Diffusion processes driven by Brownian Motion (BM) provide a probabilistic framework for failure time estimation and remaining useful life (RUL) prediction, where the key quantity is the first hitting time density (FHTD) of a degradation trajectory crossing a failure threshold. In prognostics and health management (PHM), existing FHTD computations for RUL prediction are commonly based on first-order tangent approximation, which uses only local boundary information and may incur large errors under nonlinear degradation. To address this limitation, this paper develops a constructive and error-controlled method for FHTD approximation and failure time estimation under nonlinear diffusion degradation by adapting transformation-based reduction, successive approximation, and error propagation analysis. A direct stochastic differential equation (SDE) transformation converts the original threshold-crossing problem into an equivalent first hitting problem of a standard BM with a nonlinear time-varying boundary. A successive high-order approximation scheme based on Durbin’s series incorporates boundary-shape information neglected by first-order approximations. Computable error bounds are then derived by applying Banach fixed-point theory to the successive approximation and propagating the resulting FHTD error to associated failure time statistics, including the mean time to failure (MTTF) and variance of time to failure (VTTF), enabling accuracy-driven truncation order selection. Simulations and real-data case studies are evaluated against analytical, high-accuracy numerical, and empirical references to validate the proposed method. Results show clear advantages over first-order methods in MTTF and VTTF estimation accuracy, especially in strongly nonlinear degradation scenarios.

Radio frequency (RF) circuits play a crucial role in numerous fields such as communication, radar, and navigation. However, due to their high operating frequencies, they are prone to failures under the influence of environmental factors and parasitic parameters. Existing methods for diagnosing RF circuit faults are mainly based on deep learning approaches. But the limited number of internal measurement points and the large variety of fault patterns result in complex network structure design and difficulties in application. In this manuscript, a novel general feature enhancement method based on genetic programming (GP) is proposed to improve the machine learning-based RF circuit fault diagnosis. Firstly, the time-frequency analysis of the fault signal is carried out based on the Variable Mode Decomposition-Hilbert (VMD-Hilbert) transform to obtain the original feature set. Then, the feature reconstruction method based on GP is used to achieve feature enhancement. Finally, the enhanced features are combined with machine learning algorithms to realize the fault diagnosis of RF circuits. Taking the experiment of a low-noise amplifier circuit as an example, after adopting the feature enhancement method in this manuscript, the diagnostic accuracies of Support Vector Machine and Naive Bayes are increased by 10.48

RF circuits are the core of wireless communications, radar, satellite communications, etc., responsible for the transmission, reception, and processing of high-frequency signals, which are vital to the performance of modern communications and electronic equipment. Fault diagnosis is an important means to improve the reliability and maintainability of RF circuits. Aiming at the problem of scarcity of composite fault samples faced by traditional diagnostic methods in multi-fault concurrent scenarios, this paper proposes an innovative zero-sample composite fault diagnostic framework. The method constructs a two-stage intelligent diagnosis system: firstly, the mapping relationship from single-fault feature space to composite fault domain is established through conditional generative adversarial network (CGAN), and the pseudo-composite fault samples with physical interpretability are generated by using the adversarial training mechanism, which can effectively simulate the nonlinear coupling effect between real faults; then a multi-branch binary classifier architecture based on convolutional neural network is designed for co-training by fusing single fault samples and pseudo-composite fault samples containing the target fault type. The accurate identification of composite faults is finally realized by the parallel inference mechanism of multiple classifiers. The method is validated in an experiment with a RF receiver channel, and the experimental results show that the method can effectively solve the problem of exponential explosion of composite fault sample types by modeling the nonlinear coupling relationship between different single faults, which provides an alternative for the diagnosis of composite faults in complex electronic systems.

