Intelligent control method and system for production process of digital intelligent water dispersible granules

Abstract

This invention belongs to the field of intelligent control technology, specifically a digital intelligent control method and system for the production process of water-dispersible granules. The method includes: establishing basic process parameters; analyzing the influence of raw material physicochemical properties and process linkages on the basic process parameters to obtain process control parameters; adjusting the basic process parameters to obtain production control parameters; and collecting time-series production data. The method determines whether the time-series production data meets the target quality indicators; if so, it maintains the current production control parameters; otherwise, it optimizes the process control parameters. Feature extraction is performed on the time-series production data to obtain production deviation features. A hierarchical correlation analysis method is used to analyze the linkage production relationship between the basic process parameters and the production deviation features, and the process control parameters are optimized to obtain process correction data. This invention can significantly improve quality stability, production efficiency, and capacity, achieve coordinated control of the entire production process, and reduce the incidence of chain production deviations.

Description

Technical Field

[0001] This invention belongs to the field of intelligent control technology, specifically an intelligent control method and system for the digitalized production process of water-dispersible granules. Background Technology

[0002] As an environmentally friendly pesticide formulation, water-dispersible granules require high precision in process parameters during production. The quality of the finished product is easily affected by differences in the physicochemical properties of raw material batches and the coupling and linkage between processes. Current traditional production control methods are no longer suitable for the needs of high-quality industrial production.

[0003] In existing technologies, the formulation of basic process parameters largely relies on historical experience with similar products, without quantitative design incorporating actual physicochemical data of active ingredients and fillers, as well as the compatibility of the adjuvant system. When faced with differences in particle size, hygroscopicity, and compatibility between batches of raw materials, parameter adaptability is poor, easily leading to quality deviations such as dispersibility and disintegration. Furthermore, parameters for each production step are often controlled independently, ignoring the coupling effect of preceding steps on subsequent steps, and lacking an effective process linkage compensation mechanism, making it difficult to quickly resolve quality deviations. In addition, process operation data and quality monitoring data are collected separately during production, without time-sequential and process-specific integration, making real-time feedback on production status impossible. When quality deviations occur, relying solely on manual experience to identify causes and adjust parameters results in slow traceability and inaccurate control, easily leading to poor finished product quality stability and a high failure rate.

[0004] Therefore, there is an urgent need to develop an intelligent control method and system for the production process of water-dispersible granules to achieve precise, efficient, and intelligent management and control of the production process. Summary of the Invention

[0005] To overcome the shortcomings of existing technologies, this invention proposes an intelligent control method and system for the production process of water-dispersible granules. This invention primarily addresses the impact of batch-to-batch variations in raw materials and inter-process coupling effects on the quality of the finished product during the production of water-dispersible granules.

[0006] The technical solution adopted by this invention to solve its technical problem is: This invention provides an intelligent control method for the digitalized water-dispersible granule production process, comprising: Basic data on the active ingredient and filler and the compatibility data of the adjuvant system were collected. Based on the target quality indicators of the water-dispersible granules, basic process parameters were formulated. The linkage compensation and control method was used to analyze the influence of the physicochemical properties of the raw materials and the linkage relationship of the processes on the basic process parameters to obtain process control parameters.

[0007] Production control parameters are obtained by adjusting the basic process parameters based on the process control parameters. Process operation data and quality monitoring data of the entire production process are collected and fused to obtain time-series production data.

[0008] Determine whether the time-series production data meets the target quality indicators. If yes, maintain the current production control parameters; otherwise, optimize the process control parameters.

[0009] Feature extraction is performed on time-series production data to obtain production deviation features. The hierarchical correlation analysis method is used to analyze the linkage production relationship between basic process parameters and production deviation features. The process control parameters are then optimized to obtain process correction data.

[0010] This invention provides an intelligent control method for the production process of digitally-driven water-dispersible granules, the steps of which include formulating basic process parameters as follows: The target quality indicators of the finished product are determined based on the application scenarios of the water-dispersible granules, considering performance, physical stability, and compliance.

[0011] The physicochemical data and parameters of the active pharmaceutical ingredient are collected, and the basic data of the active pharmaceutical ingredient filler are obtained by combining the structural characteristics, physicochemical properties and processing adaptability of the filler carrier.

[0012] The physicochemical properties of different types of adjuvants were obtained, the compatibility of different adjuvant combinations with the target active ingredient and fillers was analyzed, and the compatibility data of the adjuvant system was obtained by combining the stability of different types of adjuvants under different temperatures and pressures.

[0013] Based on the target quality indicators, basic data of the active pharmaceutical ingredient and filler, and compatibility data of the auxiliary agent system, the initial process parameters for different stages are calculated for each process.

[0014] The historical best production parameters for similar water-dispersible granules were retrieved, compared with the initial process parameters, and the differences were analyzed and corrected to obtain the basic process parameters.

[0015] This invention provides an intelligent control method for the production process of water-dispersible granules, which includes the following steps for calculating the initial process parameters for different stages: The process is divided into four categories: feeding and mixing, grinding, granulation, and drying and sieving. Based on the target effective ingredient content and production capacity, the theoretical feeding amount of raw materials, fillers and various adjuvants is calculated. Combined with the model of the mixing equipment, the initial mixing speed and mixing time are calculated to obtain the feeding and mixing process parameters.

[0016] Based on the target powder particle size and the initial particle size of the active ingredient, combined with the grinding equipment parameters, the initial grinding speed, grinding media filling rate, and grinding flow rate are calculated as grinding process parameters. The die parameters for the corresponding granulation process are calculated according to different types of granulation methods.

[0017] Based on the target dry particle moisture content and the thermal stability of the active ingredient, the initial drying temperature, drying wind speed and drying time are set as drying parameters in stages. Based on the target particle size distribution, the initial aperture of the sieve, vibration frequency and initial wind speed of the air classifier are determined as sieving parameters.

[0018] This invention provides an intelligent control method for the production process of digitally-controlled water-dispersible granules, the steps of which include obtaining process control parameters as follows: Analyze the correspondence between the physicochemical properties of raw materials, fillers, and additives and the basic process parameters to determine the core influencing dimensions.

[0019] Based on the compatibility data of the additive system, the basic influence coefficient is obtained by adjusting the coefficients for the case of multiple raw material properties superimposed.

[0020] The linkage coupling type is determined based on the influence of the preceding process on the following process. Using the basic process parameters as a benchmark, the influence of the unit change of the preceding process parameters on the parameters of the following process is calculated to obtain the transmission coefficient.

[0021] Based on the determined linkage coupling type and transmission coefficient, construct the process linkage matrix, and determine the linkage rules and linkage trigger threshold.

[0022] Based on the raw material characteristic data and actual process parameter data of each process collected in real time during the production process, the raw material characteristic deviation is calculated by comparing it with the basic process parameters.

[0023] Based on the basic influence coefficient and the deviation of raw material characteristics, the initial control amount of the process parameters for a single process is calculated. The process control parameters are obtained by linking and correcting the initial control amount according to the process linkage matrix.

