Search

This paper proposes an improved sampling-based approach for spacecraft proximity operation path planning under Clohessy-Wiltshire-Hill dynamics. The proposed approach is based on a modified version of the FMT* (Fast Marching Tree) algorithm with safety strategy which is divided into three parts: (1) incorporating relative ellipse to simplify the sampling state space and avoid collision with target; (2) combining internal/external ellipse-based collision detection algorithms for hovering obstacle and non-coplanar ellipse obstacle; (3) using rotating hyperplane method to handle the coplanar ellipse obstacle and uncertainty obstacle. By referring the safety strategy to simplify the state space before FMT* algorithm, the approach can reduce the complexity of path planning, especially the resampling and collision avoidance detection cyclic process in FMT*, thereby improve the planning efficiency. Two simulated scenarios, a coplanar path planning with/without coplanar ellipse obstacle, are developed to illustrate the approach. As a result, the proposed approach appears to be potential for spacecraft proximity real-time path planning as well as other complex space mission path planning generalized to different dynamics and environments, such as On-Orbit Service, and enabling a real-time computation of low-cost trajectory.

Fault prediction is critically important for many mobile equipments such as vehicles, ships and spacecrafts. Sensors deployed on these equipments continuously collect the status data, which are usually multivariate time series data. It is challenging to accurately predict the failure of the equipments based on the generated multivairate time series due to the complex correlations among the variables and the dynamic operation conditions. Though many methods have been proposed, they are not effective to provide an interpretable and accurate fault prediction result. This paper proposes a two-stage Interpretable Fault Prediction method based on Anomaly Detection and Anomaly Accumulation, called IFP-ADAC. Specially, we first design an anomaly detection module based on Generative Adversarial Nets due to the lack of samples. The generator captures the correlations among multiple variables and the temporal dependency within each variable jointly. Second, we design an anomaly accumulation model based on LSTM to capture the anomaly growth pattern, and the attention mechanism has been introduced to consider the severity of the detected anomalies. Compared with the end- to-end methods, our two-stage fault prediction method based on anomaly detection and accumulation has better interpretability. Extensive experiments conducted on two real-world datasets show the superior performance of our method.

We present a framework for reduction between different observation properties of discrete event systems modeled by automata under language-based dynamic observations defined in [1]. This framework encapsulates the reduction from coobservability to codiagnosability in [1] and from k-codiagnosability to coobservability in [2]. Under this framework, we show a reduction from observability to diagnosability, but a general reduction in the opposite direction does not exist.

We investigate how to distribute large quantity of sensors in such a way that workload is evenly shared by sensors. These techniques can also be used for regulating agile mobile sensors, such as UAVs, when the workload of sensors rather than the energy for mobility is the primarily concern. An algorithm is developed for partitioning sensing field such that different partitions have different sensor densities. The goal is to minimize unevenness of the workload to sensor density ratio. The output of the algorithm proved to be the global optimum for serving this purpose.

In various sensor network applications like monitoring and target localization, the occurrence of an event can only be effectively located if it falls within the sensing range of at least k ≥ 3 sensors. In this paper, we develop an algorithm to optimize to optimize the positions and orientations of heterogeneous sensors in order to maximize the probability of an event location being k-covered. Our proposed algorithms can handle rotatable mobile sensors with directional sensing regions. We show that our proposed algorithm converges to a local optimal solution, and present numerical results to verify its performance.

In many applications of discrete event systems, in addition to observing event occurrences, controllers directly detect changes from one subset of state space to another, and such detection often involves latency. Incorporating these practical imperatives into system design, we extend the works on state observation in [1] to model state-change perception and redefine the traditional coobservability accordingly. Given the number of supervisors fixed, a P-class verifier for the updated version of coobservability is also presented when controllers' system estimates depend on both transition-based event observations and state-change perception.

This paper investigates the operational research in active debris removal and focuses on the orbit control in multi debris removal. Far-range trajectory is the crucial phase which guides chaser to proximity of target debris in sequence and provides precondition for further rendezvous control. Far-range trajectory can be divided into two modes depending on velocity requirement at the engagement instant. A novel mode, fly-by trajectory, is firstly proposed for far-range trajectory control corresponding to debris capture by compact with whipple material. Compared with general rendezvous and capture mode, there is no requirement on relative velocity in fly-by mode. At last, hybrid optimization algorithm is applied to far-range trajectory optimization based on multi-revolution Lambert algorithm. Simulations are conducted with three scenarios to illustrate effectiveness of proposed concept. This paper summarizes relevant work in this field and presents our new research in far-range orbit control.

