Steady state identification method, system and equipment in roll package production process and medium

Abstract

The invention provides a steady state identification method, system and device in a roll package production process, and a medium. The method comprises the following steps: obtaining production process data of a roll package; wherein the production process data comprises production statistical indexes, equipment parameters, process parameters and equipment failure shutdown data; preprocessing the production process data to obtain multivariate time series data; based on the multivariate time sequence data, constructing and using a multivariate time sequence change point detection model, and identifying change points in the multivariate time sequence data; and based on the positions and the number of the change points, identifying the systematic steady state of the rolling and packaging production process, and outputting a systematic steady state identification result. According to the method, the positions and the number of the change points are determined in a self-adaptive manner by using the multivariate production process data of the roll and package, the limitation by priori knowledge is avoided, and the systematic steady state and the non-steady state of the roll and package production process can be accurately identified.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of quality evaluation of cigarette production process, and particularly relates to a steady state identification method, system, medium and equipment in a cigarette production process. BACKGROUND

[0002] Cigarette packaging is a key link in cigarette production, mainly responsible for cigarette rolling and packaging. The process requires the cooperation and processing of many raw materials such as tobacco, cigarette paper, filters and packaging materials. The production process mainly relies on key equipment such as cigarette machines and packaging machines, and has strict requirements for raw materials and equipment process parameters, so it has high production complexity. Systematic steady state identification of the cigarette packaging production process refers to judging whether the system formed by multiple indicators is in a state of continuous stable control under the influence of factors such as equipment failure, raw material quality difference and manual operation intervention. Systematic steady state identification is related to time series such as quality detection data, cost consumption data, production efficiency data, equipment operation data and equipment online data acquisition data. These time series have the characteristics of multiple dimensions, complex business structure and coupled causal relationship, which cause high difficulty in analysis and modeling.

[0003] Traditional steady state identification is mostly based on change point detection methods, but when facing the problem of systematic steady state identification, such methods cannot effectively process multi-element, complex and strongly related process data, resulting in low data utilization efficiency and poor detection performance. In addition, the traditional steady state identification method also needs to pre-determine the number of change points and assume the distribution type before and after the change, which strongly depends on prior knowledge. SUMMARY

[0004] In view of the above problems, the present application is proposed to provide a steady state identification method, system, equipment and medium in a cigarette packaging production process to overcome the above problems or at least partially solve the above problems.

[0005] To achieve the above object and other related objects, the present application provides a steady state identification method in a cigarette packaging production process, which comprises:

[0006] Obtaining production process data of the cigarette packaging; wherein the production process data comprises production statistical indicators, equipment parameters and process parameters and equipment failure downtime data;

[0007] Pretreating the production process data to obtain multi-element time series data;

[0008] Based on the multi-element time series data, a multi-element time series change point detection model is constructed and used to identify the change points in the multi-element time series data;

[0009] Based on the position and number of the change points, the systematic steady state of the cigarette packaging production process is analyzed, and the systematic steady state identification result is output.

[0010] In an embodiment of the present application, the pre-processing of the production process data to obtain multivariate time series data comprises:

[0011] Detecting missing values and outliers in the production process data to obtain missing conditions and abnormal conditions of the production process data;

[0012] According to the missing conditions and the abnormal conditions, the production process data is corrected to obtain multivariate time series data.

[0013] In an embodiment of the present application, the correction of the production process data according to the missing conditions and the abnormal conditions comprises:

[0014] According to the missing conditions, the missing values of the production process data are corrected by using a preset filling method;

[0015] According to the abnormal conditions, the abnormal values of the production process data are corrected by using a preset processing method.

[0016] In an embodiment of the present application, based on the multivariate time series data, a multivariate time series change point detection model is constructed and used to identify change points in the multivariate time series data, comprising:

[0017] According to the multivariate time series data, potential change points are identified;

[0018] According to a preset condition, the potential change points are screened to identify change points in the multivariate time series data.

[0019] In an embodiment of the present application, the identification of potential change points according to the multivariate time series data comprises:

[0020] According to a preset change point set, the multivariate time series data is segmented to obtain a first segment; wherein the preset change point set has no change points;

[0021] The E-divergence statistics of each time series data point in the first segment is calculated;

[0022] The maximum point of the E-divergence statistics in the first segment is selected, and the maximum point is set as a determined potential change point.

