Dynamic measurement errors prediction for sensors based on firefly algorithm optimize support vector machine

Combustion State Recognition of Flame Images Using Radial Chebyshev Moment Invariants Coupled with an IFA-WSVM Model

“…First, two images of each state were selected as the temporary set. Then, each image of temporary set was rotated by angels ±15 • and ±20 • , translated with translation vectors [−35, −25] and [35,25], scaled with scaling factors of 0.5, 0.8, 1.2 and 1.5. Consequently, a RTS transformation set was obtained including 600 (8 × 75) images.…”

“…It can be observed that the proposed multi-feature vectors are really invariant to rotation, translation and scaling. [35,25] and scaled with 1.2. Table 5.…”

“…In the third case study, the effect of changing in size of training set on the recognition rate was observed. There were 10,15,20,25,30,35 and 40 images of each state randomly selected from the RTS transformation set as the training sets. Accordingly, the rest of images transformed by RTS composed seven corresponding testing sets.…”

“…Therefore, in the process of approaching the optimal solution, it is possible to appear that the displacement of a firefly is greater than the distance from the optimal solution. As a result, the firefly may skip the optimal solution when updating its location with Equation (25). If this happens multiple time in a row, it will have an effect on the searching ability and the convergence rate.…”

“…Support vector machine (SVM) was first proposed by Vapnik [21], in accordance with the foundation of statistical theory and the principle of structural risk minimization. Among the existing AI technologies, SVM has been proved to have superior performance over previously developed methodologies, such as decision tree, artificial neural network and other conventional statistical models [22][23][24][25].…”