Traditional Chinese medicine dispensing granule quality tracing system based on data fusion

Abstract

The application provides a traditional Chinese medicine dispensing granule quality tracing system based on data fusion, and relates to the technical field of data processing, and comprises the following steps: defining a dynamic adjustment factor according to actual conditions; multiplying the actual value of each key feature with the corresponding weight and the dynamic adjustment factor to obtain a weighted key feature value; summing all the weighted key feature values to obtain a comprehensive quality index; comparing the comprehensive quality index with the corresponding quality standard to obtain a comparison result; judging the quality of new traditional Chinese medicine dispensing granule data according to the comparison result; if the comprehensive quality index is lower than the quality standard, it is determined that the corresponding batch of traditional Chinese medicine dispensing granules has a quality problem; and analyzing the weighted key feature value to obtain a key feature that negatively affects the comprehensive quality index. The application improves the efficiency and accuracy of quality problem processing.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a quality tracing system for Chinese medicine formula granules based on data fusion. Background Art

[0002] The traditional quality traceability system for Chinese medicine granules has the following defects:

[0003] For example, some can only collect data from partial links and cannot fully cover the entire process of Chinese medicine formula granules from production to sales, which leads to incomplete data and reduced traceability accuracy.

[0004] Secondly, in traditional quality traceability, the extraction of key quality characteristics mostly relies on manual experience and simple statistical analysis. Therefore, it is difficult to accurately capture the key factors affecting the quality of Chinese medicine formula granules.

[0005] For example, some traditional methods for calculating quality indexes use fixed weights and static evaluation criteria, ignoring the dynamic nature of actual conditions. This approach may not adapt to the ever-changing production environment and market demands, resulting in inaccurate and in-time quality assessment results. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a quality traceability system for Chinese medicine formula granules based on data fusion, thereby improving the efficiency and accuracy of quality problem handling.

[0007] In order to solve the above technical problems, the technical solutions of the present invention are as follows:

[0008] First, the quality traceability method of Chinese medicine formula granules based on data fusion includes:

[0009] Data collection module, used to collect parameter data of Chinese medicine formula granules in various links of production, processing, transportation, storage and sales;

[0010] A data preprocessing module, used for preprocessing parameter data to obtain preprocessed data;

[0011] Extraction module, used to extract key features related to the quality of traditional Chinese medicine granules from preprocessed data;

[0012] The quality assessment calculation module is used to assign a weight to each key feature based on historical data; define a dynamic adjustment factor based on actual conditions; multiply the actual value of each key feature by its corresponding weight and the dynamic adjustment factor to obtain a weighted key feature value; sum all weighted key feature values ​​to obtain a comprehensive quality index; and compare the comprehensive quality index with the corresponding quality standard to obtain a comparison result.

[0013] The quality traceability module is used to determine the quality of new TCM formula granule data based on the comparison results. If the comprehensive quality index is lower than the quality standard, it is determined that the corresponding batch of TCM formula granules has quality problems. The module analyzes the weighted key feature values ​​to obtain the key features that negatively affect the comprehensive quality index. Based on the production batch information and the weights and dynamic adjustment factors of the key features, the module locates the key nodes where the problems exist.

[0014] The prediction module is used to connect key nodes with production information of each link through the established quality traceability model to form a quality traceability chain; based on the quality traceability chain, the traceability results are obtained.

[0015] Furthermore, key features related to the quality of TCM granules were extracted from the preprocessed data, including:

[0016] Receive pre-processed data related to Chinese medicine formula granules and refer to a pre-established feature library, which contains all features related to the quality of Chinese medicine formula granules, including chemical component content, processing temperature, humidity and timestamp;

[0017] According to the pre-established feature library, the corresponding feature fields are filtered out from the pre-processed data;

[0018] The correlation coefficient between the feature fields and the quality indicators is calculated to obtain the key features.

[0019] Furthermore, the correlation coefficient between the feature fields and the quality indicators is calculated to obtain key features, including:

[0020] pass Calculate the correlation coefficient between the feature field and the quality index;

[0021] Key features are obtained by screening based on correlation coefficients;

[0022] in, represents the correlation coefficient, Represents the total number of observations, which indicates the number of samples in the data set; and Indicates the first of two variables observations; From 1 to , traverse all samples; Represents the summation symbol, indicating the sum from 1 to The summation of all sample points; Representing variables The mean of Representing variables The mean of Representing variables No. The difference between a value and its mean; Representing variables No. The difference between a value and its mean; Indicates the index of the sample point currently being processed, used to traverse from 1 to All sample points; and Is the subscript in the summation, used to calculate the variable and The mean of , in the mean calculation, traverse from 1 to All and value.

[0023] Furthermore, a weight is assigned to each key feature based on historical data, including:

[0024] According to historical data, through Assign a weight to each key characteristic;

[0025] in, It is The correlation coefficient between the key features and the quality indicators; and is the adjustment factor; is the total number of key features; It is The key feature is The probability of a value; It is The number of values ​​of the key features; It is The correlation coefficient between the key features and the quality indicators; It is The number of values ​​of the key features; It is The key feature is The probability of a value; is the index of the key feature; Is calculating the The index used when calculating the information entropy of a key feature; is the index used when calculating the information entropy of each key feature in the denominator; is the index used in the denominator to iterate over all key features.

[0026] Furthermore, the calculation formula of the dynamic adjustment factor is:

[0027] ;

[0028] in, and is the weight coefficient; and are the weights of temperature and humidity respectively; and are the ideal temperature and humidity values ​​respectively; and are the actual measured temperature and humidity values ​​respectively; and are the maximum possible ranges of temperature and humidity, respectively; is the number of critical process parameters; It is The actual value of the process parameters; It is target values ​​of the process parameters; It is the maximum possible range of variation of process parameters.

