- 标题
- 摘要
- 关键词
- 实验方案
- 产品
-
Dynamic Behavioral Modeling of RF Power Amplifier Based on Time-Delay Support Vector Regression
摘要： A new, dynamic behavioral modeling technique, based on a time-delay support vector regression (SVR) method, is presented in this paper. As an advanced machine learning algorithm, the SVR method provides an effective option for behavioral modeling of radio frequency (RF) power amplifiers (PAs), taking into account the effects of both device nonlinearity and memory. The basic theory of the proposed modeling technique is given, along with a detailed model extraction procedure. Unlike traditional artificial neural network (ANN) techniques, which take time to determine the best configuration of the model, the SVR method can obtain the optimal model in short time, using the grid-search technique. An example of an optimal SVR model selection applied to an RF PA is also given; the performance of the selected model presents a big improvement when compared with the default SVR model. Experimental validation is performed using an LDMOS PA, a single device gallium nitride (GaN) PA, and a Doherty GaN PA, revealing that the new modeling methodology provides very efficient and extremely accurate prediction. Compared with traditional Volterra models, canonical piecewise linear models, and ANN-based models, the proposed SVR model gives improved performance with reasonable complexity. In addition, it is shown that the model can predict accurately the behavior of the PA under input power levels that are different from those under which it is extracted.
关键词： time delay,radio frequency (RF) power amplifiers (PAs),machine learning,Dynamic behavioral model,support vector regression (SVR)
更新于2025-09-23 15:23:52
-
Improved measurement on quantitative analysis of coal properties using laser induced breakdown spectroscopy
摘要： It is of great significance to realize the rapid or online analysis of coal properties for combustion optimization of thermal power plants. In this work, a set of calibration schemes based on laser-induced breakdown spectroscopy (LIBS) was determined to improve the measurement on quantitative analysis of coal properties, including proximate analysis (calorific value, ash, volatile content) and ultimate analysis (carbon and hydrogen). Firstly, different normalization methods (channel normalization and normalization with the whole spectral area) combined with two regression algorithms (partial least-squares regression [PLSR] and support vector regression [SVR]) were compared to initially select the appropriate calibration method for each indicator. Then, the influence of de-noising by the wavelet threshold de-noising (WTD) on quantitative analysis was further studied, thereby the final analysis schemes for each indicator were determined. The results showed that WTD coupled SVR can be well estimated calorific value and ash, the root mean square error of prediction (RMSEP) were 0.80 MJ kg?1 and 0.60%. Coupling WTD and PLSR performed best for the measurement of volatile content, the RMSEP was 0.76%. For the quantitative analysis of carbon and hydrogen, normalization with the whole spectral area combined with SVR can get better measurement results, the RMSEP of the measurements were 1.08% and 0.21%, respectively. The corresponding average standard deviation (RSD) for calorific value, ash, volatile content, carbon and hydrogen of validation sets were 0.26 MJ kg?1, 0.57%, 0.79%, 0.47% and 0.08%, respectively. The results demonstrated that the selection of appropriate spectral pre-processing coupled with calibration strategies for each indicator can effectively improve the accuracy and precision of the measurement on coal properties.
关键词： partial least-squares regression (PLSR),quantitative analysis,normalization,Laser-induced breakdown spectroscopy (LIBS),coal properties,support vector regression (SVR),wavelet threshold de-noising (WTD)
更新于2025-09-19 17:13:59
-
Pipeline leakage identification and localization based on the fiber Bragg grating hoop strain measurements and particle swarm optimization and support vector machine
摘要： A pipeline's safe usage is of critical concern. In our previous work, a fiber Bragg grating hoop strain sensor was developed to measure the hoop strain variation in a pressurized pipeline. In this paper, a support vector machine (SVM) learning method is applied to identify pipeline leakage accidents from different hoop strain signals and then further locate the leakage points along a pipeline. For leakage identification, time domain features and wavelet packet vectors are extracted as the input features for the SVM model. For leakage localization, a series of terminal hoop strain variations are extracted as the input variables for a support vector regression (SVR) analysis to locate the leakage point. The parameters of the SVM/SVR kernel function are optimized by means of a particle swarm optimization (PSO) algorithm to obtain the highest identification and localization accuracy. The results show that when the RBF kernel with optimized C and γ values is applied, the classification accuracy for leakage identification reaches 97.5% (117/120). The mean square error value for leakage localization can reach as low as 0.002 when the appropriate parameter combination is chosen for a noise‐free situation. The anti‐noise capability of the optimized SVR model for leakage localization is evaluated by superimposing Gaussian white noise at different levels. The simulation study shows that the average localization error is still acceptable (≈500 m) with 5% noise. The results demonstrate the feasibility and robustness of the PSO–SVM approach for pipeline leakage identification and localization.
关键词： pipeline leakage localization,method of characteristics (MOC),FBG hoop strain sensor,support vector regression (SVR),particle swarm optimization (PSO) algorithm,support vector machine (SVM)
更新于2025-09-04 15:30:14