[0023] In an embodiment of the present application, the screening of the potential change points according to a preset condition to identify change points in the multivariate time series data comprises:

[0024] Using a preset permutation test method, the significance level of the potential change point is obtained;

[0025] if the significance level of the potential change point is greater than a preset level, adding the potential change point to the change point set and updating the first segment;

[0026] The step of calculating the E-divergence statistic of each time series data point in the first segment is repeatedly performed until the significance level of the potential change point is not greater than the preset level, and the change point set of the multivariate time series data is determined and output.

[0027] In an embodiment of the present application, the system stability of the cigarette production process is analyzed based on the position and number of the change points, including:

[0028] According to the position, a data segment subinterval of the multivariate time series data is determined;

[0029] According to the data segment subinterval and the number, the system stability of the cigarette production process is analyzed.

[0030] To achieve the above object and other related objects, the present application provides a system for identifying the stability of a cigarette production process, comprising:

[0031] An acquisition module is configured to acquire production process data of the cigarette, wherein the production process data includes production statistics, equipment parameters and process parameters, and equipment failure downtime data;

[0032] A processing module is configured to pre-process the production process data to obtain multivariate time series data;

[0033] A construction module is configured to construct and use a multivariate time series change point detection model based on the multivariate time series data, and identify the change points of the multivariate time series data;

[0034] An identification module is configured to analyze the system stability of the cigarette production process based on the position and number of the change points, and output the system stability identification result.

[0035] To achieve the above object and other related objects, the present application provides an electronic device, comprising a memory and a processor;

[0036] The memory is configured to store a computer program;

[0037] The processor is configured to execute the computer program stored in the memory, so that the electronic device performs the above-mentioned method for identifying the stability of a cigarette production process.

[0038] To achieve the above object and other related objects, the present application further provides a computer readable storage medium as described above, which stores a computer program, and the program is executed by a processor to realize the steady state identification method in the cigarette production process as described above.

[0039] As described above, the steady state identification method, system, device and medium in the cigarette production process of the present application, the method comprises: obtaining production process data of the cigarette; wherein the production process data comprises production statistical indicators, equipment parameters and process parameters, and equipment failure downtime data; pre-processing the production process data to obtain multivariate time series data; based on the multivariate time series data, constructing and using a multivariate time series change point detection model to identify the change points in the multivariate time series data; based on the position and number of the change points, identifying the systematic steady state of the cigarette production process, and outputting the systematic steady state identification result. The present application determines the change point position and number in an adaptive manner by using the multivariate production process data of the cigarette, and is no longer limited by prior knowledge, and can accurately identify the systematic steady state and non-steady state of the cigarette production process. BRIEF DESCRIPTION OF DRAWINGS

[0040] Figure 1 The flowchart shows the steady state identification method in the cigarette production process of an embodiment of the present application;

[0041] Figure 2 The schematic diagram shows the cigarette consumption change point detection result of a certain time period in an embodiment of the present application;

[0042] Figure 3 The schematic diagram shows the process section data change point detection result of a certain process in an embodiment of the present application;

[0043] Figure 4 The schematic diagram shows the multivariate time series data change point position calculation in an embodiment of the present application;

[0044] Figure 5 The schematic diagram shows the total flow of the multivariate change point generation in an embodiment of the present application;

[0045] Figure 6 The functional module schematic diagram of the steady state identification system in the cigarette production process of an embodiment of the present application is shown. DETAILED DESCRIPTION

[0046] Following make the embodiments of the present application more clear through specific, specific examples, the person skilled in the art can easily understand other advantages and effects of the present application from the disclosure of the specification. The present application can also be implemented or applied by other different specific embodiments, and various modifications or changes can be made based on different views and applications without departing from the spirit of the present application. It should be noted that the following examples and features in the examples can be combined with each other without conflict.

[0047] It should be noted that the diagrams provided in the following examples only illustrate the basic concept of the present application in a schematic manner, and only the components related to the present application are shown in the diagrams, not the number, shape and size of the components when actually implemented. The actual implementation of each component may be a random change, and the component layout pattern may be more complex.

