Cigarette production quality fluctuation detection method

Abstract

The invention discloses a method for detecting cigarette production quality fluctuation in the technical field of cigarette production detection, and aims to solve the problem that whether the cigarette production quality fluctuation is abnormal or not is difficult to detect and judge due to the lack of a method for detecting the cigarette production quality fluctuation in the prior art. The method comprises the following steps: obtaining index data, and dividing the index data to obtain first divided data; acquiring historical data, and calculating a first adaptive threshold and a second adaptive threshold according to the historical data; obtaining first comparison data of the first division data under the same time and different production lines, comparing the first comparison data with a first adaptive threshold value, and obtaining a first judgment result according to a comparison result; determining a first analysis result according to the first judgment result and the second judgment result; the method detects and judges whether the fluctuation of the cigarette production quality is abnormal or not, and ensures the uniformity of the cigarette production quality.

Description

Technical Field

[0001] The invention relates to a method for detecting fluctuations in cigarette production quality, belonging to the technical field of cigarette production detection. Background Art

[0002] At present, in cigarette production, "homogenization" is no longer satisfied with the management of qualified products, but focuses on fluctuations. Under a larger monitoring range and higher standards, the product quality is ultimately guaranteed to be uniform throughout the entire manufacturing cycle through consistent process conditions, stable equipment performance, and consistent processing intensity between production lines. To achieve the above goals, homogenization management has entered the fluctuation management stage with more stringent standards, requiring full coverage and continuous monitoring of the environment, process, equipment, materials, and teams involved in the silk production process to eliminate the inherent quality inconsistencies caused by fluctuations, especially the prevention of defective products, and to locate the root causes.

[0003] The prior art lacks a method for detecting fluctuations in cigarette production quality, and it is difficult to detect and determine whether the fluctuations in cigarette production quality are abnormal, thereby affecting the uniformity of cigarette production quality. Summary of the invention

[0004] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a method for detecting fluctuations in cigarette production quality, so as to solve the problem that the prior art lacks a method for detecting fluctuations in cigarette production quality, makes it difficult to detect and determine whether fluctuations in cigarette production quality are abnormal, and affects the uniformity of cigarette production quality.

[0005] In order to solve the above technical problems, the present invention is implemented by adopting the following technical solutions: The present invention provides a method for detecting fluctuations in cigarette production quality, comprising the following steps: Obtaining indicator data, dividing the indicator data, and obtaining first divided data; Acquire historical data, and calculate a first adaptive threshold and a second adaptive threshold according to the historical data; Obtain first comparison data of the first divided data at the same time and under different production lines, compare the first comparison data with the first adaptive threshold, and obtain a first judgment result according to the comparison result; use a normal distribution test method and a multi-peak test method to test the first comparison data to obtain a second judgment result; determine a first analysis result according to the first judgment result and the second judgment result; Obtain second comparison data of the first divided data in the same production line and at different times, compare the second comparison data with the second adaptive threshold, and obtain a third judgment result according to the comparison result; use a normal distribution test method and a multi-peak test method to test the second comparison data to obtain a fourth judgment result; determine the second analysis result according to the third judgment result and the fourth judgment result; Acquire the batch data according to the indicator data, divide the batch data according to time, obtain the second divided data, perform time series data anomaly detection on the second divided data based on the eight SPC abnormality judgment criteria, and determine the third analysis result; Based on the first analysis result, the second analysis result and the third analysis result, it is determined whether the fluctuation of cigarette production quality is abnormal.

[0006] Furthermore, the step of dividing the indicator data to obtain first divided data specifically includes: The index data is divided according to the preset material brand, process BOM and material BOM to obtain first divided data.

[0007] Furthermore, the first comparison data and the second comparison data both include a minimum value, a maximum value, a mean value, a standard deviation, and a Wasserstein distance.

