A system and method for detecting faults in a pill making process

Abstract

The present invention discloses a method for detecting faults in the dripping process of drop pills, which is used to detect excessive quality indicators such as pill weight or pill shape caused by various factors during the dripping process. This method is based on the laser detection system described in CN112903508A and analyzes the detected droplet width sequence. This method calculates the corresponding characteristic index for the width sequence corresponding to each droplet generated during the dripping process; establishes a PCA model using the characteristic index under normal dripping conditions; simulates multiple abnormal dripping conditions, and inputs the characteristic indexes calculated under abnormal conditions into the PCA model for prediction to verify the fault detection performance of the model. This method can be used for fault detection in the actual production process of drop pills. When an abnormality is found in a drop pill index, an alarm will be automatically issued, prompting a check of the process parameters.

Description

Technical Field

[0001] The present invention belongs to the field of droplet feature extraction and process parameter fault detection and alarm in the process of dripping pill preparation, and particularly relates to a fault detection system and method in the process of dripping pill preparation. Background Art

[0002] Dropping pills are a widely used traditional Chinese medicine formulation. Their preparation involves dropping a uniformly mixed, molten liquid of raw materials and excipients into an immiscible condensate under specific process conditions. The liquid cools and solidifies, forming a spherical solid. Changes in process parameters can lead to variations in the weight and morphology of the dropping pills, but evaluation of the pills themselves is often delayed and cannot reflect these changes in process parameters in a timely manner.

[0003] This application's prior patent application, CN112903508A, discloses an online detection method for droplet formulations. This method utilizes a laser detection system to measure the width of droplets as they pass through the detector in real time, followed by post-processing to characterize the droplets. Using the raw droplet width data measured by this method, this application provides a method for analyzing this raw droplet width data to derive a series of characteristic indicators, and employs a PCA model to predict in real time whether process conditions are abnormal. Summary of the Invention

[0004] To overcome the above technical problems, the present invention provides a method for detecting faults in the dripping pill production process. The method is easy to use, has a fast response speed, strong anti-interference ability, and good stability.

[0005] To achieve the above objectives, the technical solutions provided by the present invention are as follows:

[0006] A fault detection system for a pill making process, comprising:

[0007] (1) a detection system, wherein the detection system is used to detect droplet information during the dripping process;

[0008] (2) a processing system, the processing system being used to process the detected droplet information;

[0009] (3) An alarm system, wherein the alarm system is used to output an alarm when the information of the droplets evaluated by the model system is abnormal.

[0010] The fault detection system for the dripping pill making process is used in conjunction with the dripping system, which is a system that drips materials into dripping pills;

[0011] The processing system is used to analyze the information of droplets produced under normal dripping conditions and establish a droplet characteristic model, and to process the droplet information data produced during the dripping production process and obtain the same characteristic indicators; the detection system is a laser detection system;

[0012] Furthermore, the information of the droplet is the width of the droplet;

[0013] The laser detection system includes a voltage input module, a laser transmitter, a laser receiver, a sensor amplifier, a data acquisition card, and a data acquisition computer;

[0014] The processing system uses characteristic indicators under normal dripping conditions to establish a PCA model; simulates multiple abnormal dripping conditions, inputs the characteristic indicators calculated under abnormal conditions into the PCA model for prediction to verify the fault detection performance of the model; and processes the droplet width data generated by the current dripping production process to obtain the same characteristic indicators;

[0015] The alarm system inputs the current droplet characteristics into the PCA model to evaluate whether the process conditions are abnormal and output an alarm;

[0016] The dripping system includes a set value operating system, and the set value operating system is used to change the set value according to the operator's operation.

[0017] The dripping system includes a set value operating system, which is used to change the set value according to the operator's operation;

[0018] The PCA model is established using characteristic indices obtained by analyzing width data of droplets produced under normal dripping conditions, and is verified using characteristic indices obtained by analyzing width data of droplets produced under simulated abnormal process conditions.

[0019] Furthermore, the dripping process is a process in which a mixed liquid medicine drips from a dripper and falls into a condensate in a dripping pill preparation process; the method for processing information includes: droplet node selection and segmentation, and characteristic index calculation.

