A method and system for optimizing the control of multi-process drums in the processing of traditional Chinese medicine decoction pieces.

Abstract

This application proposes a multi-stage roller optimization control method and system for processing traditional Chinese medicine decoction pieces, belonging to the field of intelligent control. The method includes: collecting healthy samples based on roller processing modes, decoction piece specifications and varieties, and roller models; performing correlation analysis on each evaluation index to obtain the correlation attributes and correlation degrees; decomposing processing deviation parameters based on the correlation attributes to obtain the correlation deviation parameters of each index; performing weighted calculations using the correlation degree as weight to obtain deviation characteristic values; dynamically configuring the weights of evaluation indicators using the deviation characteristic values ​​to calculate the sample fitness; and finally storing the samples and fitness in an optimization particle space to perform optimization to obtain the desired control parameters. By dynamically allocating index weights, the performance of weaker indicators is enhanced while the performance of stronger indicators is weakened, solving the optimization bias problem caused by existing fixed-weight evaluations, achieving balanced optimization of multi-stage roller control parameters, and improving the processing quality of traditional Chinese medicine decoction pieces.

Description

Technical Field

[0001] This invention relates to the field of intelligent control, and in particular to a multi-process drum optimization control method and system for processing Chinese herbal medicine slices. Background Technology

[0002] In the industrial production of Chinese herbal medicine (TCM) decoction pieces, the drum processing typically involves multiple steps such as cleaning, screening, and stir-frying. Each step requires precise control of corresponding processing parameters to ensure that the quality of the decoction pieces meets standards. In actual production, the quality assessment of TCM decoction pieces involves multiple indicators, such as moisture content, active ingredient content, appearance and color, and breakage rate. To obtain high-quality decoction piece products, the drum processing parameters need to be optimized and controlled. However, existing optimization control methods typically use a fixed-weight approach to comprehensively evaluate various assessment indicators. During the optimization process, this fixed-weight evaluation method tends to favor indicators that perform well, while neglecting those that perform poorly. This results in control parameters that, while excellent in some indicators, may have significant shortcomings in others, failing to achieve balanced optimization across all indicators. Summary of the Invention

[0003] This invention addresses the technical problem in existing technologies where the fixed-weight evaluation method in the multi-process roller processing of traditional Chinese medicine decoction pieces leads to biased optimization results towards indicators with superior performance while neglecting those with inferior performance. The invention provides a multi-process roller optimization control method and system for this purpose.

[0004] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0005] In a first aspect, the present invention provides a multi-process roller optimization control method for processing traditional Chinese medicine decoction pieces, comprising: collecting a first healthy sample based on the roller processing mode, the specifications and varieties of traditional Chinese medicine decoction pieces, and the roller model, wherein the first healthy sample includes a first processing deviation parameter and a preset evaluation index detection value; traversing the preset evaluation indexes, performing correlation analysis on the processing control attributes to obtain the correlation attributes and correlation degrees of the first evaluation indexes, up to the Nth evaluation index correlation attribute and the Nth evaluation index correlation degree; decomposing the first processing deviation parameter according to the correlation attributes up to the Nth evaluation index correlation attribute to obtain the correlation deviation parameters up to the Nth evaluation index correlation parameter; and converting the first evaluation index into a single data set. The correlation degree of the indicators up to the Nth evaluation indicator is used as the normalized deviation weight. The weighted mean of the correlation deviation parameters of the first evaluation indicator up to the Nth evaluation indicator is calculated to obtain the correlation attribute deviation feature value of the first evaluation indicator up to the Nth evaluation indicator. The weight of the preset evaluation indicator detection value is configured based on the correlation attribute deviation feature value of the first evaluation indicator up to the Nth evaluation indicator. The first fitness of the first healthy sample is calculated. The first healthy sample and the first fitness are associated and stored in the optimization particle space. When the number of healthy samples in the optimization particle space meets the preset number, optimization is performed to obtain the desired processing parameters for initialization of the roller processing mode.

[0006] Secondly, the present invention provides a multi-process roller optimization control system for processing traditional Chinese medicine decoction pieces, comprising: a health sample collection module, used to collect a first health sample based on the roller processing mode, the specifications and varieties of traditional Chinese medicine decoction pieces, and the roller model, wherein the first health sample includes a first processing deviation parameter and a preset evaluation index detection value; a correlation analysis module, used to traverse the preset evaluation indexes, perform correlation analysis on the processing control attributes, and obtain the correlation attributes and correlation degrees of the first evaluation indexes, up to the Nth evaluation index correlation attribute and the Nth evaluation index correlation degree; a deviation parameter decomposition module, used to decompose the first processing deviation parameter according to the correlation attributes up to the Nth evaluation index, and obtain the first evaluation index related deviation parameters up to the Nth evaluation index related deviation parameters; and a deviation feature calculation module. The system is configured to use the correlation degree of the first evaluation index up to the correlation degree of the Nth evaluation index as a normalized deviation weight, and to calculate the weighted mean of the correlation deviation parameters of the first evaluation index up to the correlation deviation parameters of the Nth evaluation index to obtain the correlation attribute deviation feature value of the first evaluation index up to the correlation attribute deviation feature value of the Nth evaluation index; the fitness calculation module is configured to use the correlation attribute deviation feature value of the first evaluation index up to the correlation attribute deviation feature value of the Nth evaluation index to configure the weight of the preset evaluation index detection value, and calculate the first fitness of the first healthy sample; the processing mode initialization module is configured to store the first healthy sample and the first fitness in an optimization particle space, and when the number of healthy samples in the optimization particle space meets the preset number, to perform optimization and obtain the desired processing parameters to initialize the roller processing mode.

[0007] The beneficial effects of this invention are:

[0008] Based on the roller processing mode, the specifications and varieties of Chinese herbal medicine slices, and the roller model, a first healthy sample was collected. This first healthy sample included the first processing deviation parameter and the detection values ​​of preset evaluation indicators, thus establishing the basic data for optimized control and ensuring that the collected samples could reflect the processing characteristics under specific working conditions. The preset evaluation indicators were traversed, and correlation analysis was performed on the processing control attributes to obtain the correlation attributes and correlation degrees of the first evaluation indicator, up to the Nth evaluation indicator's correlation attributes and correlation degrees. Through correlation analysis, it was identified which control attributes each evaluation indicator was related to and the degree of correlation, providing a basis for subsequent parameter decomposition. Based on the correlation attributes of the first evaluation indicator up to the Nth evaluation indicator, the first processing deviation parameter was decomposed, obtaining the related deviation parameters of the first evaluation indicator up to the Nth evaluation indicator. This decomposed the overall processing deviation parameters into sub-parameters related to each evaluation indicator, achieving refined parameter management.

