Search

Abstract: Automatic performance tuning of computationally intensive kernels in scientific applications is a promising approach to achieving good performance on different machines while preserving the kernel implementation''s readability and portability. A major bottleneck in automatic performance tuning is the computation time required to test a large number of possible code variants, which grows exponentially with the number of tuning parameters. Consequently, the design, development, and analysis of effective search techniques capable of quickly finding high-performing parameter configurations have gained significant attention in recent years. An important element needed for this research is a collection of test problems that allow performance engineering and mathematical optimization researchers to conduct rigorous algorithmic development and experimental studies. In this paper, we describe a set of extensible and portable search problems in automatic performance tuning (SPAPT) whose goal is to aid in the development and improvement of search strategies. SPAPT is a test suite that contains representative serial code implementations from a number of lower-level performance-tuning tasks in scientific applications. We present an illustrative experimental study on several problems from the test suite. We discuss important issues such as modeling, search space characteristics, and performance objectives. [Copyright &y& Elsevier]

Abstract: We report an autotuning technique for feed systems of a CNC machine tool using a least-square parametric system identification, a frequency-domain design method, and a fine-tuning method based on an optimal search algorithm. The feed system of a movable-column-type vertical machining center has a large moving mass because spindle and z-axis servo systems are housed in the column. Therefore, perturbation signal is carefully designed. Using a reasonably “smooth” multiharmonic signal, system identification is completed rapidly (8s) without causing excessive vibration or violating travel limits. Accurate information on the plant dynamics is obtained up to 30Hz. Feed systems (i.e., x, y, z axis) are modeled as 3rd-order transfer functions in a discrete domain, and compared with the identification results obtained using a Gaussian random sequence and a frequency-domain system-identification method. A proportional (P) controller is designed using numerical search in frequency domain that maximizes the tracking bandwidth and still keeps the system well damped. The frequency response is improved compared to that of a pole-placement method (ζ =0.707). P controllers of all the three axes that minimize contour error for three-dimensional a 20-mm-diameter circular trajectory are fine-tuned using a fast optimal-search method (440s). The contour error is significantly improved (average error of 2.25μm), compared to the results of the pole-placement method (37.89μm) and the frequency domain design method (12.37μm) when feed rate is 0.5m/min. The calculated stability margins of the controller gains are satisfactory. [Copyright &y& Elsevier]

Abstract: To minimize cross-talk in high-precision motion systems, the possibilities of data-based dynamic decoupling are studied. Atop a model-based and static decoupling, a multi-input multi-output (MIMO) and finite impulse response (FIR) dynamic decoupling structure is considered for machine-specific and performance-driven fine tunings. The coefficients of the FIR filters are obtained via data-based optimization, whilst the machine operates under nominal and closed-loop conditions. The FIR filters provide the ability to generate zeros outside the origin. These zeros are needed in the description of the low-frequency inverted plant dynamics. In addition, a low-pass filter structure supports the ability to generate poles outside the origin as to account for plant zeros. Both filter structures are effectively used in the high-precision motion control of a state-of-the-art scanning stage system and an industrial vibration isolation system. [Copyright &y& Elsevier]

This paper proposes a model reference-based autotuning technique for digitally controlled voltage-mode dc–dc converters. The proposed solution performs a comparison between the measured system impulse response and a reference derived from the desired dynamic performances, and minimizes the error function acting on the regulator parameters. Two different approaches are investigated to evaluate the impulse response: a deterministic tuning, injecting a duty-cycle impulse, and a statistical tuning, based on white noise injection in the control loop. Compared to existing approaches, this solution has the advantages of simplicity, small-signal processing, and on-line tuning capabilities. Experimental investigation is performed on a 20-A, 1.5-V digitally controlled synchronous buck converter, and both simulations and experimental results confirm the effectiveness of the proposed solution. [ABSTRACT FROM AUTHOR]

Abstract: Autotuning using relay feedback is widely used to identify low order integrating plus dead time (IPDT) systems as the method is simple and is operated in closed-loop without interrupting the production process. Oscillatory responses from the process due to ideal relay input are collected to calculate ultimate properties of the system that in turn are used to model the responses as functions of system model parameters. These theoretical models of relay response are validated. After adjusting the phase shift, input and output responses are used to find land mark points that are used to formulate algorithms for parameter estimation of the process model. The method is even applicable to distorted relay responses due to load disturbance or measurement noise. Closed-loop simulations are carried out using model based control strategy and performances are calculated. [Copyright &y& Elsevier]

