particles are in a state of heat and mass exchange, place the sample fertilizer particles in the bypass detection chamber and obtain the time-series temperature and humidity data and time-series air pressure data in the bypass detection chamber. The micro-temperature and humidity transition signals in the bypass detection chamber are determined based on time-series temperature and humidity data and time-series air pressure data. The temperature and humidity dynamic index is calculated based on the micro-temperature and humidity transition signal. The temperature and humidity dynamic index is used to characterize the intensity of moisture absorption or release of the sample fertilizer particles. The interval stability indicator is calculated based on the temperature and humidity dynamic index, and the interval stability indicator is compared with the preset process stage characteristic interval to determine the current process stage of the sample fertilizer particles. The interval stability indicator is used to determine whether the strategy switching zone has been entered. If the strategy switching zone has been entered, the target control strategy is selected from the preset strategy library according to the fault tolerance envelope corresponding to the process stage. The target control strategy is distributed to the execution mechanisms within the production line.

[0007] In one possible implementation of the first aspect, determining whether the sample fertilizer particles are in a state of heat and mass exchange based on the continuous sample time-series image includes: Image preprocessing is performed on continuous sample time-series images; Extract the overall particle contour and matrix contour of each sample fertilizer particle from the preprocessed continuous sample time series image. The matrix contour is used to characterize the outer boundary contour of the fertilizer particle after removing the surface cavity. The mean equivalent particle size of the current continuous sample time series image is calculated based on the overall particle outline of each sample fertilizer particle, and is used as the equivalent particle size value of the image frame corresponding to the current continuous sample time series image. Calculate the total effective void area and particle matrix surface area corresponding to the current continuous sample time series image based on the matrix profile; The proportion of void area is calculated by the ratio between the total effective void area and the surface area of ​​the granular matrix. The inverse relationship between the equivalent particle size and the proportion of void area in each continuous sample time series image is determined by the corresponding equivalent particle size value and void area proportion in the continuous sample time series image. Whether a state of heat and mass exchange is in the inverse relationship between the equivalent particle size and the proportion of void area in each consecutive time series image is determined.

[0008] In one possible implementation of the first aspect, determining whether a state of heat and mass exchange is in progress by the inverse correlation between the equivalent particle size and the proportion of void area in each consecutive time-series image includes: The equivalent particle size and the proportion of void area corresponding to the first image frame in the continuous sample time series image are used as the benchmark parameter values; Calculate the grain size change rate and porosity change rate between each adjacent image frame in a continuous sample time series image based on the baseline parameter values; For any continuous sample time series image, when the particle size change rate is greater than zero and the porosity change rate is less than zero, the time corresponding to the continuous sample time series image is taken as the candidate time for moisture absorption. When the particle size change rate is less than zero and the porosity change rate is greater than zero, the time corresponding to the continuous sample time series image is taken as the candidate time for dehumidification. Whether a state of heat and mass exchange is in place is determined by the candidate times for moisture absorption and moisture release.

[0009] In one possible implementation of the first aspect, determining the micro-temperature and humidity transition signal within the bypass detection cavity based on time-series temperature and humidity data and time-series air pressure data includes: Normalize the time-series temperature and humidity data and the time-series air pressure data; The first-order and second-order differences of the time-series temperature and humidity data were used to determine the rate of change of temperature and humidity and the acceleration of temperature and humidity transitions at each sampling time point. Calculate the variance of temperature and humidity fluctuations at each sampling time point for the time-series temperature and humidity data within a preset short time window; Calculate the rate of pressure change at each sampling time point based on time-series air pressure data; The instantaneous feature vector sequence is constructed from the temperature and humidity change rate value, temperature and humidity transition acceleration value, temperature and humidity fluctuation variance value, and pressure change rate corresponding to each sampling time point. The confidence level of the transition signal is determined by verifying the feature synchronization of the instantaneous feature vector sequence; After passing the confidence verification, the instantaneous feature vector sequence is transformed into the transition signal activation score at each sampling time point using the transition signal activation function; The micro-temperature and humidity transition signal at each sampling time point is determined by combining the activation score of the transition signal at each sampling time point. The micro-temperature and humidity transition signal is used to indicate the occurrence of transient moisture absorption or release events of sample fertilizer particles.

[0010] In one possible implementation of the first aspect, the step of verifying the feature synchronization of the instantaneous feature vector sequence to determine the confidence level of the transition signal includes: Within a preset short time window, determine whether the peak value of the humidity transition acceleration in the instantaneous feature vector sequence exceeds a preset humidity peak value threshold and whether the temperature and humidity fluctuation variance exceeds a preset fluctuation threshold within the same preset short time window. If the peak value of the humidity transition acceleration exceeds the preset humidity peak value threshold and the variance of temperature and humidity fluctuations exceeds the preset fluctuation threshold within the same preset short time window, then the main criterion is determined to be valid. If the main criterion is met, determine whether the absolute value of the pressure change rate exceeds the preset pressure change threshold. If the absolute value of the pressure change rate exceeds the preset pressure change threshold, the auxiliary enhancement criterion is determined to be valid. For any sampling time point, a confidence level is generated based on the main criterion and the auxiliary enhancement criterion; Once the confidence level passes the continuity check, the confidence level of the transition signal is determined.

[0011] In one possible implementation of the first aspect, the calculation of temperature and humidity dynamics parameters based on micro-temperature and humidity transition signals includes: The transition initiation point, peak point, and stable point of fertilizer granule moisture absorption / release events are determined based on micro-temperature and humidity transition signals. The initiation gradient value of the hygroscopic / dehumidification event of fertilizer particles is calculated based on the initiation point and peak point of the transition. The gradient values ​​of the hygroscopic / dehumidification events of fertilizer particles are calculated by using the peak and stable points of the transitions. Determine the peak initiation time and peak decay time corresponding to the initiation gradient value and the decay gradient value, respectively. The phase lag index is calculated by the peak start time and the peak decay time. Within the time interval from the transition initiation point to the transition stability point, the transition energy index of fertilizer particle moisture absorption / release events is obtained by integrating the time-series temperature and humidity data within the interval. Within the time interval from the transition initiation point to the transition stability point, the spectral entropy value of the fertilizer particle moisture absorption / release event is calculated based on time-series temperature and humidity data. The spectral entropy value is used to characterize the transition complexity of the event stability. The phase lag index, transition energy index, and spectral entropy value were correlated and fused to determine the temperature and humidity kinetics of fertilizer particle hygroscopic / dehumidification events.

[0012] In one possible implementation of the first aspect, the correlation and fusion of the phase hysteresis index with the transition energy index and the spectral entropy value to determine the temperature and humidity kinetics of the fertilizer particle hygroscopic / dehumidification event includes: The phase lag index, transition energy index, and spectral entropy value are normalized to obtain the normalized phase lag value, transition energy value, and spectral entropy value. The energy transfer direction is defined by the phase lag value, and the energy transfer direction is used to characterize the heat absorption or heat release process of fertilizer particles. The direction is corrected based on the energy transfer direction and the normalized transition energy value to obtain the effective energy after direction correction. The effective energy after direction correction is stable by performing stability correction on the normalized spectral entropy value to obtain the stable corrected energy. The temperature and humidity kinetic index is calculated using a preset fusion formula based on the stabilized corrected energy, normalized phase lag value, and normalized spectral entropy value. The temperature and humidity kinetic index characterizes the total energy, energy direction, and process stability of the moisture absorption / release event of fertilizer particles.

[0013] In one possible implementation of the first aspect, the step of calculating the interval stability indicator based on temperature and humidity kinetic indices and comparing the interval stability indicator with a preset process stage characteristic interval to determine the current process stage of the sample fertilizer particles includes: Acquire the temperature and humidity dynamics indicators corresponding to all transition events within a preset time interval; The temperature and humidity dynamic indicators were normalized and directionally adjusted to unify the contribution direction of each temperature and humidity dynamic indicator to instability. Generate weights for each transition event and calculate the instability score for each transition; Aggregate and analyze the instability scores of all transition events within the preset time interval to determine the overall instability of the preset time interval. The interval stability indicator is determined by the overall instability after aggregation; The interval stability indicator is compared with the preset stability threshold to determine the current process stage of the sample fertilizer particles.

[0014] Secondly, this application provides a fertilizer production adaptive control system based on strategy selection, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the fertilizer production adaptive control method based on strategy selection as described in any one of the first aspects.

[0015] Thirdly, this application provides a computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to perform the fertilizer production adaptive control method based on strategy selection according to any one of the first aspects.

[0016] By employing the aforementioned technical solution, continuous time-series images of samples are acquired to obtain real-time dynamic evolution characteristics of particle appearance. This solves the technical problem of traditional methods that rely solely on temperature and humidity sensor data, which cannot accurately identify whether particles are in a state of heat and mass exchange. This provides a reliable basis for subsequent calculation of temperature and humidity kinetic indicators and process stage determination. Simultaneously, acquiring production process data allows for a comprehensive understanding of the operating parameters of each process stage in the fertilizer production line. Constructing a fault-tolerant envelope for each process type based on the production process data corresponding to each process stage helps identify and define normal and abnormal production states. It also provides a reference standard for selecting target control strategies based on interval stability indicators. When the interval stability indicator of sample fertilizer particles is compared with the preset process stage characteristic interval, the fault-tolerant envelope can be used to determine whether a control strategy needs to be switched and which strategy to select, thereby achieving adaptive control. Identifying whether sample fertilizer particles are in a state of heat and mass exchange based on continuous time-series images allows for timely detection of critical state changes, preventing product quality problems caused by abnormal heat and mass exchange. If the sample fertilizer granules are in a state of heat and mass exchange, the time-series temperature and humidity data and time-series air pressure data acquired in the bypass detection chamber can accurately reflect the microenvironmental changes of the sample fertilizer granules during the heat and mass exchange process. Furthermore, detection through the bypass detection chamber does not affect the normal fertilizer production process, achieving an organic integration of detection and production, ensuring both production efficiency and the accuracy of the detection data. Determining the micro-temperature and humidity transition signals within the bypass detection chamber based on the time-series temperature and humidity data and time-series air pressure data allows for a deeper understanding of the dynamic characteristics of the heat and mass exchange process, improving the accuracy of strategy switching. Calculating temperature and humidity kinetic indicators based on the micro-temperature and humidity transition signals establishes a direct link between the heat and mass exchange state during production and product quality indicators, providing intuitive data for subsequent analysis. Based on temperature and humidity dynamics indicators, an interval stability indicator is calculated. This indicator is then compared with preset process stage characteristic intervals to determine the current process stage of the sample fertilizer granules. This accurate assessment of the sample fertilizer granules' current process stage allows for more refined monitoring of the production process, facilitating timely detection of deviations or anomalies and ensuring production proceeds according to the predetermined process flow. The interval stability indicator also determines whether to enter a strategy switching zone, enabling intelligent decision-making in the production process. When the process stage of the sample fertilizer granules changes and enters the strategy switching zone, a suitable target control strategy is automatically selected from the preset strategy library without manual intervention, improving automation and response speed. The target control strategy is promptly and accurately transmitted to the actuators within the production line, ensuring rapid and effective execution of control commands. This allows adjustments and optimizations to the production process to be quickly implemented, avoiding production problems caused by delayed or inaccurate control command execution and improving the accuracy of strategy selection during fertilizer production.

