A Holistic Data Mining Methodology for Engineering Applications (DMME)

Motivation

In cyber-physical production systems, data analyses are of fundamental importance for tasks such as system and process optimization and predictive maintenance. For the application of data-driven methods in industrial applications, the CRISP-DM approach represents a de facto standard. However, CRISP-DM does not specify a data acquisition phase within engineering scenarios.

Therefore, CRISP-DM has been extended to DMME (Data Mining Methodology for Engineering

Applications) has been extended. DMME provides a communication and planning basis for the application of data-driven methodologies in manufacturing, including the design and evaluation of the infrastructure for process-integrated data acquisition. In addition, the methodology includes design of experiments functions for systematic and efficient identification of relevant interactions.

Methodical Workflow of DMME

Figure 1: Structure of the methodical workflow of DMME

Figure 2: Steps for building technical understanding

Figure 3: Steps for designing the technical solution

The methodological procedure of DMME is summarized in the following steps. Details have been published in [1, 2, 3].

• Business Understanding
  • Definition of the business goals
  • Definition of data mining goals
  • Creation of a project plan
• Technical Understanding & Conceptualization
  • Definition of technical goals
  • Analysis of the technical situation
  • Conceptualization of the technical solution
  • Test planning and project plan
• Technical Realization & Testing
  • Prototypical realization of the test environment
  • Test of the concept
  • Conduct experiments and collect data
• Data understanding
  • Examine data in terms of data formats, readability, data quality, etc.
  • Analysis of the data with regard to the intended physical effects
  • Derive areas for improvement in data collection design
• Data preparation
  • Selection and integration of data sets for modeling.
  • Data cleaning. Correct or remove erroneous values
  • Construct new parameters from existing parameters
  • Format data so that it can be processed by algorithms or software
• Model Building
  • Selection of modeling techniques or algorithms
  • Creation of the model in the analysis software
  • Creating a test design with training, testing, and validation data sets
  • Calculation of the models and iterative model fitting
• Model Evaluation
  • Evaluate models for target values
  • Improve model accuracy by iterating the project flow and through continuous training
  • Evaluate models and decide on the model that best meets business success criteria
• Deployment
  • Implementation of the prototypes into productive operation
  • Planning and managing the monitoring and maintenance of the data-based application
  • Project review and further improvement of the application

Current research projects

In our research and development projects we use DMME as a basis for a systematic way of working. The subject of current work is not only the solution of the technical, engineering task, but also supporting elements such as workshop concepts, checklists or assistance for the practical implementation of DMME.

For companies we offer consulting as well as qualification formats.

References

3	Drowatzky, L.; Wiemer, H.; Ihlenfeldt, S.: Data Mining Suitable Digitization of Production Systems – A Methodological Extension to the DMME. In: Liewald, M., Verl, A., Bauernhansl, T., Möhring, HC. (eds) Production at the Leading Edge of Technology. WGP 2022. Lecture Notes in Production Engineering. Springer, Cham. https://doi.org/10.1007/978-3-031-18318-8_53
2	Wiemer, H. ; Drowatzky, L. ; Ihlenfeldt, S. : Data Mining Methodology for Engineering Applications (DMME) – A Holistic Extension to the CRISP-DM Model. In: MDPI Applied Sciences 9 (2019), Nr. 12, S. 2407 DOI: 10.3390/app9122407
1	Huber, S. ; Wiemer, H. ; Schneider, D. ; Ihlenfeldt, S. : DMME: Data Mining Methodology for Engineering Applications – A Holistic Extension to the CRISP-DM Model. 12th CIRP Conference on Intelligent Computation in Manufacturing Engineering, 18-20 July 2018, Gulf ofNaples, Italy, 2018 https://doi.org/10.1016/j.procir.2019.02.106