Abstract: An autotuning framework based on an algorithm description language dedicated to array processing is introduced. The array processing language allows algorithm developers, may not be equipped with non-trivial knowledge of the increasingly complex architecture of today''s processors, to easily perform extensive platform-specific tuning to fully extract performance. A given array processing program is translated into candidate parallel C codes, the best of which can then be selected by empirical evaluation. The high-level abstraction nature of our language allows a unique array processing program to be exposed to wide range of high-level program transformations, thus raising chances of obtaining high performance code. Furthermore, it also enables to extract algorithm-level information which can then be used in heuristic methods for efficient optimization parameter space exploration in the autotuning process. The results of preliminary evaluations show that autotuning using the parameterized C code variants generated with high-level program transformation is useful. [Copyright &y& Elsevier]

Abstract: The popularity of Partitioned Global Address Space (PGAS) languages has increased during the last years thanks to their high programmability and performance through an effcient exploitation of data locality. This paper describes the implementation of effcient parallel dense triangular solvers in the PGAS language Unified Parallel C (UPC). The solvers are built on top of sequential BLAS functions and exploit the particularities of the PGAS paradigm. Furthermore, the numerical routines developed implement an automatic process that adapts the algorithms to the characteristics of the system where they are executed. The triangular solvers have been experimentally evaluated in two different multicore clusters and compared to message-passing based counterparts, demonstrating good scalability and effciency. [Copyright &y& Elsevier]

Abstract: Empirical performance optimization of computer codes using autotuners has received significant attention in recent years. Given the increased complexity of computer architectures and scientific codes, evaluating all possible code variants is prohibitively expensive for all but the simplest kernels. One way for autotuners to overcome this hurdle is through use of a search algorithm that finds high-performing code variants while examining relatively few variants. In this paper we examine the search problem in autotuning from a mathematical optimization perspective. As an illustration of the power and limitations of this optimization, we conduct an experimental study of several optimization algorithms on a number of linear algebra kernel codes. We find that the algorithms considered obtain performance gains similar to the optimal ones found by complete enumeration or by large random searches but in a tiny fraction of the computation time. [Copyright &y& Elsevier]

In this paper, we present Patus, a code generation and auto-tuning framework for stencil computations targeted at multi- and manycore processors, such as multicore CPUs and graphics processing units. Patus, which stands for ' Parallel Auto tuned Stencils,' generates a compute kernel from a specification of the stencil operation and a strategy which describes the parallelization and optimization to be applied, and leverages the autotuning methodology to optimize strategy-specific parameters for the given hardware architecture. [ABSTRACT FROM AUTHOR]

Sustained oscillations are obtained by putting relays in the feedback loop, which provide ultimate periods and gains of processes that can be used to tune PID controllers. The ultimate periods obtained by conventional relay autotuning have relative errors up to 5% for first order plus time delay processes. To improve the estimation of ultimate periods, modified relays such as a saturation relay and a relay with preload have been available. However, these modifications lose the binary (on-off) property of the conventional relay. Here relays with multiple switching which produce pulse-widthmodulation (PWM) signals are proposed. They retain the binary property and show improved identification accuracies of ultimate parameters. [ABSTRACT FROM AUTHOR]

Abstract: Tuning an ideal PID and then using its parameters in a real one, can yield undesired results. Nonetheless, many rules conceived for the ideal PID are successfully used in practice, owing to their greater flexibility. A systematic method would thus be useful to extend ideal PID tuning rules to the real PID, so as to preserve said flexibility while synthesising a realistic controller. This manuscript proposes a simple solution, and some examples to witness its validity. [ABSTRACT FROM AUTHOR]

Due to interactions between input and output, closed-loop multivariable systems yield undesirable responses. To characterize these responses through modelling, a systematic approach is followed to derive analytical expressions for relay feedback responses on 2×2 Multiple-Input Multiple-Output (MIMO) systems. This is based on the observation that a relay feedback test consists of a series of step inputs and a stable limit cycle implies a convergent infinite series. These analytical expressions in a closed-form solution are useful to identify unknown system parameters, understand interactions and diagnose faults at an early stage, and subsequently for autotuning. [ABSTRACT FROM PUBLISHER]

The article presents a study that proposed a process identification approach to estimate the frequency responses of the process from the activated process input and output. According to the authors, the proposed method offers exact estimates and can incorporate all three cases of process activation. A simulation research validated the proposed method's ability to address the limitations of the past methods and totally eradicate the effects of static disturbances.