[0009] Using the correlation degree of the first evaluation index up to the correlation degree of the Nth evaluation index as normalized deviation weights, a weighted average is calculated on the correlation deviation parameters of the first evaluation index up to the Nth evaluation index to obtain the correlation attribute deviation characteristic value of the first evaluation index up to the Nth evaluation index. Through weighted calculation, deviation characteristic values ​​reflecting the characteristics of each evaluation index are obtained. Using the correlation attribute deviation characteristic values ​​of the first evaluation index up to the Nth evaluation index, the weights of preset evaluation index detection values ​​are configured, and the first fitness of the first healthy sample is calculated. This achieves dynamic weight allocation, allowing indicators with poor performance to receive higher weights, while indicators with good performance have correspondingly lower weights, thus achieving balanced optimization. The first healthy sample and the first fitness are associated and stored in the optimization particle space. When the number of healthy samples in the optimization particle space meets the preset number, optimization is performed to obtain the desired processing parameters for initializing the roller processing mode. This process searches for the optimal solution in the constructed optimization particle space, obtaining control parameters that achieve a relative balance among the evaluation indicators.

[0010] The above technical solution achieves dynamic balance optimization of control parameters for multi-process drum processing of Chinese herbal medicine pieces. This allows for the automatic enhancement of the weight of indicators with poor performance and the weakening of the weight of indicators with good performance during the optimization process. This avoids the optimization bias problem caused by the traditional fixed weight method. In order to find the optimal expected processing parameters with relatively balanced evaluation indicators in the constructed optimization particle space, the overall processing quality of Chinese herbal medicine pieces can be improved. Attached Figure Description

[0011] Figure 1 A flowchart illustrating a multi-stage drum optimization control method for processing traditional Chinese medicine decoction pieces provided by the present invention;

[0012] Figure 2 This is a schematic diagram of the structure of a multi-process drum optimization control system for processing Chinese herbal medicine slices provided by the present invention.

[0013] In the attached diagram, the components represented by each number are as follows:

[0014] The module includes a health sample collection module 11, a correlation analysis module 12, a deviation parameter decomposition module 13, a deviation feature calculation module 14, a fitness calculation module 15, and a processing mode initialization module 16. Detailed Implementation

[0015] 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0017] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0018] Example 1, as Figure 1 As shown, this embodiment of the invention provides a multi-process drum optimization control method for processing Chinese herbal medicine pieces, which is applied to the processing drum of Chinese herbal medicine pieces. The processing drum of Chinese herbal medicine pieces includes a cleaning processing mode, a screening processing mode, and a stir-frying processing mode.

[0019] Specifically, the method provided in this application embodiment is applied to a processing drum for traditional Chinese medicine (TCM) decoction pieces. This processing drum is an integrated, multi-functional processing device capable of performing multiple processes in the processing of TCM decoction pieces. Specifically, the TCM decoction piece processing drum has three main processing modes: a cleaning processing mode, a screening processing mode, and a stir-frying processing mode. The cleaning processing mode is used to perform pretreatment operations such as washing, impurity removal, and dust removal on the raw materials of TCM decoction pieces to remove mud, impurities, and other contaminants from the surface of the medicinal materials, ensuring that the cleanliness of the TCM decoction pieces meets pharmaceutical standards. The screening processing mode is used to grade and screen the cleaned TCM decoction pieces, classifying them according to their size, shape, quality, and other characteristics to meet the production needs of TCM decoction pieces with different specifications. The stir-frying processing mode is used to stir-fry the screened TCM decoction pieces, achieving different degrees of processing such as yellowing, charring, and carbonization by controlling parameters such as temperature, time, and rotation speed, thereby altering the medicinal properties or enhancing the efficacy of the medicinal materials.

[0020] By integrating the above three processing modes onto the processing roller of Chinese herbal medicine slices, continuous and automated production of Chinese herbal medicine slices can be achieved, improving production efficiency and ensuring the stability and consistency of product quality.

[0021] This multi-process roller optimization control method includes:

[0022] S1. Based on the roller processing mode, the specifications and varieties of Chinese herbal medicine slices, and the roller model, a first health sample is collected, wherein the first health sample includes a first processing deviation parameter and a preset evaluation index detection value.

[0023] Specifically, the processing mode of the roller refers to the current working state of the processing roller for Chinese herbal medicine pieces, which can be one of the cleaning processing mode, screening processing mode, or stir-frying processing mode; the specifications of Chinese herbal medicine pieces correspond to the size parameter requirements of the pieces; the varieties of Chinese herbal medicine pieces correspond to different types of Chinese herbal materials; and the roller model corresponds to the specific equipment model used for processing the Chinese herbal medicine pieces.

[0024] Based on the drum processing mode, specifications of Chinese herbal medicine (TCM) slices, varieties of TCM slices, and drum model, historical processing samples are retrieved from the historical processing database of TCM slices to obtain the first healthy sample that meets the criteria. Meeting the criteria means that the drum processing mode, specifications of TCM slices, varieties of TCM slices, and drum model of the historical sample match the current process conditions to be optimized, and the processing result corresponding to the historical sample meets the quality standard requirements; thus, it is considered a healthy sample. The first processing deviation parameter refers to the difference between various actual control parameters and standard control parameters during processing, including but not limited to temperature deviation, speed deviation, time deviation, and pressure deviation. The preset evaluation index detection value refers to the actual measured value of various indicators used to evaluate the processing quality of TCM slices, including but not limited to moisture content, breakage rate, color uniformity, effective ingredient content, and appearance morphology score.

[0025] By using a multi-dimensional sampling method, we ensured that the first healthy sample could accurately reflect the processing status and quality performance under specific process conditions, laying a data foundation for subsequent optimization and control analysis.

[0026] S2. Traverse the preset evaluation indicators, perform correlation analysis on the processing control attributes, and obtain the correlation attributes and correlation degree of the first evaluation indicator, until the correlation attributes and correlation degree of the Nth evaluation indicator.

[0027] Specifically, the preset evaluation indicators refer to various indicators used to evaluate the processing quality of Chinese herbal medicine slices, such as moisture content, breakage rate, color uniformity, content of effective ingredients, and appearance morphology score; the processing control attributes refer to various parameter attributes that can be adjusted and controlled during the processing of Chinese herbal medicine slices, including temperature, rotation speed, time, pressure, and humidity.

[0028] By iterating through the data, each preset evaluation indicator is analyzed sequentially, with one indicator selected as the analysis object at a time, designated as the first evaluation indicator. For the first evaluation indicator, its correlation with each processing control attribute is analyzed to determine which control attributes have a significant impact on it; these significantly influential control attributes are the correlation attributes of the first evaluation indicator. Simultaneously, the degree of correlation between these correlation attributes and the first evaluation indicator is calculated, yielding the correlation degree of the first evaluation indicator. Following the same method, correlation analysis is performed on the second, third, and so on up to the Nth preset evaluation indicator, obtaining the correlation attributes and correlation degrees for the second and third evaluation indicators, respectively, until the correlation attributes and correlation degree for the Nth evaluation indicator are obtained. Here, N represents the total number of preset evaluation indicators.

[0029] By using traversal analysis, it is possible to comprehensively identify which processing control attributes are related to each preset evaluation index, as well as the strength of the correlation, providing accurate attribute mapping relationships for subsequent parameter optimization.

[0030] S3. Based on the association attributes of the first evaluation index up to the Nth evaluation index, decompose the first processing deviation parameter to obtain the association deviation parameters of the first evaluation index up to the Nth evaluation index.

