Learning with case-injected genetic algorithms

This paper presents a new approach to acquiring and using problem specific knowledge during a genetic algorithm (GA) search. A GA augmented with a case-based memory of past problem solving attempts learns to obtain better performance over time on sets of similar problems. Rather than starting anew on each problem, we periodically inject a GA's population with appropriate intermediate solutions to similar previously solved problems. Perhaps, counterintuitively, simply injecting solutions to previously solved problems does not produce very good results. We provide a framework for evaluating this GA-based machine-learning system and show experimental results on a set of design and optimization problems. These results demonstrate the performance gains from our approach and indicate that our system learns to take less time to provide quality solutions to a new problem as it gains experience from solving other similar problems in design and optimization.     

Learning control of process systems with hard input constraints

In this paper, a novel and simple learning control strategy based on using a bounded nonlinear controller for process systems with hard input constraints is proposed. To enable the bounded nonlinear controller to learn to control a changing plant by merely observing the process output errors, a simple learning algorithm for parameter updating is derived based on the Lyapunov stability theorem. The learning scheme is easy to implement, and does not require any a priori process knowledge except the system output response direction. For demonstrating the effectiveness and applicability of the learning control strategy, the control of a once-through boiler, as well as an open-loop unstable continuously stirred tank reactor (CSTR), were investigated. Furthermore, extensive comparisons of the proposed scheme with the conventional PI controller and with some existing model-free intelligent controllers were also performed. Due to significant features of simple structure, efficient algorithm and good performance, the proposed learning control strategy appears to be a promising and practical approach to the intelligent control of process systems subject to hard input constraints.     

Optimisation of algorithm control parameters in cultural differential evolution applied to molecular crystallography

Evolutionary search and optimisation algorithms have been used successfully in many areas of materials science and chemistry. In recent years, these techniques have been applied to, and revolutionised the study of crystal structures from powder diffraction data. In this paper we present the application of a hybrid global optimisation technique, cultural differential evolution (CDE), to crystal structure determination from powder diffraction data. The combination of the principles of social evolution and biological evolution, through the pruning of the parameter search space shows significant improvement in the efficiency of the calculations over traditional dictates of biological evolution alone. Results are presented in which a range of algorithm control parameters, i.e., population size, mutation and recombination rates, extent of culture-based pruning are used to assess the performance of this hybrid technique. The effects of these control parameters on the speed and efficiency of the optimisation calculations are discussed, and the potential advantages of the CDE approach demonstrated through an average 40% improvement in terms of speed of convergence of the calculations presented, and a maximum gain of 68% with larger population size.     

An adaptive system for process control using genetic algorithms

Researchers at the U.S. Bureau of Mines have developed adaptive process control systems in which genetic algorithms (GAs) are used to augment fuzzy logic controllers (FLCs). GAs are search algorithms that rapidly locate near-optimum solutions to a wide spectrum of problems by modeling the search procedures of natural genetics. FLCs are rule based systems that efficiently manipulate a problem environment by modeling the “rule-of-thumb” strategy used in human decision making. Together, GAs and FLCs possess the capabilities necessary to produce powerful, efficient, and robust adaptive control systems. To perform efficiently, such control systems require a control element to manipulate the problem environment, an analysis element to recognize changes in the problem environment, and a learning element to adjust to the changes in the problem environment. Details of an overall adaptive control system are discussed. A specific laboratory acid-base pH system is used to demonstrate the ideas presented.     

Process program change control in a process environment

Process programs may change during enactment. The change of an enacting process program should be controlled. This paper describes process program change control in the CSPL (Concurrent Software Process Language) environment, which consists of techniques to (1) define change plans, (2) analyze and handle change impacts, and (3) resume process programs. Moreover, we provide an editor to guide process program change, with which the consistency between a change plan and the actual change can be enforced. We have used the environment in an experimental setting, and that experience shows that change control provided by the environment depends significantly upon process programs. If the dependencies among process components are well‐specified in a process program, the change control mechanism will detect all the affected components when the process program is changed. Copyright © 2000 John Wiley & Sons, Ltd.     