[0029] Furthermore, the weighted key feature values ​​are analyzed to obtain the key features that negatively impact the comprehensive quality index, including:

[0030] Performing data standardization on the weighted eigenvalues ​​to obtain standardized data;

[0031] Calculate the statistics of each weighted eigenvalue in the standardized data, including the mean, standard deviation, median, and quartiles, to understand the distribution of the standardized data and the range of variation of the eigenvalues;

[0032] Based on the statistics, a threshold is set to determine whether the weighted eigenvalue is abnormal;

[0033] Traverse all weighted eigenvalues, compare each weighted eigenvalue with the threshold, and mark abnormal weighted eigenvalues;

[0034] For each weighted eigenvalue marked as abnormal, the corresponding weighted eigenvalue is removed from the normalized data, and the comprehensive quality index is recalculated based on the remaining weighted eigenvalues;

[0035] Compare the difference between the original comprehensive quality index and the comprehensive quality index after removing abnormal features;

[0036] All abnormal weighted eigenvalues ​​are sorted according to their respective impact on the comprehensive quality index, from large to small; a negative impact threshold is set. When the impact of the abnormal weighted eigenvalue on the comprehensive quality index exceeds the negative impact threshold, the abnormal weighted eigenvalue is regarded as a key feature that has a negative impact on the overall quality; the key features that exceed the negative impact threshold are screened out from the sorted abnormal weighted eigenvalue list.

[0037] Furthermore, based on the production batch information and the weights and dynamic adjustment factors of key features, the key nodes with problems are located, including:

[0038] For each production batch, the actual measured value of each key characteristic is multiplied by its corresponding weight to obtain a weighted score;

[0039] Multiply the weighted value by the corresponding dynamic adjustment factor to reflect the actual situation in the current production environment;

[0040] Compare the weighted scores of different production batches to identify batches with abnormal scores;

[0041] For batches with abnormal scores, analyze each link in the production process, including raw material quality and equipment status, to obtain production process analysis results;

[0042] Based on production process analysis and weighted scores, the key nodes that lead to product quality decline are located.

[0043] The second aspect is the quality traceability method of Chinese medicine formula granules based on data fusion, including:

[0044] Collect parameter data of Chinese medicine formula granules in each link of production, processing, transportation, storage and sales;

[0045] preprocessing the parameter data to obtain preprocessed data;

[0046] Extract key features related to the quality of TCM granules from preprocessed data;

[0047] Assign a weight to each key characteristic based on historical data; define a dynamic adjustment factor based on actual conditions; multiply the actual value of each key characteristic by its corresponding weight and the dynamic adjustment factor to obtain a weighted key characteristic value; sum all weighted key characteristic values ​​to obtain a comprehensive quality index; compare the comprehensive quality index with the corresponding quality standard to obtain a comparison result;

[0048] Based on the comparison results, the new TCM formula granule data is quality-judged. If the comprehensive quality index is lower than the quality standard, the corresponding batch of TCM formula granules is determined to have quality problems. The weighted key feature values ​​are analyzed to obtain the key features that negatively affect the comprehensive quality index. Based on the production batch information, the weights of the key features, and the dynamic adjustment factors, the key nodes with problems are located.

[0049] By establishing a quality traceability model, key nodes are connected with production information of each link to form a quality traceability chain; based on the quality traceability chain, traceability results are obtained.

[0050] According to a third aspect, a computing device includes:

[0051] one or more processors;

[0052] The storage device is used to store one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the method.

[0053] In a fourth aspect, a computer-readable storage medium stores a program, which implements the method when executed by a processor.

[0054] The above solution of the present invention includes at least the following beneficial effects.

[0055] Through the data acquisition module, the system can comprehensively and automatically collect parameter data for each stage of the production, processing, transportation, storage, and sales of Chinese herbal granules. This comprehensive data collection ensures data integrity and accuracy. The data preprocessing module can effectively clean, convert, and standardize the collected parameter data to obtain high-quality preprocessed data, which not only improves data quality but also reduces the complexity and error rate of subsequent processing.

[0056] The extraction module utilizes data analysis technology to accurately extract key features related to the quality of TCM granules from preprocessed data. The quality assessment calculation module combines historical data with actual conditions, dynamically adjusting factors and assigning weights to achieve dynamic quality assessment of TCM granules. The quality traceability module rapidly identifies quality issues with TCM granules based on a comparison of the comprehensive quality index and quality standards. By analyzing weighted key feature values, it accurately identifies critical nodes with issues, improving the efficiency and accuracy of quality issue resolution.

[0057] The prediction module can predict potential quality problems through the established quality traceability model, and connect key nodes with production information of each link to form a complete quality traceability chain. This prediction and prevention capability helps enterprises take measures in advance to avoid the occurrence of quality problems or reduce their impact. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a schematic diagram of a quality traceability system for Chinese herbal formula granules based on data fusion provided by an embodiment of the present invention.

[0059] Figure 2 The present invention provides a flow chart of a method for tracing the quality of Chinese medicine formula granules based on data fusion. DETAILED DESCRIPTION

[0060] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0061] like Figure 1 As shown, an embodiment of the present invention proposes a Chinese medicine formula granule quality traceability system based on data fusion, including:

[0062] Data collection module, used to collect parameter data of Chinese medicine formula granules in various links of production, processing, transportation, storage and sales;

[0063] A data preprocessing module, used for preprocessing parameter data to obtain preprocessed data;

[0064] Extraction module, used to extract key features related to the quality of traditional Chinese medicine granules from preprocessed data;

[0065] The quality assessment calculation module is used to assign a weight to each key feature based on historical data; define a dynamic adjustment factor based on actual conditions; multiply the actual value of each key feature by its corresponding weight and the dynamic adjustment factor to obtain a weighted key feature value; sum all weighted key feature values ​​to obtain a comprehensive quality index; and compare the comprehensive quality index with the corresponding quality standard to obtain a comparison result.

