HTM, hierarchical temporal memorythe LCA data machine
About the LCA data machine
Life cycle assessment studies are data intense, modern studies integrate life cycles that cover > 10,000 individual process data sets.
Data collection and modelling is done manually; integrating expert judgement and many other sources is a lot of effort, error-prone, and difficult for quality assurance.
Data collection and modelling is currently a bottleneck for LCA, due to missing information, use of outdated information, and use of incorrect information without being aware of it.
The LCA data machine (LCADM) is a system to create data sets for LCA, automatically and fast, demand-driven, with controlled fitness for purpose, currently developed at GreenDelta, in a German research project “Entwicklung eines autodidaktischen Data Mining Algorithmus mit hierarchischen temporalen Speicher (HTM) zur Interpolierung des […] Footprints von Produkten“, 2016-2018.
1st presentation about the project at the SETAC Brussels conference, May 8 2017:
Ciroth, A., Srocka, M.: Introducing the LCA data machine, presentation, SETAC Europe 2017, Brussels
LCA data machine, principles
The LCA data machine takes data from various sources (raw data, outside of the LCA domain), combines Shannon entropy and information about the fitness for purpose for a to be created data set, filtering, refactoring and adding information, and calculates a goodness of fit indicator to determine whether the created data set is already fit for purpose. Result is a core data set.
Hundreds of sources are used in this context, and various rule sets are applied. The rules are reviewed by experts.
The core data sets are then adapted to meet specifc requirements, such as: being PEF compliant, or using a certain allocation rule, in a next step. More details are available in the presentation, under resources.
LCA data machine, outlook
Plan is to establish LCADM as one LCA data source.
Not as competition to the existing LCA data sources, but as addition, to satisfy a broader need, to create data sets faster, for more widespread uses.
A first version of the machine is to be expected for end of 2017.
Project duration: 2016-2018. HTM and the LCA data machine are currently developed by GreenDelta, in a German, independent research project. Support of the German ministry for energy and the economy is gratefully acknowledged.