Kinetic feedback design for polynomial systems

New computational methods are proposed in this paper to construct polynomial feedback controllers for the stabilization of polynomial systems with linear input structure around a positive equilibrium point. Using the theory of chemical reaction networks (CRNs) and previous results on dynamical equivalence, a complex balanced or weakly reversible zero deficiency closed loop realization is achieved by computing the gain matrix of a polynomial feedback using optimization. It is shown that the feedback resulting in a complex balanced closed loop system having a prescribed equilibrium point can be computed using linear programming (LP). The robust version of the problem, when a convex set of polynomial systems is given over which a stabilizing controller is searched for, is also solvable with an LP solver. The feedback computation for rendering a polynomial system to deficiency zero weakly reversible form can be solved in the mixed integer linear programming (MILP) framework. It is also shown that involving new monomials (complexes) into the feedback does not improve the solvability of the problems. The proposed methods and tools are illustrated on simple examples, including stabilizing an open chemical reaction network.     

A network dynamics approach to chemical reaction networks

A treatment of a chemical reaction network theory is given from the perspective of nonlinear network dynamics, in particular of consensus dynamics. By starting from the complex-balanced assumption, the reaction dynamics governed by mass action kinetics can be rewritten into a form which allows for a very simple derivation of a number of key results in the chemical reaction network theory, and which directly relates to the thermodynamics and port-Hamiltonian formulation of the system. Central in this formulation is the definition of a balanced Laplacian matrix on the graph of chemical complexes together with a resulting fundamental inequality. This immediately leads to the characterisation of the set of equilibria and their stability. Furthermore, the assumption of complex balancedness is revisited from the point of view of Kirchhoff's matrix tree theorem. Both the form of the dynamics and the deduced behaviour are very similar to consensus dynamics, and provide additional perspectives to the latter. Finally, using the classical idea of extending the graph of chemical complexes by a ‘zero’ complex, a complete steady-state stability analysis of mass action kinetics reaction networks with constant inflows and mass action kinetics outflows is given, and a unified framework is provided for structure-preserving model reduction of this important class of open reaction networks.     

深空探测航天器姿态控制系统故障定位

执行机构与敏感器故障检测与定位是深空探测任务卫星平台可靠运行的前提和保障.本文从数据的角度出发,结合姿控系统工作机理,提出一种基于神经网络和支持向量机结合的故障诊断方法用于检测并定位故障.故障诊断方法分为3步,首先采集姿控系统的状态信息,采用神经网络对闭环姿控系统中未知动态特性建模并进行预测;然后将姿控系统敏感器信号与神经网络预测输出比较生成残差并提取故障特征;最后采用支持向量机辨识残差特征检测故障,并结合运动学特性分析定位故障.仿真结果表明本文所提方法可以有效提取、辨识故障特征,实现执行器与敏感器的故障检测定位.     

Positive compositional algorithms in chemical reaction systems

Algorithms which relate to the generation of positive compositional structures in chemical reaction systems are presented. The algorithms are applied to the following problems: (1) determine if a specified reaction system is proper; (2) given a reaction system, determine a minimum initial mix which allows all reactions to proceed, or conversely given an initial mix, determine the number of free reactions; (3) given a reaction system and a consistent set of reactant molecular weights, determine a proper moiety structure for the reactants; (4) given a steady-state network and a consistent set of one-way reaction rates, determine a proper reaction path structure for the network.     

The Complete Stoichiometer

The Complete Stoichiometer, a computer program used in solving stoichiometric problems in chemical, biochemical or metabolic systems, is described. The program contains general facilities for handling the stoichiometric and real matrices used in chemical network problems. The program contains a number of utility and application subprograms. The utility programs provide for input of matrices in both standard form and symbolic equation form, editing and combination of matrices; storing and retrieving stoichiometric matrices from the disk; printing of reports; a BlackBoard subprogram which allows stoichiometric matrices to be manipulated on the screen with a command-driven language. The application programs are: standard processing of network stoichiometry to obtain reaction trees and reaction loops, the independent driving potentials, and the conserved moieties; the standard processing of an atomic composition matrix to obtain the set of independent reactions interconnecting the reactants, and a possible moiety structure; factoring of an atomic composition matrix to obtain the structure of reactants in terms of moieties and of moieties in terms of atoms; generation of reduced equivalent networks when some reactions are in rapid equilibrium and some reactants are in steady-state; a robust mass action chemical equilibrium solver.     

