A method and system for constructing a multi-source printed data fusion and visualization operation analysis platform

Abstract

The application discloses a kind of multi-source printing data fusion and visual operation analysis platform construction method and system, it is related to printing production data fusion and visual analysis technical field, the present application includes data acquisition processing, multi-dimension feature extraction, data fusion association and operation analysis report, acquisition multi-source data;Extract printing equipment efficiency feature, order execution feature, quality defect feature and material consumption feature;Four kinds of features are fused in multilevel, construct four-dimensional association matrix and carry out dimension reduction and cluster analysis, generate fusion feature set;Based on fusion feature set, generate printing operation visual instrument panel, carry out early warning to abnormal operation index and generate operation analysis report, the present application constructs equipment state transition probability matrix, multi-color set printing collaborative analysis and cross-dimension causal conduction chain, improve the depth fusion range of printing production multi-source data and abnormal attribution analysis precision.

Description

Technical Field

[0001] This invention relates to the field of printing production data fusion and visualization analysis technology, specifically to a method and system for constructing a multi-source printing data fusion and visualization operation analysis platform. Background Technology

[0002] The printing industry is a typical example of discrete manufacturing, involving various equipment such as printing presses, die-cutting machines, hot stamping machines, and laminating machines. The production processes are complex, and production data is scattered across different systems and equipment. Printing companies need to simultaneously monitor multiple dimensions, including equipment utilization efficiency, order delivery schedules, product quality levels, and material consumption costs. These dimensions are interconnected and mutually influential. Therefore, effectively integrating dispersed, multi-source data to achieve comprehensive and visualized analysis of printing production operations is a crucial issue for printing companies seeking to improve their management. This necessitates the development of a multi-source printing data fusion and visualization operation analysis platform.

[0003] In the existing technology, printing companies typically use enterprise resource planning systems to manage orders and materials, manufacturing execution systems to manage production schedules, and quality management systems to record inspection data. These systems operate relatively independently. Some companies summarize cross-system data by importing and exporting data tables, and then managers manually create reports for analysis.

[0004] The existing technology has the following technical problems: 1. The analysis dimensions of the existing system are relatively simple, focusing on a single independent indicator, such as equipment utilization rate and order completion rate. It lacks correlation analysis of the four dimensions of equipment, order, quality and materials, and cannot discover potential problems across dimensions, such as the correlation between equipment performance degradation and quality defect rate increase, or the causal relationship between abnormal material loss and equipment operating status.

[0005] 2. The existing system's analysis of quality data is limited to basic statistical indicators such as defect rate and process capability index. It lacks in-depth trend analysis of key printing quality parameters such as color difference and registration deviation, making it difficult to discover the overall drift of color difference and the coordinated change of registration deviation of each color group.

[0006] 3. Existing systems mostly rely on fixed report formats for visualization, lacking dynamic interactive dashboards and the ability to attribute anomalies based on association rule reasoning. As a result, managers cannot quickly grasp the overall operational situation and promptly locate the root cause of anomalies.

[0007] 4. Existing systems only focus on outcome indicators such as delivery time and output when evaluating order execution efficiency, lacking the ability to explore process efficiency indicators such as waiting time between processes and frequency of abnormal interruptions, making it difficult to identify bottlenecks in the production process. Summary of the Invention

[0008] In view of the above-mentioned technical deficiencies, the purpose of this invention is to provide a method and system for constructing a multi-source printing data fusion and visualization operation analysis platform.

[0009] To solve the above technical problems, the present invention adopts the following technical solution: The present invention provides a method for constructing a multi-source printing data fusion and visualization operation analysis platform, including the following steps: Step 1, data acquisition and processing: Collect multi-source data in the printing production process, including printing equipment operation data, order data, quality inspection data and material consumption data.

[0010] Step 2: Multi-dimensional feature extraction: Extract printing equipment efficiency features, order execution features, quality defect features, and material consumption features from multi-source data.

[0011] Step 3: Data Fusion and Association: Multi-level fusion of printing equipment performance characteristics, order execution characteristics, quality defect characteristics, and material consumption characteristics is performed to construct a four-dimensional association matrix. Dimensionality reduction and cluster analysis are then performed on the four-dimensional association matrix to generate a fused feature set.

[0012] Step 4: Operational Analysis Report: Based on the fusion feature set, generate a visual dashboard for printing operations, issue early warnings for abnormal operational indicators, and generate an operational analysis report.

[0013] Preferably, the efficiency characteristics of the printing equipment are extracted as follows: the printing equipment operation data includes the equipment operation status, which includes running, standby, fault shutdown and planned shutdown.

[0014] The operating status of the equipment is encoded, and the status encoding values ​​at each sampling time are arranged sequentially according to the sampling time to obtain the equipment status encoding sequence. The changes in equipment status between adjacent time points are statistically analyzed in chronological order, and the occurrence frequency of each type of status change is counted to construct a matrix. The rows correspond to the state before the transition, and the columns correspond to the state after the transition. The element value is equal to the cumulative frequency of the state change. The diagonal elements of the matrix represent the frequency of the state not changing, and the off-diagonal elements reflect the switching frequency between different states. By performing row normalization on this matrix, that is, dividing each element in each row by the sum of the elements in that row, the state transition probability matrix is ​​obtained.

[0015] Divide the transition frequency of each state change by the total state transition frequency to obtain the response frequency ratio of each state change.