Degradation modeling and remaining useful life (RUL) prediction are key techniques in prognostics and health management (PHM). However, existing frameworks driven by Brownian Motion struggle to effectively integrate nonlinear degradation features or accurately separate degradation state from the inherent model noise and observation noise, leading to prediction inaccuracies. To address these limitations, a multivariate adaptive Brownian Motion-Generic Particle Filter framework is proposed. The framework enhances prediction accuracy and reduces uncertainty by comprehensively coordinating the three stages of RUL prediction. In degradation modeling stage, multiple drift terms are introduced to accommodate complex nonlinear degradation patterns, while the first-order difference with nonlinear least square (NLS) provides accurate and robust initial parameter estimation. In parameter update stage, a state-noise coupled discrete state-space model is constructed, which considers the uncertainty of all hidden variables, enabling real-time adjustment of parameter distributions based on degradation data. Hidden variables are adaptively updated using two particle filter strategies tailored to different dimensions, mitigating particle degeneracy and improving the separation of true degradation state. In RUL solution stage, the RUL distribution is extended into a stochastic process to quantify prediction uncertainty. Simulation experiments demonstrate the framework’s ability to switch particle filter strategies based on dimensionality and illustrate its overall workflow. Validations on the mechanical torsion bars dataset and the MOSFET dataset demonstrate a 31.5 % improvement in average prediction accuracy over the baseline method.

Aiming at the difficult problem of sample scarcity in the composite fault diagnosis of RF circuits, this paper proposes a few shot learning method based on a semantic matrix. The method constructs a feature layer and a semantic layer. In the feature layer, a cascade architecture of variational mode decomposition (VMD) and convolutional neural network (CNN) is used to extract highly discriminative time-frequency features from RF time-domain response. In the semantic layer, a semantic matrix generation algorithm based on greedy search is designed to construct a single-fault semantic matrix through uniqueness constraints and combinatorial anti-resistant constraints. A composite fault modeling is carried out using the bit or operation. The mapping of single-fault samples from the feature layer to the semantic layer provides basic features, and a few number of composite fault samples are used to model the relationship from single-fault semantics to composite fault semantics. A greedy search algorithm is used in the testing phase to obtain the optimal matching of the composite fault components. Experiments show that the diagnostic accuracy is improved by 13.34% with only 2 samples for each type of composite fault.

The remaining useful life (RUL) prediction of RF circuits is an important tool for circuit reliability. Data-driven-based approaches do not require knowledge of the failure mechanism and reduce the dependence on knowledge of complex circuits, and thus can effectively realize RUL prediction. This manuscript proposes a novel RUL prediction method based on a gated recurrent unit–convolutional neural network (GRU-CNN). Firstly, the data are normalized to improve the efficiency of the algorithm; secondly, the degradation of the circuit is evaluated using the hybrid health score based on the Euclidean and Manhattan distances; then, the life cycle of the RF circuits is segmented based on the hybrid health scores; and finally, an RUL prediction is carried out for the circuits at each stage using the GRU-CNN model. The results show that the RMSE of the GRU-CNN model in the normal operation stage is only 3/5 of that of the GRU and CNN models, while the prediction uncertainty is minimized.

Life prediction of RF circuits can greatly improve the reliability of RF systems. But most literature has been focused on the life prediction of RF devices, which is not applicable to RF circuits. Therefore, a novel RF circuit life prediction method is proposed based on an improved recurrent broad learning system (RBLS). First, a new feature matrix is proposed to characterize a RF circuit at each moment. Then, an improved RBLS model is used to predict the life of the circuit, in which the RBLS model is used to predict the feature vector and then the extreme learning machine (ELM) is used to expand the feature vector into a feature matrix. The method is validated in a low-noise amplifier with classical GRU, BLS-ELM, BLS, RBLS, ELM, and LSTM as a control group. The analysis results show that RBLS-ELM has the highest prediction accuracy with an RMSE of only 2.5959, the smallest prediction uncertainty of 1.2767, and a very short prediction time of 0.4251 s.

