Printing Unit Condition Monitoring

This data set contains raw sensor signals of four analogue sensors and five features derived from them. These are used to monitor the condition of a printing unit in a demonstrator application and to detect a sensor defect, which is simulated in the data set. The demonstrator is used to simulate the wiping process of an Intaglio printing process. Intaglio is the major printing process to produce security prints like banknotes. Engraved structures in the printing plates, which are mounted on a rotating plate cylinder, are filled with ink, which is transferred onto the printing substrate under high pressure. A second cylinder denoted by wiping cylinder, which is working in the printing unit, is lubricated with a solvent to wipe off surplus ink from the printing plates by rotating in the direction opposite to the plate cylinder. This process is crucial as wiping errors immediately lead to print errors. The printing unit demonstrator contains models of the two cylinders, which are turned by electric drives. Pressure between the wiping cylinder having a rubber surface and the steel-surfaced plate cylinder is freely adjustable. A set of four analogue sensors (contact force, solid-bourne sound, electric current of wiping and plate cylinder drives) continuously acquire data during operation to monitor the process. The sensors each output a continuous voltage signal in the range of [-10,10] V, which is proportional to the respective quantity the sensor is observing. Thus, each signal's unit is irrelevant and abandoned as changes of the original quantity of interest are reflected also in the respective voltage signal. All output time-domain signals are synchronously and equidistantly sampled at a frequency of 20 kHz and quantised with a resolution of 16 bit. The acquired data is then split into non-overlapping batches of 50000 samples (corresponding to 2.5 sec of operation), respectively. The length of the time frame was chosen to ensure that 3 revolutions of the plate cylinder are captured in each signal data batch. The solid-bourne sound signal is treated by the FFT to determine its frequency spectrum per signal batch. Altogether, 5 features per plate cylinder revolution are extracted. This results in 15 feature values per signal data batch. The extracted features are: contact force mean: arithmetic mean of the contact force, solid-bourne sound intensity: root mean square of the solid-bourne sound, solid-bourne sound maxPowFreqInd: index of the frequency component with largest power, motor current wiping cylinder mean: arithmetic mean of the wiping cylinder motor current, motor current plate cylinder mean: arithmetic mean of the plate cylinder motor current. Each plate cylinder revolution is represented by one instance in the feature data sets. That is, every instance in the data set is described by a vector of 5 feature values. The raw data and feature data sets are divided into two parts, each containing data of one of the two experiments under different operation conditions: Static printing unit demonstrator operation: The static experiment observes the printing unit demonstrator during 20:13 min of operation. The printing unit demonstrator was started immediately before the data acquisition began. No additional manipulations or events occurred during the experiment. Therefore, only data representing the demonstrator's normal condition is contained in the data set. It contains 10,000,000 raw signal samples resulting in 600 instances (plate cylinder revolutions), which are in summary described by 3,000 feature values. Manipulated printing unit demonstrator operation: The printing unit demonstrator was started ca. 23:00 min before the data acquisition began. During this 10:31 min long experiment, the demonstrator application was intentionally manipulated. In addition, the solid-bourne sound sensor signal was manipulated through low-pass filtering in order to simulate a defect of this sensor. An unintended incident also occurred during this experiment. Therefore, data representing both the demonstrator's normal and abnormals conditions are contained in the data set. The sequence of events along with an objective classification of the demonstrator condition by the human experimenter is summarised in the file PrintingUnit_manip_events.txt. The data set contains 5,950,000 raw signal samples, which are in summary described by 1,785 feature values. File name conventions and contents The files in the data set are organised such that each row represents a data set instances, columns represent the respective sensor or feature: PrintingUnitData*.csv: These files contain raw sensor signals. PrintingUnitFeatures*.csv: These files contain the features extracted from the sensor signals. Additional files contain information about *_condition.csv: The condition if the printing unit is indicated by 'n' (normal condition) or 'a' (abnormal condition). These are the labels of the data set instances. *_filter.csv: The solid-bourne sound filter status is indicated by '1' (filter activated) or '0' (filter deactivated). *_time.csv: The relative time, at which the respective instance of the data set was determined. It is represented as the number of days from January 0, 0000 as is returned from MATLAB's datenum function (cf. http://www.mathworks.com/help/matlab/ref/datenum.html for details). The operation conditions (with respect to the experiment, cf. above) are distinguished by *_static*.csv: Static printing unit demonstrator operation. *_manip*.csv: Manipulated printing unit demonstrator operation