[0024] This invention provides an intelligent control method for the production process of digitally-controlled water-dispersible granules, the steps of adjusting and obtaining production control parameters including: The process control parameters are broken down by process dimension and control type, and matched one by one with the parameters of the same process and type in the basic process parameters to determine the correlation and correspondence.

[0025] Based on historical production data of water-dispersible granules and the rules governing the influence of raw material characteristics, the sensitivity coefficient of each type of control parameter is calculated, and the corresponding initial adjustment value is calculated in combination with the basic process parameters.

[0026] Based on the importance of preset quality indicators, the priority order of adjustment items is defined. Combined with the equipment safety operation parameters and process range, the maximum adjustment amount and adjustment rate for a single step are set, and the initial adjustment is corrected to obtain the compliant adjustment value.

[0027] Preliminary production control parameters are calculated based on the baseline and compliant adjustment values ​​of the basic process parameters, and then combined with the corresponding relationships to obtain the final production control parameters.

[0028] This invention provides an intelligent control method for the production process of digitally-driven water-dispersible granules, the steps of which include obtaining time-series production data as follows: The sampling frequency is set according to the production line speed and the response accuracy of the testing equipment. The equipment operation parameters and process operation parameters of the entire process are collected as process operation data, and the material status and quality indicators are collected as quality monitoring data.

[0029] Multi-source processed data is obtained by standardizing, deduplicating, reducing noise, and filling in missing data in process operation data and quality monitoring data.

[0030] Using the clock pulse of the main system of the production line as the sole time axis reference, synchronous calibration is performed, a unified timestamp is added to the multi-source treated water, and time series data is obtained by global sorting from morning to night.

[0031] Time-series data is binned, categorized, and associated according to production process nodes and production batches to obtain time-series production data.

[0032] The present invention provides an intelligent control method for the production process of digitally-driven water-dispersible granules, the step of extracting production deviation characteristics includes: Based on the basic process parameters, equipment safety operation thresholds and target quality indicators, dual benchmark values ​​are determined, and time-series production data are split into multi-dimensional corresponding data according to process dimension, parameter type and data attribute.

[0033] Based on multidimensional corresponding data, deviation features are extracted from the time series data of each process and each monitoring parameter in three dimensions: amplitude, time and frequency. Deviation attributes and deviation directions are labeled, and parameter deviation features are obtained by organizing them by process.

[0034] Based on parameter deviation characteristics, time series analysis is performed on the time-series deviation data of the same parameter and the same process to extract the trend, volatility and suddenness characteristics of deviation changes with production time to form time deviation characteristics.

[0035] Spatial decomposition of parameter deviation features based on process horizontality and parameter category is performed to extract process distribution features, parameter clustering features, and local defect features to form process deviation features.

[0036] Based on the material transfer relationships and process linkage relationships of each production process, cross-process deviation characteristics are extracted from the time-series production data.

[0037] By fusing and deduplicating time deviation features, process deviation features, and cross-process deviation features, a fused deviation feature is obtained. Weight levels and analysis priorities are assigned according to preset principles to form production deviation features.

[0038] This invention provides an intelligent control method for the production process of digitally integrated water-dispersible granules, the steps of which to obtain the linked production relationship include: The analysis dimensions are determined from three aspects: causality, quantification, and coupling, and a dual-object correlation analysis table between basic process parameters and production deviation characteristics is established.

[0039] The controlled variable method was used to analyze the causal relationship between changes in a single basic process parameter and a single production deviation characteristic. Causal associations were screened, and the association attributes and association levels were labeled to obtain a list of causal associations.

[0040] Based on the causal relationship list, related data samples are extracted from time-series production data, linear fitting is performed to construct a quantitative relationship model, and the change in the intensity of deviation characteristics is calculated as the linkage production relationship.

[0041] The present invention provides an intelligent control method for the production process of digitally-driven water-dispersible granules, the steps of which include obtaining process correction data are as follows: The production deviation characteristics are broken down by process dimension, deviation type and impact scope to obtain the current deviation characteristics, which are then matched with the causal relationship list to screen the inducing process parameters.

[0042] Based on the correlation level, the inducing process parameters are classified into inducing levels, and the direction of the deviation influence is marked to generate a deviation parameter mapping list.

[0043] Based on the quantitative correlation model and the change in the intensity of deviation characteristics, the initial correction amount of each inducing process parameter is calculated in reverse according to the actual intensity value of the production deviation characteristics.

[0044] For each initial correction amount, the equipment process constraint boundary is marked. For different coupling types, corresponding correction strategies are adopted, and the correction is performed with the target quality index as the priority to obtain process correction data.

[0045] This invention provides an intelligent control system for the digitalized production process of water-dispersible granules, comprising: The control parameter generation module is used to collect basic data of active pharmaceutical ingredients and fillers and data on the compatibility of adjuvant systems. It combines the target quality indicators of water-dispersible granules to formulate basic process parameters. The linkage compensation control method is used to analyze the influence of the physicochemical properties of raw materials and the linkage relationship of processes on the basic process parameters to obtain process control parameters.

[0046] The execution data fusion module is used to adjust the basic process parameters according to the process control parameters to obtain production control parameters, collect process operation data and quality monitoring data of the entire production process, and perform fusion processing to obtain time-series production data.

[0047] The quality judgment and decision module is used to determine whether the time-series production data has reached the target quality indicators. If so, the current production control parameters are maintained; otherwise, the process control parameters are optimized.

[0048] The parameter optimization and analysis module is used to extract features from time-series production data to obtain production deviation features. It uses a hierarchical correlation analysis method to analyze the linkage production relationship between basic process parameters and production deviation features, and optimizes process control parameters to obtain process correction data.

[0049] The beneficial effects of this invention are as follows: 1. This invention collects comprehensive raw material data, calculates initial process parameters for each process step, and corrects them using historical data, ensuring a high degree of compatibility between basic process parameters and target quality indicators and raw material characteristics. Furthermore, through a linkage compensation control method and a hierarchical correlation analysis method, it achieves quantitative calculation of process control and correction parameters, eliminating reliance on manual experience and significantly improving the scientific rigor and adaptability of process parameters. By constructing a process linkage matrix, it quantifies the transmission coefficient from preceding to subsequent processes, achieving cross-process collaborative compensation for parameter adjustments. This effectively avoids subsequent chain deviations caused by single-process parameter deviations, significantly reducing the incidence of deviations in the production process and improving its stability. By extracting production deviation characteristics from multiple dimensions and combining the hierarchical correlation analysis method, it accurately locates the core inducing parameters of deviations from a causal and quantitative perspective, achieving precise and rapid resolution of deviations and significantly improving the consistency and stability of the finished water-dispersible granules. Attached Figure Description

[0050] The invention will now be further described with reference to the accompanying drawings.