[0016] The efficiency feature vector of printing equipment is formed by arranging the proportion of fault recovery frequency, the proportion of sudden fault frequency, and the proportion of standby response frequency in that order.

[0017] Preferably, the order execution features are extracted as follows: the order data includes the actual start time of each process, the actual completion time of each process, the number of times the equipment fails and stops during production, the number of times the equipment is idle during production, the current actual output, and the planned output.

[0018] For completed orders, the actual completion time of each process in each order is used as the node. The interval between adjacent processes is taken as the waiting time between adjacent processes. The waiting time between each adjacent process is added up and divided by the total number of processes minus one to obtain the average waiting time of each order. The arithmetic mean is then used to obtain the average waiting time of the order. The sum of the number of equipment downtimes and standby times during the production process of each order is counted and divided by the total production time of the corresponding order to obtain the frequency of abnormal interruption per unit time for each order. The arithmetic mean is then used to obtain the average frequency of abnormal interruption per order.

[0019] For incomplete orders, divide the current actual output of each incomplete order by the planned output to obtain the current order completion rate. Calculate the required production rate to complete each incomplete order before the delivery deadline. Divide the required production rate by the current actual production rate of the corresponding order to obtain the order completion difficulty coefficient. Calculate the proportion of incomplete orders with completion difficulty coefficients greater than the preset difficulty threshold to the total number of incomplete orders to obtain the proportion of high-risk delayed orders.

[0020] The order execution feature vector is formed by arranging the average waiting time for each process, the average frequency of abnormal interruptions, and the proportion of high-risk delayed orders in that order.

[0021] Preferably, the quality defect features are extracted as follows: the quality inspection data includes color difference values ​​and overprinting deviation values.

[0022] The color difference values ​​are summarized in chronological order to obtain a color difference value sequence. The entire sampling interval is traversed with a preset window length in the color difference value sequence. The mean and standard deviation of the color difference values ​​in each color difference window are calculated. The color difference window with the smallest deviation between the mean and the preset target color difference value and the smallest standard deviation is taken as the standard color difference interval, thus obtaining the standard color difference value interval.

[0023] Each color difference point whose color difference value falls within the standard color difference value range is recorded as a standard color difference point. The sampling time of each standard color difference point is used as the independent variable and the color difference value is used as the dependent variable to perform curve fitting, thereby obtaining the color difference reference curve.

[0024] For each color difference point within the non-standard color difference value range, the vertical distance from each color difference point to the curve is collected. The squares of the vertical distances of all color difference points in the non-standard color difference value range are summed and then the square root is taken to obtain the overall color difference deviation.

[0025] The overprinting deviation values ​​were analyzed to obtain the multi-color group overprinting coordination deviation, comprehensive deviation drift rate, deviation drift difference between color groups, and deviation mutation frequency.

[0026] The overall color difference deviation, multi-color group overprinting coordination deviation, comprehensive deviation drift rate, inter-color group deviation drift difference, and deviation mutation frequency are arranged in order to form a quality defect feature vector.

[0027] The beneficial effects of the present invention are as follows: 1. By constructing the state transition frequency matrix and state transition probability matrix in the performance characteristics of printing equipment, the present invention extracts the proportion of fault recovery frequency, the proportion of sudden fault frequency and the proportion of standby response frequency, and quantifies the dynamic change law of equipment status into the feature vector. Compared with only using static equipment comprehensive efficiency indicators, it can better reflect the stability and responsiveness of equipment operation.

[0028] 2. This invention introduces a color difference reference curve and a vertical point distance calculation method to quantify the deviation of color difference points in non-standard intervals into an overall color difference deviation. This can capture the overall offset of color difference relative to the standard fluctuation pattern. At the same time, by performing multi-color group collaborative analysis and deviation drift analysis on the overprinting deviation value, the overprinting collaborative deviation degree of multi-color group, the comprehensive deviation drift rate, and the deviation mutation frequency are extracted, realizing in-depth trend analysis of key parameters of printing quality.

[0029] 3. This invention generates a set of association rules by calculating the Pearson correlation coefficient, association direction, and optimal lag order among the feature columns in the four-dimensional association matrix, and further constructs a cross-dimensional causal transmission chain. When an anomaly warning is issued, the information of the strongly correlated features in the set of association rules is used as an attribution reference for the anomaly, providing data support for managers to locate the root cause of anomalies.

[0030] 4. This invention analyzes the waiting time between processes and the frequency of abnormal interruptions in an order, extracts the average waiting time between processes and the average frequency of abnormal interruptions in an order, and calculates the proportion of high-risk delayed orders. It evaluates the order execution status from the perspective of process efficiency and can identify bottlenecks and potential delay risks in the production process. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0032] Figure 1 This is a schematic diagram of the implementation steps of the method of the present invention. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] according to Figure 1 As shown, the present invention provides a method for constructing a multi-source printing data fusion and visualization operation analysis platform, including the following steps: Step 1, data acquisition and processing: acquire multi-source data in the printing production process, including printing equipment operation data, order data, quality inspection data and material consumption data.

[0035] Step 2: Multi-dimensional feature extraction: Extract printing equipment efficiency features, order execution features, quality defect features, and material consumption features from multi-source data.

[0036] In one specific embodiment, the efficiency characteristics of the printing equipment are extracted as follows: the printing equipment operation data includes the equipment operation status, which includes running, standby, fault shutdown, and planned shutdown.

