DigiSpek - Digital lærende spekepølseprosess

The project will develop and implement a digitized and learning process and decision support system that provides more predictable production time, reduced waste, predictable cured sausage quality, increased employee involvement and better overall economy. Automatic collection of data from sensors and analyses using algorithms in the control systems are key to calculating more accurately when the various cured sausage batches will be finished. DigiSpek will combine the company's artisanal knowledge and methods, scientifically designed model systems and existing and new production data to determine optimal control variables and select suitable sensors for measurement and analysis. The data will be integrated into user-friendly digital production management and reporting solutions. By adapting sensor technology to the processes and collecting data for intelligent digital process support and process learning, one will form the basis for better overall economy and investment analyses for a more automated and industrialized cured sausage process. In H1, our main goal is to map the extent and effect of raw material and process variation, propose available measurement systems suitable for data collection and develop sensors for measuring water activity. Through several experiments, with close collaboration between H1, H2 and H6, the chemical composition of both minced meat and finished cured sausage has been measured. Variations during the production process were recorded. The test results have provided significantly better insight into the raw material variation over time, and they also provide a basis for the choice of measurement systems suitable for data collection in the factory. IRTA has contributed to the mapping work through discussions about which parameters are most important to have control over to achieve the best and most consistent quality for the end product. To improve production planning and provide decision support for the operators, we at H2 have developed a mathematical model that describes water transport through and out of the cured sausage. The model is based on a mass and an energy balance. Several experiments have been carried out to identify the various model parameters. The nominal model describes well how a typical cured sausage dries, but due to variation between sausages, there will be deviations. To monitor current drying processes, we have also developed automated scales that measure in real time (prototype). By using these, we can identify model parameters for relevant sausages online, and we can predict future drying processes with much higher accuracy. More work remains to be done on the model update algorithms before they can be used in a production setting. We also investigate the relationships between the different process steps and how this affects the drying process and model parameters. By measuring/estimating the properties in production, you will be able to get good predictions earlier. This is important as the usefulness of the predictions is directly related to how early you have them. In the project, we have also looked more generally at the management of production in the factory in Sogndal. We have carried out a survey of production with a focus on how it is organised, planned and managed. The survey provides a description of the current governance (AS-IS) as a basis for describing a desired future governance (TO-BE). Based on the mapping of the current production at Nortura Sogndal, we have identified possible improvement proposals that relate to planning, management and control of production - especially using decision support (the right information at the right time) for the manager and operator. The project maps and evaluates existing technological solutions for automation, with a particular focus on flexible systems that can be adapted to gradual implementation in the production of cured sausages. Key areas investigated include automated suspension and transport using robots. Furthermore, work is being done on the development of IoT-based methods and karakuri techniques to increase both digital competence and process optimization among operators, which will facilitate more efficient and future-oriented solutions. The operators who handle products in smoking rooms and climate rooms at Nortura Sogndal need to register data and obtain information in an efficient way. Rocketfarm has made observations of the current workflow, and interviews have been conducted with operators and management. A report has been prepared that highlight challenges related to manual registration of data and analogue reporting. The report discusses how to simplify the operators' working day by automating the measurement of key parameters and making this information available through digital solutions.

DigiSpek skal gjennom digitalisering av spekepølseproduksjonen gjøre produksjonstiden mer forutsigbar, redusere variasjonen, øke utbyttet og forbedre sluttkvaliteten. Prosjektet skal legge grunnlaget for hensiktsmessig datafangst for å utvikle en kontinuerlig lærende digital modell av spekepølseproduksjon som skal brukes til å overvåke og styre produksjonen (dashboard). Styringssystemet skal gjøres brukervennlig og underbygge bedriftens kultur for medarbeiderinvolvering, omstilling og modernisering. Forskningsutfordringene: 1 Kartlegge råstoff- og prosessvariabler og deres betydning for total og sammensatt prosessutvikling. Identifisere gode automatiserte metoder ved bruk av tilgjengelige og nyutviklede sensorer, blant annet prototyp for måling av vannaktivitet. Data output benyttes til å utvikle robuste statistiske forklaringsmodeller som gir presis prediksjon. 2 Benytte en datadrevet arbeidsprosess basert på Exploratory Data Analysis (EDA), hvor man iterativt undersøker hypoteser for å finne nye årsakssammenhenger og øke prosessforståelsen. Eksisterende kunnskap om råvare og prosess skal kombineres med produksjonsdata for økt prosessforståelse og derav bedre overvåking av tørkeforløpet. Ambisjonen er å predikere framtidig prosess- og tørkeforløp. 3 Utvikle et konsept hvor nødvendig informasjon for beslutninger blir operasjonalisert i et støttesystem for planlegging og oppfølgning. Dette skal også danne grunnlag for fremtidige automatiseringsløsninger. 4 Utvikle konsept for hvordan operatører selv kan foreslå og utvikle enkle, automatiserte og digitaliserte lavkost-implementasjoner av produksjonsforbedringer blant annet gjennom implementering av Karikuri (LEAN). 5 Utvikle og implementere en dashboard-løsning for styring av prosessene.