[0051] Figure 1 This is a flowchart illustrating an intelligent control method for the digitalized water-dispersible granule production process provided in an embodiment of the present invention. Figure 2 This is a flowchart illustrating the process of obtaining production control parameters in an intelligent control method for the production process of digitally intelligent water-dispersible granules provided in an embodiment of the present invention. Figure 3 This is a schematic diagram of a module of an intelligent control system for a digitalized water-dispersible granule production process provided in an embodiment of the present invention. Detailed Implementation

[0052] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0053] like Figures 1 to 3 As shown in the figure, an intelligent control method for the production process of digitally-controlled water-dispersible granules provided by an embodiment of the present invention includes: Basic data on the active ingredient and filler and the compatibility data of the adjuvant system were collected. Based on the target quality indicators of the water-dispersible granules, basic process parameters were formulated. The linkage compensation and control method was used to analyze the influence of the physicochemical properties of the raw materials and the linkage relationship of the processes on the basic process parameters to obtain process control parameters.

[0054] The steps for determining basic process parameters include: The target quality indicators of the finished product are determined based on the application scenarios of the water-dispersible granules, considering performance, physical stability, and compliance.

[0055] Performance aspects may include: target particle size distribution, disintegration time, suspension rate, and dry particle moisture content.

[0056] Physical stability can include: particle hardness, dust-free properties, non-caking during storage, and sieve pass rate.

[0057] Compliance aspects may include: deviations in the content of active ingredients, impurity content, and compliance with relevant pesticide registration standards.

[0058] The physicochemical data and parameters of the active pharmaceutical ingredient are collected, and the basic data of the active pharmaceutical ingredient filler are obtained by combining the structural characteristics, physicochemical properties and processing adaptability of the filler carrier.

[0059] The physicochemical data of the original drug may include: initial particle size, crystal form, such as whether it is stable, whether there is a risk of polymorphic transformation, solubility such as water insolubility / hydrophobicity, thermal stability, and hygroscopicity.

[0060] Parameters may include: active ingredient content, density, flowability, and compatibility with common additives.

[0061] Structural characteristics can include: specific surface area, such as the large specific surface area and strong adsorption of silica, porosity, and particle size distribution.

[0062] Physicochemical properties: density, flowability, such as the density difference with the original drug needs to be controlled to avoid mixing and layering; chemical inertness, such as whether it reacts weakly with the original drug / adjuvant.

[0063] Processing compatibility: Formability and dispersibility. For example, kaolin has good formability but poor dispersibility, so it needs to be matched with corresponding additives.

[0064] The physicochemical properties of different types of adjuvants were obtained, the compatibility of different adjuvant combinations with the target active ingredient and fillers was analyzed, and the compatibility data of the adjuvant system was obtained by combining the stability of different types of adjuvants under different temperatures and pressures.

[0065] Different types of adjuvants can include: dispersants (e.g., type, dosage range), disintegrants (e.g., water absorption and swelling rate, compatible with active ingredient type), wetting agents (e.g., surface tension regulating ability), binders (e.g., molding effect, without affecting disintegration), and anti-caking agents (e.g., physicochemical properties of moisture-proof effect). The compatibility of different adjuvant combinations with the target active ingredient and fillers is also important, such as the dispersion effect of a certain dispersant on a specific hydrophobic active ingredient, the impact of adjuvant dosage on quality indicators (e.g., excessive binder prolongs disintegration time), and interfacial parameters between adjuvants and active ingredients to reduce the risk of particle agglomeration.

[0066] Based on the target quality indicators, basic data of the active pharmaceutical ingredient and filler, and compatibility data of the auxiliary agent system, the initial process parameters for different stages are calculated for each process.

[0067] The steps for calculating the initial process parameters for different stages of a process include: The process is divided into four categories: feeding and mixing, grinding, granulation, and drying and sieving. Based on the target effective ingredient content and production capacity, the theoretical feeding amount of raw materials, fillers and various adjuvants is calculated. Combined with the model of the mixing equipment, the initial mixing speed and mixing time are calculated to obtain the feeding and mixing process parameters.

[0068] Based on the target powder particle size and the initial particle size of the active ingredient, combined with the grinding equipment parameters, the initial grinding speed, grinding media filling rate, and grinding flow rate are calculated as grinding process parameters. The die parameters for the corresponding granulation process are calculated according to different types of granulation methods.

[0069] Calculate the initial viscosity and shear rate of the slurry, as well as the atomization pressure, initial hot air flow rate, and initial gap of the granulation die.

[0070] Dry granulation: Calculate the initial screw speed and die orifice diameter based on the target particle hardness and particle size to avoid hard cores or particle breakage.

[0071] Based on the target dry particle moisture content and the thermal stability of the active ingredient, the initial drying temperature, drying wind speed and drying time are set as drying parameters in stages. Based on the target particle size distribution, the initial aperture of the sieve, vibration frequency and initial wind speed of the air classifier are determined as sieving parameters.

[0072] The historical best production parameters for similar water-dispersible granules were retrieved, compared with the initial process parameters, and the differences were analyzed and corrected to obtain the basic process parameters.

[0073] The steps to obtain process control parameters include: Analyze the correspondence between the physicochemical properties of raw materials, fillers, and additives and the basic process parameters to determine the core influencing dimensions.

[0074] The core influencing dimensions may include the crystal stability, thermal stability, hygroscopicity, hydrophobicity, and initial particle size of the active ingredient; the specific surface area, density, and flowability of the filler; and the dispersion efficiency, disintegration and expansion rate, and viscosity-modifying ability of the adjuvant.

[0075] The corresponding relationship can include: strong hydrophobicity of the active ingredient → pulping / granulation process → amount of dispersant added, viscosity of the liquid → positive effect, need to increase dispersant and increase viscosity → base coefficient 0.15, for each grade of hydrophobicity increase, the amount of dispersant added increases by 15%.

[0076] Based on the compatibility data of the additive system, the basic influence coefficient is obtained by adjusting the coefficients for the case of multiple raw material properties superimposed.

[0077] The linkage coupling type is determined based on the influence of the preceding process on the following process. Using the basic process parameters as a benchmark, the influence of the unit change of the preceding process parameters on the parameters of the following process is calculated to obtain the transmission coefficient.

[0078] The linkage coupling types are divided into material status transmission type and equipment operation adaptation type.

[0079] Based on the determined linkage coupling type and transmission coefficient, construct the process linkage matrix, and determine the linkage rules and linkage trigger threshold.

[0080] Based on the raw material characteristic data and actual process parameter data of each process collected in real time during the production process, the raw material characteristic deviation is calculated by comparing it with the basic process parameters.

[0081] Based on the basic influence coefficient and the deviation of raw material characteristics, the initial control amount of the process parameters for a single process is calculated. The process control parameters are obtained by linking and correcting the initial control amount according to the process linkage matrix.

[0082] Initial control amount = raw material characteristic deviation amount × corresponding basic influence coefficient × basic process parameter benchmark value. For example: raw material hydrophobicity deviation amount + 20% × basic influence coefficient 0.15 × basic dispersant addition amount 5kg = initial control amount + 0.15kg, requiring an additional 0.15kg of dispersant.

