Process identification with multiple neural network models

This paper describes the identification of chemical processes with multiple neural network models. This concept is called a 'multimodel' approach. The multimodel approach presents a flexible framework which allows the integration of other model paradigms. Three different methods to construct neural network multimodels are presented. First, a priori knowledge is used to decompose the input domain into operating regimes. Each regime is modelled by a different neural network. In the second multimodel approach, unsupervised learning in form of clustering and SOMs is used to split the input domain. The third approach uses a gating neural network to divide the input space. In contrast to the first two approaches, the non-linear gated network approach allows the multim odel to simultaneously learn a suitable decomposition and the mapping in each regime. All three approaches are evaluated for a fed-batch fermentation.     

Particle-induced Phagocytic Cell Responses are Material Dependent: Foreign Body Giant Cells Vs. Osteoclasts from a Chick Chorioallantoic Membrane Particle-implantation Model

There is increasing concern about particles generated from wear-prone implants that are placed in body tissues, including artificial hip, knee, and jaw joints. Although phagocytes and foreign body giant cells are associated with inhaled or embedded particulate debris, some particles also induce bone digestion by eliciting the differentiation and proliferation of highly specialized osteoclastic cells. This report describes the differential phagocytic cellular responses to four implant-related types of ground, model wear particles in a live-egg cell-response model, as implants to the chick chorioallantoic membrane (CAM): polymethylmethacrylate (PMMA), a main constituent of some temporomandibular joint (TMJ) implants and orthopedic cements used to retain artificial hips and knees; Proplast-HA, an implantable composite of polytetrafluoroethylene (PTFE) and degradable mineral (hydroxyapatite) that has been associated with bone erosion around failed TMJ implants; talc, a nondegradable mineral sometimes found in tissues as a contaminant from talc-coated surgical gloves; and authentic bone, known to induce the formation of osteoclastic cells. Light and electron microscopy of CAM tissues harvested, sectioned and stained with special reagents for the enzymes tartrate-resistant acid phosphatase (TRAP) and tartrate-resistant adenosine triphosphatase (TrATPase), and for the osteoclast-specific antigen 121F, showed that only authentic bone and the degradable HA-rich particles induced osteoclast formation. From these results, and supporting data with polypropylene particles, it is concluded that nonbiodegradable polymer particles, alone, do not induce bone dissolution. Inert polymers do induce foreign body giant cells without the external mineral digestion qualities unique to osteoclasts, however. The chick embryo model system allows quick and affordable examination of material-dependent differences in phagocytic cellular responses to implant wear debris and particles from various occupational environments.     

Harnstoffhydrolyse für die selektive katalytische Reduktion von NOx: Vergleich der Flüssig‐ und Gasphasenzersetzung

In order to reduce the NO_x concentration in car exhausts usually the selective catalytic reduction with ammonia is used. However, to avoid the transport of ammonia in vehicles urea is applied as NH₃ precursor. Controlled urea decomposition before the injection into the exhaust gas system itself may be accomplished by the use of a separate reactor. Urea decomposition to ammonia in the liquid phase under pressure in a heated reactor was compared to its decomposition in the gas phase. In the liquid phase, higher conversion rates relative to the reactor volume were realized than in the gas phase. Catalysts which showed high activity in the gas phase influenced the decomposition in the liquid phase only slightly.     

Evaluation of Performance of Diesel Particulate Filters Through Integrated Multi-Scale Computer Calculations

Wall-flow channel models and soot deposition models based on micro scale considerations are integrated into global 3D diesel particulate filter simulations. In addition, transient and steady-state simulations are combined to understand at the same time short- and long-time behaviour of the diesel particulate filter (DPF). The functionality of the simulation tool is achieved and correlations with measured data encourage the use of the model as a tool to predict DPF behaviour.