The magnetoelectric sensor has the advantages of high sensitivity, low noise and simple preparation process. It has great application prospects in weak magnetic field detection such as geomagnetic field, biomagnetic field and so on. What‘s more, the resolution of detection for low-frequency AC magnetic field can reach pT magnitude [1]. Because the piezoelectric material will produce a continuous output only when it is in a periodic vibration. so it needs to use modulation technology when measuring the statics magnetic field. The modulation techniques commonly used today are AC modulation and electric field modulation [2, ~3].But now, a new modulation method will be introduced in this paper, that is what referred to as rotation modulation. As shown in Fig.1(a), the component of the measured direct current projection on the shaft is a sinusoidal signal whose frequency is equal to the frequency of rotation, thus, we can measure constant magnetic field directly. More exciting, we may apply the rotation to MEMS structure to achieve miniaturization and integration. The test platform shown in Fig.1(b) is consist of rotation system and measurement system. Rotation system inclueds a ultrasonic motor and a fixture used for installing the magnetoelectric sensor and induction coil providing DC bias. The measurement system is composed of Signal Generator, Regulated DC Power Supply, Spectrum Analyzer, Magnetic Shield Tube. The magnetoelectric sensor used in the experiment was composed of Metglas/PZT/Metglas, which top layer and bottom layer are composed of two-layers of magnetostrictive Metglas. In order to obtain the best bias magnetic field, we firstly design a series of experiments with linearly increasing bias DC field at 2Hz(because the maximum rotation frequency of ultrasonic motor is 2Hz). Fig.2(a) shows that the optimal DC bias is about 8Oe. Then our group has tested the ME output voltage of magnetoelectric sensor with 10Oe measured DC field providing by coil, and the four groups experiments had been processed under condition of statics or rotation at three different speeds shown in Fig.2(b). There are three spikes whose value far greater than other‘s, and their position is just corresponding to the frequency of rotation. Fig.2(c) has shown almost linear change between ME voltage and the measured DC field. Thus, it can be conclued that the proposed modulation method is really effective. In these experiments, our team have used coils to ensure the DC bias can be controlled accurately. The next step we will use ferromagnet for bias. Using the proposed modulation method can realize high-precision DC magnetic field measurement like AC modulation and electric field modulation after enchancing ME out voltage through the selection of magnetic composite materials, the optimization of the rotating structure and the using of charge amplifier.

Opacity is an important information-flow property that arises in security and privacy analysis of cyber-physical systems. It captures the plausible deniability of the system's “secret” in the presence of a malicious intruder modeled as a passive observer. As a specific type of opacity, infinite-step opacity requires that the intruder can never determine unambiguously that the system was at a secret system for any specific instant in the past. Existing works on the analysis of infinite-step opacity only provide a binary characterization, i.e., a system is either opaque or non-opaque. However, a non-infinite-step-opaque system may only have a small probability of violation; this may be still tolerable in many applications. To analyze infinite-step opacity more quantitatively, in this paper, we investigate the verification of infinite-step opacity in the context of stochastic discrete-event systems. A new notion of opacity, called almost infinite-step opacity, is proposed to capture whether or not the probability of violating infinite-step opacity is smaller than a given threshold. This notion is weaker than its purely logical counter-part as it takes the transition probability of the system into account. We also provide an effective algorithm for the verification of almost infinite-step opacity.

In order to operate discrete event systems, agents utilize information from observing event sequences. This information is often incomplete due to communication defects, sensor malfunction, etc. Thus, understanding partial observation is essential to operating these systems. We develop, in this paper, an algorithm that reduces relative observability problems to observability problems in the context of centralized systems. Our extended results are in the context of a general language-based dynamic observation setup in which event-based observation in literature is a special case.

