Patrik Kennes

Local Linear Mapping Neural Networks for Pattern Recognition of Plants

Abstract A new neural network architecture for classification purposes is proposed. The Self-Organizing Map (SOM) Neural Network is used in a supervised way for a classification task. The neurons of the SOM become associated with Local Linear Mappings. Error information obtained during training is used in a novel learning algorithm to train the classifier. The proposed method achieves fast convergence and good generalisation. The classification method is then applied in a precision farming application, the classification of crops and weeds using spectral properties. The classification performance of the proposed method is proven superior compared to other statistical and neural classifiers.

Comparison of Different Passive Horizontal Sprayer Boom Suspension Set-Ups

Abstract Existing boom suspensions are effective to reduce rolling boom motions, but reduction of horizontal boom vibrations is negligible. A prototype passive horizontal suspension is developed and different suspension set-ups are compared with regard to boom vibrations and spray pattern homogeneity. Boom vibrations are simulated with a non-linear finite element program. The corresponding spray pattern is obtained by combining the static spray pattern with the simulated boom movements. An appropriate passive suspension is able to almost halve the CV of spray deposition compared to a fixed boom.

Neural Network Based System Identification of Agricultural Machinery

Abstract A new method for on-line system identification based on the Self-Organizing Map is presented. The standard Self-Organizing Map (SOM) is extended with Local Linear Mappings. To every node in the SOM along with the input weight two output weights are assigned: one that stores the output part of an input-output pair and one that stores the local gradient matrix (Jacobian) that is calculated from the training pairs. A training algorithm for the Jacobian matrices is derived. The method is tested in system identification of two Agricultural Machines: a flexible Spray Boom and a shaker with a nonlinear spring.

Evaluation of a Non-Linear Finite Element Model to Simulate Vibrations of a Small Sized Sprayer Boom

Abstract The efficiency of crop treatments depends greatly on the homogeneity of the spray distribution. Boom vibrations cause local under- and over-applications. This paper describes how boom movements can be simulated with a non-linear finite element program for a 12-meter working width sprayer. Special attention must be paid to the dynamic characteristics of the hinges because they are detenninative for horizontal boom vibrations. The model is validated on a test rig that reproduces horizontal tractor vibrations. Boom velocities are measured by a laser and fit well with the simulated velocities.

Modelling and Optimization of the Dynamic Behaviour of Sprayer Booms

Abstract Finite element simulations show that existing suspensions approximately halve vertical sprayer boom movements compared to a fixed boom. The reduction of horizontal boom vibrations however is negligible. This shows the need for a performant horizontal suspension. Because of the high mass and inertia of sprayer booms active suspensions would require too much engine power. Therefore the possibilities of a passive suspension are investigated. Optimal suspension settings (spring stiffness, damping constant) are determined by optimization of the singular value plot derived from a parameterized linear multi-body model of the boom and suspension. Validation experiments confirm that the used modelling techniques are reliable.