[0017] Other features and advantages of the embodiments of this application will be described in detail in the following detailed description section. Attached Figure Description

[0018] Figure 1 A flowchart illustrating an adaptive control method for fertilizer production based on strategy selection, provided in an embodiment of this application; Figure 2 This is a schematic diagram of a structure for determining whether sample fertilizer particles are in a state of heat and mass exchange, provided in an embodiment of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for illustration and explanation of the embodiments of this application and are not intended to limit the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0020] It should be noted that if the embodiments of this application involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0021] Furthermore, if the embodiments of this application involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed in this application.

[0022] Figure 1 The illustration schematically shows a flow chart of an adaptive control method for fertilizer production based on strategy selection according to an embodiment of this application. Figure 1 As shown in the figure, this application provides an adaptive control method for fertilizer production based on strategy selection, applied to a fertilizer production line. The fertilizer production line includes a bypass detection chamber, and the method may include the following steps.

[0023] S110. Collect continuous time-series images of sample fertilizer granules and production process data corresponding to each process stage in the fertilizer production line. S120. Construct a fault tolerance envelope for each process type based on the production process data corresponding to each process stage. S130. Determine whether the sample fertilizer particles are in a state of heat and mass exchange based on the continuous sample time series image. S140. If the sample fertilizer particles are in a state of heat and mass exchange, place the sample fertilizer particles in the bypass detection chamber and obtain the time-series temperature and humidity data and time-series air pressure data in the bypass detection chamber. S150. Determine the micro-temperature and humidity transition signal in the bypass detection chamber based on the time-series temperature and humidity data and the time-series air pressure data. S160. Calculate the temperature and humidity dynamic index based on the micro temperature and humidity transition signal. The temperature and humidity dynamic index is used to characterize the intensity of moisture absorption or release of the sample fertilizer particles. S170. Calculate the interval stability indicator based on the temperature and humidity dynamic index, and compare the interval stability indicator with the preset process stage characteristic interval to determine the current process stage of the sample fertilizer particles. S180. Determine whether to enter the strategy switching zone by the interval stability indicator. If it is determined that the strategy switching zone has been entered, select the target control strategy from the preset strategy library according to the fault tolerance envelope corresponding to the process stage. S190. Distribute the target control strategy to the execution mechanism within the production line.

[0024] In this embodiment, time-series images of sample fertilizer granules are first acquired over a continuous period to obtain physical characteristic information of the granules changing over time during the production process. This acquisition can be performed using a machine vision module pre-installed in the production line. Simultaneously, production process data corresponding to each stage of the fertilizer production line are collected, including but not limited to temperature, humidity, pressure, material flow rate, and mechanical operating parameters, to establish the correlation between granule state and process stage.

[0025] For each technological stage in a fertilizer production line, historical data sets are established for each stage by collecting production process data. Based on these historical data sets, statistical analysis is performed on each process type to determine the allowable fluctuation range of each key process parameter under normal and stable operating conditions. By calculating the minimum, maximum, or mean value of each parameter ± a certain standard deviation, a corresponding tolerance envelope is constructed to characterize the acceptable fluctuation boundary of each process stage under normal operating conditions. During actual production, the collected real-time process data can be compared with the tolerance envelope. When the real-time data falls within the tolerance envelope, the process stage is considered to be operating normally; when the real-time data exceeds the tolerance envelope boundary, a potential anomaly or deviation is identified in the process stage. The system can then issue early warning signals or adjust control strategies accordingly, thereby achieving stability monitoring and process optimization of the fertilizer production process.

[0026] By analyzing continuous time-series images of samples, it is possible to determine whether the sample fertilizer particles are in a state of heat and mass exchange. First, the continuous time-series images need to be preprocessed, including image denoising, enhancement, and particle segmentation. Then, the overall particle contour and matrix contour of each fertilizer particle are extracted from the preprocessed images. The overall particle contour represents the particle's external boundary, while the matrix contour represents the particle's external boundary after removing surface cavities. Based on the overall particle contour, the equivalent particle size of all particles in each continuous time-series image is calculated, and the average value is taken as the equivalent particle size value for that image frame. Based on the matrix contour, the total effective void area and particle matrix surface area of ​​each image are calculated, thus obtaining the void area ratio. Furthermore, by analyzing the relationship between the equivalent particle size value and the void area ratio in each frame of the continuous time-series images, an inverse trend between the equivalent particle size and the void area ratio is identified. When this inverse trend continues to appear in the continuous image frames, it can be determined that the fertilizer particles are in a state of heat and mass exchange, i.e., the fertilizer particles are undergoing a moisture absorption or release process. This method allows for the direct determination of heat and mass exchange behavior based on image changes in particle physical characteristics without relying on external sensors, providing a basis for real-time monitoring and adaptive control of fertilizer production processes.

[0027] When the sample fertilizer particles are determined to be in a heat and mass exchange state based on continuous sample time-series images, they are placed in a bypass detection chamber for controlled monitoring. The bypass detection chamber is a closed or semi-closed independent detection space used to isolate the particles from direct interference from the main production line environment. After the particles are placed in the bypass detection chamber, the system continuously collects temperature, humidity, and air pressure data within the chamber, obtaining time-series temperature and humidity data and time-series air pressure data that change over time. Specifically, an automatic sampling device installed on the production line can separate the selected particles from the continuous flow. The automatic sampling device can use mechanical sorting, pneumatic suction, or other methods to collect the particles in real time. The separated particles are automatically sent into the bypass detection chamber for further analysis through a closed transmission channel or a miniature conveyor mechanism. The bypass detection chamber can achieve stable detection of single particles or small batches of particles, ensuring a one-to-one correspondence between image recognition results and heat and mass exchange analysis results.

[0028] Based on the time-series temperature and humidity data and time-series air pressure data of the sample fertilizer granules within the bypass detection chamber, the micro-temperature and humidity transition signals of the granules can be determined to indicate the occurrence of transient moisture absorption or release events. Specifically, the system first acquires the initial humidity, initial temperature, and initial pressure data within the bypass detection chamber, and normalizes the temperature, humidity, and pressure data that change over time to eliminate the influence of differences in dimensions and ranges. Subsequently, first-order and second-order differences are performed on the normalized temperature and humidity data to calculate the rate of temperature and humidity change and the transition acceleration at each sampling time point; simultaneously, the variance of temperature and humidity fluctuations is calculated within a preset short time window, and the rate of pressure change at each sampling time point is calculated based on the air pressure data. The system constructs an instantaneous feature vector sequence from the temperature and humidity change rates, transition accelerations, fluctuation variances, and pressure change rates at each time point, and performs feature synchronization verification on this sequence to determine the confidence level of the transition signal at each sampling time point. After confidence verification, the instantaneous feature vector sequence is transformed into transition signal activation scores using a transition signal activation function. These scores are then combined with the activation scores at each sampling time point to ultimately determine the micro-temperature and humidity transition signal. The micro-temperature and humidity transition signal can be used to accurately indicate transient moisture absorption or release events of sample fertilizer particles within the bypass detection chamber.

[0029] Based on the micro-temperature and humidity transition signals of sample fertilizer granules, their temperature and humidity kinetic indices can be calculated to characterize the intensity of hygroscopic absorption or release. Specifically, the transition initiation point, peak point, and stable point of the fertilizer granule's hygroscopic absorption or release event are determined based on the micro-temperature and humidity transition signals. On this basis, the initiation gradient value is calculated using the initiation point and peak point, and the decay gradient value is calculated using the peak point and stable point. The peak times corresponding to the initiation and decay gradient values ​​are also determined. Furthermore, a phase lag index is calculated based on the peak initiation and decay times to reflect the temporal characteristics of the hygroscopic absorption or release process. Within the time interval from the initiation point to the stable point, the time-series temperature and humidity data are integrated to obtain the transition energy index, which quantifies the total energy of the event. Simultaneously, the spectral entropy value is calculated based on the time-series temperature and humidity data within the interval to characterize the event's stability and transition complexity. Finally, the phase lag index, transition energy index, and spectral entropy value are correlated and fused to form a temperature and humidity dynamic index, which can comprehensively reflect the intensity, energy direction, and process stability of the sample fertilizer particles during the moisture absorption or release process, providing a basis for subsequent process status assessment and adaptive control.

[0030] The stability of the fertilizer granule production process is assessed based on the interval stability indicator. When the interval stability indicator falls below a preset threshold, the current process state is determined to have entered the strategy switching zone, indicating that the instability of the production process has reached a level requiring adjustment of the control strategy. Subsequently, the system searches for the fault tolerance envelope corresponding to the current process stage of the sample fertilizer granules to determine the acceptable range of process parameters. Based on this fault tolerance envelope, a target control strategy matching the current process state and fault tolerance requirements is selected from the preset strategy library to adjust the production process in real time, thereby ensuring that the hygroscopic / dehumidifying behavior of the fertilizer granules remains within a safe and stable process range.

[0031] Once the target control strategy is determined from the preset strategy library, the system will issue the control strategy to each actuator in the production line (such as fans that regulate airflow, heating or cooling devices, conveying mechanisms, etc.) through control commands. After receiving the control commands, the actuators will adjust the corresponding process parameters in real time according to the command content, so as to keep the moisture absorption / desorption process of fertilizer granules in the target process state, ensuring that the production process is stable and meets the preset process requirements.