[0031] Specifically, the first processing deviation parameter is the set of deviations of various control parameters included in the first healthy sample, including deviations of all processing control attributes such as temperature deviation, speed deviation, time deviation, and pressure deviation. The first evaluation index association attribute to the Nth evaluation index association attribute respectively represent the set of control attributes that are associated with each preset evaluation index.

[0032] Based on the obtained association attributes of the first evaluation index up to the Nth evaluation index, deviation parameters related to each preset evaluation index are extracted from the first processing deviation parameters. For example, if the association attributes of the first evaluation index (such as moisture content) include temperature and time, then temperature deviation and time deviation are extracted from the first processing deviation parameters to form the first evaluation index association deviation parameters; if the association attributes of the second evaluation index (such as breakage rate) include rotation speed and pressure, then rotation speed deviation and pressure deviation are extracted from the first processing deviation parameters to form the second evaluation index association deviation parameters. Following this decomposition method, all preset evaluation indicators are processed sequentially to finally obtain the first evaluation index association deviation parameters, the second evaluation index association deviation parameters, and so on up to the Nth evaluation index association deviation parameters. Each evaluation index association deviation parameter only contains the control attribute deviation values ​​that affect that preset evaluation index, achieving accurate classification and targeted extraction of deviation parameters.

[0033] By using the above decomposition method, the original comprehensive deviation parameters are grouped according to their correlation with each preset evaluation index, providing a structured data foundation for subsequent targeted deviation analysis and weight calculation.

[0034] S4. Using the correlation degree of the first evaluation indicator up to the correlation degree of the Nth evaluation indicator as the normalized deviation weight, calculate the weighted average of the correlation deviation parameters of the first evaluation indicator up to the correlation deviation parameters of the Nth evaluation indicator to obtain the correlation attribute deviation characteristic value of the first evaluation indicator up to the correlation attribute deviation characteristic value of the Nth evaluation indicator.

[0035] Specifically, the correlation degree from the first evaluation index to the Nth evaluation index represents the strength of the correlation between each processing control attribute and the corresponding preset evaluation index; the correlation deviation parameter from the first evaluation index to the Nth evaluation index includes the control attribute deviation value related to each preset evaluation index.

[0036] First, the correlation scores of the first evaluation indicator up to the Nth evaluation indicator are normalized. Specifically, taking the first evaluation indicator as an example, its correlation score includes the correlation scores of multiple related attributes. These correlation scores are summed to obtain a total, and then each correlation score is divided by this total to obtain the normalized correlation score, which is used as the normalization bias weight. For example, if the first evaluation indicator has three related attributes with correlation scores of 0.6, 0.4, and 0.3 respectively, the total correlation score is 1.3, and the normalized weights are 0.6 / 1.3≈0.46, 0.4 / 1.3≈0.31, and 0.3 / 1.3≈0.23 respectively. Normalization ensures that the sum of the weight coefficients is 1, giving control attributes with high correlation scores a greater influence weight in subsequent calculations.

[0037] Then, using normalized deviation weights, a weighted average is calculated for the corresponding correlation deviation parameters. Specifically, for the first evaluation indicator, each correlation deviation parameter is multiplied by its corresponding normalized deviation weight, and all products are summed to obtain the correlation attribute deviation characteristic value of the first evaluation indicator. This characteristic value comprehensively reflects the weighted influence of all relevant control attribute deviations on the evaluation indicator. Following the same processing method, the correlation degrees of the second to Nth evaluation indicators are normalized sequentially, and the corresponding correlation attribute deviation characteristic values ​​are calculated, thereby obtaining the correlation attribute deviation characteristic values ​​of the first to Nth evaluation indicators.

[0038] By normalizing and using weighted average calculation, the deviation values ​​of multiple related attributes are weighted and synthesized according to their correlation strength to obtain feature values ​​that can accurately reflect the degree of influence of the deviation of control attributes on each evaluation index, thus providing a standardized data foundation for subsequent fitness calculation.

[0039] S5. Using the first evaluation index related attribute deviation feature value up to the Nth evaluation index related attribute deviation feature value, configure the weight of the preset evaluation index detection value, and calculate the first fitness of the first healthy sample.

[0040] Specifically, the characteristic value of the deviation of the associated attribute of the first evaluation indicator to the characteristic value of the deviation of the associated attribute of the Nth evaluation indicator reflects the comprehensive influence of the deviation of the controlled attribute on each evaluation indicator; the preset evaluation indicator detection value is the actual measurement value of each quality indicator included in the first health sample.

[0041] First, weights are configured based on the correlation attribute deviation characteristic values ​​of each evaluation indicator. Specifically, the weights of the preset evaluation indicator detection values ​​are configured based on the correlation attribute deviation characteristic values ​​of the first evaluation indicator up to the Nth evaluation indicator. The configuration principle is: evaluation indicators with larger correlation attribute deviation characteristic values ​​indicate that they are more affected by control attribute deviations and are at a relative disadvantage in quality control; therefore, the corresponding preset evaluation indicator detection values ​​are assigned higher weights. Conversely, evaluation indicators with smaller correlation attribute deviation characteristic values ​​indicate that they are less affected by control attribute deviations and are at a relative advantage; therefore, the corresponding preset evaluation indicator detection values ​​are assigned lower weights. This weight configuration method reflects a focus on disadvantaged indicators to achieve dynamic balance optimization.

[0042] Then, the fitness of the preset evaluation index detection values ​​is calculated using the configured weights. Specifically, the detection values ​​of each preset evaluation index are compared with their expected values ​​(or standard values), the degree of deviation is calculated, multiplied by the corresponding weight, and finally all weighted deviations are combined to obtain the first fitness of the first healthy sample. This first fitness reflects the overall quality performance of the first healthy sample after considering the dynamic weights of each index.

[0043] By dynamically configuring weights based on deviation eigenvalues, the importance of each evaluation indicator in fitness calculation is automatically adjusted, so that indicators with poor performance receive more attention and indicators with good performance have relatively lower weights. This promotes balanced optimization of various quality indicators and avoids the problem that some indicators may be over-optimized while others are ignored, which may be caused by traditional fixed-weight methods.

[0044] S6. The first healthy sample and the first fitness are associated and stored in the optimized particle space. When the number of healthy samples in the optimized particle space meets the preset number, optimization is performed to obtain the desired processing parameters and initialize the roller processing mode.

[0045] Specifically, firstly, the first healthy sample and its first fitness are stored together as a single data pair in the optimization particle space. This associated storage ensures that each healthy sample has a corresponding fitness value, facilitating subsequent optimization calculations. As data collection progresses, more healthy samples and their fitness values ​​are stored in the optimization particle space. Then, the number of healthy samples in the optimization particle space is continuously monitored. When the number of healthy samples reaches a preset quantity, it indicates that sufficient data has been accumulated, at which point the optimization process is triggered. The setting of the preset quantity requires a balance between optimization accuracy and computational efficiency, ensuring that there are enough samples to support the optimization calculations without causing excessive computational burden due to an excessive number of samples.