[0048] In the following description, a large number of details are discussed to provide a more thorough explanation of the embodiments of the present application, however, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details, and in other embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail, to avoid making the embodiments of the present application difficult to understand.

[0049] The terms "first", "second", and the like in the description and claims of the present disclosure and the above-described figures are used to distinguish similar objects, and do not necessarily have to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present disclosure described herein can be implemented. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion.

[0050] Unless otherwise specified, the term "a plurality of" means two or more.

[0051] In the embodiments of the present disclosure, the character " / " represents a "or" relationship between the objects before and after it. For example, A / B means: A or B.

[0052] The term "and / or" is a description of the relationship between the objects, which means that there can be three relationships. For example, A and / or B means: A or B, or, A and B, the three relationships.

[0053] The technical solutions in the embodiments of the present application will be described in detail below with reference to the drawings in the embodiments of the present application.

[0054] Please refer to Figure 1 An embodiment of the present application provides a steady state recognition method in a roll packaging production process, which can include the following steps:

[0055] Step S10, obtaining production process data of the cigarette package; wherein the production process data comprises production statistical indicators, equipment parameters and process parameters, and equipment failure downtime data.

[0056] The production process data can be composed of several long time series, and there is a specific business relationship between each time series, which is manifested as correlation of the same processing level sequence or causal relationship between the processes before and after. The production process data can include, but is not limited to, production statistical indicators, equipment parameters and process parameters, and equipment failure downtime data in the cigarette package production process.

[0057] The production statistical indicators can be composed of three indicators of cigarette output, cigarette quality and cigarette consumption, which can macroscopically define the entire cigarette package production level.

[0058] The equipment parameters and process parameters can be indicators composed of equipment operation data, equipment online data acquisition data, process parameters, associated cigarette material measurement indicators, and associated rejection rates, which are used to evaluate the stability of a certain cigarette quality indicator control.

[0059] As an example, the equipment parameters and process parameters can include VE segment core process parameters, cigarette weight control level and stability indicators, and cigarette weight related rejection rates. The VE segment core process parameters, cigarette weight control level and stability indicators, and cigarette weight related rejection rates are closely related, and each indicator has a certain qualified fluctuation range, forming a system of cigarette package process control and result response.

[0060] The equipment failure downtime data can be data generated when the cigarette package equipment fails to stop, which can include, but is not limited to, the number of cigarette downtime, equipment failure downtime, equipment failure downtime rate, etc.

[0061] In addition, the embodiment does not limit the specific implementation of obtaining the production process data of the cigarette package.

[0062] In a specific implementation, the production process data of the cigarette package can include, but is not limited to, production statistical indicators, equipment parameters and process parameters, and equipment failure downtime data in the cigarette package production process; thereby, by taking into account multiple types of information such as yield, quality, loss and process parameters with business association, systematic steady-state changes are obtained; in this way, the analysis of the cigarette package production process can be more comprehensive and in-depth, and therefore, it is more accurate in identifying systematic steady state, and important change information is not easily missed.

[0063] Step S20, pre-processing the production process data to obtain multivariate time series data.

[0064] The multivariate time series data can be data columns recorded in time sequence by a plurality of time series samples.

[0065] As an example, the multivariate time series data can be a plurality of time series classified and arranged according to production machine information and the like for roll package production process data, and can be data columns including cigarette production, package production, cigarette downtime, cigarette rejection rate, aluminum rod consumption, and cigarette paper consumption, and can specifically be multivariate time series samples Z1, Z2, … Z T ∈R d T is a sample length, and d is a total dimension of the sample.

[0066] In a specific implementation, the production process data can be preprocessed by defect and anomaly detection, and then when defects and anomalies are detected, the production process data with defects and / or anomalies can be corrected, so that the multivariate time series data can be obtained.

[0067] In step S30, a multivariate time series change point detection model is constructed and used based on the multivariate time series data to identify change points in the multivariate time series data.

[0068] The multivariate time series change point detection model can be a model constructed from multivariate time series data for detecting whether there is a change point in the multivariate time series data.

