Production quality control method of carbonized hair

Abstract

The invention provides a carbonized hair production quality control method which comprises the following steps: acquiring real-time thermal weight loss curve data in a carbonized hair production process, and extracting thermal weight loss abnormal characteristics to obtain an abnormal characteristic set; on the basis of the abnormal feature set, quantifying the relationship among the weight loss rate change, the feature temperature point offset, the carbon residue rate fluctuation and the weight loss platform time, and combining the association between the weight loss rate abnormal change and the medicinal component content to form a feature combination; inputting a feature combination, training a dynamic mapping relationship between abnormal features and carbonized hair product quality indexes, predicting the product quality indexes of the current production batch of carbonized hair, and generating a quality evaluation result; if the storage stability deviates from the standard due to the fact that the weightlessness platform time is abnormal, quality deviation characteristics are obtained by analyzing the relation between the weightlessness platform time and storage stability parameters.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a production quality control method for blood residue charcoal. Background Art

[0002] Xueyu charcoal, a key processed product of traditional Chinese medicine, plays an irreplaceable role in clinical applications such as hemostasis, blood stasis removal, and anti-inflammatory treatment. Its production quality directly impacts its efficacy and clinical safety. Thermogravimetric analysis (TGA), a key technology for monitoring the carbonization process during the production of Xueyu charcoal, effectively characterizes product quality by analyzing the characteristics of its thermogravimetric curve. However, its current application in Xueyu charcoal production still faces certain limitations. Existing quality assessment methods often rely on post-analysis, making it difficult to capture the dynamic relationship between abnormal changes in the TGA curve and product quality indicators in real time. Furthermore, there is a lack of systematic guidance for optimizing production process control parameters, resulting in a need for improved product quality consistency. A key challenge lies in accurately identifying abnormal features in the TGA curve, such as unusual fluctuations in the second-stage weight loss rate, shifts in characteristic temperature points, abnormal carbon residue rates, and changes in the duration of the weight loss plateau, and in establishing a mapping model between these features and product quality indicators. Furthermore, the dynamic adjustment mechanism for process parameters during production needs to be improved, making it difficult to quickly respond to abnormal changes. Furthermore, the impact of parameter adjustments on the clinical efficacy and shelf life of Xueyu charcoal remains unclear. Therefore, how to construct a real-time quality assessment model based on thermogravimetric analysis data, accurately map the relationship between the characteristic points of the thermogravimetric curve and product quality indicators, and dynamically optimize the production process parameters according to abnormal changes to ensure the efficacy stability and storage safety of Xueyu charcoal has become a key issue in this study. Summary of the Invention

[0003] The present invention provides a production quality control method for blood residue charcoal, which mainly comprises:

[0004] Obtain real-time thermogravimetric curve data during the production of blood charcoal, extract abnormal thermogravimetric features, and obtain an abnormal feature set;

[0005] Based on the abnormal feature set, the relationship between the weight loss rate change, characteristic temperature point offset, carbon residue rate fluctuation and weight loss platform time is quantified. Combined with the correlation between the abnormal change of weight loss rate and the content of active ingredients, a feature combination is formed.

[0006] Input feature combinations, train the dynamic mapping relationship between abnormal features and Xueyu charcoal product quality indicators, predict the product quality indicators of the current production batch of Xueyu charcoal, and generate quality assessment results;

[0007] If the abnormal weightlessness platform time causes the storage stability to deviate from the standard, the quality deviation characteristics are obtained by analyzing the relationship between the weightlessness platform time and the storage stability parameters;

[0008] If the duration of the plateau period exceeds the preset time threshold, the abnormal time interval of uneven heating of the material and the equipment temperature control deviation are determined by analyzing the thermal weight loss curve data and the equipment operating status within the abnormal time window, and the abnormal equipment status characteristic data is generated;

[0009] Based on the abnormal characteristic data of equipment status, the relationship between the weightlessness platform time anomaly and the material conveying rate and heating rate is analyzed to determine the key parameters that cause batch consistency deviations. The control parameter set is generated in combination with the temperature rise gradient constraint relationship.

[0010] If the real-time quality assessment results or quality deviation characteristics deviate from the preset quality standards, the degree of influence of the heating rate on the content of the medicinal ingredients and the storage stability is calculated, and the key parameters are adjusted proportionally based on the influence degree ranking results. Combined with the control parameter set, a control parameter set for the blood charcoal production process is generated;

[0011] Based on the control parameter set of the blood charcoal production process, a virtual model of the blood charcoal production equipment was established, and simulation results were generated. The simulation results were compared and corrected with the actual equipment operation data, and equipment control instructions were generated and sent to the field bus. The PID control algorithm was used to adjust the operating status of the blood charcoal production equipment in real time, and thermal weight loss curve data was generated.

[0012] By comparing the deviation between the adjusted thermal gravimetric curve data and the preset standard curve, the particle swarm optimization algorithm is used to minimize the deviation, update the weights of abnormal features in the mapping relationship, and combine the association between the weight loss platform time and the storage stability parameters to obtain the optimized dynamic mapping relationship. The product quality indicators are predicted and the control parameters are adjusted repeatedly until the product quality indicators stabilize at the target level, and the process optimization results of blood charcoal are obtained.

[0013] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0014] The present invention discloses a quality control method for the production process of blood charcoal. By acquiring thermal gravimetric curve data, extracting abnormal features and establishing a dynamic mapping relationship with product quality indicators, real-time quality assessment is achieved. When an abnormality occurs, the correlation between the weightlessness platform time and storage stability is analyzed, the uneven heating of the material and the temperature control deviation of the equipment are determined, and an abnormal feature vector of the equipment state is generated. Based on this, key production parameters are adjusted, a virtual model is established for simulation optimization, and the operating state of the equipment is adjusted in real time through the PID control algorithm. Finally, a particle swarm algorithm is used to optimize the mapping relationship, and it is iterated repeatedly until the product quality is stable. The present invention can effectively improve the quality control accuracy and production efficiency of the blood charcoal production process, and ensure the stability and consistency of product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1The present invention is a flow chart of a method for controlling the production quality of blood residue charcoal. DETAILED DESCRIPTION

[0016] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0017] like Figure 1 The production quality control method of blood residue charcoal in this embodiment may specifically include:

[0018] Step S101, obtaining real-time thermogravimetric curve data during the production process of blood residue charcoal, extracting abnormal thermogravimetric features, and obtaining an abnormal feature set.

[0019] Real-time gas flow data at a preset temperature point collected by a gas flow sensor is received, and the gas flow data is combined with mass change data obtained by a mass change detector to obtain thermal weight loss curve data of the residual carbon; temperature change rate data and mass change rate data are obtained by a thermal parameter detector based on the thermal weight loss curve data, and the mass change rate data are analyzed by a carbonization speed monitor to obtain an initial weight loss rate data set; a three-parameter thermal analyzer is used to obtain characteristic temperature point data for the initial weight loss rate data set, and the characteristic temperature point data are analyzed by a thermal weight loss plateau detection algorithm to obtain abnormal fluctuation data of the residual carbon rate; if the weight loss rate change value in the abnormal fluctuation data of the residual carbon rate exceeds a preset threshold range, a time series analysis method is used to perform a sliding time window calculation on the weight loss rate data to obtain an abnormal feature set of thermal weight loss.