[0020] Furthermore, the method of selecting and segmenting the droplet nodes is as follows: the continuous droplet width values measured by the laser detector are divided into individual droplets. For each droplet, four nodes are taken and divided into three segments, where point A is the bottom of the droplet, point B is the point with the maximum width, point C is where the material begins to draw, and point D is where the liquid breaks.

[0021] The method for selecting point C is to traverse the data points backward from point B until the i-th point meets the conditions:

[0022]

[0023] Then the i-th point satisfies the selection of point C. In the formula, d i is the droplet width at point i, d i+N is the droplet width at N points after the i-th point, and Δ is the judgment threshold. After optimization, it is determined that N = 15 and Δ = 0.05.

[0024] 7. The fault detection system for the dripping pill production process according to claim 5, wherein the characteristic index comprises: the width at point B and point C, denoted as d B d C ; The length of AB segment and BC segment, that is, the number of data points contained between nodes, is recorded as len AB 、len BC The slope of segment AB and segment BC, that is, the ratio of the width at point B to the length of segment AB and segment BC, is expressed as The half-peak width of the AC segment, that is, the number of data points between the half-maximum width of the AB segment and the half-maximum width of the BC segment, is recorded as half_peak_width; the width of the AB segment and the BC segment at the three-division point is recorded as The drop rate, that is, the length between points A of adjacent droplets, is recorded as rate.

[0025] A method for detecting faults in a pill drop-making process includes the following steps:

[0026] (1) The dripping system performs the dripping process of the pills, and uses the detection system to detect the droplet information during the dripping process;

[0027] (2) The information of each droplet is processed and the characteristic index is calculated. The characteristic index is input into the PCA model to predict whether the process conditions are abnormal. If an abnormality occurs, an alarm is issued.

[0028] The material temperature in the dripping system is adjustable; the material input flow rate is adjustable; the material liquid level is adjustable; the detection system is a laser detection system; the dripping process is the process in which the mixed liquid drips from the dripper into the condensate in the dripping pill preparation process; the droplet information is the droplet width;

[0029] Furthermore, the temperature of the material in the dripping system is controlled at a set value by adjusting the heating power; the laser detection system is composed of a voltage input module, a laser transmitter, a laser receiver, a sensor amplifier, a data acquisition card, and a data acquisition computer.

[0030] Furthermore, the laser detection system detects the width of the droplets passing through the laser detector in real time during the dripping process and transmits it to the data acquisition computer;

[0031] Furthermore, the method for extracting droplet information includes: droplet node selection and segmentation, and characteristic index calculation.

[0032] The method for selecting and segmenting droplet nodes is as follows: the continuous droplet width values measured by the laser detector are divided into individual droplets. For each droplet, four nodes are taken and divided into three segments, where point A is the bottom of the droplet, point B is the point with the maximum width, point C is the point where the material begins to draw, and point D is the point where the liquid breaks.

[0033] The method for selecting point C is to traverse the data points backward from point B until the i-th point meets the conditions:

[0034]

[0035] Then the i-th point satisfies the selection of point C. In the formula, d i is the droplet width at point i, d i+N is the droplet width at N points after the i-th point, and Δ is the judgment threshold. After optimization, it is determined that N = 15 and Δ = 0.05.

[0036] Furthermore, the characteristic index includes: the width at point B and point C, denoted as d B d C ; The length of AB segment and BC segment, that is, the number of data points contained between nodes, is recorded as len AB 、len BC The slope of segment AB and segment BC, that is, the ratio of the width at point B to the length of segment AB and segment BC, is expressed as The half-peak width of the AC segment, that is, the number of data points between the half-maximum width of the AB segment and the half-maximum width of the BC segment, is recorded as half_peak_width; the width of the AB segment and the BC segment at the three-division point is recorded as The drop rate, that is, the length between points A of adjacent droplets, is recorded as rate.

[0037] This method, based on the laser detection system described in CN112903508A, analyzes the detected droplet width sequence. It calculates characteristic indicators for each droplet width sequence generated during the dripping process. A PCA model is established using these indicators under normal dripping conditions. A variety of abnormal dripping conditions are simulated, and the characteristic indicators calculated under these conditions are input into the PCA model for prediction to verify the model's fault detection performance. This method can be used for fault detection in the actual dripping pill production process. If an abnormality is detected in a pill indicator, an automatic alarm is triggered, prompting a check of process parameters.

