| Gas turbine,a type of engine with high efficiency,high power density,and a long service life,has been widely applied in many fields,such as aviation,shipping,power generation and mechanical drives.In natural gas pipelines,the use of gas turbine-driven compressor sets can help reduce the load of local power grid significantly,thus making it the first choice for power equipment in underdeveloped areas with a lack of electric supply.However,harsh working environments tend to give rise to a range of problems in compressors,combustion chambers and turbines,such as fouling,wearing,increase in the tip clearance and carbon deposition on nozzle,which finally result in the gas-path performance degradation of each component.In the early stage of such degradation,the output power of gas turbine and its thermal efficiency will decrease in varying degrees.Once the further degradation leads to an unscheduled shutdown,it may pose significant safety risks to the operation of natural gas pipeline.To avoid unscheduled shutdowns caused by gas path failures,an operator needs to monitor the health condition of each gas turbine component timely and accurately.On this basis,a reasonable maintenance plan should be made.An effective way to realize the above goal is to research the gas-path performance diagnosis of gas turbines,alternatively aiming to promote the shift of maintenance strategy from scheduled maintenance to condition-based maintenance.In the actual operation,for the performance degradation of each component,the economy and safety factors should be comprehensively considered before the decision of downtime maintenance is made.However,the slight performance degradation will reduce the power output of gas turbine,potentially causing a further negative impact on the gas transmission capacity of the booster station.Therefore,it is also of important significance to conduct the performance self-healing research to enhance the gas turbine output when it plunges into the state of sub-health.The two research contents above intend to:(1)establish the models with different purposes for gas-path performance analysis of gas turbine;(2)solve the problem of inaccurate and unstable component characteristic map modification due to poor field data quality;(3)solve the problem of reduction of diagnosis precision when the number of gas path measurable parameters is less than that of the components' health factors;(4)mitigate the negative impact of slight degradation of component performance on the output power of the gas turbine.The specific research contents are as follows:(1)Three gas-path performance analysis models with different purposes were established for the gas turbine.Specifically,the thermodynamic properties of air,fuel and gas were calculated by programming based on NASA Glenn Coefficients for Calculating Thermodynamic Properties of Individual Species.The performance calculation module of each component with the same input and output data format was set up and adopted to build the entire gas path model of the GE LM2500+SAC gas turbine.Under the framework of Newton-Raphson algorithm and according to the different purposes of performance analysis,the gas-path performance calculation model,gas-path performance simulation model and gas-path performance diagnosis model were established by replacing the independent variables and governing equations in the iterative process.Their accuracies were verified afterwards.(2)A genetic-algorithm-based improved performance adaptation method was proposed for improving the inaccurate and unstable components' characteristic maps modification caused by bad field data quality.The improved method contains two measures: on the one hand,the similarity between the predicted performance curves and the performance regression curves of each component was used to evaluate the adaptation accuracy;on the other hand,the performance parameters at design point would be recalibrated in off-design performance adaptation.The result indicates that conducting performance simulation using the components' characteristic maps modified with the improved method,the proportion of performance parameters with prediction errors exceeding 0.5% was only 4.9%.Furthermore,adaptation results were not sensitive to the selection of error control point(ECP).The computational time for adaptation based on eight ECPs(200 population size,30 generations)was approximately 12 min,which satisfies the requirement of engineering application.(3)A degradation diagnosis method fusing the extreme learning machine(ELM)with the gas-path performance analysis models was proposed to solve the reduction of diagnosis precision when the number of gas-path measurable parameters is less than that of the components' health factors.In the fusion method,the simulation model was employed to build a symptom set under typical performance degradation patterns,and provide the gas-path measurable parameters of the gas turbine in a healthy state.ELM was used to determine the similarity between the sample to be recognized and all performance degradation patterns.The gas-path diagnosis model was used for quantifying the health factors to be determined under the corresponding degradation pattern.Only the diagnosis that meets the requirement of reasonableness can be accepted.Based on the recognition results of the pattern of the three-component performance degradation,the false recognition rates of BP Neural Network,ELM and Support Vector Machine were all about 4%,while that of the fusion method was only0.64%.Even if the training set was improperly selected or the hyper parameters was improperly set,the false recognition rate of the fusion method was still lower than 1.6%.Considering the better recognition performance,more accurate diagnosis results can be achieved using the fusion method.(4)A performance self-healing method based on the variable-geometry power turbine was proposed for mitigating the negative impact of slight degradation of component performance on the gas turbine output.Under the precondition of ensuring the gas-path measurable parameters to be within the safety limit,by improving the flow capacity of power turbine and fuel supply,a higher cycle pressure ratio and a greater working medium flow can be obtained,and the goal of power output recovery can be realized.The simulation results showed that the upper limit of power output recovery depends on the outlet temperature margin of high-pressure compressor(HPC).Therefore,when the atmospheric temperature rises or the flow capacity of high-pressure turbine(HPT)decreases,the margin will be reduced and the effect of performance self-healing will be weakened.Conversely,when the flow capacity of HPT increases or its isentropic efficiency decreases,the margin will rise and the effect of performance self-healing will be enhanced.Furthermore,the HPC keeps operating at high isentropic efficiency when its rotor speed exceeds the rated value,which will also help to improve the effect of performance self-healing.Under the combined pattern of performance degradation,the changes of each health factor will have an independent impact on the HPC outlet temperature margin.Thus,the combined pattern has a comprehensive influence on performance self-healing. |
PDF Full Text Request