Search

The hyperspectral images(HSI) consist of many closely spaced bands carrying the most object information. While due to its high dimensionality and high volume nature, it is hard to get satisfactory classification performance. In order to reduce HSI data dimensionality preparation for high classification accuracy, it is proposed to combine a band selection method of artificial immune systems (AIS) with a hybrid kernels support vector machine (SVM-HK) algorithm. In fact, after comparing different kernels for hyperspectral analysis, the approach mixed radial basis function kernel (RBF-K) with sigmoid kernel (Sig-K) and applied the optimized hybrid kernels in SVM classifiers. Then the SVM-HK algorithm used to induce the bands selection of an improved version of AIS. The AIS was composed of clonal selection and elite antibody mutation, including evaluation process with optional index factor (OIF). Experimental classification performance was on a San Diego Naval Base acquired by AVIRIS, the HRS dataset shows that the method is able to efficiently achieve bands redundancy removal while outperforming the traditional SVM classifier.

For the purpose of meeting the rapid development of blood purification in China, improve the level of blood purification treatment, and get rid of the plight of the foreign technology monopolization to promise patients’ medical safety, a parameter-calibrator for the hemodialysis unit, which can detect simultaneously multi-parameter, is designed. The instrument includes a loop, which connects to the hemodialysis unit. Sensors are in the loop in series, so that the dialysis can flow through this loop and the sensors can acquisitive data of various parameters. In order to facilitate detection and carrying, the integrated circuit part modularly based on the ultralow-power microcontrollers,TI MSP430 is designed. High-performance and small-packaged components are used to establish a modular, high-precision, multi-functional, portable system. The functions and the key technical indexes of the instrument have reached the level of products abroad.

Influence of atmospheric pressure supplied on permittivity of the air-film is researched based on the capacitive testing method of the air film thickness of aerostatic bearing. An experiment platform is designed. The experimental results illustrate that permittivity has significant negative correlation with atmospheric pressure which varies from 0.1MPa to 0.48MPa when other environmental conditions remain unchanged. The curves conform to the fourth-order polynomial approximately. All of the values of R 2 are beyond 0.944 which means that trend lines fit the data curves well. Relative permittivity of the air film is between 0.996 and 1.324. This interval shows that weak current exists between restrictor and flat of the experiment which are not absolutely insulating and atmosphere of the air film is not pure. This result provides a basis both for establishing accurate mathematical model of air film thickness and capacitance value of the aerostatic bearing and for other exploratory experiments later.

In order to test the interface conditions between concrete and rebar, the propagation properties of ultrasonic guided waves (UGW) are explored. Numerical methods are employed to calculate the disperse curves. Optimal excitation signal and frequency are selected. Then UGW test installation is designed to test the prepared specimens, which include different artificial defects. Time-domain and frequency-domain analysis are employed in processing the received signals. In order to obtain more accurate correlation, energy attenuation parameter is proposed, and the linear curve was obtained. This study indicates UGW is an accurate tool in evaluating the condition of interfaces between concrete and rebar on the condition that frequency and modes are properly selected. Also promising is the UGW technology being used in non-destructive test of reinforced concrete.

On-line calibration is widely used no matter in science research or industrial production. In this paper, an on-line calibration approach for industrial displacement transducers has been developed. We used laser displacement sensor as standard and improved its accuracy to meet the requirements in different applications. In order to achieve the objective, laser sensor must be calibrated by automatic interference comparator. Laser sensor and automatic interference comparator measured the same displacement simultaneously and obtained the laser displacement sensor's calibration curve by analyzing the measurement data. Then the on-line calibration system including mechanical device and calibration platform was developed with the calibrated laser sensor. Numerous factors, such as verticality, colors, reflector material and vibration, contribute to errors in the compact laser displacement sensor measurements under field conditions. Analyzing the data obtained from the device's application of air spring system in automatic interference comparator, the presented paper outlines the error compensation methods and mathematical model. Experimental results show that laser sensor on-line calibration system can solve factory calibrating problems with accuracy requirement over 0.3%.

A 2D high-precision motion unit plays an important role in an automatic optical inspection (AOI) system, and its positioning accuracy and speed have an decisive effect on the inspection accuracy, efficiency and stability of the AOI system, but the present AOI systems cannot ensure the high speed and high-precision at the same time. To solve this problem, an automatic optical platform is developed, using the whole-closed-loop, fast, high-precision positioning technology to construct the mathematical models of its motion control unit, including models of current loop, velocity loop and position loop from inside to outside. Experiments were made on the motion platform using a laser interferometer. Experimental results show that the 2D motion platform can be used to reach a scanning speed of 20m/s and a positioning accuracy of 9im in a work area of 300mm * 3000mm with a load of 5kg.

This paper presents an overview of current research and development in the field of sensor placement methods in structural health monitoring. Due to the limitations of equipment facilities and cost, the number of sensors to be installed in a structure is relatively few. In many cases, three are always degrees of freedoms in a structure are not easily accessible, or eventually inaccessible. Moreover, actual structures are all continuous and have essentially infinite degrees of freedoms, therefore, it is impossible to instrument on all the degrees of freedoms. The methods of sensor placement are mainly dealing with accelerometers for global dynamic testing, or structural health monitoring. Three major concerns are discussed, the necessary number of sensors to be installed, where to deploy these sensors, and how to evaluate the effectiveness of these deployed sensors. Finally, a new loading dependent sensor placement method is proposed, and verified by the experimental data of a simple cantilever beam structure.