[0038] The present invention can simultaneously evaluate the indicators of the dropping pills themselves and can timely reflect the changes in process parameters.

[0039] This application's prior patent application, CN112903508A, discloses an online detection method for droplet formulations. This method utilizes a laser detection system to measure the width of droplets as they pass through the detector in real time, followed by post-processing to characterize the droplets. Using the raw droplet width data measured by this method, this application provides a method for analyzing this raw droplet width data to derive a series of characteristic indicators, and employs a PCA model to predict process anomalies in real time. This rapid and sensitive method reduces production costs and improves the yield rate of finished products. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is the score graph of the PCA model established based on the characteristic indicators of normal dripping conditions;

[0041] Figure 2 is the Hotelling's T of the PCA model established based on the characteristic indicators of normal dripping conditions 2 Control charts;

[0042] Figure 3 The DModX control chart of the PCA model established based on the characteristic indicators of normal dripping conditions;

[0043] Figure 4 The PCA score diagram of the model’s prediction of the characteristic index of abnormal dripping conditions of valve opening;

[0044] Figure 5 Hotelling's T is the characteristic index of the valve opening abnormal dripping condition predicted by the model 2 Control charts;

[0045] Figure 6 DModX control chart for model prediction of characteristic index of abnormal dripping condition of valve opening;

[0046] Figure 7 The PCA score diagram of the model predicting the characteristic index of the dripping condition when the material temperature is abnormally low;

[0047] Figure 8 Hotelling's T is the characteristic index of the dripping condition for predicting abnormal low material temperature. 2 Control charts;

[0048] Figure 9 The DModX control chart for the model predicting the characteristic index of the dripping condition when the material temperature is abnormally low;

[0049] Figure 10 The PCA score diagram of the model predicting the characteristic index of the dripping condition when the material temperature is abnormally high;

[0050] Figure 11Hotelling's T is the characteristic index of the dripping condition for predicting abnormal high material temperature. 2 Control charts;

[0051] Figure 12 The DModX control chart for the model to predict the characteristic index of the dripping condition when the material temperature is abnormally high;

[0052] Figure 13 The PCA score diagram of the model’s prediction of the characteristic index of abnormal material level dripping conditions;

[0053] Figure 14 Hotelling's T is the characteristic index of the model for predicting abnormal material level dripping conditions. 2 Control charts;

[0054] Figure 15 DModX control chart for the model to predict the characteristic index of abnormal material level dripping conditions;

[0055] Figure 16 Schematic diagram of the droplet node selection and segmentation method;

[0056] Figure 17 A flow chart was established for the fault detection method of the dripping pill making process. DETAILED DESCRIPTION

[0057] The following experimental examples and embodiments are intended to further illustrate but not limit the present invention.

[0058] The following is a clear and complete description of the technical solutions in the embodiments of the present invention, in conjunction with the accompanying drawings in the examples of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0059] The present invention provides a method for detecting faults in the pill making process. Based on the laser detection system described in CN112903508A, the method analyzes the detected pill width sequence. The specific application process is as follows:

[0060] The raw materials and auxiliary materials for the dropping pills are melted and mixed evenly, and then transported to the storage tank of the dropping device. The temperature and liquid level in the storage tank are maintained at the set values by adjusting the heating power and material delivery flow rate. The dropping pills are dropped during the dropping process, and the droplet width is detected by a laser detection system. The width sequence of each droplet is processed and the characteristic index is calculated:

[0061] The method of selecting and segmenting the droplet nodes is as follows: Figure 16, the continuous droplet width values measured by the laser detector are divided into individual droplets. For each droplet, four nodes are taken and divided into three segments. Point A is the bottom of the droplet, point B is the point with the maximum width, point C is where the material starts to draw, and point D is where the liquid breaks.

[0062] The method for selecting point C is to traverse the data points backward from point B until the i-th point meets the conditions:

[0063]

[0064] Then the i-th point satisfies the selection of point C. In the formula, d i is the droplet width at point i, d i+N is the droplet width at N points after the i-th point, and Δ is the judgment threshold. After optimization, it is determined that N = 15 and Δ = 0.05.

