Intelligent fault diagnosis technology based on the deep neural network has shown significant advancements in recent years. However, it is difficult and expensive to deploy a fault diagnosis neural network with a huge number of parameters to an embedded computing platform with limited hardware resources. To address this issue, a practical bearing failure detection method using a new efficient deep network with the knowledge self-adaptive evolution, named autonomous compression method based on network pruning and knowledge distillation (AMC-NPKD), is proposed in this paper. In the proposed method, the reinforcement learning technique based on the deep deterministic policy gradient (DDPG) is employed to iteratively prune the network’s structure. The knowledge distillation (K-D) process is employed to fine-tune the pruned network after each pruning iteration. The results based on two datasets demonstrate that the proposed method effectively optimizes the structure of fault diagnosis networks. The proposed AMC-NPKD method is meaningful for promoting the engineering development of the intelligent fault diagnosis technology.

This study proposes an accurate dead zone compensation control method for electro-hydrostatic actuators (EHAs) under low-speed conditions. Specifically, the nonlinear dead zone characteristics under low-speed conditions are summarized based on numerous EHA experiments. An adaptive compensation function (ACF) is then constructed for the dead zone. Next, this study proposes an adaptive dead zone compensation control method for EHAs by integrating the ACF with a virtual decomposition controller (VDC) based on the established EHA model. The stability of the proposed control method is also proven. Finally, the proposed control method is verified using an EHA platform. The test results show that the dead zone trajectory tracking errors of EHAs are significantly reduced when combined with the ACF. Furthermore, since most EHAs are controlled by adjusting the motor speed, the method presented in this study is simpler and easier to use than methods that employ flow compensation.

Flow mismatch, strong nonlinearity, and complex disturbance are the main challenges that electro-hydrostatic actuator (EHA) control faces. This article proposes a three-port piston pump to replace the two-port piston pump in traditional EHA to solve the flow mismatch problem caused by asymmetric cylinders. Then, for the complex disturbance problem in EHA control, an extended state observer and a nonlinear disturbance observer are designed to estimate matched and mismatched disturbances, respectively. On this basis, the virtual decomposition control method is adopted. The EHA is virtually decomposed into three subsystems, and the controllers of each subsystem are designed separately to realize the high-precision nonlinear control of the EHA. Next, the stability of the proposed control method and observer is proved. Finally, the proposed three-port pump, observer, and controller are systematically verified based on the simulation model and the EHA experimental platform. Simulation and experimental results show that the proposed three-port pump solves the flow mismatch problem without increasing the weight and volume of the EHA. The designed two observers can accurately estimate the matching and mismatching disturbances during the EHA operation. By combining designed observers and the virtual decomposition control method, the accuracy and robustness of EHA control can be greatly improved.

Deep learning has become a popular approach for fault diagnosis due to its powerful feature extraction and adaptability. However, its reliance on extensive annotations poses challenges in real-world applications. To confront this issue, this article proposes the CLTrans, a contrastive learning-based knowledge transfer method for semi-supervised fault diagnosis. CLTrans utilizes large-scale unlabeled data to benefit downstream tasks by simply performing unsupervised similarity matching. A feature encoder pre-trained by CLTrans can extract discriminative representations of vibration signals and can efficiently adapt to various tasks, even with data under different distributions. Experimental results of inner-dataset and inter-dataset knowledge transfer demonstrate that CLTrans outperforms conventional deep learning and state-of-the-art semi-supervised fault diagnosis approaches in terms of accuracy and domain adaptability, especially under limited labels. The capability of unsupervised knowledge mining and transfer allows for reducing the burden of data collection and annotation.