[0020] Specifically, a gas flow sensor is used to obtain real-time gas flow data at preset temperature points in the blood char production line based on the thermal decomposition temperature and material feed rate. Heat loss data and mass change data are obtained using a mass change detector and a thermal temperature sensor to generate real-time thermal weight loss curve data for the blood char. Based on the thermal weight loss curve data, a thermal parameter detector is used to obtain sample temperature change rate and mass change rate data. A carbonization rate monitor is used to monitor the sample carbonization process in real time, and the residual carbon rate is calculated to obtain an initial weight loss rate change dataset. Based on the initial weight loss rate change dataset, a three-parameter thermal analyzer is used to obtain thermal gas flow rate data, temperature change rate data, and mass change rate data. Characteristic temperature point data are obtained through temperature and mass curve fitting. A thermal weight loss plateau detection algorithm is used to calculate the weight loss plateau time data for the characteristic temperature point data. A residual carbon rate fluctuation detector is used to obtain abnormal residual carbon rate fluctuation data to generate a thermal weight loss anomaly feature dataset. If the weight loss rate change value in the thermal weight loss anomaly feature dataset exceeds a preset threshold range, a time series analysis method is used to perform a sliding time window calculation on the weight loss rate data to obtain abnormal weight loss rate time window data and fluctuation amplitude data. According to the abnormal time window data and fluctuation amplitude data of the weight loss rate, the reaction temperature is dynamically adjusted through the thermal temperature controller, and the gas flow control valve is used to implement real-time adjustment of the air intake volume to obtain the corrected thermal weight loss curve data. In the production process of blood residue carbon, the thermal weight loss curve records the mass change of the material during the heating process. The gas flow rate change during the reaction process is monitored by a gas flow sensor. When the gas flow rate is between 300 and 500 milliliters per minute, the gas flow data generated by the material at different temperature points is recorded. At the same time, the thermal temperature sensor collects the reaction temperature change data in real time. In response to the acquisition of the thermal weight loss curve data, the thermal parameter detector can accurately obtain the temperature change rate data of the sample in the range of 350 to 850 degrees Celsius. The carbonization rate monitor records the change in carbonization percentage per minute during the carbonization process of the sample. When the temperature reaches 600 degrees Celsius, the carbonization rate reaches its maximum value. At this time, the residual carbon rate fluctuates significantly, and the mass change rate shows an obvious fluctuation trend. In order to deeply analyze the abnormal characteristics of thermal weight loss, a three-parameter thermal analyzer simultaneously collects data on thermal airflow rate, temperature change rate, and mass change rate. When the temperature rises from room temperature to 850 degrees Celsius, a set of data is recorded every 50 degrees Celsius, and the temperature-

[0021] Quality curve, extract characteristic temperature points from the curve, including the starting temperature point of weight loss and the temperature point of maximum weight loss rate. The thermal weight loss plateau detection algorithm focuses on the mass change of the sample in the range of 400 to 600 degrees Celsius. When the mass change rate is less than 0.5% per minute, it is considered to have entered the weight loss plateau period, and the duration of the plateau period is recorded. At the same time, the residual carbon rate fluctuation detector monitors the amplitude of the change in the residual carbon rate. When the fluctuation exceeds 3%, an abnormal feature warning is triggered. When a sudden change in the weight loss rate is detected that exceeds the preset threshold of 2% per minute, a 120-second sliding time window is used to analyze the weight loss rate data, calculate the mean and standard deviation of the weight loss rate in the window, determine the time period of the abnormal occurrence, and record the peak value and duration of the fluctuation. According to the abnormal time window data, the thermal temperature controller dynamically adjusts the reaction temperature to the optimal range, and the gas flow control valve synchronously adjusts the air intake to restore the thermal weight loss curve to a normal change trend. In actual production, when the material feed rate is 500 grams per batch, by adjusting the gas flow rate to approximately 400 milliliters per minute, the thermal weight loss curve can be smoothly changed, the carbon residue rate is stabilized between 28% and 32%, the characteristic temperature point offset is controlled within a range of plus or minus 10 degrees Celsius, and the weight loss plateau duration is maintained between 15 and 20 minutes, ensuring the stable quality of the blood charcoal product. Step S102 quantifies the relationship between the change in weight loss rate, the characteristic temperature point offset, the fluctuation in carbon residue rate, and the weight loss plateau duration based on the abnormal feature set. Combined with the correlation between the abnormal change in weight loss rate and the content of the active ingredient, a feature combination is formed.

[0022] The method involves obtaining a weight loss rate value and a temperature point offset, the weight loss rate value being collected by a thermogravimetric curve analyzer during the reaction process; obtaining a carbon residue rate value using a carbon residue detector based on the weight loss rate value and the temperature point offset, and forming a thermogravimetric feature dataset with the carbon residue rate value and the active ingredient content data analyzed by an active ingredient detector; performing normalization calculations on the weight loss rate value and the carbon residue rate value using a data normalization processor for the thermogravimetric feature dataset to obtain a standardized feature dataset; and processing the standardized feature dataset using a random forest algorithm to obtain a feature combination for input into a machine learning model. Specifically, characteristic temperature data is obtained using a thermogravimetric curve analyzer based on the weight loss rate value and the temperature point offset, a carbon residue content detector is used to monitor the mass change rate and reaction temperature value in real time, and an airflow rate sensor is used to record airflow rate data during the reaction process to obtain a thermogravimetric process correlation dataset. For the thermogravimetric process correlation dataset, a weight loss plateau duration data is calculated using a weight loss plateau detector, a carbon residue rate value is obtained using a carbon residue detector, and active ingredient content data is analyzed using an active ingredient detector to obtain a thermogravimetric feature dataset. Based on the thermogravimetric characteristic dataset, a data normalization processor was used to normalize all parameters. Principal component analysis was then used to extract characteristic parameters, resulting in a standardized characteristic dataset. A correlation quantizer was used to calculate the correlation between the weight loss rate and the plateau duration, and a temperature offset quantizer was used to measure the offset amplitude of characteristic temperature points, resulting in a parameter correlation dataset. Based on the parameter correlation dataset, a pharmacological component correlater was used to calculate the correspondence between abnormal changes in weight loss rate and changes in pharmacological component content. A carbon residue rate fluctuation detector was used to obtain carbon residue rate fluctuation amplitude data, resulting in a feature association dataset. A parameter prediction model was constructed using a random forest algorithm for the feature association dataset, and cross-validation was used to verify the model's accuracy, resulting in machine learning feature combination data. A thermogravimetric curve analyzer monitored the weight loss rate during the blood charcoal production process. The weight loss rate values ​​were recorded at each temperature point as the temperature increased from room temperature to 850°C. It was found that the weight loss rate peaked at 5.8% per minute between 400°C and 600°C, and the characteristic temperature point shifted from the standard value of 550°C to 580°C. The mass change rate was monitored in real time by a residual carbon content detector, and data was recorded every 30 seconds during the pyrolysis process. When the reaction temperature reached 600 degrees Celsius, the mass change rate was the largest, reaching 0.8% per second. At this time, the air flow rate sensor showed that the reaction gas flow rate was 450 ml per minute, and it began to decrease after 120 seconds at this temperature point.The weight loss plateau detector recorded the duration of the plateau phase within the sample's temperature range of 450 to 550 degrees Celsius. The plateau phase began when the mass change rate dropped below 0.2% per minute, and under normal operating conditions, the plateau phase lasted 18 minutes. The carbon residue detector indicated a final carbon residue rate of 30%, and the active ingredient detector measured an active ingredient content of 85%. The collected data were normalized, with the weight loss rate, temperature, and carbon residue values ​​mapped to a range of 0 to 1. Principal component analysis (PCA) extracted eigenvectors, revealing a significant negative correlation between weight loss rate and plateau duration, with a correlation coefficient of -0.82. Each 10-degree Celsius shift in the characteristic temperature point resulted in a 2% fluctuation in carbon residue rate. Further analysis of the active ingredient fluctuation patterns revealed that an abnormal increase in the weight loss rate by 1% per minute was associated with a 3% decrease in active ingredient content. The carbon residue fluctuation test revealed that fluctuations in the carbon residue rate between 28% and 32% corresponded to fluctuations in active ingredient content between 82% and 88%. The random forest algorithm established 500 decision trees, using abnormal changes in weight loss rate, temperature point offsets, residual carbon rate fluctuations, and plateau duration as input features, and the active ingredient content as the prediction target. The model was validated using cross-validation, achieving a prediction accuracy of 92%. In practice, when a sudden change in weight loss rate exceeding 1% per minute was detected, the model accurately predicted the changing trend of the active ingredient content, providing a basis for adjusting process parameters. In production practice, by monitoring changes in weight loss rate and characteristic temperature point offsets, process parameters were immediately adjusted when abnormal fluctuations were detected, maintaining the residual carbon rate at around 30% and the plateau duration between 15 and 20 minutes, ensuring that the active ingredient content remained above 80%.

