MRC Laboratory of Molecular Biology
CUED - Cambridge University Engineering DepartmentNot a member yet
45551 research outputs found
Sort by
Optimization by moving ridge functions: derivative-free optimization for computationally intensive functions
No full textA novel derivative-free algorithm, called optimization by moving ridge functions (OMoRF), for unconstrained and bound-constrained optimization is presented. This algorithm couples trust region methodologies with output-based dimension reduction to accelerate convergence of model-based optimization strategies. The dimension-reducing subspace is updated as the trust region moves through the function domain, allowing OMoRF to be applied to functions with no known global low-dimensional structure. Furthermore, its low computational requirement allows it to make rapid progress when optimizing high-dimensional functions. Its performance is examined on a set of test problems of moderate to high dimension and a high-dimensional design optimization problem. The results show that OMoRF compares favourably with other common derivative-free optimization methods, even for functions in which no underlying global low-dimensional structure is known
IoT and Fog Computing based Predictive Maintenance Model for Effective Asset Management in Industry 4.0 using Machine Learning
No full textThe assets in Industry 4.0 are categorised into physical, virtual and human. The innovation and popularisation of ubiquitous computing enhance the usage of smart devices: RFID tags, QR codes, LoRa tags, etc. for assets identification and tracking. The generated data from Industrial Internet of Things (IIoT) eases information visibility and process automation in Industry 4.0. Virtual assets include the data produced from IIoT. One of the applications of the industrial big data is to predict the failure of manufacturing equipment. Predictive maintenance enables the business owner to decide such as repairing or replacing the component before an actual failure which affects the whole production line. Therefore, Industry 4.0 requires an effective asset management to optimise the tasks distributions and predictive maintenance model. This paper presents the Genetic Algorithm (GA) based resource management integrating with machine learning for predictive maintenance in fog computing. The time, cost and energy performance of GA along with MinMin, MaxMin, FCFS, RoundRobin are simulated in the FogWorkflowsim. The predictive maintenance model is built in two-class logistic regression using real-time datasets. The results demonstrate that the proposed technique outperforms MinMin, MaxMin, FCFS, RoundRobin in execution time, cost and energy usage. The execution time is 0.48% faster, 5.43% lower cost and energy usage is 28.10% lower in comparison with second-best results. The training and testing accuracy of the prediction model is 95.1% and 94.5%, respectively
Detection in molecular communications with ligand receptors under molecular Interference
No full textMolecular Communications (MC) is a bio-inspired communication technique that uses molecules to transfer information among bio-nano devices. In this paper, we focus on the detection problem for biological MC receivers employing ligand receptors to infer the transmitted messages encoded into the concentration of molecules, i.e., ligands. In practice, receptors are not ideally selective against target ligands, and in physiological environments, they can interact with multiple types of ligands at different reaction rates depending on their binding affinity. This molecular cross-talk can cause a substantial interference on MC. Here we consider a particular scenario, where there is non-negligible concentration of interferer molecules in the channel, which have similar receptor-binding characteristics with the information molecules, and the receiver employs single type of receptors. We investigate the performance of four different detection methods, which make use of different statistics of the ligand-receptor binding reactions: instantaneous number of bound receptors, unbound time durations of receptors, bound time durations of receptors, and combination of unbound and bound time durations of receptors within a sampling time interval. The performances of the introduced detection methods are evaluated in terms of bit error probability for varying strength of molecular interference, similarity between information and interferer molecules, number of receptors, and received concentration difference between bit-0 and bit-1 transmissions. We propose synthetic receptor designs that can convert the required receptor statistics to the concentration of intracellular molecules, and chemical reaction networks that can chemically perform the computations required for detection
Condition and criticality-based predictive maintenance prioritisation for networks of bridges
No full textNumerous bridges are exposed to increasingly frequent and intense extreme events due to climate change, while serving more traffic than originally designed due to population growth. Thus, predictive maintenance of bridges is of paramount importance for securing structural safety and bridge network reliability. However, the application of predictive maintenance for networks of bridges, considering condition and criticality of bridges within the network, has not seen much attention in practice and literature. Presented herein is a maintenance prioritisation method for networks of bridges, based on: a deterioration model for individual components considering uncertainty; life-cycle cost analysis; grouping of components maintenance to reduce traffic interruption and setup cost; and criticality evaluation of bridges using a specifically tailored version of closeness vitality and traffic simulation. This method has been applied to a network of 21 bridges in Portugal, composed of several heterogeneous elements. It showed a substantial decrease in maintenance cost, compared to the Structures Investment Toolkit, and significant differences between criticality of bridges within a network. The proposed methodology can be applied to any networks of bridges or serve as the basis for updated maintenance decision support systems for infrastructure asset networks
Pilot study of a non-invasive real-time optical backscatter probe in liver transplantation.
