A Method for Detecting Abnormal Changes in the Temperature Field of Grain Bulk Based on HSV Features of Cloud Maps

Abstract

HighlightsAbnormal grain temperature changes were detected by calculating the similarity of HSV features in cloud maps.The F-measures were higher for the improved method than for methods based on HSV and LBP feature similarity.The improved method can detect abnormal heating of grain due to mold activity or spoilage and the fluctuation in grain temperature caused by aeration.The temperature field of the grain bulk in adjacent time periods has high similarity during normal storage.Abstract. Analyses of grain temperature data are time-consuming and labor-intensive, and thorough analyses are difficult to perform. This article proposes an improved method based on the hue, saturation, and value (HSV) color feature similarity of temperature field cloud maps to detect abnormal changes in grain temperature. Historical grain temperature data are preprocessed to generate temperature field cloud maps. The improved method based on HSV feature similarity is used to calculate the similarity of temperature field cloud maps for two successive days during normal storage, and a similarity threshold is set. Five types of grain bulk temperature anomalies are then simulated. Additionally, a comparative experiment is carried out that considers traditional methods based on HSV feature similarity and local binary pattern (LBP) feature similarity. The results show that the average recall rates of the F-measures of the improved method, the traditional method based on HSV feature similarity, and the method based on LBP feature similarity are 96.2%, 89.3%, and 95.4%, respectively, and the processing speeds are 340, 300, and 690 ms per group, respectively. Finally, an abnormal grain temperature experiment is carried out. The experimental results show that the improved method can detect sudden changes in the temperature field due to mold activity or spoilage and the fluctuations in grain temperature caused by aeration. Keywords: Grain storage, HSV feature, LBP feature, Similarity, Temperature.