[0032] By acquiring continuous time-series images of samples, the dynamic evolution characteristics of particle appearance can be obtained in real time. This solves the technical problem that traditional methods relying solely on temperature and humidity sensor data cannot accurately identify whether particles are in a state of heat and mass exchange, thus providing a reliable basis for subsequent calculation of temperature and humidity kinetic indicators and process stage determination. Simultaneously, acquiring production process data allows for a comprehensive understanding of the operating parameters of each process stage in the fertilizer production line. Constructing a fault-tolerant envelope for each process type based on the production process data corresponding to each process stage helps identify and define normal and abnormal production states, and provides a reference standard for selecting target control strategies based on interval stability indicators. When the interval stability indicator of sample fertilizer particles is compared with the preset process stage characteristic interval, the fault-tolerant envelope can be used to determine whether a control strategy needs to be switched and which strategy to select, thereby achieving adaptive control. Identifying whether sample fertilizer particles are in a state of heat and mass exchange based on continuous time-series images allows for timely detection of critical state changes, preventing product quality problems caused by abnormal heat and mass exchange. If the sample fertilizer granules are in a state of heat and mass exchange, the time-series temperature and humidity data and time-series air pressure data acquired in the bypass detection chamber can accurately reflect the microenvironmental changes of the sample fertilizer granules during the heat and mass exchange process. Furthermore, detection through the bypass detection chamber does not affect the normal fertilizer production process, achieving an organic integration of detection and production, ensuring both production efficiency and the accuracy of the detection data. Determining the micro-temperature and humidity transition signals within the bypass detection chamber based on the time-series temperature and humidity data and time-series air pressure data allows for a deeper understanding of the dynamic characteristics of the heat and mass exchange process, improving the accuracy of strategy switching. Calculating temperature and humidity kinetic indicators based on the micro-temperature and humidity transition signals establishes a direct link between the heat and mass exchange state during production and product quality indicators, providing intuitive data for subsequent analysis. Based on temperature and humidity dynamics indicators, an interval stability indicator is calculated. This indicator is then compared with preset process stage characteristic intervals to determine the current process stage of the sample fertilizer granules. This accurate assessment of the sample fertilizer granules' current process stage allows for more refined monitoring of the production process, facilitating timely detection of deviations or anomalies and ensuring production proceeds according to the predetermined process flow. The interval stability indicator also determines whether to enter a strategy switching zone, enabling intelligent decision-making in the production process. When the process stage of the sample fertilizer granules changes and enters the strategy switching zone, a suitable target control strategy is automatically selected from the preset strategy library without manual intervention, improving automation and response speed. The target control strategy is promptly and accurately transmitted to the actuators within the production line, ensuring rapid and effective execution of control commands. This allows adjustments and optimizations to the production process to be quickly implemented, avoiding production problems caused by delayed or inaccurate control command execution and improving the accuracy of strategy selection during fertilizer production.

[0033] In one embodiment of this example, Figure 2 This application provides a schematic diagram of a structure for determining whether sample fertilizer particles are in a state of heat and mass exchange. The determination, based on a continuous sample time-series image, includes: S210. Perform image preprocessing on the continuous sample time series images; S220. Extract the overall particle contour and matrix contour of each sample fertilizer particle in the preprocessed continuous sample time series image. The matrix contour is used to characterize the outer boundary contour of the fertilizer particle after removing the surface cavity. S230. Calculate the mean equivalent particle size corresponding to the current continuous sample time series image based on the overall particle outline of each sample fertilizer particle, and use it as the equivalent particle size value of the image frame corresponding to the current continuous sample time series image. S240. Calculate the total effective void area and particle matrix surface area corresponding to the current continuous sample time series image based on the matrix profile. S250, The proportion of void area is calculated by the ratio between the total effective void area and the surface area of ​​the particle matrix; S260. Determine the inverse relationship between the equivalent particle size and the proportion of void area in each continuous sample time series image by using the corresponding equivalent particle size value and void area proportion in the continuous sample time series image. S270. Determine whether the sample is in a state of heat and mass exchange by the inverse correlation between the equivalent particle size and the proportion of void area in each consecutive time series image.

[0034] First, image preprocessing is performed on the continuous sample time-series images. This involves constructing a time-series image sequence from the acquired multiple sample images and then performing quality enhancement and normalization processing on the time-series image sequence. The timestamps of each frame are corrected and aligned, and missing and duplicate frames are removed. Subsequently, image enhancement algorithms such as filtering and denoising, brightness equalization, and contrast enhancement are used to improve image quality. Finally, geometric correction and region cropping are performed on the time-series images. Through perspective correction, stabilization compensation, and target region localization, the sample images acquired at different times are kept consistent in terms of viewpoint, size, and position. Further image normalization and background subtraction are performed to eliminate illumination variations, background interference, and scale differences.

[0035] Secondly, the contour information of each fertilizer particle is extracted from the preprocessed continuous time-series images, specifically including the extraction of the overall particle contour and the matrix contour. An overall contour containing all surface details of the particle is constructed, representing the complete surface morphology of the fertilizer particle at the time of imaging, including features such as local depressions, pores, cavities, and fine cracks on the particle surface. Further, morphological reconstruction processing is performed on the overall contour. Through steps such as pore filling, local depression removal, boundary smoothing, and contour reconstruction, cavities and locally collapsed areas on the particle surface are removed to obtain the matrix contour of the particle. Specifically, firstly, connected component analysis is performed on the overall contour to identify internal closed regions. When the area of ​​a connected component is less than a preset proportion of the overall projected area of ​​the particle, it is identified as a pore region and filled using a closing operation to eliminate small pores and cavities on the particle surface. The preset proportion of the overall projected area of ​​the particle is obtained through statistical analysis of the connected component area distribution of a large number of sample particle images, and can be set according to the accuracy requirements of the enterprise. Subsequently, curvature analysis is performed on the boundary of the overall contour to identify abrupt changes in boundary curvature. When both the depth and width of the depression are less than the relative proportion threshold of the average equivalent particle size, it is identified as a local depression region and removed using an opening operation to eliminate local collapses on the particle surface. The relative proportion threshold of the average equivalent particle size is obtained through statistical analysis of the defect size distribution of a large number of sample particle images, and can be set according to the accuracy requirements of the enterprise. Next, the contour boundary after hole filling and depression removal is smoothed using Gaussian filtering to eliminate jagged edges caused by noise and ensure the continuity of the contour. Finally, the smoothed boundary is reconstructed as a continuous closed curve to form the matrix contour. The matrix contour is used to characterize the outer boundary shape of the particle body after excluding surface cavity interference, and can accurately reflect the overall volume change and shape evolution of fertilizer particles during moisture absorption, moisture release, or heat and mass exchange. By extracting the overall contour and the matrix contour separately, the separation analysis of particle surface structure changes and particle body deformation can be achieved, providing geometric variable input for the calculation of temperature and humidity dynamic indicators.

[0036] The average equivalent particle size of the current continuous sample time series image is calculated based on the overall contour of each sample fertilizer particle, and this average equivalent particle size is used as the equivalent particle size value of the corresponding image frame of the continuous sample time series image. Specifically, based on the aforementioned contour extraction results, the overall contour region of each fertilizer particle in the image frame is obtained, and the effective area enclosed by this region is calculated. Calculating the effective area enclosed by this region involves treating the extracted fertilizer particle contour as a closed polygon composed of several pixel boundary points, and performing topological sorting on the contour point set in sequence to construct a continuous boundary. Subsequently, an area calculation method based on polygon geometric integral is used to calculate the area of ​​the closed contour. For example, the area calculation formula derived from the contour point coordinates is used to perform integral calculation on the region enclosed by the overall contour to obtain the effective area corresponding to the fertilizer particle. Next, combined with the effective area corresponding to the fertilizer particle, the corresponding equivalent particle size value is calculated using a calculation method based on the diameter of an equal-area circle. The equivalent particle size can be calculated using the following formula:

[0037] Where A is the effective area of ​​the overall particle outline, and D eq This is the diameter of a circle equivalent to that area. For example, when the effective area of ​​a particle is A = 2.5 mm. 2 When the equivalent particle size is, it can be calculated as follows:

[0038] The equivalent particle size of all particles within an image frame is obtained. Furthermore, the equivalent particle size of all particles in the same image frame is statistically averaged to obtain the mean equivalent particle size, which characterizes the overall particle size characteristics of the image frame. The mean equivalent particle size reflects the average size of the sample particles at the current moment and can characterize the overall expansion or contraction trend of fertilizer particles during moisture absorption or release.

[0039] Based on the extracted matrix contour, the total effective void area and particle matrix surface area corresponding to the current continuous sample time series image are calculated. That is, the matrix contour is considered as the outer boundary of the particle body after removing cavities, pores, and local depressions on the particle surface. The area of ​​the closed region enclosed by the matrix contour is calculated to obtain the particle matrix surface area. This can be calculated using the polygon area formula or integral method in existing technology. The particle matrix surface area refers to the area of ​​the closed region enclosed by the overall particle contour, used to characterize the geometric dimensions of the particle solid part. Subsequently, using the overall particle contour as the outer boundary and the matrix contour as the inner boundary, the area of ​​the difference region between the two is calculated to obtain the total effective void area occupied by the cavities and depressions on the particle surface. The total effective void area refers to the sum of the void areas formed on the particle matrix surface or between particles, used to characterize the porosity features of the particle surface or stacked structure that are not filled by the solid. By obtaining the particle matrix surface area and the total effective void area respectively, the geometric dimension changes of the particle body and the development of the surface void structure can be reflected.

[0040] Next, based on the previously obtained total effective void area of ​​the granules, which corresponds to the area of ​​the difference between the overall outline of the granules and the outline of the matrix, and simultaneously obtaining the surface area of ​​the granule matrix, which is the main area enclosed by the matrix outline, the ratio of the total effective void area to the granule matrix surface area is calculated to obtain a void area ratio index that reflects the proportion of void structure on the granule surface. The void area ratio characterizes the distribution density of cavities on the fertilizer granule surface.

[0041] By processing continuously acquired time-series images, the equivalent particle size and void area ratio for each image frame were obtained. The equivalent particle size characterizes the overall geometric size of the particles at that moment, while the void area ratio characterizes the porosity of the particle surface or packing structure. Furthermore, a correlation analysis was performed on the equivalent particle size and void area ratio to determine their inverse relationship: as the equivalent particle size increases, the void area ratio decreases, and vice versa.