[0065] The characteristic indicators include: the width at point B and point C, denoted as d B d C ; The length of AB segment and BC segment, that is, the number of data points contained between nodes, is recorded as len AB 、len BC The slope of segment AB and segment BC, that is, the ratio of the width at point B to the length of segment AB and segment BC, is expressed as The half-peak width of the AC segment, that is, the number of data points between the half-maximum width of the AB segment and the half-maximum width of the BC segment, is recorded as half_peak_width; the width of the AB segment and the BC segment at the three-division point is recorded as The drop rate, that is, the length between points A of adjacent droplets, is recorded as rate.

[0066] Keep the process conditions at normal values, perform dripping and data analysis according to the above analysis process, establish the PCA model, and obtain the following Figure 1 The score graph of the PCA model established by the normal dripping condition characteristic index is shown as Figure 2 Hotelling's T of the PCA model established based on the characteristic index of normal dripping conditions shown 2 Control charts, such as Figure 3 The DModX control chart of the PCA model established by the characteristic indicators of the normal dripping conditions is shown. By simulating the abnormal conditions of the four process parameters, the following four embodiments are obtained.

[0067] Example 1

[0068] The dripper valve opening is reduced to 1 / 2 of the normal process conditions, while other process parameters are kept to meet the normal process conditions. According to the above analysis process, dripping and data analysis are carried out, and the established PCA model is used for prediction, and the following is obtained: Figure 4 The PCA score diagram of the model predicting the characteristic index of abnormal valve opening dripping condition is shown in the figure. Figure 5 The model shown predicts Hotelling's T of the characteristic index of valve opening abnormal dripping condition 2 Control charts, such as Figure 6 The model shown is a DModX control chart for predicting the characteristic index of the valve opening abnormal dripping condition.

[0069] Example 2

[0070] The material temperature in the storage tank is lowered to 15°C lower than that under normal process conditions, while other process parameters are kept to meet normal process conditions. According to the above analysis process, dripping and data analysis are carried out, and the established PCA model is used for prediction, and the following is obtained: Figure 7 The model shown in the figure predicts the PCA score diagram of the characteristic index of the dripping condition with abnormal low material temperature, as shown in Figure 8 The model shown predicts Hotelling's T of the characteristic index of the dripping condition with abnormally low material temperature. 2 Control charts, such as Figure 9 The model shown is a DModX control chart for the characteristic index of dripping conditions when the material temperature is abnormally low.

[0071] Example 3

[0072] The material temperature in the storage tank is raised to 20°C higher than that under normal process conditions, while other process parameters are kept to meet normal process conditions. According to the above analysis process, dripping and data analysis are carried out, and the established PCA model is used for prediction, and the following is obtained: Figure 10 The model shown in the figure predicts the PCA score diagram of the characteristic index of the dripping condition with abnormally high material temperature, as shown in Figure 11 The model shown predicts Hotelling's T of the characteristic index of the dripping condition with abnormally high material temperature. 2 Control charts, such as Figure 12 The model shown is a DModX control chart for predicting characteristic indicators of dripping conditions with abnormally high material temperature.

[0073] Example 4

[0074] Increase the amount of material in the storage tank to 200g more than the normal process conditions, that is, the liquid level is increased, while keeping other process parameters to meet the normal process conditions. According to the above analysis process, dripping and data analysis are carried out, and the established PCA model is used for prediction, and the following is obtained: Figure 13 The PCA score diagram of the model predicting the characteristic index of abnormal material level dripping conditions is shown in the figure. Figure 14 The model shown in the figure predicts Hotelling's T of the characteristic index of abnormal material level dripping condition. 2 Control charts, such as Figure 15The model shown is a DModX control chart for predicting characteristic indicators of abnormal material level dripping conditions.

[0075] In the actual production process, the calculated characteristic indicators are input into the PCA model in real time to predict whether the process conditions will produce abnormalities. If abnormalities occur, an alarm will be issued.

