With the growing complexity of modern micro-architectures, processor design must accommodate a wide array of parameters, resulting in vast design spaces. Identifying the optimal trade-offs among various design metrics poses a significant challenge. Performance is inherently difficult to model analytically, while area can be represented by analytical models. Performance evaluation relies on expensive and time-consuming cycle-accurate simulators (CAS), which puts forward strict requirements on the convergence speed and data dependence of the optimization methods. In this paper, based on the characteristics of the design metrics, the processor design problem is modeled as a hybrid black-box and white-box multi-objective discrete optimization problem (BWMO-DOP). In engineering applications, parallel simulation is a common acceleration technology. Therefore, white-box objective, area, is formulated as parallel constraints, while black-box objective, performance, is approximated by order-preserving surrogate objective. And then, BWMO-DOP is simplified into parallel single-objective expensive black-box optimization problems, which are solved by an efficient SOP-MOOA. SOP-MOOA iteratively explores more design points, enhancing the accuracy of the surrogate model while simultaneously updating the Pareto set. Experimental results demonstrate that the proposed algorithm outperforms baseline algorithm in an engineering case and three general numerical cases. In the engineering experiment for performance-area optimization, the proposed algorithm outperforms the baseline algorithm by a factor of more than two when considering the combined effect of performance improvement percentage and area reduction percentage. In numerical tests conducted on three 40-dimensional black-box functions under the same evaluation overhead, the proposed algorithm consistently identified Pareto set of superior quality.

The virtual machine (VM) placement considering the topology constraints is difficult because the unpredictable topological VMs raise additional structural requirements (including the affinity, anti-affinity and fault-domain) on the resource pool. Thus, the service level agreement (SLA) can be violated even when the occupancy of the resource pool is quite modest. In order to solve this problem, we propose an efficient buffer-migration-based heuristic online algorithm. First, we build an integer programming model for the topology-aware VM placement problem. Second, we propose a hierarchical resource-preserving online approach, where the Rack and physical machine (PM) nodes are selected in the upper and lower layers respectively. Finally, we utilize the buffer to place and migrate the unfitted VMs to enhance the capacity of the resource pool. The proposed approach is tested with high proportional topological VM requests (nearly 60%) in the resource pool with the scale of 500, 1000 and 1500 PMs. The results show that our online approach (with unknown upcoming VM information) can achieve more than 85% of the performance for the offline approach (with complete upcoming VM information). The latency is lower than 5ms per VM.

Existing target segmentation methods are typically considered easy to implement and yield satisfactory results. However, they generally cannot adapt to a new environment without tiring parameter tuning, which leads to poor performance, including issues such as over-segmentation, under-segmentation, missing segmentation, and false positives. To avoid the wrong segmentation in a parameter-tuning-free and user-friendly fashion, we propose a generalized target segmentation (GTS) method based on the image-view representation of point clouds. Specifically, the method avoids devising a Euclidean space-based algorithm that is sensitive to surrounding objects and to the varied point cloud density and intensity in a new environment. The target segment produced by GTS can be used with any target-based extrinsic calibration architecture, based on which this paper further proposes a generalized target-based (in this case, chessboard) extrinsic calibration framework called GTSCalib for a non-repetitive scanning LiDAR and a camera. GTSCalib additionally introduces a novel intensity threshold method based on kernel density estimation (KDE) for 3D corner detection and the SQPnP solver for optimization to achieve more generalized and robust performance. Extensive simulations and experiments demonstrate that GTSCalib has high generalization ability, robustness, and accuracy. The code is released at https://github.com/Natsu-Akatsuki/GTSCalib.Note to Practitioners—Calibration is necessary for many non-repetitive scanning LiDAR-camera systems to enable sensor fusion in the fields of mapping, localization, and perception. Unfortunately, existing target-based (in this case, chessboard) calibration methods are weakly adaptable to the surrounding environment with variable density or intensity of point clouds, resulting in unstable performance, particularly for the target segmentation submodule. To solve this problem, we introduce a new target segmentation approach, GTS, and a more generalized and robust extrinsic calibration framework, GTSCalib. The proposed GTSCalib is very suitable for practitioners looking for a robust and accurate target-based calibration without limits on the target’s pose or its surrounding environment and without the need for time-consuming parameter tuning.