The internal leakage fault-tolerant control problem of the electro-hydraulic servo actuator under the influence of multiple unmodeled dynamics is investigated in this paper, and an adaptive fault-tolerant control scheme based on unmodeled dynamics estimation and compensation is proposed. The model of the actuator is divided into two subsystems, which extended-state observers are respectively constructed to estimate the matched and mismatched unmodeled dynamics. Combined with the estimation results of the unmodeled dynamics, an adaptive fault-tolerant controller is designed by using the backstepping method. In which a controller reconfiguration mechanism based on internal leakage fault parameter online adaptation is used to accommodate the fault, and a feedforward compensation strategy is used to suppress the influence of unmodeled dynamics. Semi-physical simulation test of the proposed scheme is conducted under serious cylinder internal leakage. The test result shows that when the maximum internal leakage flow reaches 10.53 L/min, accounting for about 56.77% of the load flow, at the moment the opening of the servovalve is close to the maximum and the proposed scheme can still achieve high-precision position tracking control, where the maximum position tracking errors of fault transient and post-fault steady state are both limited within ±1.5% of the given position.

This paper proposes an accurate dead zone compensation control method for electro-hydrostatic actuators (EHAs) under low-speed conditions. Specifically, the nonlinear dead zone characteristics under low-speed conditions are summarized based on many EHA experiments. Then, an adaptive compensation function (ACF) for the dead zone is constructed. Next, based on the established EHA model, this paper proposes an adaptive dead zone compensation control method for EHAs by combining ACF with the virtual decomposition controller (VDC). The stability of the proposed control method is also proved. Finally, the proposed control method is verified by the EHA platform. Test results show that the dead zone trajectory tracking error of EHA is significantly reduced when combined with ACF. Furthermore, since most EHAs are controlled by adjusting the motor speed, the method presented in this paper is simpler and easier to use than methods that employ flow compensation.

During take-off and landing of airplanes, landing gear control system features have huge influence on airplane safety. In this paper, Labview software, a virtual instrument from National Instruments Corporation is used as a development platform to apply its powerful G language programming ablity and friendly program design interface to develop system application in order to enable control of landing gear extension and retraction and perform landing gear-related ground tests under laboratory conditions.

电动静液作动系统(Electro?Hydrostatic Actuator System,EHAS)是多电飞机关键子系统之一,系统结构紧凑,换热能力差,影响飞行安全,因此EHA热特性研究是一个具有现实意义的课题.通过对EHA整体结构进行简化处理,得到热特性分析模型,运用有限元仿真软件进行热特性仿真.对比实验结果,验证了该分析方法的准确性,并提出了结构优化方案.通过优化建模,运用仿真得到优化后模型的热特性,并与原模型特性对比得出,EHA结构整体稳态热分析最高点温度降低29.5％,瞬态分析最高点温度降低22.3％.该研究为求解EHA热特性提供了一种方法,对现有EHA的结构布局提出一种优化改进方案,提升了EHA的热特性,有助于电作动新产品的开发设计.

分别对无刷直流电动机在Ⅰ,Ⅲ象限以及Ⅱ,Ⅳ象限运行时三相绕组的导通及续流过程进行了分析,推导了驱动电路占空比与绕组电流的函数关系,提出了适用于电机四象限PWM驱动的变参数可重置PI控制方法.实验结果表明该控制方法可以使电机在四象限下良好地运行,且具有较好的电流和转速限踪精度.

with the increasing market demand and rapid development of aircraft technology, more electric concept has become an important trend for the future development of aircraft. Therefore, motors are widely applied in the aviation field increasingly. This paper mainly introduces the hardware environment and LabVIEW implementation process in the development of aviation high speed and high voltage BLDCM control system based on NI hardware and software platform. The test shows the control system is able to realize effective control over aviation BLDCM. This system is highly flexible and can lay a foundation for the algorithms used in subsequent related research efforts.

Trawl nets are mostly flexible structures working in the water. In order to investigate the effect of the fluid-structure interaction on the trawl net's numerical model, we modeled the trawl net and the flow field based on lumped mass method and finite volume method, separately; then we adopted a hybrid volume method (HVM) to model the fluid-structure interaction between the net and surrounding water. Since the gridding of trawl net is independent of its shape within the proposed HVM, large mesh can be used for the calculation of fluid-structure coupling model for higher efficiency. First, existing flume tank experimental data were used to verify the accuracy of the HVM. Then, trawl net states were analyzed based on the HVM by taking the fluid-structure coupling into consideration. The simulation reveals that difference between the trawl net model with and without fluid-structure interaction is about 6%, the main reason for that is the relative velocity of water flow around the codend is only 1/5 of the towing speed because of the flow blocking by the trawl mesh upstream. The above results indicate that fluid-structure interaction is very important for the analysis of trawl net which should not be ignored when numerical modelling.