[0046] During optimization, all healthy samples and their fitness in the optimization particle space are used to search for the optimal solution through an optimization algorithm (such as particle swarm optimization) to obtain the desired processing parameters. The desired processing parameters refer to a set of control parameter values ​​that maximize the fitness, including temperature, rotational speed, time, and pressure. Then, the obtained desired processing parameters are used to initialize the drum processing mode, that is, these parameters are set as the initial control parameters for the corresponding processing mode, providing an optimized process parameter benchmark for actual production.

[0047] By using a sample accumulation and optimization method, the transformation from historical data to optimized parameters was realized, providing a data-driven parameter optimization scheme for the roller processing mode. This allows for adaptive adjustment of control parameters based on different varieties, specifications, and processing modes of Chinese herbal medicines, improving the stability and consistency of Chinese herbal medicine processing. At the same time, the dynamic weighting mechanism avoids the problem of over-optimization of a single indicator, achieving a balanced improvement in various quality indicators and enhancing the overall processing quality and production efficiency of Chinese herbal medicines.

[0048] Furthermore, the preset evaluation indicators are traversed, and a correlation analysis is performed on the processing control attributes to obtain the correlation attributes and correlation degrees of the first evaluation indicator, up to the correlation attributes and correlation degrees of the Nth evaluation indicator, including:

[0049] S21. Extract the first evaluation index of the preset evaluation index, and extract the first processing control attribute of the processing control attribute;

[0050] S22. Based on the roller processing mode, the specifications of the Chinese herbal medicine slices and the varieties of the Chinese herbal medicine slices, and in combination with the roller model, a processing anomaly sample set is collected, wherein any processing anomaly sample in the processing anomaly sample set includes an abnormal processing control attribute set and an abnormal evaluation index set.

[0051] S23. Calculate the ratio of the trigger frequency of the first evaluation index and the first processing control attribute in the same processing abnormal sample to the total number of processing abnormal samples.

[0052] S24. When the ratio is greater than or equal to a preset ratio, the first processing control attribute is added to the first evaluation index association attribute. After the processing control attribute has been traversed, the association attribute of the first evaluation index is analyzed based on the first evaluation index to obtain the association degree of the first evaluation index.

[0053] S25. Update the evaluation index cyclical analysis until the correlation attribute of the Nth evaluation index and the correlation degree of the Nth evaluation index are obtained.

[0054] In one feasible implementation, an evaluation index is extracted from preset evaluation indicators as the current analysis object, denoted as the first evaluation index, such as moisture content; simultaneously, a control attribute is extracted from processing control attributes, such as temperature, as the first processing control attribute. Then, under four-dimensional conditions (roller processing mode, Chinese herbal medicine slice specifications, Chinese herbal medicine slice varieties, and roller model), sample data of historically occurring quality anomalies are collected to obtain a processing anomaly sample set. Each processing anomaly sample in this sample set records the abnormal processing control attribute set (i.e., which control parameters are abnormal) and the abnormal evaluation index set (i.e., which quality indicators fail to meet the standards) at the time of the anomaly.

[0055] Subsequently, all abnormal processing samples in the abnormal processing sample set are iterated through, and the number of times the first evaluation indicator and the first processing control attribute are simultaneously abnormal is counted. That is, in the same abnormal processing sample, the evaluation indicator fails to meet the standard and the control attribute is abnormal. This number of simultaneous triggers is divided by the total number of abnormal processing samples in the abnormal processing sample set to obtain the trigger frequency ratio. If this ratio reaches a preset ratio (e.g., 0.6), it indicates that there is a strong correlation between the first processing control attribute and the first evaluation indicator, and the first processing control attribute is added to the first evaluation indicator's associated attributes. The process of judging S21-S24 is repeated for other control attributes. After all processing control attributes have been iterated through, all control attributes related to the first evaluation indicator are obtained. Then, in-depth correlation analysis is performed on these associated attributes to calculate the correlation degree of the first evaluation indicator.

[0056] After completing the analysis of the first evaluation indicator, update to the second evaluation indicator and repeat the process of S21-S24. Continue in this manner until the correlation attributes and correlation degrees of all N preset evaluation indicators are identified, obtaining a complete result from the correlation attributes and correlation degrees of the first evaluation indicator to the correlation attributes and correlation degrees of the Nth evaluation indicator.

[0057] By using correlation analysis based on outlier sample statistics, we can accurately identify the intrinsic relationship between each evaluation indicator and control attribute, providing a precise dependency mapping for subsequent parameter optimization.

[0058] Furthermore, based on the first evaluation indicator, a correlation analysis is performed on the correlation attributes of the first evaluation indicator to obtain the correlation degree of the first evaluation indicator, which also includes:

[0059] S241. Based on the roller processing mode, the specifications of the Chinese herbal medicine slices and the varieties of the Chinese herbal medicine slices, and in conjunction with the roller model, collect several first evaluation index related attribute record values ​​and several first evaluation index record values.

[0060] S242. Calculate the same-attribute deviation of the recorded values ​​of the related attributes of the several first evaluation indicators to obtain the deviation modulus of the related attributes of the several first evaluation indicators.

[0061] S243. Perform deviation calculation on the recorded values ​​of several first evaluation indicators to obtain the deviation modulus of the recorded values ​​of several first evaluation indicators.

[0062] S244. Extract the first associated attribute deviation modulus sequence up to the Qth associated attribute deviation modulus sequence from the deviation modulus of the associated attributes of the plurality of first evaluation indicators, combine it with the deviation modulus of the recorded values ​​of the plurality of first evaluation indicators, construct a gray relational degree matrix, perform correlation analysis, obtain the correlation degree of the first associated attributes up to the Qth associated attributes, and add it to the correlation degree of the first evaluation indicators.

[0063] In a preferred embodiment, firstly, under the same four-dimensional conditions as described above (roller processing mode, Chinese herbal medicine specifications, Chinese herbal medicine varieties, and roller model), multiple sets of data are collected from the historical processing database to obtain several first evaluation index related attribute record values ​​and several first evaluation index record values. The first evaluation index related attribute record values ​​refer to the actual measured values ​​of each processing control attribute related to the first evaluation index at different times; the first evaluation index record values ​​refer to the actual measured values ​​of the first evaluation index at the corresponding time. These data constitute the original dataset for the correlation analysis.

[0064] Then, for the first evaluation indicator, deviation calculations are performed on the recorded values ​​of each processing-related attribute. Specifically, for a processing-related attribute (such as temperature), the deviation of each recorded value from the standard value or mean of that processing-related attribute is calculated, and the modulus value is taken to obtain a sequence of deviation modulus values ​​for that processing-related attribute. This process is repeated for all related attributes to obtain several deviation modulus values ​​for the first evaluation indicator's related attributes. Similarly, deviation calculations are performed on the sequence of recorded values ​​for the first evaluation indicator, calculating the deviation of each recorded value from the standard value or mean of that indicator, and the modulus value is taken to obtain several deviation modulus values ​​for the recorded values ​​of the first evaluation indicator.

