Optimal reliable constraints-based design space exploration in VLSI for power grid design

Optimal reliable constraints-based design space exploration in VLSI for power grid design

Reliable design space and complete single-loop reliability-based design optimization

This paper proposes a methodology to explore the optimal design space of sports equipment, where non-linear dynamics play a key role, with the goal of gaining creative insights. In the presented example, this space encompasses both the optimal launch conditions and the optimal sizes of a discus to maximize flight distance. The optimization procedure is divided into four steps. The first step calculates the aerodynamic forces acting on a discus using computational fluid dynamics. The second step optimizes the discus flight distance for launch conditions. This step provides the maximum attainable flight distance and the optimal launch conditions for each candidate discus design, that is, the optimal skill of the athlete. The third step is Bayesian optimization of the flight distance. Bayesian optimization identifies the promising design set of design variables, that is, equipment (discus). Finally, the fourth step explores the optimal space using self-organizing maps, providing creative insights for an optimal aerodynamic design space. It was found that discus thickness is the most crucial design variable, with the thinner discus achieving longer flight distances due to the greater maximum lift coefficient. Regarding launch conditions, the initial flight path angle is the most important parameter, ideally about 40 degrees with a narrow tolerance window.

Early design decisions have higher potential to influence building performance compared with the decisions made at later design stages. Performance simulation and optimization algorithms have been integrated to assist early design in reducing carbon emissions, improving indoor thermal comfort, etc. However, early decision making within a limited time frame is still challenging due to the large number of design options, the lack of decision-making guidance, and the trade-offs among various requirements. Selecting appropriate methods to explore design space is the key to find an ideal solution. This paper reviewed the challenges and identified the key questions to access the ability of existing decision-making methods to cope with different challenges. It is concluded that the interactive exploration of design space could be more effective and efficient by (1) combining the surrogate models and the automated optimization algorithms to improve the efficiency of the building performance calculation and the optimal design space position; and by (2) extending the optimal design space to increase the solution diversity, and (3) filtering the near optimal design space with consideration of the stakeholders’ preferences and values. Further integration of tools for building performance simulation, diversity description and decision-making guidance is needed to support the decision -making process.

We present a framework for area optimal system design space exploration for hypermedia applications. We focus on a category of processors that are programmable yet optimized to a hypermedia application. The key components of the framework presented in this paper are a retargetable instruction-level parallelism compiler, instruction level simulators, a set of complete media applications written in a high level language and a media processor synthesis algorithm. The framework addresses the need for area optimal system design by exploiting the instruction-level parallelism found in media applications by compilers that target multiple-instruction-issue processors. Using the framework we conduct an extensive exploration of area optimal system design space for a hypermedia application. We found that there is enough ILP in the typical media and communication applications to achieve highly concurrent execution when throughput requirements are high. On the other hand, when throughput requirements are low, there is no need to use multiple-instruction-issue processors.

In this paper, the connotation and an extension of the mechanical reliability design theory and method are presented via charts and texts. The abstruse mechanical reliability theories and methods are illustrated intuitively and scientifically via easy-to-understand language and simple and clear illustrations. Topics such as mechanical strength design, mechanical reliability design, dynamic and gradual reliability design, reliability optimization design, reliability sensitivity design and reliability robust design are covered to help the readers understand the wonder and practicality of mechanical reliability engineering. Diagrams of reliability design, dynamic and gradual reliability theory and technology, reliability optimization design, reliability sensitivity design and reliability robust design are illustrated for the first time. Additionally, the relationship between mechanical product reliability design and conventional safety factor design is elaborated in a table. This paper provides a solid mechanical reliability engineering foundation for design, manufacturing, usage and evaluation of mechanical products.

Iterative algorithms are widely applied in reliability analysis and design optimization. Nevertheless, phenomena of failed convergence, such as periodic oscillation, bifurcation, and chaos, are oftentimes observed in iterative procedures of solving some nonlinear problems. In the present paper, the essential causes of numerical instabilities including periodic oscillation and chaos of iterative solutions are revealed by the eigenvalue-based stability analysis of iterative schemes. To understand and control these instabilities, the stability transformation method (STM), which is capable of tackling numerical instabilities of iterative algorithms in reliability analysis and design optimization, is proposed. Finally, several benchmark examples of convergence control of PMA (performance measure approach) for probabilistic analysis and the SORA (sequential optimization and reliability assessment) for reliability-based design optimization (RBDO) are presented. The observations from the benchmark examples indicate that the STM is a promising approach to achieve convergence control for iterative algorithms in reliability analysis and design optimization.

Reliability analysis and design optimization of nonlinear structures

Design space optimization for topology based on fixed grid is proposed and numerical verification for high probability to a global optimum is shown. In conventional optimization, only a portion of structural parameters are designated as design variables while the remaining set of other parameters related to the design space are often taken for granted. It is desirable to make the design space evolve into a better one during optimization process by increasing or decreasing the number of design variables, which is called design space optimization. To have this capability, it is needed to obtain sensitivities when design space expands. A method has been developed recently with the necessary mathematical background, but due to coupling effect between neighboring elements during implementation sensitivity results have not been satisfactory. To solve this problem, a new method of decoupling nearby elements based on FE model is proposed. Also, to avoid the deletion of newly created elements during the optimization process, mixed artificial material is proposed. Seeking optimal design space for topology is very tedious and time-consuming. To accelerate this process, new method for design space expansion is proposed and superior performance over conventional method illustrated. Although there are many local optimums in the usual formulations, this is hardly mentioned or studied in the past. Two kinds of numerical examples are taken to demonstrate that the proposed design space optimization can give higher probability of getting the global optimum. Nomenclature * , u u→ + ψ = sensitivity of a functional using directional derivative at the pivot phase