Learning algorithms for neural networks with the Kalman filters

Based on various approaches, several different learing algorithms have been given in the literature for neural networks. Almost all algorithms have constant learning rates or constant accelerative parameters, though they have been shown to be effective for some practical applications. The learning procedure of neural networks can be regarded as a problem of estimating (or identifying) constant parameters (i.e. connection weights of network) with a nonlinear or linear observation equation. Making use of the Kalman filtering, we derive a new back-propagation algorithm whose learning rate is computed by a time-varying Riccati difference equation. Perceptron-like and correlational learning algorithms are also obtained as special cases. Furthermore, a self-organising algorithm of feature maps is constructed within a similar framework.     

Learning control algorithms for tracking “slowly”varying trajectories

To date, most of the available results in learning control have been utilized in applications where a robot is required to execute the same motion over and over again, with a certain periodicity. This is due to the requirement that all learning algorithms assume that a desired output is given a priori over the time duration t in [0,T]. For applications where the desired outputs are assumed to change "slowly", we present a D-type, PD-type, and PID-type learning algorithms. At each iteration we assume that the system outputs and desired trajectories are contaminated with measurement noise, the system state contains disturbances, and errors are present during reinitialization. These algorithms are shown to be robust and convergent under certain conditions. In theory, the uniform convergence of learning algorithms is achieved as the number of iterations tends to infinity. However, in practice we desire to stop the process after a minimum number of iterations such that the trajectory errors are less than a desired tolerance bound. We present a methodology which is devoted to alleviate the difficulty of determining a priori the controller parameters such that the speed of convergence is improved. In particular, for systems with the property that the product matrix of the input and output coupling matrices, CB, is not full rank. Numerical examples are given to illustrate the results.     

The evolution of control algorithms in artificial pancreas: A historical perspective

Blood glucose control algorithms have evolved since the beginnings of the artificial pancreas in diabetes treatment. Although the main problem to solve remains as the regulation of blood glucose into the healthy physiological range, the schemes have evolved over time from on-off schemes to the data-based personalized schemes. The evolution has been in accordance with the understanding of glucose metabolism, the theoretical background to model it, and the availability of sensor technology. The algorithms have allowed the calculation of insulin infusion (sometimes glucagon or glucose), by the highly invasive intravenous route, up to schemes based on the minimally invasive subcutaneous route. Solutions have also been proposed to deal with delays in insulin action and glucose measurement, as well as robust schemes to reject disturbances due to meal intake, exercise, non-modeled dynamics, and parametric variations due to inter- and intravariability of metabolism. Other problem that control schemes have solved is the safety in insulin infusion, including the calculation of insulin on board to avoid episodes of hypoglycemia, guaranteeing glucose regulation in normoglycemia, and decreasing the time in hyperglycemia. To summarize the role of control algorithms in the development of the artificial pancreas, this paper presents a historical review of the proposed control algorithms, from the establishment of the paradigm of artificial pancreas to the present date.     

Multilevel optimization algorithms in computer aided production control in process industry

The use of optimization techniques in production control is discussed. Two optimization problems in relation to typical process industry complexes are formulated. A modified version of Tamura's algorithm is reviewed. The promising performance of the computerised algorithms is illustrated by numerical results. DISPATCHER, a practical operative decision support system, is described.     

基于遗传算法的污水处理过程优化控制方法

分析和研究了在限制有机物排放总量条件下,污水处理运行费用最低的优化控制问题,给出污水处理过程中多变量最优控制的数学模型,提出一种采用遗传算法求解污水处理过程优化控制问题新的计算方法,论述了二进制编码、适应度函数计算的设计与实现方法.该方法避免了对迭代初值进行猜测的困难,提高了计算效率.数值仿真说明在污水处理过程中采用遗传算法寻优是可行和有效的.     