[0008] Further, the calculating the first adaptive threshold according to the historical data specifically includes: Divide historical data according to preset material grades, process BOMs, and material BOMs to obtain historical divided data. Divide historical divided data based on the same time and different production lines to obtain historical data of line A, line B, and line C. Performing statistical analysis on the historical data of line A, the historical data of line B, and the historical data of line C to obtain third comparative data of the historical data of line A, the historical data of line B, and the historical data of line C; the third comparative data includes a minimum value, a maximum value, a mean value, a standard deviation, and a Wasserstein distance; The continuous variable formula is used to calculate the Wasserstein distance between the historical data of line A and line B, the Wasserstein distance between the historical data of line B and line C, and the Wasserstein distance between the historical data of line A and line C; Calculate the relative difference between the historical data of line A and the historical data of line B, the relative difference between the historical data of line B and the historical data of line C, and the relative difference between the historical data of line A and the historical data of line C according to the third comparison data of each of the historical data of line A, the historical data of line B, and the historical data of line C; The relative difference between the historical data of line A and the historical data of line B is calculated by the following formula: Relative difference %=abs(historical data of line A-historical data of line B) / max(historical data of line A, historical data of line B) For the minimum value, maximum value, mean, standard deviation and Wasserstein distance, all relative differences between the historical data of line A, the historical data of line B and the historical data of line C are counted to obtain the first historical difference, and the preset quantile is used as the threshold of the first historical difference to obtain the first adaptive threshold.

[0009] Further, comparing the first comparison data with the first adaptive threshold and obtaining a first judgment result according to the comparison result specifically includes: The first divided data is divided according to the same time and different production lines to obtain line A data, line B data and line C data; Perform statistical analysis on the A-line data, the B-line data, and the C-line data to obtain first comparative data of the A-line data, the B-line data, and the C-line data; The continuous variable formula was used to calculate the Wasserstein distance between A-line data and B-line data, the Wasserstein distance between B-line data and C-line data, and the Wasserstein distance between A-line data and C-line data; Calculating the relative difference between the A-line data and the B-line data, the relative difference between the B-line data and the C-line data, and the relative difference between the A-line data and the C-line data according to the first comparison data of each of the A-line data, the B-line data, and the C-line data; For the minimum value, maximum value, mean, standard deviation and Wasserstein distance, the relative difference between the A-line data and the B-line data, the relative difference between the B-line data and the C-line data, and the relative difference between the A-line data and the C-line data are respectively compared with the first adaptive threshold. If one of the relative differences is greater than the first adaptive threshold, the first judgment result is abnormal; if none of the relative differences is greater than the first adaptive threshold, the first judgment result is normal.

[0010] Further, the obtaining of the second judgment result, and determining the first analysis result according to the first judgment result and the second judgment result, specifically includes: The first comparison data is tested using a normal distribution test method and a multi-peak test method. If the normal distribution test confidence is less than a preset value or the multi-peak test finds multiple peaks, the second judgment result is abnormal; otherwise, the second judgment result is normal; If both the first judgment result and the second judgment result are normal, then the first analysis result is normal; Otherwise, the first analysis result is abnormal.

[0011] Further, determining the second analysis result according to the third judgment result and the fourth judgment result specifically includes: The relative differences between the second divided data are compared with the second adaptive threshold respectively. If one of the relative differences is greater than the second adaptive threshold, the third judgment result is abnormal; if none of the relative differences is greater than the second adaptive threshold, the third judgment result is normal; The second comparison data is tested using a normal distribution test method and a multi-peak test method. If the normal distribution test confidence is less than a preset value or the multi-peak test finds multiple peaks, the fourth judgment result is abnormal; otherwise, the fourth judgment result is normal; If the third judgment result and the fourth judgment result are both normal, then the second analysis result is normal; Otherwise, the second analysis result is abnormal.

[0012] Furthermore, dividing the data in the batch according to time specifically includes: The data within the batch is divided according to the preset production line, batch number, start time and end time to obtain second divided data.

[0013] Further, the determining of the third analysis result specifically includes: When the second partitioned data meets the eight SPC abnormality criteria, the third analysis result is abnormal; When the second divided data does not show any of the eight SPC abnormality criteria, the third analysis result is normal.

