Research On Knowledge Based Energy Efficiency Optimization Method For Discrete Manufacturing System

Discrete manufacturing industry includes aerospace,automotive industry,electronic devices,equipment manufacturing and other industries.The number of discrete manufacturing industries is huge,in which case its overall energy consumption and CO2 emissions can not be ignored.The energy consumption of discrete manufacturing system is closely related to many elements,such as equipment state and its transmission structure,processing parameters,process feature ranking,product structure,production scheduling and start-stop control.In particular,complex discrete manufacturing processes,such as flexible manufacturing and mixed-flow manufacturing,are affected by process characteristics,such as changing processing tasks,equipment status and process dynamics,which leads to complex energy consumption mechanism.The unknown parameters in the processing parameter/thermodynamic energy consumption model,the highly nonlinear characteristics of the equipment energy consumption process and the inevitable randomness and dynamics in the complex workflow are all challenges for the modeling,analysis and optimization of system energy consumption process.The existing research results focused on the optimization of the energy efficiency index based on the static mechanism model.In the system modeling and optimization,the dynamic uncertainty factors in the production operation process are not fully considered.They also ignored the vast amount of information provided by the discrete manufacturing environment.Based on the knowledge automation,this thesis proposes the energy consumption knowledge representation and update method for discrete manufacturing systems and present the process parameter energy-saving optimization method,process route energy-saving optimization method and dynamic scheduling energy-saving optimization method based on the hierarchy of manufacturing energy consumption system and association analysis.Besides,an example workshop is used to develop a discrete manufacturing system energy efficiency optimization software platform.The main research contents of this thesis are as follows:(1)Based on the energy flow analysis of discrete manufacturing systems,an ontologybased energy consumption knowledge representation and update method is proposed.According to energy flow analysis,the energy consumption of discrete manufacturing system consists of four levels: working step layer,process layer,part layer and product layer and have complex dynamic correlation with the state of the device,process route,planned scheduling and other factors,which is difficult to uniformly describe by mathematical model.To solve this multi-source multi-level dynamic system modeling problem,this thesis builds an ontology-based multi-granular hierarchical description model for energy consumption knowledge based on top-down method.This model semantically labels various types of knowledge through ontology,correlates ontology and knowledge according to different levels and different energy-consuming elements of energy consumption system,and finally forms an energy knowledge network to reflect the energy consumption flow and elements.Besides,since the existing knowledge update method is difficult to guarantee the timeliness of knowledge,a knowledge automatic update mechanism is proposed based on the memory and forgetting mechanism which ensures the validity of knowledge by increasing the proportion of energy consumption knowledge reuse in knowledge management.(2)Based on energy consumption data and knowledge,a process parameter energy-saving optimization method based on CBR(Case Based Reasoning)and CMOPSO(Competitive mechanism based Multi-objective Particle Swarm Optimizer)is proposed.Considering that the working-step energy consumption is affected by various factors,such as equipment type and processing parameters,traditional static mechanism modeling method has high technical requirements and it is difficult to describe the dynamic changes of the working step energy consumption characteristics caused by the aging of the equipment.This thesis builds an energy consumption knowledge instance library based on the knowledge modelling method mentioned before,quantifies the importance of heterogeneous energy-consuming elements based on input power fluctuation,and uses hierarchical case retrieval method and CBR method to retrieve similar working-steps to predict energy consumption.Based on this prediction model,a CMOPSO method is used to optimize processing parameters with the aim of processing energy efficiency and processing time.Experimental results show that the proposed prediction method has higher prediction accuracy than existing methods.Besides,process parameter optimization can improve effective energy consumption,reduce idle energy consumption,thereby improve the working-step energy efficiency.(3)Considering the correlation between process route and part energy efficiency,an energy efficiency oriented typical process route knowledge automatic discovery and push method is proposed.Because of insufficient consideration of process information for existing process route similarity indicators,a novel process similarity coefficient is presented combined with improved psudo-LCS(Longest Common Subsequence)and Jaccard similarity measure.Besides,the energy consumption information and process similarity information are fused by entropy weight method to form a comprehensive similarity index of process route.In order to ensure the clustering validity,quantity soft constraint and size soft constraint are introduced in the clustering part.Considering the situations that the number of clusters is known or unknown,the K-medoids algorithm,the AP(Affinity Propagation)and their hybrid algorithm are improved to find effective clustering results that meet the soft constraint requirements.Besides,an automatic push method for typical process route knowledge based on comprehensive similarity is also present.Experiments show that the proposed algorithm can find a more reasonable typical process route,avoid production bottleneck,reduce standby energy consumption and improve the energy efficiency of parts production.(4)Because of the influence of production scheduling on energy efficiency of discrete manufacturing systems,an automatic discovery method for energy-saving scheduling rules based on GPHH(Genetic Programming Hyper Heuristic)is proposed.The traditional meta-algorithm in GPHH deals with scheduling decisons in a non-delayed manner,which has its disadvantages in dealing with the dynamic environment.Based on this,this thesis proposes a novel meta-algorithm which delays the routing decisions.This novel meta-algorithm ensures that all the scheduling decisions can consider the latest system information to make the most reasonable decisions.Based on this idea,three queue selection strategies are also proposed,and GPHH is used to automatically generate the DRs used in the novel meta-algorithm.Based on the proposed algorithm,the energy efficiency and the average delay time are chosen as the objective to solve the dynamic flexible discrete manufacturing system scheduling problem.The results show that the proposed algorithm can achieve higher energy efficiency than the existing algorithms.(5)Taking the large part workshop of a machine tool manufacturing enterprise in Wuxi as a demonstration workshop,the energy efficiency optimization methods proposed in this thesis are applied to the production process of large parts of machine tools.Under the Windows system,the energy efficiency optimization software platform of discrete manufacturing system is designed and developed.The manufacturing process supervision module,process parameter optimization module,typical process route query module and scheduling rule model are built to realize the energy consumption monitoring and energy efficiency optimization of the demonstration workshop manufacturing process.The effectiveness of the optimization method is analyzed from four levels: process,equipment,parts and products.The results show that the software platform can effectively improve the energy efficiency of the discrete manufacturing workshop.