| Long-distance pipeline transferring system is principal means of crude oil storage and transportation in Chinese; direct furnace is one of three main parts of it. Furnace is used to increase temperature of crude oil, and to reduce viscosity of oil, so as to assurance oil normal transferation.System has the properties of big inertia, long time delay and parameter variance, and its multi-input and multi-output parameters are strongly coupled, which make automatic control difficulties.There are many problems in stationarity, safety and efficiency of system, such as poor capability of fault diagnosis and fault-tolerant, low accuracy of temperature of crude oil outlet of furnace, instability, inefficiency, hidden safety troubles, difficulty of measure and long time-lag for oxygen content, etc. How to improve control level and work performance of furnace, not only yield good economic returns, but also have academic research value. Thus, thesis researches automatic control technique for furnace.The main contributions are listed as follows:(1) During furnace working, Crude oil evaporation in the pipeline is one of hidden safety trouble, which maybe be caused by severe fire deflecting. According to redundancy between parameters of furnace, redundancy select strategy was firstly proposed, which can effectively diagnose fire deflect and sensors fault. A fault diagnosis and tolerant system based on fuzzy observer was designed accordingly in the paper, which improves safety and fault-tolerant capacity of furnace, and stronger robustness is also obtained.(2) Crude oil temperature of furnace outlet is key parameter for furnace control system, in present system, Crude oil temperature of outlet adopted arithmetical mean of two side measured value of crude oil temperature ofoutlet, the result was not accurate. Trial showed that there were three factors which influence accurate measured value of crude oil temperature of outlet, which include sensor accuracy, crude oil drifting and fire deflecting. In view of the above, a novel double-fusion algorithm was designed, which can improve measure precision of crude oil temperature of outlet0.5℃.If one-sided temperature sensor is false, after judged by model of multi-sensors fault diagnosis and management, single-fusion algorithm should be in place of double-fusion algorithm. The result also agrees accuracy of crude oil temperature of outlet. The measure can reduce faulty operation and shutting down for furnace, and improve stationarity of furnace.(3) Oxygen content of funnel flue is key parameter to judge whether furnace is working on optimal state. Its long time-lag and short life of its detection device influence optimization operation of furnace. A novel Elman Neural Network model was proposed for measuring oxygen content of furnace. The model used a novel category method to design input parameter of Elman NN, which reduced numbers of input parameter of neural network, therefore meet the challenge of real-time. In the paper, studying andtraining neural network by selecting input value on t(k) and measuredvalue of oxygen content on t(k)+τ (τ is delay time of oxygen content offunnel flue), which change soft sensing of oxygen content of funnel flue to chamber. Trial results show that good dynamic regulation performance of system can be obtained, and fuel efficiency is improved greatly.(4) It is another hidden safety trouble for crude oil evaporation that crude oil flux of furnace entry declining abruptly. Based on integrating rarefaction wave method with real time transient modeling, a leak detection system for crude oil pipeline based on SCADA system was designed. If crude oil leak happened, the system can not only accurately detect leakage time, leakageposition and leakage rate on-line, but also shut down furnace in advance bynetwork control system to avoid oil evaporation. (5) Fuel atomization is important influencing factor for whether fuel combustingcompletely, a simplified control method was proposed, which implementautomatic control on pressure of atomizing air in response to f...
|
PDF Full Text Request