[0014] Furthermore, the determination of whether the fluctuation in cigarette production quality is abnormal specifically includes: If the first analysis result, the second analysis result and the third analysis result are all normal, it is determined that there is no abnormality in the fluctuation of cigarette production quality; Otherwise, it is judged that the fluctuation in cigarette production quality is abnormal.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The method for detecting fluctuations in cigarette production quality divides the index data to obtain first comparison data at the same time and on different production lines and second comparison data at the same production line and at different times, and compares the two with the first adaptive threshold and the second adaptive threshold respectively, so as to detect and judge whether the fluctuations in cigarette production quality are abnormal. At the same time, the data within the batch is divided according to time to obtain the second divided data, and the second divided data is subjected to time series data anomaly detection based on the eight SPC abnormality judgment criteria, so as to further judge whether the fluctuations in cigarette production quality within the same batch are abnormal, thereby ensuring the uniformity of cigarette production quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1The figure is a flow chart of a method for detecting fluctuations in cigarette production quality provided according to an embodiment of the present invention. DETAILED DESCRIPTION

[0017] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations on the technical solution of the present invention. The embodiments of the present invention and the technical features in the embodiments may be combined with each other unless there is a conflict.

[0018] The term "and / or" is only a description of the association relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " generally indicates that the related objects are in an "or" relationship.

[0019] like Figure 1 As shown, the present invention provides a method for detecting fluctuations in cigarette production quality, comprising the following steps: Obtaining index data, dividing the index data, and obtaining first divided data; wherein the index data is an operating parameter of the equipment: including loose outlet moisture, loose outlet temperature, etc.; Acquire historical data, and calculate a first adaptive threshold and a second adaptive threshold according to the historical data; Obtain first comparative data of the first divided data at the same time and under different production lines, compare the first comparative data with the first adaptive threshold, and obtain a first judgment result according to the comparison result; use a normal distribution test method and a multi-peak test method to test the first comparative data and obtain a second judgment result; determine a first analysis result according to the first judgment result and the second judgment result; optionally, the production line includes three groups, namely, line A, line B and line C, and the first divided data is divided into three groups of A, B and C, and the data containing letters A, B and C based on the batch number is divided into line A data, line B data and line C data; Obtain second comparison data of the first divided data in the same production line and at different times, compare the second comparison data with the second adaptive threshold, and obtain a third judgment result according to the comparison result; use a normal distribution test method and a multi-peak test method to test the second comparison data to obtain a fourth judgment result; determine the second analysis result according to the third judgment result and the fourth judgment result; Acquire the batch data according to the indicator data, divide the batch data according to time, obtain the second divided data, perform time series data anomaly detection on the second divided data based on the eight SPC abnormality judgment criteria, and determine the third analysis result; Based on the first analysis result, the second analysis result and the third analysis result, it is determined whether the fluctuation of cigarette production quality is abnormal.

[0020] In one embodiment, the step of dividing the indicator data to obtain first divided data specifically includes: The index data is divided according to the preset material brand, process BOM and material BOM to obtain first divided data.

[0021] Optional: Since different material grades have different standard values ​​for the upper and lower limits of production indicators, it is necessary to perform analysis under the same grade when performing fluctuation analysis. After communication and confirmation, they are divided into: Liqun (soft long mouth), Liqun (long mouth A), Liqun (soft red A), Liqun (new version A), and Liqun (sunshine); the process BOM version is divided into leaf making BOM and silk making BOM. It is necessary to filter the data of the corresponding BOM version for the process segment where the indicator is located. The specific rules are: the leaf making indicator filters the BOM version containing the data of B0 (leaf making process segment), and the silk making indicator filters the BOM version containing the data of S0 (silk making process segment); the data is divided according to the change points of the three dimensions of material grade, process BOM and material BOM. Any change in any dimension is regarded as a change point. If the amount of divided data is less than 30, the data is padded to the previous change point to ensure that the data amount is sufficient for fluctuation analysis, so that the data analyzed each time is based on the same production environment as much as possible; if the analysis data changes across production environments, a prompt will be given in the analysis results.

[0022] In one embodiment, after statistical analysis and calculation of line data of A, B, and C, the difference between the lines can be obtained. In order to answer the question of "how much difference is inconsistent", it is necessary to find a suitable threshold. When the difference exceeds the threshold, it is determined to be inconsistent; the comparison data that need to be calculated by threshold include: minimum value, maximum value, mean, standard deviation and Wasserstein distance, among which all comparison data (i.e., statistical indicators) use the "average value" field of batch data, and are obtained by statistical analysis of a group of data (average value) after data division.