[0042] The presence of sample particles in a state of heat and mass exchange is determined by the inverse relationship between the equivalent particle size and the proportion of void area in each consecutive time-series image. Specifically, the equivalent particle size and void area proportion corresponding to the first image frame in the consecutive time-series images are used as baseline parameters. Subsequently, based on these baseline parameters, the particle size change rate and void area change rate between each adjacent image frame in the consecutive time-series images are calculated to characterize the changing trends of the particle's main body size and surface void structure. For any consecutive time-series image, when the particle size change rate is greater than or equal to zero and the void area change rate is less than or equal to zero, the time point corresponding to that image frame is marked as a candidate moment for moisture absorption; when the particle size change rate is less than or equal to zero and the void area change rate is greater than or equal to zero, the time point corresponding to that image frame is marked as a candidate moment for moisture release. By analyzing the candidate moments for moisture absorption and moisture release in each frame, it is determined whether the particles are in a state of heat and mass exchange, that is, the inverse relationship between the equivalent particle size and the proportion of void area reflects the heat and mass exchange behavior of the particles during the moisture absorption or release process.

[0043] Determining whether fertilizer granules are in a state of heat and mass exchange by using continuous sample time-series images can significantly improve the real-time nature and accuracy of heat and mass exchange state monitoring, ensuring the targeted and timely adjustment of fertilizer production process parameters, and helping to improve the quality stability of fertilizer products.

[0044] In one embodiment of this invention, determining whether a state of heat and mass exchange is in progress is achieved by inversely correlated with the equivalent particle size and the proportion of void area in each consecutive time-series image of samples, including: S310. The equivalent particle size and the proportion of void area corresponding to the first image frame in the continuous sample time series image are used as the benchmark parameter values. S320. Calculate the particle size change rate and porosity change rate between each adjacent image frame in a continuous sample time series image based on the baseline parameter values. S330. For any continuous sample time series image, when the particle size change rate is greater than zero and the porosity change rate is less than zero, the time corresponding to the continuous sample time series image is taken as the moisture absorption candidate time. S340. When the particle size change rate is less than zero and the porosity change rate is greater than zero, the time corresponding to the time series image of the continuous sample is taken as the candidate time for dehumidification. S350: Determine whether the state is in heat and mass exchange by the candidate time of moisture absorption and the candidate time of moisture release.

[0045] This embodiment addresses the problem that relying solely on temperature and humidity sensor data in traditional fertilizer production processes is insufficient to accurately identify whether granules are in a state of heat and mass exchange. A determination mechanism based on the inverse relationship between image particle size and porosity is proposed. The equivalent particle size and porosity ratio corresponding to the first image frame in a continuous sample time-series image are used as baseline parameters. Based on these baseline parameters, the particle size change rate and porosity change rate between adjacent image frames are calculated to determine candidate moments for moisture absorption and release. By comprehensively determining these candidate moments, it is possible to accurately identify whether sample fertilizer granules are in a state of heat and mass exchange. This allows for the assessment of subtle moisture absorption or release reactions of fertilizer granules under the current environment, enabling the implementation of corresponding measures to prevent phenomena such as excessive return of fertilizer granules, sticking to trays, and excessive dryness followed by moisture reabsorption.

[0046] First, the equivalent particle size and particle matrix porosity ratio extracted from the first frame of the continuous sample time series image are used as reference benchmarks for subsequent analysis. The benchmark parameter values ​​can be used to calculate the particle size change rate and porosity change rate between adjacent image frames, thereby establishing the dynamic change relationship between the particle body size and surface porosity structure in the continuous sample time series image. By using the equivalent particle size and porosity area ratio of the first frame image as benchmarks, it can be ensured that the change analysis of the continuous sample time series image has a unified starting point and comparability, thereby improving the accuracy of heat and mass transfer state determination.

[0047] Next, using the equivalent particle size and void area ratio corresponding to the first image frame of the continuous sample time series images as benchmark parameters, each adjacent image frame is analyzed sequentially. The rate of change of the equivalent particle size relative to the previous frame and the rate of change of the void area ratio relative to the previous frame are calculated for each frame, thereby obtaining the changing trends of the particle size and surface void structure in continuous time series. Through these particle size change rates and void change rates, the hygroscopic or dehumidification behavior in the sample time series images can be determined, thus realizing the analysis and monitoring of the particle heat and mass exchange process. Among them, the particle size change rate refers to the relative change of the equivalent particle size in two consecutive image frames. Its value can be calculated by the ratio of the difference between the equivalent particle size of the current frame and the previous frame to the equivalent particle size of the previous frame, and is used to reflect the trend of particle expansion or contraction. The void area change rate refers to the relative change of the void area ratio in two consecutive image frames. Its value can be calculated by the ratio of the difference between the void area ratio of the current frame and the previous frame to the void area ratio of the previous frame, and is used to reflect the increase or decrease of the surface cavity of the particle.

[0048] For any consecutive time-series image, when the particle size change rate is greater than zero and the porosity change rate is less than zero, the time point corresponding to that image frame is marked as a candidate moment for moisture absorption. That is, for each frame of the consecutive time-series image, the rate of change of the equivalent particle size and the rate of change of the porosity area ratio are calculated; then, it is determined whether each frame simultaneously satisfies the conditions that the equivalent particle size change rate is greater than zero and the porosity area ratio change rate is less than zero. Image frames that satisfy this condition indicate that the particle's main size is either expanding or remaining constant, while the surface porosity of the particles is compressed or remains unchanged, thus this time point can be used as a candidate moment for particle moisture absorption behavior.

[0049] When the particle size change rate is less than zero and the porosity change rate is greater than zero in a continuous sample time series image, the time point corresponding to that image frame is marked as a candidate moment for moisture release. That is, for each frame of the continuous sample time series image, the rate of change of the equivalent particle size and the rate of change of the porosity area ratio are calculated; then, it is determined whether each frame simultaneously satisfies the conditions that the equivalent particle size change rate is less than zero and the porosity area ratio change rate is greater than zero. Image frames that meet this condition indicate that the particle bulk size is either shrinking or remaining constant, while the surface porosity of the particles is increasing or remaining constant, thus this time point can be used as a candidate moment for particle moisture release behavior.

[0050] By analyzing continuously acquired time-series images of samples, sampling points showing a rapid increase in humidity are first identified, and combined with corresponding changes in temperature, these are marked as candidate moments for moisture absorption. Simultaneously, sampling points showing a rapid decrease in humidity are identified, and combined with corresponding changes in temperature, these are marked as candidate moments for moisture release. Furthermore, when both candidate moments for moisture absorption and candidate moments for moisture release exist simultaneously in the temperature and humidity data corresponding to the continuous sample time-series images, and they satisfy a preset time interval and causal order relationship, the current sample is determined to be in a state of heat and mass exchange. Through this determination method, the moisture absorption and release behavior of particles in a dynamic process can be identified, and the heat and moisture coupling exchange process between particles and the environment can be quantitatively characterized.

[0051] In this embodiment, by acquiring continuous time-series images of sample fertilizer particles and combining them with temperature, humidity, and air pressure data in the bypass detection chamber, real-time identification of the heat and mass exchange state of the particles is achieved. This effectively avoids the problem of lag in particle state determination in traditional control methods, thereby solving the problem that fertilizer particles in the production process of highly hygroscopic fertilizers such as urea, ammonium phosphate, and ammonium chloride will produce violent hygroscopic or dehumidification reactions under slight fluctuations in environmental parameters, leading to a surge in return material, sticking to trays, excessive dryness and re-moistening, etc., which seriously affect the continuity of the production line and the stability of product quality.

[0052] In one embodiment of this invention, determining the micro-temperature and humidity transition signal within the bypass detection cavity based on time-series temperature and humidity data and time-series air pressure data includes: S410. Normalize the time-series temperature and humidity data and the time-series air pressure data; S420. Perform first-order and second-order differences on the time-series temperature and humidity data to determine the rate of change of temperature and humidity and the acceleration of temperature and humidity transition at each sampling time point. S430. Calculate the variance of temperature and humidity fluctuations at each sampling time point for the time-series temperature and humidity data within a preset short time window. S440. Calculate the rate of pressure change at each sampling time point based on the time-series air pressure data; S450. Construct an instantaneous feature vector sequence from the temperature and humidity change rate, temperature and humidity transition acceleration, temperature and humidity fluctuation variance, and pressure change rate corresponding to each sampling time point. S460. Verify the feature synchronization of the instantaneous feature vector sequence to determine the confidence level of the transition signal; S470. After passing the confidence verification, the instantaneous feature vector sequence is transformed into the transition signal activation score at each sampling time point using the transition signal activation function. S480. Combine the activation score of the transition signal at each sampling time point to determine the micro-temperature and humidity transition signal at each sampling time point. The micro-temperature and humidity transition signal is used to indicate the occurrence of transient moisture absorption or release events of the sample fertilizer particles.

[0053] Normalization of continuously acquired time-series temperature, humidity, and pressure data involves using the initial humidity, temperature, and pressure data from the bypass detection chamber as baseline reference values. Subsequently sampled temperature, humidity, and pressure data are then converted into relative changes or standardized values, thereby eliminating differences between different sampling conditions, dimensions, and magnitudes. Specifically, the mean-standard deviation normalization method can be used, standardizing the data around its mean.

[0054] Subsequently, the continuously acquired time-series temperature and humidity data were processed using first-order and second-order differencing. First-order differencing was used to calculate the temperature and humidity changes between adjacent sampling points to determine the rate of change at each sampling time point, thus reflecting the speed of temperature and humidity change over time. Second-order differencing was used to calculate the difference in the rate of change between adjacent sampling points to determine the acceleration value of temperature and humidity transitions at each sampling time point, thus reflecting the acceleration characteristics of the temperature and humidity change trend. First-order and second-order differencing operations can be implemented using data processing software such as MATLAB and Python.

[0055] Within a preset short time window, the variance of temperature and humidity fluctuations at each sampling time point is calculated to characterize the fluctuation characteristics of temperature and humidity over a short time range. First, the length of the short time window is set, which can be determined based on the duration of transient moisture absorption or release events of the sample fertilizer particles in the time-series temperature and humidity data. At each sampling time point, a local temperature and humidity data sequence is extracted within this window. Then, the variance of the temperature and humidity data within the window is calculated to obtain the variance of temperature and humidity fluctuations, which can be achieved using existing variance calculation formulas. The variance of temperature and humidity fluctuations is used to capture the occurrence characteristics of transient moisture absorption or release events of the sample fertilizer particles and can be combined with the rate of change of temperature and humidity and the transition acceleration value to improve the accuracy and stability of the determination of micro-temperature and humidity transition signals.