[0083] If the process is a preceding process, calculate the linkage control amount of its initial control amount on the subsequent process based on the linkage transmission matrix (linkage control amount = initial control amount of the preceding process × transmission coefficient). If the process is a subsequent process, the passive control amount brought about by the deviation transmission of the preceding process needs to be added to obtain the actual control amount of the single process.

[0084] For each individual process, the actual control quantities are marked with the equipment safety threshold and the process feasible range.

[0085] For process parameters that have overlapping effects, such as the amount of dispersant added affecting both pulping and granulation processes, the actual control amounts of each process are superimposed to calculate the comprehensive control amount.

[0086] If there are conflicts in the control requirements of different processes, such as increasing the binder to improve granulation formability, while reducing the binder to ensure disintegration time, the core quality indicators of water-dispersible granules should be prioritized. If disintegration time is a key indicator, then the control should be adjusted in the direction of reducing the binder. The adjustment should be made by combining the data on the compatibility of additives.

[0087] Production control parameters are obtained by adjusting the basic process parameters based on the process control parameters. Process operation data and quality monitoring data of the entire production process are collected and fused to obtain time-series production data.

[0088] The steps to adjust and obtain production control parameters include: The process control parameters are broken down by process dimension and control type, and matched one by one with the parameters of the same process and type in the basic process parameters to determine the correlation and correspondence.

[0089] The process dimension corresponds to processes such as mixing, grinding, and pulping. Control types are categorized into raw material proportioning (e.g., dispersant addition amount, feed ratio), equipment operation (e.g., rotation speed, pressure, temperature, air velocity), and material state (e.g., liquid viscosity, solids content). Corresponding relationships exist; for example, the "grinding particle size optimization" control parameter requires simultaneous adjustment of two basic parameters: grinding speed and grinding media filling rate.

[0090] Based on historical production data of water-dispersible granules and the rules governing the influence of raw material characteristics, the sensitivity coefficient of each type of control parameter is calculated, and the corresponding initial adjustment value is calculated in combination with the basic process parameters.

[0091] Initial adjustment value = adjustment range of process control parameter × corresponding sensitivity coefficient × baseline value of basic process parameter.

[0092] Based on the importance of preset quality indicators, the priority order of adjustment items is defined. Combined with the equipment safety operation parameters and process range, the maximum adjustment amount and adjustment rate for a single step are set, and the initial adjustment is corrected to obtain the compliant adjustment value.

[0093] Prioritization can include: prioritizing all control parameters according to "direct impact on core quality indicators, process response speed, and difficulty of equipment adjustment", and classifying them into first-level controls that must be adjusted immediately, such as excessive grinding particle size or drying temperature exceeding the heat tolerance temperature of the original drug.

[0094] Secondary control allows for gradual adjustments, such as for minor deviations in the viscosity of the liquid.

[0095] Three-level control requires only monitoring and no immediate adjustment, such as for minor fluctuations in the vibration frequency during the screening process.

[0096] Maximum single-step adjustment, such as maximum single-step adjustment of grinding speed ≤ 5 r / min, maximum single-step adjustment of drying temperature ≤ 3℃. Adjustment rate, such as temperature parameters ≤ 1℃ per minute, speed parameters ≤ 2 r / min per minute.

[0097] If the overall initial adjustment value of a certain basic process parameter exceeds the maximum adjustment amount in a single step, it will be corrected to the maximum adjustment amount in a single step. The remaining unadjusted part will be marked as "to be adjusted in subsequent iterations" and included in the next round of parameter optimization, rather than being adjusted all at once.

[0098] Preliminary production control parameters are calculated based on the baseline and compliant adjustment values ​​of the basic process parameters, and then combined with the corresponding relationships to obtain the final production control parameters.

[0099] Preliminary production control parameter value = baseline value of basic process parameters + compliance adjustment value. When the adjustment direction is to decrease, the compliance adjustment value is negative, that is, the baseline value minus the corresponding value.

[0100] If it is a "one-to-many" correspondence, the adjustment values ​​of multiple related basic process parameters are synergistically superimposed according to the correlation of the control parameters. For example, when adjusting the die head aperture and screw speed simultaneously for the "optimize granulation particle size" control parameter, it is necessary to ensure that the adjustment ranges of the two are matched to avoid the particle size fluctuating, rather than being independently superimposed.

[0101] The steps to obtain time-series production data include: The sampling frequency is set according to the production line speed and the response accuracy of the testing equipment. The equipment operation parameters and process operation parameters of the entire process are collected as process operation data, and the material status and quality indicators are collected as quality monitoring data.

[0102] Process operation data can include: equipment operating parameters covering the entire process, such as mixer speed / current, grinding media filling rate / cooling water temperature of sand mill, process operation parameters, such as feed rate / ratio, pulping shear rate, granulation atomization pressure, drying hot air temperature / speed, and material conveying parameters, such as feed / discharge speed and material level for each process, which are numerical continuous data.

[0103] Quality monitoring data can include: key indicators covering intermediate and finished products at each stage of the process, such as uniformity of powder after mixing, D90 particle size / dispersibility of powder after grinding, viscosity / solid content of slurry, particle size / moisture content of wet granules, moisture content / hardness of dry granules, disintegration time / suspension rate / sieve pass rate of finished products, including numerical continuous data such as viscosity and temperature, and detection discrete data such as disintegration time and uniformity rating.

[0104] Multi-source processed data is obtained by standardizing, deduplicating, reducing noise, and filling in missing data in process operation data and quality monitoring data.

[0105] Data standardization processing: Numerical data: Standardize the dimensions according to the normalization / standardization formula, such as converting temperature (°C), rotation speed (r / min), and viscosity (mPa·s) into normalized values ​​between 0 and 1, or converting them into standardized values ​​according to industry standards, to avoid the impact of dimensional differences on subsequent analysis.

[0106] Discrete / character data: Digital encoding is performed, such as equipment status "normal=0, fault=1", powder uniformity "excellent=3, good=2, qualified=1, unqualified=0", to realize the quantification of non-numerical data.

[0107] Unified data format: All data is named according to the format of **"Collection Timestamp-Process Name-Location Number-Data Type-Value / Code-Collection Device"**, generating a unified raw data structure table.

[0108] Deduplication: Using "collection timestamp + location number" as a unique identifier, duplicate data is removed, and the first collection value is retained.

[0109] Noise reduction: For continuous data acquired by the sensor, moving average / median filtering methods are used to remove sudden noise points, such as abnormally high / low values ​​caused by instantaneous fluctuations in the sensor, while preserving the true trend of data change. For offline detection data, manually entered errors that significantly deviate from the reasonable range are removed.

[0110] Random minor missing data: Use adjacent data interpolation methods, such as linear interpolation or moving average interpolation, to fill in the missing data.

