moncler online Least squares support vector machin

2n2sale160ad

Least squares support vector machine parameter optimization and its application

Into the local minimum and other shortcomings, this paper used method to improve it additional momentum, combined with neural network toolbox of Mat-lab simulation was carried out. For this 3-layer feedforward neural network, hidden layer neuron number is 10, the initial weights and thresholds using Nguyen-Widrow (referred to as the N-w method) method to find out. Training simulation 30,000 times the temperature curve and error curve shown in Figure 8 and Figure 9. Genetic algorithm and PSO algorithm for optimization of support vector machine after the BP neural network with simulation results in Table 2 by several performance indicators to quantify. MAXPE the largest positive error. An acridine J. A state =- s! grapefruit 4 {i ~ one by one 282 East Manager 1 - University (Natural Science) Vol.34 MAXNE the maximum negative error, RMSE the root mean square error. Figure 8, the true temperature simulation curves and temperature curves (BtNN method) Fig. 8CurveofsimulationandrealtemperaturewithBP-NN0.5, -,[link widoczny dla zalogowanych], 0 stems. . 0.5 �� a 0l00200300400500600SamplesDJ1eStll error curve in Figure 9 (BP-NN method) Fig. 9Curveoferror (BP-NN) Table2ComparisonofperformanceErroroftrainingMAXPEMAXNERMSEErroroftestingMAXPEMAXERMSE Table 2 Comparison of performance parameters can be seen from the table, the parameters of the LS-SVM optimized simulation results better than the BP neural network, in particular the root mean square error of the key performance indicators; this A parameter optimization method is applied to soft sensor Texaco Gasifier, although the increased complexity of the algorithm, the corresponding computation time also increases,[link widoczny dla zalogowanych], however,[link widoczny dla zalogowanych], Texaco Gasifier Soft measurement Temperature measurements are selected and the range of each secondary variable has a strong similarity, therefore, the initial simulation algorithm can be applied to future applications in soft sensor modeling to reduce the range of parameters and parameter optimization significantly reduce the time searching, algorithm and application portability is also relatively strong. 5 conclusions will be the introduction of evolutionary class support vector machine algorithm parameter optimization, can effectively carry out parameter optimization. Texaco temperature from the actual policy of soft sensor modeling examples can be seen, the use of genetic algorithms and Ps () algorithm for parameter optimization have achieved a better simulation results,[link widoczny dla zalogowanych], the project can meet the actual needs; such method can be a test of error method more efficient approach to the optimal value, to solve this problem. Support vector machine optimal adjustment of the parameters of Support Vector Machine is still a priority, how to apply advanced algorithms to parameter optimization of support vector machines in the study will remain a hot topic.

相关的主题文章：

louis vuitton outlet Determination of viscosity of

[link widoczny dla zalogowanych]

dgsdf Since the spread of fused ceramic coating fo