A data analysis-based nasal allergy-resistant gel production optimization method

Abstract

The application discloses an anti-nasal allergy gel production optimization method based on data analysis, and particularly relates to the field of resource allocation and production optimization, and is used for solving the problem that the response to regional and seasonal demand changes in the existing anti-allergy gel production process is lagging behind, and the raw material supply constraint and the production line switching cost are difficult to balance. The method collects raw material supply chain data, market sales records and environmental allergen time series data, constructs a unified space-time correlation data basis, predicts the anti-allergy response intensity in the future production cycle by using a recurrent neural network, and forms product dosage form preference weights based on association rule mining. On this basis, combined with the raw material inventory and the delivery cycle constraint, a plurality of executable candidate production schemes are generated, and further, the schemes are selected according to the real-time state of the production line and the switching cost, so as to realize the fine configuration and efficient execution of the anti-nasal allergy gel production process.

Description

Technical Field

[0001] This invention relates to the field of resource allocation and production optimization technology, and more specifically, to a data analysis-based method for optimizing the production of anti-nasal allergy gel. Background Technology

[0002] As a commonly used topical anti-allergy product targeting seasonal and regional allergy sufferers, the market demand for anti-nasal allergy gels exhibits significant spatiotemporal fluctuations, often influenced by factors such as pollen dispersal, changes in weather conditions, and regional population allergy exposure levels. In actual production and operation, manufacturers typically face complex situations including significantly different demand across multiple sales regions, concurrent production of multiple product formulations, unstable raw material supply and delivery cycles, and high production line switchover costs. Especially during peak allergy seasons, rapid changes in anti-allergy demand in different regions can lead to frequent adjustments to production plans. Relying solely on manual experience or historical averages for production scheduling often fails to respond promptly to demand fluctuations, resulting in problems such as insufficient supply in certain areas, unbalanced inventory structures, or decreased efficiency due to frequent production line switchovers.

[0003] Meanwhile, anti-nasal allergy gels typically involve multiple raw material components, and their supply is significantly constrained by inventory levels and delivery cycles, making supply chain tensions more pronounced during peak demand periods. Existing production management methods often focus on single-dimensional data statistics or simple rule-based judgments, lacking the systematic analysis and comprehensive decision-making capabilities to understand the correlation between environmental allergens, market response behavior, and production execution conditions. This makes it difficult to balance production costs and production line stability while ensuring timely delivery.

[0004] Therefore, a method is needed that can integrate environmental allergy information, market sales behavior, and production resource constraints to systematically analyze and optimize the production process of anti-nasal allergy gel, so as to improve the scientific nature and execution efficiency of production organization. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a data analysis-based optimization method for the production of anti-nasal allergy gel to address the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A data analysis-based method for optimizing the production of anti-nasal allergy gel includes the following steps:

[0008] S1. Collect raw material supply chain data and sales records of anti-nasal allergy gel, and simultaneously obtain allergen time-series data from meteorological monitoring in the sales area;

[0009] S2. The collected data is structured to construct a unified spatiotemporal relational data table that includes the dimensions of raw material supply, environmental allergens, and market.

[0010] S3. Input the allergen time series data from the spatiotemporal correlation data table into the trained recurrent neural network model to predict the anti-allergy response intensity of each sales region within the specified production cycle in the future.

[0011] S4. Based on the intensity of anti-allergic response and sales records of anti-nasal allergy gel, product dosage form preference weights for each sales region are generated through association rule mining.

[0012] S5. Based on the product dosage form preference weights and real-time raw material supply chain data for the future specified production cycle, verify the supply constraints of the anti-nasal allergy gel raw material components in the production benchmark formulation, and generate candidate execution plans to be put into production.

[0013] S6. Based on the candidate implementation plans to be put into production and the production line status of the anti-nasal allergy gel, select the implementation plan with the lowest switching cost.

[0014] As a further aspect of the present invention, in step S1, collecting raw material supply chain data and sales records of the anti-nasal allergy gel, and simultaneously acquiring allergen time-series data from meteorological monitoring in the sales area, specifically includes:

[0015] The raw material supply chain data refers to the real-time inventory and promised delivery cycle of the raw materials for anti-nasal allergy gel obtained from raw material suppliers. The sales records are extracted from the company's historical orders and cover all the raw materials required for different dosage forms of anti-nasal allergy gel products.

[0016] The allergen time-series data is pollen concentration data recorded in time series by sales area, obtained through regional meteorological monitoring. The pollen concentration data is special monitoring data of various types of highly prevalent allergenic pollen released in the sales area.

[0017] The raw material supply chain data, sales records, and allergen time-series data are integrated into a structured data set by using timestamp alignment. Based on the sales region corresponding to the allergen time-series data, each data point is assigned a corresponding regional code.