[0037] The operating status of the equipment is encoded, and the status encoding values ​​at each sampling time are arranged sequentially according to the sampling time to obtain the equipment status encoding sequence. The changes in equipment status between adjacent time points are statistically analyzed in chronological order, and the occurrence frequency of each type of status change is counted to construct a matrix. The rows correspond to the state before the transition, and the columns correspond to the state after the transition. The element value is equal to the cumulative frequency of the state change. The diagonal elements of the matrix represent the frequency of the state not changing, and the off-diagonal elements reflect the switching frequency between different states. By performing row normalization on this matrix, that is, dividing each element in each row by the sum of the elements in that row, the state transition probability matrix is ​​obtained.

[0038] Divide the transition frequency of each state change by the total state transition frequency to obtain the response frequency ratio of each state change.

[0039] The frequency proportions of responses to various state changes are arranged in order to form a feature vector of printing equipment efficiency.

[0040] It should be noted that the efficiency feature vector of printing equipment also includes overall equipment efficiency, time utilization rate, performance utilization rate, and yield rate.

[0041] The time utilization rate is obtained by dividing the duration marked as running in the equipment's operating status by the planned running time of the equipment within the statistical period; the performance utilization rate is obtained by dividing the actual average printing speed of the equipment within the statistical period by the rated printing speed of the equipment; the pass rate is obtained by dividing the number of finished products that pass quality inspection within the statistical period by the total output of the equipment; and the overall efficiency of the equipment is obtained by multiplying the time utilization rate, performance utilization rate, and pass rate together.

[0042] The planned operating time and rated printing speed of the equipment during the statistical period are preset values ​​by the staff. The actual average printing speed of the equipment during the statistical period and the number of finished products that pass quality inspection during the statistical period are collected additionally by the corresponding equipment.

[0043] In one specific embodiment, the order execution features are extracted as follows: the order data includes the actual start time of each process, the actual completion time of each process, the number of times the equipment fails and stops during production, the number of times the equipment is idle during production, the current actual output, and the planned output.

[0044] For completed orders, the actual completion time of each process in each order is used as the node. The interval between adjacent processes is taken as the waiting time between adjacent processes. The waiting time between each adjacent process is added up and divided by the total number of processes minus one to obtain the average waiting time of each order. The arithmetic mean is then used to obtain the average waiting time of the order. The sum of the number of equipment downtimes and standby times during the production process of each order is counted and divided by the total production time of the corresponding order to obtain the frequency of abnormal interruption per unit time for each order. The arithmetic mean is then used to obtain the average frequency of abnormal interruption per order.

[0045] It should be noted that for each completed order, the actual completion time of each process in the order is used as the node, and the difference between the actual completion time of the previous process and the actual start time of the next process is used as the waiting time between adjacent processes.

[0046] For incomplete orders, divide the current actual output of each incomplete order by the planned output to obtain the current order completion rate. Calculate the required production rate to complete each incomplete order before the delivery deadline. Divide the required production rate by the current actual production rate of the corresponding order to obtain the order completion difficulty coefficient. Calculate the proportion of incomplete orders with completion difficulty coefficients greater than the preset difficulty threshold to the total number of incomplete orders to obtain the proportion of high-risk delayed orders.

[0047] The order execution feature vector is formed by arranging the average waiting time for each process, the average frequency of abnormal interruptions, and the proportion of high-risk delayed orders in that order.

[0048] It should be noted that the order execution feature vector also includes order completion rate, average production cycle, order delay rate, output achievement rate, delay difference between product types, and periodic fluctuation difference between product types.

[0049] Read the product types of all orders within the statistical period, group the orders by product type, calculate the order delay rate for each product type, which is the number of orders whose actual completion time is later than the delivery deadline divided by the total number of completed orders for that product type. Take the standard deviation of the order delay rates for different product types to obtain the delay difference between product types.

[0050] Read the product types of all orders within the statistical period, group the orders by product type, and for each completed order, subtract the planned start time from the actual completion time to obtain the actual production cycle. Calculate the arithmetic mean of the actual production cycles of orders for each product type as the actual production cycle of that product type. Divide the difference between the actual production cycle and the standard production cycle of each product type by the standard production cycle to obtain the production cycle deviation rate of each product type. Take the standard deviation of the production cycle deviation rates of different product types to obtain the periodic fluctuation difference degree between product types.

[0051] In one specific embodiment, the quality defect features are extracted as follows: the quality inspection data includes color difference values ​​and overprinting deviation values.

[0052] The color difference values ​​are summarized in chronological order to obtain a color difference value sequence. The entire sampling interval is traversed with a preset window length in the color difference value sequence. The mean and standard deviation of the color difference values ​​in each color difference window are calculated. The color difference window with the smallest deviation between the mean and the preset target color difference value and the smallest standard deviation is taken as the standard color difference interval, thus obtaining the standard color difference value interval.

[0053] Each color difference point whose color difference value falls within the standard color difference value range is recorded as a standard color difference point. The sampling time of each standard color difference point is used as the independent variable and the color difference value is used as the dependent variable to perform curve fitting, thereby obtaining the color difference reference curve.

[0054] For each color difference point within the non-standard color difference value range, the vertical distance from each color difference point to the curve is collected. The squares of the vertical distances of all color difference points in the non-standard color difference value range are summed and then the square root is taken to obtain the overall color difference deviation.

