Probabilistic Control for Agricultural Machines

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Agricultural machines, such as combines, are difficult to operate due to complex settings that vary based on multiple criteria like crop type, weather, and field topology, and current deterministic control systems are inadequate in handling these complexities.

Innovation Solution

A probabilistic control system, specifically a Bayesian network control system, is implemented in agricultural machines to receive sensor inputs, compute posterior probability distributions, and generate control signals to optimize settings such as concave clearance, rotor speed, and grain separation, accommodating uncertainties and varying conditions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If deterministic control systems with lookup tables are used, then the control system structure is simple, but the system cannot handle uncertainties and varying conditions effectively

Engineering Contradiction:
Improvecontrol accuracy under varying conditionsVSAvoidcontrol system structure
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent transforms the deterministic control approach into a probabilistic control framework by changing the fundamental parameter representation from fixed lookup table values to probability distributions. The Bayesian network model represents control parameters (concave clearance, rotor speed, etc.) as probabilistic variables that can adapt to varying conditions, resolving the contradiction between reliability under uncertainty and system complexity.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent replaces the traditional deterministic control mechanism (lookup tables with fixed values) with a probabilistic reasoning system (Bayesian network). This substitution allows the system to handle uncertainties in crop conditions, weather, and machine state by computing posterior probability distributions, thereby improving reliability without requiring excessive complexity.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Adaptability or versatility

If multiple operator controllable settings are provided, then the machine can adapt to different crop types and conditions, but the machine becomes difficult to operate

Engineering Contradiction:
Improveadaptation to crop type and conditionsVSAvoidoperator complexity
Core Design Contradiction:
Adaptability or versatilityVSEase of operation

Solution Approach 1:

The probabilistic control system with Bayesian network enables the machine to automatically adjust its settings based on sensor inputs and computed probability distributions. The system performs self-optimization of parameters like concave clearance and rotor speed without requiring the operator to manually configure multiple settings, thereby maintaining adaptability while significantly improving ease of operation.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system continuously receives sensor inputs about crop conditions, machine state, and environmental factors, processes this information through the Bayesian network to update probability distributions, and automatically adjusts control parameters. This closed-loop feedback mechanism allows the machine to adapt to varying conditions while keeping the operator interface simple.

Inventive Principle:
Principle #23Feedback

3Reliability

If probabilistic control system is implemented, then the system can handle uncertainties and optimize performance, but the computational complexity increases

Engineering Contradiction:
Improvecontrol robustness under uncertaintiesVSAvoidcomputational complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The Bayesian network control system segments the complex probabilistic reasoning task into modular components representing different physical subsystems (threshing, separation, cleaning). Each node in the network represents a specific parameter or subsystem, and the modular structure allows efficient computation of posterior distributions by processing local dependencies rather than requiring exhaustive global calculations, thus managing computational complexity while maintaining robustness.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP3150049B1Probabilistic control of an agricultural machine
Publication Date: 2022.08.24 DEERE & CO
  • EP3150049B1 patent drawingFigure 1
  • EP3150049B1 patent drawingFigure 2
  • EP3150049B1 patent drawingFigure 3

AI summary

A set of sensor inputs (102-104) are received in an agricultural machine (100). The sensor inputs (102-104) are indicative of sensed or measured variables. A probabilistic control system (108) probabilistically infers values for another set of variables and generates a set of control signals based on the probabilistically inferred values.