[0018] As a further aspect of the present invention, in step S2, the collected data undergoes structured processing to construct a unified spatiotemporal relational data table encompassing raw material supply, environmental allergen, and market dimensions. Specifically, this includes:

[0019] The structured dataset is divided into supply dimension data, allergen dimension data, and market dimension data based on data source and field attributes;

[0020] Using shared timestamps and regional codes as association keys, a database table with unified field definitions is created for the three dimensions of data. The split data of each dimension is mapped and inserted into the corresponding fields of the database table according to their association key values, generating a unified spatiotemporal related data table.

[0021] As a further aspect of the present invention, in step S3, the trained recurrent neural network model is specifically as follows:

[0022] The recurrent neural network model includes an input layer, at least one hidden layer, and an output layer. The hidden layer is composed of long short-term memory units and is deployed and initialized independently in each sales region.

[0023] The training input features are time-series data of environmental allergens within a historical time period, and the training labels and outputs are verified anti-allergy response data, which are the number of clinical allergy cases or sales data of anti-allergy drugs in the same sales area.

[0024] The network is trained under supervision using training input features and training labels. The network weight parameters are adjusted using a backpropagation algorithm over time until the error between the model's predicted output and the training labels is lower than a set threshold. The trained network weight parameters are then saved to form the trained recurrent neural network model.

[0025] As a further aspect of the present invention, in step S3, predicting the anti-allergy response intensity in each sales region within a specified future production cycle specifically includes:

[0026] From the environmental allergen dimension of the unified spatiotemporal correlation data table, pollen concentration in each sales region within a set continuous time unit is extracted to form historical time series data fragments.

[0027] Historical time-series data fragments are converted into fixed-dimensional input vectors as specified by the model and input into a trained recurrent neural network model to perform the forward propagation operation of the model. The output layer of the model generates anti-allergy response data for future consecutive time periods and converts it into an anti-allergy response intensity index.

[0028] As a further aspect of the present invention, in step S4, generating product dosage form preference weights for each sales region specifically includes:

[0029] By integrating historical anti-allergy response intensity indicators from various sales regions with sales records of anti-nasal allergy gels in the market dimension of a unified spatiotemporal correlation data table, an analytical dataset with region and time as the primary keys is formed.

[0030] An association rule mining algorithm was applied to the dataset to scan the co-occurrence patterns between different anti-allergy response intensity index ranges and specific anti-nasal allergy gel product dosage forms, and to identify association rules that meet the minimum support and confidence thresholds.

[0031] Based on the support and confidence of the association rules, the rule scores are calculated and converted into normalized weights, which serve as the preference weight coefficients for each product dosage form. Based on the association rules and their corresponding product dosage form preference weights, a demand mapping table for product dosage forms indexed by sales region and anti-allergy response intensity index is compiled and generated.

[0032] As a further aspect of the present invention, in step S5, generating a candidate execution scheme to be put into production specifically includes:

[0033] Based on the predicted anti-allergy response intensity in each sales region within a specified future production cycle, product dosage form preference weights for the specified future production cycle are extracted from the product dosage form demand mapping table and then vectorized.

[0034] The raw material components of anti-nasal allergy gel in the production benchmark formulation of different product dosage forms are obtained and matrix-converted, labeled as dosage form-raw material component matrix. The weight vector after vectorization is multiplied with the dosage form-raw material component matrix to calculate the multi-dimensional raw material requirements for each sales region.

[0035] A multi-objective optimization model is constructed with real-time raw material inventory and promised delivery cycle as constraints. The optimization objectives include maximizing the raw material demand satisfaction rate and minimizing the supply chain delivery urgency rate. By solving the multi-objective optimization model, a set of non-dominated solutions is obtained. Each solution defines the executable production batch combination under the constraints, the specific product dosage form corresponding to each batch, and the range of raw material allocation ratios, which serve as candidate execution plans for production.

[0036] As a further aspect of the present invention, in step S6, selecting the production execution plan with the lowest switching cost based on the candidate execution plan to be put into production and the production line status of the anti-nasal allergy gel specifically includes:

[0037] Obtain the real-time status of each anti-nasal allergy gel production line, including the product dosage form currently being produced, the list of alternative dosage forms that have completed cleaning validation, and the standard switchover time between each dosage form;

[0038] For each candidate execution plan, the production dosage form allocation is determined based on the real-time status of the production line. The total production line switchover cost required for the plan to be executed is calculated. The total production line switchover cost of all candidate execution plans is compared, and the candidate execution plan with the lowest total switchover cost is selected as the production execution plan.

