S. J. Wilcox

Neural network and on-off control of bicarbonate alkalinity in a fluidised-bed anaerobic digester

Abstract A laboratory-scale fluidised-bed anaerobic digester with a sintered glass carrier, Siran®, was operated for 8 months on a simulated bakers yeast wastewater (12,000 mg soluble COD l−1) at a loading rate of 27 kg COD m−3 d−1, giving 75% removal of soluble COD. Percentage CO2, H2 concentration, gas flow rate and pH were measured continuously. An on-line bicarbonate alkalinity (BA) monitor was used in experiments comparing two control strategies, adjusting digester buffering by addition of NaHCO3 solution during organic overloads. The first, an on-off controller with a set point at the steady-state level (2700 mg CaCO3 l−1), maintained BA concentration but resulted in levels above the upper set point. Thus, to avoid consuming excess NaHCO3 the rate of delivery and solution strength must be carefully adjusted. The second was a controller developed from a neural network trained on BA data from an anaerobic filter operating on ice-cream processing wastewater (alkalinity around 1400 mg CaCO−1). Without re-training, despite the different steady-state BA levels and reactor type, the neural network based controller was capable of maintaining stable BA levels during overload without overshoot. Control of BA during overloads did not prevent changes in gaseous CO2 and H2 concentrations and gas flow rate.

A neural network, based on bicarbonate monitoring, to control anaerobic digestion

Abstract The use of a neural network simulation for monitoring and controlling an anaerobic digestion process by utilising data from an on-line bicarbonate alkalinity (BA) sensor is presented. The results collected from a series of induced disturbances, in the form of an increase in influent concentration, suggest that the instrument is very sensitive to BA changes and that the neural network is capable of rapid recognition of these disturbances. A scheme for the use of these changes is suggested that would allow on-line control of the digestion process in cases where the wastewater has a low pH buffering capacity.

The Link Between Intravenous Multiple Pump Flow Errors and Infusion System Mechanical Compliance

IV drug delivery in intensive care often takes the form of simultaneous multiple infusions from separate infusion devices via either shared or individual fluid pathways. Because of the potency of the drugs administered and the acuity of the patients, accurate drug delivery is required. Instances of unexpected and unacceptable accuracy degradation have been reported when the equilibrium of the infusion system is disturbed. We describe a mathematical model of a simple infusion system used to investigate and verify results reported from a simple experimental multiple pump fault scenario. The results suggest that flow degradation is attributable to small changes in infusion system extracorporeal volume, referred to as “compliance.” The model may, by expansion, be used to determine the nature of fluid flow within other multiple pump systems, be applied to the design of future IV systems, and explain the need for small-volume infusion systems with small mechanical compliance.

A comparison of the ability of black box and neural network models of ARX structure to represent a fluidized bed anaerobic digestion process

The performance of three black box models which were parameterized and validated using data collected from a laboratory scale fluidized bed anaerobic digester, were compared. The models investigated were all ARX (auto regressive with exogenous input) models, the first being a linear single input single output (SISO) model, the second a linear multi-input multi-output (MIMO) model and the third a nonlinear neural network based model. The performance of the models were compared using correlation analysis of the residuals (one-step-ahead prediction errors) and it was found that the SISO model was the least able to predict the changes in the reactor parameters (bicarbonate alkalinity, gas production rate and % carbon dioxide). The MIMO and neural models both performed reasonably well. Though the neural model was shown to be superior overall to the MIMO model, the simplicity of the latter should be a consideration in choosing between them. A simulation with an horizon approaching 48 h was performed using this model and showed that although the absolute values differed significantly, there were encouraging similarities between the dynamic behavior of the model and that of the fluidized bed reactor.

Use of artificial intelligence techniques for optimisation of co-combustion of coal with biomass

AbstractThe optimisation of burner operation in conventional pulverised-coal-fired boilers for co-combustion applications represents a significant challenge This paper describes a strategic framework in which Artificial Intelligence (AI) techniques can be applied to solve such an optimisation problem. The effectiveness of the proposed system is demonstrated by a case study that simulates the co-combustion of coal with sewage sludge in a 500-kW pilot-scale combustion rig equipped with a swirl stabilised low-NOx burner. A series of Computational Fluid Dynamics (CFD) simulations were performed to generate data for different operating conditions, which were then used to train several Artificial Neural Networks (ANNs) to predict the co-combustion performance Once trained, the ANNs were able to make estimations of unseen situations in a fraction of the time taken by the CFD simulation. Consequently, the networks were capable of representing the underlying physics of the CFD models and could be executed efficien...