[0056] The pressure change rate is calculated at each sampling time point based on continuously collected time-series air pressure data. The pressure change rate characterizes the increase or decrease in air pressure per unit time. For sampling time point t, the pressure change rate is obtained by dividing the difference between the air pressure value at that time and the air pressure value at the previous sampling time point by the sampling interval. By performing the above calculation point-by-point on the entire time-series air pressure data, a pressure change rate sequence is formed. The pressure change rate sequence reflects the dynamic changes in air pressure over time, providing auxiliary information for determining micro-temperature and humidity transition signals, thereby improving the accuracy of identifying the heat and mass exchange behavior of sample fertilizer particles during transient moisture absorption or release processes.

[0057] For each sampling time point, the rate of change of temperature and humidity, the acceleration of temperature and humidity transitions, the variance of temperature and humidity fluctuations, and the rate of change of pressure are obtained. These indicators are combined in chronological order to form an instantaneous feature vector corresponding to each sampling time point. Furthermore, the instantaneous feature vectors of all sampling time points are arranged in chronological order to construct an instantaneous feature vector sequence. The instantaneous feature vector sequence refers to the time sequence of multi-dimensional feature vectors constructed for the preset sampling time points.

[0058] To determine the confidence level of a transition signal, a feature synchronization verification is performed on the instantaneous feature vector sequence. First, within a preset short time window, it is determined whether the peak value of the humidity transition acceleration in the instantaneous feature vector sequence exceeds a preset humidity peak threshold, and simultaneously, whether the temperature and humidity fluctuation variance exceeds a preset fluctuation threshold within the same time window. If both the peak value of the humidity transition acceleration and the temperature and humidity fluctuation variance exceed the preset fluctuation threshold within the same time window, the primary criterion is considered valid. If the primary criterion is valid, the absolute value of the pressure change rate is further determined to exceed a preset pressure change threshold; if so, the auxiliary enhancement criterion is considered valid. Subsequently, for any sampling time point, a corresponding confidence value is generated based on the primary and auxiliary enhancement criteria, and the confidence values ​​for all sampling time points are smoothed and their continuity is verified. When the confidence values ​​for consecutive smoothed time points pass the continuity verification, the final confidence level of the transition signal is determined.

[0059] After passing the confidence level verification, the system further processes the instantaneous feature vector sequence using a transition signal activation function to transform multidimensional feature information into transition signal activation scores corresponding to each sampling time point. The transition signal activation function takes the instantaneous feature vectors at each sampling time point and the confidence level obtained through feature synchronization verification as input. It integrates multidimensional data such as the rate of change of temperature and humidity, the acceleration of temperature and humidity transitions, the variance of temperature and humidity fluctuations, and the rate of change of pressure into a transition signal activation score through a preset nonlinear mapping expression. This score characterizes the activation degree of the transition signal at that sampling time point. The preset nonlinear mapping expression is shown below:

[0060] in, This represents the activation score of the transition signal; Represents the step function; Indicates the variance of humidity fluctuation; This represents the variance of temperature fluctuations; This represents the rate of change of air pressure. The multidimensional data of dynamic characteristics such as the rate of change of temperature and humidity, the acceleration of temperature and humidity transitions, the variance of temperature and humidity fluctuations, and the rate of change of pressure are normalized data.

[0061] After processing by the transition signal activation function, the transition signal activation score sequence corresponding to all sampling time points can be obtained, providing a unified and quantitative feature index for subsequent transition event identification, boundary determination or control strategy triggering.

[0062] The micro-temperature and humidity transition signal for each sampling time point is determined by combining the activation score of the transition signal. Specifically, based on experimental calibration results, an activation threshold is set to characterize weak transient changes in temperature and humidity, and the activation score of the transition signal is used as the criterion. When the activation score of the transition signal at a certain sampling time point reaches or exceeds the activation threshold, it is determined that there is a weak but real temperature and humidity transition behavior at that time point, and a corresponding micro-temperature and humidity transition signal is generated; when the activation score is lower than the activation threshold, it is determined that no micro-transition event has occurred at that time point. Next, the continuity and trend of the activation score of the transition signal within the local time window are further verified. Only when the activation score shows a continuous increase or stably exceeds the threshold in adjacent time points is the micro-transition event confirmed as valid. The finally generated micro-temperature and humidity transition signal is used to indicate whether the sample fertilizer particles have undergone transient moisture absorption or release events at the corresponding sampling time point, thereby achieving accurate identification and characterization of the particle heat and mass exchange behavior.

[0063] In this embodiment, the micro-temperature and humidity transition signals within the bypass detection chamber are determined using time-series temperature and humidity data and time-series air pressure data, thereby identifying these signals with high confidence. These micro-temperature and humidity transition signals can accurately indicate transient moisture absorption or release events of the sample fertilizer granules during production. This avoids the shortcomings of traditional control methods that rely solely on macroscopic environmental parameters and cannot capture the intrinsic dynamic behavior of the granules. It achieves real-time and refined monitoring of the granule's heat and mass exchange state, improving the accuracy and stability of process stage determination and providing a reliable basis for subsequent calculation of interval stability indicators and strategy switching.

[0064] In one embodiment of this invention, verifying the feature synchronization of the instantaneous feature vector sequence to determine the confidence level of the transition signal includes: S510. Within a preset short time window, determine whether the peak value of the humidity transition acceleration in the instantaneous feature vector sequence exceeds a preset humidity peak value threshold and whether the temperature and humidity fluctuation variance exceeds a preset fluctuation threshold within the same preset short time window. S520. If the peak value of the humidity transition acceleration exceeds the preset humidity peak value threshold and the temperature and humidity fluctuation variance exceeds the preset fluctuation threshold within the same preset short time window, then the main criterion is determined to be valid. S530. If the main criterion is met, determine whether the absolute value of the pressure change rate exceeds the preset pressure change threshold. S540. If the absolute value of the pressure change rate exceeds the preset pressure change threshold, the auxiliary enhancement criterion is determined to be valid. S550. For any sampling time point, generate confidence based on the main criterion and the auxiliary enhancement criterion; S560. Once the confidence level passes the continuity check, determine the confidence level of the transition signal.

[0065] In this embodiment, the humidity transition acceleration at each sampling time point is analyzed within a selected short time window to identify local peak values, which are then compared with a preset humidity peak threshold to determine the degree of transient moisture absorption or release events. Subsequently, the temperature and humidity fluctuation variance is calculated within the same window and compared with a preset fluctuation threshold to eliminate spurious signals caused by random noise or background fluctuations. Only when both the humidity transition acceleration peak value and the temperature and humidity fluctuation variance exceed the corresponding thresholds are the transition events within that time window considered valid micro-temperature and humidity transition signals. The preset humidity peak threshold, obtained through experimental calibration or statistical analysis, is used to determine whether the peak value of the humidity transition acceleration is sufficiently significant to identify transient moisture absorption or release events in the sample fertilizer granules. The preset short time window is used for local analysis of temperature and humidity signals to cover the typical duration of transient transition events and can be determined based on the sampling frequency and experimental observations. The preset fluctuation threshold is used to determine whether the temperature and humidity fluctuation variance is significant and can be obtained through experimental measurement or historical data analysis.

[0066] The primary criterion is established when the peak value of the humidity transition acceleration exceeds a preset humidity peak threshold, and the temperature and humidity fluctuation variance exceeds a preset fluctuation threshold within the same preset short time window. Specifically, the humidity transition acceleration at each sampling time point in the instantaneous feature vector sequence is analyzed within the selected short time window, and local peak values ​​are identified. Subsequently, the temperature and humidity fluctuation variance within the same window is calculated and compared with a preset fluctuation threshold. When both the peak value of the humidity transition acceleration and the temperature and humidity fluctuation variance exceed their respective thresholds, it indicates a significant transient change in the temperature and humidity signal within that time window. This reliably indicates the moisture absorption or release event of the sample fertilizer particles, thus classifying the transition event as a valid micro-temperature and humidity transition signal. The preset humidity peak threshold was obtained through multiple benchmark experiments on the humidity response curves of sample fertilizer granules under standard environmental conditions: different amplitude humidity step perturbations were applied under constant temperature conditions, the humidity change curves over time were recorded, and the humidity transition acceleration was calculated; the peak accelerations exhibiting hygroscopic or hygroscopic behavior in different experiments were statistically analyzed and their distribution was determined, with the lower bound of the fixed confidence interval serving as the minimum significant change in humidity transition acceleration, thereby determining the numerical range of the humidity peak threshold. Those skilled in the art can then select the final threshold based on the moisture content sensitivity of different fertilizer types. The preset fluctuation threshold was set based on the reference range of temperature and humidity fluctuation variance under static steady-state conditions: continuous temperature and humidity data were collected during the stable phase without heat and mass exchange, and the temperature and humidity fluctuation variance within multiple short time windows was calculated; the variance obtained during the stable phase was used as the noise benchmark, and its mean plus twice the standard deviation was selected as the fluctuation threshold, ensuring that conditions exceeding this threshold can be significantly distinguished from background noise. In addition, the preset pressure change threshold is obtained by conducting pressure disturbance tests on typical material conveying scenarios in the production line, including: recording the pressure change rate distribution under normal conveying conditions, and measuring the pressure rate characteristics when local expansion or contraction is caused by moisture absorption or desiccation events; using the minimum separable point of the two distributions as the pressure change threshold, thereby ensuring that the pressure change threshold has consistent repeatability on different batches of samples.

[0067] If the primary criterion is met, it is also necessary to determine whether the absolute value of the pressure change rate exceeds a preset pressure change threshold. Specifically, for each sampling time point, the pressure change rate is first calculated, and its absolute value is compared with the preset pressure change threshold. When the absolute value of the pressure change rate does not exceed the threshold, it can be considered that the pressure factor has a small impact on the temperature and humidity transition signal, thus confirming the validity of the micro-temperature and humidity transition event. Conversely, if the absolute value of the pressure change rate exceeds the threshold, it indicates that the pressure change may interfere with the temperature and humidity signal, and the transition event should be excluded or its confidence level reduced. Through this pressure-assisted criterion, the accuracy of micro-temperature and humidity transition signal identification can be further improved, and misjudgments caused by pressure interference can be effectively eliminated.