[0055] It should be noted that for each color difference point within the non-standard color difference value range, the curve point corresponding to the target color difference point on the time axis is taken as the initial point, and the search is performed along both sides of the curve at a preset step size. The distance from the color difference point to each search point is calculated, and the search point corresponding to the minimum distance is taken as the approximate perpendicular point. This minimum distance is used as the perpendicular distance from the color difference point to the curve.

[0056] The overprinting deviation values ​​were analyzed to obtain the multi-color group overprinting coordination deviation, comprehensive deviation drift rate, deviation drift difference between color groups, and deviation mutation frequency.

[0057] The overall color difference deviation, multi-color group overprinting coordination deviation, comprehensive deviation drift rate, inter-color group deviation drift difference, and deviation mutation frequency are arranged in order to form a quality defect feature vector.

[0058] It should be noted that the quality defect feature vector also includes the product defect rate, the defect category proportion vector, and the process capability index.

[0059] The product defect rate is obtained by dividing the number of defective products detected within the statistical period by the total number of products detected within that period. The detected defective products are classified according to the defect type, and the proportion of the number of defective products of each type to the total number of defective products is calculated to obtain the defect category proportion vector. The process capability index is calculated for the overprinting deviation value. The process capability index is the difference between the upper limit and the lower limit of the overprinting deviation tolerance divided by six times the standard deviation of the overprinting deviation.

[0060] In one specific embodiment, the analysis of the overprinting deviation value is carried out as follows: the overprinting deviation value sequence is grouped by color group, with each color group corresponding to a set of overprinting deviation value sequences. The mean, standard deviation, and range of the overprinting deviation value sequences of each color group are calculated respectively. The correlation coefficient between each pair of overprinting deviation value sequences of each color group is calculated to obtain the overprinting deviation correlation coefficient matrix between color groups. The mean of the off-diagonal elements in the overprinting deviation correlation coefficient matrix between color groups is taken as the multi-color group overprinting cooperative deviation degree.

[0061] For the overprinting deviation value sequence of each color group, a linear regression is performed with sampling time as the independent variable and overprinting deviation value as the dependent variable. The slope of the regression line for each color group is extracted as the deviation drift rate of that color group. The absolute value of the mean of the deviation drift rate of each color group is calculated as the comprehensive deviation drift rate. The deviation drift rates of each color group are sorted, and the difference between the maximum and minimum deviation drift rates is taken as the deviation drift difference between color groups.

[0062] The deviation change of adjacent sampling points is calculated point by point according to the sampling time sequence of the overprinting deviation value sequence. The deviation change is divided by the corresponding sampling time interval to obtain the instantaneous change rate of deviation. The proportion of the number of sampling points exceeding the preset change rate threshold to the total number of sampling points is calculated for the instantaneous change rate sequence of deviation to obtain the deviation change frequency.

[0063] In one specific embodiment, the material consumption characteristics are extracted as follows: the material consumption data includes paper consumption, ink consumption, and printing plate consumption.

[0064] The paper consumption sequence is segmented according to a preset time window. The ratio of paper consumption in each time window to finished product output in the corresponding time period is calculated to obtain the paper consumption sequence per unit product for each time window. The coefficient of variation is calculated for the paper consumption sequence per unit product to obtain the paper consumption volatility. The ink consumption volatility is obtained according to the method for obtaining the paper consumption volatility.

[0065] The material consumption of printing equipment under different operating conditions within the statistical period is classified according to the equipment operating condition. The material consumption per unit time is calculated when the equipment is in operation, standby, and debugging state. The material consumption per unit time in standby state is divided by the material consumption per unit time in operation state to obtain the standby material consumption ratio. The material consumption per unit time in debugging state is divided by the material consumption per unit time in operation state to obtain the debugging material consumption ratio.

[0066] The ink consumption volatility and the ratio of adjustment material consumption are arranged in order to form a material consumption feature vector.

[0067] It should be noted that the material consumption feature vector also includes paper loss rate, ink loss rate, plate loss rate, waste material ratio, and downtime material retention ratio.

[0068] The paper consumption per unit product is obtained by dividing the total paper consumption within the statistical period by the total output of finished products. The ink consumption per unit product and the plate consumption per unit product are obtained in the same way. The paper loss rate is obtained by dividing the difference between the actual paper consumption and the standard paper consumption by the standard paper consumption. Based on the method for obtaining the paper loss rate, the ink loss rate and the plate loss rate are obtained.

[0069] The percentage of scrapped materials is obtained by dividing the amount of scrapped materials due to quality defects during the statistical period by the total amount of material consumed; the percentage of material left behind during equipment downtime is obtained by dividing the amount of material left behind during downtime by the total amount of material consumed.

[0070] Step 3: Data Fusion and Association: Multi-level fusion of printing equipment performance characteristics, order execution characteristics, quality defect characteristics, and material consumption characteristics is performed to construct a four-dimensional association matrix. Dimensionality reduction and cluster analysis are then performed on the four-dimensional association matrix to generate a fused feature set.

[0071] In one specific embodiment, the generation process of the fusion feature set is as follows: the values ​​of the printing equipment efficiency feature vector, order execution feature vector, quality defect feature vector, and material consumption feature vector within the same statistical period are taken as a row, and the data from multiple statistical periods are stacked vertically to construct a four-dimensional correlation matrix. The rows of the matrix are the statistical periods, and the columns are the elements of each feature dimension.

[0072] The data in each column of the four-dimensional correlation matrix are standardized. The Pearson correlation coefficient between each column is calculated in the standardized correlation matrix to obtain the correlation coefficient matrix. The absolute value of the correlation coefficient is recorded as the correlation strength. The column pairs in the correlation coefficient matrix whose absolute value of the correlation coefficient is greater than the preset strong correlation threshold are marked as strong correlation feature pairs.