On-Line Monitoring of Anaerobic-Aerobic Biotreatment of a Simulated Textile Effluent for Selection of Control Parameters

Parameters monitored on-line in a laboratory biological wastewater plant treating simulated textile effluent were examined for use in a control strategy. The plant, an anaerobic (UASB) reactor with an aerobic stage and overall HRT 1.8 days, was operated at steady-state and transient conditions with varying ratios of azo dye and starch. Gas flow rate, %CO2, biogas hydrogen levels and effluent bicarbonate alkalinity were assessed on-line in the anaerobic stage, together with on-line measurements of effluent total oxygen demand (TOD), colour, pH, and dissolved oxygen. Their ability to indicate process instability and efficiency, the reliability/maintenance of the instrumentation and delay in response were evaluated. On-line TOD measurement failed due to the wastewaters mineral content; biogas hydrogen content varied unpredictably. It is recommended that the other measurements be used together in a control system for textile effluent biotreatment, providing complementary information and redundancy.

Pelvis feature extraction and classification of Cardiff body match rig base measurements for input into a knowledge-based system.

The purpose of this paper is to determine whether it is possible to use an automated measurement tool to clinically classify clients who are wheelchair users with severe musculoskeletal deformities, replacing the current process which relies upon clinical engineers with advanced knowledge and skills. Clients’ body shapes were captured using the Cardiff Body Match (CBM) Rig developed by the Rehabilitation Engineering Unit (REU) at Rookwood Hospital in Cardiff. A bespoke feature extraction algorithm was developed that estimates the position of external landmarks on clients’ pelvises so that useful measurements can be obtained. The outputs of the feature extraction algorithms were compared to CBM measurements where the positions of the client’s pelvis landmarks were known. The results show that using the extracted features facilitated classification. Qualitative analysis showed that the estimated positions of the landmark points were close enough to their actual positions to be useful to clinicians undertaking clinical assessments.

Damage Characterisation of Carbon Fibre Reinforced Composite Plate Using Acoustic Emission

Acoustic Emission (AE) is a sensitive technique which can be used to characterise damage in high strength composite plate. This paper describes an extension to an earlier piece of research work carried out by the ERC which resulted in the successful development of a novel source location methodology for the said material. The previous work concentrated on the source location in plate-like composite structures using acoustic emission. The work presented in this paper focuses on establishing the correlation between the different damage types suffered in the material namely de-lamination, matrix cracking, fibre rupture and stringer to skin debonding with key signal features of the AE activities. Controlled bending tests were initially carried out on laterally grooved slender composite specimens to progressively propagate damage in the weakened region of these specimens. The composite laminate plate itself is made from 16 plies of carbon fibre twill weaved in an epoxy matrix with bidirectional fibre alignments in the 0° and 90° directions with 60/40 fibre-matrix volume composition. These prepared samples were fully instrumented with broad band (100 kHz to 1MHz) Physical Acoustic AE sensors linked to the necessary signal conditioning hardware. The AE events were recorded using a high speed DAQ card accessed by customised software written in LabVIEWTM. Gathered raw data were analysed off-line for key signal features including energy and frequency contents and subsequently correlated to actual damage types. It can be concluded from the empirical evidence that feature vectors are distinct to the type of damage. Results gathered from additional test on the progressive skin-stringer debonding of the same material to failure confirmed the uniqueness of the AE feature trends. An integrated system which is capable of both in-situ location of compromised sites and the diagnostic of flaw types in composite plate can potentially find engineering applications including the structural health monitoring of composite aircraft parts.

Development of an intelligent flame monitoring system for steel reheating burners

This article describes the development of a system to indirectly monitor the combustion characteristics of individual burners based on measurement and analysis of the signals detected from photodiodes detecting flame radiation signals. A series of experiments were conducted on a 500 kW pilot-scale furnace and on two 4 MW industrial burners located in two steel reheating furnaces. The flame radiation signals were monitored using a lens that transmitted the flame radiation to ultraviolet, visible and infrared photodiodes through a trifurcated optical fibre. The experiments covered a wide range of burner operating conditions including; variations in the burner load and excess air levels and simulations of burner imbalance. The relationships between the dynamic flame radiation signals and the burner operating parameters and conditions were made off-line using neural network models. The present work indicates that the measurement of flame radiation characteristics, coupled with neural networks, provides a promising means of monitoring and adjusting burner performance.

Development of fuzzy based methodology to commission co-combustion of unprepared biomass on chain grate stoker fired boilers

AbstractThis paper describes the development of an intelligent commissioning system to enable operators to maximise the utilisation of unprepared biomass by combusting the biomass with the minimum amount of support fuel to achieve a desired boiler output and thermal efficiency on chain grate stoker fired boilers. Tests were conducted on a 0·8 MWth chain grate stoker fired hot water boiler to investigate the combustion of different types of biomass blended with a support fuel over a wide range of boiler operating conditions and biomass moisture contents. The commissioning system was developed using fuzzy logic and expert system type rules developed while gathering the experimental data. The system was validated on untested blends of unprepared biomass with two support fuels where it was shown that it is possible to efficiently burn unprepared, high moisture content biomass with a support fuel on a chain grate stoker. This system could enable operators of chain grate stoker fired boilers to maximise the use...