[0069] Specifically, when detecting whether there is a change point in the multivariate time series data, the multivariate time series data is taken as initial input, and the change points in the multivariate time series data are output, which can include positions and quantities of the change points.

[0070] In a specific implementation, after obtaining the multivariate time series data, a potential change point in the multivariate time series data can be first identified according to the multivariate time series data; then a change point generation judgment can be performed to determine whether the potential change point is to be added to a change point set (at this time, the change point set is an empty change point set); specifically, if the potential change point is retained, the change point set is updated, and the next potential change point is identified; if the potential change point is discarded, the change point division is not performed, and an updated change point set is output; wherein the updated change point set is the change points identified in the multivariate time series data.

[0071] In step S40, a system stability of the roll package production process is analyzed based on the positions and quantities of the change points, and a system stability identification result is output.

[0072] The system stability can refer to a system composed of multiple indexes in the cigarette production process. The indexes in the system are closely related to a certain stable state, and are intended to ensure the homogeneity of the cigarette production quality. The system stability can be used to continuously improve the production process stability and quality according to the system stability change of the cigarette production process.

[0073] As an example, the system stability of the cigarette production process can be identified by judging whether there is a production system stability change caused by the production behaviors (indexes or data) such as raw material quality, process parameter adjustment, equipment failure, operation error, and the like.

[0074] The system stability identification result can be a result of identifying the system stability and non-stability in the cigarette production process. Specifically, multiple change points in the process can be detected by means of change point calculation and generation judgment. The data sub-intervals are output according to the change point positions, and the system stability identification result is divided based on the interval indexes.

[0075] As an example, if there are a large number of change points in a time period, it indicates that the system index fluctuation is high, and the stability control ability is poor.

[0076] In a specific implementation, after obtaining the change points in the multivariate time series data, the system stability of the cigarette production process can be judged according to the positions and quantities of the change points, and then the system stability identification result is output.

[0077] As an example, please refer to Figure 2 , Figure 2 which is a schematic diagram of the change point detection result of the cigarette production process in a time period. The cigarette production process data of a batch is selected, the cigarette paper consumption data and the cigarette downtime data are extracted, and two-dimensional multivariate time series data are generated. The input data particles are work shifts, and a total of 163 shifts are included. Two change points are identified by constructing a multivariate time series change point detection model (the change points are represented by “x”), and then the data in the time period is divided into three subgroups as shown in Figure 2 . Figure 2

[0078] As can be seen from Figure 2 , after the first change point, the cigarette paper consumption and the filter rod consumption both have an overall upward trend. After the second change point, the fluctuation of the cigarette paper consumption and the filter rod consumption is obviously increased. The identified change points indicate that there are two system changes in the cigarette production process in the time period. According to the positions (time, shift information) of the identified change points, the cigarette production manager can make in-depth investigation and identify the important factors affecting the cigarette production process.

[0079] As another example, please refer to Figure 3 , Figure 3 ​A schematic diagram of the detection result of parameter change point of a certain process section of a certain process; select the production process data of a certain batch of cigarette packaging, extract the end density waste smoke rejection data and VE section important process parameter needle roller proportion coefficient data, generate two-dimensional multi-time series data. The input data granularity is the work shift, which contains 200 shifts. By constructing a multi-time series change point detection model, 5 change points are identified Figure 3 The change point is represented by "x". Then the time period data is divided into 6 subgroups, as shown in Figure 3

[0080] As shown in Figure 3 It can be seen that the stability of the first half of the system is weak, and the average length of the subgroups is slightly lower than that of the second half. In addition, the third change point is important, after the third change point, the VE needle roller proportion coefficient setting value changes from stable to continuous fluctuation, and the end density waste smoke rejection rate changes from continuous decline to sudden overall rise, indicating that the system stability has changed. The change point detection result shows high business significance, and the stability of the process parameter control affects the control level of the process quality.

[0081] In this embodiment, by obtaining the production process data of the cigarette packaging; wherein the production process data includes production statistics indicators, equipment parameters and process parameters, and equipment failure downtime data; pre-processing the production process data to obtain multi-time series data; based on the multi-time series data, a multi-time series change point detection model is constructed and used to identify the change points in the multi-time series data; based on the position and number of the change points, the system stability of the cigarette packaging production process is identified, and the system stability identification result is output. By using the multi-production process data of the cigarette packaging to determine the position and number of change points in an adaptive manner, it is no longer limited by prior knowledge, and the system stability and non-stability of the cigarette packaging production process can be accurately identified.