[0073] It should be noted that the standardized correlation matrix is ​​obtained by subtracting the mean of each column from the data in that column and then dividing by the standard deviation of that column.

[0074] The association direction is determined for strongly correlated feature pairs. The lag correlation coefficient is calculated for each strongly correlated feature pair to obtain the optimal lag order. The strongly correlated feature pairs, association direction, association strength and optimal lag order are combined to form an association rule set.

[0075] It should be noted that the direction determination process is as follows: calculate the linear regression slope of the two features. If the slope is positive, it is determined to be positively correlated; if the slope is negative, it is determined to be negatively correlated.

[0076] The process of calculating the lag correlation coefficient is as follows: shift one feature forward by a preset time step and calculate the correlation coefficient with the other feature. Take the shift step corresponding to the maximum absolute value of the correlation coefficient as the optimal lag order for the feature pair.

[0077] Principal component analysis is performed on the standardized correlation matrix to obtain the dimensionality-reduced fusion feature set. Hierarchical clustering analysis is then performed on the fusion feature set to obtain the category feature profile.

[0078] It should be noted that the first few principal components whose cumulative variance contribution rate reaches the preset contribution rate threshold are extracted, and the original standardized data are projected onto the principal component space to obtain the dimensionality-reduced fusion feature set.

[0079] Different statistical periods are divided into several operational status categories. For each category, the mean of the feature vectors of each dimension of each period is calculated to obtain the category feature profile.

[0080] The strongly correlated feature pairs in the association rule set are grouped according to the source dimension. For each association group, the mean of the best lag order of all strongly correlated feature pairs in the group is calculated and denoted as the average time lag of each dimension. The average time lag of each dimension is sorted from shortest to longest to obtain the cross-dimensional causal transmission chain.

[0081] It should be noted that each group is divided into equipment order association group, equipment quality association group, equipment material association group, order quality association group, order material association group, and quality material association group.

[0082] Step 4: Operational Analysis Report: Based on the fusion feature set, generate a visual dashboard for printing operations, issue early warnings for abnormal operational indicators, and generate an operational analysis report.

[0083] In one specific embodiment, the process of generating the printing operation visualization dashboard is as follows: generate a bar chart of the overall efficiency of each printing equipment with the equipment name as the horizontal axis and the overall efficiency as the vertical axis; generate each equipment at each standard level according to the preset efficiency target range; and mark each equipment corresponding to each standard level with different colors to obtain a capacity utilization rate view.

[0084] It should be noted that a preset efficiency target line is superimposed on the bar chart. Equipment with an overall efficiency lower than the preset lower limit is marked with the first color, equipment with an overall efficiency between the preset lower limit and the preset upper limit is marked with the second color, and equipment with an overall efficiency higher than the preset upper limit is marked with the third color.

[0085] The capacity utilization rate view also includes a radar chart of the average equipment efficiency characteristics. The radar chart of the average equipment efficiency characteristics corresponding to each operating status category in the corresponding category feature profile is embedded in the capacity utilization rate view, and the actual value of the equipment efficiency characteristic vector for the current period is marked on the radar chart.

[0086] Read the planned output and actual output completed for each order, divide the actual output completed by the planned output to get the delivery progress percentage, generate a progress bar chart with the order number as the horizontal axis and the delivery progress percentage as the vertical axis, and obtain the order delivery progress view.

[0087] It should be noted that orders with a delivery progress percentage lower than the preset progress warning line are marked with a warning color. The order delivery progress view displays feature pairs that are strongly correlated with the order delay rate in the association rule set, and a delay risk dashboard is generated using the proportion of high-risk delayed orders as an indicator.

[0088] Read the defect category percentage vector, generate a Pareto chart with defect type as the horizontal axis and percentage as the vertical axis, and mark the cumulative percentage curve on the chart to obtain a quality defect distribution view.

[0089] It should be noted that the time-series change curve of the overall color difference deviation is overlaid in the quality defect distribution view, and a heat map of the overprinting coordination status is generated with the multi-color group overprinting coordination deviation as an indicator.

[0090] Read the unit product material consumption for multiple consecutive statistical periods, generate a line chart with the statistical period as the horizontal axis and the consumption as the vertical axis, and overlay a moving average trend line on the line chart to obtain a material consumption trend view.

[0091] It should be noted that the stacked area of ​​the scrap material ratio and the downtime material retention ratio is marked in the material consumption trend view, and a pie chart of the non-production state material consumption ratio is generated using the standby material consumption ratio and the debugging material consumption ratio as indicators.

[0092] In one specific embodiment, the process of issuing early warnings for abnormal operation indicators is as follows: when the overall efficiency of the equipment is lower than the preset efficiency warning lower limit, an abnormal equipment efficiency warning is generated, and information on features strongly correlated with the overall equipment efficiency in the association rule set is used as an attribution reference for the abnormality.

[0093] When the percentage of order delivery progress at the current time point is lower than the preset percentage of progress that should be completed at that time point, or when the percentage of high-risk delayed orders exceeds the preset high-risk percentage threshold, an order delay risk warning will be generated.

[0094] When the product defect rate exceeds the preset defect rate limit, or the overall color difference deviation exceeds the preset deviation limit, a quality anomaly warning is generated, and information on features strongly correlated with the corresponding indicators in the association rule set is used as an anomaly attribution reference.

