DSDesri Kristina Silalahi

Desri Kristina Silalahi

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Desri Kristina Silalahi has a total of 18 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    LAGADS: Lag-Aware and Gated Adaptive Dynamic Selection Transformer for Multivariate Long-Term Forecasting

    by Chao-Lung Yang, Desri Kristina Silalahi, Hsin-Chun Lin, Yung-Yao Chen

    Long-term multivariate time series forecasting (LTSF) plays a pivotal role in facilitating long-term planning across domains such as energy management, transportation, and finance. While Transformer-based models that replace cross-time attention with cross-variable attention have demonstrated strong forecasting performance, three limitations remain. First, cross-variable attention assumes zero-lag dependency between variables, ignoring propagation delays common in real-world systems. Second, correlation estimation is vulnerable to bias from shared non-stationary components. Third, the channel-dependent or channel-independent processing decision is applied uniformly across all variables. To address these limitations, we propose LAGADS (Lag-Aware and Gated Adaptive Dynamic Selection), comprising three complementary modules: (1) a Lag-Aware Cross-Variable Attention module that learns a temporal shift matrix via differentiable frequency-domain phase modulation; (2) a Pattern-Adaptive Frequency Normalization (PA-FAN) module that separates non-stationary spectral components before correlation estimation; and (3) a Dynamic Variable Selection (DVS) Gate that adaptively fuses channel-dependent and channel-independent representations per variable based on correlation, spectral entropy, and temporal variability. Experiments on nine benchmark datasets show that LAGADS improves forecasting accuracy over several state-of-the-art Transformer-based baselines, with the largest gains observed on high-dimensional datasets exhibiting structured propagation delays and heterogeneous per-variable correlation structure. Beyond forecasting accuracy, LAGADS produces physically interpretable lag structure and per-variable routing decisions that offer actionable insight for engineering decision-making in domains such as power grid monitoring and transportation management, while requiring substantially lower computational resources than comparable Transformer-based baselines. Ablation studies confirm that all three components contribute independently and complementarily to forecasting accuracy.

  • Open Access

    Pemasangan dan Penyuluhan Sistem Pengolahan Limbah Batik di Kampung Batik Pasundan Desa Sayati

    by Desri Kristina Silalahi, Jaspar Hasudungan, Jangkung Raharjo, Nabila Hatami Putri, Nadya Ainun Avrilya, Sudiana Sudiana

    Beberapa Industri batik menggunakan bahan yang lebih fleksibel dan mudah didapatkan untuk meningkatkan efisiensi produksi, salah satunya menggunakan bahan kimia sebagai pewarna sintetis. Penggunaan bahan-bahan ini untuk efisiensi dan biaya produksi yang dapat merusak kualitas air dan mengancam keseimbangan ekosistem. Kampung Batik Pasundan yang terletak di Desa sayati belum memiliki instalasi pengolahan air limbah. Melalui kegiatan pengabdian kepada masyarakat dilakukan pemasangan sistem pengolahan air limbah yang dapat digunakan komunitas batik di Kampung Batik Pasundan. Â Air limbah batik disaring dalam media penyaringan sebagai alat untuk mengolah air limbah. Air limbah yang sudah melewati pengujian laboratorium menghasilkan penurunan kandungan pH yakni dari 8.08 sebelum dilakukan penyaringan menjadi 7.75. Selain itu, dilakukan juga sosialisasi terkait bahaya limbah batik kepada komunitas batik serta pengoperasionalan sistem pengolahan air limbah yang telah terpasang. Komunitas ini dapat merasakan manfaat secara langsung.

  • Open Access

    Design of passive voltage balancer system for lead acid battery

    by Bandiyah Sri Aprillia, Desri Kristina Silalahi, Kharisma Bani Adam, Putu Krishna Bhaskara Perteka Yuda

    The use of rechargeable batteries for electrical energy storage requires a voltage balancing system. The voltage balancing system requires monitoring in the use of rechargeable batteries so that they can be utilized properly and can improve the electrical energy storage system. In this research, a monitored lead acid battery voltage balancing system was designed so that the management of the battery voltage balance level and the storage of electrical energy in rechargeable batteries can be stored and used optimally. In this research a series of lead acid battery voltage detection and power dissipation circuits were designed. The power dissipation circuit uses the controlled shunt resistor method which is used when the voltage of the lead acid battery being charged exceeds the maximum voltage. This method is easy to implement and can display the value of the lead acid battery voltage and other parameters, so that it can be monitored by the user. The results obtained show that the average voltage error for batteries 1 and 2 is 0.09% and 0.3% respectively. The power dissipation circuit can dissipate lead acid battery power above the reference voltage VREF=±7 V, with a balanced voltage of 6.8 V at 140 minutes and 160 minutes.


  • Open Access

    Pelatihan Kewirausahaan Secara Daring untuk Meningkatkan Motivasi Berwirausaha Siswa-Siswi MA Ishlahul Amanah

    by Desri Kristina Silalahi, