In practical aerospace engineering, due to the high reliability and low failure probability, using the traditional Monte Carlo method to obtain accurate solution needs huge computation. The paper combines Latin Hypercube Sampling and Kriging surrogate model to get the minimum fitting error model by using the EFF learning mechanism for active learning INITIAL Kriging model, then uses Monte Carlo Method for reliability calculation, which greatly reduces the computation. At last, the method is verified by three numerical examples.

Aiming at reliability of pipeline connection and seal of reusable liquid rocket engine, combining with the development of a liquid rocket engine, this paper compares and analyzes the reliability requirements of one-time use engine and reusable engine from the aspects of mission profile, environment profile, life profile, functional requirements and structure change. Based on the analysis of the present situation of the reliability research of the pipeline connection and seal of the reused engine, this paper points out the deficiency existing at present, gives the reliability design and reliability verification direction of the pipeline connection and seal, and puts forward the technical roadmap of the pipeline connection and seal reliability. It can provide reference and inspiration for future general planning and development of reusable liquid rocket engine .

At present, the actual combat application concept of "Can fight and can win" is getting deeper and stronger. The requirements of complex battlefield environment adaptability and operational performance robustness of missile weapons are extremely harsh, and one single design cycle often cannot fully meet the actual combat needs. In view of the above problems, it is inevitable and necessary to explore the application and implementation path of aerospace product reanalysis, redesign and reverification (hereinafter referred to as "three-Re") in improving the reliability of missile weapons and equipment. This paper adopts logic decision principle (LDP) analysis, method of quality function deployment (QFD), ultimate tensile partial validation, etc., for missile weapons product design, production process control, environment risk identification. It can carry out the product risk control measures, improve the product reliability and maturity, and solve the problems existing in the missile weapons. The proposed method can guide the subsequent model improvement, research and design.

In order to improve the quality of products and process design, this paper mainly aims to eliminate the failure rate of initial products, improve their mean time between failures and further save product costs. Based on the theoretical requirements of HASS profile design, the paper presents the general screening profile conditions for electronic product HASS screening through specific failure analysis and test profile design optimization, which could be used as a reference for the HASS of electronic products and improve the screening test efficiency.

According to the principle of facing the troops and facing the actual combat, equipment reliability, as one of the important capabilities, is always in the working link of equipment life cycle test and identification. Considering that complex equipment systems usually go through multiple mission profiles after being delivered for use, and the environmental profiles and stress levels between different missions vary greatly, the orderer pays more attention to the mission reliability facing the actual combat conditions during the equipment in-service assessment. This paper presents an in-service evaluation method for the multi-mission reliability of complex equipment systems, taking the failure rate of equipment under different mission profiles as a classification index. When the failure rate under the multi-mission profile is the same, the evaluation model is given by using the principle of truncated test and Gamma distribution; when the failure rate of multi-mission profile is different, the law of large numbers and analytic method are used to model the reliability, and the corresponding evaluation method is given. Finally, an example shows that the method presented in this paper can provide a reasonable and feasible model and method for the in-service assessment of the mission reliability of complex equipment.

In order to reduce the safety risks caused by the difference in situation awareness in the interaction process between human and artificial intelligence agents, this paper has carried out research on human-machine system safety modeling method based on multi-agent and situation awareness. Firstly, the three-level model of situation awareness of artificial intelligence agents is constructed based on situation awareness. Based on this, the situation awareness error mode and the evolution process of situation awareness accidents are analyzed. Secondly, a human-machine system safety modeling method based on multi-agent and situation awareness is proposed. Finally, the effectiveness of the proposed method is verified by taking automatic driving as an example. The simulation results are consistent with the objective facts, which proves that the proposed safety modeling method is reasonable and effective. Keywords: artificial intelligence； safety of human-machine system； human-machine interaction； multi-agent； situation awareness； accident evolution

In order to solve the unreasonable problem that may exist in the establishment of support resource indicators in the process of equipment RMS indicator demonstration, and based on the common spare parts model in the process of maintenance, the utilization rate and satisfaction rate of spare parts are interpreted. Combined with the typical life distribution of spare parts prediction model and the experience of performance measurement of machine learning method, some of the comprehensive tradeoff of supportability is presented. The application analysis of the model is carried out with a practical case, and the results of parameter analysis are given. The results show that the comprehensive tradeoff method of support resource indicators based on machine learning performance measurement theory is practical in engineering and has significant theoretical advantages in comprehensive balancing, which can improve the efficiency of support resource indicator demonstration.

In order to improve the debugging efficiency and maintainability of missile products under missile-in-tube (MIT) test conditions, a software upgrade method is designed under MIT test conditions. This method does not need to disassemble the missile, and can be upgraded online under the condition of full missile. In this method, different CAN_ID numbers are designed to distinguish each node product in the system to complete the software upgrade of the node product. This method provides an upgrade method for product debugging personnel and maintenance personnel.

To solve the non-deterministic polynomial hard (NP-hard) problem of test selection in the design for testability of weapon system, an optimal test selection method based on simulated annealing-improved discrete particle swarm optimization (SA-DPSO) algorithm is proposed to acquire the best complete test set. This algorithm is on the basis of discrete particle swarm optimization (DPSO), and uses asynchronous dynamic learning divisors to obtain time-varying contraction factor, which facilitates the global searching speed, guarantees the convergence of DPSO, and abrogates the boundary constraint of particle velocity in DPSO. And the simulated annealing algorithm with probabilistic jumping ability is combined to prevent DPSO from converging to local optimum. Simulation test shows that compared with other algorithms, the proposed algorithm is more effective in acquiring global optimal solution to optimal test selection.