[0066] The quality traceability module is used to determine the quality of new TCM formula granule data based on the comparison results. If the comprehensive quality index is lower than the quality standard, it is determined that the corresponding batch of TCM formula granules has quality problems. The module analyzes the weighted key feature values ​​to obtain the key features that negatively affect the comprehensive quality index. Based on the production batch information and the weights and dynamic adjustment factors of the key features, the module locates the key nodes where the problems exist.

[0067] The prediction module is used to connect key nodes with production information of each link through the established quality traceability model to form a quality traceability chain; based on the quality traceability chain, the traceability results are obtained.

[0068] In this embodiment of the present invention, comprehensive and automated data collection is achieved across the entire life cycle of Chinese herbal granules (production, processing, transportation, storage, and sales), ensuring the timeliness and accuracy of the data and providing a rich data source for subsequent quality traceability. Preprocessing the collected raw data, including cleaning, conversion, and standardization, effectively eliminates noise and outliers, improving data quality and usability. Using data mining techniques, key features closely related to the quality of Chinese herbal granules are accurately extracted from the preprocessed data. These features can directly reflect the quality status of the product.

[0069] By combining historical data and dynamic adjustment factors and assigning reasonable weights to each key feature, a dynamic and quantitative evaluation of the quality of Chinese medicine formula granules is achieved. This evaluation method is not only flexible, but also can more truly reflect the actual quality level of the product, thereby improving the accuracy and credibility of the quality assessment.

[0070] Based on the comparison results between the comprehensive quality index and the quality standards, it is possible to quickly and accurately determine whether there are quality problems with the Chinese medicine formula granules, and by analyzing the weighted key characteristic values, accurately locate the specific links and factors that affect product quality.

[0071] The established quality traceability model connects production information from all links with key nodes, forming a complete quality traceability chain. This not only helps companies fully understand the process of product quality formation but also predicts potential quality risks. Furthermore, through feedback from traceability results, the quality traceability model can be continuously improved and optimized, enhancing the accuracy and effectiveness of predictions.

[0072] In another preferred embodiment of the present invention, in the production link, data such as raw material information, production process parameters, and equipment operating status are collected; in the processing link, key data such as processing methods, processing environment, and processing time are collected; the transportation link focuses on transportation methods, transportation time, and changes in temperature and humidity during transportation; the storage link collects warehouse environment data (such as temperature, humidity, and light), inventory status, and inventory time; and the sales link records information such as sales time, location, and customer feedback.

[0073] Use sensor technology to monitor key environmental parameters such as temperature, humidity, and pressure in real time; use RFID (radio frequency identification) or QR code technology to track the movement paths of materials and products; integrate information systems such as enterprise resource planning (ERP) and manufacturing execution system (MES) to automatically capture data in the production and processing processes; transmit the collected data to the data center in real time via wired or wireless networks, and establish a secure and reliable database in the data center to store and manage this data.

[0074] The data preprocessing module is immediately followed by the data acquisition module and is responsible for cleaning, converting and standardizing the collected raw data to improve data quality and availability; identifying and removing duplicate records to ensure data uniqueness; detecting and processing missing values, and using deletion, mean filling, interpolation and other methods to complete them according to data characteristics and business rules; using statistical methods (such as Z-score, IQR, etc.) to identify outliers and perform corresponding processing (such as replacement, deletion or retention but marking); performing data type conversion, such as converting text data into numerical data to facilitate subsequent analysis and calculation; performing data normalization or standardization to eliminate dimensional differences and make different features comparable, encoding categorical data (such as unique hot encoding), formulating data standards according to business needs, formatting data, and extracting key features related to the quality of Chinese medicine formula granules, such as raw material quality indicators, production process stability indicators, etc.

[0075] For example, taking a Chinese medicine granule manufacturer as an example, its data collection and preprocessing process is as follows:

[0076] Temperature and humidity sensors are installed on the production line to monitor and record temperature and humidity changes in the production environment in real time. RFID tags are used to track the movement of raw materials from storage to the production line to ensure traceability of their sources. Key process parameters in the production process, such as mixing time and drying temperature, are collected through the MES system. GPS and temperature and humidity monitoring equipment are installed on transport vehicles to record transport trajectories and environmental changes. The collected temperature and humidity data is cleaned to remove outliers caused by sensor failures. RFID tag data is associated with raw material quality information to form a complete raw material traceability data set. The process parameter data in the MES system is normalized and combined with GPS data and temperature and humidity monitoring data to analyze the impact of environmental stability during transportation on product quality.