RF (radio frequency) circuits have been widely used in radar systems, communication systems, cellular phones, etc. It is extremely important to ensure the proper operation of RF circuits and fault prediction is an important means to evaluate the reliable operation of RF circuits. Current fault prediction methods can usually predict the health state of a circuit at the next time step with high prediction accuracy. However, more often than not, circuit operators want to know if the circuit will operate properly in the next period, i.e., they need to know the health state of the circuit for the next multiple time steps. The manuscript proposes a GRU network based on adversarial training to implement RF circuit fault multi-step prediction, which can greatly reduce the prediction error accumulated in multi-step prediction using a recursive strategy. The core of the method is to use adversarial training to make the distribution of the time series generated by the multi-step prediction as close as possible to the data distribution of the series in the training set. Firstly, the GRU network is trained to achieve single-step prediction of RF circuit faults. Second, a recursive strategy is used to perform multi-step prediction. Finally, the GRU network is trained adversarially with a discriminator to determine whether the prediction series belongs to the training set; if it does not, the optimization of the GRU network continues; if it does, the prediction series of the network is shown to have been tuned and its distribution has been essentially identical to that of the training set. The method is validated in a low-noise amplifier circuit, experimental results show that when the prediction time step increases, the method greatly slows down the decreasing trend of the model performance. The RMSE and MAPE are higher than that of the traditional GRU model, and R2 is also the higher one, indicating that the method has the highest prediction accuracy and the best fit.

In order to meet the urgent demand for civil aviation talents in China, this paper takes the construction of civil aviation major in Changzhou Institute of Technology as an example, and studies the development status of the education mechanism of the modern integration of enterprises with vocational schools and universities for civil aviation major. This paper summarizes the development experience of the integration of enterprises with vocational schools and universities in foreign countries, and combined with the development status of China's civil aviation industry, it is suggested to build a semi-practice-semi-employment practice platform, and the government should grasp the baton of integration of enterprises with vocational schools and universities to achieve a win-win-win situation of students, schools and enterprises.

The remaining useful life (RUL) prediction is one of the core technologies in the field of reliability engineering. In recent years, the emergence of long-life and high-reliability electronic equipment has posed a challenge to the life prediction of such equipment. Reliability data obtained alone by experimental tests or field observations have limitations. When we use data fusion methods, we can make full use of different types of data, save test cost, and improve prediction accuracy. At present, the commonly used data fusion methods are to merge multi-source data such as failure time information and degradation data under multi-stress conditions. Electronic devices have the characteristics of high cost of test execution and extensive sources of reliability data. For such characteristics, this paper presents a method that combines simulation data with measured data, The simulation data can be used as a priori information for reliability assessment of electronic equipment. When reliability tests are not conducted. In the absence of reliability test, the life distribution of the equipment can be approximated by simulation information, and the life distribution of the device can be updated according to the measurement data after a small amount of measurement information is obtained. The above idea can be realized by Bayes method. Firstly, we need to obtain the degradation/failure information of functional parameters of electronic equipment based on the degradation of key components in them. The linear Wiener process is used to model the degradation curve and inverse Gaussian distribution to model the failure-time data, then, we can get the prior distribution of model parameters under different simulation conditions. Secondly, to get some data in reality, we need to carry out reliability tests (ALT, ADT). Finally, the MCMC method is used to obtain the parameter estimation and confidence intervals, the correlation between parameters and the smoothness of the martingale chain need to test. After all these work, we give the remaining life prediction of the equipment and its reliability function. Sallen-Key bandpass filter is used to verify the practicality of the method and compare it with the extant models: (1) Self-contrast with the present model using only a small amount of field information or only simulation information, and the results show that the confidence interval for predicting the remaining life by combining the two sources of information is the narrowest at the same confidence level. The accuracy of prediction is higher than those two models. (2) Compared with the time series prediction models ARIMA and recurrent neural networks RNN, LSTM, GRU, the outstanding advantage of this method is that it can give an explicit expression of the degradation process of electronic equipment affected by stress-time, which can better solve the problems of overfitting of the training set and accumulation of multi-step prediction errors of the above-mentioned models, besides the sacrificing part of the training set fitting accuracy can be exchanged for higher test set accuracy, which can be better used in engineering practice.

A large number of carbon fiber reinforced polymers have been applied to aircraft and automobiles, and many nondestructive testing methods have been studied to detect their defects. Eddy current magneto-optical imaging nondestructive testing technology has been widely used in the detection of metal materials such as aircraft skin, but it usually requires a large excitation current and, at present, can only detect metal materials with high conductivity. In order to take full advantage of the innate benefits and efficiency of eddy current magneto-optic imaging and enable it to detect defects in carbon fiber reinforced polymers with weak conductivity, it is necessary to improve the magnetic field response of the eddy current magneto-optic imaging system and explore suitable excitation and detection methods. The scanning eddy current magneto-optical imaging nondestructive testing device built in this study has improved the magnetic field response of the system, and the eddy current magneto-optical phase imaging testing method has been proposed to detect the crack defects of carbon fiber reinforced polymers. The effectiveness of the method has been verified by simulation and experiment.

