基于遗传算法的神经网络在爆破振动预测中的应用

Application of Neural Network based on Genetic Algorithm in Prediction of Blasting Vibration

针对BP神经网络对工程爆破振动的预测存在精度不够高的缺点,建立遗传算法优化神经网络的模型,并介绍了它的原理。最后通过爆破振动预测实例的介绍,应用MATLAB编程,将总装药量Q、测点与爆源的高差h、孔间微差时间t、最大药包距离L这4个参数作为模型参数,对爆破振动幅值v、振动主频f和振动持续时间T进行预测,得出基于遗传算法的神经网络预测的结果比BP神经网络更为精确,克服了BP神经网络的缺点。

In view of the shortcoming of bad precision of forecasting the blasting vibrationi by BP neural network, the genetic algorithm optimization neural network the model is established and principles is introduced. Through introduction of the example of blasting vibration forecast, the total charge amount, and height, delay time between holes, the maximum distance to blast source these four parameters are used as the model parameters to torecast blasting vibration amplitude, vibration frequency and vibration duration by applying the MATLAB programming, more precise based on the genetic algorithm neural network forecast result compared with BP neutral network is obtained, which overcome the BP neural network shortcoming.