[0023] Step S103, input feature combinations, train the dynamic mapping relationship between abnormal features and blood charcoal product quality indicators, predict the product quality indicators of the current production batch of blood charcoal, and generate quality assessment results. According to the feature combination data, the blood charcoal quality indicator database is used to obtain the activity value data of historical batch samples, and the activity value data of the historical batch samples includes storage stability benchmark parameters; a parameter normalization processor is used to perform standardized calculations on the activity value data of the historical batch samples, and abnormal fluctuation values ​​are eliminated through a data filter to obtain a standardized quality parameter set; a long short-term memory network is used to construct a time series prediction model based on the standardized quality parameter set, and the trend of changes in the content of the active ingredient is modeled through a sliding time window method to obtain a quality indicator mapping model; a pharmacological content analyzer is used to measure the real-time active ingredient content data for the quality indicator mapping model, and the prediction accuracy is verified through a cross-validator to obtain quality assessment data.

[0024] Specifically, based on the feature combination data, the activity values ​​and storage stability benchmark data of historical batch samples are obtained from the blood charcoal quality index database. A time series analyzer is used to perform time series calculations on the active ingredient content ratio. A quality parameter fluctuation detector is used to record all indicator change data to obtain a historical quality parameter set. A parameter normalization processor is used to normalize the value range of the historical quality parameter set. A sample amplifier is used to randomly sample and supplement boundary condition data. A data filter is used to remove abnormal fluctuations to obtain a standardized quality parameter set. Based on the standardized quality parameter set, a time series prediction model is constructed using a long short-term memory network. A sliding time window method is used to model the trend of active ingredient content changes. A storage stability calculator is used to obtain product stability parameters to obtain a quality index mapping model. A dynamic calibrator is used to adjust the prediction weights of the quality index mapping model in real time, and a residual compensator is used to correct prediction deviations to obtain a calibrated prediction model. Based on the calibrated prediction model, a quality predictor is used to calculate the quality index values ​​of the current batch of products. A content analyzer is used to measure real-time active ingredient content data. A stability evaluator is used to obtain storage stability prediction values ​​to obtain real-time quality assessment results. The prediction accuracy of the real-time quality assessment results was verified using a cross-validator, and the prediction results were calibrated using an indicator corrector to obtain the final quality assessment data. The Xueyu Carbon Quality Index Database records the sample activity values ​​and storage stability data of 500 batches of products in the past five years. The sample activity values ​​include key parameters such as the content of active ingredients, residual carbon rate, and characteristic temperature points. The storage stability benchmark data includes regular test results within 12 months, which show that the product stability is good when the attenuation rate of the active ingredient content is within 0.5% per month. The time series analyzer calculates the active ingredient content ratio, using a time window of 30 days to record the content change trend of each batch of products for 12 consecutive months after production. It was found that the active ingredient content of normal products started from an initial value of 85% and dropped to more than 80% after 12 months, while the content decay rate of abnormal products exceeded 1% per month. The parameter normalization processor maps all quality indicators to a range of 0 to 1. The sample amplifier randomly samples boundary conditions to generate new data points to fill in blank areas. When the active ingredient content of a batch of products is detected to be less than 75%, interpolation is used to supplement intermediate data points, enhancing the model's ability to identify anomalies. The long short-term memory network adopts a three-layer structure. The input layer receives normalized quality parameters, the hidden layer contains 128 neurons, and the output layer predicts the trend of quality indicators over the next three months. When the predicted results show a monthly decline in active ingredient content exceeding 0.8%, an early warning signal is triggered. The dynamic calibrator adjusts the prediction weights based on the actual test results. When the prediction deviation exceeds 3%, the residual compensation mechanism is activated. The prediction accuracy is improved by introducing temperature and humidity influencing factors, keeping the error between the predicted results and the actual test value within 2%.The quality predictor calculates the product indicators of the current batch in real time, including the predicted value of the active ingredient content and the predicted value of storage stability. When the active ingredient content analyzer detects that the real-time content is 82% and predicts that the content will drop to 78% after 6 months, a quality warning is issued in time. The stability evaluator predicts based on historical data that under normal storage conditions, the annual attenuation rate of the active ingredient content should be controlled within 4%. The cross-validator uses a 5-fold cross-validation method to evaluate the prediction accuracy. By comparing the difference between the predicted value and the actual detection value, it is calculated that the prediction accuracy is over 90%. The indicator corrector corrects the abnormal prediction value according to the verification result to ensure the reliability of the final evaluation result. In actual application, this prediction method accurately identifies 95% of batches with abnormal quality, providing strong support for quality control.

[0025] Step S104: If the storage stability deviates from the standard due to the abnormal weightlessness platform time, a quality deviation feature is obtained by analyzing the relationship between the weightlessness platform time and the storage stability parameter.

[0026] A real-time temperature curve and a mass change curve are obtained based on the weightlessness platform time data, and the real-time temperature curve and the mass change curve are analyzed and processed by a curve fitter to obtain a numerical value of the plateau duration; the efficacy content decay rate and the moisture content change rate are measured by a stability index analyzer for the plateau duration value, and the efficacy content decay rate is compared with the preset standard limit by a standard limit comparator to obtain stability deviation data; if the efficacy content decay rate in the stability deviation data exceeds the preset standard limit, the weightlessness platform time and the storage stability parameter are normalized, and a mapping relationship between the weightlessness platform time and the storage stability parameter is established by a support vector regression method to obtain parameter association data; a mass deviation feature vector is obtained by a feature extractor for the parameter association data, and the mass deviation feature vector is combined by a feature combiner to obtain a mass deviation feature data set, including the plateau duration.

