The diesel engine, as the main power source of equipment, faces practical problems in the maintenance process, such as difficulty in fault location and a lack of preventive maintenance techniques. Currently, breakdown maintenance and cyclical preventive maintenance are the main means of maintenance support after a diesel engine failure, but these methods require professional maintenance personnel to carry out manual fault diagnosis, which is time-consuming. Prognostics and health management (PHM), as a new technology in the field of equipment maintenance support, has significant advantages in improving equipment reliability and safety, enhancing equipment maintenance support capability, and reducing maintenance support costs. In view of this, when introducing PHM into diesel engine maintenance support, the research progress and development trend of the key technologies of PHM for diesel engines are carried out with the objective of achieving precise maintenance and scientific management of diesel engines, and the key technologies demand traction. Firstly, the development history of PHM technology is reviewed, and its basic concept and main functions are introduced. Secondly, the system architecture of PHM for diesel engines is constructed, and its key technologies are summarized. Then, the research progress in the field of PHM for diesel engines is reviewed from four aspects: data acquisition, data processing, fault diagnosis, and health status assessment. Finally, the challenges faced by diesel engine PHM in engineering applications are analyzed, effective solutions to address these challenges are explored, and the future development trend is foreseen.

Regarding the difficulty of extracting fault information in the faulty status of UAV (unmanned aerial vehicle) engines and the high time cost and large data requirement of the existing deep learning fault diagnosis algorithms with many training parameters, in this paper, a small-sample transfer learning fault diagnosis algorithm is proposed. First, vibration signals under the engine fault status are converted into a two-dimensional time-frequency map by multiple simultaneous squeezing S-transform (MSSST), which reduces the randomness of manually extracted features. Second, to address the problems of slow network model training and large data sample requirement, a transfer diagnosis strategy using the fine-tuned time-frequency map samples as the pre-training model of the ResNet-18 convolutional neural network is proposed. In addition, in order to improve the training effect of the network model, an agent model is introduced to optimize the hyperparameter network autonomously. Finally, experiments show that the algorithm proposed in this paper can obtain high classification accuracy in fault diagnosis of UAV engines compared to other commonly used methods, with a classification accuracy of faults as high as 97.1751%; in addition, we show that it maintains a very stable small-sample migratory learning capability under this condition.

In order to solve the problem of the subtraction-average-based optimizer (SABO), which is difficult to effectively balance the local development and global search capability, golden sine SABO integrating multiple strategies (GSABO) is proposed. The improved Sine chaos mapping is introduced to refine the population initialization strategy of SABO in order to enrich the diversity of the population and improve the algorithm's search accuracy and speed. The position update method of SABO is improved by the golden sine algorithm to further enhance the local development and global search capability of SABO. The specific implementation steps of GSABO are described in detail, and the optimization ability of GSABO is tested using 23 benchmark functions. The test results show that the GSABO method is able to solve the optimization problem effectively compared with the current novel optimization algorithms, and has more excellent performance under most benchmark functions.