[0111] Continuous missing: Complete by combining the linkage rules of the process (e.g., if the grinding cooling water temperature is missing, it can be deduced and completed based on the changing trends of grinding speed and feed flow rate).

[0112] Missing offline detection data: Mark it as "not detected" and record the reason for the missing data, such as equipment calibration or delays in manual operation. Do not force the data to be filled in, and preserve the authenticity of the data.

[0113] Using the clock pulse of the main system of the production line as the sole time axis reference, synchronous calibration is performed, a unified timestamp is added to the multi-source treated water, and time series data is obtained by global sorting from morning to night.

[0114] Time-series data is binned, categorized, and associated according to production process nodes and production batches to obtain time-series production data.

[0115] Each batch of water-dispersible granules is assigned a unique batch number according to the production plan, and the start and end timestamps of the production for that batch are marked in the data system. On the timeline of each batch, the start and end timestamps and node identifiers of each process are marked.

[0116] The time series data is hierarchically binned according to the hierarchy of "batch number → process name → collection point". That is, all the collected data of the same batch and the same process are grouped into a data subset. Each subset retains the original timestamp and data value, forming a process-based data subset set.

[0117] Based on the time delay of material transportation, establish a correlation mapping between the data of the preceding and following processes. For example, the discharge time of the mixing process corresponds to the feeding time of the grinding process, and the discharge time of wet granules in the granulation process corresponds to the feeding time of the drying process. Mark the material transfer timestamps between processes to enable the data sets of each process to achieve orderly linkage according to the production process chain.

[0118] Determine whether the time-series production data meets the target quality indicators. If yes, maintain the current production control parameters; otherwise, optimize the process control parameters.

[0119] Feature extraction is performed on time-series production data to obtain production deviation features. The hierarchical correlation analysis method is used to analyze the linkage production relationship between basic process parameters and production deviation features. The process control parameters are then optimized to obtain process correction data.

[0120] The steps for extracting production deviation characteristics include: Based on the basic process parameters, equipment safety operation thresholds and target quality indicators, dual benchmark values ​​are determined, and time-series production data are split into multi-dimensional corresponding data according to process dimension, parameter type and data attribute.

[0121] Based on multidimensional corresponding data, deviation features are extracted from the time series data of each process and each monitoring parameter in three dimensions: amplitude, time and frequency. Deviation attributes and deviation directions are labeled, and parameter deviation features are obtained by organizing them by process.

[0122] Amplitude categories: peak deviation, average deviation, and deviation range.

[0123] Time-related: duration of deviation, time of occurrence of deviation, and period of deviation interval.

[0124] Frequency-based: Number of times deviations occurred, percentage of serious deviations.

[0125] Deviation attributes can include process operation deviations / quality monitoring deviations and deviation directions, which can include positive deviations (actual value > reference value, such as drying temperature 75℃ > reference 70℃) and negative deviations (actual value < reference value, such as dispersant addition amount 3kg < reference 4kg).

[0126] Based on parameter deviation characteristics, time series analysis is performed on the time-series deviation data of the same parameter and the same process to extract the trend, volatility and suddenness characteristics of deviation changes with production time to form time deviation characteristics.

[0127] Extracting trend deviation features: For single-parameter deviation time-series data within a preset production period, the deviation change curve is fitted using linear regression / moving average methods to extract trend features. Trend type, such as linear increase / linear decrease / no obvious trend, such as grinding speed slowly decreasing over time, and particle size deviation increasing linearly.

[0128] Trend slope, quantifying the rate of change of deviation, such as a 2% increase in relative particle size deviation every 10 minutes, with a slope of 0.2% / min.

[0129] The duration of the trend is the total time during which the deviation changes according to a fixed trend, such as the viscosity deviation continuously increasing for 30 minutes.

[0130] For biased data exhibiting periodic fluctuations, Fourier transform / analysis of variance are used to extract fluctuation characteristics: Fluctuation type: regular periodic fluctuation / irregular random fluctuation, such as regular fluctuations in granulation atomization pressure every 5 minutes due to unstable equipment air pressure.

[0131] Fluctuation characteristics, fluctuation frequency, fluctuation amplitude, variance / standard deviation, such as amplitude ±3% and frequency 0.2 times / min.

[0132] Consistency of fluctuations: Whether the fluctuations of multiple parameters within the same process are synchronized, such as whether the fluctuations of drying temperature and the moisture content of dry particles are synchronized.

[0133] For sudden and abrupt deviations in data, such as instantaneous occurrences of excessive grinding particle size or sudden pressure increases due to clogging of the granulation die, extract the sudden characteristics: Sudden triggering nodes, precise timestamps + process operation nodes, such as a sudden deviation in particle size after the die head is changed in the granulation process at 10:20.

[0134] Sudden deviation magnitude, the difference in deviation before and after the jump, such as when the pressure suddenly rises from 0.5MPa to 0.8MPa, the difference is 0.3MPa.

[0135] Sudden recovery characteristics: whether the system recovers automatically or requires manual intervention after a sudden deviation, and the recovery time.

[0136] Spatial decomposition of parameter deviation features based on process horizontality and parameter category is performed to extract process distribution features, parameter clustering features, and local defect features to form process deviation features.

[0137] Extract process distribution characteristics: Statistically analyze the percentage of effective deviations across the entire process range, such as 40% for grinding, 30% for granulation, and 20% for drying. Also analyze the severity of process deviations, such as the number / percentage of severe deviations in each process, and mark processes with high deviation incidence and critical deviation processes.

[0138] Extracting parameter clustering features: performing correlation analysis on deviation parameters within each process, and extracting parameter clustering patterns: independent deviation of a single parameter, such as only the grinding cooling water temperature deviating while other parameters are normal.

[0139] Multi-parameter correlated aggregation deviations, such as simultaneous deviations in viscosity, solids content, and shear rate during the pulping process, are considered correlated aggregation deviations. For correlated aggregation deviations, the core triggering parameter is marked; for example, viscosity deviation triggers solids content and shear rate deviations, with viscosity being the core triggering parameter.

[0140] Extracting local defect features: For local operation nodes / equipment points in WDG production processes, extract deviation features caused by local defects, such as uneven mixing due to wear of mixer blades, particle size distribution deviation due to local blockage of granulation die head, and excessive moisture content due to uneven distribution of hot air in drying fluidized bed: local defect location, local deviation range, and local deviation transferability.

[0141] Based on the material transfer relationships and process linkage relationships of each production process, cross-process deviation characteristics are extracted from the time-series production data.

[0142] Based on the production process chain of water-dispersible granules (WDG), three core deviation transmission links are identified: mixing → grinding → pulping, pulping → granulation → drying, and drying → screening → finished product. The material transmission time delay of each link is marked, such as the deviation in the granulation process being transmitted to the drying process after 5 minutes.

[0143] For cross-process deviations in each transmission link, the transmission characteristics can be extracted, which may include: The source process / parameter is the initial process / parameter of the deviation, such as the grinding particle size deviation as the source.