One of the major challenges in facial recognition is to find an efficient method to extract the necessary features from face images under various lighting conditions. To deal with this problem, a great deal of existing algorithms have focused on analyzing the first-order gradients of a face image. This method has proven to be mildly resistant to lighting variations. However, these algorithms fail to make adequate use of potential information obtained from high-order gradients. In this paper, we propose to explore such high-order gradients along with an algorithm framework that performs the following: (1)utilize not only the information obtained from a first-order derivative, but also make use of the information provided by a second-order derivative; (2)a directional second-order gradient operator that captures a face image's edges and orientations; (3)an image-processing technique developed to suppress the noise in computed gradients using a structure tensor to obtain the second-order gradient. Four operators; First-order Gradient Operator(FGO), Second-order Gradient Operator(SGO), Directional Second-order Gradient Operator( DSGO), and Multi-scale Directional Second-order Gradient Operator(Mul-DSGO) are taken into consideration. Extensive experimental results using Yale B and Extended Yale B databases show that SGO performs better than FGO, DSGO performs better than SGO, and Mul-DSGO performs better than DSGO for facial recognition under various lighting conditions.

This paper presents a guidance and control scheme based on three dimensional Augment Proportion Navigation (APN) for rendezvous with uncooperative target based on rotating line of sight coordinates. Application of classic proportion navigation is expanded to rendezvous with introduction of sliding mode control along line-of-sight. The algorithm can deal with the maneuvering target by measuring only elevation and azimuth angles between each chaser and target. Cooperative scheme provides the distance measurement by estimation which improves the observability of angle only navigation. Simulation scenarios with various maneuverability and control parameter settings are conducted to demonstrate the efficiency of proposed guidance and control system.

A controller activates sensors to acquire information in order to correctly implement a supervisory control law. Sensor activation is costly because systems often operate in restrictive environments with limited resources. In order to reduce unnecessary sensor activation, along with controlling the system, the supervisory controller continuously adjusts its selecting information in response to its observation of changing circumstances. Building upon existing methods for computing a transition-based minimal sensor activation policy, this paper presents algorithms for computing all minimal transition-based sensor activation policies. This result is also extended for minimizing the numerical sensor activation cost for systems modeled by stochastic automata.

Based on geographic information systems (GIS) and remote sensing (RS), five primary factors affecting soil erosion in the ErHai Lake Basin are extracted rainfall erosivity, soil erodibility, terrain, vegetation coverage, and soil and water conservation measures. Meanwhile, the amount of soil erosion is estimated using the revised universal soil loss equation (RUSLE) model. Furthermore, the spatial distribution characteristics were also analyzed with the application of GIS in the view of elevation, slope and land use types. The results also showed that the soil erosion modulus was 64.68t/(hm2·a), which belonged to the grade of intensity erosion, and a erodible area of 1404.279km2(61.12% of the whole area). The results also revealed that soil erosion, especially soil erosion with high grade, mostly distributed in the following areas, that is elevation area from 2000 to 2500 meters, slope area from 25 to 45 degree, and unused land. The findings from the study are valuable for the soil erosion control and decision-making for the ErHai Lake Basin.

Precision analysis is an important task of the exoatmospheric endgame guidance. The conventional method Monte Carlo is time-consuming. Covariance analysis description equation technique (CADET) needs to linearize the nonlinear model, which is complicated and the precision is difficult to be guaranteed. In order to overcome these problems, a new method based on the Unscented Transformation (UT) for nonlinear endgame guidance error analysis is proposed. The error transformation model and the error propagation model are deduced for the nonlinear exoatmospheric endgame guidance system. And they are testified by simulation. Simulation results are compared with that of Monte Carlo method and indicate that the guidance high precision can be guaranteed by the proposed method high at low time-cost and the process is simple.

This note studies two closely related decision problems from the area of decentralized synthesis. The first problem is about deciding the existence of a non-empty decomposable sublanguage of a regular language, called Decomposable Sublanguage Problem; the second is Joint Observability Problem, which is known in the literature. We provide characterizations of the decidability of both decision problems. Then, we study the undecidability of related problems such as Distributed Supervisory Control Problem and Parametrized Control Synthesis Problem. To address these undecidability results, we also propose a couple of heuristics that can provide solutions to the two studied problems, but also to more general problems from control or trace theory.