The relationship between the strains measured by a fiber Bragg grating sensor and the actual structural strains is deduced, then the average strain transfer rate computed by the formulation developed in this paper is compared with available experimental data. The critical adherence length of an optical fiber sensor is determined by a strain lag parameter, which contains both the effects of the geometry and the relative stiffness of the structural components. The analyses shows that the critical adherence length of a fiber sensing segment is the minimum length with which the fiber has to be tightly glued to a structure for adequate sensing. The strain transfer rate of an optical fiber sensor embedded in a multi-layered structure is developed in a similar way, and the factors that influence the efficiency of optical fiber sensor strain transferring are discussed. It is concluded that the strains, sensed by a fiber Bragg grating, have to be magnified by a factor (strain transfer rate) to equal exactly to the actual structural strains.

A new method for identifying dynamic parameters of a structure by measuring its first several frequencies and mode shapes is presented in this paper. Accurate estimates of the dynamic characteristics of a structure using limited measurement data are necessary for applications such as damage localization and finite element model verification. This method is developed from the unconstrained simplex method by introducing some pratical constraint conditions. The approach is not limited by either the minimum order of required eigenvalue and eigenvector pairs or the completeness of each order, for instance, even only a subset of frequencies or mode shapes are sufficient for a given order to yield acceptable identification results. Two numerical examples for identifying the stiffness distribution of a ten-story shear building under different health conditions are provided. The identified stiffness agrees with the physical model quite well. It is shown that this method has the advantages of good accuracy in calculation and great potentiality for most applications in similar structures or areas due to the fact that it is not only very robust but also not so strict with measurement methods and the amount of required information in field tests.

A method is presented for the selection of a set of sensor locations from a larger candidate sent for the purpose of structural loading identification. The method ranks the candidate sensor locations according to their effectiveness for identifying the given known loadings. Measurement locations that yield abnormal jumps in identification results or increase the condition number of the frequency response function are removed. The final sensor configuration tends to minimize the error of the loading identification results and the condition number of the frequency response function. The initial candidate set is selected based on the modal kinetic energy distribution that gives a measure of the dynamic contribution of each physical degree freedom to each of the target mode shapes of interest. In addition, excitation location is considered when selecting appropriate response measurement locations. This method was successfully applied to the optimal sensor location selection and loading identification of a uniform cantilever beam in experiment. It is shown that computer simulation is a good way to select the optimal sensor location for loading identification.

Combining with the characteristics of contactiess sensor and according to the Donaldson principle the paper puts forward the error separation technique of " two points and two setting method" . This technique can not only realize the separation for roundness error of intermediary from rotating error movement of spindle but also reduce the time of a measuring period and avoid the " drift" error which is brought by measuring system. Besides according to above error separation technique the authous of the paper develop a error separation system which can be used in ultra-high--precision contactless roundness measuring instrument. The maximal " drift" value of the system is less than 10 nm.

In order to reflect the shape of measured flat fully and reasonably authors of the paper apply the least square method to the processing of original data. First of all the paper expounds the establishment of mathematical model by means of simple examples. Then the difference of accuracy of two patterns is compared under the condition when the number of measured points are same and the patterns of mensuring pathes are different. Thereby the scheme of measuring path which can reach higher accurary is determined. Finally through calculating two groups of data and comparing the results with that of traditianal method the correctness rcasonableness and practicality of the method are verified.

This article expounds the principle and constitution of the visual and intelligent detecting system of geometrical parameters in the microscopic view. This system is used mainly on manufactory site. It fits in with the micro geometrical parameters detecting on all sorts of planes and curved surfaces such as the depth and width of furrow the surface degree of roughness etc. This system consists of mechanical part optical part photographic sensing unit and chip microprocessors. The function of mechanical part is pointing allocation adsorption adjusting the focal length and scanning on the plate of all sorts of planes and curved surfaces. The principle of optical part is parting with light. The flaw on the surface of the plate which is detected will be amplified and imaged on the focal plane of the objector in the reflect light path. CCD photographic sensing unit will change the flaw of the image to digital electrical signal. The function of microprocessor system is image recognition and diagnoses. This article expounds emphatically the theory and technology on the engagement of the optical peculiarity focal plane fuzzy recognition and compensation. On the basis of common image processing this detecting instrument uses the technology of background separation gate circuit trigger highspeed S/H circuit parallel AID change and the software of image processing makes use of not only the technology of space and frequency domain and the modularity constitution but

The paper analyses the torque error which is brought by zero drift and searches for the effective method to compensate the error. In accordence with the characteristics of wave form of signal two kinds of reasonable compensation method have been put forward. Experiments show that these methods possess ideal compensation effect. The zero drift of magneto-sensing torque sensor has been decreased to permissible extent.

This paper explored the restraint and recognition for offset components and the estimation of sphericity error. Two methods for processing offset components are put forward. The "meridian method" is applied to measure sphere comtour ; The "sequence elimination offset method" is applied to recognize and eliminates horizontal offset component. The restraint methods for other offset componenets are also discussed in this paper. Besides, according to the conception of least zone, the paper also puts forward "optimum method" to evaluate the position of sphere center and sphere error.

This paper introduces the working principle of an on - line thickness measuring system for large area ribboi. It presents a new design method for the thickness probe of the system. On this basis, it establishes the air -cushion grate thickness probe s mathematical model , analyses the relation between each parameter and the probe , and summarizes the probe design criteria. This measuring system can be used for various ribbons thickness on - line intelligent measurement.