Abstract: The use of an OpenMP compiler optimized for the corresponding multicore system is a good option, but it is possible in a system to have access to more than one compiler and different compilers can appropriately optimize different parts of the code. In this paper we present a proposal for an autotuning system for linear algebra routines that decides the best compiler for each situation, as well as other parameter values, as, for example, the number of threads to generate. [ABSTRACT FROM AUTHOR]

Abstract: In model-based PI/PID tuning regulators, the same set of I/O data and the same tuning rule can produce very different results, depending just on the procedure used to parametrise the process model. The problem is seldom addressed, but extremely relevant for the acceptability of model-based autotuners in the applications. This manuscript proposes a methodology to treat the model parametrisation and regulator tuning phases jointly, so as to circumvent said problem with affordable process stimulation and computational effort. The methodology can be generalised to different regulator structures, and even employed to devise new tuning rules. [Copyright &y& Elsevier]

In this article, a new on-line process identification method for the automatic tuning of the proportional–integral–derivative (PID) controller is proposed and applied to a full-scale wastewater treatment plant. The proposed method can identify general time-invariant linear systems with a frequency weighting, while retaining simplicity. The frequency weighting makes it possible to improve the robustness to underparameterization and uncertainties such as disturbances and noise. Under the proposed scheme, the high-order model is reduced to a lower order using a model reduction method to enable the tuning of the PID controller using the well-established tuning rules based on the first- or the second-order plus time delay model. Because the suggested process identification method uses only process input/output, it overcomes many of the disadvantages of previous identification methods; in particular, it does not require a special test signal generator such as a relay or a P controller. The properties of well-known signal generators are discussed, and guidelines are also outlined for choosing an appropriate signal generator. The proposed method showed better control performance in the simulation studies than previous methods, and was successfully applied to dissolved oxygen (DO) control in a full-scale wastewater treatment plant. [ABSTRACT FROM AUTHOR]

Abstract: This manuscript is part of a long-term research, aimed at establishing methodologically grounded relationships between model- and relay-based tuning of industrial regulators, and at consequently deriving synthesis procedures that couple the advantages of model-based tuning to the simplicity and clarity of relay-based identification. In this work, the addressed controller structure is the “PI+p”, i.e., a PI augmented with an additional, stable pole. The advantages of using the combined model-relay-based approach on that structure are evidenced, by means of both simulation and experimental results. [Copyright &y& Elsevier]

The control of various processes using a proportional-integral (PI) controller is investigated in this work. A new tuning rule based on the ultimate gain and phase margin specification is proposed. The tuning rule is given by analytical approximations. To obtain the tuning rule, a control signature is formulated using the ultimate properties of the system. The PI controller parameters are related with process model parameter to yield a model based tuning rule. Simulation examples are provided to demonstrate the performance and robustness of the present tuning rule. The new tuning rule requires a relatively small computational effort and is particularly useful for online applications. [ABSTRACT FROM AUTHOR]

For the automatic tuning of PID controllers, a new identification method is proposed to estimate the frequency responses of the process from the activated process input and output. It can extract many more frequency responses as well as guarantee better accuracy compared with the previous describing function analysis algorithm. Also, the proposed method can be applied to the case that the initial part of the activated process input and output is periodic (cyclic-steady-state), whereas the previous method using the modified Fourier transform cannot incorporate the case. [ABSTRACT FROM AUTHOR]

This paper presents industrial experience of process identification, monitoring, and control in a full-scale wastewater treatment plant. The objectives of this study were (1) to apply and compare different process-identification methods of proportional-integral-derivative (PID) autotuning for stable dissolved oxygen (DO) control, (2) to implement a process monitoring method that estimates the respiration rate simultaneously during the process-identification step, and (3) to propose a simple set-point decision algorithm for determining the appropriate set point of the DO controller for optimal operation of the aeration basin. The proposed method was evaluated in the industrial wastewater treatment facility of an iron- and steel-making plant. Among the process-identification methods, the control signal of the controller's set-point change was best for identifying low-frequency information and enhancing the robustness to low-frequency disturbances. Combined automatic control and set-point decision method reduced the total electricity consumption by 5% and the electricity cost by 15% compared to the fixed gain PID controller, when considering only the surface aerators. Moreover, as a result of improved control performance, the fluctuation of effluent quality decreased and overall effluent water quality was better. [Copyright &y& Elsevier]