[0027] Specifically, according to the weightlessness platform time data, the real-time temperature curve and mass change curve are obtained through the platform monitor, the platform duration value is calculated using a curve fitter, and the heat loss rate and mass loss rate during the platform are recorded using a data collector to obtain the weightlessness platform process data. The efficacy content decay rate and the moisture content change rate are measured by a stability index analyzer for the weightlessness platform process data, and the measured value is compared with the preset standard limit using a standard limit comparator to obtain stability deviation data. If the efficacy content decay rate in the stability deviation data exceeds the preset standard limit, the weightlessness platform time and storage stability parameter are normalized using a data normalization processor, and the data noise is eliminated using an outlier identifier to obtain a standardized parameter set. Based on the standardized parameter set, the support vector regression method is used to establish a mapping relationship between the weightlessness platform time and the storage stability parameter, and the sliding window method is used to calculate the parameter correlation coefficient to obtain parameter association data. The mass deviation feature vector is obtained by a feature extractor for the parameter association data, and the deviation feature is classified using a decision tree method to obtain classification result data. Based on the classification result data, a quality deviation feature set is constructed through a feature combiner, and a weight coefficient is assigned to each feature parameter using a weight allocator to obtain a weighted feature data set. The platform period monitor records the temperature and mass change data in real time during the production process of blood charcoal. Through monitoring, it is found that under normal working conditions, the temperature curve shows an obvious platform feature in the range of 450 to 550 degrees Celsius, and the duration is maintained at 15 to 20 minutes. The mass loss rate is kept below 0.2% per minute, and the heat loss rate is stable at about 5% per minute. In the stability index analysis, the efficacy content decay rate and the moisture content change rate are two key parameters. The standard limit comparator shows that the efficacy content decay rate of normal products should be controlled within 0.5% per month, and the moisture content change rate should not exceed 0.3% per month. When the efficacy content decay rate of a batch of products is detected to reach 0.8% per month, it is judged to exceed the preset standard limit. Data normalization normalized the weightlessness plateau time and storage stability parameters, mapping the plateau duration to a range of 0 to 1, with 15 minutes corresponding to 0.5 and 20 minutes corresponding to 0.8. An outlier identifier was set at a threshold of ±3 standard deviations, revealing that 5% of the batch had plateaus less than 10 minutes or greater than 25 minutes. Support vector regression was used to establish the parameter mapping, and correlation coefficients were calculated using a 180-day sliding window. Results showed a significant negative correlation between weightlessness plateau time and the rate of potency decay, with a correlation coefficient of -0.85. For every 5-minute decrease in plateau duration, the rate of potency decay increased by 0.2% per month. A feature extractor extracted feature vectors from the mass deviation data, encompassing three dimensions: plateau duration deviation, temperature curve slope, and mass loss rate fluctuation. A decision tree approach categorized the deviation features into three categories: mild, moderate, and severe. A decrease in plateau duration exceeding 8 minutes was classified as severe.The feature combiner combines deviation features of different categories, and the weight allocator sets weight coefficients based on the degree of influence of each parameter on product quality. The weight of the plateau duration deviation value is 0.5, the weight of the temperature curve slope is 0.3, and the weight of the mass loss rate fluctuation amplitude is 0.2. The weighted calculation results in a comprehensive quality score. Practical applications have shown that when the abnormal weight loss platform time causes the storage stability to deviate from the standard, it is often accompanied by coordinated changes in multiple quality indicators. For example, in batches where the plateau period is shortened to 12 minutes, not only does the drug efficacy content decay faster, but the moisture content change rate also increases to 0.4% per month. The comprehensive score is less than 80 points, indicating that there is a potential risk in product storage stability. Through comprehensive analysis of these relevant parameters, more than 90% of batches with abnormal quality were successfully identified.

[0028] Step S105: If the duration of the plateau period exceeds the preset time threshold, the abnormal time interval of uneven heating of the material and the equipment temperature control deviation are determined by analyzing the thermogravimetric curve data and the equipment operating status within the abnormal time window, and the equipment status abnormal characteristic data is generated.

[0029] Based on the duration of the plateau period, a thermogravimetric curve recorder is used to obtain temperature change and mass loss data, and the thermogravimetric curve recorder records the surface temperature distribution data of the material within the abnormal time period; based on the temperature distribution data, a heat conduction meter is used to obtain heat transfer rate data, and the material heating unevenness value is calculated by the heat conduction meter; based on the heat transfer rate data, a temperature control state prediction model is constructed through a deep neural network, and the temperature deviation value is quantified and calculated through the deep neural network. The support vector machine classifies the equipment state according to the temperature deviation value, extracts the equipment temperature control characteristics, and obtains the equipment state abnormality feature data.

[0030] Specifically, based on the weightlessness plateau duration data, a time series monitor is used to obtain real-time plateau duration data. A threshold determiner is used to compare the duration with a preset threshold. A thermal weight loss curve recorder is used to collect temperature change and mass loss data to obtain plateau abnormality characteristic data. Based on the plateau abnormality characteristic data, a time window extractor is used to determine the abnormal period range. An infrared temperature sensor is used to obtain material surface temperature distribution data, and a thermal imager is used to record material heating uniformity data to obtain material heating distribution data. Based on the material heating distribution data, a temperature difference calculator is used to obtain the difference between the highest and lowest temperature points. A heat conduction meter is used to record heat transfer rate data, and a temperature uniformity evaluator is used to calculate the material heating non-uniformity value to obtain heat distribution characteristic data. Based on the heat distribution characteristic data, heater temperature control parameters are obtained from the equipment thermostat. An operating status monitor is used to record equipment operating data. A sensor network is used to collect multi-point temperature deviation data to obtain equipment operating characteristic data. Based on the equipment operating characteristic data, a temperature control state prediction model is constructed using a deep neural network. A temperature deviation quantizer is used to calculate the control error value to obtain temperature control abnormality data. For temperature control abnormal data, the equipment status is classified by support vector machine, and the temperature control characteristics of the equipment are extracted using feature vector generator to obtain abnormal equipment status feature data. The time series monitor monitors the duration of the plateau period during the production process of blood charcoal. Under normal working conditions, the plateau period is maintained in the range of 15 to 20 minutes. When the duration is less than 10 minutes or more than 25 minutes, the threshold determiner triggers an abnormal alarm. At the same time, the thermal gravimetric curve recorder shows that the temperature curve has obvious fluctuations in the range of 450 to 550 degrees Celsius. The time window extractor locks the abnormal period to the range of 5 to 15 minutes after the start of the plateau period. At this time, the infrared temperature sensor is arranged at 16 detection points on the surface of the material. The detection results show that the highest temperature point reaches 580 degrees Celsius and the lowest temperature point is only 420 degrees Celsius. The thermal imager scanning results show a clear temperature gradient distribution. A temperature difference calculator processed the test data and found that the temperature difference between the highest and lowest points on the material surface exceeded 150 degrees Celsius. The heat conductivity meter recorded that the heat transfer rate in the edge area was only 65% ​​of that in the center. The temperature uniformity assessment showed a nonuniformity coefficient of 0.35, far exceeding the standard limit of 0.15. Data from the equipment thermostat indicated that the heater temperature setpoint was 500 degrees Celsius, but the actual controlled temperature fluctuated between 480 and 520 degrees Celsius. The operating status monitor recorded unstable heater power output, and temperature data collected by the sensor network showed that the temperature deviation around the heating zone exceeded 30 degrees Celsius. A deep neural network uses a three-layer structure to process temperature data. The hidden layer contains 128 neurons. The input layer receives temperature data from 16 test points and heater operating parameters. The output layer predicts the temperature distribution trend over the next 30 minutes and issues a warning signal if the prediction results indicate that the temperature nonuniformity is deteriorating.The support vector machine classifies the equipment's operating status into three categories: normal, sub-healthy, and abnormal. The feature vectors encompass three dimensions: temperature control accuracy, heater power fluctuation, and temperature distribution uniformity. When heater power fluctuations exceeding 10% and a temperature control deviation exceeding 20 degrees Celsius are detected, the equipment is considered abnormal. Practical applications have shown that the simultaneous occurrence of abnormal plateau duration and uneven material heating often indicates an abnormal equipment condition. For example, in one production batch, the plateau was shortened to 8 minutes, while the material surface temperature difference reached 160 degrees Celsius. Further investigation revealed a 30% decrease in local heater power output, resulting in severely uneven heat distribution. By analyzing this multi-parameter coordinated variation, 85% of equipment abnormalities were successfully predicted and detected.

[0031] Step S106: Analyze the relationship between the weightlessness platform time anomaly and the material conveying rate and heating rate based on the abnormal equipment status characteristic data, determine the key parameters that cause batch consistency deviation, and generate a control parameter set in combination with the temperature rise gradient constraint relationship.

[0032] The material conveying rate and heating rate data are normalized by a data normalization processor, and the normalization processing obtains a standardized data set; a time series correlator is used to calculate the correlation coefficient between the weightlessness platform time and the standardized data set for the standardized data set, and the correlation coefficient is used to obtain a key parameter data set; a parameter change prediction model is constructed based on the key parameter data set using a random forest algorithm, and the prediction model outputs parameter influence data; a temperature gradient monitor is used to record the temperature rise data of the material heating process for the parameter influence data, and a temperature rise gradient constraint relationship is constructed between the temperature rise data and the residence time data obtained by the residence time detector; the power change data of the heating equipment is obtained based on the temperature rise gradient constraint relationship, and the power change data is processed by a curve fitter and integrated with the temperature rise gradient constraint relationship to generate a control parameter set.

