A fault detection method based on orthogonality for integrated navigation system

Improving the reliability of integrated navigation systems requires effective fault detection and isolation strategies. In response to false alarms in jump soft fault detection grounded on the orthogonality principle, this study proposes refinements to the approach for data selection within a sliding window. The refined method introduces an improved orthogonality fault detection algorithm, which enhances detection accuracy and reduces the likelihood of false alarms. By harnessing data within the sliding window optimally, the algorithm computes the orthogonal average as the test statistic without necessitating an increase in the sliding window length. Furthermore, by combining the traditional residual chi-square test and extrapolation chi-square test, more precise judgments can be made regarding fault type and occurrence time. Ultimately, through the integration of traditional residual chi-square and residual extrapolation chi-square detection, this methodology enhances fault classification capabilities, reduces false alarm rates, and enables the effective identification of pulse, abrupt, and gradual faults even in scenarios with limited fault signal strength.