[0082] Based on the foregoing embodiment, a second embodiment of the stability identification method in the cigarette packaging production process is proposed. In this embodiment, step S20 can include the following sub-steps:

[0083] Sub-step A10, detecting missing values and abnormal values in the production process data to obtain the missing and abnormal conditions of the production process data.

[0084] Wherein, the missing value can be a value that is lost or cannot be obtained at a certain time point in the production process data.

[0085] As an example, from Table 1 (VE needle roller proportion coefficient and end density waste smoke ratio), it can be seen that the end density waste smoke ratio collected on January 13, 2023 is a missing value.

[0086] Time VE needle roller proportionality factor Tip density waste smoke ratio 2023-01-10 a1 b1 2023-01-13 a2 ​

[0087] Table 1 VE needle roller proportion coefficient and end density waste smoke ratio

[0088] Outlier, which can be some data obviously deviating from other values in production process data.

[0089] As an example, from Table 2 (collected rejection rate and cigarette moisture content), it can be seen that the collected rejection rate (300%) on January 17, 2023 is an outlier.

[0090]

[0091]

[0092] Table 2 Rejection rate and cigarette moisture content

[0093] Missing condition, which can be a missing value condition in production process data, which can include missing value position (for example, time, shift information).

[0094] Abnormal condition, which can be an outlier condition in production process data, which can include outlier position (for example, time, shift information).

[0095] In a specific implementation, the missing condition and the abnormal condition of the production process data can be obtained by performing missing value and outlier detection and other preprocessing operations on the production process data.

[0096] Sub-step A20, correcting the production process data according to the missing condition and the abnormal condition to obtain multivariate time series data.

[0097] In a specific implementation, after obtaining the missing condition and the abnormal condition, the production process data can be corrected according to the missing condition and the abnormal condition; and then the multivariate time series data can be obtained.

[0098] As an example, the missing values in the production process data can be deleted or filled (mean, average, median, etc.) according to the missing condition.

[0099] As another example, the outliers in the production process data can be deleted or corrected (mean, average, median, etc.), retained, and converted according to the abnormal condition.

[0100] Further, in an embodiment, sub-step A20 can include the following sub-steps:

[0101] Sub-step A201, correcting the missing values of the production process data using a preset filling method according to the missing condition.

[0102] Sub-step A202, according to the abnormal situation, the preset processing method is used to correct the abnormal value of the production process data.

[0103] The preset filling method can be a method of filling the missing value by using the average value.

[0104] The preset processing method can be a method of correcting the abnormal value by using deletion or average value.

[0105] In this embodiment, by using the preset filling method to correct the missing value of the production process data according to the missing situation, and by using the preset processing method to correct the abnormal value of the production process data according to the abnormal situation, the accuracy of the systematic steady state identification of the cigarette production process can be improved by correcting the missing and abnormal values in the production process data.

[0106] In this embodiment, by detecting the missing value and the abnormal value of the production process data, the missing situation and the abnormal situation of the production process data are obtained, and the production process data is corrected according to the missing situation and the abnormal situation to obtain the multivariate time series data. Through a series of data preprocessing operations on the production process data, the accuracy of the systematic steady state identification of the cigarette production process can be improved.

[0107] Based on the foregoing embodiments, a third embodiment of the steady state identification method in the cigarette production process is provided. In this embodiment, step S30 can include the following sub-steps:

[0108] Sub-step B10, according to the multivariate time series data, a potential change point is identified.

[0109] The potential change point can be a time series data point in the multivariate time series data that can be a change point.

[0110] As an example, the maximum value (corresponding position point) of the divergence statistics of each time series data point E can be determined as the potential change point.

[0111] The divergence statistics of E can be defined as It can be measured by the divergence of the multivariate distribution, combined with the definition of the Euclidean distance, and can reflect the difference between the time series.

