Periodic Reporting for period 1 - Pythia (Artificial Intelligence to predict and control the behavior even in the most complex industrial processes)
Okres sprawozdawczy: 2019-07-01 do 2019-09-30
The “process break”, an error or halt in production flow, is a major problem that the automated manufacturing industry faces today. This is because each process break event costs enterprises up to hundreds of thousands of euros in wasted resources. The manufacturing industry is thus undergoing dramatic digitalisation and automation to collect vast amounts of manufacturing process data across hundreds of conditions and variables. These data contain all of the information needed to optimise processes, reduce downtime and minimise waste. However, no methods currently exist to process and interpret all of this information in real-time. Effective in operando break prevention for complex manufacturing industries (e.g. paper, steel, pharma, chemical) is thus impossible. These manufacturing industries are consequently currently limited adopters of AI-based control systems, because today they cannot see clear economic advantages from an investment into today’s process break prevention solutions. A solution, which could provide effective and efficient break prediction, prevention and process control to afford manufacturers noticeable savings, is thus needed.
The fundamentally new mathematical model behind our innovation, Pythia, and its easy integration, make it universal to any machine and factory (e.g. paper, steel, chemical, pharma). Nevertheless, to build a proof-of-concept system we started with the process-break-prone paper producing industry. The overall objectives of the Pythia project, to reproducible guarantee effective break prevention with 90% justification accuracy, 30 minutes ahead of time, have been determined to be: 1) development of features for Pythia usability (i.e. data connectors, integration, user interface and application platform); 2) core Pythia technology development and optimisation (i.e. counteracting, automatic model adaption and missing and faulty data detection); 3) external validation of Pythia with pilot partners in paper and steel manufacturing; and 4) effective deployment and execution of our commercialisation, communication and dissemination strategies.
The fundamentally new mathematical model behind our innovation, Pythia, and its easy integration, make it universal to any machine and factory (e.g. paper, steel, chemical, pharma). Nevertheless, to build a proof-of-concept system we started with the process-break-prone paper producing industry. The overall objectives of the Pythia project, to reproducible guarantee effective break prevention with 90% justification accuracy, 30 minutes ahead of time, have been determined to be: 1) development of features for Pythia usability (i.e. data connectors, integration, user interface and application platform); 2) core Pythia technology development and optimisation (i.e. counteracting, automatic model adaption and missing and faulty data detection); 3) external validation of Pythia with pilot partners in paper and steel manufacturing; and 4) effective deployment and execution of our commercialisation, communication and dissemination strategies.
Technical: a detailed work plan has been developed, describing the steps needed to optimise Pythia capabilities, whilst demonstrating its usability with live data in paper and steel manufacturing environments. We have determined that this will allow us to demonstrate Pythia genericity, accuracy as well as build an application platform for effective Pythia deployment in any manufacturing industry, thus ensuring its uptake. Potential risks were analysed, and mitigation and corrective measures outlined.
Commercial: the paper and steel manufacturing markets were analysed in line with Internet of Things and predictive maintenance market trends. Our strategy and business model was assessed within this market, whilst taking into account the competitor landscape. This has allowed us to determine realistic commercialisation, communication and dissemination strategies, tailored specifically to the manufacturing industry market.
IP and Legal: we thoroughly analysed the IP landscape, which showed that the Pythia project will not infringe any IPR and we are therefore free to operate globally. Steps towards protecting our IP through several software IP safeguarding mechanisms (e.g. licensing code, obfuscation and native compiling language selection) have been outlined, with preparations made for exploring future patenting options.
Financial: we assessed the investment/resources needs for Pythia’s development and full market launch by Q1-2022, as well as quantified the expected profitability and payback period of the project. We confirmed financial projections to 2025, allowing us to cement a viable business plan, ensuring the profitability and high ROI of Pythia.
Commercial: the paper and steel manufacturing markets were analysed in line with Internet of Things and predictive maintenance market trends. Our strategy and business model was assessed within this market, whilst taking into account the competitor landscape. This has allowed us to determine realistic commercialisation, communication and dissemination strategies, tailored specifically to the manufacturing industry market.
IP and Legal: we thoroughly analysed the IP landscape, which showed that the Pythia project will not infringe any IPR and we are therefore free to operate globally. Steps towards protecting our IP through several software IP safeguarding mechanisms (e.g. licensing code, obfuscation and native compiling language selection) have been outlined, with preparations made for exploring future patenting options.
Financial: we assessed the investment/resources needs for Pythia’s development and full market launch by Q1-2022, as well as quantified the expected profitability and payback period of the project. We confirmed financial projections to 2025, allowing us to cement a viable business plan, ensuring the profitability and high ROI of Pythia.
PerfectPattern presents Pythia, the first AI technology, able to process many different streams of unordered incoming data from various sensors to create a real-time dynamic model of an entire factory, leveraging continuously incoming information. The fundamentally new mathematical model behind Pythia and its easy integration will allow it to be universal to any machine and therefore factory. Based on desired targets, Pythia learns how to optimise the global system within seconds, without supervision and requiring only a fraction of the computing power of more conventional AI approaches. With Pythia, for the first time, up to 80% of process breaks can be predicted and justified 30 minutes ahead of time, allowing for the effective optimisation of global manufacturing operations.
Pythia’s capabilities will deliver a positive impact on the economy, the environment and society by: 1) catalysing significant predictive maintenance market growth, surpassing current forecasts; 2) maximising process and resource efficiency indiscriminately across a multitude of industries and sectors, with potentially trillions in savings; 3) forcing the manufacturing industry to digitalise to improve the efficiency of operations to remain competitive in line with climate protection, and thus secure a significant boost in terms of societal favour and support.
Pythia’s capabilities will deliver a positive impact on the economy, the environment and society by: 1) catalysing significant predictive maintenance market growth, surpassing current forecasts; 2) maximising process and resource efficiency indiscriminately across a multitude of industries and sectors, with potentially trillions in savings; 3) forcing the manufacturing industry to digitalise to improve the efficiency of operations to remain competitive in line with climate protection, and thus secure a significant boost in terms of societal favour and support.