Combine Harvester Low Crushing Rate Operation Strategy Research by Using Bayesian Network

Abstract

As the main harvesting machinery, the combine harvester is often due to improper adjustment of its operating parameters resulting in increased crushing rate and grain waste during the harvesting process. To quickly obtain the working range of key operating parameters under low crushing rate, this study conducted field tests on the relevant parameters affecting the crushing rate and finally selected the travel speed, feed rate, threshing drum speed, concave clearance, and crushing rate as node variables for the construction of the Bayesian network model. Based on the “search-and-score” algorithm, the best network structure can be obtained using the combination of the Akaike Information Criterion (AIC) scoring function and the hill-climbing method. In the obtained network, adjusting the proportion of the lowest level of the crushing rate nodes to 100 %, the operation strategy under the condition of low broken rate obtained by the network reasoning was: feed rate < 6 kg/s, travel speed < 5 km/h, concave clearance = 10 mm, threshing drum speed < 900 rpm. Three field trials were carried out using this optimized operation strategy, and the measured crushing rates were 0.93 %, 0.95 %, and 1.07 %, respectively, and the average crushing rate was 0.98 %. At the same time, when the optimized strategy was not used, the crushing rates were, respectively, 1.12 %, 1.41 %, and 1.93 %, and the average crushing rate was 1.48 %. The test results prove that the operation strategy based on Bayesian network inference can effectively reduce the crushing rate in the harvesting process.