[0112] In specific implementation, first, the Euclidean distance of each time series data point can be calculated according to the multivariate time series data; then, the divergence statistics of the time series data point can be calculated according to the Euclidean distance of each time series data point; and finally, the potential change point can be identified according to the divergence statistics.

[0113] Further, in an embodiment, sub-step B10 can comprise the following sub-steps:

[0114] Sub-step B101, segmenting the multi-variate time series data according to the preset set of change points to obtain a first segment; wherein the preset set of change points has no change point;

[0115] Sub-step B102, calculating the E-divergence statistic of each time series data point in the first segment;

[0116] Sub-step B103, selecting the maximum point of the E-divergence statistic in the first segment, and setting the maximum point as a potential change point.

[0117] Wherein the preset set of change points has no change point, i.e., the set of change points is an empty set of change points, at this time the multi-variate time series data has only one segment (i.e., the first segment includes all the multi-variate time series data).

[0118] Calculating the E-divergence statistic of each time series data point k in the first segment Using the formula:

[0119]

[0120] Wherein X k ={Z1,Z2,…,Z k} is the time series data before the change point, with a length of n; is the time series data after the change point, with a length of m.

[0121] In this embodiment, first, the multi-variate time series data can be divided according to the preset set of change points to generate first segments C1, C2, …, C n ; further, the E-divergence statistic of each time series data point k in the first segments C1, C2, …, C n can be calculated Then the maximum point of the E-divergence statistic in the first segment (and the corresponding position point) is selected, and the maximum point is set as a potential change point.

[0122] As an example, please refer to Figure 4 , Figure 4 a schematic diagram of the calculation of the change point position of the multi-variate time series data; the calculation process of the potential change point position can be:

[0123] Step 1: dividing the multi-variate time series data according to the preset set of change points to generate first segments C1, C2, …, C n , wherein the length of the first segments C1, C2, …, C n is T (i) .

[0124] Step 2: initialize i = 1.

[0125] Step 3: if i < n, go to Step 4, otherwise go to Step 6.

[0126] Step 4: calculate the E-divergence statistics of each time series data point k in the first segment C1, C2, …, C n

[0127] Step 5: record the maximum value of the E-divergence statistics of each time series data point and the corresponding position point k (i) , and return to Step 3.

[0128] Step 6: determine the maximum E-divergence statistics in all first segments C1, C2, …, C n .

[0129] Step 7: output the corresponding potential change point k (i*) .

[0130] Sub-step B20: screening the potential change point according to a preset condition to identify the change point in the multivariate time series data.

[0131] The preset condition can be a condition for screening the potential change point as a change point.

[0132] As an example, a statistical test can be used to determine whether to add the potential change point to the preset change point set.

[0133] If the potential change point is retained (i.e., the potential change point is added to the change point set), the change point set is updated and the next potential change point is identified; if the potential change point is discarded, the change point division is not performed and the updated change point set is output.

[0134] In a specific implementation, the potential change point can be screened according to the preset condition, and then the change point in the multivariate time series data can be identified.

[0135] Further, in an embodiment, sub-step B20 can further include the following sub-steps:

[0136] Sub-step B201: using a preset permutation test method to obtain the significance level of the potential change point;

[0137] Sub-step B202: if the significance level of the potential change point is greater than a preset level, the potential change point is added to the change point set, and the first segment is updated.

[0138] ​Sub-step B203, the step of calculating the E-divergence statistic of each time series data point in the first segment is performed in a loop until the significance level of the potential change point is not greater than the preset level, the potential change point is discarded, and the iteration process is stopped.

[0139] The permutation test method can be performed by randomly permuting the order of the potential change point, recalculating the E-divergence statistic of each time series data point, and repeating the permutation process multiple times.

[0140] The preset level p0 can be a preset significance level of the change point.

[0141] In this embodiment, after obtaining the potential change point, the E-divergence statistic of the potential change point is first obtained using the permutation test method to obtain the significance level p of the potential change point. When the significance level p of the potential change point is greater than the preset level p0, the potential change point is added to the change point set and the first segment is updated. Then, the potential change point calculation process can be repeated, and when the significance level of the potential change point is not greater than the preset level, the potential change point is discarded and the iteration process is stopped. Finally, the change point set of the multivariate time series data is output.