No full textTransplantation of severely steatotic donor livers is associated with early allograft dysfunction and poorer graft survival. Histology remains the gold standard diagnostic of donor steatosis despite the lack of consensus definition and its subjective nature. In this prospective observational study of liver transplant patients, we demonstrate the feasibility of using a handheld optical backscatter probe to assess the degree of hepatic steatosis and correlate the backscatter readings with clinical outcomes. The probe is placed on the surface of the liver and emits red and near infrared light from the tip of the device and measures the amount of backscatter of light from liver tissue via two photodiodes. Measurement of optical backscatter (Mantel Cox p<0.0001) and histopathological scoring of macrovesicular steatosis (Mantel Cox p=0.046), were predictive of 5-year graft survival. Recipients with early allograft dysfunction defined according to both Olthoff (p=0.0067) and MEAF score (p=0.0097) had significantly higher backscatter levels from the donor organ. Backscatter was predictive of graft loss (AUC 0.75, p=0.0045). This study demonstrates the feasibility of real-time measurement of optical backscatter in donor livers. Early results indicate readings correlate with steatosis and may give insight to graft outcomes such as early allograft dysfunction and graft loss
A Framework for Using Data as an Engineering Tool for Sustainable Cyber-Physical Systems
No full textSmart infrastructure has the potential to revolutionise how infrastructure is delivered, managed and automatically controlled. Data and digital twins offer an opportunity to enable this revolution and secure sustainable future smart infrastructure. In this article, we discuss data as an engineering tool and propose to use data throughout the asset's whole life cycle from identifying the need, planning and designing to construction, operation, integration and maintenance. This requires systems thinking where focus is not limited to the problems but rather constructs a systemic perspective to understand the interrelationships between components and systems. Future infrastructure is connected, intelligent and data-driven. To enable more sustainable decision-making, we should not only consider how to integrate different infrastructure elements but also use data to monitor, learn from and inform decisions. To this end, we present a case study where several assets, such as bridges, railways and transport systems are integrated, and data are curated for the purpose of aiding climate-conscious, sustainable decision-making. An example systems architecture for integration of different digital twins is explained and benefits of this data-driven, systemic perspective are discussed
The effect of fine droplets on laminar propagation speed of a strained acetone-methane flame: Experiment and simulations
No full textIn this study, we investigate the effect of the presence of fuel droplets, their size and concentration, on stretched laminar flame speeds. We consider premixed strained methane/air mixtures, with the addition of small acetone droplets, and compare the flame velocity field behaviour to that of the fully vaporized mixture. An impinging stagnation flame configuration is used, to which a narrowly distributed polydisperse mist of acetone droplets is added. Total acetone molar concentrations between 9% and 20% per mole of methane are used, corresponding to 18.6% and 41.4% of the total fuel energy. The Sauter Mean Diameter (SMD) of acetone droplets is varied from 1.0 to 4.7 μm by carefully tuning the air flow rate passing through an atomizer. The droplet size distribution is characterized by a Phase Doppler Anamometry (PDA) system at the outlet of the burner. The flame propagation speed is measured using Particle Image Velocimetry (PIV) for overall equivalence ratios ranging from 0.8 to 1.4 at various strain rates, and the result is compared with a reference case in which acetone was fully vaporized. Unlike the fully vaporized flame, a two-stage reaction flame structure is observed for all droplet cases: a blue premixed flame front followed by a reddish luminous zone. Comparison of the results between gas-only and droplet-laden cases shows that the mean reference burning velocity of the mixture is significantly enhanced when droplets are present under rich cases, whereas the opposite is true for stoichiometric and lean cases. The mean droplet size also changes the relationship between flame speed and strain rate, especially for rich cases. The result suggests that with typical conditions found in laminar strained flames, even for the finest droplets that may have been vaporized before reaching the flame front, the resulting inhomogeneities may lead the flame to behaves differently from the well-premixed gaseous counterpart. Simulations at similar conditions are performed using a two-phase counterflow flame model to compare with experimental data. Model results of reference velocities do not compare well with observations, and the possible reasons for this behaviour are discussed, including the difficulties in determining the pre-vaporization process and thus the boundary conditions, as well as the fidelity of the current point-source based 1D model
Reducing Passive Drug Diffusion from Electrophoretic Drug Delivery Devices through Co-Ion Engineering
No full textImplantable electrophoretic drug delivery devices have shown promise for applications ranging from treating pathologies such as epilepsy and cancer to regulating plant physiology. Upon applying a voltage, the devices electrophoretically transport charged drug molecules across an ion-conducting membrane out to the local implanted area. This solvent-flow-free “dry” delivery enables controlled drug release with minimal pressure increase at the outlet. However, a major challenge these devices face is limiting drug leakage in their idle state. Here, a method of reducing passive drug leakage through the choice of the drug co-ion is presented. By switching acetylcholine's associated co-ion from chloride to carboxylate co-ions as well as sulfopropyl acrylate-based polyanions, steady-state drug leakage rate is reduced up to sevenfold with minimal effect on the active drug delivery rate. Numerical simulations further illustrate the potential of this method and offer guidance for new material systems to suppress passive drug leakage in electrophoretic drug delivery devices
Model reduction of dominant feedback systems
No full textThe paper presents a model reduction framework geared towards the analysis and design of systems that switch and oscillate. While such phenomena are ubiquitous in nature and engineering, the problem of approximating the global behavior of systems away from equilibrium is largely open. We address this problem in the framework of dominance theory. Classical balanced truncation is applied to systems that can be decomposed as the feedback interconnection of a large-scale linear system and a comparably low-dimensional nonlinear system. The method is illustrated on a classical feedback model of a biological oscillator