Understanding ART-based neural algorithms as statistical tools for manufacturing process quality control

Neural networks have recently received a great deal of attention in the field of manufacturing process quality control, where statistical techniques have traditionally been used. In this paper, a neural-based procedure for quality monitoring is discussed from a statistical perspective. The neural network is based on Fuzzy ART, which is exploited for recognising any unnatural change in the state of a manufacturing process. Initially, the neural algorithm is analysed by means of geometrical arguments. Then, in order to evaluate control performances in terms of errors of Types I and II, the effects of three tuneable parameters are examined through a statistical model. Upper bound limits for the error rates are analytically computed, and then numerically illustrated for different combinations of the tuneable parameters. Finally, a criterion for the neural network designing is proposed and validated in a specific test case through simulation. The results demonstrate the effectiveness of the proposed neural-based procedure for manufacturing quality monitoring.     

随程序规模动态调整的通道优化布线算法

针对常规层次型布图算法在大规模程序中布线混乱的缺点,借鉴Sugiyama层次布局算法,提出了一种随着程序规模动态调整的通道优化布线算法。通过将节点的通道数目与程序规模建立函数关系,以解决现有算法在布图时出现的线路重叠和效率低下的问题;在布图中结合广义张量平衡思想,以减少交叉并实现布图的美观性;并根据调用节点之间的相对位置关系,给出了相应的线路分配和申请策略,实现了布线的有序性。实践证明,该算法能够提高布图效率,有效地减少交叉,实现节点的有序布线和实现简单等优点。     

Dynamically discovering likely program invariants to supportprogram evolution

Explicitly stated program invariants can help programmers by identifying program properties that must be preserved when modifying code. In practice, however, these invariants are usually implicit. An alternative to expecting programmers to fully annotate code with invariants is to automatically infer likely invariants from the program itself. This research focuses on dynamic techniques for discovering invariants from execution traces. This article reports three results. First, it describes techniques for dynamically discovering invariants, along with an implementation, named Daikon, that embodies these techniques. Second, it reports on the application of Daikon to two sets of target programs. In programs from Gries's work (1981) on program derivation, the system rediscovered predefined invariants. In a C program lacking explicit invariants, the system discovered invariants that assisted a software evolution task. These experiments demonstrate that, at least for small programs, invariant inference is both accurate and useful. Third, it analyzes scalability issues, such as invariant detection runtime and accuracy, as functions of test suites and program points instrumented     

数据平滑处理算法的编程

本文所写的程序是为了我所试验所用的电源控制的测试工作,是对电压值进行平滑处理。只要给出时间递增量就能计算出平滑处理后的电压值。在本文中我们使用了抛物线平滑处理方法,该处理方法我们将基于组件对象模型COM(Component Object Module)来实现,在客户端程序里在对该组件方法进行调用。经实践证明该方法有很好的平滑效果。     

数据平滑处理算法的编程

本文所写的程序是为了我所试验所用的电源控制的测试工作，是对电压值进行平滑处理。只要给出时间递增量就能计算出平滑处理后的电压值。在本文中我们使用了抛物线平滑处理方法，该处理方法我们将基于组件对象模型COM（Component Object Module）来实现，在客户端程序里在对该组件方法进行调用。经实践证明该方法有很好的平滑效果。     

Learning sparse classifiers with difference of convex functions algorithms

Sparsity of a classifier is a desirable condition for high-dimensional data and large sample sizes. This paper investigates the two complementary notions of sparsity for binary classification: sparsity in the number of features and sparsity in the number of examples. Several different losses and regularizers are considered: the hinge loss and ramp loss, and ?₂, ?₁, approximate ?₀, and capped ?₁ regularization. We propose three new objective functions that further promote sparsity, the capped ?₁ regularization with hinge loss, and the ramp loss versions of approximate ?₀ and capped ?₁ regularization. We derive difference of convex functions algorithms (DCA) for solving these novel non-convex objective functions. The proposed algorithms are shown to converge in a finite number of iterations to a local minimum. Using simulated data and several data sets from the University of California Irvine (UCI) machine learning repository, we empirically investigate the fraction of features and examples required by the different classifiers.