[0077] In another preferred embodiment of the present invention, key features related to the quality of Chinese medicine formula granules are extracted from the preprocessed data, including:

[0078] Receive pre-processed data related to Chinese medicine formula granules and refer to the pre-established feature library. The feature library contains all the features related to the quality of Chinese medicine formula granules, including chemical composition content, processing temperature, humidity and timestamp; According to the pre-established feature library, filter out the corresponding feature fields from the pre-processed data, including:

[0079] A secure data interface is established to receive TCM granule data transmitted from the data preprocessing module. Once the data preprocessing module completes data preprocessing, a data reception operation is triggered. The preprocessed data is received and stored in a designated data warehouse or database through the data interface to ensure data integrity and security. Before feature screening, a comprehensive feature library is established. This library should contain all features relevant to TCM granule quality, such as chemical content (e.g., active ingredients, impurities), processing temperature, humidity, and timestamp. Each feature should be clearly defined and described in the library. The preprocessed data is traversed field by field, and each field is matched against features in the library to determine whether it is a quality-related feature of TCM granules. If a field successfully matches a feature in the library, it is marked as a key feature field. If a field does not match any features in the library, it is considered a non-key feature field and can be retained or discarded, depending on business needs and data redundancy. The selected key feature fields are organized into a new dataset or data table.

[0080] The correlation coefficient between the feature fields and the quality indicators is calculated to obtain the key features.

[0081] In an embodiment of the present invention, by referencing a pre-established feature library, the system can accurately filter out feature fields directly related to the quality of Chinese herbal formula granules from massive amounts of preprocessed data. This targeted data extraction method greatly reduces the computational complexity of subsequent analysis and improves data processing efficiency. Calculating the correlation coefficient between feature fields and quality indicators helps identify the key features that have the greatest impact on the quality of Chinese herbal formula granules. Quality assessment based on these key features can more accurately reflect the actual quality status of the product, thereby improving the reliability of quality assessment. After understanding which features are critical to quality, companies can allocate resources more rationally, prioritizing and improving those production links and process parameters that have a significant impact on quality. This optimized resource allocation strategy helps companies improve production efficiency while reducing quality risks. The extracted key features and their correlation information with quality indicators provide strong data support for companies to formulate quality management strategies and optimize production processes. During the quality traceability process, the extraction of key features helps quickly locate specific links and factors that may affect product quality, improving traceability efficiency.

[0082] In another preferred embodiment of the present invention, the correlation coefficient between the feature field and the quality index is calculated to obtain the key feature, including:

[0083] pass Calculate the correlation coefficient between the feature field and the quality index;

[0084] Key features are obtained by screening based on correlation coefficients;

[0085] in, represents the correlation coefficient, Represents the total number of observations, which indicates the number of samples in the data set; and Indicates the first of two variables observations; From 1 to , traverse all samples; Represents the summation symbol, indicating the sum from 1 to The summation of all sample points; Representing variables The mean of Representing variables The mean of Representing variables No. The difference between a value and its mean; Representing variables No. The difference between a value and its mean; Indicates the index of the sample point currently being processed, used to traverse from 1 to All sample points; and Is the subscript in the summation, used to calculate the variable and The mean of , in the mean calculation, traverse from 1 to All and value.

[0086] In an embodiment of the present invention, by calculating the correlation coefficient, it is possible to accurately identify which feature fields have a significant impact on the quality of Chinese medicine formula granules, which helps companies focus on improving key production links. After understanding the key features, companies can optimize production processes in a targeted manner, improve product quality by adjusting these key features, and achieve more accurate quality control. By reducing excessive attention to non-critical links, companies can allocate resources more efficiently, thereby improving overall production efficiency. Accurately identifying key features helps companies avoid investing too many resources in unnecessary links, thereby reducing production costs.

[0087] In another preferred embodiment of the present invention, a weight is assigned to each key feature based on historical data, including:

[0088] According to historical data, through Assign a weight to each key characteristic;

[0089] in, It is The correlation coefficient between the key features and the quality indicators; and is the adjustment factor; is the total number of key features; It is The key feature is The probability of a value; It is The number of values ​​of the key features; It is The correlation coefficient between the key features and the quality indicators; It is The number of values ​​of the key features; It is The key feature is The probability of a value; is the index of the key feature; Is calculating the The index used when calculating the information entropy of a key feature; is the index used when calculating the information entropy of each key feature in the denominator; is the index used in the denominator to iterate over all key features.

[0090] In an embodiment of the present invention, by combining the correlation coefficient and information entropy, the weight allocation not only takes into account the direct correlation between the feature and the quality indicator, but also takes into account the diversity of the feature's values, thereby more comprehensively evaluating the importance of the feature. The allocation of weights is based on historical data, which makes the evaluation more objective and accurate. By quantifying the impact of each key feature, enterprises can optimize production processes in a more targeted manner. After understanding the weight of each key feature, enterprises can allocate resources more reasonably and give priority to improving those features with larger weights and more significant impact on quality. The weight allocation process is based on clear mathematical formulas and historical data, which enhances the transparency and credibility of decision-making.

[0091] In another preferred embodiment of the present invention, the calculation formula of the dynamic adjustment factor is:

[0092] ;

[0093] in, and is the weight coefficient; and are the weights of temperature and humidity respectively; and are the ideal temperature and humidity values ​​respectively; and are the actual measured temperature and humidity values ​​respectively; and are the maximum possible ranges of temperature and humidity, respectively; is the number of critical process parameters; It is The actual value of the process parameters; It is target values ​​of the process parameters; It is the maximum possible range of variation of process parameters.

[0094] In the embodiment of the present invention, the formula comprehensively considers multiple influencing factors, including temperature, humidity and key process parameters, and provides a comprehensive evaluation index for the dynamic adjustment of the production environment. and , and the weights of temperature and humidity and , this formula can flexibly adapt to different production needs and environmental conditions. Enterprises can adjust these weights according to actual conditions to reflect the different impacts of various factors on product quality. The formula introduces the difference between ideal values ​​and actual measured values, as well as the ratio of these values ​​to the maximum possible range of variation, thereby achieving precise control of the production environment. This helps to reduce fluctuations in the production process and improve the stability of product quality. By comparing the actual values ​​of process parameters with the target values, the formula helps enterprises identify deviations in the production process, thereby improving production efficiency and product quality. Regularly calculating and monitoring changes in dynamic adjustment factors, timely identifying problems in the production process and making improvements, will help enterprises achieve continuous improvement and continuously improve production efficiency and product quality.