[0065] Next, the deviation modulus values ​​of several first evaluation index related attributes are grouped by attribute to obtain the first related attribute deviation modulus value sequence to the Qth related attribute deviation modulus value sequence, where Q is the total number of related attributes among the several first evaluation index related attributes. Subsequently, using the deviation modulus values ​​of several first evaluation index record values ​​as reference sequences and the deviation modulus values ​​of the first related attribute deviation modulus values ​​to the Qth related attribute deviation modulus values ​​as comparison sequences, a grey relational degree matrix is ​​constructed. In this matrix, rows represent different time points, and columns represent the reference sequence and each comparison sequence. Then, based on this grey relational degree matrix, correlation analysis is performed to obtain the correlation degree of the first related attributes up to the Qth related attributes, which is added to the first evaluation index correlation degree. Specifically, the correlation degree is calculated for each related attribute deviation modulus value and the deviation modulus value of the first evaluation index record value: first, the absolute difference between the two sequences at each time point is calculated, and the minimum and maximum differences are found; then, the correlation coefficient at each time point is calculated; finally, the correlation coefficients at all times are averaged to obtain the correlation degree of that related attribute, thus obtaining the correlation degree of the first related attribute up to the Qth related attribute.

[0066] By using the grey relational analysis method based on deviation sequences, we can delve deeper into the dynamic relationship between the correlation attributes of the first evaluation index and the first evaluation index, providing more accurate correlation quantification results.

[0067] Furthermore, using the correlation degree of the first evaluation indicator up to the correlation degree of the Nth evaluation indicator as a normalized deviation weight, a weighted average is calculated on the correlation deviation parameters of the first evaluation indicator up to the correlation deviation parameters of the Nth evaluation indicator to obtain the correlation attribute deviation characteristic value of the first evaluation indicator up to the correlation attribute deviation characteristic value of the Nth evaluation indicator, including:

[0068] S41. Extract several related attribute deviation parameters from the first evaluation index related deviation parameters;

[0069] S42. Traverse the several correlation attribute deviation parameters and perform normalization processing to obtain several correlation attribute normalization deviations.

[0070] S43. Extract the correlation of several related attributes from the first evaluation index correlation degree;

[0071] S44. Based on the correlation degree of the several correlation attributes, calculate the weighted mean of the normalized deviation of the several correlation attributes to obtain the characteristic value of the correlation attribute deviation of the first evaluation index.

[0072] In a preferred embodiment, firstly, the first evaluation index-related deviation parameters include the deviation values ​​of all control attributes related to the first evaluation index. Several related attribute deviation parameters are extracted from these parameters, that is, the deviation parameter values ​​corresponding to each specific related attribute (such as temperature, rotational speed, time, etc.) are extracted. Then, the extracted several related attribute deviation parameters are normalized one by one. Normalization transforms deviation parameters with different dimensions and numerical ranges into a unified [0,1] interval, ensuring that each deviation parameter is comparable in subsequent calculations. After normalization, several normalized deviations of the related attributes are obtained.

[0073] The correlation degree of the first evaluation index includes the correlation strength value between each related attribute and the first evaluation index. Several correlation degrees of related attributes are extracted from this, and these correlation degrees correspond one-to-one with the correlation attribute deviation parameters extracted in step S41. Subsequently, the correlation degrees of these several related attributes are used as weighting coefficients to calculate the weighted average of the corresponding normalized deviations of the related attributes. Specifically, the normalized deviation of each related attribute is multiplied by its corresponding correlation degree, and then all products are summed and divided by the sum of the correlation degrees to obtain the characteristic value of the correlation attribute deviation of the first evaluation index. This characteristic value comprehensively reflects the weighted influence of all correlation attribute deviations on the first evaluation index.

[0074] Through the processing steps S41-S44 described above, the deviation features of the first evaluation indicator are extracted. The second to Nth evaluation indicators are processed using the same method to obtain the deviation feature values ​​of the associated attributes of the first evaluation indicator up to the Nth evaluation indicator, providing feature data for subsequent dynamic weight configuration.

[0075] Furthermore, the weights of the preset evaluation indicator detection values ​​are configured based on the deviation feature values ​​of the first evaluation indicator's associated attributes up to the deviation feature values ​​of the Nth evaluation indicator's associated attributes, including:

[0076] S51. Summing up the deviation feature values ​​of the first evaluation index's associated attribute until the deviation feature value of the Nth evaluation index's associated attribute is obtained, to obtain the sum of deviation feature values.

[0077] S52. Traverse the deviation feature values ​​of the first evaluation index associated attributes up to the deviation feature value of the Nth evaluation index associated attributes, and compare the sum of the deviation feature values ​​to obtain the weights of the first evaluation index up to the Nth evaluation index.

[0078] In a preferred embodiment, the deviation characteristic values ​​of the first evaluation index's associated attributes to the Nth evaluation index's associated attribute deviation characteristic values ​​respectively reflect the comprehensive influence of the control attribute deviations on each evaluation index. These N deviation characteristic values ​​are summed: first evaluation index associated attribute deviation characteristic value + second evaluation index associated attribute deviation characteristic value + ... + Nth evaluation index associated attribute deviation characteristic value, resulting in a sum of deviation characteristic values. This sum represents the total number of deviation characteristic values ​​for all evaluation indices.

[0079] The deviation feature values ​​of the associated attributes of each evaluation indicator are iterated and processed. Specifically, the deviation feature value of the associated attribute of the first evaluation indicator is divided by the sum of the deviation feature values ​​to obtain the weight of the first evaluation indicator; the deviation feature value of the associated attribute of the second evaluation indicator is divided by the sum of the deviation feature values ​​to obtain the weight of the second evaluation indicator; and so on, until the deviation feature value of the associated attribute of the Nth evaluation indicator is divided by the sum of the deviation feature values ​​to obtain the weight of the Nth evaluation indicator.

[0080] By calculating ratios, the weights are normalized, ensuring that the sum of the weights of all evaluation indicators is 1. Evaluation indicators with larger deviation eigenvalues ​​have higher weights, indicating that these indicators are more susceptible to deviations under current control conditions and require more attention in fitness calculations. Conversely, evaluation indicators with smaller deviation eigenvalues ​​have lower weights, indicating relative stability. This dynamic weight allocation method based on deviation eigenvalues ​​automatically prioritizes weaker indicators and appropriately weakens stronger ones, contributing to the balanced optimization of various quality indicators.

[0081] Further, calculating the first fitness of the first healthy sample includes:

[0082] S53. Obtain the expected value of the preset evaluation index, and calculate the deviation modulus between the preset evaluation index detection value and the preset evaluation index deviation modulus value.

[0083] S54. Based on the weights of the first evaluation index up to the weights of the Nth evaluation index, the preset evaluation index deviation modulus is weighted and summed to obtain the first fitness.

[0084] In a preferred embodiment, the expected value of the preset evaluation index is the ideal target value or standard value of each evaluation index, representing the quality target of the processing of Chinese herbal medicine slices; the detected value of the preset evaluation index is the actual measured value of each quality index contained in the first healthy sample. For each preset evaluation index, the difference between its detected value and the expected value is calculated, and the absolute value (i.e., the deviation modulus) is taken to obtain the preset evaluation index deviation modulus value. This preset evaluation index deviation modulus value reflects the gap between the actual processing quality and the target quality.