[0068] For each sampling time point, its absolute value is compared with a preset pressure change threshold. When the absolute value exceeds the threshold, it indicates a significant pressure change at that time point, thus determining the micro-temperature and humidity transition signal at that time point as a valid auxiliary enhancement criterion. The auxiliary enhancement criterion, as a supplement to the main criterion, further improves the reliability and confidence of micro-temperature and humidity transition signal identification. It also assists in verifying or correcting transition events when pressure mutations may affect the temperature and humidity signals, thereby providing a reliable basis for the accurate detection of transient moisture absorption or release events in sample fertilizer granules. The preset pressure change threshold is used to determine the significance of the pressure change rate and can be obtained through experimental calibration. The experimental calibration method involves collecting time-series pressure data from the bypass detection chamber under conditions without moisture absorption or release events, statistically analyzing the fluctuation range of the pressure change rate, and selecting the high percentile value as the threshold.

[0069] For any given sampling time point, a confidence level can be generated based on the primary criterion and the auxiliary enhancement criterion to determine whether the primary criterion and the auxiliary enhancement criterion are valid at that sampling time point. Then, based on the states of the primary and auxiliary enhancement criters, a confidence value is generated according to the confidence function C(t). The confidence level can be a discrete level or a continuous value. For example, a higher confidence value can be assigned when both the primary and auxiliary enhancement criters are valid; a medium confidence value can be assigned when the primary criterion is valid but the auxiliary enhancement criterion is invalid; and a low confidence value can be assigned when the primary criterion is invalid. The confidence function C(t) is as follows:

[0070] Among them, I z (t)=1 indicates that the principal criterion is true, otherwise it is 0; f (t)=1 indicates that the auxiliary enhancement criterion is valid, otherwise it is 0; α and β are weighting coefficients that can adjust the importance of the main criterion and the auxiliary criterion. The weighting coefficients can be obtained through experimental calibration, statistical analysis, or empirical setting. The experimental calibration method collects temperature, humidity, and pressure data under different environmental and sample conditions, analyzes the contribution of the main criterion and the auxiliary criterion to the determination of the transition event, and adjusts α and β to maximize the match between the confidence level and the actual transient event.

[0071] When post-processing the continuously acquired instantaneous feature vector sequence, a continuity check is further performed on the confidence sequence. This involves determining whether the confidence level continuously exceeds a preset threshold within a pre-defined time window. If the continuity condition is met, the transition signal within that time window is considered a valid signal. Once the confidence sequence passes the continuity check, the confidence level of the transition signal within that time window is determined as a valid confidence level, used to characterize the transient moisture absorption or release events occurring in the sample fertilizer particles at that moment.

[0072] By generating the confidence level for each sampling time point using the main criterion and auxiliary enhancement criteria, and performing smoothing and continuity verification on the confidence level, noise interference can be effectively filtered out, ensuring accurate identification of transition signals and improving the system's reliability for signals in complex environments.

[0073] In one embodiment of this invention, the calculation of temperature and humidity dynamics indicators based on micro-temperature and humidity transition signals includes: S610. Determine the transition start point, transition peak point, and transition stability point of fertilizer granule moisture absorption / release events based on micro-temperature and humidity transition signals. S620. Calculate the initiation gradient value of the fertilizer particle hygroscopic / dehumidification event based on the initiation point and peak point of the transition. S630. Calculate the gradient value of the hygroscopic / dehumidification event of fertilizer particles through the peak point and the stable point of the transition. S640. Determine the peak start time and peak decay time corresponding to the rise start gradient value and rise decay gradient value, respectively. S650: The phase lag index is calculated by the peak start time and the peak decay time; S660. Within the time interval from the transition start point to the transition stability point, the transition energy index of fertilizer particle moisture absorption / release events is obtained by integrating the time-series temperature and humidity data within the interval. S670. Within the time interval from the transition start point to the transition stability point, calculate the spectral entropy value of the fertilizer particle moisture absorption / release event based on time-series temperature and humidity data. The spectral entropy value is used to characterize the transition complexity of the event stability. S680. The phase lag index is correlated and fused with the transition energy index and the spectral entropy value to determine the temperature and humidity kinetic index of fertilizer particle hygroscopic / dehumidification events.

[0074] Based on the identified micro-temperature and humidity transition signals, the starting point, peak point, and stable point of the fertilizer granule's moisture absorption or release events are further determined. Specifically, the starting point is the time when the micro-temperature and humidity transition signal is first triggered, corresponding to the moment when the rate of temperature and humidity change or the acceleration of the transition first significantly deviates from the stable state, marking the beginning of moisture absorption or release behavior; the peak point is the time when the rate of temperature and humidity change or the acceleration of the transition reaches its maximum value, representing the most intense stage of moisture absorption or release behavior; the stable point is the time when the rate of temperature and humidity change returns to the stable range and remains stable, marking the end of the transition process. A preset time window and threshold are set, which can be set according to specific production needs, and the activation score sequence of the transition signal and its corresponding rate of temperature and humidity change and acceleration value are extracted within this window; when the activation score of the transition signal first exceeds the set threshold, this time point is determined as the start of the transition. Subsequently, within the time window after the starting point, the time point corresponding to the maximum value of the rate of temperature and humidity change or the acceleration value of the transition is identified and determined as the peak point. When the rate of change of temperature and humidity remains within a stable range for multiple consecutive time points, and the activation score of the transition signal is lower than the trigger threshold, the time point is determined to be the stable transition point.

[0075] Subsequently, using the identified transition initiation and peak points, the initiation gradient value of the fertilizer granules' hygroscopic or desiccation events is calculated. Between the transition initiation and peak points, the corresponding temperature and humidity change rate values ​​or temperature and humidity transition acceleration values ​​are extracted. The initiation gradient value is obtained by dividing the change amplitude of this characteristic value by the time interval. The initiation gradient value is used to characterize the response intensity and rate of change of fertilizer granules during the initiation phase of hygroscopic or desiccation events, reflecting the transition characteristics of granules from a stable state to a state of intense hygroscopic or desiccation within a short period of time.

[0076] Next, after obtaining the peak transition point tp and the stable transition point ts in the temperature and humidity transition signal, the temperature and humidity amplitudes Tp (peak temperature amplitude), Ts (steady-state temperature amplitude), RHp (peak humidity amplitude), and RHs (steady-state humidity amplitude) at the two time points can be extracted respectively. Based on this, the amplitude difference between the peak and the stable interval is calculated: temperature amplitude difference ΔT = Tp - Ts, and humidity amplitude difference ΔRH = RHp - RHs. Combined with the time difference Δt = ts - tp, the decay rate per unit time, i.e., the temperature decay gradient G, is obtained. d,T and humidity attenuation gradient G d,RH Temperature decay gradient G d,T= ΔT / Δt, humidity attenuation gradient G d,RH=ΔRH / Δt. The temperature and humidity decay gradient value is used to characterize the rate at which fertilizer granules recover to a steady state after experiencing a peak in moisture absorption or release. A large gradient value indicates a rapid recovery rate after moisture absorption / release and a more vigorous water exchange reaction; a small gradient value or close to zero indicates a slow or insufficient recovery of the water exchange process. To further simplify or generate a unified index, G can be used... d,RH and G d,T Normalization was performed, and the results were summed to obtain the gradient values ​​of the hygroscopic decay. These gradient values ​​characterize the rate at which fertilizer granules recover their temperature and humidity deviations to a steady state after experiencing a hygroscopic or dehumidification peak. A larger gradient value indicates a faster recovery rate after hygroscopic / dehumidification and a more vigorous water exchange reaction; a smaller gradient value or value close to zero indicates a slow or insufficient recovery of the water exchange process.

[0077] Based on the interval in the micro-temperature and humidity transition signal from steady state to rapid change phase, the gradient of the rate of change of temperature and humidity per unit time is calculated, and the time point corresponding to the maximum value of this gradient rate of change is determined as the initiation peak moment. The initiation peak moment is used to characterize the time position when the rapid transition phase of the moisture absorption / desorption event reaches its maximum change intensity. Subsequently, in the interval from the transition peak point to the transition stability point, based on the changing trend of the temperature and humidity offset decay gradient, the time point corresponding to the maximum value of the decay gradient is identified and determined as the decay peak moment. The decay peak moment is used to indicate the key node with the strongest decay rate during the recovery of the moisture absorption / desorption event to steady state.

[0078] The initiation peak moment characterizes the time point when the rate of temperature and humidity change reaches its maximum increase at the beginning of an event, while the decay peak moment indicates the time point when the rate of temperature and humidity change reaches its maximum decrease during the decline phase. Since moisture absorption / desorption events typically exhibit an asymmetric rise-decay process in their kinetic characteristics, the timing of these two events differs. By calculating the time interval between the initiation peak moment and the decay peak moment, a quantitative index for measuring the phase delay characteristics of moisture absorption or desorption events can be obtained, namely the phase lag index. When the phase lag index is positive, it indicates that the decay peak moment is later than the initiation peak moment, meaning that after a strong initiation of moisture absorption or desorption, the decay process of the particles is significantly delayed, reflecting a strong duration of the event. When the phase lag index is negative, it indicates that the decay peak moment is earlier than the initiation peak moment, meaning that the particles enter the decay phase prematurely before fully entering the peak stage during moisture absorption or desorption, reflecting the short duration or rapid decline of the event. When the phase lag index is zero, it indicates that the initiation peak and the decay peak occur simultaneously, and the moisture absorption or desorption process of the particles is completed almost synchronously, exhibiting an instantaneous rapid transition.

[0079] The transition energy index of this event is obtained by integrating the time-series temperature and humidity data within the time interval from the transition initiation point to the transition stability point. Specifically, the transition initiation point represents the time position when the temperature and humidity dynamics transition from a steady state to a significant hygroscopic or hygroscopic trend, while the transition stability point characterizes the time position when the temperature and humidity stabilize again after completing a significant change. Within this time interval, the temperature and humidity data exhibit a non-linear trajectory, and both the amplitude and duration of the changes reflect the strength of the particle's hygroscopic / hygroscopic behavior. By integrating the curves of temperature and humidity offsets over time within this interval, both the amplitude and duration of the changes can be included in the measurement range, thus forming a transition energy index that characterizes the overall energy consumption or energy release degree of the hygroscopic / hygroscopic event. The transition energy index can comprehensively reflect the intensity of the particle moisture exchange event.