[0076] The PCA model described in the present invention is established using characteristic indices derived from the analysis of droplet width data generated under normal dripping conditions. The established model produces a PCA score graph, Hotelling's T2 control graph, and DModX control graph under normal dripping conditions. The model is verified using characteristic indices derived from the analysis of droplet width data generated under simulated abnormal process conditions. The model produces a PCA score graph, Hotelling's T2 control graph, and DModX control graph under various abnormal dripping conditions, including heating, cooling, excessive liquid volume, and insufficient valve opening. Abnormal dripping conditions include but are not limited to the above. It can be seen that the model has very few false positives for normal dripping conditions, i.e., Hotelling's T2 control graphs. 2 The control chart rarely exceeds the control limit; it is more sensitive to abnormal drip conditions, i.e. Hotelling's T 2 The control chart clearly exceeds the control limit and can distinguish abnormal process parameters. In the 3 / 4 principal component space of the PCA score chart, the four abnormal situations form four obvious clusters, based on which the abnormal process parameters can be specifically judged and alarm information can be provided.

[0077] Example 5

[0078] A fault detection system for a pill making process, comprising:

[0079] (1) a detection system, wherein the detection system is used to detect droplet information during the dripping process;

[0080] (2) a processing system, the processing system being used to process the detected droplet information;

[0081] (3) an alarm system, wherein the alarm system is used to evaluate the information generation of droplets in the model system

[0082] When abnormal, output alarm;

[0083] Example 6

[0084] A fault detection system for a pill making process, comprising:

[0085] (1) a detection system, wherein the detection system is used to detect droplet information during the dripping process;

[0086] (2) a processing system, the processing system being used to process the detected droplet information;

[0087] (3) an alarm system, wherein the alarm system is used to output an alarm when an abnormality occurs in the information of the droplets evaluated by the model system; the fault detection system is used in conjunction with the dripping system, which is a system for dripping materials into dripping pills;

[0088] The processing system is to establish a droplet feature model after analyzing the information of the droplets produced under normal dripping conditions, and process the droplet information data produced during the dripping production process to obtain the same feature index; the detection system is a laser detection system

[0089] Example 7

[0090] A fault detection system for a pill making process, comprising:

[0091] (1) a detection system, wherein the detection system is used to detect droplet information during the dripping process;

[0092] (2) a processing system, the processing system being used to process the detected droplet information;

[0093] (3) an alarm system, wherein the alarm system is used to output an alarm when an abnormality occurs in the information of the droplets evaluated by the model system; the fault detection system is used in conjunction with the dripping system, which is a system for dripping materials into dripping pills;

[0094] The processing system is used to analyze the information of droplets produced under normal dripping conditions and establish a droplet characteristic model, and to process the droplet information data produced during the dripping production process and obtain the same characteristic indicators; the detection system is a laser detection system;

[0095] The information of the droplet is the droplet width;

[0096] The laser detection system includes a voltage input module, a laser transmitter, a laser receiver, a sensor amplifier, a data acquisition card, and a data acquisition computer;

[0097] The processing system uses characteristic indicators under normal dripping conditions to establish a PCA model; simulates multiple abnormal dripping conditions, inputs the characteristic indicators calculated under abnormal conditions into the PCA model for prediction to verify the fault detection performance of the model; and processes the droplet width data generated by the current dripping production process to obtain the same characteristic indicators;

[0098] The alarm system inputs the current droplet characteristics into the PCA model to evaluate whether the process conditions are abnormal and output an alarm;

[0099] The dripping system includes a set value operating system, which is used to change the set value according to the operator's operation;

[0100] The dripping system includes a set value operating system, which is used to change the set value according to the operator's operation

[0101] Example 8

[0102] A fault detection system for a pill making process, comprising:

[0103] (1) a detection system, wherein the detection system is used to detect droplet information during the dripping process;

[0104] (2) a processing system, the processing system being used to process the detected droplet information;

[0105] (3) an alarm system, wherein the alarm system is used to output an alarm when an abnormality occurs in the information of the droplets evaluated by the model system; the fault detection system is used in conjunction with the dripping system, which is a system for dripping materials into dripping pills;

[0106] The processing system is used to analyze the information of droplets produced under normal dripping conditions and establish a droplet characteristic model, and to process the droplet information data produced during the dripping production process and obtain the same characteristic indicators; the detection system is a laser detection system;

[0107] The information of the droplet is the droplet width;

[0108] The laser detection system includes a voltage input module, a laser transmitter, a laser receiver, a sensor amplifier, a data acquisition card, and a data acquisition computer;