[0085] Then, using the obtained weights of the first to Nth evaluation indicators, the corresponding preset evaluation indicator deviation moduli are weighted and summed. Specifically, the first preset evaluation indicator deviation moduli is multiplied by the first evaluation indicator weight, the second preset evaluation indicator deviation moduli is multiplied by the second evaluation indicator weight, and so on, until the Nth preset evaluation indicator deviation moduli is multiplied by the Nth evaluation indicator weight. Then, all weighted deviation moduli are summed to obtain the first fitness.

[0086] A smaller first fitness value indicates that the overall quality of the first healthy sample is closer to the expected target; conversely, a larger first fitness value indicates that it deviates further from the target. By introducing dynamic weights, indicators with larger deviations occupy a higher proportion in the fitness calculation, prompting the optimization process to pay more attention to weak links and achieve a balanced improvement in various quality indicators.

[0087] Furthermore, the first healthy sample and the first fitness are associated and stored in an optimized particle space. When the number of healthy samples in the optimized particle space meets a preset quantity, optimization is performed to obtain the desired processing parameters, including:

[0088] S61. Perform particle swarm optimization based on the healthy sample set and the fitness set to obtain expanded samples;

[0089] S62. Retrieve the set of samples with the same modality as the expanded sample;

[0090] S63. When the proportion of abnormal samples in the same modality sample set is greater than or equal to the abnormal sample proportion threshold, the expanded sample is deleted.

[0091] S64. When the proportion of abnormal samples in the same modal sample set is less than the abnormal number proportion threshold, the average processing deviation parameter of the same modal sample set is set as the expanded sample processing deviation parameter, and the average preset evaluation index detection value of the same modal sample set is set as the expanded sample preset evaluation index detection value.

[0092] S65. Perform iterative optimization based on the expanded sample processing deviation parameters and the expanded sample preset evaluation index detection values.

[0093] In a preferred embodiment, the healthy sample set contains all stored healthy samples in the optimized particle space, and each healthy sample contains processing deviation parameters and preset evaluation index detection values; the fitness set contains fitness values ​​corresponding one-to-one with each healthy sample. In the particle swarm optimization algorithm, each healthy sample is regarded as a particle, and its position is defined by the processing deviation parameters, with the fitness value serving as the evaluation criterion for the quality of that position. During algorithm execution, each particle updates its velocity and position based on its own historical best position and the group's best position. Specifically, particles move in the parameter space by learning individual and group experiences, exploring new combinations of processing parameters. After a certain number of iterations, the algorithm generates new position points, i.e., expanded samples. The expanded samples contain new combinations of processing deviation parameters, representing potential optimization schemes explored by the algorithm during the optimization process. Then, the expanded samples are analyzed in depth, and historical samples with the same control state as the expanded samples are retrieved from the historical processing database. Specific retrieval conditions include: the same drum processing mode (cleaning, screening, or roasting), the same specifications of Chinese herbal medicine slices, the same variety of Chinese herbal medicine slices, and the same drum model. All historical samples that meet these conditions are grouped into a set of samples with the same modality. These samples share the same technological background as the expanded samples, and their historical performance can provide a reference for the reliability assessment of the expanded samples.

[0094] Subsequently, a quality analysis was performed on the same modal sample set, and the number of abnormal samples was counted. Abnormal samples refer to historical samples where the processing quality did not meet the standards or where quality problems occurred. The ratio of the number of abnormal samples to the total number of samples in the same modal sample set was calculated to obtain the abnormal sample proportion. If the abnormal sample proportion reaches or exceeds a preset abnormal sample proportion threshold (e.g., 0.7), it indicates that quality problems have frequently occurred historically under similar control conditions, suggesting that the parameter combination corresponding to the expanded sample has a high risk. Therefore, the expanded sample is removed from the candidate solutions to avoid outputting high-risk parameters as optimization results.

[0095] If the proportion of abnormal samples is lower than the threshold for the proportion of abnormal samples, it indicates that the control condition has been relatively stable and reliable in historical production. In this case, the expanded samples are empirically corrected using statistical information from the same modal sample set. Specifically, the correction method is as follows: calculate the arithmetic mean of the processing deviation parameters (such as temperature deviation, rotational speed deviation, etc.) of all samples in the same modal sample set, and use these mean values ​​as the new processing deviation parameters for the expanded samples; simultaneously, calculate the arithmetic mean of the preset evaluation index detection values ​​(such as moisture content, breakage rate, etc.) of all samples, and use these mean values ​​as the preset evaluation index detection values ​​for the expanded samples. This correction method based on historical data mean makes the expanded samples closer to actual production experience, improving their reliability and practicality. Then, using the corrected expanded sample parameters from step S64, namely the expanded sample processing deviation parameters and the expanded sample preset evaluation index detection values, the complete dynamic weight fitness calculation process is re-executed to obtain the fitness value of the expanded samples. The calculated expanded samples and their fitness values ​​are added to the optimization particle space to update the state of the particle swarm. Continue executing the next iteration of the particle swarm optimization algorithm, repeating processes S61 to S65 until the algorithm converges or reaches the maximum number of iterations. Finally, select the sample with the best fitness from the optimized particle space; its processing parameters are the desired processing parameters.

[0096] By combining historical sample verification, experience correction, and iterative optimization, not only is the exploratory and innovative nature of the optimization process guaranteed, but the reliability and practicality of the optimization results are also ensured, effectively improving the accuracy of the optimization of processing parameters for traditional Chinese medicine decoction pieces.

[0097] Example 2, as Figure 2 As shown, based on the same inventive concept as the multi-process drum optimization control method for processing Chinese herbal medicine slices provided in Embodiment 1, this embodiment of the invention also provides a multi-process drum optimization control system for processing Chinese herbal medicine slices, including:

[0098] The health sample collection module 11 is used to collect a first health sample based on the roller processing mode, the specifications and varieties of Chinese herbal medicine pieces, and the roller model. The first health sample includes a first processing deviation parameter and a preset evaluation index detection value.

[0099] The correlation analysis module 12 is used to traverse the preset evaluation indicators, perform correlation analysis on the processing control attributes, and obtain the correlation attributes and correlation degree of the first evaluation indicator, until the correlation attributes and correlation degree of the Nth evaluation indicator.

[0100] The deviation parameter decomposition module 13 is used to decompose the first processing deviation parameter according to the first evaluation index association attribute up to the Nth evaluation index association attribute, and obtain the first evaluation index associated deviation parameter up to the Nth evaluation index associated deviation parameter.

[0101] The deviation feature calculation module 14 is used to use the correlation degree of the first evaluation index up to the correlation degree of the Nth evaluation index as a normalized deviation weight, and to calculate the weighted mean of the correlation deviation parameter of the first evaluation index up to the correlation deviation parameter of the Nth evaluation index to obtain the correlation attribute deviation feature value of the first evaluation index up to the correlation attribute deviation feature value of the Nth evaluation index.