[0039] The technical effects and advantages of the data analysis-based optimized production method for anti-nasal allergy gel of this invention are as follows:

[0040] This invention, by introducing a multi-source data fusion and decision optimization mechanism, transforms the production organization process of anti-nasal allergy gel from a traditional experience-driven model to an optimization mode combining data-driven and rule-driven approaches. This effectively improves the responsiveness of production planning to regional and seasonal fluctuations in anti-allergy demand. Through systematic mining of the correlation between changes in environmental allergens and market sales behavior, this invention can generate dosage form preference weights that better align with actual demand structures before the production cycle begins, reducing the risk of inventory backlog or stockouts caused by biased demand assessments. Simultaneously, by combining real-time raw material inventory and delivery cycle constraints to generate candidate execution plans, it helps maintain production continuity and feasibility under complex supply conditions, avoiding production interruptions caused by raw material shortages or delivery delays. Furthermore, by quantitatively assessing production line switchover costs and determining the corresponding production execution plan, unnecessary production line switchover frequency can be significantly reduced, switchover time consumption decreased, and equipment utilization and overall production efficiency improved.

[0041] Overall, this invention effectively balances the relationship between demand fulfillment, raw material supply, and production line stability while ensuring timely delivery, providing an feasible and scalable optimization method for the large-scale and refined production of anti-nasal allergy gel. Attached Figure Description

[0042] Figure 1 This is a schematic diagram of an optimized production method for an anti-nasal allergy gel based on data analysis according to the present invention. Detailed Implementation

[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0044] Example 1

[0045] Figure 1 This invention presents a data analysis-based optimization method for the production of anti-nasal allergy gel, comprising the following steps:

[0046] S1. Collect raw material supply chain data and sales records of anti-nasal allergy gel, and simultaneously obtain allergen time-series data from meteorological monitoring in the sales area;

[0047] S2. The collected data is structured to construct a unified spatiotemporal relational data table that includes the dimensions of raw material supply, environmental allergens, and market.

[0048] S3. Input the allergen time series data from the spatiotemporal correlation data table into the trained recurrent neural network model to predict the anti-allergy response intensity of each sales region within the specified production cycle in the future.

[0049] S4. Based on the intensity of anti-allergic response and sales records of anti-nasal allergy gel, product dosage form preference weights for each sales region are generated through association rule mining.

[0050] S5. Based on the product dosage form preference weights and real-time raw material supply chain data for the future specified production cycle, verify the supply constraints of the anti-nasal allergy gel raw material components in the production benchmark formulation, and generate candidate execution plans to be put into production.

[0051] S6. Based on the candidate implementation plans to be put into production and the production line status of the anti-nasal allergy gel, select the implementation plan with the lowest switching cost.

[0052] In step S1, raw material supply chain data and sales records of the anti-nasal allergy gel are collected, and allergen time-series data from meteorological monitoring in the sales area are acquired simultaneously.

[0053] Raw material supply chain data for the anti-nasal allergy gel is obtained from its suppliers. This data uses raw material batches as the basic recording unit and clearly includes real-time inventory information and corresponding promised delivery cycles for various anti-nasal allergy gel raw materials. Real-time inventory refers to the actual available quantity for delivery from the supplier at the time of data collection. The promised delivery cycle refers to the time span from the date the company confirms its purchase request to the actual arrival of the raw materials at the production company's warehouse. This cycle is confirmed and provided by the supplier based on its production cycle, logistics arrangements, and historical performance records. Simultaneously, sales records for the anti-nasal allergy gel are extracted from the company's internal historical order database. These sales records use order time, sales region, product specifications, and dosage form as core fields, covering all anti-nasal allergy gel product specifications actually sold by the company. To ensure a direct correlation between sales records and subsequent production optimization, the production formulas corresponding to each product dosage form are reverse-mapped during sales record extraction. This clarifies the actual range of raw materials involved in these sales records, such as antihistamine functional materials for relieving nasal irritation and moisturizing matrix materials for improving nasal mucosa conditions.

[0054] Environmental allergen time-series data were acquired through regional meteorological monitoring channels. This data, using the sales area as the statistical unit, continuously records changes in allergen exposure levels in the regional environment at a uniform time granularity. Priority was given to acquiring specialized monitoring data released by local meteorological or public environmental monitoring agencies. This monitoring data uses pollen concentration as the primary indicator, further subdividing and recording various types of allergenic pollen prevalent in the area, such as spring tree pollen and summer / autumn herbaceous pollen, and forming time-series concentration records using a standardized measurement method. In addition to pollen allergens, other common indoor and outdoor allergens were further included in the allergen time-series data collection scope. For example, dust mite concentration indices and biological particulate indicators related to allergic reactions in airborne particulates were obtained through environmental monitoring or public health data channels, and recorded and organized using the same time stamp as pollen concentration.

