Prediction of Increasing Production Activities using Combination of Query Aggregation on Complex Events Processing and Neural Network

Authors

  • Achmad Arwan

DOI:

https://doi.org/10.26594/register.v2i2.550

Abstract


Produksi, order, penjualan, dan pengiriman adalah serangkaian event yang saling terkait dalam industri manufaktur. Selanjutnya hasil dari event tersebut dicatat dalam event logComplex Event Processing adalah metode yang digunakan untuk menganalisis apakah terdapat pola kombinasi peristiwa tertentu (peluang/ancaman) yang terjadi pada sebuah sistem, sehingga dapat ditangani secara cepat dan tepat. Jaringan saraf tiruan adalah metode yang digunakan untuk mengklasifikasi data peningkatan proses produksi. Hasil pencatatan rangkaian proses yang menyebabkan peningkatan produksi digunakan sebagai data latih untuk mendapatkan fungsi aktivasi dari jaringan saraf tiruan. Penjumlahan hasil catatan event log dimasukkan ke input jaringan saraf tiruan untuk perhitungan nilai aktivasi. Ketika nilai aktivasi lebih dari batas yang ditentukan, maka sistem mengeluarkan sinyal untuk meningkatkan produksi, jika tidak, sistem tetap memantau kejadian. Hasil percobaan menunjukkan bahwa akurasi dari metode ini adalah 77% dari 39 rangkaian aliran event.

Kata kunci: complex event processing, event, jaringan saraf tiruan, prediksi peningkatan produksi, proses.

 

 

Productions, orders, sales, and shipments are series of interrelated events within manufacturing industry. Further these events were recorded in the event log. Complex event processing is a method that used to analyze whether there are patterns of combinations of certain events (opportunities / threats) that occur in a system, so it can be addressed quickly and appropriately. Artificial neural network is a method that we used to classify production increase activities. The series of events that cause the increase of the production used as a dataset to train the weight of neural network which result activation value. An aggregate stream of events inserted into the neural network input to compute the value of activation. When the value is over a certain threshold (the activation value results from training process), the system will issue a signal to increase production, otherwise system will keep monitor the events. Experiment result shows that the accuracy of this method is 77% for 39 series of event streams.

Keywords: complex event processing, event, neural networks, process, production increase prediction.

References

Bishop, C. M. (1995). Neural networks for pattern recognition. New York: Oxford University Press.

Chakravarthy, S., & Jiang, Q. (2009). Stream data processing: a quality of

Published

2016-07-01

How to Cite

[1]
A. Arwan, “Prediction of Increasing Production Activities using Combination of Query Aggregation on Complex Events Processing and Neural Network”, Register: Jurnal Ilmiah Teknologi Sistem Informasi, vol. 2, no. 2, pp. 79–85, Jul. 2016.

Issue

Section

Article