[0102] Fitness calculation module 15 is used to configure the weight of the preset evaluation index detection value based on the first evaluation index correlation attribute deviation feature value up to the Nth evaluation index correlation attribute deviation feature value, and calculate the first fitness of the first healthy sample.

[0103] The processing mode initialization module 16 is used to associate and store the first healthy sample and the first fitness in the optimization particle space. When the number of healthy samples in the optimization particle space meets the preset number, optimization is performed to obtain the desired processing parameters for initializing the roller processing mode.

[0104] Furthermore, the correlation analysis module 12 includes the following execution steps:

[0105] Extract the first evaluation index of the preset evaluation index, and extract the first processing control attribute of the processing control attribute;

[0106] Based on the roller processing mode, the specifications of the Chinese herbal medicine slices and the varieties of the Chinese herbal medicine slices, and in combination with the roller model, a processing anomaly sample set is collected. Each processing anomaly sample in the processing anomaly sample set includes an abnormal processing control attribute set and an abnormal evaluation index set.

[0107] The ratio of the trigger frequency of the first evaluation index and the first processing control attribute in the same processing abnormal sample to the total number of processing abnormal samples is calculated.

[0108] When the ratio is greater than or equal to the preset ratio, the first processing control attribute is added to the first evaluation index association attribute. After the processing control attribute has been traversed, the association attribute of the first evaluation index is analyzed based on the first evaluation index to obtain the association degree of the first evaluation index.

[0109] The evaluation indicators are updated and analyzed iteratively until the correlation attributes and correlation degree of the Nth evaluation indicator are obtained.

[0110] Furthermore, the correlation analysis module 12 also includes the following execution steps:

[0111] Based on the roller processing mode, the specifications of the Chinese herbal medicine slices and the varieties of the Chinese herbal medicine slices, and in combination with the roller model, collect the associated attribute records of several first evaluation indicators and the recorded values ​​of several first evaluation indicators.

[0112] The same attribute deviation is calculated for the recorded values ​​of the associated attributes of the several first evaluation indicators to obtain the magnitude value of the deviation of the associated attributes of the several first evaluation indicators.

[0113] Deviation calculation is performed on the recorded values ​​of several first evaluation indicators to obtain the deviation modulus of the recorded values ​​of several first evaluation indicators;

[0114] From the deviation modulus values ​​of the associated attributes of the several first evaluation indicators, extract the sequence of deviation modulus values ​​of the first associated attributes up to the Qth associated attribute deviation modulus value sequence, combine it with the deviation modulus values ​​of the recorded values ​​of the several first evaluation indicators, construct a gray relational degree matrix, perform correlation analysis, obtain the correlation degree of the first associated attributes up to the Qth associated attribute correlation degree, and add it to the correlation degree of the first evaluation indicators.

[0115] Furthermore, the deviation feature calculation module 14 includes the following execution steps:

[0116] Extract several related attribute deviation parameters from the first evaluation index related to the deviation parameters;

[0117] The normalization process is performed on the aforementioned correlation attribute deviation parameters to obtain the normalized deviation of the correlation attributes.

[0118] From the correlation degree of the first evaluation index, extract the correlation degree of several related attributes;

[0119] Based on the correlation degree of the aforementioned correlation attributes, a weighted average of the normalized deviations of the aforementioned correlation attributes is calculated to obtain the characteristic value of the correlation attribute deviation of the first evaluation index.

[0120] Furthermore, the fitness calculation module 15 includes the following execution steps:

[0121] Summing the deviation feature values ​​of the associated attributes of the first evaluation index up to the deviation feature values ​​of the associated attributes of the Nth evaluation index, we obtain the sum of the deviation feature values.

[0122] Iterate through the deviation feature values ​​of the associated attributes of the first evaluation index up to the deviation feature value of the associated attributes of the Nth evaluation index, and compare the sum of the deviation feature values ​​to obtain the weights of the first evaluation index up to the Nth evaluation index.

[0123] Furthermore, the fitness calculation module 15 also includes the following execution steps:

[0124] Obtain the expected value of the preset evaluation index, and calculate the deviation modulus between it and the detected value of the preset evaluation index to obtain the deviation modulus value of the preset evaluation index.

[0125] Based on the weights of the first evaluation index up to the weights of the Nth evaluation index, the deviation modulus of the preset evaluation index is weighted and summed to obtain the first fitness.

[0126] Furthermore, the processing mode initialization module 16 includes the following execution steps:

[0127] Particle swarm optimization is performed based on the healthy sample set and the fitness set to obtain expanded samples;

[0128] Retrieve the set of samples with the same modality as the augmented sample;

[0129] When the proportion of abnormal samples in the same modality sample set is greater than or equal to the abnormal sample proportion threshold, the expanded sample is deleted.

[0130] When the proportion of abnormal samples in the same modal sample set is less than the threshold of the proportion of abnormal samples, the average value of the processing deviation parameter of the same modal sample set is set as the processing deviation parameter of the expanded sample, and the average value of the preset evaluation index detection value of the same modal sample set is set as the preset evaluation index detection value of the expanded sample.

[0131] Based on the processing deviation parameters of the expanded sample and the detection values ​​of the preset evaluation index of the expanded sample, a cyclic optimization is performed.

[0132] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0133] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0134] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0135] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0136] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0137] Although preferred embodiments of the invention have been described, those skilled in the art, once they have learned the basic inventive concept, can make other changes and modifications to these embodiments.

[0138] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A multi-stage drum optimization control method for processing traditional Chinese medicine decoction pieces, characterized in that, An application to a processing drum for traditional Chinese medicine decoction pieces, the processing drum for traditional Chinese medicine decoction pieces includes a cleaning processing mode, a screening processing mode, and a stir-frying processing mode, including: Based on the roller processing mode, the specifications and varieties of Chinese herbal medicine pieces, and the roller model, a first health sample is collected. The first health sample includes a first processing deviation parameter and a preset evaluation index detection value. Iterate through the preset evaluation indicators, perform correlation analysis on the processing control attributes, and obtain the correlation attributes and correlation degrees of the first evaluation indicator, up to the correlation attributes and correlation degrees of the Nth evaluation indicator, including: Extract the first evaluation index of the preset evaluation index, and extract the first processing control attribute of the processing control attribute; Based on the roller processing mode, the specifications of the Chinese herbal medicine slices and the varieties of the Chinese herbal medicine slices, and in combination with the roller model, a processing anomaly sample set is collected. Each processing anomaly sample in the processing anomaly sample set includes an abnormal processing control attribute set and an abnormal evaluation index set. The ratio of the trigger frequency of the first evaluation index and the first processing control attribute in the same processing abnormal sample to the total number of processing abnormal samples is calculated. When the ratio is greater than or equal to the preset ratio, the first processing control attribute is added to the first evaluation index association attribute. After the processing control attribute has been traversed, the association attribute of the first evaluation index is analyzed based on the first evaluation index to obtain the association degree of the first evaluation index. The evaluation indicators are updated and analyzed iteratively until the correlation attributes and correlation degree of the Nth evaluation indicator are obtained. Based on the association attributes of the first evaluation index up to the Nth evaluation index, the first processing deviation parameter is decomposed to obtain the association deviation parameters of the first evaluation index up to the Nth evaluation index. Using the correlation degree of the first evaluation indicator up to the correlation degree of the Nth evaluation indicator as the normalized deviation weight, the weighted mean of the correlation deviation parameter of the first evaluation indicator up to the correlation deviation parameter of the Nth evaluation indicator is calculated to obtain the correlation attribute deviation characteristic value of the first evaluation indicator up to the correlation attribute deviation characteristic value of the Nth evaluation indicator. The weights of the preset evaluation index detection values ​​are configured based on the first evaluation index correlation attribute deviation feature value up to the Nth evaluation index correlation attribute deviation feature value, and the first fitness of the first healthy sample is calculated. The first healthy sample and the first fitness are associated and stored in the optimized particle space. When the number of healthy samples in the optimized particle space meets the preset number, optimization is performed to obtain the desired processing parameters for initializing the roller processing mode.

