Joseph McGhee

Neural networks applied for the identification and fault diagnosis of process valves and actuators

Abstract This paper is concerned with the instrumentation and technology of fault detection and isolation (FDI) in process valves and actuators. A classification of faults in process valves and actuators is followed by a brief review of EDI techniques. Artificial neural networks (ANNs) are classified and introduced as an effective way of modelling valves and actuators, which are severely nonlinear components. Experimental results obtained from tests conducted on a double acting, twin piston rack-and-pinion actuator, are presented.

Sensor science : essentials for instrumentation and measurement technology

Abstract The need for an analysis of the constitution of SENSOR SCIENCE is justified in the paper. A systematic ordering in Sensor Science is then shown to be possible using Classical Taxonomy, also called classification science, which provides the general principles for organisation in science and philosophy. After describing the four orders or problems of taxonomy, their application, using temperature sensors as an example, is illustrated. In this way, it is shown that taxonomy provides the enabling key to defining the nature and scope of Sensor Science. Applying Classical Taxonomy to ordering in Sensor Science is also shown to depend upon the cardinal link between humans and machines. This provides a context for describing the information flow interfaces between humans, machines and their universal environment. It is asserted that Sensor Science, which is pivotal in Information Technology, is an identifiable discipline with its own unique core of knowledge. By giving a critical analysis of sensors and sensor systems this paper contributes to the organisation of the science of sensors. This is necessary if sensor science is to be recognised as a prime contributor to the advancement and development of all fields of scientific activity. As sensors extend the human faculties of sensing and perception by performing information handling operations, it is proposed that SYSTEMS ENGINEERING is fundamental to sensor science. Moreover, as sensors are the front end elements in machines for measurement, calculation, communication and control, Sensor Science is proposed as the core not only of INFORMATION TECHNOLOGY but more specifically of Measurement Science. The importance of other contributory disciplines such as Mathematical Modelling , Materials Science and Artificial Intelligence among others is also emphasised.

Modelling and MBS experimentation for temperature sensors

Proposals for a simplified dynamic model for a Resistance Thermometer Detector, or RTD, which is based upon a priori and a posteriori knowledge, are made. A new self-heating method of in situ sensor testing is presented. Experimentation using Multifrequency Binary Signals for both this new self-heating and the traditional external input immersion tests are described. Different transfer functions were obtained for these two testing methods. Mathematical relationships between them were determined from the laws of thermokinetics.

Functions and structures in measuring systems: A systems engineering context for instrumentation

Abstract The importance of Information Technology has focused attention upon that class of machines now commonly referred to as information machines. This technology has many implications. One of these is the manner in which the body of principles, upon which the technology is based, may be presented for educational purposes. An outline of the scope of teaching in measurement is given. Subsequently, it is proposed that the milieu of Systems Engineering is a pertinent, broad-based context for the teaching of instrumentation at all levels. An important justification for this view is the holistic attitude which the methodology of Systems Engineering affords. As man-machine systems are particularly important, a boundary perspective of these systems is considered. This view, which classifies inputs and outputs into one of three main types, indicates the place occupied by information within this systems environment. A reticulation of man-machine systems reveals that their structure consists of four main sub-systems. By this method, an incisive top-down view of the functions and structures of information systems in general and measurement systems in particular is achieved. In conclusion, energy bandwidth modelling is proposed as the link between modelling and information.

Symbolic codes for multifrequency binary testing of control systems

Abstract Symbolic codes allow a simple description of binary test signals for the identification of control systems. These symbolic descriptions of digital shift keyed modulation using compact binary codes are highlighted as the basis for designing new multifrequency binary sequence (MBS) test signals. Their energy may be concentrated in frequency arrays which measure specified spectral information. Phase shift keyed (PSK) MBS signals are designed with a zoom attribute to identify narrowband frequency information with a high spectral resolution. Frequency shift keyed (FSK) MBS use an extended range attribute to evaluate wideband frequency information. The application of these test signals is illustrated by two examples.