[0023] The calculating of the first adaptive threshold according to the historical data specifically includes: Divide historical data according to preset material grades, process BOMs, and material BOMs to obtain historical divided data. Divide historical divided data based on the same time and different production lines to obtain historical data of line A, line B, and line C. Performing statistical analysis on the historical data of line A, the historical data of line B, and the historical data of line C to obtain third comparative data of the historical data of line A, the historical data of line B, and the historical data of line C; the third comparative data includes a minimum value, a maximum value, a mean value, a standard deviation, and a Wasserstein distance; The continuous variable formula is used to calculate the Wasserstein distance between the historical data of line A and line B, the Wasserstein distance between the historical data of line B and line C, and the Wasserstein distance between the historical data of line A and line C; Calculate the relative difference between the historical data of line A and the historical data of line B, the relative difference between the historical data of line B and the historical data of line C, and the relative difference between the historical data of line A and the historical data of line C according to the third comparison data of each of the historical data of line A, the historical data of line B, and the historical data of line C; The relative difference between the historical data of line A and the historical data of line B is calculated by the following formula: Relative difference %=abs(historical data of line A-historical data of line B) / max(historical data of line A, historical data of line B) For the minimum value, maximum value, mean, standard deviation and Wasserstein distance, all relative differences between the historical data of line A, the historical data of line B and the historical data of line C are counted to obtain the first historical difference, and the preset quantile is used as the threshold of the first historical difference to obtain the first adaptive threshold.

[0024] Among them, for the minimum value, maximum value, mean, standard deviation and Wasserstein distance, all their relative differences are counted to obtain the first historical difference, and the 80% quantile is taken as the threshold, that is, 80% of the differences in history are lower than the threshold, and 20% of the differences are higher than the threshold; the quantile threshold algorithm is a commonly used statistical method for identifying outliers in a data set; it is based on the quantile of the data and determines the threshold of the outlier by dividing the data into multiple intervals; in statistics, the quantile is the value that divides the data set into several equal parts, for example: the median is the value that divides the data into two parts, which is also the 50% quantile; the 20%, 40%, 60%, and 80% quantiles divide the data into five equal parts, and we use the 80% quantile as the threshold for the first historical difference.

[0025] In one embodiment, comparing the first comparison data with the first adaptive threshold and obtaining the first judgment result according to the comparison result specifically includes: The first divided data is divided according to the same time and different production lines to obtain line A data, line B data and line C data; Perform statistical analysis on the A-line data, the B-line data, and the C-line data to obtain first comparative data of the A-line data, the B-line data, and the C-line data; The continuous variable formula was used to calculate the Wasserstein distance between A-line data and B-line data, the Wasserstein distance between B-line data and C-line data, and the Wasserstein distance between A-line data and C-line data; Calculating the relative difference between the A-line data and the B-line data, the relative difference between the B-line data and the C-line data, and the relative difference between the A-line data and the C-line data according to the first comparison data of each of the A-line data, the B-line data, and the C-line data; For the minimum value, maximum value, mean, standard deviation and Wasserstein distance, the relative difference between the A-line data and the B-line data, the relative difference between the B-line data and the C-line data, and the relative difference between the A-line data and the C-line data are respectively compared with the first adaptive threshold. If one of the relative differences is greater than the first adaptive threshold, the first judgment result is abnormal; if none of the relative differences is greater than the first adaptive threshold, the first judgment result is normal.

[0026] Specifically, for the first comparison data of the first divided data at the same time and under different production lines; Perform statistical analysis on the A-line data, the B-line data and the C-line data, obtain the first comparison data of the A-line data, the B-line data and the C-line data, perform statistical analysis on the A-line data, and calculate the maximum value, minimum value, mean value and standard deviation respectively. The sample is as follows;

[0027] According to the first comparison data of the A-line data, the B-line data and the C-line data, the relative difference between the A-line data and the B-line data, and the relative difference between the A-line data and the C-line data are calculated. Taking the relative difference between the A-line data and the B-line data as an example, the relative difference is calculated by the following formula: Relative difference %=abs(A line data-B line data) / max(A line data, B line data) Based on the first adaptive threshold, the maximum value, minimum value, mean value, standard deviation, and Wasserstein distance are compared with the first adaptive threshold. If they are greater than the first adaptive threshold, the first judgment result is abnormal:

[0028] Optionally, the first comparison data distribution may be tested by a normal distribution test method and a multi-peak test method. When the confidence of the normal distribution test is less than 5%, it is determined that the distribution is abnormal. When the multi-peak test finds multiple peaks, it is determined that the distribution is abnormal. Multi-peak analysis algorithm steps: Perform histogram statistics on the data in 10 segments, find the largest maximum value in the 5 neighborhoods, which is the peak value; the output chart results are as follows:

[0029] In the table, N test is the confidence value of the normal distribution test.