Design Space Exploration (DSE) in processors is an expensive discrete simulation optimization problem. The data requirements of the regular data-driven methods are so large that it is challenging to converge to a satisfactory solution within a limited simulation budget. Based on binary integer linear programming (BILP), an iteratively piecewise linear approximate method (IPLAM) is proposed for this kind of problems to reduce the dependence of simulation data. IPLAM starts from an initial reference point. Each iteration generates a set of trial points based on the most promising reference point by piecewise shift method. After evaluating the trial points, a local surrogate model is constructed for the unit neighborhood of the reference point. The surrogate model is then used to guide the exploration of the next reference point. In theory, IPLAM can converge to the global optimal point under mild assumptions, which is further verified by the numerical experiments. Meanwhile, the numerical experiments demonstrate that IPLAM outperforms the advanced Bayesian optimization and differential evolution methods on high-dimensional discrete black-box test functions. Besides, the practical effectiveness of IPLAM is validated by an industrial case for processor DSE.

The efficiency and profitability of cloud data centers are significantly influenced by virtual machine (VM) placement. However, the Non-Uniform Memory Access (NUMA), which has been practically applied to reduce the memory bandwidth competition, is often neglected in the existing research. Actually, the incorporation of NUMA may change the traditional resource allocation mechanism, and demands for a new VM placement model. Hence, considering the multi-NUMA architecture, this paper studies the NUMA-aware VM placement (NAVMP) problem in a cloud computing system, where the resource pool is composed of enormous number of heterogeneous servers with diverse multi-resource remains. The NAVMP problem is analytically formulated as an integer program (IP). Also, for the first time, the incarnations of VM types are introduced to simplify the VM deployment rules originated from complex NUMA architecture. We aim to maximize the VM provision ability (VPA) of the resource pool, and thus propose a novel Value Function to describe servers’ VPA. The resulting formulation, which is a new variant of the multiple-choice multiple multi-dimensional knapsack (MMMK) problem, is of significant computational challenges. So we customize a decomposition approach based on Column Generation (CG) to support the offline optimization. Numerical experiments on a practical dataset demonstrate the validity and scalability of the customized CG-based approach. Our approach outperforms a professional IP solver, i.e., Cbc, and a popular meta-heuristic algorithm, i.e., genetic algorithm (GA), and can efficiently address large-scale NAVMP instances with ten thousands of VM demands and servers.Note to Practitioners—This paper proposes a novel IP model for NAVMP. To cope with the complicated deployment logic associated with the complex multi-NUMA architecture of modern multi-core systems, we present an NAVMP formulation from the perspective of incarnations of VM types. Different from the traditional VM placement problem that aims to minimize the number of activated servers, i.e., the vector bin packing (VBP)-based model, we adopt the objective that maximizes the VPA of a resource pool for further improving the resource utilization. The resulting formulation is an MMMK problem, which is computational very challenging for a practical scale resource pool. Hence, to mitigate the computation burden, we design and implement a CG-based decomposition approach to support the offline optimization for NAVMP. Parallelization scheme and nontrivial heuristic strategies are applied to promote the computation efficiency. According to our numerical experiments, the proposed decomposition approach demonstrates a much superior solution capacity to the Cbc solver and GA. In particular, to achieve a comparable solution precision with Cbc, the computing time can be reduced by orders of magnitude. Also the CG-based approach outperforms GA in both the solution quality and computation time for large-scale instances. Besides, compared to the VBP model, our MMMK-based NAVMP model has improved the VPA up to 44.39%. Practically, the proposed offline approach can be leveraged to guide online VM allocation decisions, and perform efficient results evaluation.

Battery swapping station (BSS) can provide fast battery swapping and flexible battery charging in off-peak hours, it is thus beneficial for electric vehicles (EVs) and power grid in terms of battery life extension and power grid regulation. However, this increases the charging scheduling complexity in BSS since the batteries are not necessarily required to be charged immediately as they arrived. This problem becomes challenging in the presence of nonlinear battery charging characteristics and demand/supply uncertainties. Since it is difficult for the traditional learning-based methods to deal with the constraints caused by nonlinear charging characteristics, low sampling efficiency and unstable training issues can occur. In order to solve these issues, we present a novel deep reinforcement learning (DRL) approach. In contrast to the traditional approaches where the battery charging characteristic is simplified to a constant-current or constant-power process, we propose an equivalent circuit model (ECM) to capture the nonlinear charging characteristics. In ECM, the battery’s open-circuit voltage (OCV) is a function of its state of charge (SoC), as a result, the upper bound of the charging/discharging power of battery is influenced by its SoC. Then we construct a constrained Markov decision process (CMDP) model and propose a Beta distribution-based DRL approach with a continuous action mask (AM) to improve the sampling efficiency and consistency of the training process. Numerical experiments show that our new approach can provide better results in terms of operation cost and quality of service (QoS) in comparison with other state-of-the-art DRL methods.

