INTELLIGENT SYSTEM FOR ACCELEROGRAPHIC PROCESSING IN PERU

Peru is located in a high seismic hazard zone as its territory is subject to the interaction of the Nazca and South American Plates, as well as surface faulting which also generate large earthquakes along the continent. These seismic events create seismic forces that impact structures, frequently causing numerous material and human losses to vulnerable buildings throughout Peruvian territory. The last major seismic events to have occurred in Peru were in 2001, in Arequipa, and in 2007, in Pisco, reaching magnitudes Mw 8.4 and 7.9, respectively. When these earthquakes took place, Peru did not have enough instrumentation to measure seismic events so they were only recorded by few stations. This lack of information led the National Engineering University (UNI) in 2014, through the Faculty of Civil Engineering’ Accelerographic Network, to acquire the necessary instruments to implement eighty accelerograph stations which are now installed in different cities of our country. An Intelligent System for Accelerographic Processing, known as SIPA, (for its acronym in Spanish) was also developed additionally to this instrumentation. This system uses intelligent algorithms (neural networks) that help process accelerograms such a degree that reports are created automatically and in real time. Such data includes history time records, response spectra and Fourier spectra, spectral ratio. This benefits researches greatly as they now have more time to focus on correctly interpreting the information obtained, thus allowing them to contribute to improving seismic risk management in our country. SIPA is an information system capable of determining whether or not an accelerogram contains a seismic event. To perform this task, neural networks are used. This paper will describe the topology of the neural network, as well as the performance tests obtained during in recent years. Training and evaluation of the neural network used 2016 to 2018 accelerograms, and the backpropagation method was used for the training. The paper also includes as an example the Mw 8 Lagunas earthquake report, dated May 26, 2019 at 2:41 a.m., which occurred in Lagunas, Alto Amazonas, Loreto, in the northeastern part of Peru. The Accelerographic Network’s Intelligent System for Accelerographic Processing (SIPA) recorded the event at 44 different accelerographic stations, the highest recorded acceleration being 95.84 gal, which allowed to learn about the amplification and attenuation of the accelerations, as well as the dynamic behavior of the soil of the different cities in which this equipment was installed.