[0055] A unified timestamp alignment process is performed on the collected raw material supply chain data, sales records, and environmental allergen time-series data. Specifically, firstly, based on the company's unified time standard, the time format differences between different data sources are converted, mapping the timestamps of various data types to discrete time points or continuous time intervals on the same time axis. Then, using the sales region division used in the environmental allergen time-series data as the regional division benchmark, regional association processing is performed on the raw material supply chain data and sales records to ensure that each raw material record and sales record corresponds to a specific sales region. For raw material data involving cross-regional supply or centralized warehousing, regional attribution is confirmed by recording the final production or sales region of the raw material. After completing both time and regional alignment, the above multi-source data is integrated into a structured dataset. Each record in this structured dataset simultaneously includes a timestamp, regional code, raw material supply status, sales behavior information, and the corresponding allergen exposure level at that time point. The regional code is generated using a unified coding rule and is permanently stored as the basic field for subsequent data association and indexing.

[0056] In step S2, the collected data is processed in a structured manner to construct a unified spatiotemporal relational data table that includes the dimensions of raw material supply, environmental allergens, and market.

[0057] For the existing structured dataset, the data was explicitly split into dimensions based on its source and its functional attributes in the production optimization process. First, each field in the structured dataset was analyzed, and fields reflecting raw material inventory status and promised delivery cycles were assigned to the supply dimension. This dimension characterizes the availability and supply stability of raw materials under different time and regional conditions. Next, fields from environmental monitoring channels, such as pollen concentration and dust mite index, describing changes in allergen exposure levels within the sales area, were assigned to the allergen dimension. This dimension fully preserves the dynamic characteristics of environmental factors in a time-series format. Simultaneously, fields reflecting actual sales behavior of anti-nasal allergy gels, such as order quantity, product dosage form, and sales region, were assigned to the market dimension to depict the actual market demand for different product dosage forms. During the splitting process, each field was only allowed to be assigned to one dimension to avoid duplicate fields across multiple dimensions.

[0058] After splitting the data into supply, allergen, and market dimensions, the common timestamps and region codes across all dimensions are used as unique association keys to perform unified association processing on the three types of data. In practice, firstly, a consistent data table framework with uniform field structures is defined for each of the three dimensions, ensuring complete consistency in field name, data type, and precision for the timestamp and region code fields. Then, based on the matching relationship between timestamps and region codes, raw material inventory and delivery cycle information from the supply dimension data, environmental allergen time-series records from the allergen dimension data, and sales behavior records from the market dimension data are mapped and written to their respective data table fields. During the mapping process, a unified data filling rule is used to handle cases where data for a particular dimension is missing under a specific timestamp or region.

[0059] In S3, the intensity of anti-allergy response in each sales region is predicted within a specified future production cycle.

[0060] To address the regional differences involved in optimizing the production of anti-nasal allergy gel, a recurrent neural network (RNN) model was constructed for modeling the intensity of the anti-allergy response. The model consists of an input layer, at least one hidden layer, and an output layer. The hidden layer explicitly employs Long Short-Term Memory (LSTM) units as the core computational structure to simultaneously characterize the long-term trends and short-term fluctuations in environmental allergen time-series data. In implementation, an independent RNN model instance was deployed for each sales region, maintaining structural consistency while maintaining parameter independence to avoid interference from regional differences in allergen environments during training. During initialization, the weight parameters of the models for each sales region were initialized using a unified rule, ensuring all models started from the same point. This independent deployment and initialization by sales region allows the model structure to better reflect the rhythms of environmental allergen changes and population response characteristics in different regions.

[0061] The training process of the recurrent neural network model involves the explicit construction and mapping of training input features and training labels. The training input features are selected from time-series data on environmental allergens within historical time periods. This data is organized sequentially by sales region, comprehensively reflecting the dynamic changes in allergen exposure levels such as pollen concentration and dust mite index within the region. To ensure the stability of the training data, a fixed time span rule is adopted for the historical time periods, such as several consecutive weeks or months as a training sample interval, ensuring that the input features cover typical high-incidence and low-incidence periods of allergies. The training labels and model outputs use the same data format, both being validated anti-allergy response data derived from actual statistical results within the same sales region. Either the number of clinical allergy cases or sales data of anti-allergy drugs can be used as the label caliber, and this caliber must be maintained consistently during model training; mixing is not allowed. By matching each time-series input of environmental allergens with the corresponding anti-allergy response data within that time period, a clear correspondence between input and output is established, enabling the recurrent neural network to learn the inherent mapping pattern between changes in environmental allergens and the intensity of regional anti-allergy responses during training.

