Detection And Analysis System Of Arc Welding Based On ARM Microcontroller And USB Interface

Welding voltage and current can reflect behaviors and states of welding arc. It is an effective method that picking voltage and current signal to achieve real-time monitoring of weld process and evaluate welding quality. And continuous acquisition is necessary because of unpredictability of arc behavior. But cost and portability of current continuous acquisition system based on DAC (Data Acquisition Card) restrict employment of this method. This article analyzed existing acquisition and analysis means of welding electrical-parameters relative comprehensively. On this basis, employing 32-bit ARM microcontroller and USB interface, research were conducted regarding constructing continuous acquisition system based on single microcontroller; choice of USB transmission-mode; achievement of USB data-flow on target chip; key factor of continuous acquisition system; constructing queue-model based data buffer with on-chip memory in embedded microcontroller to replace dedicated FIFO or FPGA chip; priority level assignment of time-critical mission and effect of interrupt nesting upon acquisition speed; performance of data-flow store etc. Software of acquisition module on embedded microcontroller and analysis module on PC were programmed with C and LabVIEW respectively. Experiments were used to verify functions of acquisition-analysis system and integrity-and-correctness of data with oscilloscope and analysis module. Finally, the actual welding signal was used to compute characteristic parameters totaled 22 and perform waveform analysis, statistical analysis, U-I analysis etc. Furthermore, on the basis of modern analytical methods for non-stationary random signal, Short Time Fourier Transform (STFT) and Gabor expansion were introduced further from traditional analytical methods to analyze welding electrical signal.Results of study and experiments show that: (1) continuous acquisition system suit for arc weld can be built on single-chip microprocessor. Such a system can be used to execute real-time display and off-line analysis of data with the cooperation of PC. The cost of system is considerably lower than that of DAC. It is portable exceedingly with the help of USB interface. Continuous sampling frequencies that the study achieves are 70K in the case of two-channel and 140K in the case of one-channel. Only the capacity of hard drive can restrict the duration of acquisition. (2) The USB bulk transmission mode is the best mode for data acquisition. (3) The critical factor for continuous acquisition is a data buffer to balance the speed between data in and out system. (4) The "Soft FIFO"2 that based on queue model (a classic data structure of computer science) is constructed in on-chip memory in embedded microcontroller by means of program. The Soft FIFO can avoid the use of dedicated FIFO or FPGA chip. And the use of Soft FIFO can not only reduce the cost of acquisition system, but also improve the stability of system. The amounts of memory used relate to sampling frequency and interface speed. (5) Interrupt nesting has little effect on sampling frequency when appropriate USB transmission mechanism is employed. (6) Parallel circle in LabVIEW based on queue is an excellent method to store data flow. But better performance of data flow store needs better hardware platform. (7) JTFA (Joint Time-Frequency Analysis) can be used to analyze welding electrical signal. JTFA can recognize the point of arc burning, short-circuit, and the point of instantaneous short-circuit accurately without filtering the raw signal. Time-frequency resolution of Gabor expansion is better than STFT in the case of same window width.