[0030] In one embodiment, obtaining the second judgment result and determining the first analysis result according to the first judgment result and the second judgment result specifically includes: The first comparison data is tested using a normal distribution test method and a multi-peak test method. If the normal distribution test confidence is less than a preset value or the multi-peak test finds multiple peaks, the second judgment result is abnormal; otherwise, the second judgment result is normal; If both the first judgment result and the second judgment result are normal, then the first analysis result is normal; Otherwise, the first analysis result is abnormal.

[0031] In one embodiment, determining the second analysis result according to the third judgment result and the fourth judgment result specifically includes: The relative differences between the second divided data are compared with the second adaptive threshold respectively. If one of the relative differences is greater than the second adaptive threshold, the third judgment result is abnormal; if none of the relative differences is greater than the second adaptive threshold, the third judgment result is normal; The second comparison data is tested using a normal distribution test method and a multi-peak test method. If the normal distribution test confidence is less than a preset value or the multi-peak test finds multiple peaks, the fourth judgment result is abnormal; otherwise, the fourth judgment result is normal; If the third judgment result and the fourth judgment result are both normal, then the second analysis result is normal; Otherwise, the second analysis result is abnormal.

[0032] Specifically, in order to verify whether the indicator does not change with time, the ABC line data of the indicator is pulled for the past year and divided into four parts by quarter to obtain one-segment data, two-segment data, three-segment data and four-segment data. A similar method of comparing line differences is used to calculate the relative differences between one-segment data, two-segment data, three-segment data and four-segment data, and each relative difference is compared with the second adaptive threshold to obtain the third judgment result, and the seasonal data is subjected to normal distribution test and multi-peak test to verify whether the data in the four quarters are consistent and the distribution is stable.

[0033] The calculating of the second adaptive threshold according to the historical data specifically includes: The first historical data is divided based on the same production line and different time to obtain one historical data segment, two historical data segments, three historical data segments and four historical data segments; wherein the production line is the same as the production line of the second comparison data; Performing statistical analysis on one historical data segment, two historical data segments, three historical data segments, and four historical data segments to obtain fourth comparative data of each of the one historical data segment, two historical data segments, three historical data segments, and four historical data segments; the fourth comparative data includes a minimum value, a maximum value, a mean value, a standard deviation, and a Wasserstein distance; The continuous variable formula is used to calculate the Wasserstein distance between one historical data segment and the second historical data segment, the Wasserstein distance between one historical data segment and the third historical data segment, the Wasserstein distance between one historical data segment and the fourth historical data segment, the Wasserstein distance between two historical data segments and the third historical data segment, the Wasserstein distance between two historical data segments and the fourth historical data segment, and the Wasserstein distance between three historical data segments and the fourth historical data segment; Calculate, based on the fourth comparison data of each of the first historical data, the second historical data, the third historical data, and the fourth historical data, the relative difference between the first historical data and the second historical data, the relative difference between the first historical data and the third historical data, the relative difference between the first historical data and the fourth historical data, the relative difference between the second historical data and the third historical data, the relative difference between the second historical data and the fourth historical data, and the relative difference between the third historical data and the fourth historical data; Taking the relative difference between the first segment of historical data and the second segment of historical data as an example, it is calculated by the following formula: Relative difference % = abs (one historical data - two historical data) / max (one historical data, two historical data) For the minimum value, maximum value, mean, standard deviation and Wasserstein distance, all relative differences between one historical data segment, two historical data segments, three historical data segments and four historical data segments are counted to obtain the second historical difference, and the preset quantile is used as the threshold of the second historical difference to obtain the second adaptive threshold.

