AutoUniMo

Automotive Production Engineering Unified Perspective based on Data Mining Methods and Virtual Factory Model

MES for Short Series Production (EMBA2.0)

Prototyping process

There are two main kinds of production carried out by Prototyping Department of Continental Ingolstadt. First one is focused on prototypes of new automotive electronics devices that have to be produced to allow testing of new Advanced Driver Assistance Systems (ADAS) before starting of their mass production. This kind of production has obviously low product volume. The quality data collected during the process by means of MES is an essential source of information used in the improved product versions development. However, the product quality must be ensured on the same level like in the case of mass production. The second kind of production is short series production. In this case the advanced car electronic equipment is produced in short series often with many variants.

All orders are collected by electronic order acquisition system. The Prototyping Department in Ingolstadt serves to different Continentals centers all over the world which forces the exchanged information to have flexible format and content. Each order is supervised by dedicated product manager. This process is partially manual since product manager has to contact with customer and agree both technical issues and logistic aspects of production as well. Afterwards orders are qualified to execution. The logistics team is responsible for preparation of all necesary materials together with technical documentation for production process. Production is splitted between SMD production, that is focussed on electronics components and backend production related to housing and connectors assembling.

In the Prototyping Department the test section consists of tree operations: AOI (Automatic Optical Inspection(, Flying Probe and  Rework station. AOI tests allow for detection of any irregularities on SMD surface by automatic visual inspection of particular electronic components. Tests are carried out by automated stations that use vision recognition algorithms. In the case of error detection the additional verification and error validation is carried out manually by operator. Error information is processed by MES.

Depending on the type of production and type of error, products can be scrapped or directed to Rework Station. Reworking can be also used for additional product modifications in  the case of prototype production. Test devices or additional connectors not used in the final products can be added on Rework Station. Next, products are examined due to the correctness of electrical connections. Electrical tests are performed by the automatic Flying Probe station. In the case of error product can be classified as scrap or rework.

The product flow on Test section is depictured in the figure below. The maximum possible number of repetitions of each production phase is limited by product description. It causes that actual production cycle is not predetermined by product definition but is emerging during production realisation. Therefore, classical MES cannot be used in such variable production production and product supervisory operations have to be executed manually.

EMBA Inspection Process

EMBA 2.0 architecture and implementation

The existing MES system for Prototyping Department EMBA 1.0 did not comply to the new functional and non-functional requirements of the countinusly developing process.

The new version of the system introduce not only new functionalities required by the users and management but also implements new software architecture allowing for futher extensions and integration with other systems.

The main system requirements are as follows:

  • Application architecture
    • More scalability
    • Better administration functionalities
    • .NET technology as Continental standard
    • Interface to external systems
    • Agents and services support
    • Extensibility
  • Porcess implementation
    • Flexible process flow
    • Improved tracebility
    • Data validations
    • Data model optimization
  • User interface
    • OS independent
    • Modern Look & Feel
    • Continental standards

In order to implement the architecture requirements ASP .Net MVC 4.0 technology has been choosen.

The Model-View-Controler design pattern allows for splitting the system into independent components joined by interfaces which makes the system more flexible for extensions as well as introduces a better division of work and roles. The system architecture provides also better access and administration. The optimized data model allows for automatic data entry and validation which shorten the operation time (especially in case of Rework station).

The model of information is directly connected to Oracle Data Base which stores the data together with procedures allowing for accesing the data according to implemented business logic. The controller is a bridge between the model and the view which is HTML5 user interface as well as data interface to external systems. The ASP .NET MVC service is executed by means of IIS 7.5 server on the platfrom of Windows Server 2008. 

 EMBA Architecture

The new user interface with implemented input control and modern design has made the operator‘s work more fluent and error resistant. Every work station has similar action screen flow and the Look & Feel is impemented according to the Continental’s standards. The client application can be accessed by any HTML5 compatible web browser.

 EMBA Client

The current status of EMBA 2.0 system implemetation is as presented in the figure below. The inspection work stations operate productively. Currently the two new work places are developed: SMD production tracking and intelocking and the station where the process technology for a given order is setup. The ordering application is out of scope of the project.

EMBA Implementation

Distributed agent based architecture for QM support

As the MVC archtecture allows for joining different components we decided to implement MES functions according to Holon and Agent paradigm. The controller in this case implements the .NET SignalR interface opening the system for message based communication to external agents. The first implemented application supports Quality Management system.

Proposed architecture consist of three levels as depicted in the figure below. There are holons on the bottom, that are directly connected with devices used for production (EMBA Workplaces). Three types of holons were realized: AOI holons are directly linked to Automatic Optical  Inspection, FP Holons with Flying Prob stations and RU Holons with Rework Units.

EMBA Agents
The Product Supervisory Agents are placed above Holons level and perform operations that are related to product supervisor activities. Because of the flexible mode of production product path is not fixed before production but its actual implementation is traced by Product Supervisor Agents basing on the information sent by Holons.

The third level of agents is necessary to aggregate information from different orders and to support finding about origin of errors. In the former system this operations were performed manualy. In proposed system information is exchanged automatically by events exchanged between agents. Error analysis is executed in on-line mode by Quality Management Agent.

To implement and test our Distributed Agent Based Architecture concept we decide to use the open-source ASP .NET SignalR library which let us to build flexible application layer based on the simple network communication framework. The popular Java based agent frameworks (e.g. JADE, WADE) could not be used in our solution due to the development policy in the corporation which excludes Java applications from productive solutions.

EMBA Agents SignalR