[0033] Specifically, a data normalization processor normalizes material conveying rate and heating rate data based on abnormal equipment status data. A time series correlator calculates the correlation coefficient between the weightlessness platform time and the processed data. A parameter filter extracts correlation significance data to obtain a key parameter dataset. A random forest algorithm is used to construct a parameter change prediction model for this key parameter dataset. The prediction results are verified using a cross-validator, and the weight coefficients of each parameter are obtained using a feature importance calculator to obtain parameter influence data. Based on this parameter influence data, a temperature gradient monitor records the temperature rise of the material during heating. A residence time detector collects the material's residence time in the heating area. A constraint generator constructs a constraint relationship between the temperature rise gradient and residence time to obtain dynamic constraint data. For this dynamic constraint data, an energy consumption collector collects heating equipment power change data. An energy consumption curve fitter performs polynomial fitting on the collected data. A curve feature extractor extracts energy consumption trend characteristics to obtain energy consumption curve data. Based on the energy consumption curve data, a constraint fusion module integrates the dynamic constraint data into the energy consumption curve. The fused data is processed using a parameter optimizer to obtain an optimized parameter set. The batch consistency evaluator calculated the variance of product quality parameters for the optimized parameter set, and the control parameter generator extracted batch consistency control parameters to generate a control parameter dataset. The data normalizer mapped the material conveying rate from 150 to 350 grams per minute to the interval 0 to 1, and the heating rate from 5 to 15 degrees Celsius per minute to the same interval. Time series correlator analysis showed that the weightlessness plateau time was significantly negatively correlated with the material conveying rate, with a correlation coefficient of -0.82, and positively correlated with the heating rate, with a correlation coefficient of 0.75. A random forest algorithm constructed 500 decision trees to predict parameter changes. Cross-validation results showed that the model achieved an accuracy of 88%. Feature importance calculations indicated that the weight of the material conveying rate was 0.45, the weight of the heating rate was 0.35, and the weight of the equipment status feature was 0.20, indicating that the material conveying rate had the greatest impact on the weightlessness plateau time. The temperature gradient monitor records the temperature change trend of the material within the heating zone. When the material conveying rate is 250 grams per minute, the temperature gradient is maintained at 8 degrees Celsius per minute. The residence time detector shows that the material stays in the heating zone for 25 minutes. The constraint generator establishes a constraint relationship based on this data: the product of the temperature gradient and the residence time should be maintained at approximately 200 degrees Celsius. The energy consumption collector records the power data of the heating equipment every 30 seconds. Under standard operating conditions, the power curve exhibits a typical three-stage change: the power is maintained at 90% during the heating stage, the power is reduced to 75% during the plateau period, and the power is reduced to 30% during the cooling stage. The energy consumption curve fitter uses a cubic polynomial to fit the power changes, and the curve feature extractor obtains the rate of change of energy consumption in each stage.The constraint fusion tool integrates the constraint relationship between temperature rise gradient and residence time into the energy consumption curve. It found that when the material conveying rate is too fast, resulting in insufficient residence time, the equipment power will increase abnormally. For example, when the conveying rate is increased to 300 grams per minute, the power rises to 85% during the plateau period, while the temperature rise gradient only increases to 9 degrees Celsius per minute, indicating a decrease in heating efficiency. Batch consistency assessment calculations show that under standard operating conditions, the variance of product quality parameters is controlled within 3%. When the material conveying rate fluctuates by more than 30 grams per minute, the variance of product quality parameters increases to more than 5%. Based on this, the control parameter generator sets the material conveying rate fluctuation threshold to ±20 grams per minute and the heating rate fluctuation threshold to ±1 degree Celsius per minute. In actual production, by adjusting the material delivery rate in the range of 230 to 270 grams per minute and maintaining the heating rate in the range of 7 to 9 degrees Celsius per minute, the product of the temperature rise gradient and the residence time is stabilized in the range of 190 to 210 degrees Celsius. The product batch consistency control parameters show that the quality fluctuation is reduced to less than 2%, the equipment energy consumption curve is stable, and the power utilization rate is increased to 85%.

[0034] Step S107: If the real-time quality assessment results or quality deviation characteristics deviate from the preset quality standards, the degree of influence of the heating rate on the content of the medicinal ingredients and the storage stability is calculated, and the key parameters are adjusted proportionally based on the influence degree ranking results. Combined with the control parameter set, a control parameter set for the blood charcoal production process is generated.

[0035] The deviation value between the quality index and the preset quality standard is obtained through a standard comparator, and the process parameter change curve is recorded by a heating rate monitor to obtain quality abnormality characteristic data; according to the quality abnormality characteristic data, the abnormal characteristics are weighted by a weight calculator, and the correlation between the heating rate and the content of the medicinal ingredient is calculated by a correlation analyzer to obtain characteristic weight data; for the characteristic weight data, a heating rate influence model for the quality index is constructed by an influence calculator, and the degree of influence is quantitatively calculated by the hierarchical analysis method to obtain a parameter influence matrix; according to the parameter influence matrix, the process parameters are corrected according to the degree of influence by a parameter adjuster, and the out-of-range parameters are corrected by a parameter calibrator to obtain verification parameter data; for the verification parameter data, the batch consistency control requirements are integrated into the control parameter set by a consistency matcher to generate a control parameter set for the blood charcoal production process.

[0036] Specifically, based on the real-time quality assessment results, a standard comparator is used to determine the deviation between quality indicators and preset quality standards. A heating rate monitor is used to record the process parameter change curve. An anomaly detector is used to extract parameter data that exceeds a preset threshold range to obtain quality anomaly feature data. A weight calculator is used to assign weights to each anomaly feature within the quality anomaly feature data. A correlation analyzer is used to calculate the correlation between heating rate and active ingredient content. A stability evaluator is used to determine the impact of heating rate on storage stability to obtain feature weight data. Based on the feature weight data, an impact calculator is used to construct a model for the impact of heating rate on quality indicators. The degree of impact is quantified using the analytic hierarchy process (AHP). A deep learning network is used to train feature mapping relationships to obtain a parameter impact matrix. A sorter is used to sort the impact values ​​in the parameter impact matrix in descending order. A ratio calculator is used to generate parameter adjustment coefficients for each parameter. A parameter adjuster is used to modify the process parameters according to the degree of impact to obtain optimized parameter data. Based on the optimized parameter data, a boundary checker is used to verify the validity of the parameter correction values. A constraint checker is used to ensure that the parameters meet process requirements. A parameter calibrator is used to correct any out-of-range parameters to obtain verified parameter data. For the verification parameter data, the batch consistency control requirements are integrated into the parameters through the consistency matcher, and the parameter fusion device is used to generate a complete control parameter set to obtain the production control parameter data. The standard comparator shows that the quality indicators in the production process of blood charcoal deviate from the preset standards. The content of medicinal ingredients has dropped to 78%, which is lower than the standard limit of 80%. The heating rate monitor records show that the heating rate fluctuates by 2 degrees Celsius per minute in the range of 400 to 600 degrees Celsius, exceeding the preset threshold range of ±1 degree Celsius per minute. The weight calculator assigns weights to abnormal features, where the weight of heating rate anomaly is 0.4, the weight of uneven temperature distribution is 0.3, and the weight of material residence time anomaly is 0.3. Correlation analysis shows that for every 1 degree Celsius per minute increase in heating rate, the content of medicinal ingredients decreases by 2% and storage stability decreases by 1.5%. The impact model constructed by the impact calculator shows that when the heating rate reaches 10 degrees Celsius per minute, the active ingredient content drops to 75%, and the storage stability index drops to 85% of the standard value. Hierarchical analysis results indicate that the heating rate has an impact weight of 0.6 on the active ingredient content, and a weight of 0.4 on the storage stability. The deep learning network uses a three-layer structure: the input layer receives three sets of parameters: heating rate, temperature distribution, and dwell time. The hidden layer contains 64 neurons, and the output layer predicts the active ingredient content and storage stability index. Training results show a prediction accuracy of 90%. The sorter ranks the impact from greatest to least, with the heating rate impact coefficient of 0.48 ranking first. The scale calculator uses this information to generate parameter adjustment coefficients: 0.4 for heating rate, 0.3 for temperature distribution, and 0.3 for dwell time. After these adjustments, the heating rate drops to 8 degrees Celsius per minute.The boundary checker verifies the corrected parameters to ensure that the heating rate is maintained in the range of 7 to 9 degrees Celsius per minute, the temperature distribution uniformity deviation is controlled within the range of ±20 degrees Celsius, the material residence time is maintained between 22 and 28 minutes, and the constraint condition check shows that all parameters meet the process requirements. The consistency matching results show that with the corrected parameter combination, the fluctuation of the active ingredient content between batches is reduced to within 3%, and the fluctuation of the storage stability index is controlled within the range of 2%. The control parameter set generated by the parameter fusion includes: heating rate of 8±1 degrees Celsius per minute, temperature distribution uniformity of more than 90%, and material residence time of 25±3 minutes. Practical application verification shows that when the optimized control parameter combination is used, the active ingredient content is increased to 83%, the storage stability index is restored to 95% of the standard value, the quality fluctuation between batches is significantly reduced, and the quality indicators of 10 batches of products produced in succession remain stable, and all parameters are within the control range.

