Process Mining And Optimization Base On Controlled Log

With the widespread use of business systems,the complexity of system models and the diversity of log attributes have increased.Analysis and research of system log and business processes are usually normalized.The main problems in the three aspects of process discovery,consistency detection and process improvement are as follows:1)Because the company or system owner needs to ensure user privacy and part of the system process is invisible,there is often a lack of effective data in log analysis and research,which is very important to the analysis algorithm.The verification and evaluation of the company are under a greater impact.The current log generation research will generate a large number of redundant logs,resulting in uncontrollable constraints between events;2)Process mining is to extract effective information from the logs available in today’s information systems to detect,simulate and improve the actual process.Existing process mining algorithms are built on event logs,in which only the execution of tasks is registered,and the remaining attributes of the logs are less used.3)Consistency checking is an effective method to check the deviation between the log and the model in the field of process mining,and alignment is one of many innovative methods.At this stage,most of the cost calculations for optimal alignment are only related to the number of moves in the alignment,and lack of analysis of activity dependence,that is,lack of consideration of the importance of activities;4)The current business process optimization is mainly for managers or developers.The developed business process model may deviate slightly from the actual operational business process,which affects the feasibility of the optimization result.Aiming at some of the problems in log generation,process mining,alignment calculation and model optimization,this Dissertation makes the following contributions based on the research of Petri net theory and application:(1)For dynamic arithmetic calculations,the Petri model is too complicated to be simulated by simulation software.This Dissertation proposes a method based on Java programming to dynamically construct a Petri net model and simulate the operation,and is used in the analysis and verification of the power model.This method makes dynamic The computer simulation of the structural Petri net model is realized and provides a basic framework for the log generation below.(2)The process mining algorithm is verified against the lack of operational logs or system logs containing the specified structure.The existing method is used by random spanning trees and generates random logs.Here,a method for generating controlled logs based on augmented Petri nets is proposed.This method can generate editable controlled logs for the specified structure model,support the conversion of multiple sets of logs,and apply it to the following process mining and alignment calculations.(3)In view of the fact that the existing lots have a large number of available attributes in addition to the active tags,most of the existing process mining algorithms lack the use of attributes other than the operational tags in the log.This Dissertation offers a way to use the additional information of the enhanced log to identify the organizational relationship between tasks.This algorithm simplifies the mining steps,and utilizes colored Petri nets to represent the scene information of the obtained process model.(4)In view of the unusual activity labels appearing in the existing alignment calculation process,the existing alignment calculation methods use the same cost for separate activities.An optimal alignment calculation method based on dynamic programming to enhance activity dependence is proposed and used in the previous The generated log and the calculation of the model obtained by the process mining,this method can differentiate the weight of different activities in the calculation of the optimal alignment and reflect it in the calculation results(5)In view of the inaccuracy of the model and the ignoring of the data flow during the process optimization process,most of the existing process optimization methods only optimize from the control flow structure.Here,a parallel optimization algorithm based on process mining is proposed.The addition of mining addresses the problem of outdated process models,and considers data interaction between activities to optimize the parallel structure,which improves the feasibility of actual optimization results to a certain extent.Figure[74]Table[13]Reference[94]...