[0080] Spectral analysis was performed on time-series temperature and humidity data within the time interval from the transition initiation point to the transition stability point, and the corresponding spectral entropy value was calculated based on the spectral characteristics. The interval from the transition initiation point to the transition stability point encompasses the complete process of a moisture absorption / desorption event gradually returning to a steady state from a rapid change phase. This process often exhibits multi-scale, non-stationary dynamic characteristics. By performing Fourier transforms or power spectral density estimation on the temperature and humidity change curves within the interval, the distribution of energy at different frequency components representing the event can be obtained. Furthermore, the spectral entropy value is obtained by calculating the entropy value through the normalization of the frequency energy distribution. A higher spectral entropy value indicates a more dispersed energy distribution and more complex dynamic changes, suggesting stronger irregularities or disturbances in the moisture absorption / desorption process; a lower spectral entropy value indicates a higher concentration of frequency energy, a more singular and stable event change, and stronger regularity in the moisture absorption / desorption process. Therefore, the spectral entropy value can be used as an indicator of the transition complexity characterizing the stability of moisture absorption / desorption events. The temperature and humidity kinetics of fertilizer granule moisture absorption / release events are determined by correlating and fusing the phase lag index, transition energy index, and spectral entropy value. Specifically, the phase lag index, transition energy index, and spectral entropy value are normalized to obtain normalized phase lag values, transition energy values, and spectral entropy values. Then, the energy transfer direction is defined based on the normalized phase lag value to characterize the energy flow direction of the granules during moisture absorption or release; the normalized transition energy value is then directionally corrected based on the energy transfer direction to obtain the directionally corrected effective energy. On this basis, the directionally corrected effective energy is stability-corrected using the normalized spectral entropy value to obtain the stable corrected energy. Finally, the stable corrected energy, normalized phase lag value, and normalized spectral entropy value are calculated according to a preset fusion formula to obtain a temperature and humidity kinetic index that simultaneously reflects the total energy, energy direction, and process stability of the event. This index can be used to quantitatively characterize the comprehensive kinetic characteristics of fertilizer granule moisture absorption / release events.

[0081] By determining the temperature and humidity kinetics of fertilizer granules' hygroscopic / dehumidification events, it is possible to comprehensively and quantitatively describe the dynamic characteristics of fertilizer granules during the hygroscopic / dehumidification process, significantly improving the accuracy of fertilizer performance evaluation.

[0082] In one embodiment of this invention, the phase lag index is correlated and fused with the transition energy index and the spectral entropy value to determine the temperature and humidity kinetics of the fertilizer particle hygroscopic / dehumidification event, including: S710. Normalize the phase lag index, transition energy index and spectral entropy value to obtain the normalized phase lag value, transition energy value and spectral entropy value. S720. The energy transfer direction is defined by the phase lag value. The energy transfer direction is used to characterize the heat absorption or heat release process of fertilizer particles. S730. Based on the energy transfer direction and the normalized transition energy value, the direction is corrected to obtain the effective energy after direction correction. S740. Based on the normalized spectral entropy value, the effective energy after direction correction is stable to obtain the energy after stable correction. S750: The temperature and humidity kinetic index is calculated by using a preset fusion formula to calculate the stable corrected energy, normalized phase lag value, and normalized spectral entropy value. The temperature and humidity kinetic index characterizes the total energy, energy direction, and process stability of the fertilizer particle moisture absorption / release event.

[0083] First, the phase lag index, transition energy index, and spectral entropy value are processed according to a preset normalization method to obtain normalized phase lag values, normalized transition energy values, and normalized spectral entropy values. After normalization, the data of each index are mapped to a unified standard range, enabling subsequent comparisons, weighted fusions, and comprehensive analyses of different indices to be performed under the same dimensions, thereby improving the accuracy of quantification of temperature and humidity dynamic characteristics.

[0084] The energy transfer direction is determined by the phase lag value to characterize the heat absorption or release process of fertilizer particles during hygroscopic or dehumidification events. Specifically, the phase lag value is calculated from the time difference between the peak initiation time and the peak attenuation time. When the phase lag value is positive, it indicates that the peak attenuation time is later than the peak initiation time, the energy transfer direction is from the environment to the particles, and the particles are in an endothermic state. When the phase lag value is negative, it indicates that the peak attenuation time is earlier than the peak initiation time, the energy transfer direction is from the particles to the environment, and the particles are in an exothermic state. By determining the energy transfer direction, the qualitative classification and dynamic characterization of the heat and mass exchange behavior of fertilizer particles during transient hygroscopic or dehumidification processes can be achieved.

[0085] Based on the previously obtained energy transfer direction and the normalized transition energy values, the transition energy values ​​are directionally corrected to obtain the directionally corrected effective energy. Specifically, when the fertilizer particles are in an endothermic state, the normalized transition energy value is taken as positive; when the fertilizer particles are in an exothermic state, the normalized transition energy value is taken as negative, aligning the sign of the effective energy value with the actual energy flow direction. This directional correction not only reflects the energy change amplitude of the fertilizer particles during transient heat and mass exchange but also accurately characterizes the energy flow direction. This provides a basis for subsequent calculations of temperature and humidity kinetic indicators and analysis of particle moisture absorption or release behavior, thereby improving the accuracy of quantifying particle heat and mass exchange characteristics.

[0086] The effective energy after direction correction is stabilized based on the normalized spectral entropy value to obtain the stabilized energy. Specifically, a higher normalized spectral entropy value indicates a more dispersed energy distribution; a lower value indicates a more concentrated and stable energy. The normalized spectral entropy value is then converted into a stability correction factor, which can be achieved through a linear inverse transformation using existing techniques. The stability correction factor is inversely proportional to the entropy value. This stability correction factor is then multiplied by the effective energy after direction correction. In other words, when the spectrum is unstable, the stability correction factor reduces the effective energy, thereby mitigating the impact of noise or transient interference; while when the spectrum is stable, the correction factor approaches 1, preserving the effective energy.

[0087] The stabilized corrected energy is used to characterize the total energy of fertilizer particles during hygroscopic or desiccation events; the normalized phase lag value is used to characterize the energy transfer direction; and the normalized spectral entropy value is used to characterize the stability of the event process. By weighting and combining these three types of indicators using a preset fusion formula, the resulting temperature and humidity kinetic index can simultaneously reflect the energy magnitude, energy direction, and process stability of hygroscopic or desiccation events. The preset fusion formula is expressed as follows:

[0088] Where D represents the temperature and humidity dynamic index; Esta represents the energy after stability correction; This represents the normalized phase lag value; The value represents the normalized spectral entropy; a, b, and c represent weighting coefficients, which can be directly assigned based on engineering experience and the relative importance of the indicators to the output. The contribution ratio can be determined based on the physical meaning of the indicators and experimental observations. For example, the temperature and humidity kinetic indicator is most sensitive to the response to moisture absorption / release events and can be set to 0.5; the stability correction energy Esta is next and can be set to 0.3; the normalized phase lag value or spectral entropy value is set to 0.2. Ensure the weight sum is 1 or scaled according to system requirements. The above weight values ​​are only for reference and are intended to illustrate the implementation of the present invention, and do not limit the scope of protection of the present invention.

[0089] By determining the temperature and humidity kinetics of fertilizer granule hygroscopic / dehumidification events, we can comprehensively reflect the total energy, energy direction, and process stability of these events. This not only improves the accuracy of the index calculations but also significantly enhances the ability to assess the dynamic changes in fertilizer performance.

[0090] In one embodiment of this invention, a range stability indicator is calculated based on a temperature and humidity kinetic index, and the range stability indicator is compared with a preset process stage characteristic range to determine the current process stage of the sample fertilizer granules, including: S810. Obtain the temperature and humidity dynamic indicators corresponding to all transition events within the preset time interval; S820. Normalize the temperature and humidity dynamic indicators and make directional adjustments to unify the contribution direction of each temperature and humidity dynamic indicator to instability. S830. Generate weights for each transition event and calculate the instability score for a single transition. S840. Aggregate and analyze the instability scores of all transition events within the preset time interval to determine the overall instability of the preset time interval. S850, Determine the interval stability indicator by the overall instability after aggregation; S860. Compare the interval stability indicator with the preset stability threshold to determine the current process stage of the sample fertilizer particles.

[0091] Within a pre-defined time interval, the temperature and humidity signals of fertilizer granules are analyzed, and the temperature and humidity kinetic indices corresponding to all transition events within that time interval are obtained. By acquiring the kinetic indices of all transition events, a comprehensive characterization of the heat and mass exchange behavior of fertilizer granules within a specific time period can be achieved.

[0092] The acquired temperature and humidity kinetic indices were first normalized. Then, based on the relationship between each index and process instability, the normalized indices were directionally adjusted to ensure that the contribution direction of each temperature and humidity kinetic index to instability was consistent. For example, when a larger index value indicates a more stable process, its sign was adjusted so that a larger value in the fusion calculation indicates higher instability. Through normalization and directional adjustment, the obtained temperature and humidity kinetic indices unified both their dimensions and physical meaning, thus providing a reliable basis for subsequent comprehensive evaluation of the stability of fertilizer granule moisture absorption or release processes.

[0093] For each temperature and humidity transition event detected within a preset time interval, a corresponding weight is generated based on its kinetic characteristics. The weights can be generated based on parameters such as the energy magnitude of the transition event, the effective energy after direction correction, and the energy after stability correction, reflecting the relative importance of each transition event to the overall instability of the heat and mass transfer process. Subsequently, combining the temperature and humidity kinetic indices of each transition event and its assigned weight, the instability score of a single transition is calculated, thereby quantifying the contribution of the event to the instability of the fertilizer granule's moisture absorption or release process throughout the entire time interval. By assigning weights and calculating instability scores for each transition event, the instability evaluation of the granule's heat and mass transfer process can be achieved, providing a reliable basis for the formulation of subsequent control strategies.

[0094] The instability scores of all transition events within a preset time interval are aggregated and analyzed to determine the overall instability level of that time interval. Specifically, all moisture absorption or release events that meet the transition conditions are identified within the preset time interval, and an instability score is calculated for each event. The score can be determined based on characteristic parameters such as spectral entropy, signal fluctuation amplitude, and duration. Subsequently, the instability scores of all events are aggregated and analyzed. The aggregation method may include averaging to obtain the overall instability level of that time interval. The overall instability level is used to characterize the overall stability of the moisture absorption or release process of the fertilizer granules within that time interval. After obtaining the aggregated overall instability level, it is used as an interval stability indicator. The interval stability indicator is a stability characterization parameter obtained by aggregating and mapping the instability scores of all transition events within the preset time interval. The preset stability threshold is a reference value set based on process experience, experimental data, or quality requirements. When the interval stability indicator is greater than or equal to the stability threshold, the sample fertilizer granules are determined to be in a stable process stage; when the interval stability indicator is less than the stability threshold, the sample fertilizer granules are determined to be in an unstable process stage. The above comparison and judgment mechanism can be used to identify the process stage of fertilizer granules during the hygroscopic or dehumidification process.