[0034] In one embodiment, dividing the data in a batch according to time specifically includes: The data within the batch is divided according to the preset production line, batch number, start time and end time to obtain second divided data.

[0035] Determining the third analysis result specifically includes: When the second partitioned data meets the eight SPC abnormality criteria, the third analysis result is abnormal; When the second divided data does not show any of the eight SPC abnormality criteria, the third analysis result is normal.

[0036] Specifically, data division for intra-batch fluctuation detection: Since the time series data does not contain batch information, only the data records collected by the sensors at each time point are available, and data fluctuation analysis cannot be performed directly, data division processing is required. The process is as follows: Read the batch number, start time, end time, and production line (A, B, C) of the batch from the batch data table Filter the time series data within the time period from the start time to the end time of the batch based on the indicator name and the production line locked data source.

[0037] In one embodiment, the determining whether the cigarette production quality fluctuation is abnormal specifically includes: If the first analysis result, the second analysis result and the third analysis result are all normal, it is determined that there is no abnormality in the fluctuation of cigarette production quality; Otherwise, it is judged that the fluctuation in cigarette production quality is abnormal.

[0038] The present invention divides the index data to obtain first comparison data at the same time and different production lines and second comparison data at the same production line and different times, and compares the two with the first adaptive threshold and the second adaptive threshold respectively, so as to detect and judge whether the fluctuation of cigarette production quality is abnormal. At the same time, the data within the batch is divided according to time to obtain the second divided data, and the second divided data is detected for time series data anomaly based on the eight SPC abnormality judgment criteria, so as to further judge whether the fluctuation of cigarette production quality in the same batch is abnormal, thereby ensuring the uniformity of cigarette production quality.

[0039] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0040] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0041] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0042] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0043] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for detecting fluctuations in cigarette production quality, characterized in that: The following steps are involved: Obtaining indicator data, dividing the indicator data, and obtaining first divided data; Acquire historical data, and calculate a first adaptive threshold and a second adaptive threshold according to the historical data; Obtain first comparison data of the first divided data at the same time and under different production lines, compare the first comparison data with a first adaptive threshold, and obtain a first judgment result according to the comparison result; Using a normal distribution test method and a multi-peak test method to test the first comparison data and obtain a second judgment result; determining the first analysis result according to the first judgment result and the second judgment result; Obtain second comparison data of the first divided data in the same production line and at different times, compare the second comparison data with the second adaptive threshold, and obtain a third judgment result according to the comparison result; use a normal distribution test method and a multi-peak test method to test the second comparison data to obtain a fourth judgment result; determine the second analysis result according to the third judgment result and the fourth judgment result; Acquire the batch data according to the indicator data, divide the batch data according to time, obtain the second divided data, perform time series data anomaly detection on the second divided data based on the eight SPC abnormality judgment criteria, and determine the third analysis result; Based on the first analysis result, the second analysis result and the third analysis result, it is determined whether the fluctuation of cigarette production quality is abnormal.

2. The method for detecting fluctuations in cigarette production quality according to claim 1, characterized in that: The dividing the indicator data to obtain first divided data specifically includes: The index data is divided according to the preset material brand, process BOM and material BOM to obtain first divided data.

3. The method for detecting fluctuations in cigarette production quality according to claim 1, characterized in that: The first comparison data and the second comparison data both include a minimum value, a maximum value, a mean value, a standard deviation, and a Wasserstein distance.

4. The method for detecting fluctuations in cigarette production quality according to claim 1, characterized in that: The calculating of the first adaptive threshold according to the historical data specifically includes: Divide historical data according to preset material grades, process BOMs, and material BOMs to obtain historical divided data. Divide historical divided data based on the same time and different production lines to obtain historical data of line A, line B, and line C. Performing statistical analysis on the historical data of line A, the historical data of line B, and the historical data of line C to obtain third comparative data of the historical data of line A, the historical data of line B, and the historical data of line C; the third comparative data includes a minimum value, a maximum value, a mean value, a standard deviation, and a Wasserstein distance; The continuous variable formula is used to calculate the Wasserstein distance between the historical data of line A and line B, the Wasserstein distance between the historical data of line B and line C, and the Wasserstein distance between the historical data of line A and line C; Calculate the relative difference between the historical data of line A and the historical data of line B, the relative difference between the historical data of line B and the historical data of line C, and the relative difference between the historical data of line A and the historical data of line C according to the third comparison data of each of the historical data of line A, the historical data of line B, and the historical data of line C; The relative difference between the historical data of line A and the historical data of line B is calculated by the following formula: Relative difference %=abs(historical data of line A-historical data of line B) / max(historical data of line A, historical data of line B) For the minimum value, maximum value, mean, standard deviation and Wasserstein distance, all relative differences between the historical data of line A, the historical data of line B and the historical data of line C are counted to obtain the first historical difference, and the preset quantile is used as the threshold of the first historical difference to obtain the first adaptive threshold.