Landing Auto‐Pilots for Aircraft Motion in Longitudinal Plane using Adaptive Control Laws Based on Neural Networks and Dynamic Inversion

This paper presents two new automatic landing systems (ALSs) for aircraft motion in longitudinal plane; the model of the landing geometry determines the flight trajectory and the aircraft calculated altitude; the flight trajectory during landing consists of two parts: the glide slope and the flare. Both designed ALSs have an adaptive system (ACS) for the aircraft output's control; for the first ALS, the output vector consists of the flying altitude and the longitudinal velocity, while, for the second ALS, the output variables are the pitch angle and the longitudinal velocity of aircraft. The second variant of ALS also contains an altitude controller providing the calculated pitch angle. The calculated altitude (for the first ALS), the calculated pitch angle (for the second ALS), and the desired flight velocity are provided to the ACS by means of a block consisting of two reference models. ACS is based on the dynamic inversion concept and contains an adaptive controller which includes a linear dynamic compensator, a state observer, a neural network, and a Pseudo Control Hedging block. The paper is focused both on the design of the two ALSs and on their complex software implementation and validation.     

Hamiltonian perspective on compartmental reaction–diffusion networks

Inspired by the recent developments in modeling and analysis of reaction networks, we provide a geometric formulation of the reversible reaction networks under the influence of diffusion. Using the graph knowledge of the underlying reaction network, the obtained reaction–diffusion system is a distributed-parameter port-Hamiltonian system on a compact spatial domain. Motivated by the need for computer-based design, we offer a spatially consistent discretization of the PDE system and, in a systematic manner, recover a compartmental ODE model on a simplicial triangulation of the spatial domain. Exploring the properties of a balanced weighted Laplacian matrix of the reaction network and the Laplacian of the simplicial complex, we characterize the space of equilibrium points and provide a simple stability analysis on the state space modulo the space of equilibrium points. The paper rules out the possibility of the persistence of spatial patterns for the compartmental balanced reaction–diffusion networks.     

Simulation of the oregonator model by using deterministic ARMS

We propose a deterministic method for simulating chemical reactions, deterministic abstract rewriting system on multisets (DARMS), which is based on the concept of mass action low. The feasibility and utility of DARMS are demonstrated by applying it to the oregonator, which is a well-known model of the Belousov-Zhabotinskii (BZ) reaction. This work was presented in part at the 13th International Symposium on Artificial Life and Robotics, Oita, Japan, January 31–February 2, 2008     

Constructing and visualizing chemical reaction networks from pi-calculus models

The π-calculus, in particular its stochastic version the stochastic π-calculus, is a common modeling formalism to concisely describe the chemical reactions occurring in biochemical systems. However, it remains largely unexplored how to transform a biochemical model expressed in the stochastic π-calculus back into a set of meaningful reactions. To this end, we present a two step approach of first translating model states to reaction sets and then visualizing sequences of reaction sets, which are obtained from state trajectories, in terms of reaction networks. Our translation from model states to reaction sets is formally defined and shown to be correct, in the sense that it reflects the states and transitions as they are derived from the continuous time Markov chain-semantics of the stochastic π-calculus. Our visualization concept combines high level measures of network complexity with interactive, table-based network visualizations. It directly reflects the structures introduced in the first step and allows modelers to explore the resulting simulation traces by providing both: an overview of a network’s evolution and a detail inspection on demand.     

An ant colony algorithm for the multi-compartment vehicle routing problem

We demonstrate the use of Ant Colony System (ACS) to solve the capacitated vehicle routing problem associated with collection of recycling waste from households, treated as nodes in a spatial network. For networks where the nodes are concentrated in separate clusters, the use of k-means clustering can greatly improve the efficiency of the solution. The ACS algorithm is extended to model the use of multi-compartment vehicles with kerbside sorting of waste into separate compartments for glass, paper, etc. The algorithm produces high-quality solutions for two-compartment test problems.     

基于门控图卷积神经网络的有机化学反应预测

随着现代医药技术和计算机技术的发展,采用人工智能技术来加速药物的研发进度成为了研究热点,而对有机化学反应产物的高效预测是药物逆合成路线设计中的关键问题。针对样本数据集中化学反应类型分布不均匀的问题,提出了一种主动采样训练下的门控图卷积神经网络（ASGGCN）模型。首先,输入化学反应物的简化分子线性输入规范（SMILES）编码,通过门控图卷积神经网络（GGCN）以及注意力机制预测反应中心所在位置;然后,根据化学约束条件和候选反应中心枚举出可能的化学键组合来生成候选产物,再通过门控图卷积差分网络对候选产物进行筛选;最终,得到反应产物。门控图卷积神经网络拥有三个权重参数矩阵并通过门控对信息加以融合,与传统的图卷积神经网络相比,它能获取更加丰富的原子隐藏特征信息。通过主动采样的方式进行训练,使得该模型能够兼顾较差样本和普通样本的分析能力。实验结果表明,所提模型对化学反应产物的Top-1预测准确率可达87.2%,对比Weisfeiler-Lehman差分网络（WLDN）模型提高了1.6个百分点,可见模型能够更准确地预测有机化学反应产物。     