A low noise amplifier (LNA) is one of the key circuits of RF integrated circuits and RF systems, and it is prone to failure due to external interference, so it is necessary to carry out fault diagnosis. At present, LNA's reliable operation is mainly guaranteed through the testing of multiple performance parameters. The testing cost is high and the testing time is long. Moreover, the coupling relationship between each component and each test parameter is not considered, so it is difficult to isolate the faults. The multi-signal model uses a directed graph to describe the propagation dependence of faults, which is a very effective method in electronic system-level fault diagnosis. In this paper, a multi-signal model is proposed to describe the propagation dependence between the LNA fault modes and each test parameter, then a multi-core parallel diagnosis algorithm based on the multi-signal model and OpenMP is proposed for the rapid diagnosis of RF circuits. ADS is used to simulate the LNA circuit based on ATF54143. For the proposed multi-core parallel diagnosis method, the fault coverage rate can reach 100% and the fault isolation rate can reach 81.8% for a single fault, and the diagnosis efficiency can reach 2.033 under 4 threads.

This paper proposed a high-sensitivity phase imaging eddy current magneto-optical (PI-ECMO) system for carbon fiber reinforced polymer (CFRP) defect detection. In contrast to other eddy current-based detection systems, the proposed system employs a fixed position excitation coil while enabling the detection point to move within the detection region. This configuration effectively mitigates the interference caused by the lift-off effect, which is commonly observed in systems with moving excitation coils. Correspondingly, the relationship between the defect characteristics (orientation and position) and the surface vertical magnetic field distribution (amplitude and phase) is studied in detail by theoretical analysis and numerical simulations. Experiments conducted on woven CFRP plates demonstrate that the designed PI-ECMO system is capable of effectively detecting both surface and internal cracks, as well as impact defects. The excitation current is significantly reduced compared with traditional eddy current magneto-optical (ECMO) systems.

Fault diagnoses and prognostics are important tools to improve system reliability [...]

In order to improve the reliability, operational readiness and system safety of equipment, testability should be seriously considered in the design stage. As an important part of design for testability, test sequence generation is a binary identification problem because a minimal expected cost testing procedure must be developed in order to determine the amount of possible failure sources, if any, are present. Many algorithms have been proposed, but the generation time is long or the test cost is high when dealing with a large-scale dependency matrix. To address this issue, we propose a general enhancement method based on the SVM, the ECA* and the Monte Carlo. It can be applied to any existing algorithm and can effectively improve the performance. The available tests are classed based on the SVM according to the information of nodes, the ECA* is used to cluster states, and the morpho-logical function of the test sequence is obtained through the Monte Carlo simulation. All this information is fused to dynamically adjust the scale of the dependency matrix and selected to modify the parameters. Experiments show that the existing algorithms have shorter calculation time and lower costs because the information is considered more comprehensively after enhancement.

The remaining useful life (RUL) prediction for hydrogen fuel cells is an important part of its prognostics and health management (PHM). Artificial neural networks (ANNs) are proven to be very effective in RUL prediction, as they do not need to understand the failure mechanisms behind hydrogen fuel cells. A novel RUL prediction method for hydrogen fuel cells based on the gated recurrent unit ANN is proposed in this paper. Firstly, the data were preprocessed to remove outliers and noises. Secondly, the performance of different neural networks is compared, including the back propagation neural network (BPNN), the long short-term memory (LSTM) network and the gated recurrent unit (GRU) network. According to our proposed method based on GRU, the root mean square error was 0.0026, the mean absolute percentage error was 0.0038 and the coefficient of determination was 0.9891 for the data from the challenge datasets provided by FCLAB Research Federation, when the prediction starting point was 650 h. Compared with the other RUL prediction methods based on the BPNN and the LSTM, our prediction method is better in both prediction accuracy and convergence rate.