Abstract: This manuscript deals with the automatic tuning of cascade control systems, that are frequently encountered in process control. A tuning method is presented, the main goal of which is to be simple to operate, and introduce a moderate process upset. The method combines an ad hoc relay-based identification procedure based on a single test, a specific choice of the controller blocks’ structure, and a set of tuning rules based on a particular use of the well known IMC principle. Simulations are reported to illustrate the effectiveness of the proposal within its applicability limits, that are characterised as rigorously as possible, and may be widened by some of the envisaged extensions. [Copyright &y& Elsevier]

A methodology of the first-order plus dead time process transfer function identification based on the measurement of the locus of a perturbed relay system from a single relay feedback test is proposed. Simple analytical formulas and rules that do not require iterative calculations and are suitable for the use within distributed control systems or stand-alone proportional-integral-differential controllers are obtained for this process model. [ABSTRACT FROM AUTHOR]

Abstract: A model-based autotuning method consists of an identification and a regulator tuning phase. To achieve satisfactory performance and robustness, it is advisable that both phases be tailored a priori to the characteristics of the observed process dynamics. Such characteristics include, but are not limited to, the model structure. For example, overdamped and underdamped models with the same pole-zero structure are parametrised and controlled in different ways. Step response data, that are typically used for the identification phase in the autotuning context, can also be pre-processed to reveal those characteristics. This paper presents a step response classification method suitable for the above purpose. The method is based on a polygonal curve approximation technique for data pre-processing, followed by a neural network classifier. Only normalised I/O data are employed, so that the neural network can be trained off-line with simulated data. Simulation results are reported to show the effectiveness of the proposed classification method in terms of the achievable tuning results. [Copyright &y& Elsevier]

The article proposes new relay feedback methods, which use integrals of the relay response instead of the point data. It is stated that the proposed methods guarantee advantages in obtaining the ultimate information process because effects of the high order harmonic terms are suppressed significantly using the integrals of relay responses. They include proposed methods for ultimate data estimation, nyquist point data estimation, steady state data estimation, first-order plus time delay model estimation, and critically damped plus time delay model estimation.

A signal processing circuit using autocorrelation statistics that intelligently selects the optimal hold-off time in an active quench and reset integrated circuit (AQR-IC) is presented. This circuit optimizes the hold-off time to reduce afterpulsing effects in any Geiger-mode avalanche photodiode (GM-APD). The described AQR-IC has a maximum count rate of 20 Mcounts/s, includes a subnanosecond comparator for quickly detecting incident photons, and incorporates a novel technique, based on silicon delay lines, for selecting 1 of 16 predefined, discrete, hold-off times between 5 and 660 ns. The accompanying hold-off time autotuning circuit sets the optimal hold-off time by assigning a 4-b output to the AQR-IC corresponding to the minimization of afterpulse effects in the GM-APD. The configuration sequence completes in a maximum time of 20 s, without operator intervention, and only requires repeating if the GM-APD is changed. [ABSTRACT FROM PUBLISHER]

Abstract: A simple approach to the automatic tuning of PID process controllers is proposed. Like the relay-based autotuner, its objective is to attain a design-point on the Nyquist diagram. By injecting sinewaves and employing a phase/frequency estimator, closed-loop adaptive tuning is possible and there is exact convergence to the design-point without the approximations of describing-function theory. The variant discussed here achieves a required phase margin and imposes a carefully chosen constraint on the controller parameters, leading to consistent behaviour for a wide variety of generic test-cases. A real-life demonstration on a non-linear flow rig is provided. [Copyright &y& Elsevier]

This paper presents the design and implementation of an advanced digital controller for a 1-kW H-bridge dc-dc power converter. A new control algorithm based on the active disturbace rejection concept is developed to cope with the highly nonlinear dynamics of the converter and the disturbances. An experimental digital control system is used to implement the new control strategy. It consists of a digital control board based on the TMS320C6711 digital signal processor chip, an analogy I/O board, and a complex programmable logic device pulsewidth-modulation generation board. Using a newly developed bandwidth-paramerization technique, an autotuning method based on noise quantification is also developed and tested. Experimental results show the advantages and flexibilities of the new control method for the H-bridge dc-dc power converter. [ABSTRACT FROM AUTHOR]

