Back to search

FFL-JA-Forskningsmidlene for jordbruk og matindustri

KORNMO - produksjonsoptimalisering, kvalitetsstyring og bærekraft gjennom verdikjeden for korn

Alternative title: KORNMO - production optimization, quality management and sustainability across the grain value chain.

Awarded: NOK 7.4 mill.

Small grains form the very basis for the Norwegian agriculture sector. Felleskjøpet Agri?s (FKA) new small grain strategy emphasizes volume, quality and rationality as crucial for profitability and sustainability in the value chain. FKA is the dominant small grain purchaser and user in Norway. FKA wants to utilize this position to strengthen interaction across the value chain. By sharing insight and facilitating best practices with producers, FKA aims to increase crops through efficiency and productivity gains while achieving higher product quality and an improved market value of small grain. A unique feature of the agriculture sector is the large amount of data points being generated from various agricultural equipment and sources, ranging from soil data, weather data, input factor data, disease and pest data, etc. In KORNMO our goal is to link this information to a small grain delivery and thus create valuable insights down to an individual field level. Connecting all the data points related to a small grain delivery, we want to use machine learning and AI to find the most crucial variables in optimizing production. Furthermore, new insight will allow the individual small grain producer to improve quality management. Linking data to individual deliveries will improve traceability, enable declaration of quality and allow for segmentation of crops. Finally, linking of data to a delivery will form the basis for establishing individual black/green carbon accounts. Through this the farmer is given action-triggering insights into how sustainable the operations are compared to others, by measuring and comparing total energy consumption in production (black carbon) with the amount of energy produced through crops (green carbon). Such an approach gives Norwegian plant-producing farmers an opportunity to take ownership to the photosynthesis and showcase the farmer in the public debate as a driver in creating sustainable climate solutions for the future. Our focus in this project period has been to establish automatic data capture and set up a data plattform in Azure that forms the basis for piloting in the project. The project has also designed and completed research questions for the period and prepared the implementation of pilots. The project has established a good base for data collection, with a number of data sources from public data sources collected and established in the cloud platform. In this work, the project has had a number of meetings and established collaborations with several players in the industry, such as: NIBIO, the Norwegian Directorate of Agriculture, NLR and the Agricultural Climate Calculator. Our efforts are reflected in a number of data points that together form the basis for execution of the project. In order to be able to carry out pilots, the project has defined two user scenarios related to each of the three main value extractions in the project. For Production optimization, the following user scenarios are defined; 1) Benchmark for the farmer and 2) Advice for the farmer for production optimization. For Quality management, the following user scenarios are defined; 1) Quality management of reception and production and 2) Tracking in the value chain. For Sustainability, the following user scenarios are defined; 1) The manufacturer's own footprint and 2) Accumulated footprint for FKA's products. All user scenarios are detailed and a progressive plan for development / extraction has been prepared over the project period. The project have leveraged the collected data to create a machine-learning model that combines historical data from sattelites, temperatures, rainfall, and harvest yield in a neural network. The model can be used to predict the harvest yield and harvest date for the season. This knowledge is relevant for several of the user scenarios that we have defined.

I innovasjonsprosjektet KORNMO skal vi utvikle modeller for å optimalisere bærekraftig kornproduksjon, gjennom maskinlæring anvendt på data om driften av hvert skifte, om eksempelvis jordsmonn, vær, innsatsfaktorer og maskinering, samt avlingens volum og kvalitet. Prosjektet skal bygge ny innsikt og gi verdi for enkeltbonden, NLR og landbruksnæringen som helhet gjennom verdiuttak langs de tre aksene; i) produksjonsoptimalisering, ii) kvalitetsstyring og iii) bærekraft. Gjennom å utnytte de store mengdene data som allerede finnes, i kombinasjon med ny data og ny teknologi, skal forskningen i KORNMO bygge fundamentet for verdiøkende tjenester, produkter og rådgivning for bonden og NLRs øvrige kunder 1) Vi vil undersøke om satelittfoto egner seg som kilde til pålitelige data om dyrkede arealer, aktuell korntype, og maskineringsstrategi. 2) Vi vil undersøke om dype læringsalgoritmer er hensiktsmessige for prediksjon av produksjons-optimalisering og bærekraftsbehov 3) Vi vil undersøke muligheten for å applisere såkalte generative adversarial networks (GAN) for å generere data i tilfeller hvor data ikke eksisterer. Det er et åpent forskningsspørsmål om dette kan brukes til å generere gode nok data for agronomiformål 4) En annen forskningsutfordring er optimalisering mot flere mål. Å løse optimaliseringsoppgaver med flere mål er fremdeles et åpent forskningsspørsmål, siden målene kan ha flere motstridende løsninger Prosjektets funn og resultater vil gjennom NLR og dets rådgivingsnettverk komme hver enkelt kornbonde til gode, samt kunne komme til anvendelse i andre deler av den norske verdikjeden for korn. Forskningen forventes å ha overføringsverdi både til verdikjeden for korn i andre geografiske markeder og potensielt til lignende verdikjeder og optimaliseringsproblem innen landbruk (eks. dyrking av andre avlinger og melkeproduksjon) og andre sektorer. Kompetanse og erfaringer fra prosjektet vil forøvrig komme til anvendelse for å styrke produkt- og tjenestetilbudet hos prosjektets partnere.

Publications from Cristin

No publications found

Funding scheme:

FFL-JA-Forskningsmidlene for jordbruk og matindustri