[0095] When the material loss rate exceeds the preset loss rate limit, an abnormal material consumption warning is generated.

[0096] It should be noted that each warning message includes the warning type, warning object, current indicator value, preset warning threshold and related attribution information, and is displayed on the visual dashboard as a pop-up window or flashing icon.

[0097] In one specific embodiment, the process of generating the operation analysis report is as follows: at the end of each statistical period, the fusion feature set and category feature profile of the current period are read, the operation status category of the current period is compared with the category of the previous period, and if the category changes, the status migration information is recorded. The status migration information includes the category before migration, the category after migration, the migration direction, and the change in the category feature profile.

[0098] Read the association rule set of the current period, filter strongly associated feature pairs with an association strength greater than the preset report citation threshold, generate an association analysis summary, read the cross-dimensional causal transmission chain, perform root cause labeling on the features associated with the causal dimension with the shortest average time lag, and generate a causal analysis summary.

[0099] The current periodic data and historical comparison data, early warning records, status transition information, correlation analysis summary, causal analysis summary, and periodic summary values ​​of each feature vector from each visualization view are integrated into an operational analysis report.

[0100] It should be noted that the operational analysis report also includes trend analysis of order execution efficiency and trend analysis of material consumption stability.

[0101] Read the average process waiting time and average abnormal interruption frequency of orders in the current period, compare them with historical data for the same period, and generate an analysis of the trend of order execution efficiency; read the paper consumption volatility and ink consumption volatility, compare them with historical data for the same period, and generate an analysis of the trend of material consumption stability.

[0102] The data acquisition technology, data preprocessing technology, timestamp alignment technology, interpolation filling technology, outlier detection and replacement technology, state coding conversion technology, Pearson correlation coefficient calculation, linear regression analysis, principal component analysis, hierarchical clustering analysis, cubic spline interpolation curve fitting, least squares method, process capability index calculation, standardization processing method, and moving average trend line calculation described in this invention are all existing technologies and can be found on the Internet, so they will not be described in detail here.

[0103] It should also be noted that the various preset thresholds and preset parameters described in this invention are based on long-term production operation data from the printing workshop, historical quality inspection records, equipment maintenance logs, and test results from multiple rounds of process experiments. These are combined with the practical experience of printing process technicians and equipment maintenance personnel, and determined by considering the characteristics of the printing equipment model, the complexity of the product process, batch differences in raw and auxiliary materials, and the range of temperature and humidity variations in the workshop environment. Specific preset values ​​specified in the manual include: a preset time interval of 30 minutes; a preset fluctuation range of the mean plus or minus 3 times the standard deviation; a preset window length of 50 sampling points; a preset target color difference value of 1.5; a preset strong correlation threshold of 0.7; and a preset contribution... The preset thresholds for contribution rate (85%), change rate (0.05 mm / thousand sheets), difficulty (1.5), progress warning line (80%), efficiency target line (75%), lower efficiency limit (60%), upper efficiency limit (90%), defect rate limit (2%), deviation limit (2.5%), loss rate limit (5%), high-risk percentage (30%), report citation threshold (0.8), and step size (0.1 seconds) are all exemplary presets adapted to typical offset printing production scenarios. Specific adjustments can be made according to the actual product structure and equipment configuration of the printing company. The relevant detailed rules are based on mature existing technologies and will not be elaborated upon here.

[0104] The examples described in this invention are not limited to the specific embodiments listed above. The examples are merely illustrative to facilitate understanding of the invention and do not constitute a limitation on the scope of protection of this invention. Any modifications, equivalent substitutions, etc., made within the spirit and principles of this invention should be included within the scope of protection.

[0105] The above description is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined in this specification, they should all fall within the protection scope of the present invention.

Claims

1. A method for constructing a multi-source printing data fusion and visualization operation analysis platform, characterized in that, Includes the following steps: Step 1: Data Acquisition and Processing: Collect multi-source data during the printing production process, including printing equipment operation data, order data, quality inspection data, and material consumption data; Step 2: Multi-dimensional feature extraction: Extract printing equipment efficiency features, order execution features, quality defect features, and material consumption features from multi-source data; Step 3: Data Fusion and Association: Multi-level fusion of printing equipment efficiency characteristics, order execution characteristics, quality defect characteristics, and material consumption characteristics is performed to construct a four-dimensional association matrix. Dimensionality reduction and cluster analysis are then performed on the four-dimensional association matrix to generate a fused feature set. Step 4: Operational Analysis Report: Based on the fusion feature set, generate a visual dashboard for printing operations, issue early warnings for abnormal operational indicators, and generate an operational analysis report.

2. The method for constructing a multi-source printing data fusion and visualization operation analysis platform according to claim 1, characterized in that, The specific extraction process for the efficiency characteristics of the printing equipment is as follows: The printing equipment operation data includes the equipment operating status, which includes running, standby, fault shutdown, and planned shutdown. The operating status of the equipment is encoded, and the status encoding values ​​at each sampling time are arranged sequentially according to the sampling time to obtain the equipment status encoding sequence. The changes in equipment status between adjacent time points are statistically analyzed in chronological order, and the occurrence frequency of each type of status change is counted to construct a matrix. The rows correspond to the state before the transition, and the columns correspond to the state after the transition. The element value is equal to the cumulative frequency of the state change. The diagonal elements of the matrix represent the frequency of the state not changing, and the off-diagonal elements reflect the switching frequency between different states. By performing row normalization on this matrix, that is, dividing each element of each row by the sum of the elements in that row, the state transition probability matrix is ​​obtained. Divide the transition frequency of each state change by the total state transition frequency to obtain the response frequency ratio of each state change. The efficiency feature vector of printing equipment is formed by arranging the proportion of fault recovery frequency, the proportion of sudden fault frequency, and the proportion of standby response frequency in that order.

