Ion process, which can distinguish the gross errors and phase jumps
Ion method, which can distinguish the gross errors and phase jumps correctly. Following preprocessing, the amplitude spectrums primarily based on distinct FFT lengths are compared, as well as the limitation of FFT in analyzing the time-varying periodic noise is discussed. To fully analyze the periodic variations, the QPM fitting residuals inside the time domain, the spectrum analysis results in frequency domain, plus the time-frequency analysis outcomes in time and frequency domain of three constellation satellites are given. By time-frequency evaluation of STFT, the periodic variations of BDS satellite clock offsets are characterized, along with the Pinacidil In stock relationships in between the periodic variations and sun elevation angle above the orbit plane ( angle) are investigated. Right after that, the frequency variations with the primary periodic term are analyzed in detail. Lastly, the clock prediction model is modified by taking into account the periodic variations of clock offsets, and also the TFAM is proposed. two.1. Preprocessing of Clock Offsets The preprocessing of raw clock offsets is of great importance for periodic variation analysis and clock prediction. The gross errors and phase jumps will be the most important anomalies from the clock offsets, which cannot objectively reflect the traits from the satellite clocks and degrade the efficiency of clock prediction [33]. Such clock anomalies must be processed just before periodic variation evaluation and clock offset modeling. With high calculation efficiency and anti-error performance, the MAD strategy is generally applied to detect gross errors [34]. When the clock frequency series satisfies Equation (1), the related clock offsets are viewed as abnormal:| yi | k(1)where yi Goralatide medchemexpress represents the clock frequency series, i could be the epoch, and k will be the threshold value, which is often calculated by k = m + n MAD. m denotes the median of your clock frequency series, MAD = Median/0.6745, and also the aspect n may be set to 3. The conventional MAD method is usually a preprocessing method corresponding to clock frequency series. Each gross errors and phase jumps of clock offsets in time domain will bring outliers in clock frequency series, which means the outliers detected by regular MAD are caused by gross errors or phase jumps. If all the outliers are processed with out distinguishing them, some effective clock offsets may possibly be destroyed simultaneously. Consequently, it can be necessary to determine the outliers caused by gross errors or phase jumps. Within this paper, we use double MAD detection to distinguish the outliers. The detailed flowchart in the double MAD detection is shown in Figure 1. As shown in Figure 1, the double MAD detection performs two MAD detection. In the very first MAD detection, the raw clock offsets are converted to clock frequency series, along with the threshold value k is calculated. The outliers of clock frequency series could be easily detected by MAD detection. Then, retailer the clock offset outliers identified by the very first MAD detection temporarily and remove the frequency outliers. After that, the clock offset series with out gross errors are recovered by integral algorithm. Inside the second MAD detection, the new clock frequency series are recalculated, whilst the threshold k still uses the worth obtained by the very first MAD detection. In this case, the outliers of clock frequency series absolutely correspond towards the phase jumps, as well as the clock offsets, that are accidentally removed throughout the initial MAD detection, is usually recovered. Just after double MAD detection, the outliers caus.