[0037] Step S108: Based on the control parameter set of the blood charcoal production process, a virtual model of the blood charcoal production equipment is established, simulation results are generated, and compared and corrected with the actual equipment operation data. Equipment control instructions are generated and sent to the field bus. The PID control algorithm is used to adjust the operating status of the blood charcoal production equipment in real time to generate thermal weight loss curve data.

[0038] A control parameter set of the blood charcoal production equipment is obtained through a three-dimensional modeler, and virtual equipment model data is generated according to the control parameter set, and the virtual equipment model data includes the temperature field distribution data of the heating furnace and the motion parameters of the conveying device; a working condition simulator is used to generate simulated production data according to the virtual equipment model data, and the simulated production data includes temperature field change data and material transmission status parameters; the actual equipment operation parameters are obtained through a parameter collector for the simulated production data, and a comparison calculator is used to generate calibrated model data, and the calibrated model data is determined by the deviation value between the virtual and actual parameters; the equipment control data is generated according to the calibrated model data through an instruction encoder, and the equipment control data is obtained by converting the field bus message. The field bus message is transmitted to the equipment controller, and a thermal gravimetric monitor is used to record the material mass change curve to obtain thermal gravimetric curve data.

[0039] Specifically, a 3D modeler was used to construct a geometric model of the production equipment based on the control parameter set for the blood charcoal production equipment. A thermal parameter generator was used to configure the temperature field distribution of the heating furnace, and a material transport modeler was used to generate the motion parameters of the conveyor mechanism, generating virtual equipment model data. A working condition simulator was used to generate temperature field change data during the heating process. A material dynamics calculator was used to obtain material transport state parameters, and a heat and mass transfer calculator was used to generate heat transfer values, generating simulated production data. Based on the simulated production data, a parameter collector was used to obtain actual equipment temperature field data and material transport data. A comparison calculator was used to generate the deviation between virtual and actual parameters. The virtual model parameters were calibrated using an iterative corrector, generating calibrated model data. Based on the calibrated model data, a command encoder was used to generate equipment control variables. A communication protocol converter was used to convert the control variables into fieldbus messages. The command distributor was used to transmit the control commands to the equipment controller, generating equipment control data. Based on the equipment control data, a proportional-integral-differential (PI / D) arithmetic unit was used to calculate the temperature control output. The actuator driver module was used to adjust the heating power output. A feedback compensator was used to dynamically correct the temperature curve, generating temperature control data. The conveyor controller adjusts the material conveying speed based on the temperature control data. A thermal weight loss monitor records the material mass change curve, and a data processor filters the curve data to generate the thermal weight loss curve data. A 3D modeler constructs a virtual equipment model based on the actual production line dimensions. The heating furnace chamber dimensions are 2000 mm long, 800 mm wide, and 600 mm high. The thermal parameter generator arranges 16 temperature field monitoring points within the chamber. The material transport modeler sets the conveyor belt width to 600 mm and the speed range to 100 to 300 mm per minute. A working condition simulator simulates the heating process. The temperature field changes show that as the furnace chamber temperature rises from room temperature to 850 degrees Celsius, the temperature difference between each monitoring point remains within ±20 degrees Celsius. Material dynamics calculations indicate that at a conveying speed of 200 mm per minute, the material residence time in the heating zone is 25 minutes. Heat transfer calculations show that the temperature difference between the material surface and the center is less than 30 degrees Celsius. The parameter collector acquires operational data from the actual equipment, revealing that the actual temperatures at the furnace's 16 temperature measurement points deviate by 5 to 15 degrees Celsius from those in the virtual model. The material conveying speed fluctuates within ±10 mm / minute. The iterative corrector, through 10 iterations, reduces the error between the virtual model parameters and the actual equipment to within 3%. The command encoder converts the temperature control variable into a standard 4 to 20 mA current signal. The communication protocol converter generates control messages using the Modbus-RTU protocol. The command distributor transmits the control commands to the temperature controller via the RS485 bus, maintaining a 100 millisecond communication cycle.The PID controller utilizes a cascade control structure, with the inner loop controlling heating power and the outer loop controlling the temperature profile. The proportional coefficient is set to 1.2, the integral time is 120 seconds, and the differential time is 30 seconds. The actuator driver module converts the control output into a pulse-width modulated signal, controlling the conduction time of the thyristor (SCR) to adjust the heating power. The conveyor controller uses variable frequency speed regulation with an output frequency range of 10 to 50 Hz, corresponding to conveyor speeds of 100 to 300 mm / min. The thermogravimetric monitor records a material mass loss rate of 0.3% / min under standard operating conditions, with a plateau duration of 18 minutes. The data processor uses a low-pass filtering algorithm to smooth the thermogravimetric curve. Through coordinated control between the virtual model and the actual equipment, the controller automatically adjusts heating power distribution when abnormal temperature distribution is detected, ensuring temperature uniformity of over 95%. It also dynamically adjusts the conveyor speed based on the thermogravimetric curve to maintain a stable plateau duration of 15 to 20 minutes, ensuring consistent product quality. Actual operation data shows that after adopting this control method, the quality fluctuation between product batches is reduced to less than 2%, equipment energy consumption is reduced by 8%, and production efficiency is increased by 10%.

[0040] Step S109, by comparing the deviation of the adjusted thermal weight loss curve data with the preset standard curve, using the particle swarm optimization algorithm with the goal of minimizing the deviation, updating the weight of the abnormal features in the mapping relationship, combining the association between the weight loss platform time and the storage stability parameters, obtaining the optimized dynamic mapping relationship, repeating the prediction of product quality indicators and adjusting the control parameters until the product quality indicators stabilize at the target level, and obtaining the process optimization results of blood charcoal.

[0041] Characteristic point data is extracted from the thermogravimetric curve data through a curve analyzer, and the curve analyzer calculates the deviation value between the characteristic point data and the corresponding point of the preset standard curve; a particle swarm optimizer is used to construct a deviation square sum minimization objective function for the deviation value, and the particle swarm optimizer adopts a global search method to obtain characteristic weight data; based on the characteristic weight data, the time data of the weight loss platform and the storage stability index data are calculated through a time series associator, and the time series associator adopts a dynamic mapping matrix to obtain characteristic mapping data; based on the characteristic mapping data, a quality predictor is used to calculate the quality index prediction value, and the quality predictor adopts a dynamic adjustment method to obtain parameter optimization data, and the parameter optimization data is used to update the equipment operating parameters; it is checked whether the optimization termination condition is met, and the process optimization result data is generated using a result outputter.