3. The method for constructing a multi-source printing data fusion and visualization operation analysis platform according to claim 1, characterized in that, The specific extraction process for the order execution features is as follows: Order data includes the actual start time of each process, the actual completion time of each process, the number of times equipment malfunctions caused downtime during production, the number of times equipment was idle during production, the current actual output, and the planned output. For completed orders, the actual completion time of each process in each order is used as the node. The interval between adjacent processes is taken as the waiting time between adjacent processes. The waiting times between adjacent processes are added together and divided by the total number of processes minus one to obtain the average waiting time of each order. The arithmetic mean is then used to obtain the average waiting time of the order. The sum of the number of equipment downtimes and standby times during the production process of each order is counted and divided by the total production time of the corresponding order to obtain the frequency of abnormal interruption per unit time for each order. The arithmetic mean is then used to obtain the average frequency of abnormal interruption per order. For incomplete orders, divide the current actual output of each incomplete order by the planned output to obtain the current order completion rate. Calculate the required production rate to complete each incomplete order before the delivery deadline. Divide the required production rate by the current actual production rate of the corresponding order to obtain the order completion difficulty coefficient. Calculate the proportion of incomplete orders with completion difficulty coefficients greater than the preset difficulty threshold to the total number of incomplete orders to obtain the proportion of high-risk delayed orders. The order execution feature vector is formed by arranging the average waiting time for each process, the average frequency of abnormal interruptions, and the proportion of high-risk delayed orders in that order.

4. The method for constructing a multi-source printing data fusion and visualization operation analysis platform according to claim 1, characterized in that, The specific extraction process for the quality defect features is as follows: Quality inspection data includes color difference values ​​and registration deviation values; The color difference values ​​are summarized in chronological order to obtain a color difference value sequence. The entire sampling interval is traversed with a preset window length in the color difference value sequence. The mean and standard deviation of the color difference values ​​in each color difference window are calculated. The color difference window with the smallest deviation between the mean and the preset target color difference value and the smallest standard deviation is taken as the standard color difference interval, and thus the standard color difference value interval is obtained. Each color difference point whose color difference value falls within the standard color difference value range is recorded as a standard color difference point. The sampling time of each standard color difference point is used as the independent variable and the color difference value is used as the dependent variable to perform curve fitting and obtain the color difference reference curve. For each color difference point within the non-standard color difference value range, the vertical distance from each color difference point to the curve is collected. The squares of the vertical distances of all color difference points in the non-standard color difference value range are summed and then the square root is taken to obtain the overall color difference deviation. The overprinting deviation values ​​were analyzed to obtain the multi-color group overprinting coordination deviation, comprehensive deviation drift rate, deviation drift difference between color groups, and deviation mutation frequency. The overall color difference deviation, multi-color group overprinting coordination deviation, comprehensive deviation drift rate, inter-color group deviation drift difference, and deviation mutation frequency are arranged in order to form a quality defect feature vector.

5. The method for constructing a multi-source printing data fusion and visualization operation analysis platform according to claim 4, characterized in that, The analysis of the overprinting deviation value is performed as follows: The overprinting deviation value sequence is grouped by color group, with each color group corresponding to a set of overprinting deviation value sequences. The mean, standard deviation, and range of the overprinting deviation value sequences of each color group are calculated. The correlation coefficient between each pair of overprinting deviation value sequences of each color group is calculated to obtain the overprinting deviation correlation coefficient matrix between color groups. The mean of the off-diagonal elements in the overprinting deviation correlation coefficient matrix between color groups is taken as the multi-color group overprinting cooperative deviation degree. For the overprinting deviation value sequence of each color group, linear regression is performed with sampling time as the independent variable and overprinting deviation value as the dependent variable. The slope of the regression line for each color group is extracted as the deviation drift rate of that color group. The absolute value of the mean of the deviation drift rate of each color group is calculated as the comprehensive deviation drift rate. The deviation drift rates of each color group are sorted, and the difference between the maximum and minimum deviation drift rates is taken as the deviation drift difference between color groups. The deviation change of adjacent sampling points is calculated point by point according to the sampling time sequence of the overprinting deviation value sequence. The deviation change is divided by the interval of the corresponding sampling time to obtain the instantaneous change rate of deviation. The proportion of the number of sampling points exceeding the preset change rate threshold to the total number of sampling points is calculated for the instantaneous change rate sequence of deviation to obtain the deviation change frequency.

6. The method for constructing a multi-source printing data fusion and visualization operation analysis platform according to claim 1, characterized in that, The specific extraction process for the material consumption characteristics is as follows: Material consumption data includes paper consumption, ink consumption, and printing plate consumption; The paper consumption sequence is segmented according to a preset time window. The ratio of paper consumption in each time window to finished product output in the corresponding time period is calculated to obtain the unit product paper consumption sequence for each time window. The coefficient of variation is calculated for the unit product paper consumption sequence to obtain the paper consumption volatility. The ink consumption volatility is obtained according to the method of obtaining the paper consumption volatility. The material consumption of printing equipment under different operating conditions during the statistical period is classified according to the equipment operating condition. The material consumption per unit time when the equipment is in the running state, standby state, and debugging state is calculated separately. The material consumption per unit time in the standby state is divided by the material consumption per unit time in the running state to obtain the standby material consumption ratio. The material consumption per unit time in the debugging state is divided by the material consumption per unit time in the running state to obtain the debugging material consumption ratio. The ink consumption volatility and the ratio of adjustment material consumption are arranged in order to form a material consumption feature vector.

