Driving assistance: agri-food application

The expertise developed by a company's employees often constitutes a real asset for the company. Over time, as staff come and go, this expertise may remain stable, grow, or, conversely, decline if it is not formalized. It may therefore be essential for the company to capitalize on this expertise in a way that allows it to be exploited in the future or even passed on to other employees. The "corporate memory" thus concerns, a priori, all sectors of the economy, and several knowledge capitalization initiatives have been carried out. Examples include: Usinor-Sacilor for the development of a computer tool that optimizes the operation of blast furnaces based on the knowledge of blast furnace operators (SACHEM system), SNCF for the development of guides for detecting damage to tunnels, bridges, etc., the LCPC for the production of tools to assist in the detection of road damage, EDF and Cemagref for the monitoring and diagnosis of dam failures (CD-Rom Vigie BarragesÒ), and the CEA for the capitalization of knowledge and experience acquired in the field of nuclear reactors (REX project).

In the food industry, the operators who carry out the process are generally key players in product quality control, and their expertise is a valuable asset to the company. This is linked to two main points.

On the one hand, there are not always direct or indirect instrumental measures available to evaluate the properties of a product during manufacture. The development or use of measuring devices is difficult due to the properties:

• of the product itself: food products are complex biological products whose nature and composition may vary over time; furthermore, they are not always homogeneous, and the evaluation system must take this heterogeneity into account and even reflect it;

• sensors and measuring instruments:

  • Sensors must be cleanable, and it is also desirable that they be non-invasive and non-destructive.

  • they must meet strict requirements such as cleanability or operation in very humid environments,

  • The operation of certain measuring devices may require personnel trained in their use.

  • A number of measurement systems cannot be used directly online because response times are too long.

  • Criteria such as cost, metrological characteristics (accuracy, reliability, speed, etc.) and development time must be taken into account.

On the other hand, food processes are generally not robust when it comes to variations in raw material characteristics, for example. Understanding of the interactions between...