[0042] Specifically, based on the adjusted thermal weight loss curve data, a curve analyzer extracts three characteristic points: the temperature plateau point, the weight loss rate point, and the carbon residue rate point. A feature comparator calculates the deviation from the corresponding point on the preset standard curve, and a data normalizer normalizes the deviation values ​​to obtain standardized deviation data. A particle swarm optimizer constructs a squared deviation minimization objective function for the standardized deviation data. A particle position updater iteratively calculates feature weights, and a global searcher searches for the optimal weight combination to obtain feature weight data. Based on the feature weight data, a time series correlator calculates the correlation between the weight loss plateau time and the storage stability index. A weight assigner assigns weights to the correlation data, and a feature mapper constructs a dynamic mapping matrix to obtain feature mapping data. Based on the feature mapping data, a quality predictor calculates predicted product quality indicators. An indicator evaluator evaluates the prediction results, and a quality determiner determines whether the indicators meet the target levels to obtain quality assessment data. Based on the quality assessment data, a control parameter generator updates equipment operating parameters. A parameter checker verifies the validity of the updated values, and a dynamic adjuster performs parameter adjustments to obtain parameter optimization data. An iterative controller determines the stability of quality indicators based on the parameter optimization data. A termination determinator checks whether the optimization termination conditions are met, and a result outputter generates process optimization result data. A curve analyzer extracts key feature points from the thermal gravimetric analysis curve. The temperature plateau lies between 450 and 550 degrees Celsius, indicating an actual plateau temperature of 530 degrees Celsius, 30 degrees Celsius higher than the standard curve. The maximum weight loss rate is 0.8% per minute, 0.3% per minute higher than the standard, and the carbon residue is 28%, 2 percentage points lower than the standard. A particle swarm optimizer (PSO) uses a population size of 50 and a maximum number of iterations of 100. The objective function is constructed by minimizing the sum of squared deviations. Each particle represents a set of feature weights, with initial weights of 0.4 for the temperature plateau, 0.3 for the weight loss rate, and 0.3 for the carbon residue. After global search optimization, the optimal weight combination is found: 0.5 for the temperature plateau, 0.3 for the weight loss rate, and 0.2 for the carbon residue. Time series correlation analysis showed a significant positive correlation between weightlessness plateau duration and storage stability. When the plateau period was maintained at 18 minutes, the rate of decline in active ingredient content was the lowest, at 0.3% per month. The weight allocator accordingly set a weight of 0.6 for plateau time and a weight of 0.4 for temperature uniformity. The feature mapping matrix reflected that for every minute decrease in plateau time, the rate of decline in active ingredient content increased by 0.1% per month. Quality prediction results showed that, using the optimized weight parameters, the active ingredient content of the next batch of products was predicted to be 85%, and the storage stability index was expected to remain above 80% after 12 months. Comparing the results with the target values, the quality assessor found that while the active ingredient content met the target, storage stability still had room for improvement, triggering a parameter optimization signal from the quality determinant.The results of the control parameter update showed that the heating rate needed to be reduced to 7 degrees Celsius per minute and the material conveying speed was adjusted to 250 mm per minute. Parameter verification showed that all values ​​were within the process's allowable range, and the dynamic adjuster sent the new parameters to the equipment controller via the fieldbus. During the iterative control process, the quality indicators stabilized after the third iteration. The fluctuation of the efficacy content of five consecutive batches of products was controlled within the range of ±1%, and the attenuation rate of the storage stability indicator was reduced to 0.25% per month over 12 months, meeting the optimization termination conditions. The process optimization results showed that the product quality stability was significantly improved after adopting the new control parameters. Actual production application verification showed that the optimized process parameters significantly improved the consistency of product batches. Test data from 30 consecutive production batches showed that the efficacy content remained between 84% and 86%, and after 12 months of storage, the content dropped to 81% to 83%. The residual carbon rate was stable in the range of 29% to 31%. The average deviation between the thermal gravimetric loss curve and the standard curve was reduced to within 2%, achieving stable control of product quality.

[0043] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A production quality control method for blood charcoal, characterized in that: The method comprises: Obtain real-time thermogravimetric curve data during the production of blood charcoal, extract abnormal thermogravimetric features, and obtain an abnormal feature set; Based on the abnormal feature set, the relationship between the weight loss rate change, characteristic temperature point offset, carbon residue rate fluctuation and weight loss platform time is quantified. Combined with the correlation between the abnormal change of weight loss rate and the content of active ingredients, a feature combination is formed. Input feature combinations, train the dynamic mapping relationship between abnormal features and Xueyu charcoal product quality indicators, predict the product quality indicators of the current production batch of Xueyu charcoal, and generate quality assessment results; If the abnormal weightlessness platform time causes the storage stability to deviate from the standard, the quality deviation characteristics are obtained by analyzing the relationship between the weightlessness platform time and the storage stability parameters; If the duration of the plateau period exceeds the preset time threshold, the abnormal time interval of uneven heating of the material and the equipment temperature control deviation are determined by analyzing the thermal weight loss curve data and the equipment operating status within the abnormal time window, and the abnormal equipment status characteristic data is generated; Based on the abnormal characteristic data of equipment status, the relationship between the weightlessness platform time anomaly and the material conveying rate and heating rate is analyzed to determine the key parameters that cause batch consistency deviations. The control parameter set is generated in combination with the temperature rise gradient constraint relationship. If the real-time quality assessment results or quality deviation characteristics deviate from the preset quality standards, the degree of influence of the heating rate on the content of medicinal ingredients and storage stability is calculated, and the key parameters are adjusted proportionally based on the influence degree ranking results. Combined with the control parameter set, a control parameter set for the blood charcoal production process is generated.

2. The production quality control method of blood charcoal according to claim 1, characterized in that: The method of obtaining real-time thermogravimetric curve data during the production of blood charcoal, extracting abnormal thermogravimetric features, and obtaining an abnormal feature set includes: receiving real-time gas flow data at a preset temperature point collected by a gas flow sensor, combining the gas flow data with mass change data obtained by a mass change detector to obtain thermal weight loss curve data of the blood residue charcoal; According to the thermal weight loss curve data, temperature change rate data and mass change rate data are obtained through a thermal parameter detector, and the mass change rate data are analyzed by a carbonization speed monitor to obtain an initial weight loss rate data set; A three-parameter thermal analyzer is used to obtain characteristic temperature point data for the initial weight loss rate data set. The characteristic temperature point data is analyzed by a thermal weight loss plateau detection algorithm to obtain abnormal fluctuation data of the residual carbon rate. If the weight loss rate change value in the abnormal fluctuation data of the residual carbon rate exceeds a preset threshold range, a time series analysis method is used to perform a sliding time window calculation on the weight loss rate data to obtain an abnormal feature set of thermal weight loss.

3. The production quality control method of blood charcoal according to claim 1, characterized in that: The abnormal feature set is based on quantifying the relationship between the weight loss rate change, characteristic temperature point offset, carbon residue rate fluctuation and weight loss platform time, and combining the correlation between the abnormal change of weight loss rate and the content of active ingredients to form a feature combination, including: Obtaining a weight loss rate value and a temperature point offset, wherein the weight loss rate value is collected by a thermogravimetric curve analyzer during the reaction process; A carbon residue rate value is obtained by a carbon residue detector according to the weight loss rate value and the temperature point offset, and the carbon residue rate value and the active ingredient content data obtained by the active ingredient detector are combined to form a thermal weight loss characteristic data set; For the thermal weight loss characteristic data set, a data normalization processor is used to perform normalization calculation on the weight loss rate value and the carbon residue rate value to obtain a standardized characteristic data set; The standardized feature dataset is processed using a random forest algorithm to obtain a feature combination for input into a machine learning model.

