摘要:
Die Erfindung bezieht sich auf ein Verfahren zur Steuerung von Prozeßvorgängen einer technischen Anlage. Um eine gleichzeitige und zusammenhängende Bewertung relevanter Prozeßgrößen der Anlage zu ermöglichen, wird vorgeschlagen, die relevanten Prozeßgrößen durch eine neuronale Analyse auf der Grundlage selbstorganisierender neuronaler Karten in Beziehung zueinander auszuwerten, indem eine topologieerhaltende, nichtlineare Projektion von Daten der relevanten Prozeßgrößen auf eine mehrdimensionale neuronale Karte realisiert wird.
摘要:
Die Erfindung bezieht sich auf ein Prozeßdiagnosesystem und ein Verfahren zur Diagnose von Vorgängen und Zuständen eines technischen Prozesses, insbesondere zur Diagnose eines Kraftwerksprozesses. Es wird ein dem Prozeß entsprechend strukturiertes Multiagentensystem (MAS) mit mehreren autonomen Diagnoseagenten (MA, A1, A2, A) vorgeschlagen. Die Agenten (MA, A1, A2, A) enthalten jeweils ein Neuronales Netz (NN), mit dessen Hilfe ein Referenzverhalten zu überwachender Prozeßkomponenten erlernbar ist und auch eine automatisierte Anpassung an ein neues Referenzverhalten ermöglicht ist.
摘要:
There is disclosed a combined heat and power system for a building comprising a prime mover drivingly connected to an electrical generator adapted to provide building electrical power and a heat recovery arrangement connected to the prime mover recovering heat therefrom, and adapted to provide building heating characterised by comprising a control arrangement adapted automatically to control the prime mover in response to heating demand.
摘要:
In a trouble analysis apparatus, a cause of an accident which is empirically and statistically obtained, and a plurality of effects corresponding to the cause are entered in advance at a keyboard (9) and stored as a data base in a magnetic disk unit (5). In this case, a sum of basic probabilities of the plurality of effects for the cause is set to be 1. A preprocessor (3) calculates an upper (or lower) probability of a subset of the result by using the Dempster & Shafer probability theory. The pre-processor (3) uses the upper or lower probability to create Fuzzy membership functions, and also creates a hypothetical proposition for corresponding the Fuzzy membership function and the cause. An inference engine (7) matches accident phenomena with the effects when an accident occurs and infers the cause by a modified Fuzzy inference method using the hypothetical proposition.
摘要:
Complex industrial processes are involved for example in the operation of a nuclear power station, in particular in emergency situations. A digital computer which implements by means of its program so-called automatic agents which interact with sensors or regulators is a prerequisite. Process information is obtained or the process is influenced by the activities of the automatic agents (programmable controllers) and operators, progress in carrying out activities being marked by actions. In detail: a) information supplied by the programmable controllers and operators is collected in a protocol of the computer (both the history of plant behaviour and that of actions are acquired in the protocol); b) a number of specifications of activities contain process descriptions which describe which actions in which types of situations (situation types) can be triggered when carrying out the activity concerned; c) a comparator unit compares the situation types with the real situation indicated by the protocol and ascertains whether the given situation corresponds to a situation type from which action possibilities are derived; d) each specification of an activity contains a number of references to information for carrying out the activities (implementation information), aimed at the programmable controllers or operators. A process control system for implementing such a method is disclosed.
摘要:
A plant with a gas turbine such as a combined cycle power plant of a gas turbine cycle and a steam turbine cycle is provided with a nitrogen oxide removal device for removing NO x by injecting ammonia to an exhaust gas of the gas turbine. The device reduces NO x concentration to a certain value or less before the exhaust gas is released to the air. An ammonia flow amount is rapidly controlled such that a mole ratio of ammonia to NO x coincides with a set mole ratio value. The mole ratio is calculated from a predicted NO x concentration at an inlet of the nitrogen oxide removal device predicted by calculating operation conditions of the gas turbine, an ammonia flow amount value, and an exhaust gas flow amount. The set mole ratio value is calculated from a deviation of a measured NO x concentration value at an outlet of the nitrogen oxide removal device and a set NO x concentration value, and an amount of water injected to a combustor. Finally, a controlled system is stabilized in a state where a measured NO x concentration value coincides with the set NO x concentration value at a high speed.