[0095] In another preferred embodiment of the present invention, the actual value of each key feature is multiplied by its corresponding weight and the dynamic adjustment factor to obtain a weighted key feature value; all weighted key feature values ​​are summed to obtain a comprehensive quality index; and the comprehensive quality index is compared with the corresponding quality standard to obtain a comparison result, including:

[0096] Collect the actual values ​​of all key characteristics, ensuring they are up-to-date and accurate. Obtain the weights associated with each key characteristic, determined through previous analysis. Calculate or obtain the current dynamic adjustment factor, which should be based on real-time environmental conditions (such as temperature and humidity) and process parameters. For each key characteristic, multiply its actual value by its corresponding weight to obtain a preliminary weighted value. Multiply each preliminary weighted value by the dynamic adjustment factor to obtain the final weighted key characteristic value. This step allows for real-time adjustment of characteristic values ​​to account for current environmental and process conditions.

[0097] Sum all weighted key characteristic values ​​to obtain an overall value, the comprehensive quality index. This index comprehensively reflects the performance of all key characteristics under current environmental and process conditions. Set or obtain corresponding quality standards. These can be industry-wide standards or standards developed internally based on actual company needs. Compare the calculated comprehensive quality index with the quality standards. The result of this comparison can be a specific numerical difference or a rating (such as "qualified," "unqualified," "excellent," etc.). Output the comparison results in an appropriate format for easy review and understanding by relevant personnel. This output can be in the form of a report, chart, or other visualization tool. Interpret the comparison results to clarify the current quality status of the product or service and any areas that may require improvement or optimization. Based on the comparison results, develop appropriate improvement measures or optimization plans. These measures may include adjusting process parameters, improving the production environment, and strengthening quality control. Repeat this process regularly to continuously monitor and improve product or service quality.

[0098] In another preferred embodiment of the present invention, quality assessment is performed on the new TCM formula granules data based on the comparison results. If the comprehensive quality index is lower than the quality standard, it is determined that the corresponding batch of TCM formula granules has quality problems, including:

[0099] Collect relevant data on new TCM granules, including actual test values ​​for key characteristics such as active ingredient content, dissolution rate, and impurity content, to ensure data accuracy and completeness. Perform necessary data preprocessing, such as cleaning and formatting. Apply the previously determined weights and dynamic adjustment factor calculation formulas, combined with the newly collected data, to calculate weighted key characteristic values. Sum all weighted key characteristic values ​​to obtain the comprehensive quality index for the batch of TCM granules.

[0100] Consult relevant quality standard documents or databases to obtain the quality standard values ​​for TCM granules. This standard can be set based on historical data requirements. Compare the calculated comprehensive quality index with the obtained quality standard. If the comprehensive quality index is lower than the quality standard, the batch of TCM granules is deemed to have quality issues. If the comprehensive quality index meets or exceeds the quality standard, the batch of TCM granules is deemed qualified. Record the results of this determination in the quality management system or relevant documentation. If quality issues are identified, immediately generate a quality report detailing the issues and possible causes. For batches with quality issues, based on the analysis in the quality report, implement appropriate corrective measures, such as reprocessing, formula adjustment, and production process improvements. Monitor the effectiveness of corrective measures and make necessary adjustments to ensure the quality of subsequent production batches. Analyze the causes of quality issues, identify weak links in the production process, and implement continuous improvements in these areas, such as optimizing production processes, improving employee skills, and upgrading equipment.

[0101] In another preferred embodiment of the present invention, the weighted key feature values ​​are analyzed to obtain key features that negatively impact the comprehensive quality index, including:

[0102] Perform data standardization on the weighted eigenvalues ​​to obtain standardized data; calculate the statistics of each weighted eigenvalue in the standardized data, including the mean, standard deviation, median, and quartiles, to understand the distribution of the standardized data and the range of variation of the eigenvalues; set a threshold based on the statistics to determine whether the weighted eigenvalue is abnormal; traverse all weighted eigenvalues, compare each weighted eigenvalue with the threshold, and mark abnormal weighted eigenvalues; for each weighted eigenvalue marked as abnormal, remove the corresponding weighted eigenvalue from the standardized data, and recalculate the comprehensive quality index based on the remaining weighted eigenvalues; compare the difference between the original comprehensive quality index and the comprehensive quality index after removing the abnormal features, specifically including:

[0103] For each weighted eigenvalue, first calculate its overall mean and standard deviation. Using the z-score standardization method, subtract the mean from each weighted eigenvalue and then divide by the standard deviation to obtain the standardized data. For each standardized weighted eigenvalue, calculate its mean and standard deviation to confirm that the standardization is correct (the mean should be close to 0 and the standard deviation should be close to 1). Calculate the median to understand the central tendency of the data, and the quartiles (Q1, Q2, and Q3) to identify the distribution range and potential outliers. Based on these calculated statistics, particularly the quartiles, set a threshold for outlier detection. For example, 1.5 or 3 times the interquartile range (IQR) (Q3 - Q1) can be used as the outlier detection threshold. Any weighted eigenvalue below (Q1 - 1.5 × IQR) or above (Q3 + 1.5 × IQR) is considered an outlier. Iterate through all standardized weighted eigenvalues ​​and compare each value to the thresholds set above. If a weighted eigenvalue exceeds the threshold range, it is marked as an outlier. For each weighted feature value marked as abnormal, it is removed from the standardized data set, and the remaining normal weighted feature values ​​are used to recalculate the comprehensive quality index according to the original comprehensive quality index calculation formula; the original comprehensive quality index is compared with the comprehensive quality index calculated after removing the abnormal features, and the difference between the two is analyzed to understand the degree of influence of the abnormal feature value on the comprehensive quality index.