[0062] Based on the constructed training input features and training labels, a supervised training process is performed on recurrent neural network models independently deployed for each sales region. Specifically, historical time-series data on environmental allergens are sequentially input into the model's input layer. Each time step corresponds to an input vector containing exposure levels to multiple allergens, such as daily pollen concentration and dust mite exposure index. This data is then processed sequentially through a hidden layer composed of long short-term memory units. In the example implementation, the hidden layer can be configured with one or more layers, where the number of hidden units in each layer can be a fixed value of 32, 64, or 128 to balance model expressive power and training stability. The model output layer generates anti-allergy response prediction results consistent with the training label format. These prediction results output anti-allergy response values ​​for the corresponding time period, organized by sales region. During supervised training, model parameters are updated using backpropagation over time. Training batches use time-series data spanning several days or weeks as a single training sample unit. The number of training epochs can be set to any preset value among 50, 100, or 200 epochs. When the error variation between the predicted output and the training label stabilizes and falls below a pre-set error threshold (e.g., relative error below 5%) across multiple training epochs, the model is considered to have reached training convergence. After training, the hidden layer weight matrix, gating parameters, and output layer weight parameters of the corresponding sales region model are saved as a whole, along with the sales region code and the training data time range, forming a set of parameters for the trained recurrent neural network model. This ensures that the model can be directly called for predicting anti-allergy response intensity under the same regional conditions, achieving reproducibility and applicability of the model training results.

[0063] Starting from the environmental allergen dimension in the unified spatiotemporal correlation data table, data is filtered according to sales region to extract the time-series records of environmental allergens corresponding to the target sales region. For each sales region, pollen concentration data is extracted continuously from the environmental allergen dimension according to a pre-set continuous time unit length, forming a historical time-series data segment for model prediction. The continuous time unit length is set using a fixed rule, such as using 7, 14, or 30 consecutive days as a time-series segment window, so that the historical data segment can cover a complete cycle of allergy exposure changes. During the extraction process, strict consistency of time order is maintained to ensure that the pollen concentration values ​​at each time point in the time-series data segment are arranged in the actual order of occurrence. Subsequently, the extracted historical time-series data segments undergo dimension unification processing, converting them into fixed-dimensional input vectors that meet the input requirements of the recurrent neural network model. The dimension-transformed historical time-series input vectors are input into the trained recurrent neural network model for the corresponding sales region to perform the model's forward propagation operation. The model sequentially receives data from each time step in the historical time-series input vector. It updates the state time-series through a hidden layer composed of Long Short-Term Memory (LSTM) units. The output layer generates anti-allergy response data for consecutive future time periods corresponding to a set prediction time range. This anti-allergy response data maintains the same numerical form as the anti-allergy response labels used during training, such as the number of allergy cases or the demand for anti-allergy medications within the predicted time period. The model's output anti-allergy response data is further processed by intensity index transformation. Specifically, it is normalized based on the statistical distribution interval of the historical anti-allergy response data, and the result is used as the production standardized intensity index value.

[0064] In step S4, product dosage form preference weights for each sales region are generated.

[0065] By integrating historical anti-allergy response intensity indicators from various sales regions with sales records of anti-nasal allergy gels corresponding to the market dimension in a unified spatiotemporal association data table, an analytical dataset is formed with sales region and time as the joint primary key. Each record in this dataset explicitly includes an identifier for the anti-allergy response intensity index range within a specific time point or time period, as well as the dosage form information of the anti-nasal allergy gel products actually sold within the corresponding time period. This approach establishes a direct correspondence between environmental response status and market sales behavior at the same record level. Based on this, association rule mining methods are applied to the analytical dataset to scan for co-occurrence between different anti-allergy response intensity index ranges and specific product dosage forms. To ensure the statistical significance and business representativeness of the mining results, minimum support thresholds and minimum confidence thresholds are simultaneously set during the association rule identification process. The minimum support threshold is used to limit the minimum frequency ratio of a certain intensity range and product dosage form combination in the analysis dataset. For example, it can be set to no less than 5% or 10% of the total number of records in the analysis dataset to avoid occasional sales behavior from interfering with the rule formation. The minimum confidence threshold is used to limit the stability of the actual occurrence of the corresponding product dosage form under the condition that a certain anti-allergy response intensity index range appears. For example, it can be set to no less than 60% or 70% to ensure that the identified association has a high degree of consistency in a statistical sense.

