To improve the navigation performance of the navigation system combining inertial navigation system (INS) and global positioning system (GPS) under complicated environments, especially GPS outages, a navigation method - wavelet neural network based on random forest regression (RFR-WNN) to assist adaptive Kalman filter (AKF) - is proposed. AKF is employed to correct INS errors, the Kalman filter is improved by introducing adaptive factor, to suppress the influence of the complex environment and random errors on the filtering accuracy; RFR-WNN is used to construct a high-precision prediction model when GPS works well, and to provide the required observations for AKF update when GPS outages. To solve the problem that the single neural network structure is easy to cause the overfitting, unstable and low prediction accuracy due to the lack of comprehensive training samples, RFR is introduced to optimize the single WNN, which can improve the generalization ability and prediction accuracy. In order to verify the effectiveness and advancement of the proposed method, vehicle navigation experiments were carried out, the results indicate that the proposed method has better navigation accuracy and performance than compared methods during GPS outages, and this advantage is more obvious in the case that fewer samples are collected.

Due to the vehicle’s severe maneuver and abnormal measurements of GPS in practical applications, the statistic of process noise in SINS/GPS integrated navigation system may be unknown and the measurement noise may not follow the Gaussian distribution, which results in a deteriorated performance for the conventional cubature Kalman filter. To address this issue, we propose in this paper a new robust cubature Kalman filter based on the adaptive information entropy theory. In the proposed filter, the process uncertainty and non-Gaussian measurement noise are simultaneously suppressed based on a new constructed cost function using the maximum correntropy and residual orthogonal principle based weighted least squares technology, which is independent of noise distribution and more insensitive to the non-Gaussian noise. Moreover, a multiple-channel adaptive strategy for the better process uncertainty suppression is given. Furthermore, some improvements are proposed to avoid the numerical problem and implement the proposed robust filter effectively. Extensive simulation and car-mounted experiment demonstrate that the proposed filter can achieve higher estimation accuracy and better robustness as compared with the related state-of-the-art methods.

In modern warfare, in order to destroy the target to largest extent, missile warhead is required to explode after penetrating inside the target, therefore, the measurement of mechanical signal during projectile penetration is critical. This paper studies the dynamic design principle of projectile penetration overload test system based on charge sensor, according to characteristics of steep front and a great number of contained installing structure impulse response, designs the measuring circuit based on charge sensor, conducts detailed theoretical deduction on dynamic model of sensor, charge-amplifier, filter and single conversion differential circuit, and provides theoretical basis for design. In order to more objectively evaluate the dynamic characteristics of this kind of test system, the dynamic uncertainty estimation of charge signal measured by this test system is considered as the standard for evaluating its dynamic characteristics, instead of unflatness of amplitude-frequency characteristic and its width in frequency domain, and the comprehensive evaluation on uncertainty of this test system is realized. In order to improve the measurement accuracy of dynamic test system designed, in this paper, an improved de-noising algorithm based on Ensemble Empirical Mode Decomposition (EEMD) is proposed, different low-pass filters are formed on the basis of Intrinsic Mode Function (IMF), de-noising effect of signal with noise is analyzed, and the evaluation indicators of dynamic uncertainty and approximation are established. Finally, the effectiveness of the designed method is verified through experiment.

In recent years, with the development of hyper-velocity weapon, the research focus of penetration effect has switched gradually from high-velocity to hyper-velocity, with insufficient studies on penetration effect and its internal mechanism. Besides, the study on penetration effect of hyper-velocity weapon depends on the measurement of mechanical signal during penetration of projectile. In this paper, dynamic test of hyper-velocity impact penetration acceleration signal is studied in depth. Firstly, the penetration velocity and acceleration of projectile under the impact load are analyzed, the influence of the movement of the testing unit on the measured acceleration signal is studied, and the factors to be considered when measuring the hyper-velocity impact load are given. Further, this paper designs a measurement circuit and dynamic models of each measurement module, details the influence of dynamic characteristics of power delivery network on test accuracy, and provides a method to obtain dynamic characteristics of power delivery network by virtue of S parameter. Finally, to extract the real penetration overload signal, a superior noise reduction algorithm adopting Modified Ensemble Empirical Mode Decomposition (MEEMD) is proposed, taking dynamic uncertainty and degree of approximation as evaluation indicators, and the measured sensor data is analyzed and processed with the proposed superior noise reduction algorithm. The experiment results show that, this method has a better noise reduction effect, and can be adopted to effectively extract real penetration overload signal. The results have theoretical research and application value to evaluate the reasonableness and effectiveness of measurement method of hyper-velocity penetration overload signal, and seek for a superior processing method for the noisy penetration overload signal.

The performance of integrated Global Navigation Satellite System and Inertial Navigation System (GNSS/INS) navigation often declines in complex urban environments due to frequent GNSS signal blockages. This poses a significant challenge for autonomous driving applications that require continuous and reliable positioning. To address this limitation, this paper presents a novel hybrid framework that combines a deep learning architecture with an adaptive Kalman Filter. At the core of this framework is a Temporal Convolutional Network and Bidirectional Long Short-Term Memory (TCN-BiLSTM) model, which generates accurate pseudo-GNSS measurements from raw INS data during GNSS outages. These measurements are then fused with the INS data stream using an Adaptive Noise-Regulated Iterated Extended Kalman Filter (ANR-IEKF), which enhances robustness by dynamically estimating and adjusting the process and observation noise statistics in real time. The proposed ANR-IEKF + TCN-BiLSTM framework was validated using a real-world vehicle dataset that encompasses both straight-line and turning scenarios. The results demonstrate its superior performance in positioning accuracy and robustness compared to several baseline models, thereby confirming its effectiveness as a reliable solution for maintaining high-precision navigation in GNSS-denied environments. Validated in 70 s GNSS outage environments, our approach enhances positioning accuracy by over 50% against strong deep learning baselines with errors reduced to roughly 3.4 m.

Calibrating Micro-Electro-Mechanical System (MEMS) gyroscopes can significantly improve the accuracy of projectile attitude perception in highly dynamic and thermally shocked environments. Gyroscopes calibrated using mainstream least squares methods experience significant performance degradation in terms of accuracy, while intelligent algorithms are unsuitable for projectile environments due to their "black box" nature and implementation complexity. To address the existing limitations, this paper presents a novel sensor calibration method based on dendritic neural networks and exponentially weighted moving average (EWMA) algorithms for gyroscopes in a projectile environment. The method combines the advantages of model-driven and data-driven techniques. First, a three-axis gyroscope calibration model was developed for highly dynamic flight environments. By introducing temperature terms, bias terms, and corresponding composite terms, the model accurately quantifies the effects of high rotation and sudden temperature changes on the gyroscope output. Second, the input-output recognition capability of dendritic neural networks is innovatively utilized to solve the complex coefficients in the model. Finally, the EWMA algorithm is applied to assign weights to data points, suppressing low-weight noise points. Semi-physical simulations were conducted on a three-axis temperature-controlled turntable and a high-speed flight simulation turntable. The experimental results demonstrate that this method effectively suppresses sensor zero drift caused by temperature, reduces noise and outliers in the sensor output, and enhances the accuracy of the sensor output. Compared to high-order least squares and LSTM algorithms, the RMSE value of the sensor output calibrated by this method improved by 40.7% and 19.9%, respectively.