In view of the problems brought from current traditional maintenance method for complex system, such as the lack of maintenance, the excessive maintenance, the unreasonable allocation of spare parts, the belated deploying of support resources, and the low efficiency of maintenance. This paper studies the condition-based maintenance of complex system based on performance degradation, a decision process of CBM combing the preventive maintenance and corrective maintenance are combed. Based on the performance degradation date of complex system, the early fault warning model of trend filtering and the fault diagnosis model of data-driven are proposed. The aspects of early fault warning, fault diagnosis and maintenance decision are considered, a framework of the condition-based maintenance of complex system based on performance degradation is constructed. This has great significance for promoting the ability of maintenance support and combat to complex system.

In order to solve the problem of insufficient research on the failure mechanism of key components in the prediction and health management system (PHM), this paper proposes a simulation analysis method of rotating machinery failure mechanism based on HHT. First, the working conditions and failure mechanism of the turret bearing are analyzed and calculated. Then vibration data are denoised and simulated. Finally, the theoretical calculation results and simulation results are compared. The validity of the fault mechanism analysis method is verified by fault simulation analysis, which can lay the foundation for accurate diagnosis of weapon faults, and improve the accuracy and stability of fault diagnosis prediction.

In the process of predicting the remaining life of key system components with a large amount of operational observation performance data in the product, it is difficult to establish the life distribution model due to the scarcity of life data, and traditional degradation process analysis models have poor adaptability of product performance observation data, which leads to low accuracy and weak validity of product life prediction. Fully excavating component degradation data information, based on relevant degradation analysis techniques and the filtering prediction method in the statistical model and the regression convolutional neural network prediction method in the machine learning technology, a fusion model of product remaining life prediction is established. The fusion model combines the filtering forecasting model’s ability to mine product degradation status, the ability to express uncertainty, and the data adaptability and forecasting accuracy of the regression convolutional neural network model, which improves the accuracy and effectiveness of product degradation data analysis, and can effectively predict the life of key product components, and provides auxiliary reference for health management of key system components with large amount of operational observation data in the product.

In view of the question of whether the performance of the high-sensitivity receiver put into mass production is guaranteed, according to the telemetry parameter characteristics of the high-sensitivity receiver, the channel AGC and SNR parameters sensitive to the environmental characteristics are analyzed, and using the test data under different conditions obtained by the test specified in the environment test specification, the performance consistency of the receiving channel is analyzed. The analysis of on-orbit data and ground environment test data shows that the channel performance of receiver in the two different environments is consistent, and the channel performance of the high-sensitivity receiver is stable, which ensures the broad adaptability of the de-mission product.

This paper evaluates the security and validity of triaxial vibration test for the connecting pieces in the typical section, for example, standard parts and connectors, etc. Through the impacting analysis and simulation analysis about the failure mode, the failure mechanism of all kinds of connecting pieces in the section during triaxial vibration, and the stress distribution of loose failure, strength failure and fatigue failure are obtained, which provide a foundation for the analysis of triaxial vibration and defects implant. Then, the composite clipping spectrum is obtained by analyzing the section’s triaxial vibration from three aspects of stress, acceleration and fatigue failure. The paper analyzes the security of the composite clipping spectrum and the validity of triaxial vibration. Finally, the composite clipping spectrum is validated effectively and could motivate the potential faults of the connecting pieces, at the same time, the triaxial vibration could increase the test efficiency and reduce the test period, comparing to the uni-axial vibration.

In order to improve the detection accuracy and efficiency of flexible solar array hinge defect detection, a two-stage algorithm combining target detection and classification network is proposed by carrying out the research on the surface defect detection method of flexible solar array hinge based on deep learning. The method adopts an automatic camera device to take photos of the hinges and record the surface state of hinges. At the same time, transfer learning and data enhancement algorithms are introduced to solve the problem of lack of defect samples. At the same time, the requirements for computing resources are low and the computational performance is high, so as to achieve the effect of high accuracy and efficiency of surface defect detection.

This paper investigates the thermal environment about Lunar surface. Based on the requirements of landing mission near the equator of the moon, the heat rejection scheme of the Manned Lunar Lander Module was designed under the high temperature environment during the moon day. The Lunar Lander thermal control system(TCS) accomplished the heat acquisition function using various coldplates and heat exchangers, and the heat transport function was accomplished using a double-loop system architecture. For the majority of the mission, heat rejection was accomplished using deployable radiator. A sublimator was chosen as the supplemental heat-rejection device. The instantaneous radiator heat rejection capability and heat rejection requirement by the sublimator during moon day was analyzed. The simulation program of TCS was established based on Python language, and the performance of the TCS during lunar day was simulated. The results show that when the radiator area is 29m2, the thermal control system can meet the temperature control requirements of the equipment inside and outside the cabin. Sublimator as a heat-rejection device can reduce the radiator area.

For a new generation of launch vehicle using four-parallel liquid oxygen kerosene engine, FLUENT software is applied to study the influence of the engine jet on the thermal environment of rocket bottom. The study shows that: at low flight height, the jet flow interaction is weak; the backflow impacts the bottom center of the rocket at subsonic speed; and the air mole fraction in the backflow accounts for more than 90%. At high flight height, the backflow flows to the bottom plate at supersonic speed; The high-temperature jet gas in the backflow accounts for more than 50%. With the increase of flight height, the backflow point is closer to the vehicle afterbody, and the thermal environment of the afterbody is more severe. The maximum heat flux point at the bottom occurs at the height of about 21 km.