Submitted by admin on Wed, 08/10/2022 - 15:11

Inadequate storage and maintenance leading to 6% crop loss annually

Sequence Number 4 Industry Agriculture Banner Inadequate storage and maintenance leading to 6% crop loss annually How 5G enabled

Optimise storage levels and improve equipment availability through predictive maintenance by collection of data on storage facilities and equipment.

Data Flows
Title Devices Icon Devices Description
  • For maintenance: Sensors to collect timeseries data from critical equipment
  • For storage facility: Sensors providing information on storage availability (levels)
Title Connectivity Icon Connectivity Description
  • Equipment (timeseries) data transmission
  • Information on storage facilities is sent to Enterprise storage
Title Edge Compute Icon Edge Compute Description
  • Not time-critical at present, but this is expected to change to real-time in the future
Title Cloud Compute & Storage Icon Cloud Compute & Storage Description
  • Equipment data stored to train AI algorithms
  • All data stored in Enterprise storage
Title Applications & Services Icon Applications & Services Description
  • AI (ML/MV) models with inputs from all relevant (timeseries) data to:
  • (ML) predict failure of critical equipment
  • (MV) forecast storage availability
Title Inform Decision Makers Icon Inform Decision Makers Description
  • Information and alerts of critical equipment predicted to fail so action can be taken to replace it in time
Title Support Decision Making Icon Support Decision Making Description
  • Overview of storage options
Application Logic
Description
  • Sensors installed on selected critical equipment (e.g. rotating equipment and equipment that would impact crop growth and/or harvesting if not available).
  • As a result, timeseries data will be collected every X seconds and sent to Enterprise storage.
  • Cameras are installed at each storage facility to monitor stock levels, and images will be sent to Enterprise storage for MV monitoring.
Description
  • All data collected is non-critical since it contains predictions of days ahead.
  • Over time, this could change to real-time when fault rectification of certain equipment is time sensitive.
  • Developing AI – ML/MV models to predict will take a number of iterations where accuracy of these models (predictions) will improve over time.
  • SME involvement working with data scientists is required to develop the models.
Description
  • The models will be able to predict potential failure of critical equipment and send triggers and alerts to rectify the failure before it happens.
  • Identifying standard images and training of the MV model will be required and ultimately be able to trigger alerts such as low stock levels, predict storage capacity, etc.
Expected benefits Key value created