The initial alignment is a critical stage for a strapdown inertial navigation system (SINS) and global positioning system (GPS) integrated navigation system. Currently, two major factors degrade the performance of SINS/GPS in-motion initial alignment, i.e., outliers in GPS measurements and cumulative low-accuracy inertial measurement unit (IMU) bias errors. This article considers both factors and proposes GISEIA-EMM: a high-accuracy GPS-inertial state estimator for in-motion alignment based on extended magnitude matching (EMM) method. First, we use the full integral method and non-interpolation procedure to construct the vector observation, which reduces the number of outliers and improves the accuracy of outlier detection. Second, we use an error-state extended Kalman filter (ESEKF), based on an augmented state-space model where the reference vector is regarded as a state, to suppress cumulative IMU bias errors, which improves the alignment accuracy. Third, we propose an EMM method, with the non-drifted expected normalized magnitude error, to detect and eliminate outliers in GPS measurements, which makes the alignment process stable. Simulation and field test results demonstrate that GISEIA-EMM can effectively address the negative impact of the two factors.

One of the major sources of conservativeness in min-max model predictive control (MPC) is the handling of constraints, where the ellipsoidal robust invariant set is utilized for the proposal of sufficient and conservative conditions for the satisfaction of the constraints. In this article, in order to reduce the conservativeness due to the constraint handling, we add additional relaxation variables to the physical constraints and propose a two-step approach through relaxing the constraints by the amount which is determined by the calculating of the maximal admissible set (MAS). The constraints are relaxed in an iterative manner to avoid the constraints violation, and the constraint relaxation variables are degrees of freedom for relaxing the constraints and improving the control performance. Moreover, we show that under certain circumstance, the physical constraints can be removed without the constraint violation. The proposed approach is shown to be recursively feasibility and its effectiveness is verified through an air conditioning control in a building energy system. Note to Practitioners—Buildings are account for large percentage of worldwide energy consumptions. One of the most applicable method for optimization of building energy system subject to multiple constraints is the model predictive control (MPC). However, the industrial MPC is usually not recursively feasible, which implies that the optimization problem can become infeasible and the software will be terminated at some time. In order to apply the MPC synthesis approach (MPC with recursive feasibility guarantee), we have to overcome the conservativeness problem due to the handling of constraints. We propose a useful approach in this work by introducing relaxation variables which act as degrees of freedom for improving the control performance, while the physical constraints are still satisfied. The proposed approach is verified through an example of a 24m2 office room located in Cyber-Physical Energy System (CPES) lab in Western China Science and Technology Innovation Harbour in Xianyang, China. The numerical results show the performance improve of the proposed approach.

In real-world industrial control systems, the extended usage of diverse equipment and instruments over extended periods increases the likelihood of malfunctions in any process unit. Such malfunctions can significantly impact the entire system, leading to substantial economic losses. To address these challenges, this paper proposes an actuator-sensor fault estimation and self-healing control scheme aimed at ensuring the stable and efficient operation of a discrete T-S fuzzy system. Firstly, a dual observer fault estimation method is introduced to overcome the limitations of highly conservative stability conditions and limited applicability encountered when estimating actuator and sensor faults with a single observer. Secondly, selfhealing controllers based on integral sliding mode and state feedback are individually designed. Lastly, leveraging the TS fuzzy model of the wastewater treatment plant, simulation experiments are conducted to validate the efficacy of the proposed methods. Comparative analysis of simulation results reveals that the dual observer fault estimation methods exhibit faster response speed for both faults estimation. Furthermore, in comparison to other self-healing controllers, the fuzzy-weighted self-healing controller exhibits superior overall performance

This article considers the dynamic output feedback model predictive control (DOFMPC) for the constrained Takagi-Sugeno (T-S) model with bounded disturbance. Unlike the existing approach where the robust positively invariant set is characterized by a single ellipsoid, we characterize it by the intersection of multiple ellipsoids, each corresponds to a vertex sub-model of the T-S model realization. The previous single ellipsoid is then an inner approximation of the intersection of multiple ellipsoids in this article. Therefore, the performance can be improved. We also generalize the multi-ellipsoid approach to the previous multi-step approach and formulate the so-called multi-step multi-ellipsoid approach in this article, which can further enlarge the feasibility region and enhance the performance. The recursive feasibility and the convergence of the approach are guaranteed. The proposed approaches are compared through a numerical problem to show their effectiveness.