[0144] The receiving process / parameters are transmitted, and the subsequent receiving process / parameters for deviations are received, such as the viscosity of the slurry.

[0145] Deviation transmission coefficient is the ratio of the source deviation amplitude to the receiving deviation amplitude, quantifying the magnitude of the transmitted influence. For example, if a 10% deviation in grinding particle size leads to an 8% viscosity deviation, the deviation transmission coefficient is 0.8.

[0146] Transmission delay time is the time difference between the occurrence of source deviation and the occurrence of reception deviation.

[0147] The transmission attenuation / amplification characteristics mean that the amplitude is attenuated / maintained / amplified during the deviation transmission process. For example, if the particle size deviation is 10% after grinding, the particle size deviation will be 15% after pulping and granulation, which is deviation amplification.

[0148] For the phenomenon of deviation linkage between multiple processes, such as the simultaneous deviation of particle size-related parameters in grinding, pulping, and granulation processes, the linkage characteristics are extracted: linkage trigger source, such as the grinding particle size as the linkage trigger source.

[0149] Synchronization and linkage: whether the deviations in each process occur at the same time, and whether there is a time difference.

[0150] Linkage amplification factor is the ratio of the total deviation amplitude after multi-process linkage to the single-source deviation amplitude.

[0151] By fusing and deduplicating time deviation features, process deviation features, and cross-process deviation features, a fused deviation feature is obtained. Weight levels and analysis priorities are assigned according to preset principles to form production deviation features.

[0152] Cross-validation of deviation features across the three dimensions is performed to eliminate redundant features, such as the trend deviation of grinding particle size in the time dimension and the particle size deviation transmission source features in the cross-process dimension, which are redundant features. The core transmission source features are retained, and complementary features are fused, such as the fusion of local defect features in granulation in the spatial dimension and sudden deviation features in particle size in the time dimension, thus fusing deviation features.

[0153] Core Level 1 characteristics: Deviation characteristics that directly determine the quality of finished products, have clear root causes, and affect multiple processes, such as excessive grinding particle size D90 leading to particle size-related deviations in subsequent processes, or clogged granulation die head causing unqualified particle size distribution in finished products.

[0154] Important secondary characteristics: deviation characteristics caused by factors within the process and local root causes, such as excessive moisture content in the drying area and deviation of air velocity in the screening air classifier.

[0155] Typical Level 3 characteristics: independent deviation of a single parameter, deviation characteristics without cross-process influence, such as small deviation of mixer current and small fluctuation of cooling water temperature.

[0156] The steps to obtain the linkage production relationship include: The analysis dimensions are determined from three aspects: causality, quantification, and coupling, and a dual-object correlation analysis table between basic process parameters and production deviation characteristics is established.

[0157] Causal dimension: Determine whether the change in a single basic process parameter is the direct / indirect cause of a certain production deviation characteristic. For example, a decrease in grinding speed → excessive grinding particle size D90 is a direct deviation → an increase in the viscosity of the slurry is an indirect deviation.

[0158] Quantitative dimension: Calculate the unit change of basic process parameters and the corresponding intensity change of the production deviation characteristics. For example, for every 5 r / min decrease in grinding speed, the relative deviation of particle size D90 increases by 3%.

[0159] Coupling dimension: Analyze how the synergistic changes of multiple basic process parameters lead to the coupling, superposition, or amplification of production deviation characteristics. For example, an increase in the diameter of the granulation die head + an increase in the screw speed → an amplification of the deviation of wet particle size exceeding the standard, with the relative deviation increasing from 5% to 12%.

[0160] The controlled variable method was used to analyze the causal relationship between changes in a single basic process parameter and a single production deviation characteristic. Causal associations were screened, and the association attributes and association levels were labeled to obtain a list of causal associations.

[0161] If, after a parameter change, a certain deviation characteristic changes from "no deviation" to "effective deviation," or the deviation intensity increases significantly, such as the relative deviation increasing from 3% to 8%, then the parameter is determined to be the direct cause of the deviation characteristic.

[0162] If the deviation characteristics do not change significantly or only fluctuate slightly after the parameter is changed, it is determined that the parameter and the deviation characteristics are not directly causally related.

[0163] If a parameter change indirectly causes other deviations, such as increased pulp viscosity → poor granulation discharge → wet particle size deviation, then the parameter is determined to be an indirect cause of subsequent deviations, and the deviation transmission chain is marked.

[0164] If the time point at which the deviation occurs in the case happens to correspond to the change in a certain basic process parameter from the baseline value, and the deviation disappears after the parameter is restored to the baseline, the causal relationship is verified.

[0165] If the parameter change and deviation are not synchronized in time, or if the deviation still exists after the parameter is restored, the causal relationship of the parameter should be ruled out, and other possible causes should be traced back.

[0166] The correlation attributes can include direct and indirect causes. The correlation level can be categorized as follows: Core cause: parameter change is the main reason for the deviation, accounting for ≥70%. Secondary cause: parameter change is a secondary cause of the deviation, accounting for 30%~70%.

[0167] Based on the causal relationship list, related data samples are extracted from time-series production data, linear fitting is performed to construct a quantitative relationship model, and the change in the intensity of deviation characteristics is calculated as the linkage production relationship.

[0168] The steps to obtain process correction data include: The production deviation characteristics are broken down by process dimension, deviation type and impact scope to obtain the current deviation characteristics, which are then matched with the causal relationship list to screen the inducing process parameters.

[0169] Core inducing parameters: The main parameters that cause deviations, with a correlation ratio of ≥70%. For example, a decrease in grinding speed is a core inducing factor for particle size exceeding the standard.

[0170] Secondary contributing parameters: auxiliary parameters that cause deviations, accounting for 30% to 70% of the total. For example, a low grinding media filling rate is a secondary contributing factor to excessive particle size.

[0171] Indirectly related parameters: Adaptation parameters transferred across processes, such as pulping shear rate, which is a subsequent adaptation adjustment parameter for grinding particle size deviation. The direction of influence of each parameter's deviation is also indicated; for example, grinding speed and particle size deviation are negatively correlated: decreasing speed → increasing particle size deviation.

[0172] Based on the correlation level, the inducing process parameters are classified into inducing levels, and the direction of the deviation influence is marked to generate a deviation parameter mapping list.

[0173] Based on the quantitative correlation model and the change in the intensity of deviation characteristics, the initial correction amount of each inducing process parameter is calculated in reverse according to the actual intensity value of the production deviation characteristics.

[0174] Initial relative correction of parameter = actual relative strength of production deviation / linkage coefficient between the parameter and the deviation.

[0175] Initial absolute correction of parameters = basic process value of parameters × initial relative correction of parameters.

[0176] Among them: the correction direction is determined by the direction of the parameter's influence. For example, for negatively correlated parameters, the deviation increases in a positive direction → the parameter needs to be corrected in a positive direction: particle size deviation +8%, grinding speed linkage coefficient -0.4 → initial relative correction amount of speed = +8% / -0.4 = +20%, that is, the speed increases by 20%.

