Application Of Parameters Identification And Optimization Algoirthm On Porous Media Thermal Physical Property Measurement

Abstract：With the further development of computer science and intelligentoptimization algorithms, more and more engineering parameter identificationproblem can be solved with computers. Intelligent optimization algorithm isused for a lot of engineer problems more and more widely, and become to bean important tool to these problems. In the porous medium thermal parametersidentification process, it is difficult to identify the parameters because thenumber of parameters to be identified is many and the optimized solution tomulti-dimensional space is difficult, approximate formula and line sourcemodel are used for identification. But the recognition accuracy is not high andthe result of identification needs to be modified, the efficiency ofidentification is low. In this paper, in order to solve these problems above,engineering physic model and its solution methods are deeply thought,intelligent optimization algorithm and its parameters setting problem areresearched, the error from the project is thought. Finally the thermal physicalparameters are identified with the practical porous medium data. Overall, thispaper presents the four major innovative results:(1) Selection of engineering model and solution method, use a thermalprobe model which could reflect the influence of the test system to theidentification effectively in practical project problems for the first time at thisfield, avoiding modification problem at later period since using approximateformula and line source model. Meanwhile inherent relation of parameters isconsidered, propose a solution method to the model.(2) Arithmetic research to the identification of t hermal propertiesparameters, use some intelligent optimization algorithms to solve the problem. First improve the range research algorithm, propose a new algorithm namedrange particle swarm search improved algorithm solution method. By themechanism of generating particle randomly and mechanism of judging newparticle, this algorithm discretize and reduce search range constantly, analyseif the interval should integrate the best particle in history, and finally identifythe multi-dimensional particle successfully. Use the parallel computing withthe fitness function value of every particle, raising the speed of compute. Set amechanism of feedback of initial interval to adjust the initial interval,improving the robustness of algorithm. Second, research the genetic algorithmand simulated annealing algorithm, propose genetic algorithm-simulatedannealing algorithm mixed solution method by their characteristics, and givesome improvement to the solution method. Finally, combine the analog dataand practical measured data, prove and compare the veracity, rapidity, andstability of these solution methods.(3)Parameter setting of parameters identity algorithms, the parametersneeded to set are often more than one, the current parameter setting is basedon experience or experiments. In this paper, on research of parameter setting,make it into an optimization problem. Design a training method and useintelligent optimization algorithm to search the best parameters. Finally,combining the experience and the large amount of test data, determine theparameters of the algorithm, enhance the efficiency of algorithm to thethermal physical parameters identification.(4) Error analysis, by discussing the error factor in actual project,analyse the influence of error factor to parameters identification with theanalog data, meanwhile give the method of reducing the influence, make theaccuracy maintain in the order of engineering calculation, provide somereferences to the identification of parameters.