Loading...
Please wait, while we are loading the content...
Similar Documents
Fault Detection of Liquid-Propellant Rocket Engines Based on LSSVM
| Content Provider | Semantic Scholar |
|---|---|
| Author | Li, Ning-Ning Guo, Xiang Zhao, Song-Bo Xu, Liang |
| Copyright Year | 2018 |
| Abstract | In terms of fault diagnosis of liquid-propellant rocket engine, the fault diagnosis accuracy of the traditional method is low the characteristics of engine fault data are small sample and nonlinear variation. In order to improve the accuracy of sensor fault diagnosis and overcome the scarcity of related samples, the particle swarm optimization (PSO) with strong global searching ability is used to optimize the LS-SVM parameters, and the LSSVM optimal parameter values are obtained by iteration to improve the model fitting accuracy and generalization ability. At the same time, the traditional support vector machine and BP neural network model are used for the detection. The simulation results show that the least squares support vector machine (SVM) detection method based on particle swarm optimization has the advantages of high precision and high speed. It has certain effect and positive significance for improving the safety of liquid-propellant rocket engine test and engine failure loss. Introduction Aerospace technology is one of the most important achievements in the development of human science and technology in the 20th century. With the development of science and technology, more and more countries have joined the research on aerospace technology [1]. The United States, Russia, China and Europe have all launched their own manned space projects and space station construction projects, while Japan and India have also made great efforts to develop launch vehicle technology. Aerospace engineering involves many fields of technology and production. It is a concrete manifestation of a country's comprehensive national strength. The development of aerospace industry is of great significance to the comprehensive development of each country. In recent years, with the development of artificial intelligence technology, a large number of intelligent methods such as artificial neural networks, fuzzy theory, mixing algorithms, genetic algorithms, rough set theory, etc. have been introduced into the field of fault diagnosis. The most widely used are neural networks and supporting vector machines. Support vector machine (SVM) overcomes the difficulty of determining neural network structure and converges to local minima, and solves the problems of high dimension and nonlinearity. The least squares support vector machine replaces the inductive loss function in the SVM with the quadratic loss function, and the quadratic optimization of the algorithm in the original SVM is changed to solve the linear equation, which reduces the computational complexity and has better noise immunity, and faster operation speed. However, the kernel function parameters and normalization parameters of LSSVM have a significant impact on the classification performance of LSSVM. Based on the improvement of particle swarm optimization algorithm, the structural parameters of LSSVM are optimized and optimized, so that the particles can be guaranteed in the parameter optimization process, to enhance the ability to jump out of local optimal value, to find the optimal kernel functions and regularization parameters, and then improve the classification performance of LSSVM, accurately identify whether there is fault. |
| File Format | PDF HTM / HTML |
| DOI | 10.12783/dtcse/iece2018/26614 |
| Alternate Webpage(s) | http://dpi-proceedings.com/index.php/dtcse/article/viewFile/26614/26027 |
| Alternate Webpage(s) | https://doi.org/10.12783/dtcse%2Fiece2018%2F26614 |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