[0177] The actual strength value of the deviation is taken as the relative deviation value / trend slope / deviation peak value (selected according to the deviation characteristic type, such as the relative deviation value for amplitude-type deviations and the trend slope for trend-type deviations).

[0178] The correction amount is calculated separately for each parameter type. For cases where multiple deviation characteristics share the same inducing parameter, such as grinding particle size deviation simultaneously causing pulp viscosity deviation and granulation particle size deviation, a weighted summation of deviation intensity is used to calculate the comprehensive correction amount. The formula is: Comprehensive initial correction amount = Σ (Actual intensity value of each deviation × Deviation weight) / Linkage coefficient. The deviation weight is set according to the degree of influence of the deviation on the finished product quality; the deviation weight for core quality indicators is 1, and for secondary indicators it is 0.5.

[0179] The initial correction amount for each parameter is marked with the equipment process constraint boundary, that is, the maximum / minimum allowable adjustment value of the parameter. For example, the maximum adjustment of the grinding speed in a single step is ≤5r / min, and the maximum adjustment of the drying temperature is ≤3℃. The correction amount exceeding the boundary is marked as "exceeding the threshold" and will be corrected uniformly later.

[0180] For each initial correction amount, the equipment process constraint boundary is marked. For different coupling types, corresponding correction strategies are adopted, and the correction is performed with the target quality index as the priority to obtain process correction data.

[0181] Deviation amplification coupling: The correction amount of each parameter is reduced by the amplification factor. The formula is: Coupling correction amount = Initial correction amount / Amplification factor. This avoids excessive correction of deviation caused by the coordinated adjustment of multiple parameters, which may lead to new reverse deviations.

[0182] Deviation cancellation type coupling: The correction amount of the core parameter is increased by the amplification factor. The formula is: Core parameter coupling correction amount = initial correction amount / amplification factor. The secondary parameters are kept at the initial correction amount to make up for the cancellation of the correction effects between parameters.

[0183] No obvious coupling: directly retain the initial correction amount of each parameter, and mark the synchronous adjustment requirements, such as the pulp solids content and viscosity need to be adjusted synchronously, to avoid new deviations caused by the adjustment of a single parameter.

[0184] Based on the same general inventive concept, this invention also protects an intelligent control system for a digitalized water-dispersible granule production process, comprising: The control parameter generation module is used to collect basic data of active pharmaceutical ingredients and fillers and data on the compatibility of adjuvant systems. It combines the target quality indicators of water-dispersible granules to formulate basic process parameters. The linkage compensation control method is used to analyze the influence of the physicochemical properties of raw materials and the linkage relationship of processes on the basic process parameters to obtain process control parameters.

[0185] The execution data fusion module is used to adjust the basic process parameters according to the process control parameters to obtain production control parameters, collect process operation data and quality monitoring data of the entire production process, and perform fusion processing to obtain time-series production data.

[0186] The quality judgment and decision module is used to determine whether the time-series production data has reached the target quality indicators. If so, the current production control parameters are maintained; otherwise, the process control parameters are optimized.

[0187] The parameter optimization and analysis module is used to extract features from time-series production data to obtain production deviation features. It uses a hierarchical correlation analysis method to analyze the linkage production relationship between basic process parameters and production deviation features, and optimizes process control parameters to obtain process correction data.

[0188] In summary, this embodiment provides an intelligent control method and system for the digitalized production process of water-dispersible granules. This method enables the scientific and quantitative formulation and control of process parameters, improves parameter adaptability, and provides coordinated control across the entire production process, reducing the incidence of chain-like production deviations. The precise quantitative control of process parameters avoids problems such as raw material waste, equipment idling, and rework of finished products caused by improper parameters. The reduction in deviation rate and the improvement in resolution efficiency effectively shorten the production cycle and reduce the generation of defective products. While improving production efficiency, it significantly reduces production costs such as raw materials, energy consumption, and labor. The target quality indicators can be flexibly adjusted according to the application scenario of the water-dispersible granules, and it is adaptable to the production of water-dispersible granules with different combinations of active ingredients, fillers, and additives, exhibiting good process adaptability and raw material compatibility.

[0189] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0190] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent control of a digitalized water-dispersible granule production process, characterized in that: Collect basic data on the active ingredient and filler and data on the compatibility of the adjuvant system. Combine the target quality indicators of the water-dispersible granules to formulate basic process parameters. Use the linkage compensation and control method to analyze the influence of the physicochemical properties of the raw materials and the linkage relationship of the process on the basic process parameters to obtain process control parameters. Based on the process control parameters, the basic process parameters are adjusted to obtain production control parameters. Process operation data and quality monitoring data of the entire production process are collected and fused to obtain time-series production data. Determine whether the time-series production data meets the target quality index. If yes, maintain the current production control parameters; otherwise, optimize the process control parameters. The production deviation features are obtained by extracting features from the time-series production data. The hierarchical correlation analysis method is used to analyze the linkage production relationship between the basic process parameters and the production deviation features. The process control parameters are then optimized to obtain process correction data.

2. The intelligent control method for the production process of digitally-controlled water-dispersible granules according to claim 1, characterized in that: The steps for determining the basic process parameters include: The target quality indicators of the finished product are determined based on performance, physical stability, and compliance, according to the application scenario of the water-dispersible granules. The physicochemical data and parameters of the active pharmaceutical ingredient are collected, and the basic data of the active pharmaceutical ingredient filler are obtained by combining the structural characteristics, physicochemical properties and processing adaptability of the filler carrier. The physicochemical properties of different types of adjuvants are obtained, the compatibility of different adjuvant combinations with the target active ingredient and filler is analyzed, and the compatibility data of the adjuvant system is obtained by combining the stability of different types of adjuvants under different temperatures and pressures. Based on the target quality indicators, the basic data of the active pharmaceutical ingredient and filler, and the compatibility data of the adjuvant system, the initial process parameters for different stages are calculated for each process step. The historical best production parameters for similar water-dispersible granules are retrieved, compared with the initial process parameters, and the differences are analyzed and corrected to obtain the basic process parameters.

3. The intelligent control method for the production process of digitally-controlled water-dispersible granules according to claim 2, characterized in that: The steps for calculating the initial process parameters for different stages of a process include: The process is divided into four categories: feeding and mixing, grinding, granulation and drying and sieving. Based on the target effective ingredient content and production capacity, the theoretical feeding amount of raw materials, fillers and various adjuvants is calculated. Combined with the model of the mixing equipment, the initial mixing speed and mixing time are calculated to obtain the feeding and mixing process parameters. Based on the target powder particle size and the initial particle size of the active ingredient, combined with the grinding equipment parameters, the initial grinding speed, grinding media filling rate and grinding flow rate are calculated as grinding process parameters. The die head parameters for the corresponding granulation process are calculated according to different types of granulation methods. Based on the target dry particle moisture content and the thermal stability of the active ingredient, the initial drying temperature, drying wind speed and drying time are set as drying parameters in stages. Based on the target particle size distribution, the initial aperture of the sieve, vibration frequency and initial wind speed of the air classifier are determined as sieving parameters.