[0109] The processing system uses characteristic indicators under normal dripping conditions to establish a PCA model; simulates multiple abnormal dripping conditions, inputs the characteristic indicators calculated under abnormal conditions into the PCA model for prediction to verify the fault detection performance of the model; and processes the droplet width data generated by the current dripping production process to obtain the same characteristic indicators;

[0110] The alarm system inputs the current droplet characteristics into the PCA model to evaluate whether the process conditions are abnormal and output an alarm;

[0111] The dripping system includes a set value operating system, which is used to change the set value according to the operator's operation;

[0112] The dripping system includes a set value operating system, which is used to change the set value according to the operator's operation;

[0113] The PCA model is established using characteristic indices obtained by analyzing droplet width data under normal dripping conditions, and is verified using characteristic indices obtained by analyzing droplet width data under simulated abnormal process conditions.

[0114] The dripping process is the process in which the mixed liquid drops from the dripper into the condensate in the dripping pill preparation process; the method for processing the information includes: droplet node selection and segmentation, and characteristic index calculation;

[0115] The method for selecting and segmenting droplet nodes is as follows: the continuous droplet width values measured by the laser detector are divided into individual droplets. For each droplet, four nodes are taken and divided into three segments, where point A is the bottom of the droplet, point B is the point with the maximum width, point C is the point where the material begins to draw, and point D is the point where the liquid breaks.

[0116] The method for selecting point C is to traverse the data points backward from point B until the i-th point meets the conditions:

[0117]

[0118] Then the i-th point satisfies the selection of point C. In the formula, d i is the droplet width at point i, d i+N is the droplet width at N points after the i-th point, Δ is the judgment threshold.

[0119] Example 9

[0120] A fault detection system for a pill making process, comprising:

[0121] (1) a detection system, wherein the detection system is used to detect droplet information during the dripping process;

[0122] (2) a processing system, the processing system being used to process the detected droplet information;

[0123] (3) an alarm system, wherein the alarm system is used to output an alarm when an abnormality occurs in the information of the droplets evaluated by the model system; the fault detection system is used in conjunction with the dripping system, which is a system for dripping materials into dripping pills;

[0124] The processing system is used to analyze the information of droplets produced under normal dripping conditions and establish a droplet characteristic model, and to process the droplet information data produced during the dripping production process and obtain the same characteristic indicators; the detection system is a laser detection system;

[0125] The information of the droplet is the droplet width;

[0126] The laser detection system includes a voltage input module, a laser transmitter, a laser receiver, a sensor amplifier, a data acquisition card, and a data acquisition computer;

[0127] The processing system uses characteristic indicators under normal dripping conditions to establish a PCA model; simulates multiple abnormal dripping conditions, inputs the characteristic indicators calculated under abnormal conditions into the PCA model for prediction to verify the fault detection performance of the model; and processes the droplet width data generated by the current dripping production process to obtain the same characteristic indicators;

[0128] The alarm system inputs the current droplet characteristics into the PCA model to evaluate whether the process conditions are abnormal and output an alarm;

[0129] The dripping system includes a set value operating system, which is used to change the set value according to the operator's operation;

[0130] The dripping system includes a set value operating system, which is used to change the set value according to the operator's operation;

[0131] The PCA model is established using characteristic indices obtained by analyzing droplet width data under normal dripping conditions, and is verified using characteristic indices obtained by analyzing droplet width data under simulated abnormal process conditions.

[0132] The dripping process is the process in which the mixed liquid drops from the dripper into the condensate in the dripping pill preparation process; the method for processing the information includes: droplet node selection and segmentation, and characteristic index calculation;

[0133] The method for selecting and segmenting droplet nodes is as follows: the continuous droplet width values measured by the laser detector are divided into individual droplets. For each droplet, four nodes are taken and divided into three segments, where point A is the bottom of the droplet, point B is the point with the maximum width, point C is the point where the material begins to draw, and point D is the point where the liquid breaks.

[0134] The method for selecting point C is to traverse the data points backward from point B until the i-th point meets the conditions:

[0135]

[0136] Then the i-th point satisfies the selection of point C. In the formula, d i is the droplet width at point i, d i+N is the droplet width at N points after point i, and Δ is the judgment threshold.