[0104] All abnormal weighted eigenvalues ​​are sorted according to their respective impact on the comprehensive quality index, from large to small. A negative impact threshold is set. When the impact of an abnormal weighted eigenvalue on the comprehensive quality index exceeds the negative impact threshold, the abnormal weighted eigenvalue is considered a key feature that has a negative impact on the overall quality. From the sorted list of abnormal weighted eigenvalues, the key features that exceed the negative impact threshold are screened, including:

[0105] Ensure that weighted eigenvalues ​​are detected for anomalies and maintain a list of all anomalous weighted eigenvalues. For each anomalous weighted eigenvalue, retain its original value, normalized value, and the corresponding change in the overall quality index (i.e., the change in the overall quality index after removing the eigenvalue). For each anomalous weighted eigenvalue in the list, calculate its impact on the overall quality index by comparing the difference between the overall quality index with the eigenvalue and the overall quality index after removing it. Save this impact (difference) as a key attribute of the anomalous eigenvalue. Sort each anomalous weighted eigenvalue by its impact on the overall quality index. Use an appropriate sorting algorithm (such as quick sort or merge sort) to arrange the eigenvalues ​​in descending order of impact. Set a negative impact threshold based on historical data, industry standards, or expert advice to determine whether the anomalous weighted eigenvalue has a significant negative impact on overall quality. Traverse the sorted list of abnormal weighted eigenvalues ​​and compare the impact of each eigenvalue with the negative impact threshold. If the impact of a certain eigenvalue exceeds the negative impact threshold, it is regarded as a key feature that has a negative impact on the overall quality. These key features are listed separately to form a new list or set.

[0106] In an embodiment of the present invention, by conducting in-depth analysis of weighted key eigenvalues, this method can accurately identify key features that negatively impact the overall quality index. By normalizing the weighted eigenvalues, the effects of dimensionality and magnitude between different eigenvalues ​​are eliminated, making them comparable and improving the accuracy and reliability of data analysis. Calculating statistics (such as mean, standard deviation, median, and quartiles) for each weighted eigenvalue in the standardized data helps to fully understand the data distribution and the range of variation of the eigenvalues. Setting thresholds based on these statistics automatically detects abnormal weighted eigenvalues. This significantly improves data processing efficiency and reduces the potential for human error. By comparing the difference between the original overall quality index and the overall quality index after removing abnormal features, this method can quantitatively assess the specific impact of each abnormal feature on overall quality. Sorting all abnormal weighted eigenvalues ​​according to their impact on the overall quality index helps companies prioritize improvement efforts. Companies can prioritize key features with the greatest impact on the overall quality index to rapidly improve product quality. Setting a negative impact threshold ensures that the identified key features have a significant negative impact on overall quality. This avoids unnecessary improvement work and improves the efficiency of enterprise resource utilization.

[0107] In another preferred embodiment of the present invention, locating critical nodes with problems based on production batch information, weights of key features, and dynamic adjustment factors includes:

[0108] For each production batch, multiply the actual measured value of each key characteristic by its corresponding weight to obtain a weighted score. Multiply the weighted value by the corresponding dynamic adjustment factor to reflect the actual situation in the current production environment. Compare the weighted scores of different production batches to identify batches with abnormal scores. Based on the batches with abnormal scores, analyze each link in their production process, including raw material quality and equipment status, to obtain production process analysis results. Based on the production process analysis and weighted scores, locate the key nodes that lead to product quality decline, including:

[0109] Collect the actual measured values ​​of each key characteristic in each production batch, ensuring that each key characteristic has a corresponding weight value that reflects the importance of the characteristic to product quality; obtain or calculate the dynamic adjustment factor under the current production environment; for each production batch, multiply the actual measured value of each key characteristic by its corresponding weight, and then add these products to obtain the weighted score of the batch; multiply the weighted score of each production batch by the corresponding dynamic adjustment factor to reflect the actual situation under the current production environment; compare the adjusted weighted scores of all production batches to identify batches with abnormally high or low scores, and use statistical methods (such as standard deviation, quartiles, etc.) to set thresholds for abnormal scores. For batches with abnormal scores, conduct in-depth analysis of each link in their production process, including quality inspection records of raw materials, equipment operation status logs, operator skill levels, etc., and identify differences that may lead to deterioration in product quality by comparing production process data of normal batches and abnormal batches.

[0110] Integrate the production process analysis results and the adjusted weighted score data to ensure they are aligned with production batches. For production process analysis, pay particular attention to key data such as raw material quality inspection reports, equipment maintenance records, and operator logs. Compare the weighted scores of different batches and flag those with unusually low scores or large fluctuations, as these may contain quality issues. Preliminarily screen these abnormal batches for any problematic aspects of the production process analysis, such as raw material quality fluctuations and equipment failure records.

[0111] For the problem links that have been initially screened out, further analysis is conducted on their correlation with abnormal weighted scores. For example, if a batch is found to have unstable quality during the raw material quality inspection, and the weighted score of the batch is also abnormally low, then the unstable raw material quality can be considered a key node.

[0112] Through statistical analysis and expert evaluation, we identify those links that have a significant impact on the weighted score and do have problems in the production process analysis as key nodes. Possible key nodes include but are not limited to: unstable raw material quality, performance degradation due to equipment aging, improper operation, unreasonable process parameter settings, etc.