[0066] After obtaining the set of association rules that meet the threshold conditions, a rule score is calculated for each association rule based on its corresponding support and confidence information. Support and confidence are used as two core factors reflecting the stability and representativeness of the rule, and a comprehensive evaluation value is assigned to each rule. For example, a single rule score is formed by jointly calculating support and confidence, reflecting the overall importance of the rule in historical data. Subsequently, the rule scores are summarized according to the product dosage form dimension. The rule scores corresponding to multiple association rules pointing to the same product dosage form are accumulated to form the original preference value of that product dosage form under specific sales regions and anti-allergy response intensity conditions. Based on this, the original preference values ​​of each product dosage form are normalized to transform them into preference weight coefficients under a unified scale. For example, each dosage form preference value is divided by the sum of all dosage form preference values ​​within the same region and intensity range, so that the final sum of the preference weight coefficients is 1, facilitating relative comparisons between different dosage forms. After completing the preference weight calculation, a product dosage form demand mapping table is compiled and generated based on the sales region, anti-allergy response intensity index range, and corresponding product dosage form preference weights involved in the association rules. This mapping table uses sales region and anti-allergy response intensity index as index conditions, and clearly lists the preference weight distribution of each dosage form of anti-nasal allergy gel product under different regions and different response intensity states, so that the mapping table can reflect the demand tendency under historical statistical significance in a structured form.

[0067] In step S5, candidate execution plans to be put into production are generated.

[0068] Based on the predicted anti-allergy response intensity in each sales region within a specified future production cycle, product dosage form preference weights for each sales region are extracted from the product dosage form demand mapping table according to their respective anti-allergy response intensity ranges. The anti-allergy response intensity level of each sales region within the future production cycle is determined based on the prediction results. For example, if a sales region is identified as being in a medium- or high-intensity anti-allergy response state during the prediction period, this intensity level is used as an index to retrieve all product dosage form preference weight records for that region from the demand mapping table at that intensity level. Subsequently, the extracted product dosage form preference weights are arranged in a uniform dosage form order, and the preference weight corresponding to each anti-nasal allergy gel product dosage form is sequentially written into a fixed position, forming a consistent weight vector. Each element in this vector uniquely corresponds to a specific product dosage form, and the vector length is equal to the total number of product dosage forms currently included in the company's production management scope. During vectorization, for cases where a dosage form preference weight is missing, it is filled according to predefined default rules in the mapping table. For example, the preference weight for a missing dosage form is set to zero or a very low fixed value to ensure the weight vector structure is complete and the dimensions are uniform. The product dosage form preference weights, which were originally in tabular form, are converted into a vectorized expression that can be directly used in subsequent calculations.

[0069] For the different dosage forms of anti-nasal allergy gel products already identified by the company, the production baseline formulation information corresponding to each dosage form was obtained, and the raw material components involved were systematically organized. Specifically, the baseline formulations for all dosage forms were expanded using the product dosage form as the row dimension and the types of raw materials involved in the formulation as the column dimension, clarifying the usage ratio or standard dosage range of each raw material in each product dosage form. Based on this, the organized formulation data was converted into a dosage form-raw material component matrix, where each row corresponds to a specific product dosage form, and each column corresponds to a specific raw material component. Matrix cells record the standard feeding ratio or baseline consumption of the corresponding raw material in that dosage form. To ensure the uniformity of the matrix structure, raw material components not included in a certain dosage form were explicitly filled with zero values ​​in their corresponding matrix positions. After completing the matrix construction, the previously vectorized product dosage form preference weight vector was calculated and processed with the dosage form-raw material component matrix. Through weighted formulation calculations, multi-dimensional raw material demand results for each sales region were obtained. The raw material demand results, organized by raw material type, reflect the overall demand scale of various raw materials in different sales regions within the predicted production cycle, taking into account product dosage form preferences. This combination of matrix and vectorization processing directly couples product preference information with formulation structure information, yielding a raw material demand distribution that aligns with the predicted demand structure.

[0070] Based on the calculated multi-dimensional raw material demand results, combined with real-time raw material inventory and promised delivery cycle information, a multi-objective optimization model is constructed to generate candidate execution schemes. In specific implementation, real-time raw material inventory is used as one constraint, clarifying the maximum available quantity of each type of raw material for production at the current point in time; simultaneously, the promised delivery cycle is used as another constraint, clarifying the time limit within which each type of raw material can be replenished in the future production cycle. Regarding the optimization objectives, the raw material demand satisfaction rate is determined by comparing the actual allocable quantity of each type of raw material under the candidate scheme with the corresponding raw material demand, for example, by reflecting the degree to which raw material demand is met through a satisfaction ratio; the supply chain delivery urgency rate is determined based on the relative relationship between the production time supported by the remaining raw material inventory and its promised delivery cycle. The closer the inventory's support time is to the upper limit of the delivery cycle, the higher the corresponding delivery urgency rate. Based on the above constraints and optimization objectives, solutions are obtained for different production batch combinations, product dosage form allocation methods, and raw material allocation ratio ranges. Through a multi-objective optimization process, a set of solutions that are not simultaneously superior to other schemes in any objective is selected, forming a non-dominated solution set. Each solution in this set of non-dominated solutions clearly provides a production batch combination scheme that can be actually executed under the current inventory and delivery conditions, including the specific product dosage form arrangement for each batch and the allocation ratio range of various raw materials, thus forming a set of candidate execution schemes to be put into production that can be directly used for subsequent production decisions.

