Since lots of systems in the industries have non-linearity and those structures are generally complicated, it is difficult to express them as mathematical models. The database-driven modeling (DDM) method which is a kind of Just-In-Time(JIT) modeling has been proposed as a method to construct a non-linear model. However, DDM method can not improve modeling accuracy in a complicated system including many needless variables. This study introduces the variable evaluation/selection method based on a random forest to improve the modeling accuracy of DDM method. The random forest can quantify the degree of contribution for variable prediction as importance. The effectiveness of the proposed scheme is numerically verified by some simulation examples.
This paper gives a method for tuning controllers using one-shot closed-loop data. The method estimates the plant impulse response from the data and the response is utilized as the plant model in a Finite Impulse Response filter form. The controller tuning is carried out so that the output fluctuation of the closed-loop system with the plant model due to step disturbance is minimized. Therefore, the proposed method does not require reference models for controller tuning that are necessary in data-driven controller tuning methods such as VRFT (Virtual Reference Feedback Tuning) and FRIT (Fictitious Reference Iterative Tuning). Finally, the effectiveness of the proposed method is shown through experiments of a position control system with a BLDC motor.
Feedback error learning (FEL) is a promising design method for two-degree-of-freedom (2DOF) control systems where feedforward control is tuned online. A tuning law has been proposed in the literature with stability proof for the single-input single-output (SISO) case under a strictly positive real condition. This paper generalizes this scheme to the multi-input multi-output (MIMO) case together with its stability proof, by making full use of left coprime fraction by polynomial matrices. Numerical simulation illustrates the effectiveness of the proposed MIMO FEL scheme.
Recently, data-driven controller tuning methods have been actively researched in the control engineering field, and some practical applications also have been reported. For Multi-Input Multi-Output (MIMO) systems, some data-driven controller tuning methods have been proposed. However, conventional methods for MIMO systems have some problems: restriction for reference models, increase of data acquisition, and restriction for use of open-loop data. This paper proposes a new data-driven controller tuning method to overcome the above problems. By shaping an input complementary sensitivity function, the number of data acquisition can be reduced and the closed-loop data can be applicable. To avoid restriction for reference models, we formulate a design problem based on the cost function of the Noniterative Correlation based Tuning (NCbT). However, since the proposed design method shapes an input complementary sensitivity function, an output complementary sensitivity function specifying the tracking performance might not be tuned well. This paper resolves this problem to impose a constraint for a condition number of a designed controller to the optimization problem. The constraint for the condition number of the designed controller is divided into two constraints to make the optimization problem convex. The effectiveness of the proposed method is verified by some numerical simulations.
This paper proposes a design method for continuous-time VRFT (Virtual Reference Feedback Tuning) for I-PD (Integral-Proportional, Differential) gain tuning using closed-loop step response data. The I-PD controllers can avoid an abrupt change of manipulated variables, and have been widely used for industrial process control. The present work derives an optimal pre-filter condition, and a design procedure of an optimal pre-filter. The efficiency of the proposed method is assured using a numerical example and a helicopter attitude control experimental result.
This letter proposes a new design method for controlling a dual-rate sampled-data control system, in which the sampling interval of a plant output is an integer multiple of the holding interval of a control input. In the proposed method, a discrete-time control law is extended such that the existing discrete-time response is maintained. As a result, the control law is re-designed independent of the reference response as well as the disturbance response in discrete time.
Fog services have been gaining momentum due to the demand generated by IoT-based services that require data and services to be provided at the edge of the network. This phenomenon creates several problems for Internet systems, such as security, and data management. Particularly, edge and fog computing addresses these problems and provides new services by solving these issues. To this end, a service-oriented router (SoR) was proposed and extended to provide edge and fog services by performing authorized stream content analysis (ASCA). Packet Capture (Libpcap) library was used to develop the initial implementation of the SoR to perform ASCA using conventional hardware and software platforms. However, Libpcap implementation together with conventional hardware demonstrated limited performance. Consequently, this paper proposes a method using Intel technologies such as the Intel Data Plane Development Kit and Hyperscan to accelerate stream processing and string matching on SoRs. The paper compares both implementations of SoRs and evaluates the performance with regard to packet throughput, CPU utilization, and memory usage. Furthermore, the latency of major functions is evaluated under Libpcap-based and DPDK-based SoR implementations. The results demonstrate an increase of more than 0.8Gbps in bandwidth in a 1Gbps link using less hardware utilization when SoR is implemented using Intel technologies compared to the original Libpcap implementation.
