Baseline Tracking in Bipolar Charge Signal Process by Dynamic Time Warping Algorithm

Abstract

Electrostatic charge of solid particles can cause problems in many handling processes and need to be evaluated in terms of charge levels and charge polarity. An inductive charge sensor is suitable for the evaluation of both levels and distributions of particle charges at the same time. However, the performance of charge sensing is critically subject to its signal process, which can result in huge errors. One of error sources is drifting of the baseline tracked, which leads raw signals generated by the sensor to be distorted. Especially in determining polarity and quantity of bipolar charges, the distorted signal leads to significant biases and errors in charge measurements, when the number of particles measured is big. Currently, the existing correction algorithms cannot produce a satisfied result in baseline tracking. In this paper, the baseline drifting problem for the charge signals has been explored according to the types of charge polarity. For unipolar charge signals, charge polarity and quantity are determined directly by a poles-pairing method without any further baseline correction. For bipolar charge signals, a new method in baseline tracking and correction has been developed based on dynamic time warping algorithm. Further optimization by a double-check process is used to remove the ‘small hump’ errors in the signal. With the results of the charge measurements, this new method shows significant advantages on accuracy and efficiency of charge detection compared to the other existing methods.