[0071] In step S6, the production execution plan with the lowest switching cost is selected based on the candidate execution plans to be put into production and the production line status of the anti-nasal allergy gel.

[0072] For multiple production lines used for the production of anti-nasal allergy gel, their current availability status was uniformly acquired and organized. Using the production line as the basic unit, the operational information of each production line at the current point in time was read, clearly recording the product dosage forms being produced on each line. This information reflects the current process status of the production line. Simultaneously, a list of alternative dosage forms that have completed cleaning validation was acquired. This list indicates the range of product dosage forms that the production line can directly switch to without re-performing deep cleaning or additional validation procedures. During this process, the cleaning validation status of each production line was consistently confirmed to ensure that the alternative dosage form list only includes dosage forms that have passed the established cleaning validation standards, preventing unfeasible switchover plans from being included in subsequent calculations. Furthermore, the standard switchover time between different product dosage forms was compiled. This switchover time, based on the production line's historical operation records and process specifications, clearly defines the time required to switch from the current production dosage form to other alternative dosage forms and records it as a fixed value.

[0073] Based on the acquired real-time production line status information, the production line switchover cost is calculated for each candidate execution plan to be put into production. In practice, for each candidate execution plan, the defined production batch combinations and corresponding product dosage form allocations are analyzed, and the dosage form allocation plan is mapped to the currently available production line set. Subsequently, combining the product dosage forms currently being produced on each production line and the list of alternative dosage forms that have completed cleaning validation, the switchable paths for each production batch in the candidate execution plan on the corresponding production line are determined. Based on the standard switchover man-hours between dosage forms, the man-hours for each necessary dosage form switchover are accumulated to obtain the total production line switchover man-hours required for the overall execution of the candidate execution plan. This total switchover man-hour is used as a quantitative indicator of production line switchover cost to reflect the degree of impact of the plan on production line stability and execution efficiency. After completing the switchover cost calculation for all candidate execution plans, the total production line switchover costs for each plan are compared uniformly and sorted in ascending order of switchover cost to clearly identify the candidate execution plan with the lowest switchover cost. Ultimately, the candidate execution plan with the lowest total switching cost and meeting the production line status constraints was selected as the production execution plan. This plan will guide the actual production arrangements of the anti-nasal allergy gel within the specified production cycle in the future, thereby effectively reducing the time consumption and management complexity caused by production line switching while ensuring the feasibility of execution.

[0074] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0075] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0076] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0077] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0078] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0079] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0080] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0081] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0082] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A data-driven optimization method for the production of anti-nasal allergy gel, characterized in that, Includes the following steps: S1. Collect raw material supply chain data and sales records of anti-nasal allergy gel, and simultaneously obtain allergen time-series data from meteorological monitoring in the sales area; S2. The collected data is structured to construct a unified spatiotemporal relational data table that includes the dimensions of raw material supply, environmental allergens, and market. S3. Input the allergen time series data from the spatiotemporal correlation data table into the trained recurrent neural network model to predict the anti-allergy response intensity of each sales region within the specified production cycle in the future. S4. Based on the intensity of anti-allergic response and sales records of anti-nasal allergy gel, product dosage form preference weights for each sales region are generated through association rule mining. S5. Based on the product dosage form preference weights and real-time raw material supply chain data for the future specified production cycle, verify the supply constraints of the anti-nasal allergy gel raw material components in the production benchmark formulation, and generate candidate execution plans to be put into production. S6. Based on the candidate implementation plans to be put into production and the production line status of the anti-nasal allergy gel, select the production implementation plan with the lowest switching cost. In step S5, generating candidate execution schemes for production deployment specifically includes: Based on the predicted anti-allergy response intensity in each sales region within a specified future production cycle, product dosage form preference weights for the specified future production cycle are extracted from the product dosage form demand mapping table and then vectorized. The raw material components of anti-nasal allergy gel in the production benchmark formulation of different product dosage forms are obtained and matrix-converted, labeled as dosage form-raw material component matrix. The weight vector after vectorization is multiplied with the dosage form-raw material component matrix to calculate the multi-dimensional raw material requirements for each sales region. A multi-objective optimization model is constructed with real-time raw material inventory and promised delivery cycle as constraints. The optimization objectives include maximizing the raw material demand satisfaction rate and minimizing the supply chain delivery urgency rate. The supply chain delivery urgency rate is determined based on the relative relationship between the production time supported by the remaining raw material inventory and its promised delivery cycle. By solving the multi-objective optimization model, a set of non-dominated solutions is obtained. Each solution defines the executable production batch combination under the constraints, the specific product dosage form corresponding to each batch, and the range of raw material allocation ratios, serving as a candidate execution plan for production.

