The past two years I have conducted an extensive literature and tool review to answer the question: “What should software engineers learn about building production-ready machine learning systems?”. During my research I noted that because the discipline of building production-ready machine learning systems is so new, it is not so easy to get the terminology straight. People write about it from different perspectives and backgrounds and have not yet found each other to join forces. At the same time the field is moving fast and far from mature. My focus on material that is ready to be used with our bachelor level students (applied software engineers, profession-oriented education), helped me to consolidate everything I have found into a body of knowledge for building production-ready machine learning (ML) systems. In this post I will first define the discipline and introduce the terminology for AI engineering and MLOps.
LINK
To avoid energy scarcity as well as climate change, a transition towards a sustainable society must be initiated. Within this context, governmental bodies and/or companies often note sustainability as an end goal, for instance as a green circular economy. However, if sustainability cannot be clearly defined as an end goal or measured uniformly and transparently, then the direction and progress towards this goal can only be roughly followed. A clear understanding of and a transparent, uniform measuring technique for sustainability are hence required for sustainable and circular (renewable) energy production pathways (REPPs), as society is asking for an integrated and understandable overview of the decision-making and planning process towards a future sustainable energy system. Therefore, within this dissertation, a new approach is proposed for measuring and optimizing the sustainability of REPPs; it is useful for the analysis, comparison, and optimization of REPP systems on all elements of sustainability. The new approach is applied and tested on a case based on farm-scale, anaerobic digestion (AD), biogas production pathways.
LINK
A transparent and comparable understanding of the energy efficiency, carbon footprint, and environmental impacts of renewable resources are required in the decision making and planning process towards a more sustainable energy system. Therefore, a new approach is proposed for measuring the environmental sustainability of anaerobic digestion green gas production pathways. The approach is based on the industrial metabolism concept, and is expanded with three known methods. First, the Material Flow Analysis method is used to simulate the decentralized energy system. Second, the Material and Energy Flow Analysis method is used to determine the direct energy and material requirements. Finally, Life Cycle Analysis is used to calculate the indirect material and energy requirements, including the embodied energy of the components and required maintenance. Complexity will be handled through a modular approach, which allows for the simplification of the green gas production pathway while also allowing for easy modification in order to determine the environmental impacts for specific conditions and scenarios. Temporal dynamics will be introduced in the approach through the use of hourly intervals and yearly scenarios. The environmental sustainability of green gas production is expressed in (Process) Energy Returned on Energy Invested, Carbon Footprint, and EcoPoints. The proposed approach within this article can be used for generating and identifying sustainable solutions. By demanding a clear and structured Material and Energy Flow Analysis of the production pathway and clear expression for energy efficiency and environmental sustainability the analysis or model can become more transparent and therefore easier to interpret and compare. Hence, a clear ruler and measuring technique can aid in the decision making and planning process towards a more sustainable energy system.
LINK
Author supplied: In a production environment where different products are being made in parallel, the path planning for every product can be different. The model proposed in this paper is based on a production environment where the production machines are placed in a grid. A software entity, called product agent, is responsible for the manufacturing of a single product. The product agent will plan a path along the production machines needed for that specific product. In this paper, an optimization is proposed that will reduce the amount of transport between the production machines. The effect of two factors that influence the possibilities for reductions is shown in a simulation, using the proposed optimization scheme. These two factors are the redundancy of production steps in the grid and the
DOCUMENT
Complex situations and systems can be studied by using adequate models in simulation. An important aspect of models and the simulation software is the ability to use a wide range of possible input parameters. The simulation described in this paper is based on agile manufacturing by using transport robots and cheap reconfigurable production platforms, called equiplets. This setup makes agile manufacturing of different products in parallel possible.The simulation showed the maximum load of such a production environment as well as a proof of concept for the distributed approach for transport used. (from IEEE) / paid version: https://doi.org/10.1109/ISADS.2017.54
DOCUMENT
This chapter discusses how to build production-ready machine learning systems. There are several challenges involved in accomplishing this, each with its specific solutions regarding practices and tool support. The chapter presents those solutions and introduces MLOps (machine learning operations, also called machine learning engineering) as an overarching and integrated approach in which data engineers, data scientists, software engineers, and operations engineers integrate their activities to implement validated machine learning applications managed from initial idea to daily operation in a production environment. This approach combines agile software engineering processes with the machine learning-specific workflow. Following the principles of MLOps is paramount in building high-quality production-ready machine learning systems. The current state of MLOps is discussed in terms of best practices and tool support. The chapter ends by describing future developments that are bound to improve and extend the tool support for implementing an MLOps approach.
