Xinmiao Wu

Application of improved PSO to optimization of gravity dam and sluice gate

In hydraulic engineering fields, optimization is an important means to save investment and shorten construction time, however, traditional optimization methods based on gradient often confront with such problems as non-convergence or convergence to local optimum when it is applied in large-scale complicated hydraulic engineering problems. It is necessary to develop a new optimization method that possesses more ability to seek a global optimum. In this paper, an improved PSO (particle swarm optimization) method is used to optimize body of gravity dam and select steel members for the tumble gate. Since the PSO method uses forward calculation process, and the process is performed by ANSYS calculation (calculation program compiled by APDL), the interface problem between exterior PSO program of C language and ANSYS should be solved. Two application examples are presented to demonstrate the applicability of the proposed method.

The optimization of mixture ratio of model sand based on Simplex Particle Swarm Optimization Algorithm

Model sand composed of several kinds of sand according to a certain proportion to meet required sand gradation is often needed in model test. In this paper, an optimization model of mixture ratio is built and mixture ratio of model sand is optimized by simplex particle swarm optimization algorithm (SMPSO). The powerful global search ability of particle swarm optimization algorithm (PSO) and the well local search ability of simplex method (SM) are combined in SMPSO. Example is presented to demonstrate that SMPSO has a better ability in searching the globally optimal solution than PSO. The optimization model and method of sand gradation introduced in this paper is not only an improvement on traditional experience method, but also a new method for other optimization problems, such as concrete aggregation gradation optimization.

Inverse Analysis of Frictional Coefficients of Pipelines Based on PSO

Frictional coefficient is an important hydraulic parameter for pipe network design and scheduling calculation. The inverse analysis to frictional coefficient has much meaning to pipe network optimization and reconstruction. Combining the particle swarm optimization with node water head method, a new inverse analysis method to frictional coefficient is put forward. The example analysis result shows that the method is rapid and easy to find the globally optimal solution.

Dam's Safety Monitoring Statistical Model Optimization Basing on The GA and AIC

It is difficult to select influence factors when the dams monitoring model is built, so the Akaike information criterion (AIC) used in the field of information statistics is introduced. Both the fitting to modeling data and prediction precision to other data are considered in the AIC formula. The optimization method basing on GA and AIC is introduced. The method is applied to practical engineering, and the comparison with multiple regression, stepwise regression and neural network model shows the monitoring model optimized by the method can reach higher fitting and prediction precision by lesser factors and data

Research on the genetic regression model of Earth-rock dam safety monitoring and its application

In this paper, the factors which influence the piezometric level of Earth-rock dams are combined randomly by using a genetic algorithm to do selective preference. A new genetic regression model is established based on considering the comprehensive influence of fitting and prediction accuracy in fitness function. Through analyzing the seepage monitoring data of Dong Wu Shi Reservoir, it is shown that the prediction accuracy of this method is better than that of the general statistical regression.

The inverse calculation of roughness coefficient in village and county circular pipe network based on Stochastic Particle Swarm Optimization

As timepsilas going on, county pipespsila aging problem is even serious, and water providing capability is coming down. Stochastic particle swarm optimization is an improvement of the standard PSO, compared with which, it can be guaranteed to converge to the global optimization solution with probability one in theory, and speed up the convergence. By using this method in the county pipe network inspection system, its running conditions will be got generally and betimes, accordingly, the control of water source is regulated for supplying the industry and citizen safely.

Calculation of Parameters of Crop Water Production Function of Jensen model based on Simplex Particle Swarm Optimization Algorithm

In this paper, simplex particle swarm optimization algorithm (SMPSO) is introduced to calculate Parameters of Crop Water Production Function directly. The powerful global search ability of particle swarm optimization algorithm (PSO) and the well local search ability of simplex method (SM) are combined in SMPSO. The dimension of space of a particulate is determined by the phases of crop growth and development period in SMPSO, and the sensitive index lambdai , an important parameter of crop water production function, is regarded as optimized variable to be calculated directly. This method can eliminate some problems existing in linear regression method, such as biased estimate and distortion of result and it can search the global optimization value more easily than the PSO. Through example analysis, we can see that the SMPSO is better than PSO in accuracy and calculation speed.

Model optimization of load - bearing capacity of macadam pile composite foundation based on genetic algorithm

In this paper, model optimization method of load - bearing capacity of composite foundation based on genetic algorithm is put forward. In this method, the chromosome bit string, which is looked as the generator, is used to complete random combination of influence factor and therefore need not be decoded. Considering both the fitting accuracy to modeling data and prediction accuracy to other data, model samples are divided into training samples and checkout samples, and in order to the balance between fitting accuracy and prediction accuracy, the multiple regression function is built and optimized. Through analyzing the static load experiment data of fifteen vibrating macadam pile, the genetic regression model of bearing capacity of macadam pile composite foundation is established, the prediction result shows, that the method has good fitting accuracy and predict accuracy and the stability of the model is satisfactory.

Irrigation system optimization under non-sufficient irrigation based on Elitist Non-dominated Sorting Genetic Algorithm

In this paper, irrigation system optimization with limited water supplies efficiency of water application. A system optimization model was developed for crop irrigation system with the every irrigation quota as the decision variable. The model was based on a simulation model of the soil water balance and the crop water-production function, and was optimized by the elitist non-dominated sorting genetic algorithm (NSGA-II). NSGA-II is introduced to optimize irrigation system under the condition of non-sufficient irrigation. The result of example proves that the NSGA-II method is equal-distributing, fast-calculation, and the astringency and the robustness is nice to get the Pareto optimal solution in objective space.

Stability Evaluation of Rock Slope Basing on Rough Set

In this paper, rough set theory is applied to evaluate the stability of rock slope through the analysis to slope failure examples. Genetic Algorithms (GA) is incorporated to reduce attributes and the corresponding evaluation knowledge is used to build an expert system on rock slope stability evaluation. The system adopts inexact reasoning and default reasoning methods to deal with incomplete match or information absence. In order to test the validity of this method, an example is evaluated and the evaluation results are compared with the calculation results of fuzzy regression method.