Development of Knowledge Integration Model for E-Maintenance

In recent years, e-maintenance provides the opportunity for a new maintenance generation. Knowledge is one of core e-maintenance resources. To improve the efficiency of knowledge management and so as to improve the quality of e-maintenance work, a knowledge integration model for e-maintenance was proposed. The proposed model was made up of the relationships among role knowledge, task knowledge and equipment knowledge, to achieve the integration of maintenance business and knowledge resources. This study involved the following tasks: (1) developed the key components of the proposed model, including e-maintenance federation ontology, emaintenance knowledge space and knowledge integration network; (2) designed the construction procedures for the proposed model; (3) presented an example to illustrate the application of the proposed model. Results of this study can improve the level of knowledge management for e-maintenance.


INTRODUCTION
The concept of e-maintenance in the industry refers to the integration of the information and communication technologies within the maintenance strategy and/or plan to service for plant maintenance (Holmberg et al., 2010).E-maintenance is characteristic of knowledge intensive and collaborative (Levrat et al., 2009).Each of Maintenance tasks requires employing of various knowledge and experience.Members of e-maintenance federation have the willing to share knowledge, so as to cooperate with other members and finish the maintenance work.Consequently, knowledge management is one of the important guarantees to the successful execution of e-maintenance.
At present, researches on knowledge management in e-maintenance mainly focus on the following three aspects: Relevant technologies to achieve knowledge sharing (Ivana et al., 2005;Ren et al., 2007); Resources sharing (include knowledge sharing) model or framework (Hung et al., 2005;Jianwen and Dejie, 2012).However, in addition to the effective sharing, knowledge in e-maintenance also requires effective integration model to rapidly respond to the demand of maintenance work.This study proposes the knowledge integration model for e-maintenance, which is made up of the relationships among role knowledge, task knowledge and equipment knowledge, to achieve the integration of maintenance business and knowledge resources.The model can support the overall view of knowledge resources in e-maintenance and to achieve the effective organization and application of emaintenance knowledge.

BUSINESS OPERATIONAL MODEL FOR E-MAINTENANCE
The business operational model for e-maintenance is shown as in Fig. 1.E-maintenance federation is a virtual organization in which different enterprise share the processes, activities and resources of maintenance to provide a value-add maintenance services.Maintenance Business (MB) is a value-add maintenance service.An e-maintenance federation can perform several MBs.A maintenance business can be divided into several Maintenance Tasks (MT).Maintenance Role (MR) is the divisions of duties in e-maintenance federation and each role reflects a position and rank in organization architecture.Maintenance Worker (MW) is a staff of federation enterprise which actual performed the MT.MR is assigned to MP to perform and MT is assigned to MR to perform.Mechanical Equipment (ME) is Mechanical device or Mechanical part in Mechanical device.Each ME is associated with one or more MTs and each MT is associated with one or more MEs.An integration model of knowledge must cover the following three elements f business:  Maintenance role: Each maintenance role has different demand for maintenance knowledge.For example, as to a fault diagnosis expert, some knowledge about fault diagnosis theory will be recommended to knowledge user.For example, when the task is about compressor, some knowledge about structure of compressor will be recommended to knowledge user.

FRAMEWORK OF KNOWLEDGE INTEGRATION MODEL
The framework of knowledge integration model is shown as in Fig. 2. E-Maintenance Federation Ontology (EMFO) provides a common understanding for operational model of e-maintenance business.E-Maintenance Knowledge Space (EMKS) provides knowledge description model, which is the base of knowledge integration.Knowledge Integration Network (KIN) is made up of the relationships among role knowledge, task knowledge and equipment knowledge, to achieve the integration of maintenance business and knowledge resources.

