Heng Yang

The mast is a critical component of rotary drilling rigs, which has a cross-section consisting of a rectangular shape formed by two web plates and two flange plates. Structural optimization of the mast is necessary to address the issue of excessive weight. The shortcomings of the traditional structural optimization algorithms are summarized as follows: the optimized steel plate thickness is a non-integer, where rounding upwards may increase the cost to a certain extent, but it can ensure the safety of the structure; rounding downwards its load carrying capacity may not satisfy the requirements, and thus a novel Salp Swarm Algorithm is proposed to solve the optimization problem. First, this study improves the initialization and update strategy in the traditional Salp Swarm Algorithm. In order to obtain a solution for engineering, an innovative multi-dimensional running comparison is carried out. Secondly, the optimization model of rotary drilling rigs is established based on the division of the working conditions. The objective function of the optimization model is to minimize the weight of the mast while considering the constraints of strength, stiffness, stability, and welding process. Finally, the proposed optimization algorithm and the established optimization model are applied to optimize the design of the mast for a rotary drilling rig. The empirical results demonstrate that the weight of the mast has been reduced by 20%. In addition, the Improved Salp Swarm Algorithm exhibits higher solution quality, faster iteration capability, and extreme stability in optimizing welded box sections compared to the conventional algorithm. The example shows that the Improved Salp Swarm Algorithm is applicable to the optimization problem of box sections.

Deep Reactive Ion Etching (DRIE) is a critical process in the fabrication of microelectromechanical systems (MEMS) and other advanced semiconductor devices. Achieving the desired trench profile with minimal defects remains a significant challenge due to the complex interplay of various process parameters. This paper focuses on optimizing parameters such as etching cycle time and passivation cycle time in the DRIE process. An improved central composite design method was employed to design 18 sets of etching recipe experiments. Actual etching profile data collection was conducted, resulting in 1800 sets of etching profile data, which serve as the foundation for subsequent AI algorithm optimization. Through real data experiments, effective etching recipes for high-precision smooth vertical sidewalls were identified. It was found that an etching/passivation ratio in the range of 1.09 to 1.4 could achieve 90-degree sidewalls, while a ratio between 1.09 and 1.5 could achieve scallop heights below 90 nm. Furthermore, based on convolutional neural networks and data-driven models, the predictive models are proposed for etching profile angles and scallop smoothness. The results demonstrated a high prediction accuracy, with only a 5.5% error in the etching profile angle prediction. The optimized etching results and recipes identified in this study achieve highly smooth sidewalls. This work not only enhances the quality of etching processes but also contributes significantly to the advancement of AI applications in semiconductor manufacturing.

This paper presents a novel adaptive visual servo controller for controlling a manipulator with unknown time-varying actuator failure and uncalibrated camera, which allows the output capacity of actuators that drive the same joint different. Since the actuators that drives the same joint in parallel are not identical, an unequal-actuation scheme is given to guarantee that every actuator’s output dose not exceed its maximum output capacity. Additionally, to identify the unknown time-varying failure, this paper adopts a widely used actuator failure model to describe the potential failure and proposes a new adaptive law based on Lyapunov’s method to compensate the unknown failure. Furthermore, the convergence of the image error and the whole signals in system are proved to be stable. Finally, The simulation results are provided to evaluate the good performance of the proposed controller.

In this paper, we reimagine volumetric representations through the lens of quadrics. We posit that rigid scene components can be effectively decomposed into quadric surfaces. Leveraging this assumption, we reshape the volumetric representations with million of cubes by several quadric planes, which results in more accurate and efficient modeling of 3D scenes in SLAM contexts. First, we use the quadric assumption to rectify noisy depth estimations from RGB inputs. This step significantly improves depth estimation accuracy, and allows us to efficiently sample ray points around quadric planes instead of the entire volume space in previous NeRF-SLAM systems. Second, we introduce a novel quadric-decomposed transformer to aggregate information across quadrics. The quadric semantics are not only explicitly used for depth correction and scene decomposition, but also serve as an implicit supervision signal for the mapping network. Through rigorous experimental evaluation, our method exhibits superior performance over other approaches relying on estimated depth, and achieves comparable accuracy to methods utilizing ground truth depth on both synthetic and real-world datasets.