LINK
From the article: Manufacturing technology can improve the turnover of a company if it enables fast market introduction for volume production. Reconfigurable equipment is developed to meet the growing demand for more agile production. Modular reconfiguration, defined as changing the structure of the machine, enables larger variation of products on a single manufacturing system; these solutions are called Reconfigurable Manufacturing Systems (RMS). The quality of RMS, and the required resources to bring it to reliable production, is largely determined by a swift execution of the reconfiguration process. This paper proposes a method to compare alternatives for the ways to implement reconfiguration. Three classes of reconfiguration are defined to distinguish the impact of the proposed alternatives. The procedure uses a recently introduced index method for development of RMS process modules. This index method is based on the Axiomatic Design theory. Weighing factors are used to calculate the resources and lead time needed to implement the reconfiguration process. Application of the method leads to quick comparison of alternatives in the early stage of development. Successful execution of the method was demonstrated for the manufacturing process of a 3D measuring probe.
MULTIFILE
We review the current body of academic literature concerning gamification of production and logistics. The findings indicate that production execution and control has been addressed most often in the current body of literature, which consists mostly of design research. Objectives and goals, points, achievements, multimedial feedback, metaphorical/fictional representations, and levels and progress are currently most often employed gamification affordances on this field. The research has focused on examining or considering motivation, enjoyment and flow as the main psychological outcomes of gamification in the given context, while individual performance and efficiency are the most commonly examined or suggested behavioral/organizational impacts. Future studies should employ more rigorous study designs and firmly ground the discussions in organization theory.
LINK
Sustainable consumption is interlinked with sustainable production. This chapter will introduce the closed-loop production, the circular economy, the steady state economy, and Cradle to Cradle (C2C) models of production. It will reflect on the key blockages to a meaningful sustainable production and how these could be overcome, particularly in the context of business education. The case study of the course for bachelor’s students within International Business Management Studies (IBMS) program at three Universities of Applied Science (vocational schools), and at Leiden University College in The Netherlands will be discussed. Student teams from these schools were given the assignment to make a business plan for a selected sponsor company in order to advise them how to make a transition from a linear to circular economy model. These case studies will illustrate the opportunities as well as potential pitfalls of the closed loop production models. The results of case studies’ analysis show that there was a mismatch between expectations of the sponsor companies and those of students on the one hand and a mismatch between theory and practice on the other hand. The former mismatch is explained by the fact that the sponsor companies have experienced a number of practical constraints when confronted with the need for the radical overhaul of established practices within the entire supply chain and students have rarely considered the financial viability of the "ideal scenarios" of linear-circular transitions. The latter mismatch applies to what students had learned about macro-economic theory and the application through micro-economic scenarios in small companies. https://www.springer.com/gp/book/9783319656076 LinkedIn: https://www.linkedin.com/in/helenkopnina/
MULTIFILE
Innovations are required in urban infrastructures due to the pressing needs for mitigating climate change and prevent resource depletion. In order to address the slow pace of innovation in urban systems, this paper analyses factors involved in attempts to introduce novel sanitary systems. Today new requirements are important: sanitary systems should have an optimal energy/climate performance, with recovery of resources, and with fewer emissions. Anaerobic digestion has been suggested as an alternative to current aerobic waste water treatment processes. This paper presents an overview of attempts to introduce novel anaerobic sanitation systems for domestic sanitation. The paper identifies main factors that contributed to a premature termination of such attempts. Especially smaller scale anaerobic sanitation systems will probably not be able to compete economically with traditional sewage treatment. However, anaerobic treatment has various advantages for mitigating climate change, removing persistent chemicals, and for the transition to a circular economy. The paper concludes that loss avoidance, both in the sewage system and in the waste water treatment plants, should play a key role in determining experiments that could lead to a transition in sanitation. http://dx.doi.org/10.13044/j.sdewes.d6.0214 LinkedIn: https://www.linkedin.com/in/karel-mulder-163aa96/
MULTIFILE