[0142] As an example, refer to Figure 5 , Figure 5 the total flowchart of the generation of multiple change points; the generation process of the multiple change points can be as follows:

[0143] Step 1: The production process data of the cigarette is classified according to the production machine and other information, such as cigarette production, packaging production, cigarette downtime, cigarette rejection, filter rod consumption, and cigarette paper consumption. The data columns are taken as the horizontal coordinates to form multivariate time series sample Z1, Z2, …, Z T ∈R d , T is the sample length, and d is the total dimension of the sample.

[0144] Step 2: Calculate the pairwise distance of the sample data to obtain the distance matrix D, where D ij represents the distance |Z i -Z j | between two samples Z i and Z j . 2

[0145] Step 3: Use the multivariate process data change point position calculation method to generate a potential change point k.

[0146] Step 4: Use the permutation test method to obtain the E-divergence statistic of the potential change point a significance level p of the potential change point, wherein the significance level p can adopt a formula: R is the total number of permutation arrangements.

[0147] Step 5: when the significance level p of the potential change point > p0, the current potential change point is added to the change point set {τ1, …, τ k-1}, and the first segment is updated, and Step 3 is returned to calculate the next potential change point position; otherwise, when the significance level p of the potential change point ≤ p0, Step 6 is entered. Wherein p0 is a preset level.

[0148] Step 6: determining and outputting the change point set {τ1, …, τ k-1} of the multivariate time series data.

[0149] In the embodiment, the potential change point is identified according to the multivariate time series data, and the change point in the multivariate time series data is identified by screening the potential change point according to a preset condition. Thus, the change point is determined in a data adaptive manner, which can reduce the dependence on prior knowledge.

[0150] Based on the foregoing embodiment, a fourth embodiment of a steady state identification method in a cigarette package production process is provided. In the embodiment, step S40 can include the following steps:

[0151] Sub-step S401: determining a data segmented sub-interval of the multivariate time series data according to the position.

[0152] The data segmented sub-interval can be one or more sub-groups of the multivariate time series data divided according to the change point position.

[0153] As an example, when two change points are identified, the multivariate time series data can be divided into three sub-groups according to the positions (for example, time, shift information) of the change points.

[0154] In specific implementation, after the change points are identified, the multivariate time series data can be divided according to the positions of the change points, and then one or more data segmented sub-intervals can be obtained.

[0155] Sub-step S402: analyzing the systematic steady state of the cigarette package production process according to the data segmented sub-interval and the number.

[0156] In specific implementation, after one or more data segmented sub-intervals are obtained, the systematic steady state and non-steady state in the cigarette package production process can be identified according to the data segmented sub-interval and the number of change points.

[0157] As an example, when 5 change points are identified, the multivariate time series data can be divided into 6 subgroups according to the change point positions (for example, time, shift information). If the average length of the first 3 change points is slightly lower than that of the last half (the last 2 change points), it indicates that the systematic steady state in the first half (the first 3 change points) of the bale production process is weaker.

[0158] In this embodiment, by determining the data segment sub-interval of the multivariate time series data according to the positions, the systematic steady state of the bale production process is analyzed according to the data segment sub-interval and the number. Thus, the systematic steady state and non-steady state in the bale production can be effectively identified according to the positions and number of change points.

[0159] Based on the same inventive concept, the fifth embodiment of the present application also provides a steady state identification system in a bale production process corresponding to the steady state identification method in the bale production process of the foregoing embodiments. Since the principle of the system in the fifth embodiment of the present application solves the problem is similar to the foregoing steady state identification method in the bale production process of the foregoing embodiments of the present application, the implementation of the device can be referred to the implementation of the method, and the repeated parts will not be described herein. Please refer to Figure 6 The steady state identification system in the bale production process of the present application can include:

[0160] The acquisition module 10 is configured to acquire production process data of the bale; wherein the production process data includes production statistical indicators, equipment parameters, process parameters, and equipment failure downtime data;

[0161] The processing module 20 is configured to pre-process the production process data to obtain multivariate time series data;

[0162] The construction module 30 is configured to construct and use a multivariate time series change point detection model based on the multivariate time series data, and identify change points of the multivariate time series data;

[0163] The identification module 40 is configured to analyze the systematic steady state of the bale production process based on the positions and number of change points, and output a systematic steady state identification result.