Reconstructing visual stimulus representation is a significant task in neural decoding. Until now, most studies have considered functional magnetic resonance imaging (fMRI) as the signal source. However, fMRI-based image reconstruction methods are challenging to apply widely due to the complexity and high cost of acquisition equipment. Taking into account the advantages of the low cost and easy portability of electroencephalogram (EEG) acquisition equipment, we propose a novel image reconstruction method based on EEG signals in this article. First, to meet the high recognizability of visual stimulus images in a fast-switching manner, we construct a visual stimuli image dataset and obtain the corresponding EEG dataset through EEG signals collection experiment. Second, we introduce the deep visual representation model (DVRM), comprising a primary encoder and a subordinate decoder, to reconstruct visual stimuli representation. The encoder is designed based on residual-in-residual dense blocks to learn the distribution characteristics between EEG signals and visual stimulus images. Meanwhile, the decoder is designed using a deep neural network to reconstruct the visual stimulus representation from the learned deep visual representation. The DVRM can accommodate the deep and multiview visual features of the human natural state, resulting in more precise reconstructed images. Finally, we evaluate the DVRM based on the quality of the generated images using our EEG dataset. The results demonstrate that the DVRM exhibits an excellent performance in learning deep visual representation from EEG signals, generating reconstructed representation of images that are realistic and highly resemble the original images.

In this article, a novel approach to hierarchical implementation of output feedback robust model predictive control is proposed for the linear polytopic uncertain model. One optimization problem for minimizing the performance index is followed with the other assessing estimation error set (EES). The two problems are posed in a lexicographic order. Since in the latter problem, the controller parametric matrices are retaken as the degrees of freedom for the optimization, a much less conservative EES is calculated. Therefore, by applying the new approach, the control performance can be greatly improved as compared with the earlier schemes without lexicographic optimization. The proposed approach is proven to be recursively feasible, and the closed-loop stability is specified by the notion of quadratic boundedness. The result is verified through two numerical examples.

In this article, we present a codesign of measurement quantized dynamic output feedback model predictive control (DOFMPC) for the Takagi–Sugeno model with bounded disturbance. The system output is quantized by a dynamic quantizer before it is transmitted to the DOFMPC controller. Hence, we utilize the dynamic output feedback control law with a quantized output signal and consider the mixed input and quantized output constraint for the controller design. By optimizing the quantizer and controller parameters online, the control performance is enhanced. Moreover, we formulate a two-leveled optimizations, with the upper level optimizing the performance index and the lower level optimizing the soft constraint in a lexicographic order, for the codesign of the DOFMPC controller and dynamic quantizer. Thus, there are more degrees of freedom for tightening the soft constraints. The recursive feasibility and stability of the proposed approaches are guaranteed. The applicability of the proposed approach is illustrated by a simulation example.

In order to ease the online computational burden of fuzzy model-predictive control for the Takagi–Sugeno model with bounded disturbance, a lookup table containing the possible mappings from the state to the input is usually constructed offline so that the online computational burden is reduced to searching in this lookup table. However, with the increase in the problem size, the computational burden of the online search in the lookup table can be large enough to influence the real-time implementation. In this article, we propose a novel offline approach to solve this problem, where the control law is online searched in a receding horizon cache, which is only a small portion of the lookup table. The cache is refreshed in a one-step-ahead fashion to guarantee that the proper one-step-ahead state-to-input mapping can be found in the cache. The recursive feasibility and stability hold. The effectiveness and the efficiency of the proposed approach are verified through two examples.

In this article, an output feedback model predictive control for fuzzy systems to track a time-varying steady-state target is presented. The steady-state target is calculated through a steady-state target calculation (SSTC) module, which is combined with the controller in the lower layer designing based on the linear time invariant model. The presence of SSTC makes the recursive feasibility as well as the stability difficult to guarantee. This article presents a promising approach by specifying the solution of SSTC as a function of estimated artificial disturbance, designing the robust invariant set offline and including the invariance condition in SSTC. Hence, the proposed approach in this article guarantees that if SSTC optimization problem is feasible, then the dynamic controller steers the system to the time-varying steady-state target. Furthermore, in order to reduce the conservativeness of the proposed approach, a heuristic online approach with N free control moves is proposed where the invariant set specified offline is acting as a terminal region for the dynamic controller. A numerical example is given to show the effectiveness of the proposed approach.

In this paper, the fuzzy model-predictive control for a nonlinear system represented by a discrete-time Takagi-Sugeno model with norm-bounded disturbance is studied. An output feedback algorithm is proposed by parameterizing the infinite horizon control moves and estimated states into one free control move and one free estimated state followed by a dynamic output feedback law. Since the introduced free control move and free estimated state are decision variables that bring more degrees of freedom for the optimization, a larger feasibility region and better control performance can be achieved. By properly designing the constraints in the optimization problem, the recursive feasibility and the convergence of the closed-loop system to the neighborhood of the equilibrium are guaranteed. A numerical example is given to illustrate the effectiveness of the proposed algorithm.