4. The production quality control method of blood residue charcoal according to claim 1, characterized in that: The input feature combination trains the dynamic mapping relationship between abnormal features and the quality indicators of Xueyu charcoal products, predicts the product quality indicators of the current production batch of Xueyu charcoal, and generates quality assessment results, including: Acquire activity value data of historical batch samples through the blood residue charcoal quality index database according to the feature combination data, wherein the activity value data of historical batch samples includes a storage stability benchmark parameter; A parameter normalization processor is used to perform normalization calculation on the activity value data of the historical batch samples, and abnormal fluctuation values ​​are eliminated through a data filter to obtain a standardized quality parameter set; A time series prediction model is constructed using a long short-term memory network according to the standardized quality parameter set, and a quality indicator mapping model is obtained by modeling the change trend of the active ingredient content using a sliding time window method; The quality index mapping model is used to measure real-time data on the content of effective ingredients using an effective content analyzer, and the prediction accuracy is verified using a cross-validator to obtain quality assessment data.

5. The production quality control method of blood residue charcoal according to claim 1, characterized in that: If the abnormal weightlessness platform time causes the storage stability to deviate from the standard, the quality deviation characteristics are obtained by analyzing the relationship between the weightlessness platform time and the storage stability parameters, including: A real-time temperature curve and a mass change curve are obtained based on the weightlessness platform time data, and the real-time temperature curve and the mass change curve are analyzed and processed using a curve fitter to obtain a plateau duration value; based on the plateau duration value, a stability index analyzer is used to measure the efficacy content decay rate and the moisture content change rate, and a standard limit comparator is used to compare the efficacy content decay rate with a preset standard limit to obtain stability deviation data; If the attenuation rate of the efficacy content in the stability deviation data exceeds the preset standard limit, the weightlessness platform time and the storage stability parameter are normalized, and a mapping relationship between the weightlessness platform time and the storage stability parameter is established by support vector regression to obtain parameter association data; A quality deviation feature vector is obtained by a feature extractor for the parameter association data, and a feature combiner is used to combine the quality deviation feature vector to obtain a quality deviation feature data set, including the duration of the plateau phase.

6. The method for controlling the production quality of blood charcoal according to claim 1, characterized in that: If the duration of the plateau period exceeds the preset time threshold, the abnormal time interval of uneven heating of the material and the equipment temperature control deviation are determined by analyzing the thermal weight loss curve data and the equipment operating status within the abnormal time window, and the abnormal equipment status characteristic data is generated, including: According to the duration of the plateau period, a thermogravimetric curve recorder is used to obtain temperature change and mass loss data, and the thermogravimetric curve recorder records the surface temperature distribution data of the material within the abnormal period; A heat transfer rate data is obtained using a heat conduction meter based on the temperature distribution data, and a value of the material heating unevenness is calculated using the heat conduction meter; A temperature control state prediction model is constructed based on the heat transfer rate data through a deep neural network, and the temperature deviation value is quantified and calculated through the deep neural network. The support vector machine classifies the device state according to the temperature deviation value, extracts the device temperature control characteristics, and obtains the device state abnormality feature data.

7. The method for controlling the production quality of blood charcoal according to claim 1, characterized in that: According to the abnormal characteristic data of the equipment status, the relationship between the weightlessness platform time anomaly and the material conveying rate and heating rate is analyzed to determine the key parameters that cause batch consistency deviation. Combined with the temperature rise gradient constraint relationship, a control parameter set is generated, including: Performing normalization processing on the material conveying rate and heating rate data through a data normalization processor, wherein the normalization processing obtains a standardized data set; A time series correlator is used to calculate a correlation coefficient between the weightlessness platform time and the standardized data set for the standardized data set, wherein the correlation coefficient is used to obtain a key parameter data set; Constructing a parameter change prediction model based on the key parameter data set using a random forest algorithm, wherein the prediction model outputs parameter influence data; A temperature gradient monitor is used to record the temperature rise data of the material heating process based on the parameter influence data, and a temperature rise gradient constraint relationship is established between the temperature rise data and the residence time data obtained by the residence time detector; The power change data of the heating equipment is obtained according to the temperature rise gradient constraint relationship. The power change data is processed by a curve fitter and integrated with the temperature rise gradient constraint relationship to generate a control parameter set.

8. The method for controlling the production quality of blood charcoal according to claim 1, wherein: If the real-time quality assessment result or the quality deviation characteristic deviates from the preset quality standard, the degree of influence of the heating rate on the content of the medicinal ingredient and the storage stability is calculated, and the key parameters are adjusted proportionally based on the influence degree ranking result. Combined with the control parameter set, a control parameter set for the blood charcoal production process is generated, including: The deviation between the quality index and the preset quality standard is obtained through a standard comparator, and the process parameter change curve is recorded using a heating rate monitor to obtain quality abnormality characteristic data; According to the quality abnormality feature data, a weight calculator is used to assign a weight to the abnormal feature, and a correlation analyzer is used to calculate the correlation between the heating rate and the content of the medicinal ingredient to obtain feature weight data; A heating rate influence model on quality index is constructed based on the characteristic weight data by using an influence calculator, and the influence degree is quantitatively calculated by using the hierarchical analysis method to obtain a parameter influence matrix; According to the parameter influence matrix, the process parameters are corrected by a parameter adjuster according to the degree of influence, and the out-of-range parameters are corrected by a parameter calibrator to obtain verification parameter data; The batch consistency control requirements are integrated into the control parameter set through the consistency matcher for the verification parameter data to generate the control parameter set for the blood charcoal production process.

9. The method for controlling the production quality of blood charcoal according to claim 1, characterized in that: The method includes: establishing a virtual model of the blood charcoal production equipment based on a control parameter set of the blood charcoal production process, generating simulation results, comparing and correcting the results with actual equipment operation data, generating equipment control instructions and issuing the equipment control instructions to a field bus, using a PID control algorithm to adjust the operating state of the blood charcoal production equipment in real time, and generating thermal weight loss curve data, specifically including: Obtaining a control parameter set of a blood charcoal production device through a three-dimensional modeler, and generating virtual device model data based on the control parameter set, wherein the virtual device model data includes temperature field distribution data of a heating furnace and motion parameters of a conveying device; Generate simulated production data using a working condition simulator according to the virtual equipment model data, wherein the simulated production data includes temperature field change data and material transmission state parameters; Actual equipment operating parameters are obtained through a parameter collector for the simulated production data, and calibrated model data is generated using a comparison calculator. The calibrated model data is determined by deviation values ​​between virtual and actual parameters. Equipment control data is generated through an instruction encoder based on the calibrated model data. The equipment control data is obtained by converting a fieldbus message. The fieldbus message is transmitted to the equipment controller, and a thermal gravimetric monitor is used to record a material mass change curve to obtain thermal gravimetric curve data.

10. The method for controlling the production quality of blood charcoal according to claim 1, characterized in that: The method comprises: comparing the deviation of the adjusted thermogravimetric curve data with a preset standard curve, using a particle swarm optimization algorithm with the goal of minimizing the deviation, updating the weight of abnormal features in the mapping relationship, combining the association between the weight loss platform time and the storage stability parameter, obtaining an optimized dynamic mapping relationship, repeatedly predicting product quality indicators and adjusting control parameters until the product quality indicators stabilize at the target level, and obtaining the process optimization results of the blood residue charcoal, specifically comprising: Extracting characteristic point data through a curve analyzer according to the thermogravimetric curve data, and calculating a deviation value between the characteristic point data and a corresponding point of a preset standard curve; A particle swarm optimizer is used to construct a deviation square sum minimization objective function for the deviation value, and the particle swarm optimizer adopts a global search method to obtain feature weight data; Calculating weightlessness platform time data and storage stability index data through a time series correlator according to the feature weight data, wherein the time series correlator uses a dynamic mapping matrix to obtain feature mapping data; Calculating a quality index prediction value for the feature mapping data using a quality predictor, wherein the quality predictor obtains parameter optimization data using a dynamic adjustment method, and the parameter optimization data is used to update the device operating parameters; Check whether the optimization termination conditions are met and use the result outputter to generate process optimization result data.

Images