7. The method for constructing a multi-source printing data fusion and visualization operation analysis platform according to claim 1, characterized in that, The specific process for generating the fused feature set is as follows: The values ​​of printing equipment efficiency feature vector, order execution feature vector, quality defect feature vector, and material consumption feature vector within the same statistical period are grouped into one row, and the data from multiple statistical periods are stacked vertically to construct a four-dimensional correlation matrix. The rows of the matrix are the statistical periods, and the columns are the elements of each feature dimension. The data in each column of the four-dimensional correlation matrix are standardized. The Pearson correlation coefficient between each column is calculated in the standardized correlation matrix to obtain the correlation coefficient matrix. The absolute value of the correlation coefficient is recorded as the correlation strength. The column pairs in the correlation coefficient matrix with the absolute value of the correlation coefficient greater than the preset strong correlation threshold are marked as strong correlation feature pairs. Determine the association direction for strongly correlated feature pairs to obtain the association direction. Calculate the lag correlation coefficient for each strongly correlated feature pair to obtain the optimal lag order. Combine the strongly correlated feature pairs, association direction, association strength, and optimal lag order to form an association rule set. Principal component analysis is performed on the standardized correlation matrix to obtain the dimensionality-reduced fusion feature set. Hierarchical cluster analysis is then performed on the fusion feature set to obtain the category feature profile. The strongly correlated feature pairs in the association rule set are grouped according to the source dimension. For each association group, the mean of the best lag order of all strongly correlated feature pairs in the group is calculated and denoted as the average time lag of each dimension. The average time lag of each dimension is sorted from shortest to longest to obtain the cross-dimensional causal transmission chain.

8. The method for constructing a multi-source printing data fusion and visualization operation analysis platform according to claim 1, characterized in that, The specific process for generating the printing operations visualization dashboard is as follows: The overall efficiency of each printing machine is plotted as a bar chart with the machine name on the horizontal axis and the overall efficiency on the vertical axis. Based on the preset efficiency target range, each machine at each target level is generated. Each machine at each target level is labeled with a different color to obtain a capacity utilization rate view. The capacity utilization rate view also includes a radar chart of the average equipment efficiency characteristics. The radar chart of the average equipment efficiency characteristics corresponding to each operating status category in the corresponding category feature profile is embedded in the capacity utilization rate view, and the actual value of the equipment efficiency characteristic vector for the current period is marked in the radar chart. Read the planned output and actual completed output of each order, divide the actual completed output by the planned output to get the delivery progress percentage, generate a progress bar chart with the order number as the horizontal axis and the delivery progress percentage as the vertical axis, and get the order delivery progress view; Read the defect classification percentage vector, generate a Pareto chart with defect type as the horizontal axis and percentage as the vertical axis, and mark the cumulative percentage curve on the chart to obtain a quality defect distribution view; Read the unit product material consumption for multiple consecutive statistical periods, generate a line chart with the statistical period as the horizontal axis and the consumption as the vertical axis, and overlay a moving average trend line on the line chart to obtain a material consumption trend view.

9. The method for constructing a multi-source printing data fusion and visualization operation analysis platform according to claim 1, characterized in that, The specific process for issuing early warnings for abnormal operational indicators is as follows: When the overall efficiency of the equipment is lower than the preset efficiency warning limit, an abnormal equipment efficiency warning is generated, and information on features that are strongly correlated with the overall equipment efficiency in the association rule set is used as an attribution reference for the abnormality. When the percentage of order delivery progress at the current time point is lower than the preset percentage of progress that should be completed at that time point, or when the percentage of high-risk delayed orders exceeds the preset high-risk percentage threshold, an order delay risk warning will be generated. When the product defect rate exceeds the preset defect rate limit, or the overall color difference deviation exceeds the preset deviation limit, a quality anomaly warning is generated, and the information of the feature pairs that are strongly correlated with the corresponding indicators in the association rule set is used as an anomaly attribution reference. When the material loss rate exceeds the preset loss rate limit, an abnormal material consumption warning is generated.

10. The method for constructing a multi-source printing data fusion and visualization operation analysis platform according to claim 1, characterized in that, The specific process for generating the operation analysis report is as follows: at the end of each statistical period, the fusion feature set and category feature profile of the current period are read, the operation status category of the current period is compared with the category of the previous period, and if the category changes, the status migration information is recorded. The status migration information includes the category before migration, the category after migration, the migration direction, and the amount of change in the category feature profile. Read the association rule set of the current period, filter strongly associated feature pairs with an association strength greater than the preset report citation threshold, generate an association analysis summary, read the cross-dimensional causal transmission chain, perform root cause labeling on the features associated with the causal dimension with the shortest average time lag, and generate a causal analysis summary. The current periodic data of each visualization view is integrated with historical comparison data, early warning records, status transition information, correlation analysis summary, causal analysis summary, and periodic summary values ​​of each feature vector into an operation analysis report.

Images