2. The data analysis-based optimized production method for anti-nasal allergy gel according to claim 1, characterized in that, In step S1, the collection of raw material supply chain data and sales records for the anti-nasal allergy gel, and the simultaneous acquisition of allergen time-series data from meteorological monitoring in the sales area, specifically include: The raw material supply chain data refers to the real-time inventory and promised delivery cycle of the raw materials for anti-nasal allergy gel obtained from raw material suppliers. The sales records are extracted from the company's historical orders and cover all the raw materials required for different dosage forms of anti-nasal allergy gel products. The allergen time-series data is pollen concentration data recorded in time series by sales area, obtained through regional meteorological monitoring. The pollen concentration data is special monitoring data of various types of highly prevalent allergenic pollen released in the sales area. The raw material supply chain data, sales records, and allergen time-series data are integrated into a structured data set by using timestamp alignment. Based on the sales region corresponding to the allergen time-series data, each data point is assigned a corresponding regional code.

3. The data analysis-based optimized production method for anti-nasal allergy gel according to claim 1, characterized in that, In step S2, the collected data undergoes structured processing to construct a unified spatiotemporal relational data table encompassing raw material supply, environmental allergens, and market dimensions. Specifically, this includes: The structured dataset is divided into supply dimension data, allergen dimension data, and market dimension data based on data source and field attributes; Using shared timestamps and regional codes as association keys, a database table with unified field definitions is created for the three dimensions of data. The split data of each dimension is mapped and inserted into the corresponding fields of the database table according to their association key values, generating a unified spatiotemporal related data table.

4. The data analysis-based optimized production method for anti-nasal allergy gel according to claim 1, characterized in that, In S3, the trained recurrent neural network model is specifically as follows: The recurrent neural network model includes an input layer, at least one hidden layer, and an output layer. The hidden layer is composed of long short-term memory units and is deployed and initialized independently in each sales region. The training input features are time-series data of environmental allergens within a historical time period, and the training labels and outputs are verified anti-allergy response data, which are the number of clinical allergy cases or sales data of anti-allergy drugs in the same sales area. The network is trained under supervision using training input features and training labels. The network weight parameters are adjusted using a backpropagation algorithm over time until the error between the model's predicted output and the training labels is lower than a set threshold. The trained network weight parameters are then saved to form the trained recurrent neural network model.

5. The data analysis-based optimized production method for anti-nasal allergy gel according to claim 1, characterized in that, In step S3, predicting the intensity of anti-allergic response in each sales region within a specified future production cycle specifically includes: From the environmental allergen dimension of the unified spatiotemporal correlation data table, pollen concentration in each sales region within a set continuous time unit is extracted to form historical time series data fragments. Historical time-series data fragments are converted into fixed-dimensional input vectors as specified by the model and input into a trained recurrent neural network model to perform the forward propagation operation of the model. The output layer of the model generates anti-allergy response data for future consecutive time periods and converts it into an anti-allergy response intensity index.

6. The data analysis-based optimized production method for anti-nasal allergy gel according to claim 1, characterized in that, In step S4, generating the product dosage form preference weights for each sales region specifically includes: By integrating the historical anti-allergy response intensity of each sales region with the sales records of anti-nasal allergy gel in the market dimension of the unified spatiotemporal correlation data table, an analytical dataset with region and time as the main keys is formed. An association rule mining algorithm was applied to the dataset to scan the co-occurrence patterns between different anti-allergy response intensity ranges and specific anti-nasal allergy gel product dosage forms, and to identify association rules that meet the minimum support and confidence thresholds. Based on the support and confidence of the association rules, the rule scores are calculated and converted into normalized weights, which serve as the preference weight coefficients for each product dosage form. Based on the association rules and their corresponding product dosage form preference weights, a product dosage form demand mapping table indexed by sales region and anti-allergy response intensity is compiled and generated.

7. The data analysis-based optimized production method for anti-nasal allergy gel according to claim 1, characterized in that, In step S6, selecting the production execution plan with the lowest switching cost based on the candidate execution plans to be put into production and the production line status of the anti-nasal allergy gel specifically includes: Obtain the real-time status of each anti-nasal allergy gel production line, including the product dosage form currently being produced, the list of alternative dosage forms that have completed cleaning validation, and the standard switchover time between each dosage form; For each candidate execution plan, the production dosage form allocation is determined based on the real-time status of the production line. The total production line switchover cost required for the plan to be executed is calculated. The total production line switchover cost of all candidate execution plans is compared, and the candidate execution plan with the lowest total switchover cost is selected as the production execution plan.

Images