2. The method as described in claim 1, characterized in that, Based on the first evaluation index, a correlation analysis is performed on the correlation attributes of the first evaluation index to obtain the correlation degree of the first evaluation index, which further includes: Based on the roller processing mode, the specifications of the Chinese herbal medicine slices and the varieties of the Chinese herbal medicine slices, and in combination with the roller model, collect the associated attribute records of several first evaluation indicators and the recorded values ​​of several first evaluation indicators. The same attribute deviation is calculated for the recorded values ​​of the associated attributes of the several first evaluation indicators to obtain the magnitude value of the deviation of the associated attributes of the several first evaluation indicators. Deviation calculation is performed on the recorded values ​​of several first evaluation indicators to obtain the deviation modulus of the recorded values ​​of several first evaluation indicators; From the deviation modulus values ​​of the associated attributes of the several first evaluation indicators, extract the sequence of deviation modulus values ​​of the first associated attributes up to the Qth associated attribute deviation modulus value sequence, combine it with the deviation modulus values ​​of the recorded values ​​of the several first evaluation indicators, construct a gray relational degree matrix, perform correlation analysis, obtain the correlation degree of the first associated attributes up to the Qth associated attribute correlation degree, and add it to the correlation degree of the first evaluation indicators.

3. The method as described in claim 1, characterized in that, Using the correlation degree of the first evaluation indicator up to the correlation degree of the Nth evaluation indicator as normalized deviation weights, a weighted average is calculated on the correlation deviation parameters of the first evaluation indicator up to the correlation deviation parameters of the Nth evaluation indicator to obtain the correlation attribute deviation characteristic value of the first evaluation indicator up to the correlation attribute deviation characteristic value of the Nth evaluation indicator, including: Extract several related attribute deviation parameters from the first evaluation index related to the deviation parameters; The normalization process is performed on the aforementioned correlation attribute deviation parameters to obtain the normalized deviation of the correlation attributes. From the correlation degree of the first evaluation index, extract the correlation degree of several related attributes; Based on the correlation degree of the aforementioned correlation attributes, a weighted average of the normalized deviations of the aforementioned correlation attributes is calculated to obtain the characteristic value of the correlation attribute deviation of the first evaluation index.

4. The method as described in claim 1, characterized in that, The weights of the preset evaluation indicator detection values ​​are configured based on the correlation attribute deviation feature values ​​of the first evaluation indicator up to the correlation attribute deviation feature values ​​of the Nth evaluation indicator, including: Summing the deviation feature values ​​of the associated attributes of the first evaluation index up to the deviation feature values ​​of the associated attributes of the Nth evaluation index, we obtain the sum of the deviation feature values. Iterate through the deviation feature values ​​of the associated attributes of the first evaluation index up to the deviation feature value of the associated attributes of the Nth evaluation index, and compare the sum of the deviation feature values ​​to obtain the weights of the first evaluation index up to the Nth evaluation index.

5. The method as described in claim 4, characterized in that, Calculating the first fitness of the first healthy sample includes: Obtain the expected value of the preset evaluation index, and calculate the deviation modulus between it and the detected value of the preset evaluation index to obtain the deviation modulus value of the preset evaluation index. Based on the weights of the first evaluation index up to the weights of the Nth evaluation index, the deviation modulus of the preset evaluation index is weighted and summed to obtain the first fitness.

6. The method as described in claim 1, characterized in that, The first healthy sample and the first fitness are associated and stored in the optimized particle space. When the number of healthy samples in the optimized particle space meets a preset number, optimization is performed to obtain the desired processing parameters, including: Particle swarm optimization is performed based on the healthy sample set and the fitness set to obtain expanded samples; Retrieve the set of samples with the same modality as the augmented sample; When the proportion of abnormal samples in the same modality sample set is greater than or equal to the abnormal sample proportion threshold, the expanded sample is deleted. When the proportion of abnormal samples in the same modal sample set is less than the threshold of the proportion of abnormal samples, the average value of the processing deviation parameter of the same modal sample set is set as the processing deviation parameter of the expanded sample, and the average value of the preset evaluation index detection value of the same modal sample set is set as the preset evaluation index detection value of the expanded sample. Based on the processing deviation parameters of the expanded sample and the detection values ​​of the preset evaluation index of the expanded sample, a cyclic optimization is performed.

7. A multi-process drum optimization control system for processing traditional Chinese medicine decoction pieces, characterized in that, For implementing the method as described in any one of claims 1 to 6, the system is applied to a traditional Chinese medicine processing drum, the traditional Chinese medicine processing drum including a cleaning processing mode, a screening processing mode, and a stir-frying processing mode, the system comprising: The health sample collection module is used to collect a first health sample based on the roller processing mode, the specifications and varieties of Chinese herbal medicine pieces, and the roller model. The first health sample includes a first processing deviation parameter and a preset evaluation index detection value. The correlation analysis module is used to traverse the preset evaluation indicators, perform correlation analysis on the processing control attributes, and obtain the correlation attributes and correlation degree of the first evaluation indicator, until the correlation attributes and correlation degree of the Nth evaluation indicator. The deviation parameter decomposition module is used to decompose the first processing deviation parameter according to the association attribute of the first evaluation index up to the association attribute of the Nth evaluation index, and obtain the first evaluation index associated deviation parameter up to the Nth evaluation index associated deviation parameter. The deviation feature calculation module is used to take the correlation degree of the first evaluation index up to the correlation degree of the Nth evaluation index as the normalized deviation weight, and to calculate the weighted mean of the correlation deviation parameter of the first evaluation index up to the correlation deviation parameter of the Nth evaluation index to obtain the correlation attribute deviation feature value of the first evaluation index up to the correlation attribute deviation feature value of the Nth evaluation index. The fitness calculation module is used to configure the weights of the preset evaluation index detection values ​​based on the first evaluation index correlation attribute deviation feature value up to the Nth evaluation index correlation attribute deviation feature value, and to calculate the first fitness of the first healthy sample. The processing mode initialization module is used to associate and store the first healthy sample and the first fitness in the optimization particle space. When the number of healthy samples in the optimization particle space meets the preset number, optimization is performed to obtain the desired processing parameters for initializing the roller processing mode.

Images