We study the problem of verification and synthesis of robust control barrier functions (CBF) for control-affine polynomial systems with bounded additive uncertainty and convex polynomial constraints on the control. We first formulate robust CBF verification and synthesis as multilevel polynomial optimization problems (POP), where verification optimizes-in three levels-the uncertainty, control, and state, while synthesis additionally optimizes the parameter of a chosen parametric CBF candidate. We then show, by invoking the KKT conditions of the inner optimizations over uncertainty and control, the verification problem can be simplified as a single-level POP and the synthesis problem reduces to a min-max POP. This reduction leads to multilevel semidefinite relaxations. For the verification problem, we apply Lasserre's hierarchy of moment relaxations. For the synthesis problem, we draw connections to existing relaxation techniques for robust min-max POP, which first uses sum-of-squares programming to find increasingly tight polynomial lower bounds to the unknown value function of the verification POP, and then call Lasserre's hierarchy again to maximize the lower bounds. Both semidefinite relaxations guarantee asymptotic global convergence to optimality. We provide an in-depth study of our framework on the controlled Van der Pol Oscillator, both with and without additive uncertainty.

Inductive Conformal Prediction (ICP) provides a practical and effective approach for equipping deep learning models with uncertainty estimates in the form of set-valued predictions which are guaranteed to contain the ground truth with high probability. Despite the appeal of this coverage guarantee, these sets may not be efficient: the size and contents of the prediction sets are not directly controlled, and instead depend on the underlying model and choice of score function. To remedy this, recent work has proposed learning model and score function parameters using data to directly optimize the efficiency of the ICP prediction sets. While appealing, the generalization theory for such an approach is lacking: direct optimization of empirical efficiency may yield prediction sets that are either no longer efficient on test data, or no longer obtain the required coverage on test data. In this work, we use PAC-Bayes theory to obtain generalization bounds on both the coverage and the efficiency of set-valued predictors which can be directly optimized to maximize efficiency while satisfying a desired test coverage. In contrast to prior work, our framework allows us to utilize the entire calibration dataset to learn the parameters of the model and score function, instead of requiring a separate hold-out set for obtaining test-time coverage guarantees. We leverage these theoretical results to provide a practical algorithm for using calibration data to simultaneously fine-tune the parameters of a model and score function while guaranteeing test-time coverage and efficiency of the resulting prediction sets. We evaluate the approach on regression and classification tasks, and outperform baselines calibrated using a Hoeffding bound-based PAC guarantee on ICP, especially in the low-data regime.

Set-membership estimation (SME) outputs a set estimator that guarantees to cover the groundtruth. Such sets are, however, defined by (many) abstract (and potentially nonconvex) constraints and therefore difficult to manipulate. We present tractable algorithms to compute simple and tight overapproximations of SME in the form of minimum enclosing ellipsoids (MEE). We first introduce the hierarchy of enclosing ellipsoids proposed by Nie and Demmel (2005), based on sums-of-squares relaxations, that asymptotically converge to the MEE of a basic semialgebraic set. This framework, however, struggles in modern control and perception problems due to computational challenges. We contribute three computational enhancements to make this framework practical, namely constraints pruning, generalized relaxed Chebyshev center, and handling non-Euclidean geometry. We showcase numerical examples on system identification and object pose estimation.

The two-stage object pose estimation paradigm first detects semantic keypoints on the image and then estimates the 6D pose by minimizing reprojection errors. Despite performing well on standard benchmarks, existing techniques offer no provable guarantees on the quality and uncertainty of the estimation. In this paper, we inject two fundamental changes, namely conformal keypoint detection and geometric uncertainty propagation, into the two-stage paradigm and propose the first pose estimator that endows an estimation with provable and computable worst-case error bounds. On one hand, conformal keypoint detection applies the statistical machinery of inductive conformal prediction to convert heuristic keypoint detections into circular or elliptical prediction sets that cover the groundtruth keypoints with a user-specified marginal probability (e.g., 90%). Geometric uncertainty propagation, on the other, propagates the geometric constraints on the keypoints to the 6D object pose, leading to a Pose UnceRtainty SEt (PURSE) that guarantees coverage of the groundtruth pose with the same probability. The PURSE, however, is a nonconvex set that does not directly lead to estimated poses and uncertainties. Therefore, we develop RANdom SAmple averaGing (RANSAG) to compute an average pose and apply semidefinite relaxation to upper bound the worst-case errors between the average pose and the groundtruth. On the LineMOD Occlusion dataset we demonstrate: (i) the PURSE covers the groundtruth with valid probabilities; (ii) the worst-case error bounds provide correct uncertainty quantification; and (iii) the average pose achieves better or similar accuracy as representative methods based on sparse keypoints.