The digital watermarking for video is an effective method to protect the video copyright. Based on the previous video watermarking techniques, this paper summarizes their theories, features, model and classic algorithms and then discusses the algorithms' advantages and disadvantages. Finally, the key techniques and the development tendency of the video watermarking are discussed.

The problem of dynamic sensor activation for event diagnosis in discrete event systems is considered. Diagnostic agents are able to activate/deactivate sensors dynamically during the operation of the system. Sensor activation policies are functions that determine which sensors are on after a system trajectory. The sensor activation policy must be sufficient to fulfill the diagnosis duty. To compute minimal policies, in previous work, we propose language partition methods to restrict the solution space to be finite. In this paper, we present an algorithm that can be used to compute a provable minimal sensor activation policy without such a restriction. The output of our algorithm, which is a deterministic finite-state automaton, can be directly used to control sensors and diagnose the system.

This work is motivated by the U.S. Air Forces Intelligence, Surveillance and Reconnaissance (ISR) mission, where an Unmanned Aerial Vehicle (UAV), or an agent, is to fly over a number of unidentified objects within a given search area, collect information using onboard sensors, and classify the objects. The problem is challenging because the mission time is limited, the agent is only provided with partial a priori information, and the amount of information that the sensor can measure is dependent on the range and the azimuth of the explorer with respect to the object. A sequential decision problem (path planning) is posed that incorporates the potential loss of the classification outcome that is made by an autonomous moving agent. The problem is solved using stochastic dynamic programming. The resulting path exploits the interaction between the agent kinematics, informatics, and classification. Numerical simulation results that validate the concept are presented.

This paper proposes a real coded hybrid local evolutionary strategy (μ + λ) (RCHLES) to predict 3D protein tertiary structure of HP model with Ab Initio method. This strategy codes each amino acid sequence by the proposed discrete real code instead of binary code. The strategy devises the genetic operators, selection strategy, the fitness function, control parameter and termination criteria. The method is implemented by C programming language on CPU 1G Htz, 512M memory. The tests' results demonstrate that the algorithm has the advantage of diversity of structure and strong computational ability. The strategy can relatively avoid prematurely converge.

Motivated by optimizing the performance of human operators of Unmanned Aircraft Systems (UAS), we consider the use of stochastic approximation algorithms in this paper. With the increasing levels of automation available for both military and civilian unmanned vehicle systems, the human operators are expected to contribute as high-level planners and decision makers more than as remote-control pilots. Humans and, to a lesser extent, unmanned vehicles, are limited by workload. To improve the performance of the mixed systems of humans and unmanned vehicles, it is important to find the workload for human operators that will achieve the best rate of correct decision making. Although the performance of human operators is known to be a concave function of their arousal level, as described by the Yerkes-Dodson law, precise descriptions of such a function and how workload related to arousal level remain unknown in general. Furthermore, assessing the correctness of decisions is difficult in practice due to uncertainties in real situations, and due to the small number of data sets available for training of operators, and cost of such training. To bypass these difficulties and optimize operators' performance, we adjusted traditional stochastic approximation formulation and developed algorithm to solve it. Our approach can be used to optimize the performance of multiple human operators without knowing the correctness of any individual's decisions.

To observe occurrences of an event in a discrete event system, a sensor must be placed and activated. Both placement and activation incur costs. Assuming sensors have been placed, to minimize costs, one would like to minimize the activation of the sensors. In this paper, an online algorithm is developed to minimize sensor activation. The discrete event system under consideration is modeled by a finite automaton. The objective of observation is to distinguish certain pairs of states in the automaton, which are called a specification. This specification is extended to account for future evolution of the system under partial observation. Using the extended specification, the online algorithm presented needs only to look one step ahead and requires only to remember the current state estimate. The algorithm is of polynomial complexity at each step. The activation policy generated by the online algorithm is proved to be feasible, minimal, and to satisfy the specification. An example is presented to illustrate the results.

Disassembly hive is a kind of model expressing topological relationships of adjacent parts in disassembly body, which reflects the integration module between parts, as well as the feasibility after the integration. Expressing the disassembly process with disassembly hive in different dimensions and different disassembly planes and comparing their movement feasibility will provide better solutions to the exponential explosion brought by simple algorithm; improve the efficiency of path planning. Modularization properties of disassembly hive makes the disassembly process hierarchical, intelligent and avoid a great deal of calculation.