5. The method for detecting fluctuations in cigarette production quality according to claim 3, characterized in that: The comparing the first comparison data with the first adaptive threshold value and obtaining the first judgment result according to the comparison result specifically includes: The first divided data is divided according to the same time and different production lines to obtain line A data, line B data and line C data; Perform statistical analysis on the A-line data, the B-line data, and the C-line data to obtain first comparative data of the A-line data, the B-line data, and the C-line data; The continuous variable formula was used to calculate the Wasserstein distance between A-line data and B-line data, the Wasserstein distance between B-line data and C-line data, and the Wasserstein distance between A-line data and C-line data; Calculating the relative difference between the A-line data and the B-line data, the relative difference between the B-line data and the C-line data, and the relative difference between the A-line data and the C-line data according to the first comparison data of each of the A-line data, the B-line data, and the C-line data; For the minimum value, maximum value, mean, standard deviation and Wasserstein distance, the relative difference between the A-line data and the B-line data, the relative difference between the B-line data and the C-line data, and the relative difference between the A-line data and the C-line data are respectively compared with the first adaptive threshold. If one of the relative differences is greater than the first adaptive threshold, the first judgment result is abnormal; if none of the relative differences is greater than the first adaptive threshold, the first judgment result is normal.

6. The method for detecting fluctuations in cigarette production quality according to claim 5, characterized in that: The obtaining of the second judgment result and determining the first analysis result according to the first judgment result and the second judgment result specifically includes: The first comparison data is tested using a normal distribution test method and a multi-peak test method. If the normal distribution test confidence is less than a preset value or the multi-peak test finds multiple peaks, the second judgment result is abnormal; otherwise, the second judgment result is normal; If both the first judgment result and the second judgment result are normal, then the first analysis result is normal; Otherwise, the first analysis result is abnormal.

7. The method for detecting fluctuations in cigarette production quality according to claim 6, characterized in that: The determining of the second analysis result according to the third judgment result and the fourth judgment result specifically includes: The relative differences between the second divided data are compared with the second adaptive threshold respectively. If one of the relative differences is greater than the second adaptive threshold, the third judgment result is abnormal; if none of the relative differences is greater than the second adaptive threshold, the third judgment result is normal; The second comparison data is tested using a normal distribution test method and a multi-peak test method. If the normal distribution test confidence is less than a preset value or the multi-peak test finds multiple peaks, the fourth judgment result is abnormal; otherwise, the fourth judgment result is normal; If the third judgment result and the fourth judgment result are both normal, then the second analysis result is normal; Otherwise, the second analysis result is abnormal.

8. The method for detecting fluctuations in cigarette production quality according to claim 1, characterized in that: The division of the data in the batch according to time specifically includes: The data within the batch is divided according to the preset production line, batch number, start time and end time to obtain second divided data.

9. The method for detecting fluctuations in cigarette production quality according to claim 7, characterized in that: Determining the third analysis result specifically includes: When the second partitioned data meets the eight SPC abnormality criteria, the third analysis result is abnormal; When the second divided data does not show any of the eight SPC abnormality criteria, the third analysis result is normal.

10. The method for detecting fluctuations in cigarette production quality according to claim 9, characterized in that: The determination of whether the fluctuation in cigarette production quality is abnormal specifically includes: If the first analysis result, the second analysis result and the third analysis result are all normal, it is determined that there is no abnormality in the fluctuation of cigarette production quality; Otherwise, it is judged that the fluctuation in cigarette production quality is abnormal.

Images