The two-stage object pose estimation paradigm first detects semantic keypoints on the image and then estimates the 6D pose by minimizing reprojection errors. Despite performing well on standard benchmarks, existing techniques offer no provable guarantees on the quality and uncertainty of the estimation. In this paper, we inject two fundamental changes, namely conformal keypoint detection and geometric uncertainty propagation, into the two-stage paradigm and propose the first pose estimator that endows an estimation with provable and computable worst-case error bounds. On one hand, conformal keypoint detection applies the statistical machinery of inductive conformal prediction to convert heuristic keypoint detections into circular or elliptical prediction sets that cover the groundtruth keypoints with a user-specified marginal probability (e.g., 90%). Geometric uncertainty propagation, on the other, propagates the geometric constraints on the keypoints to the 6D object pose, leading to a Pose UnceRtainty SEt (PURSE) that guarantees coverage of the groundtruth pose with the same probability. The PURSE, however, is a nonconvex set that does not directly lead to estimated poses and uncertainties. Therefore, we develop RANdom SAmple averaGing (RANSAG) to compute an average pose and apply semidefinite relaxation to upper bound the worst-case errors between the average pose and the groundtruth. On the LineMOD Occlusion dataset we demonstrate: (i) the PURSE covers the groundtruth with valid probabilities; (ii) the worst-case error bounds provide correct uncertainty quantification; and (iii) the average pose achieves better or similar accuracy as representative methods based on sparse keypoints.

We study the problem of verification and synthesis of robust control barrier functions (CBF) for control-affine polynomial systems with bounded additive uncertainty and convex polynomial constraints on the control. We first formulate robust CBF verification and synthesis as multilevel polynomial optimization problems (POP), where verification optimizes -- in three levels -- the uncertainty, control, and state, while synthesis additionally optimizes the parameter of a chosen parametric CBF candidate. We then show that, by invoking the KKT conditions of the inner optimizations over uncertainty and control, the verification problem can be simplified as a single-level POP and the synthesis problem reduces to a min-max POP. This reduction leads to multilevel semidefinite relaxations. For the verification problem, we apply Lasserre's hierarchy of moment relaxations. For the synthesis problem, we draw connections to existing relaxation techniques for robust min-max POP, which first use sum-of-squares programming to find increasingly tight polynomial lower bounds to the unknown value function of the verification POP, and then call Lasserre's hierarchy again to maximize the lower bounds. Both semidefinite relaxations guarantee asymptotic global convergence to optimality. We provide an in-depth study of our framework on the controlled Van der Pol Oscillator, both with and without additive uncertainty.

With the widely application of electronic transformers in smart grids, transformer faults have become a pressing problem. However, reliable fault diagnosis of electronic current transformers (ECT) is still an open problem due to the complexity and diversity of fault types. In order to solve this problem, this paper proposes an ECT fault diagnosis model based on radial basis function neural network (RBFNN) and optimizes the model parameters and the network size of RBFNN simultaneously via an improved whale optimization algorithm (WOA) to improve the classification accuracy and robustness of RBFNN. Since the classical WOA is easy to fall into a locally optimal performance, a hybrid multi-strategies WOA algorithm (CASAWOA) is proposed for further improvement in optimization performance. Firstly, we introduced the tent chaotic map strategy to improve the population diversity of WOA. Secondly, we introduced nonlinear convergence factor and adaptive inertia weight to enhance the exploitation ability of the WOA. Finally, on the premise of ensuring the convergence speed of the algorithm, we modified the simulated annealing mechanism in order to prevent premature convergence. The benchmark function tests show that the CASAWOA outperforms other state-of-the-art WOA algorithms in terms of convergence speed and exploration ability. Furthermore, to validate the performance of ECT fault diagnosis model based on CASAWOA-RBFNN, a comprehensive analysis of eight fault diagnosis methods is conducted based on the ECT fault samples collected from the detection circuit. The experimental results show that the CASAWOA-RBFNN achieves an accuracy of 97.77% in ECT fault diagnosis, which is 9.8% better than WOA-RBF and which shows promising engineering practicality.

The main host-guest conflict is universal and harmful in a tourist destination.Based on the supplement and improvement to the existing measuring scale of the residents' viewpoint.This paper constructs a structural equation model based on the survey data onto the residents of the ancient city of Langzhong, Sichuan Province, and reveals the relationship between the residents' conflict perception and the support of tourism development.The research shows that: 1) the conflict perception of the host-guest of tourist destination includes five dimensions of that perception of economic conflict, social conflict, resource environment conflict, emotional conflict, cultural conflict.Residents have the strongest perception of economic conflict and the weakest perception of social conflict.2) The five dimensions of host-guest conflict and tourism development support have a significant negative correlation, the stronger the residents' perception of the host-guest conflict, the lower their support for tourism development.