Diagnosis model of threshing cylinder blockage condition based on hybrid sparrow search algorithm and support vector machine

• Published in: Computers and electronics in agriculture Vol. 237; p. 110660
• Main Authors: Ling, Gaomin, Zhang, Luke, Liu, Wanru, Lyu, Ziwei, Xu, Hongmei, Wu, Qing, Zhang, Guozhong
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.10.2025
• Subjects: Encircling of prey; Intelligent fault diagnosis; Multi-sensor; Orca predation algorithm; Sparrow search algorithm; Threshing and separating device; Multi-sensor; Threshing and separating device; Intelligent fault diagnosis; Sparrow search algorithm; Encircling of prey; Orca predation algorithm
• ISSN: 0168-1699; 1872-7107
• DOI: 10.1016/j.compag.2025.110660

•A hybrid optimized sparrow search algorithm is designed.•A method is proposed to monitor threshing cylinder working condition based on multi-sensor information fusion.•A fault diagnosis system for the threshing cylinder of the combine harvester was designed. A hybrid optimized sparrow search algorithm was proposed as a solution to the problem of low accuracy of intelligent early warning and diagnostic analysis system for rice threshing cylinder blockage. The training cost of the algorithm was high due to the difficulty of collecting the blockage data of the threshing cylinder. Compared with other meta-heuristic algorithms, this algorithm could significantly reduce the sample size and improve the training accuracy. The algorithm commenced with the random generation of the sparrow population position, followed by the optimization of the initial position of each sparrow using Tent mapping. This process was undertaken to ensure the preservation of population diversity and to enable the population to escape local optima. Subsequently, the population was sorted, and the current optimal sparrow individual position and the best fitness value were obtained. Concurrently, the safety threshold, designated as safety threshold, undergoes a reduction in proportion to the rise in evolution algebra, thereby enhancing the algorithm’s search capability and accelerating its convergence speed. Meanwhile, utilizing the encircling of prey of orca predation algorithm, the optimal sparrow position of the previous evolution is disturbed, which is beneficial to jump out of the local optimal solution and improve the global search ability. Then the improved sparrow search algorithm is used to optimize the support vector machine. If the maximum number of iteration terminations is satisfied, the value of the optimal fitness and the best position information are output. Finally, using the optimal parameters, the support vector machine model of threshing and separating device blockage fault diagnosis is established, and the results are identified and classified. A threshing rolling blockage diagnosis system was designed based on the fundamental operational parameters of the threshing cylinder under varying rice feeding conditions. The system comprises Hall speed sensors, an acceleration vibration sensor, a dynamic torque sensor, an angular displacement sensor, a main control module, a microcontroller and a terminal display. According to the theoretical design of the machine, the feeding amount was categorized into three operating states: normal, early warning and blocking. The operating parameters of the three stages of the threshing cylinder were collected by each sensor, and the relevant data sets were established. The hybrid optimization algorithm was then used to establish the fault diagnosis model of the threshing cylinder, which can accurately identify the current operating state of the threshing cylinder. The classification diagnosis achieved 100 % accuracy, meeting the requirements for fault diagnosis of longitudinal axial flow threshing separation device. Compared to other diagnostic methods, the accuracy was improved by 33.3 %, 8.3 %, 25 %, 16.7 %, and 8.3 %, respectively. This can serve as a reference for efficient intelligent diagnosis of agricultural machinery equipment faults.