The problem of large amounts of carbon emissions causes wide concern across the world, and it has become a serious threat to the sustainable development of the manufacturing industry. The intensive research into technologies and methodologies for green product design has significant theoretical meaning and practical value in reducing the emissions of the manufacturing industry. Therefore, a low carbon-oriented product reliability optimal design model is proposed in this paper: (1) The related expert evaluation information was prepared in interval numbers; (2) An improved product failure analysis considering the uncertain carbon emissions of the subsystem was performed to obtain the subsystem weight taking the carbon emissions into consideration. The interval grey correlation analysis was conducted to obtain the subsystem weight taking the uncertain correlations inside the product into consideration. Using the above two kinds of subsystem weights and different caution indicators of the decision maker, a series of product reliability design schemes is available; (3) The interval-valued intuitionistic fuzzy sets (IVIFSs) were employed to select the optimal reliability and optimal design scheme based on three attributes, namely, low carbon, correlation and functions, and economic cost. The case study of a vertical CNC lathe proves the superiority and rationality of the proposed method.

The driving system of cutter head is the key part of shield machine, it is also a technological bottleneck in the design and production of tunneling shield currently in China. This paper combines reliability design and optimization design to produce a reliability optimization scheme of main driving gear transmission of shield machine. It uses Kriging model for approximation and Bayesian analysis algorithm for global design optimization. A mathematical model is established for the reliability optimization design. Design variables include the modulus, the tooth number of small gears, the face width and the helix angle; objective function contains the total volume of gears and the constraint conditions of reliability. Based on that, a mathematical model of the reliability optimization design of main driving gear transmission is constructed, and design parameters that satisfy the constraint conditions and minimize the total volume of gears are derived. The paper contributes to driving China's self-production of main driving gear transmission of shield machine and the design enhancement.

For the problem of evidence-theory-based reliability design optimization (EBDO), this paper presents a decoupling approach which provides an effective tool for the reliability design of some complex structures with epistemic uncertainty. The approach converts the original nested optimization into a sequential iterative process including design optimization and reliability analysis. In each iteration step, through the uniformity algorithm, the original EBDO is firstly transformed to a conventional reliability-based design optimization (RBDO) and an optimal solution is obtained by solving it. At the solution, the first-order approximate reliability analysis method (FARM) is then used to perform the evidence-theory-based reliability analysis for each constraint. In addition, the RBDO solving and the evidence-theory-based reliability analysis are carried out alternately until reaching the convergence. Finally, two numerical examples and a practical engineering application show the effectiveness of the proposed method.

An evidence theory and interval analysis-based method of reliability design optimization is proposed.For the preconcerted failure probability f_0,the plausibility measure of reliability constraint,P1(F),is calculated by using evidence theory.With P1(F)≤f_0 used as a surrogate model of reliability constraint,an evidence theory-based model of generalized reliability design optimization is formulated.If there is only random uncertainty in the formulated model,this model degenerates to a conventional model of reliability design optimization.A numerical method for calculating P1(F) is presented.The theory of interval analysis is used for approximately calculating the value range of constraint function,then the computational efficiency of P1(F) is significantly improved and the computational cost for solving the reliability optimization model is reduced.A genetic algorithm program for the evidence theory and interval analysis-based reliability design optimization is developed.The proposed method is demonstrated with a pressure vessel example.The example shows that the proposed method is effective and practical.

Foam filled thin-wall structures can effectively improve the crashworthiness of automobile thin-walled element.To design more efficient and lighter absorber structures and meet the safety design demand,a novel aluminum foam filled double square structure is presented,and deterministic optimum design of energy absorption characteristic is preformed for aluminum foam filled structures.However,the design parameters,such as the thickness,yield stress of thin wall and density of aluminum foam,can often have uncertainties in simulation and manufacturing process,it can make the deterministic optimal solution convergence to the design constraint boundaries,and lose the reliable design demand.Hence,reliability design optimization scheme based on kriging approximate models and first order reliability method is presented.Finally,reliability design optimization with parameters uncertainties of aluminum foam filled structures is performed.The optimal results show that the reliability optimal solutions not only are far from the constraint boundaries,but also better meet the safety and reliability of aluminum foam filled structure design demand.

Reliability-based design optimization (RBDO) is intrinsically a double-loop procedure since it involves an overall optimization and an iterative reliability assessment at each search point. Due to the double-loop procedure, the computational expense of RBDO is normally very high. Current RBDO research is focused on performance functions having explicit analytical expression and readily available gradients. This paper addresses a more challenging type of RBDO problem in which the performance functions are computation intensive. These computation intensive functions are often considered as a “black-box” and their gradients are not available or not reliable. Based on the reliable design space (RDS) concept proposed earlier by the authors, this paper proposes a Reliable Space Pursuing (RSP) approach, in which RDS is first identified and then gradually refined while optimization is performed. It theoretically avoids the nested optimization and probabilistic assessment loop. This approach can apply to RBDO problems with either analytical or blackbox performance functions. Three well known numerical problems from the literature are used to test and demonstrate the effectiveness of RSP.