Research On High-frequency Power Supply Process Control Based On Waveform Matching And ARM Platform Simulation

This thesis focuses on the study of high-frequency power supply process control, a method basedon fast waveform matching is proposed to predict the forecast the zero-crossing point of primarycurrent and increase the output power of high-frequency power supply. Through theoretical researchon waveform matching, three algorithms of waveform matching are proposed in this thesis andverified in the MATLAB simulation platform. And one proposed algorithm is further validated on theARM platform, which provides a reference for realizing the proposed algorithm in practicalapplication.Firstly, the related knowledge of high frequency power supply of electrostatic precipitator isintroduced, and then the current situation and existing problems are analyzed; the concept and methodof image matching theory that is referenced by waveform matching is given, and then the ARMsimulation platform is briefly introduced.Secondly, the thesis introduces the main circuit topology of the high frequency power supply, andthen analyse the switch tube conduction waveform and primary current waveform of series-parallelresonant converter with capacitor filter in a switching cycle. In view of the current power supplyprocess control, the thesis puts forward a kind of improved high frequency power supply processcontrol method.Thirdly, this thesis references image matching algorithms, and then proposes a combinationalgorithm of the two algorithms in view of the advantages and disadvantages of the Hausdorffalgorithm and SSDA(Sequential Similarity Detection Algorithm) algorithm; then aiming at theshortcomings that the speed of ergodic searching is slow，the waveform matching method that basedon non-ergodic searching is proposed. The algorithms based on hill-climbing searching strategy andparticle swarm optimization algorithm to realize non-ergodic searching is proposed respectively.Thenlastly, this thesis simulates the primary current waveform collected form site by proposed algorithms;besides，the methods proposed in this thesis are compared with other classic algorithms. Thesimulation results demonstrate that the algorithms proposed are efficient and effective.Finally, one proposed algorithm is simulated on the ARM platform. First, the waveform data aresent to ARM using serial communication, then match the waveform through the waveform matchingalgorithm proposed, and last the waveform match result is displayed on the LCD. The results showthat the proposed method has a certain potential practical value.