Development of the model: E-maintenance federation ontology:
Ontology is an explicit, shared and formal description of important concepts and their relationships (Brewster et al., 2004).E-Maintenance Federation Ontology (EMFO) is to enable the federation members to share information on federation business for knowledge integration.EMFO can be expressed as: Mechanical Equipment Ontology (MEO) describes the concept and relationship of equipment and its parts and reflects the hierarchy and structures of equipment parts.Relationships of concepts in EMFO are shown as in Table 1 (CR is the relation function).
E-maintenance knowledge space: Knowledge space model is an n-dimension space, in which each point uniquely identifies a kind of knowledge resource (Zhuge, 2002).This study has extended the knowledge space model to construct E-Maintenance Knowledge Space (EMKS) to descript knowledge of e-maintenance.EMKS is represented as follows: In EMKS, Background is sub space of knowledge background, which has the basic information of the knowledge, Content is sub space of knowledge content, which describes the content of knowledge; Application is sub space of knowledge application, which describes the applicable scope of the knowledge.
Background describes the background information of knowledge object.Background is represented as follows: Background = (Owner, Author, Creation Date, Modify Date) Level classifies knowledge into five knowledge levels of "Description", "Rule", "Procedure" and "Case".Explanation of each knowledge level is detailed as in Table 2. Logic classifies knowledge into five knowledge sort of "Know-What", "Know-Why", "Know-Who" and "Know-How".Explanation of each knowledge sort is detailed as in Table 3. Version is the version of knowledge object.Abstract is abstract description of knowledge object.Keywords is a collection of words, which descript the function of knowledge object.Evaluation is evaluation value of knowledge quality.Location is the location of knowledge object.Application is expressed as: Application = (MRS, MTS, MES) MRS is the role set, which describes the suitable roles of knowledge.MRS = {ROC 1 , ROC 2 ,…, ROC n }, ROC i (1≤i≤n) is the concept of the MRO.MTS is the task set, which describes the suitable takes of knowledge.MTS = {AOC 1 , AOC 2 ,…, AOC m }, AOC j (1≤j≤m) is the concept of MAO.MES is the equipment sets, which describes the suitable the suitable equipments of knowledge.MES = {EOC 1 , EOC 2 ,…, EOC P }, EOC k (1≤k≤p) is the concept of MEO.where, nnr ij : The relationship between two knowledge nodes nnr ij is expressed as: where, kn i & kn j : Knowledge nodes nr : Relationship between kni and knj and nr∈R KNG is used to show knowledge and the network structure by constructing the relationship among organization knowledge, business knowledge and equipment knowledge.In KNC, the concepts of maintenance role, task and equipment are all expressed as knowledge node.KNC also can be expressed as: where, oor ij : The relationship between two knowledge objects oor ij is expressed as: oor ij = (kr i , kr j , or) where, kr i & kr j : Knowledge objects or : Relationship between kr i and kr j OORC includes various associations which exist between knowledge objects.For example, the KO of a bearing fault and the abnormal knowledge of a bearing have the causal relationship.The main types of relations in OORC are shown in Table 4.   MEO are all expressed as knowledge nodes, one knowledge node can link with multiple knowledge objects; and one knowledge object can also link with multiple knowledge nodes.Each enterprise has its own KOG, so as to manage its knowledge resources.Each KOG of enterprise made up of KNG of e-maintenance federation.The relations of KN, the relations of KO and the links between KN and KO made up of KIN.

CONSTRUCTION PROCEDURES FOR THE MODEL
Procedures to set up the knowledge organization network are as follows:  Develop the EMKO: Dominance enterprise in Emaintenance federation develops the ENKO and publishes it.Expert", he can get the knowledge of "Diagnosis Expert" knowledge node (such as "Theory of Diagnosis"); when executing the diagnosis, the maintenance worker can obtain "Fault Diagnosis" knowledge node (such as "Process of Diagnosis"); if the task is to diagnose the fault of steam turbine, the "Steam Turbine" knowledge node can also be acquired (such as "Structure of Steam Turbine").From the above examples, conclusions can be that the model can achieve knowledge navigation, knowledge supply and knowledge integration by maintenance business.

CONCLUSION
Knowledge is one of core e-maintenance resources and critical support element of maintenance process.The knowledge integration model is proposed to achieve the integration of maintenance business and knowledge resources.From the application examples, it is known that the model can support the navigation knowledge, supply activities and improve the efficiency of knowledge management.

Fig. 3 :
Fig. 2: Framework of knowledge integration model EMFO = (C, R) where, C = Ontology concept set R = Ontology relation set Top level concepts and relations of EMFO are shown as in Fig. 3. Maintenance Activate Ontology (MAO) describes the activity concept and the relationship in the maintenance business.Maintenance Role Ontology (MRO) describes the role and relationship of the federation and maintenance members.
node graph KOG : Knowledge object graph NOLS : The links of knowledge node and knowledge object KNG is the mapping of the EMKO and is made up of knowledge nodes and their relations.KNG is expressed as: KNG = (KNC, NNRC) where, KNC : Collection of Knowledge Node (KN) NNRC : Collection of KN's relations and NNRC  R KN is expressed as: KN = (ID, KC, P) where, ID : Identifier of KN KC : Knowledge concept which adopts the concept of EMFO, namely KC∈C P : The collection of KN, which defied by emaintenance federation NNRC is expressed as: NNRC = (nnr ij , 1≤i≤n, 1≤j≤m)

Table 4 :
Main types of relations in OORC