Rotating Arc Current Signal Processing Based On Particle Filter And The PID Control Parameter Setting

China is an industrial country. Manual or semi-automatic welding is unable tomeet the requirement of the rapid development of petroleum industry, aviationindustry, maritime navigation, and automobile industry. Therefore, the research onwelding automation highlights the urgent. Effective noise reduction and featureextraction of welding signal plays an important role in self tracking of welding robotand welding automation. Welding signal contains various complex interference noisescaused by the factors such as short-circuit, arc welding, and wire feeding instability.Welding system is a nonlinear system. The traditional filtering methods such as meanfiltering, median filtering and wavelet filtering and so on, usually can not satisfy thefiltering requirement. The control system is the important part in welding automation.The PID control is the easiest, most practical and most widely used control method.Its effect has a direct relationship with its three parameters of proportional, integral,and differential. Aiming at the above existing problems in the automatic welding, wedo the following several aspects of research:The welding noise is divided into two categories: one is the measurement noise,the other is the process noise. Measurement noise is generated by the interference ofHall sensor, data acquisition card and the interference of sampling circuit. Processnoise is mainly composed of short-circuit, melting point splash and power instability.Signals collected without welding operation is the measure noise. We can achieve theexpectation value, the covariance, the distribution histogram, and the frequencyspectrum of the noise. These steps are to pave the way for the following signalprocessing and the profound research in the future.From the beginning of the Monte Carlo, Bayesian theory, to the development ofthe particle filtering algorithm, we are all introduced in detail. SNR (signal to noiseratio) before and after filtering is calculated by a typical nonlinear system to show theeffectiveness and advantages of the particle filtering method. By establishing theobservation equation and state equation of the welding system, we filter out thewelding current signal using the particle filtering and compare the filtering results with the result of median filtering, mean filtering, and wavelet filtering.We have introduced two methods to calculate the deviation, one is the integralmethod, the other is a kind of characteristic harmonic method. Integral method wasused to test to get the test deviation. After welding system obtained the weldingdeviation, the actuators were ordered to action and achieve tracking according to thedeviation. PID control is applied in the paper with particle swarm algorithm is usedfor PID parameters estimation. By establishing the welding model, determining thesearch space, selecting the fitness function, we have achieved the simulation responseof step signal and sine signal, and have compared the result with the Z-N method.