[0113] In an embodiment of the present invention, by comprehensively considering the weights and dynamic adjustment factors of key features, the method can more accurately locate the key nodes that affect product quality. This helps enterprises quickly find and solve problems, improve production efficiency and product quality. By comparing the weighted scores of different production batches, enterprises can allocate resources more wisely and devote more attention and resources to batches with abnormal scores and more problems, thereby improving resource utilization efficiency. By conducting in-depth analysis of each link in the production process, this method enables enterprises to more clearly understand the impact of each link in the production process on product quality, and then optimize the production process. By locating the key nodes that lead to product quality decline, enterprises can more effectively manage risks and prevent potential product quality problems. This method not only helps enterprises locate existing problems, but also provides enterprises with a framework for continuous improvement. By continuously collecting and analyzing production data, enterprises can continuously optimize the weights and dynamic adjustment factors of key features, thereby continuously improving product quality and production efficiency.

[0114] In another preferred embodiment of the present invention, a quality traceability model is established to connect key nodes with production information of each link in series to form a quality traceability chain. According to the quality traceability chain, a traceability result is obtained, including:

[0115] Obtain and analyze the production process, understand the entire production process in detail, including raw material procurement, processing, quality inspection, packaging, warehousing, logistics and other links, determine the input, output and key control points of each link, and record them in the flow chart.

[0116] Identify key nodes. Identify and record key nodes that have a significant impact on product quality, such as raw material acceptance, key process steps, and important quality inspection links, ensuring that each key node is clearly defined and identified. Design data structures and associations. Based on key nodes and business processes, design a reasonable data structure to store relevant information, such as raw material batches, production dates, process parameters, quality inspection results, etc. Use unique product identifiers (such as batch numbers and serial numbers) to associate data from each link to ensure data consistency and traceability.

[0117] Establish a data collection and transmission system, setting up data collection points at each key node. Utilize automated methods (such as barcode technology, RFID tags, and sensors) to accurately and real-timely collect data, ensuring efficient and accurate data collection and reducing human error. Build a database and traceability system, constructing a secure and reliable database to store and manage traceability data. Develop user-friendly query interfaces and interfaces to support rapid access to traceability information using product identifiers. Based on actual needs, set precise traceability query criteria, such as product batch number, production date, and quality issue type, to ensure that the query criteria accurately matches the product or production process requiring traceability.

[0118] To perform a drill-through query:

[0119] Entering predefined query criteria into the quality traceability system allows users to quickly locate relevant key nodes and production information within the quality traceability chain. The system then displays the traceability results in intuitive formats (such as tables, charts, or reports), including detailed data on key nodes and relevant information about production processes. Users can analyze the traceability results to determine the root cause and impact of product quality issues. Based on the analysis, they can develop appropriate improvement measures and preventive plans to improve product quality and production efficiency. These improvement measures can then be fed back into the production process, continuously improving and optimizing the quality traceability model.

[0120] Regularly evaluate and optimize the quality traceability system to ensure that it adapts to changes in production processes and new traceability requirements, monitor data collection and transmission at key nodes to ensure the real-time and accuracy of data, and conduct regular training for employees to improve their proficiency and efficiency in using the quality traceability system.

[0121] like Figure 2 As shown, the embodiment of the present invention also provides a method for tracing the quality of Chinese medicine formula granules based on data fusion, including:

[0122] Collect parameter data of Chinese medicine formula granules in each link of production, processing, transportation, storage and sales;

[0123] preprocessing the parameter data to obtain preprocessed data;

[0124] Extract key features related to the quality of TCM granules from preprocessed data;

[0125] Assign a weight to each key characteristic based on historical data; define a dynamic adjustment factor based on actual conditions; multiply the actual value of each key characteristic by its corresponding weight and the dynamic adjustment factor to obtain a weighted key characteristic value; sum all weighted key characteristic values ​​to obtain a comprehensive quality index; compare the comprehensive quality index with the corresponding quality standard to obtain a comparison result;

[0126] Based on the comparison results, the new TCM formula granule data is quality-judged. If the comprehensive quality index is lower than the quality standard, the corresponding batch of TCM formula granules is determined to have quality problems. The weighted key feature values ​​are analyzed to obtain the key features that negatively affect the comprehensive quality index. Based on the production batch information, the weights of the key features, and the dynamic adjustment factors, the key nodes with problems are located.

[0127] By establishing a quality traceability model, key nodes are connected with production information of each link to form a quality traceability chain; based on the quality traceability chain, traceability results are obtained.

[0128] It should be noted that this system is a system corresponding to the above method, and all implementation methods in the above method embodiment are applicable to this embodiment and can achieve the same technical effects.

[0129] An embodiment of the present invention further provides a computing device comprising: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the above-described method. All implementations in the above-described method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0130] The embodiment of the present invention further provides a computer-readable storage medium storing instructions, which, when executed on a computer, causes the computer to execute the above-described method. All implementations in the above-described method embodiment are applicable to this embodiment and can achieve the same technical effects.