[0137] The characteristic indicators include: the width at point B and point C, denoted as d B dC ; The length of AB segment and BC segment, that is, the number of data points contained between nodes, is recorded as len AB 、len BC The slope of segment AB and segment BC, that is, the ratio of the width at point B to the length of segment AB and segment BC, is expressed as The half-peak width of the AC segment, that is, the number of data points between the half-maximum width of the AB segment and the half-maximum width of the BC segment, is recorded as half_peak_width; the width of the AB segment and the BC segment at the three-division point is recorded as The drop rate, that is, the length between points A of adjacent droplets, is recorded as rate.

[0138] Example 10

[0139] A method for detecting faults in a pill making process, comprising the following steps:

[0140] (1) The dripping system performs the dripping process of the pills, and uses the detection system to detect the droplet information during the dripping process;

[0141] (2) The information of each droplet is processed and the characteristic index is calculated. The characteristic index is input into the PCA model to predict whether the process conditions are abnormal. If an abnormality occurs, an alarm is issued.

[0142] Example 11

[0143] A method for detecting faults in a pill making process, comprising the following steps:

[0144] (1) The dripping system performs the dripping process of the pills, and uses the detection system to detect the droplet information during the dripping process;

[0145] (2) The information of each droplet is processed, and characteristic indicators are calculated and input into the PCA model to predict whether the process conditions are abnormal. If an abnormality occurs, an alarm is issued; the material temperature in the dripping system is adjustable; the material input flow rate is adjustable; the material liquid level is adjustable; the detection system is a laser detection system; the dripping process is the process in which the mixed liquid drips from the dripper into the condensate in the dripping pill preparation process; the droplet information is the droplet width.

[0146] Example 12

[0147] A method for detecting faults in a pill making process, comprising the following steps:

[0148] (1) The dripping system performs the dripping process of the pills, and uses the detection system to detect the droplet information during the dripping process;

[0149] (2) Processing the information of each droplet, calculating the characteristic index, inputting it into the PCA model to predict whether the process conditions will produce abnormalities, and issuing an alarm if abnormalities occur;

[0150] The material temperature in the dripping system is adjustable; the material input flow rate is adjustable; the material liquid level is adjustable; the detection system is a laser detection system; the dripping process is the process in which the mixed liquid drips from the dripper into the condensate in the dripping pill preparation process; the droplet information is the droplet width;

[0151] The temperature of the material in the dripping system is controlled by adjusting the heating power to a set value; the laser detection system is composed of a voltage input module, a laser transmitter, a laser receiver, a sensor amplifier, a data acquisition card, and a data acquisition computer;

[0152] The laser detection system detects the width of droplets passing through the laser detector in real time during the dripping process and transmits it to the data acquisition computer; the method for extracting droplet information includes: droplet node selection and segmentation, and characteristic index calculation; the method for selecting and segmenting droplet nodes is: dividing the continuous droplet width values measured by the laser detector into single droplets, for each droplet, taking four nodes and dividing them into three segments, point A is the bottom of the droplet, point B is the point with the maximum width, point C is the point where the material starts to draw, and point D is the point where the liquid breaks.

[0153] The method for selecting point C is to traverse the data points backward from point B until the i-th point meets the conditions:

[0154]

[0155] Then the i-th point satisfies the selection of point C. In the formula, d i is the droplet width at point i, d i+N is the droplet width at N points after point i, and Δ is the judgment threshold. After optimization, it is determined that N = 15 and Δ = 0.05;

[0156] The characteristic indicators include: the width at point B and point C, denoted as d B d C ; The length of AB segment and BC segment, that is, the number of data points contained between nodes, is recorded as len AB 、len BC The slope of segment AB and segment BC, that is, the ratio of the width at point B to the length of segment AB and segment BC, is expressed as The half-peak width of the AC segment, that is, the number of data points between the half-maximum width of the AB segment and the half-maximum width of the BC segment, is recorded as half_peak_width; the width of the AB segment and the BC segment at the three-division point is recorded as The drop rate, that is, the length between points A of adjacent droplets, is recorded as rate.

[0157] The above detailed description is a specific description of one feasible embodiment of the present invention. This embodiment is not intended to limit the patent scope of the present invention. Any equivalent implementation or modification that does not depart from the present invention should be included in the scope of the technical solution of the present invention.