[0095] By determining the current process stage of the sample fertilizer granules, the stability changes of fertilizer granules at different process stages can be dynamically monitored, significantly improving the intelligence level of fertilizer production processes and the stability of product quality.

[0096] This application provides a fertilizer production adaptive control system based on strategy selection, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the fertilizer production adaptive control method based on strategy selection as described in any of the first aspects.

[0097] This application also provides a computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to perform a fertilizer production adaptive control method based on strategy selection according to any one of the first aspects.

[0098] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0099] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0100] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0101] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0102] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0103] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0104] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0105] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0106] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for policy selection based adaptive control of fertilizer production, characterized in that, The application is applied to a fertilizer production line, and the fertilizer production line comprises a bypass detection cavity, comprising: Collecting a continuous sample time sequence image of sample fertilizer particles and production process data corresponding to each process stage in the fertilizer production line; Constructing a fault-tolerant envelope line corresponding to each process type according to the production process data corresponding to each process stage; Judging whether the sample fertilizer particles are in a heat and mass exchange state according to the continuous sample time sequence image; If the sample fertilizer particles are in the heat and mass exchange state, placing the sample fertilizer particles in the bypass detection cavity and acquiring time sequence temperature and humidity data and time sequence air pressure data in the bypass detection cavity; Determining a micro temperature and humidity transition signal in the bypass detection cavity according to the time sequence temperature and humidity data and the time sequence air pressure data; Calculating a temperature and humidity dynamics index according to the micro temperature and humidity transition signal, and the temperature and humidity dynamics index is used to represent the strength of moisture absorption or moisture release of the sample fertilizer particles; Calculating an interval stability indicating quantity based on the temperature and humidity dynamics index, and comparing the interval stability indicating quantity with a preset process stage characteristic interval to determine a current process stage in which the sample fertilizer particles are located; Judging whether a strategy switching area is entered through the interval stability indicating quantity, and if it is determined that the strategy switching area is entered, selecting a target control strategy from a preset strategy library according to the fault-tolerant envelope line corresponding to the process stage; Downlinking the target control strategy to an executing mechanism in the production line.

2. The method of claim 1, wherein, The judgment whether the sample fertilizer particles are in the heat and mass exchange state according to the continuous sample time sequence image comprises: Performing image preprocessing on the continuous sample time sequence image; Extracting a particle overall contour and a substrate contour of each sample fertilizer particle in the preprocessed continuous sample time sequence image, and the substrate contour is used to represent an outer boundary contour of the fertilizer particle after removing a surface cavity; Calculating an equivalent particle diameter mean value corresponding to a current continuous sample time sequence image according to the particle overall contour of each sample fertilizer particle, and taking the equivalent particle diameter mean value as an equivalent particle diameter value of an image frame corresponding to the current continuous sample time sequence image; Calculating an effective total void area and a particle substrate surface area corresponding to the current continuous sample time sequence image according to the substrate contour; Calculating a void area proportion through a ratio between the effective total void area and the particle substrate surface area; Determining a reverse change correlation between the equivalent particle diameter and the void area proportion in each continuous sample time sequence image through the equivalent particle diameter value and the void area proportion corresponding to the continuous sample time sequence image; Determining whether the heat and mass exchange state is entered through the reverse change correlation between the equivalent particle diameter and the void area proportion in each continuous sample time sequence image.

3. The method of claim 2, wherein, The determination whether the heat and mass exchange state is entered through the reverse change correlation between the equivalent particle diameter and the void area proportion in each continuous sample time sequence image comprises: Taking an equivalent particle diameter and a void area proportion corresponding to a first image frame in the continuous sample time sequence image as a reference parameter value; Calculating a particle size change rate and a void change rate between each adjacent image frame in the continuous sample time sequence image based on the reference parameter value; For any one continuous sample time sequence image, when the particle size change rate is greater than zero and the void change rate is less than zero, taking a time corresponding to the continuous sample time sequence image as a moisture absorption candidate time. When the particle size change rate is less than zero and the void change rate is greater than zero, the time corresponding to the continuous sample time sequence image is taken as a moisture release candidate time; Whether the heat and mass exchange state is determined through the moisture absorption candidate time and the moisture release candidate time.

4. The method of claim 1, wherein, The micro-temperature and humidity transition signal in the bypass detection cavity is determined according to the time sequence temperature and humidity data and the time sequence air pressure data, and the micro-temperature and humidity transition signal includes: The time sequence temperature and humidity data and the time sequence air pressure data are normalized; The first-order difference and the second-order difference of the time sequence temperature and humidity data are respectively determined to obtain the temperature and humidity change rate value and the temperature and humidity transition acceleration value of each sampling time point; The temperature and humidity fluctuation variance value of each sampling time point is calculated in a preset short time window in combination with the time sequence temperature and humidity data; The pressure change rate of each sampling time point is calculated according to the time sequence air pressure data; The temperature and humidity change rate value, the temperature and humidity transition acceleration value, the temperature and humidity fluctuation variance value and the pressure change rate corresponding to each sampling time point are constructed into an instantaneous feature vector sequence; The confidence degree of the transition signal is determined by performing feature synchronicity verification on the instantaneous feature vector sequence; After the confidence degree is verified, the instantaneous feature vector sequence is converted into the transition signal activation score of each sampling time point by using a transition signal activation function; The micro-temperature and humidity transition signal of each sampling time point is determined in combination with the transition signal activation score of each sampling time point, and the micro-temperature and humidity transition signal is used to indicate the occurrence of the transient moisture absorption or moisture release event of the sample fertilizer particles.

5. The method of claim 4, wherein, The confidence degree of the transition signal is determined by performing feature synchronicity verification on the instantaneous feature vector sequence, and the confidence degree includes: In a preset short time window, it is judged whether the peak value of the humidity transition acceleration in the instantaneous feature vector sequence exceeds a preset humidity peak threshold value and whether the temperature and humidity fluctuation variance in the same preset short time window exceeds a preset fluctuation threshold value; If the peak value of the humidity transition acceleration exceeds the preset humidity peak threshold value and the temperature and humidity fluctuation variance in the same preset short time window exceeds the preset fluctuation threshold value, it is determined that the main criterion is established; In the case where the main criterion is established, it is judged whether the absolute value of the pressure change rate exceeds a preset pressure change threshold value; If the absolute value of the pressure change rate exceeds the preset pressure change threshold value, it is determined that the auxiliary enhancement criterion is established; For any sampling time point, the confidence degree is generated according to the main criterion and the auxiliary enhancement criterion; When the confidence degree passes the continuity check, the confidence degree of the transition signal is determined.

6. The method of claim 1, wherein, The micro-temperature and humidity transition signal is calculated, and the micro-temperature and humidity transition signal includes: The transition starting point, the transition peak point and the transition stable point of the moisture absorption / release event of the fertilizer particles are determined according to the micro-temperature and humidity transition signal; The jump start gradient value of the moisture absorption / release event of the fertilizer particles is calculated according to the transition starting point and the transition peak point; The jump decay gradient value of the moisture absorption / release event of the fertilizer particles is calculated through the transition peak point and the transition stable point; The start peak time and the decay peak time corresponding to the jump start gradient value and the jump decay gradient value are respectively determined; The phase lag index is calculated through the start peak time and the decay peak time; In the time interval from the transition starting point to the transition stable point, the time sequence temperature and humidity data in the interval are integrated to obtain the transition energy index of the moisture absorption / release event of the fertilizer particles. In the time interval from the transition starting point to the transition stable point, the spectral entropy value of the moisture absorption / desorption event of the fertilizer particles is calculated based on the time sequence temperature and humidity data, and the spectral entropy value is used to characterize the transition complexity of the event stability; The phase lag index is associated and fused with the transition energy index and the spectral entropy value to determine the temperature and humidity dynamics index of the moisture absorption / desorbing event of the fertilizer particles.

7. The method of claim 6, wherein, The phase lag index is associated and fused with the transition energy index and the spectral entropy value to determine the temperature and humidity dynamics index of the moisture absorption / desorbing event of the fertilizer particles. The phase lag index, the transition energy index and the spectral entropy value are normalized to obtain normalized phase lag values, transition energy values and spectral entropy values; The energy transmission direction is divided by the phase lag value, and the energy transmission direction is used to characterize the heat absorption or heat release process of the fertilizer particles; According to the energy transmission direction and the normalized transition energy value, the effective energy after direction correction is obtained; Based on the normalized spectral entropy value, the effective energy after direction correction is modified in stability to obtain the energy after stability correction; The stability corrected energy, the normalized phase lag value and the normalized spectral entropy value are calculated according to the preset fusion formula to obtain the temperature and humidity dynamics index, which represents the total energy, energy direction and process stability of the moisture absorption / desorbing event of the fertilizer particles.

8. The method of claim 1, wherein, The interval stability indicator is calculated based on the temperature and humidity dynamics index, and the interval stability indicator is compared with the preset process stage characteristic interval to determine the current process stage of the sample fertilizer particles, including: In a preset time interval, the temperature and humidity dynamics index corresponding to all transition events is obtained; The temperature and humidity dynamics index is normalized and directionally adjusted to unify the contribution direction of each temperature and humidity dynamics index to instability; The weight of each transition event is allocated and the instability score of single transition is calculated; The instability scores of all transition events in the preset time interval are aggregated and analyzed to determine the comprehensive instability degree of the preset time interval; The interval stability indicator is determined by the aggregated comprehensive instability degree; The interval stability indicator is compared with the preset stability threshold to determine the current process stage of the sample fertilizer particles.

9. A policy selection based adaptive control system for fertilizer production, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to realize the adaptive control method of fertilizer production based on policy selection as claimed in any one of claims 1 to 8.

10. A computer-readable storage medium having stored thereon instructions, the computer-readable storage medium comprising: The instructions, when executed by the processor, cause the processor to be configured to perform the adaptive control method of fertilizer production based on policy selection as claimed in any one of claims 1 to 8. The instructions, when executed by the processor, cause the processor to be configured to perform the adaptive control method of fertilizer production based on policy selection as claimed in any one of claims 1 to 8.

Images