Sequential fault diagnosis is a kind of important fault diagnosis method for large scale complex systems, and generating an excellent fault diagnosis strategy is critical to ensuring the performance of sequential diagnosis. However, with the system complexity increasing, the complexity of fault diagnosis tree increases sharply, which makes it extremely difficult to generate an optimal diagnosis strategy. Especially, because the existing methods need massive redundancy iteration and repeated calculation for the state parameters of nodes, the resulting diagnosis strategy is often inefficient. To address this issue, a novel fast sequential fault diagnosis method is proposed. In this method, we present a new bottom-up search idea based on Karnaugh map, SVM and simulated annealing algorithm. It combines failure sources to generate states and a Karnaugh map is used to judge the logic of every state. Eigenvalues of SVM are obtained quickly through the simulated annealing algorithm, then SVM is used to eliminate the less useful state. At the same time, the bottom-up method and cost heuristic algorithms are combined to generate the optimal decision tree. The experiments show that the calculation time of the method is shorter than the time of previous algorithms, and a smaller test cost can be obtained when the number of samples is sufficient.

Lithium-ion batteries (LiBs) are the most important part of electric vehicle (EV) systems. Because there are two different degradation rates during LiB degradation, there are many two-phase models for LiBs. However, most of these methods do not consider the randomness of the changing point in the two-phase model and cannot update the change time in real time. Therefore, this paper proposes a method based on the combination of the two-phase Wiener model and an extreme learning machine (ELM). The two-phase Wiener model is used to derive the mathematical expression of the remaining useful life (RUL), and the ELM is implemented to adaptively detect the changing point. Based on the Poisson distribution, the distribution of the changing time is derived as a gamma distribution. To evaluate the theoretical results and practicality of the proposed method, we perform both numerical and practical simulations. The results of the simulations show that due to the precise and adaptive detection of changing points, the proposed method produces a more accurate RUL prediction than existing methods. The error of our method for detecting the changing point is about 4% and the mean prediction error of RUL in the second phase is improved from 4.39 cycles to 1.61 cycles.

Accurate prediction of the remaining useful life (RUL) of lithium-ion batteries can ensure the normal and effective operation of power systems using lithium-ion batteries. However, how to select battery prediction parameters through scientific methods and how to accurately predict battery RUL values under high and low temperature conditions are still a huge challenge. Thus according to the technique for order preference by similarity to ideal solution (TOPSIS) based on information entropy, improved particle swarm optimization (PSO) and moving average filter(MAF), a novel data-driven method for predict lithium-ion batteries' RUL is proposed. The TOPSIS method based on information entropy is proposed to select the best degradation parameters; a sliding average low-pass filter is used to solve the capacity regeneration and noise problem of the battery experimental data; the improved PSO algorithm is presented to predict the battery RUL accurately. Based on the batteries experimental data from NASA and University of Maryland, we have done many simulation experiments on parameters selection and RUL accuracy comparisons among several data-driven methods. The experimental results shows:(1) compared with the other prediction methods without degradation parameters selection, the proposed method with TOPSIS and MAF filtering is more accurate;(2) our proposed algorithm has higher prediction accuracy and use less training data than other data-driven algorithms;(3) this method has high prediction accuracy under both the high and low temperature conditions.

Prognostics of the remaining useful life (RUL) of lithium-ion batteries is a crucial role in the battery management systems (BMS). An artificial neural network (ANN) does not require much knowledge from the lithium-ion battery systems, thus it is a prospective data-driven prognostic method of lithium-ion batteries. Though the ANN has been applied in prognostics of lithium-ion batteries in some references, no one has compared the prognostics of the lithium-ion batteries based on different ANN. The ANN generally can be classified to two categories: the shallow ANN, such as the back propagation (BP) ANN and the nonlinear autoregressive (NAR) ANN, and the deep ANN, such as the long short-term memory (LSTM) NN. An improved LSTM NN is proposed in order to achieve higher prediction accuracy and make the construction of the model simpler. According to the lithium-ion data from the NASA Ames, the prognostics comparison of lithium-ion battery based on the BP ANN, the NAR ANN, and the LSTM ANN was studied in detail. The experimental results show: (1) The improved LSTM ANN has the best prognostic accuracy and is more suitable for the prediction of the RUL of lithium-ion batteries compared to the BP ANN and the NAR ANN; (2) the NAR ANN has better prognostic accuracy compared to the BP ANN.