Claims

1. A fault detection system for a dripping pill production process, characterized by: include (1) A detection system, wherein the detection system is used to detect droplet information during the dripping process; (2) a processing system, wherein the processing system is used to process the detected droplet information; (3) an alarm system, wherein the alarm system is used to output an alarm when the information of the droplet evaluated by the model system is abnormal; The detection system is used in conjunction with a dripping system, which is a system that drips materials into dripping pills; The processing system is to use the information analysis of the droplets generated under normal dripping conditions to establish a droplet feature model, the droplet information data generated during the dripping production process is processed and the same characteristic indicators are obtained; the detection system is a laser detection system; The dripping process is the process in which the mixed liquid drips from the dripper into the condensate in the dripping pill preparation process; Methods for processing information include: droplet node selection and segmentation, characteristic index calculation; The method for selecting and segmenting droplet nodes is as follows: the continuous droplet width values measured by the laser detector are divided into individual droplets. For each droplet, four nodes are taken and divided into three segments: point A is the bottom of the droplet, point B is the point of maximum width, point C is the point where the material starts to draw, and point D is the point where the liquid breaks. The method for selecting point C is to traverse the data points from point B backward until the Points meet the conditions: ； Rule No. The point that satisfies the condition is selected as point C; where, For the The droplet width at point For the After The droplet width at a point, is the judgment threshold; after optimization, determine ; The characteristic indicators include: the width at point B and point C, recorded as 、 ; The length of AB segment and BC segment, that is, the number of data points contained between nodes, is recorded as 、 The slope of segment AB and segment BC, that is, the ratio of the width at point B to the length of segment AB and segment BC, is expressed as 、 The half-maximum width of the AC segment, that is, the number of data points between the half-maximum width of the AB segment and the half-maximum width of the BC segment, is recorded as ; The width of the AB segment and the BC segment at the point where they are divided into three equal parts is recorded as 、 、 、 ; Dropping speed, that is, the length between points A of adjacent droplets, is recorded as .

2. The fault detection system for the dripping pill production process according to claim 1, characterized in that: The information of the droplet is the droplet width; The laser detection system includes a voltage input module, a laser transmitter, a laser receiver, a sensor amplifier, a data acquisition card, and a data acquisition computer; The processing system uses characteristic indicators under normal dripping conditions to establish a PCA model; Simulate various abnormal dripping conditions and input the characteristic indicators calculated under these abnormal conditions into the PCA model for prediction to verify the fault detection performance of the model. Process the droplet width data generated by the current dripping production process and obtain the same characteristic indicators. The alarm system inputs the current droplet characteristic index into the PCA model to evaluate whether the process conditions are abnormal and output an alarm; The dripping system includes a set value operating system, and the set value operating system is used to change the set value according to the operator's operation.

3. The fault detection system for the dripping pill production process according to claim 2, characterized in that: The dripping system includes a set value operating system, which is used to change the set value according to the operator's operation; The PCA model is established using characteristic indices obtained by analyzing width data of droplets produced under normal dripping conditions, and is verified using characteristic indices obtained by analyzing width data of droplets produced under simulated abnormal process conditions.

4. A method for detecting faults in a dripping pill production process, characterized in that: A fault detection system for a pill making process as described in any one of claims 1 to 3 is used.

5. The method for detecting faults in the dripping pill production process according to claim 4, wherein: The information of the droplet is the droplet width; The laser detection system includes a voltage input module, a laser transmitter, a laser receiver, a sensor amplifier, a data acquisition card, and a data acquisition computer; The processing system uses characteristic indicators under normal dripping conditions to establish a PCA model; Simulate various abnormal dripping conditions and input the characteristic indicators calculated under these abnormal conditions into the PCA model for prediction to verify the fault detection performance of the model. Process the droplet width data generated by the current dripping production process and obtain the same characteristic indicators. The alarm system inputs the current droplet characteristic index into the PCA model to evaluate whether the process conditions are abnormal and output an alarm; The dripping system includes a set value operating system, and the set value operating system is used to change the set value according to the operator's operation.

6. The method for detecting faults in the dripping pill production process according to claim 5, wherein: The dripping system includes a set value operating system, which is used to change the set value according to the operator's operation; The PCA model is established using characteristic indices obtained by analyzing width data of droplets produced under normal dripping conditions, and is verified using characteristic indices obtained by analyzing width data of droplets produced under simulated abnormal process conditions.

Images