[0164] In addition, the present application also provides a computer readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steady state identification method in the bale production process as described above.

[0165] The present embodiment provides an electronic device, in detail, the electronic device at least includes a memory and a processor connected by a bus, wherein the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to execute all or part of the steps in the foregoing method embodiments.

[0166] In summary, the application proposes a kind of full range utilization of volume package production process data, overcomes the adverse effects such as data complex characteristics and prior knowledge dependence, and realizes accurate identification of system stability and non-stability in volume package production process based on multivariate time series change point detection model.

[0167] The above embodiments only exemplarily illustrate the principles and effects of the present application, and are not intended to limit the present application. Any person skilled in the art can modify or change the above embodiments without departing from the spirit and scope of the present application. Therefore, all equivalent modifications or changes made by those skilled in the art without departing from the spirit and technical concept disclosed by the present application should be covered by the claims of the present application.

Claims

1. A method for steady state recognition in a cigarette making process, characterized by, The method comprises: obtaining production process data of the bale; wherein the production process data comprises production statistics, equipment parameters and process parameters, and equipment failure downtime data; preprocessing the production process data to obtain multivariate time series data; based on the multivariate time series data, constructing and using a multivariate time series change point detection model to identify change points in the multivariate time series data; based on the positions and quantities of the change points, analyzing the systemic steady state of the bale production process and outputting a systemic steady state identification result.

2. The method of claim 1, wherein, The preprocessing of the production process data to obtain multivariate time series data comprises: detecting missing values and outliers in the production process data to obtain the missing values and outliers in the production process data; based on the missing values and outliers, correcting the production process data to obtain multivariate time series data.

3. The method of claim 2, wherein, The correction of the production process data based on the missing values and outliers comprises: based on the missing values, correcting the missing values in the production process data using a preset filling method; based on the outliers, correcting the outliers in the production process data using a preset processing method.

4. The method of claim 1, wherein, The construction and use of the multivariate time series change point detection model based on the multivariate time series data to identify change points in the multivariate time series data comprises: identifying potential change points from the multivariate time series data; screening the potential change points based on a preset condition to identify change points in the multivariate time series data.

5. The method of claim 4, wherein, The identification of potential change points from the multivariate time series data comprises: segmenting the multivariate time series data according to a preset change point set to obtain a first segment; wherein the preset change point set has no change points; calculating the E-divergence statistics of each time series data point in the first segment; selecting the maximum point of the E-divergence statistics in the first segment and setting the maximum point as a potential change point.

6. The method of claim 5, wherein, The screening of the potential change points based on a preset condition to identify change points in the multivariate time series data comprises: using a preset permutation test method to obtain the significance level of the potential change point; if the significance level of the potential change point is greater than a preset level, adding the potential change point to the change point set and updating the first segment; recursively performing the steps of calculating the E-divergence statistics of each time series data point in the first segment until the significance level of the potential change point is not greater than the preset level, determining and outputting the change point set of the multivariate time series data.

7. The method of claim 1, wherein, The analysis of the systemic steady state of the bale production process based on the positions and quantities of the change points comprises: determining data segment subintervals of the multivariate time series data based on the positions; analyzing the systemic steady state of the bale production process based on the data segment subintervals and the quantities.

8. A system for identifying a steady state in a process for producing a roll, characterized in that The system comprises: An acquisition module is configured to acquire production process data of the cigarette package, wherein the production process data comprises production statistics, equipment parameters and process parameters, and equipment failure downtime data; A processing module is configured to pre-process the production process data to obtain multivariate time series data; A construction module is configured to construct and use a multivariate time series change point detection model based on the multivariate time series data to identify change points of the multivariate time series data; An identification module is configured to analyze system stability of the cigarette package production process based on positions and quantities of the change points and output a system stability identification result.

9. An electronic device, comprising: The electronic device comprises a memory and a processor; The memory is configured to store a computer program; The processor is configured to execute the computer program stored in the memory to enable the electronic device to perform the method for identifying stability in the cigarette package production process according to any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the method for identifying stability in the cigarette package production process according to any one of claims 1 to 7.

Images