Papers by Keyword: Genetic Programming

Abstract: This paper offers a comparison of two different approaches aimed at the identification of moisture diffusivity of porous building materials as a function of moisture content. The approaches are represented by a traditional deterministic approach using the Boltzmann-Matano method and novel stochastic approach by genetic programming. The results of the comparative analysis show that genetic programming may be used as an alternative to the traditional approaches. On the basis of the very good agreement between experimental data and optimized output of genetic programming, the validation of the genetic programming method may be considered as successful.

Abstract: The aim of the paper is to demonstrate, how artificial intelligence methods, especially genetic ones, naturally combine with problems of material science. On the example of modelling a function showing how carbon concentration in steel changes its hardness it was shown how modern artificial intelligence methods can easily be adapted for solving problems of modern science. The paper presents the possibility of applying genetic programming to model properties of the steel.

Abstract: A two-run genetic programming (GP) is proposed to estimate the slump flow of high-performance concrete (HPC) using several significant concrete ingredients in this study. GP optimizes functions and their associated coefficients simultaneously and is suitable to automatically discover relationships between nonlinear systems. Basic-GP usually suffers from premature convergence, which cannot acquire satisfying solutions and show satisfied performance only on low dimensional problems. Therefore it was improved by an automatically incremental procedure to improve the search ability and avoid local optimum. The results demonstrated that two-run GP generates an accurate formula through and has 7.5 % improvement on root mean squared error (RMSE) for predicting the slump flow of HPC than Basic-GP.

Abstract: This study records the various air conditioning system parameters that affect power consumption and establishes system power consumption models for the chiller, the secondary chilled water pump, the air handling unit (AHU), and the cooling load of the AHU using artificial neural networks. The R² for each of the models are as high as 0.996. Estimations for the AHU loads in the spaces where the cooling load for the AHU are satisfied and genetic programming is used to find the optimal air conditioning system parameter set for achieving minimum power consumption. These power consumption values are then set as genetic programming end points, and the mathematical symbol (+) is used as the functional ends. Finally, the computational elements of genetic programming are used to perform iterative computation. It may be concluded from the results of the experiment that the optimal parameter set obtained from the genetic programming-based search result in a minimum power consumption that complies with the loading requirements of the location of installation result in a 22% savings in term of power consumption and an average COP increase of approximately 28%, which represent very significant improvements.

Abstract: Traditional optimization algorithms can only optimize parameters in control laws. Machine learning method can optimize parameters and evolve satellite attitude control law automatically under certain criterion. Single axis satellite attitude simulation system with noise was built up, which included satellite attitude dynamic model, sensors and actuators model. The control laws inputs were attitude error, attitude errors integral and angular velocity error, and outputs were actuators control instructions. Control laws fitness function was an attitude errors statistical function. With suitable function set selected for genetic programming (GP) and parse tree used to represent a control law expression, GP was used to evolve control law expression automatically. Simulation result shows that this method can evolve control law with uncertainties noise better. The evolved control law response and control precision are better than PID, and it can be used in satellite attitude control.

Abstract: In order to extract the fault feature validity in early fault diagnosis, method based on kernel principal component analysis and genetic programming (GP) is presented. The time domain features of the vibration signal are extracted and the initial symptom parameters (SP) are constructed. Then the combination to the initial SPs is carried on to optimize and build composite characteristics by GP. Through kernel principal component analysis (KPCA), the nonlinear principal component of the original characteristics is produced. Finally, the nonlinear principal components are selected as the feature subspace to classify the conditions of rolling bearing. Meanwhile, the within-class and among-class distance is introduced to compare and analyze the bearing condition recognition effect by using KPCA and GP plus KPCA separately. Experimental results show that the features extracted by kernel principal component analysis and genetic programming perform better ability in identifying the working states of the rolling bearing.

Abstract: Due to reliability requirements, the reaction wheel actuator of the satellite attitude control system always use traditional control method. For the satellite which has complex structure, it's difficult to build the mathematical model with classical control method. The selection of control parameters is also difficult. The design process last long and the model have poor adaptability when the parameters change. Compare to genetic algorithms, genetic programming which have the capabilities to evolve automatically, have the advantage of being able to optimize the structure of the mathematical model. Results of optimization and simulation show that design the reaction wheel actuator control law with genetic programming can simplify the design process. And the evolved control law is better than traditional PD control law.

Abstract: The disaster early-warning and corresponding prediction based on the nonlinear models is one of the important research aspects in the scope of natural disaster prediction and prevention. In this paper, the principle of applying heuristic algorithms to modify the classical neural networks so that some improved models with more efficiency can be achieved is proposed. The detailed structure and implementation of such model is also provided by building a heuristic-optimized neural network (HNN) model. Some regional flood data of China is then applied to conduct the prediction with the proposed model. Simulation results demonstrate that the proposed model can achieve higher accuracy and find an appropriate trade-off between the cost of processing time and precision, compared to the normal nonlinear models.

Abstract: Flexible Manufacturing Systems-FMS is a term with various types of definitions, each of them trying to describe the complexity and the generalized features. One of these features is their complexity, along with difficulties in building models that capture the system in all its important aspects. In a heterogeneous flexible system, the scheduling events or actions could be a combinatorial problem which claims a particular solution. Manufacturing scheduling process, in special for FMS, is a very difficult scheduling problem, because involves all the aspects of the processes: order, resources, transportation system i.e. automated vehicle guided, perturbation factors such as breakdowns of machine, etc. Typically, the scheduling problem is a NP-hard problem modeled in mathematical form. If we simulate n jobs or orders which have to be assigned to the m machines or resources, we will observe that the mathematical solution is a huge number that means (n!)m possibilities of solutions. The challenge of researchers is to solve this equation in a reasonable time with an optimal solution, and of course with minimal resources. Those scientists applied many solutions which became Operational Research-OR or Combinatorial Optimization-CO areas using a various methods: Local Search-LS, Artificial Intelligence-AI, heuristic method, priority rules, memetic or hybrid techniques which combine this techniques.

Abstract: SW-GP (Sliding Window-Genetic Programming) algorithm is provided to implement dynamic forecast of monitoring data in order to more effectively utilize coal mine monitoring data to alert and forecast safety accident. In the program, sampling data is obtained by sliding window technology and model is founded automatically by GP algorithm. The result of instance shows that forecasting values from the model well agree with the real values, which explains that employing SW-GP modeling can settle problem for alert and forecast for mine safety monitoring data satisfactorily.