Pengendalian Kualitas Statistik pada Proses Produksi Pupuk NPK Phonska di PT Petrokimia Gresik Menggunakan Diagram MEWMA Berbasis Residual Model N-BEATS

Meilisa, Elsa (2026) Pengendalian Kualitas Statistik pada Proses Produksi Pupuk NPK Phonska di PT Petrokimia Gresik Menggunakan Diagram MEWMA Berbasis Residual Model N-BEATS. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5003221038-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (13MB) | Request a copy

Abstract

Produksi pupuk NPK Phonska memiliki peran penting dalam mendukung ketahanan pangan nasional sehingga kualitas produk perlu dijaga agar memenuhi spesifikasi yang telah ditetapkan. Pengendalian kualitas pada proses produksi pupuk menghadapi tantangan berupa karakteristik kualitas yang saling berkorelasi dan adanya autokorelasi pada data proses. Penelitian ini bertujuan untuk memonitor rata-rata proses produksi pupuk NPK Phonska menggunakan diagram kontrol Multivariate Exponentially Weighted Moving Average (MEWMA) berbasis residual model Neural Basis Expansion Analysis for Time Series (N-BEATS) serta mengevaluasi kapabilitas proses produksi. Data yang digunakan meliputi karakteristik kualitas pupuk NPK Phonska, yaitu kadar H₂O, N, P₂O₅, dan K₂O. Model N-BEATS dibangun untuk mengatasi autokorelasi dan menghasilkan residual yang kemudian dimonitor menggunakan diagram kontrol MEWMA pada fase I dan fase II. Selanjutnya, dilakukan analisis kapabilitas proses menggunakan indeks kapabilitas non-normal univariat (CNp) dan multivariat (MCNp). Hasil penelitian menunjukkan bahwa model N-BEATS terbaik diperoleh dengan input_chunk_length sebesar 96 (setara 16 hari), n_epochs sebesar 150, num_stacks sebesar 20, num_blocks sebesar 2, num_layers sebesar 4, layer_widths sebesar 1024, dan batch_size sebesar 32, dengan nilai MAPE sebesar 12,24%, sMAPE sebesar 12,64%, dan RMSE sebesar 1,13. Hasil monitoring menggunakan diagram kontrol MEWMA menunjukkan masih terdapat titik out of control pada fase II untuk seluruh nilai λ yang digunakan, sehingga rata-rata proses produksi pupuk NPK Phonska belum terkendali secara statistik. Variabel yang paling dominan menyebabkan kondisi out of control adalah P₂O₅ dan N. Analisis kapabilitas menghasilkan nilai MCNp sebesar 0,23 yang menunjukkan bahwa proses produksi pupuk NPK Phonska belum kapabel dalam memenuhi spesifikasi kualitas yang telah ditetapkan oleh perusahaan. Berdasarkan diagram sebab-akibat, terdapat empat faktor yang diduga menyebabkan munculnya titik pengamatan out of control, yaitu manusia, mesin, metode, dan bahan baku.
================================================================================================================================
NPK Phonska fertilizer production plays an important role in supporting national food security; therefore, product quality must be maintained to meet established specifications. Quality control in the fertilizer production process faces challenges due to the presence of correlated quality characteristics and autocorrelation in process data. This study aims to monitor the process mean of NPK Phonska fertilizer production using a residual-based Multivariate Exponentially Weighted Moving Average (MEWMA) control chart with the Neural Basis Expansion Analysis for Time Series (N-BEATS) model and to evaluate the process capability. The data analyzed consist of four quality characteristics of NPK Phonska fertilizer, namely H₂O, N, P₂O₅, and K₂O contents. The N-BEATS model was developed to address autocorrelation and generate residuals, which were subsequently monitored using the MEWMA control chart in Phase I and Phase II. Furthermore, process capability analysis was conducted using the univariate non-normal process capability indeks CNp and the multivariate process capability indek (MCNp). The results indicate that the best N-BEATS model was obtained with an input_chunk_length of 96, n_epochs of 150, num_stacks of 20, num_blocks of 2, num_layers of 4, layer_widths of 1024, and batch_size of 32, yielding a MAPE of 12.24%, sMAPE of 12.64%, and RMSE of 1.13. The monitoring results using the MEWMA control chart show that out-of-control observations still exist in Phase II for all values of λ, indicating that the process mean of NPK Phonska fertilizer production has not yet achieved statistical control. The quality characteristics that most significantly contribute to the out-of-control condition are P₂O₅ and N. The process capability analysis produced an MCNp value of 0.23, indicating that the NPK Phonska fertilizer production process is not capable of consistently meeting the quality specifications established by the company. Based on the cause-and-effect diagram, four factors are suspected to contribute to the occurrence of out-of-control points: man, machine, method, and raw materials.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Deep Learning, MEWMA, N-BEATS, NPK Phonska Pengendalian Kualitas, Deep Learning, MEWMA, N-BEATS, NPK Phonska, Statistical Quality Control
Subjects:	H Social Sciences > HD Industries. Land use. Labor > HD9980.5 Service industries--Quality control. R Medicine > R Medicine (General) > R858 Deep Learning
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Elsa Meilisa
Date Deposited:	30 Jul 2026 02:22
Last Modified:	30 Jul 2026 02:22
URI:	http://repository.its.ac.id/id/eprint/139144

Actions (login required)

View Item