Nonlinear model-state feedback control for nonminimum-phase processes

A general nonlinear controller design methodology for continuous-time nonminimum-phase systems is presented, which utilizes synthetic outputs that are statically equivalent to the original process outputs and make the system minimum-phase. A systematic procedure is proposed for the construction of statically equivalent outputs with prescribed transmission zeros. The calculated outputs are used to construct a model-state feedback controller. The proposed method is applied to a nonminimum-phase chemical reactor control problem where a series/parallel reaction is taking place.     

Ant Colony Search Algorithm for Optimal Strategical Planning of Electrical Distribution Systems Expansion

Strategical planning is one of many research fields in the design of electrical distribution systems. The problem of strategical planning is a multiobjective combinatorial problem and the search space may often be quite large concerning to the options. The aim is to identify a strategy of expansion of a given distribution system in a given timeframe. For this problem, the search space is created beforehand by running a multiobjective optimisation algorithm for the optimal design of distribution networks for different load levels related to different years. The sets of Pareto-optimal solutions obtained for each load level at each year are equivalent in terms of the considered objectives, these being minimum losses, installation costs, and minimum unavailability. The problem of the identification of the optimal expansion strategy through these chronologically intermediate solutions leading to the final target configuration at the last year has been solved herein using an ACS (Ant Colony Search) algorithm. In order to verify the efficiency of the ACS algorithm, a small size application has been carried out and results have been compared to those obtained with enumeration. Then, a Simulated Annealing (SA) approach was used for a larger size test problem and results were compared to those obtained using the ACS. For this problem, the ACS demonstrated to be more robust than SA with higher quality results.     

基于PCS7的带搅拌釜式反应器的温度控制

带搅拌釜式反应器是一种常见的化工生产设备,为了保证生产安全、产品质量,需要对反应器的温度进行监控.针对釜式反应器的工艺特点和控制要求,研发了基于西门子PCS7系统的控制方案.讨论了控制系统相应的网络结构及硬件配置.提出了问歇反应温度控制分预热、升温、保温这三个阶段的控制策略.结果表明,控制方案使系统抗干扰能力和适应参数变化的能力都优于常规控制,达到了很好的控制效果.     

基于ACS优化BP神经网络的交通流量短时预测方法

交通流量预测是智能交通系统中非常重要的研究领域,因为交通流量的复杂性,传统的预测方法不能很好地预测。提出一种基于[t]分布自适应变异优化的布谷鸟算法,通过动态变异控制尺度和设置多个自由度来构造自适应变异算法,可以获得优于高斯变异和柯西变异的整体优化效果。在此基础上,提出改进布谷鸟搜索算法优化神经网络的交通流量预测模型（ACS-BPNN）,通过优化BP神经网络的初始权值和阈值参数,以提高短时交通流量预测精度。仿真结果表明,该方法取得比较好的预测结果。     

关于受限制的部分平衡复杂反应体系组成的计算

对于因动力学原因使某些反应不发生的受限制的复杂反应体系，本文介绍了计算组成的方法，它是在原有元素原子守恒方程组中加入了附加的“元素”，守恒条件，在新的物质守恒方程组和组成非负条件限制下，用自由能最小化方法计算受限制的部分平衡复杂反应体系的组成，根据上述原理编制了通用程序，对氨催化氧化和合成甲醇等受限制的复杂反应体系进行了计算，结果与文献值相符。     

A hybrid algorithm based on particle swarm and chemical reaction optimization

In this paper, a hybrid method for optimization is proposed, which combines the two local search operators in chemical reaction optimization with global search ability of for global optimum. This hybrid technique incorporates concepts from chemical reaction optimization and particle swarm optimization, it creates new molecules (particles) either operations as found in chemical reaction optimization or mechanisms of particle swarm optimization. Moreover, some technical bound constraint handling has combined when the particle update in particle swarm optimization. The effects of model parameters like InterRate, γ, Inertia weight and others parameters on performance are investigated in this paper. The experimental results tested on a set of twenty-three benchmark functions show that a hybrid algorithm based on particle swarm and chemical reaction optimization can outperform chemical reaction optimization algorithm in most of the experiments. Experimental results also indicate average improvement and deviate over chemical reaction optimization in the most of experiments.