ارائه مدل تغییرات سرعت امواج زلزله ( Vpn) بر اساس الگوریتم ژنتیک (مطالعه موردی- ایران)

زلزله یکی از خطرناک‌ترین بلایای طبیعی عصر حاضر به شمار می‌رود که همواره اهمیت خود را به‌طور عینی نمایان کرده است. زلزله سانحه‌ای طبیعی است که بر اساس میزان بزرگی خود می‌تواند در مدت کوتاهی فجایعی عظیم به وجود آورد. هدف این مقاله، استخراج و ارائه مدلی برای تغییرات سرعت امواج زلزله (vpn) با استفاده از الگوریتم ژنتیک می‌باشد. داده‌های مورد استفاده در این تحقیق از مرکز لرزه‌نگاری کشوری مربوط به موسسه ژئوفیزیک دانشگاه تهران دریافت شده است. در این تحقیق، سه استان کرمانشاه، آذربایجان شرقی و کرمان انتخاب شدند. به منظور استفاده از الگوریتم ژنتیک، ابتدا داده‌های دریافت شده از مرکز لرزه‌نگاری کشوری برای این سه استان با هم ادغام شده که بالغ بر تعداد 1863 رخداد زلزله برآورد گردید. پس از استخراج داده‌های مربوطه، نسبت به محاسبه سرعت امواج زلزله( vpn) اقدام شد. سپس با نادیده گرفتن حدود 25 درصد از این داده‌ها، نسبت به استخراج مدل ریاضی برای سرعت امواج زلزله، اقدام شد. در انتها، فرمول به‌دست‌آمده در مورد داده‌های چشم‌پوشی شده اولیه (25 درصد) اعمال گردید که نتایج مشابهی به دست آمد.

Presenting a Model of Earthquake Wave Velocity Changes (VPn ) Based on Genetic Algorithm (Case Study - Iran)

Earthquake is one of the most dangerous natural disasters of the present age, which has always shown its importance objectively. An earthquake is a natural disaster that, depending on its magnitude, can cause massive catastrophes in a short time. In this study, the authors seek to provide a simple analytical form for the propagation speed of waves, which despite previous studies, has not received much attention. Therefore, the purpose of this paper is to extract and present a model for earthquake wave velocity changes ( VPn) using Genetic Algorithm (GA).Research Methods The data used in this study were received from the National Seismological Center of the Institute of Geophysics, University of Tehran. In this study, three provinces of Kermanshah, East Azerbaijan and Kerman were selected. Earthquake event characteristics of each of these three provinces in the period between 2006 and the end of 2018, with a focal depth of up to 30 km and magnitude between 4 and 8 were selected. In order to use the Genetic Algorithm (GA), first the data received from the National Seismological Center for these three provinces were merged, which was estimated at 1863 earthquake events. After extracting the relevant data, the earthquake wave velocity (VPn ) was calculated. Then, ignoring about 25% of this data, a mathematical model for earthquake velocity was extracted. Finally, the obtained formula was applied to the initial ignored data (25%), which had similar results. To model the changes in wave velocity according to distance changes, a mathematical relation was considered as an exponential function and the unknown parameters of the model were determined using a Genetic Algorithm (GA). To find a suitable model between distance and speed, the following relation is considered for it.V(X)=a+bkX                                                                                                                                                        (1)         In this regard, a, b and k are constant coefficients and x is the distance from the earthquake site in terms of one thousand kilometers. In the above equation, the coefficients must be determined so that the output of this model with the recorded data has the least amount of error. For this purpose, the Genetic Algorithm (GA) optimization method is used, and the error between the model output and the actual data was considered as the objective function of the optimization problem, and the optimization variables were determined with the aim of minimizing this objective function. The objective function is defined as follows:                                                                                                                                        (2)In this connection,Vi is the velocity obtained as a measure and V(Xi) the amount of speed obtained according to the Equation (1). For implementation, the Genetic Algorithm (GA) has been used 2000 populations and 30 generations. Also the coefficient of crossover is equal to 70% and coefficient mutation is equal to 2%. The output of the model for training and test data showed that the proposed model has acceptable accuracy for modeling the velocity of longitudinal waves. This model can be used to determine the arrival time of waves of an earthquake to different points. It is also possible to estimate the location of the earthquake by recording the occurrence of the earthquake at several different points and using the provided relationship.