This paper presents an autotuning process controller aimed at providing efficient rejection of load disturbances in a class of situations that are quite typical in process control, and not easy to treat with most standard autotuning controllers, especially when not only the duration of the load disturbance response, but also the peak deviation of the controlled variable is an issue. The regulator structure is not fixed a priori; this is a peculiarity with respect to the main research stream on autotuning regulators, that refers essentially to fixed-structure (PID) regulators. Both simulation and laboratory examples are reported, to show the advantages of the proposed autotuning controller. [Copyright &y& Elsevier]

This paper presents and analyses a method for synthesizing the feedback block of a 2-d.o.f. real PID regulator, referring to the widely used ISA structure and particularly well suited for autotuning. A rigorous stability, performance, robustness and consistency analysis is provided, extending some results of the previous literature. An objective criterion is devised for selecting the method's (single) design parameter based on the process model employed and an overestimate of the corresponding process/model mismatch. Both simulation and practical application examples are reported. [ABSTRACT FROM AUTHOR]

In this work, a systematic approach to derive analytical expressions for relay feed back responses is proposed. It is based on the observation that a relay feedback test consists of a series of step inputs and a stable limit cycle implies a convergent infinite series. The second order plus dead time processes with different damping coefficients are used to illustrate the derivation. Analytical expressions for typical transfer functions up to 5th order are tabulated and a general expression for nth order system is also given. These analytical expressions are useful to identify unknown system parameters and subsequently for autotuning. [Copyright &y& Elsevier]

This paper describes a robust PID autotuner embedded in a block library for constructing both simulated and real control systems. The library has been implemented in a visual programming tool and contains all the functions for building complete systems, similar to those provided by a typical CACSD (computer-aided control system design) environment. After a brief overview of the library, the paper concentrates on describing the PID autotuning tool, its implementation and operation. The underlying theoretical background is simply sketched out to permit the necessary degree of comprehension, and for its complete illustration the interested reader is referred to more specific works. Some examples of the autotuner’s operation, one of which refers to a laboratory plant, are also reported. [ABSTRACT FROM AUTHOR]

The paper discusses the relatively new approach of relay autotuning and shows how a sound understanding of the approach requires knowledge of two approaches introduced early in the search for methods to study nonlinear control systems. The two approaches – the describing function method and procedures for evaluating limit cycles exactly in relay systems – are first reviewed. After a short section on techniques for choosing parameters of fixed controllers it is shown how relay autotuning can be used for PID controllers based on either plant critical point information or the estimation of a plant model. Finally some examples are given which illustrate the controller design techniques presented. [ABSTRACT FROM AUTHOR]

The execution of complex distributed applications in exascale systems faces many challenges, as it involves empirical evaluation of countless code variations and application runtime parameters over a heterogeneous set of resources. To mitigate these challenges, the research field of autotuning has gained momentum. The autotuning automates identifying the most desirable application implementation in terms of code variations and runtime parameters. However, the complexity and size of the exascale systems make the autotuning process very difficult, especially considering the number of parameter variations that have to be identified. Therefore, we introduce a novel approach for autotuning exascale applications based on a genetic multi-objective optimization algorithm integrated within the ASPIDE exascale computing framework. The approach considers multi-dimensional search space with support for pluggable objective functions, including execution time and energy requirements. Furthermore, the autotuner employs a machine learning-based event detection approach to detect events and anomalies during application execution, such as hardware failures or communication bottlenecks., Competing Interests: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest., (Copyright © 2021 Kimovski, Mathá, Iuhasz, Marozzo, Petcu and Prodan.)

We propose a new and simple on-line process identification method for the automatic tuning of the PID controller. It does not require a special type of test signal generators such as relay or P controller only if the signals are persistently exciting. That is, a user can choose arbitrary signal generators such as relay, a P controller, the controller itself, pulse signal and step signal generator because it needs only the measured process output and the controller output. It can incorporate nonlinearities due to actuator saturation or manual mode operation during identification work and shows a good robustness to measurement noises, nonlinearity of the process and disturbances. The proposed autotuner combined with the identification method and tuning rule using a model reduction shows good control properties compared with previous autotuning methods. [ABSTRACT FROM AUTHOR]