In this paper, we proposed a mathematical model of a mid-water trawl system. The model consisted of towing warps, otter boards, a trawl net, and other rigging components. The warp and the net were modeled with the lumped mass method. The otter board was modeled in 6 degrees of freedom, and the variation in the otter board's hydrodynamic parameters with different yaws, pitches, and roll angles was also considered. We used a first-order approximation to simplify the relationship between the otter board's hydrodynamic parameters and its working angle, and the coefficients included in the otter board model were then estimated using the sea trail data. With this approach, the movement of the otter board was simulated, and its influence on the trawl gear performance was analyzed. The final results of the door depth and the door spread with warp lengths from 200 to 900 m showed good agreements with the sea trail data, and the mean normalized absolute error between the simulation and the measurement was approximately 5%.

The lumped-mass method is one of the most fundamental methods to simulate the dynamic behavior of submerged flexible nets. However, the low calculation efficiency limits its applications. In order to improve the calculation efficiency, the numerical stability of lumped-mass method based on explicated Euler integral algorithm was investigated. The simulation stability criterion was derived as a function of simulation step size and the physical parameters of netting materials. A physical parameter optimization (PPO) method was put forward to calculate the desired values of the lumped-mass model’s physical parameters based on stability criterion; length compensation was implemented to compensate the extra deformation of mesh bars caused by the changes of their stiffness. The PPO method can ensure the stability of the lumped-mass model with a desired simulation step size while minimizing the change to the physical parameters. A lumped-mass model of trawl gear was established with the PPO method, the simulation results were compared with those of the conventional lumped-mass method (without PPO) to validate the improvements. By using the PPO method, the calculation efficiency can be accelerated by 40 times while only inducing less than 2% error to the simulation results.

In this paper we propose and discuss a numerical method to model the current loads on a net cage. In our numerical model, the fluid–structure interaction is taken into consideration. The net cage is modeled on the mass-spring model; the flow field is modeled by the finite volume method (FVM). A novel hybrid volume approach is used to add the resistance force of the net cage into the flow field for coupling the fluid and net. The net resistance to the flow is calculated directly by the net’s current load using Newton’s Third Law. The resistance force is discretized in the hybrid volume and represented in the source term of the Navier–Stokes equation. By using the hybrid volume method, the mesh grid is separated from the net shape, and sparse grid (0.1m) can be used to calculate the flow field for computational efficiency. Based on the detailed flow field, we can predict the net’s current load more accurately. The final results are derived by the segregated iterative calculation of net shape and flow field. Current forces acting on both rigid and flexible net cages are simulated at water velocity from 0 to 1m/s; the simulation results of proposed numerical method are compared with the existing experiments, good agreements are shown in both flow field and current force, the mean normalized absolute error of the current force between simulations and measurements is about 5%.

Counterbalance valves are widely used in hydraulic deck machinery to balance the overrunning loads. However, as is well known, counterbalance circuit designed with poor choice of counterbalance valve tends to introduce instability to the system. This paper investigates the dynamic behavior of a pilot operated counterbalance valve which can operate at a flow rate about 2000L/min. A linearized stability analysis of such a hydraulic circuit which consists of a slip in cartridge, a pilot counterbalance valve and a hydraulic winch is presented. Pole-zero plots are employed to reveal the effect of the volume of control cavity, the hydraulic resistance on pilot line and counterbalance valve pilot area ratio on the stability of the system. The analysis results indicate that such a system will be unstable within the normal range of each parameter. An alternative approach that guarantees system stability by adding an accumulator on the pilot line is put forward. The approach stabilizes the pilot pressure by reducing the hydro-stiffness of pilot control cavity, thus the system can reach its stability condition. Finally, a numerical optimization method is putted forward, with the optimized parameters, the dynamic performance of considered system become better.