4. The intelligent control method for the production process of digitally-controlled water-dispersible granules according to claim 1, characterized in that: The steps for obtaining the process control parameters include: Analyze the physicochemical properties of raw materials, fillers, and additives and their correspondence with the basic process parameters to determine the core influencing dimensions; Based on the compatibility data of the aforementioned auxiliary agent system, the basic influence coefficient is obtained by adjusting the coefficients for the case of superimposed characteristics of multiple raw materials. The linkage coupling type is determined based on the influence of the preceding process on the following process. Using the basic process parameters as a benchmark, the influence of the unit change of the preceding process parameters on the parameters of the following process is calculated to obtain the transmission coefficient. Based on the determined linkage coupling type and the transmission coefficient, a process linkage matrix is ​​constructed, and linkage rules and linkage trigger thresholds are determined. Based on the raw material characteristic data and actual process parameter data of each process collected in real time during the production process, the raw material characteristic deviation is calculated by comparing it with the basic process parameters. Based on the basic influence coefficient and the raw material characteristic deviation, the initial control amount of the single-process parameters is calculated, and the process control parameters are obtained by linking and correcting the initial control amount according to the process linkage matrix.

5. The intelligent control method for the production process of digitally-controlled water-dispersible granules according to claim 1, characterized in that: The steps for adjusting and obtaining the production control parameters include: The process control parameters are broken down by process dimension and control type, and matched one by one with the parameters of the same process and type in the basic process parameters to determine the correlation and correspondence. Based on historical production data of water-dispersible granules and the rules governing the influence of raw material characteristics, the sensitivity coefficient of each type of control parameter is calculated, and the corresponding initial adjustment value is calculated in combination with the basic process parameters. According to the importance of preset quality indicators, the priority order of adjustment items is defined, and the maximum adjustment amount and adjustment rate per step are set in combination with the equipment safety operation parameters and process range. The initial adjustment is then corrected to obtain the compliant adjustment value. Preliminary production control parameters are calculated based on the baseline values ​​of the basic process parameters and the compliance adjustment values, and then combined with the associated correspondence to obtain the production control parameters.

6. The intelligent control method for the production process of digitally-controlled water-dispersible granules according to claim 1, characterized in that: The steps to obtain the time-series production data include: The sampling frequency is set according to the production line speed and the response accuracy of the testing equipment. The equipment operation parameters and process operation parameters of the entire process are collected as the process operation data, and the material status and quality indicators are collected as the quality monitoring data. The process operation data and the quality monitoring data are standardized, deduplicated, denoised, and missing data is filled in to obtain multi-source processed data. Using the clock pulse of the main system of the production line as the sole time axis reference, synchronous calibration is performed, a unified timestamp is added to the multi-source treated water, and time series data is obtained by global sorting from morning to night; The time-series data is binned, categorized, and associated according to the production process nodes and production batches to obtain the time-series production data.

7. The intelligent control method for the production process of digitally-controlled water-dispersible granules according to claim 1, characterized in that: The steps for extracting the production deviation characteristics include: Based on the basic process parameters, equipment safe operation thresholds and target quality indicators, dual benchmark values ​​are determined, and the time-series production data is split into multi-dimensional corresponding data according to process dimension, parameter type and data attribute; Based on the multidimensional corresponding data, deviation features are extracted from the time series data of each process and each monitoring parameter in three dimensions: amplitude, time and frequency. Deviation attributes and deviation directions are labeled, and parameter deviation features are obtained by organizing them according to the process. Based on the parameter deviation characteristics, time series analysis is performed on the time series deviation data of the same parameter and the same process to extract the trend, fluctuation and suddenness characteristics of the deviation as production time to form time deviation characteristics. The parameter deviation features are spatially decomposed from the perspectives of process horizontality and parameter category, and process distribution features, parameter clustering features and local defect features are extracted to form process deviation features; Based on the material transfer relationship and process linkage relationship of each production process, the cross-process deviation characteristics in the time-series production data are extracted; The time deviation feature, the process deviation feature, and the cross-process deviation feature are fused and deduplicated to obtain the fused deviation feature. The weight levels and analysis priorities are divided according to preset principles to form the production deviation feature.

8. The intelligent control method for the production process of digitally-controlled water-dispersible granules according to claim 1, characterized in that: The steps to obtain the aforementioned linkage production relationship include: The analysis dimensions are determined from three aspects: causality, quantification, and coupling, and a dual-object correlation analysis table between the basic process parameters and the production deviation characteristics is established. Using the controlled variable method, the causal relationship between changes in a single basic process parameter and a single production deviation characteristic was analyzed, causal associations were screened, and a list of causal associations was obtained by labeling association attributes and association levels. Based on the causal relationship list, related data samples are extracted from the time-series production data, linear fitting is performed to construct a quantitative relationship model, and the change in the intensity of deviation features is calculated as the linkage production relationship.

9. The intelligent control method for the production process of digitally-controlled water-dispersible granules according to claim 8, characterized in that, The steps for obtaining the process correction data include: The production deviation characteristics are broken down by process dimension, deviation type and impact range to obtain the current deviation characteristics, which are then matched with the causal relationship list to screen the inducing process parameters; Based on the correlation level, the inducing process parameters are classified into inducing levels, and the direction of deviation influence is marked to generate a deviation parameter mapping list. Based on the quantitative correlation model and the change in the intensity of the deviation feature, the initial correction amount of each inducing process parameter is calculated in reverse according to the actual intensity value of the production deviation feature. For each initial correction amount, the equipment process constraint boundary is marked. For different coupling types, corresponding correction strategies are adopted, and the correction is performed with the target quality index as the priority to obtain the process correction data.

10. An intelligent control system for a digitalized water-dispersible granule production process, applied to the intelligent control method for a digitalized water-dispersible granule production process as described in any one of claims 1 to 9, characterized in that, The control system includes: The control parameter generation module is used to collect basic data of the active ingredient and filler and the compatibility data of the adjuvant system, and formulate basic process parameters in combination with the target quality index of water-dispersible granules. The linkage compensation control method is used to analyze the influence of the physicochemical properties of raw materials and the linkage relationship of processes on the basic process parameters to obtain process control parameters. The execution data fusion module is used to adjust the basic process parameters according to the process control parameters to obtain production control parameters, collect process operation data and quality monitoring data of the entire production process, and perform fusion processing to obtain time-series production data; The quality judgment and decision module is used to determine whether the time-series production data has reached the target quality index. If so, the current production control parameters are maintained; otherwise, the process control parameters are optimized. The parameter optimization and analysis module is used to extract features from the time-series production data to obtain production deviation features, and to analyze the linkage production relationship between the basic process parameters and the production deviation features using a hierarchical correlation analysis method, thereby optimizing the process control parameters to obtain process correction data.

Images