In recent years, research and development of highly accurate position detection indoors have attracted attention. In this paper, we propose a new scheme to improve accuracy of position detection indoors, identifying the direct wave between BLE transmitter (TX) and BLE receiver (RX), applying RD-CUBA-MUSIC(1) to indoor multi-path environment. In the proposed scheme, the direction of radio wave arrival is estimated in both directions of TX→RX and RX→TX. Estimation accuracy is improved by collating the arrival directions of the pair of two-way direct wave candidates. This Direction-Of-Arrival (DOA) estimation for direct wave is sequentially executed with a plurality of RXs and indoor position of TX is detected using the DOAs and the corresponding RX positions. In RD-CUBA-MUSIC, we select a beam pattern with high correlation with the sinc function beam pattern as a beam pattern of the ESPAR antenna. Computer simulation shows that the proposed scheme can achieve a detection accuracy within 1 m in error at a 91% probability. It also shows within 0.5 m in error at 67% even though it is in an indoor multi-path environment.
Recently, several estimation methods of 3D gazing points have been proposed. Many of these are measured by pupil and corneal reflection with infrared light. This conventional method required special measuring devices for estimating 3D gazing points. We previously developed an eye-gaze detection system by image analysis under natural light. This detection system uses an eye-gaze detection method based on the image analysis for horizontal and vertical eye movements. In addition, this detection system consists of a personal computer and single home video camera. In this paper, we present the new eye-gaze detection method for depth direction. We set the new depth parameter by obtained eye-gaze characteristics both eyes. Also, we proposed a depth direction identification method using it. As a result, depth eye-gaze detection became possible by our horizontal eye-gaze detection using eye images with offline process. This detection system employs input images including both eye areas by single home video camera under natural light. Similarly, horizontal and depth eye-gaze detection became possible by combining with our horizontal direction identification method.
Fall accidents at medical centers are one of the most serious problems, and many accidents happen at the bedside. In order to deal with this problem, this paper proposes a method for estimating an inpatient's fall risk using only depth data from Kinect. In this method, the obtained depth data are modified as if Kinect were above the center of the bed to estimate fall risks accurately. Then, these corrected data are divided into multiple cells, and human location is detected using gravity coordinates obtained from the data in each cell. Finally, the fall risks are estimated from the human location using fuzzy inference. Through verification using hospitalized subjects, the authors confirmed that the proposed method can estimate appropriate fall risks.
Critical dimension scanning electron microscope (CD-SEM) is widely used as an essential tool for measuring semiconductor patterns formed on a silicon wafer. For achieving a reliable measurement, CD-SEM requires to set up imaging sequence including correction of imaging position. The imaging region including unique patterns is selected manually as addressing point (AP) and positional error can be detected by a design-SEM matching at AP. In our previous work, we developed a multifactor layout analysis (MLA) method, which automatically selects AP position from design layout data of semiconductor patterns. In this paper, we propose an enhanced MLA method, which simultaneously optimizes position and size of the matching region of interest (ROI). For 100 evaluation points, the proposed method optimized the ROI region, which led to 51.3% reduction in addressing time compared to the conventional MLA method.
This research considers a service value creation methodology in IT solution service based on knowledge science. There are many business cases and researche studies of IT solution services. However, these studies do not consider knowledge itself for creating IT solution. The most important issue is how to extract and integrate different kinds of knowledge from different people for effective IT solutions. In this research, we analyze three cases based on knowledge science. The first case relates to the design office concept and MUSE. The other two are IT consulting cases. From analysing these cases, this research shows the importance of “Ba” which is knowledge space for creation the solution, and the importance of extracting explicit knowledge from implicit field knowledge by using meta-knowledge. The service value creation process for IT solution service can be described clearly and applied to various business system effectively by executing the results of this research.
Business processes are defined as functions and information relationships between functions in the phase of business requirement definition when enterprise information systems are developed. The function which is defined in the phase of business requirement definition is decomposed into the unit called a program or a module in the lower phase. The description method of the flow chart type such as DFD has been used for the visualization of the decomposition of function conventionally. In recent years, many enterprise information systems are improved by executing Business Process Re-engineering (BPR) or introducing Enterprise Resource Planning (ERP). In this case, the definition of the information relationships between functions is important. The examples utilizing the description method of the matrix type called G-RD are seen to visualize information relationships. Therefore, the comparative experiment is performed by system engineers to visualize the same business processes utilizing DFD and G-RD. In this experiment, system engineers decompose functions and visualize functions and information relationships between functions. And information relationships which are added or changed by the decomposition of functions are measured. As the result, it is proved that G-RD is able to have prevent omissions or errors of information relationships.