Papers by Keyword: Prediction

Abstract: Regression analysis, as an important branch of statistics, is an effective tool for scientific prediction. Genetic algorithm is an optimization search algorithm in computational mathematics. In this paper, a new regression model named quasi-linear regression model is established. Further, its implementation method is introduced in detail. Then by taking the population development of Hebei province as an example, we conduct the fitting problem and short-term prediction. Moreover, we compare the fitting effect and the prediction results of two models.

Abstract: Over the last decade, functional magnetic resonance imaging (fMRI) has become a primary tool to predict the brain activity.During the past research, researchers transfer the focus from the picture to the word.The results of these researches are relatively successful. In this paper, several typical methods which are machine learning methods are introduced. And most of the methods are by using fMRI data associated with words features. The semantic features (properties or factors) support words neural representation, and have a certain commonality in the people.The purpose of the application of these methods is used for prediction or classification.

Abstract: Recently, the safety of existing civil engineering structures attracts more and more attention. The long-term strength of concrete plays a key role during the assessment of safety and durability for civil engineering structures. The strength of concrete will gradually decrease during the service of civil engineering structures. It is significant to accurately predict the strength deterioration of concrete for correctly evaluating the safety of structures. The factors affecting the long-term strength of concrete include environment type, age, climate, water cement ratio, amount of cementing material and so on. In this paper, artificial neural network with powerful mapping ability has been selected to predict the long-term strength of concrete. First, there-layer BP neural network with age, water cement ratio, amount of cementing material as input and long-term strength as output was built. Then, the neural network was trained by the samples measured in real structures and the well-trained neural network was test. From the test results, the trained neural network can accurately predict the long-term strength of concrete with the error less then 9%.

Abstract: During the construction process of massive concrete, the construction must forecast and control the temperature forming of massive concrete in order to avoid the temperature crack owing to the thermal stress caused by the concrete heat. The discussion on the concept of grey system interval and grey number predication applies to the temperature rise prediction of massive concrete in the construction process. The previous temperature rise data is divided into the white part and grey part based on real numbers. And the interval grey number forecast computing models are established, so we can finish the temperature rise prediction during construction process. The forecasting results of case show that, this model has fast convergence and remarkable precision. Then, it is possible to realize the accurate temperature rise prediction of massive concrete.

Abstract: During the past decades, the microseismic technique has changed from a simple researching means for rock failure study to a practical method for daily safety monitoring at high geo-hazard tendency mines. This paper introduces the microseismic monitoring system built at a metal mine aiming at the prediction of the geological hazard. The arrangement of microseismic monitoring sensors should take consideration not only of the current mining situation, but also by the aid of experts experiences. By a successful case of prediction for hard brittle failure, the Shizhuyuan microseismic monitoring system was proven to be valid and necessary in the prediction of geological hazard.

Abstract: Blasting on large mounts of experiments, we construct mathematical model of seismic wave of blast with systems response method and predict waveform of blast seismic wave. The model can reflect the influence of seismic wave from several kinds of factors and can simulate the attenuation low of amplitude and frequency in the whole process of blast seismic wave. It not only simulates better for simple blast seismic wave, but also can get consistent result in simulating porous blast seismic wave.

Abstract: Correct resolution of ambiguities for GNSS reference network is prerequisite for generation of corrections for network RTK. Due to the presence of atmospheric delay in the double-differenced observations, the convergence time of ambiguity is about a dozen minutes and even dozens of minutes for medium and long baselines. And in the case of loss-of-lock or new rising satellites, the integer ambiguities have to be redetermined over and over again. But for the application of GNSS network RTK, the resolution of ambiguity needs to be determined real time as possible. Atmospheric delays of previous epochs are used to predict atmospheric delays of following epochs, and then wide and narrow lane combination of carrier-phase observations as well as ionosphere-free combination is used to resolve ambiguities with single epoch. Our test results show that the precision of predicted double-differenced tropospheric and ionospheric delays is about 2-3 cm using temporal-and spatial-correlation exponent model, and the success rates of L₁ ambiguities resolution with single epoch reach above 90% for medium and long distance GNSS reference station network.

Abstract: The chaotic frequency hopping sequences possesses short-term predictability. Via the phase space reconstruction approach, we can get chaos attractors, and the problem of series prediction can be transformed into the regression problem of the chaotic attractors. This paper uses SVR method to deal with the prediction of frequency hopping sequences. After analyzing the characteristics of existing kernel functions, we produce a new multi-kernel function, which is used for the prediction of frequency hopping sequences. Experiments show the fine performances of our methods.

Abstract: The wear and tear allowances (displacements) of axial thrust bearing in air compressor was diagnosed and predicted, applying the model of artificial neural network (ANN), and compared with the traditional method of diagnosis and prediction. It showed that the results of diagnosis and prediction are more precise than that of traditional method. It can diagnosis and predicts the wear and tear allowances of axial thrust bearing better.

Abstract: According to fire protection safety system, such as students' dormitories, put forward a fire prediction method based on SVM to deduce the possibility and risk level, in which peoples behavior state is combined with static fire factors .Then, the simulation results are shown the method is effective.