Search

In order to improve the reliability of The Control Rod Drive Mechanism (CRDM), a reliability trade-off method was proposed, which aims at the total cost-reliability. Firstly, the Goal-Oriented (GO) model of the CRDM was established. Then, the total cost-reliability composed of Research and Development (R&D) production cost and operation and maintenance cost was used as the objective function of the model. Finally, based on GO methodology and Genetic Algorithm (GA), the reliability trade-off optimization allocation results of the CRDM were obtained, which proves the feasibility of the proposed method.

A step-by-step fault diagnosis method based on Empirical Mode Decomposition (EMD) combined with Random Forest algorithm was proposed for actual requirements of rolling bearing vibration fault diagnosis. Firstly, the preliminary fault monitoring was carried out, and a Linear Support Vector Machine model was established by extracting the Permutation Entropy of vibration signals as characteristic parameters to judge whether the bearing was faulty or not. Then, the fault location identification and the fault degree determination were carried out, and high-dimensional characteristic parameters in time domain, frequency domain and time-frequency domain are respectively extracted as inputs of the Random Forest algorithm. Finally, through the step-by-step diagnostic test of rolling bearing vibration data, the results show that each step of diagnosis can achieve 100% diagnostic accuracy and appropriate training time, which proves that EMD and Random Forest have good effect on step-by-step fault diagnosis of rolling bearing.

In order to solve the problem of non-stationary vibration signals of rolling bearings and large sample size, a small sample fault diagnosis method based on wavelet packet theory and Random Forest algorithm was proposed in this paper. Firstly, the signal was preprocessed, then three-layer wavelet packet decomposition was carried out, and the decomposition coefficients are extracted as feature vectors, which were used to train BP Neural Network, Support Vector Machine (SVM) and Random Forest model. Through the diagnosis of multi-fault states of rolling bearings, the diagnostic accuracy rates of the three groups of classifiers are respectively 25%, 93.75% and 100%, which proves that Random Forest has better diagnosis effect and robustness for rolling bearings with small samples.

A The combat command and control system is the nerve center to carry out joint operations. It is an important part of military operations in the information age and plays the role of "amplification multiplier" in the enemy operations. In this paper, a combat command network system was taken as the research object, and the reliability modeling and analysis of the combat command network were conducted based on dynamic Bayes-GO method. Firstly, the mapping rules for transforming common operators, type 18A and type 20 operators in GO method into dynamic Bayesian networks are given. Then, the task of the brigade command vehicle 2 commanding the combat vehicle 5 to strike the target in a combat command network was modeled and analyzed. Finally, the dynamic availability curve of the system within t = 300h and the availability corresponding to the system at each moment are calculated, which can enable the combat commander to clearly judge the reliability of the system in successfully hitting the target at each moment, thus the other combat equipment can be better dispatched for cooperative operations.