Development of a blood urea monitoring system for the closed loop control of dialysis

Optimal hemodialysis prescription through real-time blood urea (BU) monitoring and closed loop control of urea removal would be of significant clinical value. Progress toward a bedside BU analyzer and a control system is described here. An Amicon Minifilter inserted into the arterial bloodline provides a 1 ml/min stream of protein free ultrafiltrate for analysis. In vitro tests with bovine blood have shown excellent correlation between plasma (CP) and ultrafiltrate (CU) urea levels: CP = 0.961CU + 0.071, (n = 34, r = 0.998). In clinical hemodialysis studies, CU accurately represented the decay in CP. The BU analyzer uses a standard UV endpoint assay with a proportioning roller pump. The absorbance of the reacted mixture is read in a spectrophotometer after a 5 min incubation. For future control system design, the transfer function (TF) of the BU analyzer was measured using multifrequency binary testing. The data indicated that the analyzer may be modeled by a second order TF with a pure time delay. The same form of TF was also found to describe the Minifilter. Control of the removal rate of NaCl (substituted for urea) through automatic dialysate flow adjustment has been achieved with a simulated dialyzer-patient circuit (using a conductivity probe in place of the BU analyzer). A modified BU analyzer using computer controlled precision syringes for improved sample processing time and accuracy is also reported.

Narrowband multifrequency binary measurement using phase shift keyed symbols

Multifrequency binary testing (MBT), which uses multifrequency binary sequence (MBS) test signals, with high signal-to-noise power ratios for their dominant harmonics, to measure frequency response is considered. Information theory, in the form of descriptive languages and codes, digital modulation, and redundancy is used to design novel phase-shift-keyed (PSK) MBS symbols. A signal zooming operation, similar to optical zooming, is used to allow narrowband frequency response measurement with a high spectral resolution. >

A Digital Frequency Shift Keyed Technique for System Identification

Abstract Maximum entropy Multifrequency Binary Sequences, or MBS, which are used to interrogate control systems, provide for the software evaluation of frequency response estimates. These signals offer the advantages of excellent signal-to-noise power ratios for their dominant harmonics, zero offset and a wide variety of power spectral distribution. A previously designed Phase Shift Keyed MBS, or PSKMBS, signal concentrates the test signal energy in a narrow range of frequencies on either side of the suppressed fundamental squarewave carrier. PSKMBS use a zoom attribute to increase the spectral resolution for specific narrowband frequency information. This paper introduces Frequency Shift Keyed MBS, or FSKMBS, which allows an extended range attribute to evaluate wideband frequency information. A novel property of these signals is that the design technique allows a perfect division of a portion of the binary energy among dominant harmonics with specified frequency ratios.

Uncorrelated multisymbol signals for MIMO system identification

This paper presents a new approach to the system identification of a two-input two-output multivariable system. In a similar manner to that used for the single-input/single-output system, the multi-input multi-output (MIMO) system response may be obtained directly from measured input output frequency responses. An investigation of compact multifrequency data measurement signals, which have been extensively cataloged by the authors, has provided compact multifrequency ternary and quaternary input signals that have uncorrelated spectra. They are powerful multisymbol, multilevel computer generated measurement signals whose signal power is concentrated in either two or three dominant harmonics. As the signals, which are given in this paper, have six, seven, or eight symbols in their measurement codes, both the computation time for the frequency estimates and the experimental time are minimized. Two multifrequency quaternary signals with uncorrelated spectra are used to identify a simulated distillation column. It is shown that the cross coupling terms between the measurement channels may be removed by these digital measurement signals with the same number of symbols but different measurement codes.

Robotics and engineering modelling: Validation of single scatter model for backscatter density gauges

Gamma-backscatter gauges are used to obtain a nonintrusive measurement of the density of bulk materials from the surface. The backscatter process has been emulated by a single scatter model and a multi-scatter version using Monte-Carlo statistics. Although the single scatter model appears to give a satisfactory explanation of the backscatter process, a full-scale investigation of the validity of this model has not previously been possible. A mathematical model, which accurately represents the detector in its working environment, has been developed. The model makes a comparison possible through the accurate calculations of both maxima and normalised range or bandwidth of the detector characteristics. This allows a comparison of design parameters by different routes and forms the basis of a validation of the single scatter model.