[0131] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. The quality traceability system of Chinese medicine formula granules based on data fusion is characterized by: include: Data collection module, used to collect parameter data of Chinese medicine formula granules in various links of production, processing, transportation, storage and sales; A data preprocessing module, used for preprocessing parameter data to obtain preprocessed data; Extraction module, used to extract key features related to the quality of traditional Chinese medicine granules from preprocessed data; A quality assessment calculation module is used to assign a weight to each key feature based on historical data; Define a dynamic adjustment factor based on actual conditions; multiply the actual value of each key characteristic by its corresponding weight and the dynamic adjustment factor to obtain a weighted key characteristic value; sum all weighted key characteristic values ​​to obtain a comprehensive quality index; compare the comprehensive quality index with the corresponding quality standard to obtain a comparison result; The quality traceability module is used to determine the quality of the new TCM formula granules based on the comparison results. If the comprehensive quality index is lower than the quality standard, it is determined that the corresponding batch of TCM formula granules has quality problems; Analyze the weighted key eigenvalues ​​to obtain the key features that have a negative impact on the comprehensive quality index, including: performing data standardization on the weighted eigenvalues ​​to obtain standardized data; calculating the statistics of each weighted eigenvalue in the standardized data, including the mean, standard deviation, median and quartiles, to understand the distribution of the standardized data and the range of variation of the eigenvalues; based on the statistics, set a threshold to determine whether the weighted eigenvalue is abnormal; traverse all weighted eigenvalues, compare each weighted eigenvalue with the threshold, and mark the abnormal weighted eigenvalues; for each weighted eigenvalue marked as abnormal, remove the corresponding weighted eigenvalue from the standardized data, and recalculate the comprehensive quality index based on the remaining weighted eigenvalues; compare the difference between the original comprehensive quality index and the comprehensive quality index after removing the abnormal features; sort all abnormal weighted eigenvalues ​​according to the degree of influence on the comprehensive quality index, from large to small; set a A negative impact threshold is set. When the degree of influence of the abnormal weighted feature value on the comprehensive quality index exceeds the negative impact threshold, the abnormal weighted feature value is regarded as a key feature that has a negative impact on the overall quality; the key features that exceed the negative impact threshold are screened out from the sorted abnormal weighted feature value list; the key nodes with problems are located according to the production batch information and the weights and dynamic adjustment factors of the key features, including: for each production batch, the actual measurement value of each key feature is multiplied by its corresponding weight to obtain a weighted score; the weighted value is multiplied by the corresponding dynamic adjustment factor to reflect the actual situation under the current production environment; the weighted scores of different production batches are compared to obtain batches with abnormal scores; according to the batches with abnormal scores, each link in the production process is analyzed, including the quality of raw materials and equipment status, to obtain the production process analysis results; based on the production process analysis and weighted scores, the key nodes that lead to product quality decline are located; The prediction module is used to connect key nodes with production information of each link through the established quality traceability model to form a quality traceability chain; obtain traceability results based on the quality traceability chain; and extract key features related to the quality of Chinese medicine formula granules from the preprocessed data, including: Receive pre-processed data related to Chinese medicine formula granules and refer to a pre-established feature library, which contains all features related to the quality of Chinese medicine formula granules, including chemical component content, processing temperature, humidity and timestamp; According to the pre-established feature library, the corresponding feature fields are filtered out from the pre-processed data; Based on historical data, each key characteristic is assigned a weight, including: According to historical data, through Assign a weight to each key characteristic; in, It is The correlation coefficient between the key features and the quality indicators; and is the adjustment factor; is the total number of key features; It is The key feature is The probability of a value; It is The number of values ​​of the key features; It is The correlation coefficient between the key features and the quality indicators; It is The number of values ​​of the key features; It is The key feature is The probability of a value; is the index of the key feature; Is calculating the The index used when calculating the information entropy of a key feature; is the index used when calculating the information entropy of each key feature in the denominator; is the index used in the denominator to traverse all key features; calculate the correlation coefficient between the feature field and the quality index to obtain the key features; calculate the correlation coefficient between the feature field and the quality index to obtain the key features, including: pass Calculate the correlation coefficient between the feature field and the quality index; Key features are obtained by screening based on correlation coefficients; in, represents the correlation coefficient, Represents the total number of observations, which indicates the number of samples in the data set; and Indicates the first of two variables observations; From 1 to , traverse all samples; Represents the summation symbol, indicating the sum from 1 to The summation of all sample points; Representing variables The mean of Representing variables The mean of Representing variables No. The difference between a value and its mean; Representing variables No. The difference between a value and its mean; Indicates the index of the sample point currently being processed, used to traverse from 1 to All sample points; and Is the subscript in the summation, used to calculate the variable and The mean of , in the mean calculation, traverse from 1 to All and value.

2. A method for tracing the quality of Chinese medicine granules based on data fusion, characterized in that: Applicable to the system as claimed in claim 1, comprising: Collect parameter data of Chinese medicine formula granules in each link of production, processing, transportation, storage and sales; preprocessing the parameter data to obtain preprocessed data; Extract key features related to the quality of TCM granules from preprocessed data; Assign a weight to each key characteristic based on historical data; define a dynamic adjustment factor based on actual conditions; multiply the actual value of each key characteristic by its corresponding weight and the dynamic adjustment factor to obtain a weighted key characteristic value; sum all weighted key characteristic values ​​to obtain a comprehensive quality index; compare the comprehensive quality index with the corresponding quality standard to obtain a comparison result; Based on the comparison results, the new TCM formula granule data is quality-judged. If the comprehensive quality index is lower than the quality standard, the corresponding batch of TCM formula granules is determined to have quality problems. The weighted key feature values ​​are analyzed to obtain the key features that negatively affect the comprehensive quality index. Based on the production batch information, the weights of the key features, and the dynamic adjustment factors, the key nodes with problems are located. By establishing a quality traceability model, key nodes are connected with production information of each link to form a quality traceability chain; based on the quality traceability chain, traceability results are obtained.

3. A computing device, characterized in that include: one or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, causes the one or more processors to implement the method according to claim 1.

4. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a program, which implements the method according to claim 1 when executed by a processor.

Images