We consider the verification of the properties of diagnosability and codiagnosability in discrete event systems where observations are dynamic. Instead of having a fixed set of observable events, it is assumed that the observability properties of an event are state-dependent: an event occurrence at a state will be observable to a diagnosing agent if that agent activates in time the sensor corresponding to the event or receives a communication about the occurrence of the event. In this context, the known polynomial-complexity tests based on verifier automata for the properties of diagnosability and codiagnosability with fixed observable event set(s) are no longer directly applicable. We develop a new testing procedure that can handle state-based dynamic observations and remains of polynomial complexity in the state space of the system. This new testing procedure employs a covering of the state space of the system based on cluster automata, which enhances its computational efficiency. Based on cluster automata, a new type of verifier automaton is built, called the C-VERIFIER. Our use of cluster automata and C-VERIFIERS also yields computational savings in the special case of fixed observable event sets.

We consider the problem of dynamic sensor activation for event diagnosis in partially-observed discrete-event systems. The observing agent is able to activate sensors dynamically during the evolution of the system. The sensor activation policy is the function that describes which sensors are to be activated after an observed string of events. The sensor activation policy must achieve the requirements of the property of diagnosability previously defined for discrete event systems. A policy is said to be minimal if there is no other policy, with strictly less sensor activation, that achieves diagnosability. For the purpose of computing minimal policies, we define language partition methods that lead to efficient computational algorithms. Specifically, we define “window-based” language partitions that lead to scalable algorithms for computing minimal policies. By increasing the size of the window in this class of partitions, one is able to refine the solution space over which minimal solutions are computed.

Given an input sentence and a target word and its frame, automatic semantic role labeling on the Chinese FrameNet (CFN) database can be divided into two subtasks. One is identification of the boundaries of semantic roles, and the other is the classification of semantic roles. In this paper, the tasks are regarded as a sequential tagging problem in the sentence at word-level, so the model of conditional random fields was adopted. The best feature templates of the model were chosen by applying orthogonal arrays. The training and testing data sets consist of the sample sentences of 25 frames selected from the current CFN corpus. The experimental results in our test data set show that the F-measure for identifying boundaries of semantic roles reached 70.42\%, and the accuracy in classifying semantic roles achieved 80.4\%, but when the two subtasks were sequential automatically processed, the result was 59.48\% F-measure.

The problem of sensor activation in a controlled discrete event system is considered. Sensors are assumed to be costly and can be turned on/off during the operation of the system. The agent activates sensors as needed in order to observe the trajectories of the system and correctly implement the given feedback control law. Different policies of sensor activation can be used by the agent. A policy is said to be minimal if any strictly less activation prevents the correct implementation of the control law. A systematic formulation of the sensor activation problem is obtained and a solution procedure is proposed. An algorithm that computes minimal sensor activation policies is presented. The algorithm is of polynomial complexity in both the number of states and the number of events in the system.

In this paper, the problem of minimizing communication in distributed systems is considered in a discrete-event formalism where the system is modeled as a finite-state automaton. There is a set of n communicating agents observing the behavior of the system for purposes of control or diagnosis. A set of communication policies for the agents is said to be minimal if communications cannot be removed without affecting the correctness of the solution. A more general formulation of the communication structure than prior works on this problem is considered. Under an assumption on the absence of cycles (other than self-loops) in the system model, an algorithm that computes a set of minimal communication policies in polynomial-time in the number of states of the system is presented. This is the first algorithm of its kind that achieves this complexity result.

The problem of minimizing communication in distributed systems is considered in a discrete-event formalism where the system is modeled as a finite-state automaton. There are a central station and a set of N communicating local agents observing the behavior of the system. The central station needs to know exactly the state of the system, whereas local agents need to disambiguate certain pre-specified pairs of states for purposes of control or diagnosis. Communication occurs only between the central station and local agents but not among local agents. A set of communication policies is said to be minimal if communications cannot be removed without affecting the correctness of the solution. Under an assumption on the absence of cycles (other than self-loops) in the system model, an algorithm that computes a set of minimal communication policies in polynomial-time in all input variables of the system is presented. These results improve upon algorithms for solving minimum communication problems.