The Research For Selective Harmonic Elimination Technique Based On Neural Network

The inverter as a quite obvious energy saving effect power electronic device, which can drive the operation of AC motor, mainly for control speed, adjust the output voltage, current and change the output frequency, it is a kind of good Energy saving effect power electronic device. The core of the frequency converter is embodied in the inverter link, to a given dc input into adjustable alternating current output, often accompanied by a harmonic wave in the process of inverter, extremely adverse impact on the motor load is very difficult to achieve stable operation.The Selective harmonic elimination technique can effectively eliminate the specified order harmonic, to eliminate higher harmonic, add the appropriate circuit filter can effectively filter, so as to improve the quality of the output voltage and current.This technology as a basic principle of the designed of the selective harmonic elimination inverter output current waveform is close to sine wave, the operation of the ac motor got very big improvement.Firstly, the basic principles of selective harmonic elimination technique through the further analysis and research, using the Fourier series transformation of three-phase voltage type inverter output voltage waveform of harmonic elimination model.The mathematical model for the output base and every harmonic wave voltage switch Angle between current and voltage and nonlinear equations.The traditional Newton iteration algorithm can realize the switch angles, but more demanding for the initial value, and in the process of solving requires repeated iteration, calculate spend time is relatively long, industrial application, only the switch Angle are deduced with the method of offline, produced by the method of look-up table PWM control signal, the downside of this approach is need mass storage, and the data exchange of discrete control points gradient makes the control precision is affected.Looking for a way to quickly solve the switch Angle, in order to realize online solving method is imminent. This paper presents a BP algorithm was applied to selective harmonic elimination technique, the algorithm using training mature BP neural network switch Angle, only one equation is solved in the process of solving the switch Angle can be obtained, greatly improve the speed to solve.Second, under the dual polarity of selective harmonic elimination model simulation analysis, using Newton iterative algorithm, the basic BP algorithm and mind evolutionary algorithm to optimize BP algorithm, particle swarm algorithm to optimize BP algorithm switch Angle, observe its switch Angle SuiDiao system change curve, compare the phase voltage harmonic distortion rate of size, by comparing the two methods of optimization of BP network to the network convergence speed and the precision of the result analysis solutions,ultimately determine the using particle swarm algorithm to optimize the BP neural network method.The algorithm in solving the switch Angle not only faster convergence speed of BP network, and solve the result more accurate, spectrum analysis, observation under different switch Angle, modulation harmonic suppression, the results show that the selective harmonic elimination technique used in the particle swarm algorithm to optimize BP algorithm is feasible.Again, the design variable voltage variable frequency synchronous control of the selective harmonic elimination inverter, divided into two parts, power circuit and control circuit.Which used TMS320F28335 DSP control circuit, its main function is to solving the selective harmonic elimination technique online switch Angle, generating PWM control signal, A/D conversion signal collection.Power circuit of inverter adopts the three-phase voltage type circuit, inverter circuit is applied in the body of the DIP - IPM (intelligent power module), realized the inverter link of the inverter.Finally, through to the inverter hardware circuit debugging, complete the testing analysis of the experimental circuit output voltage waveform, solving the selective harmonic elimination technique using BP algorithm is verified the correctness of the switch angles.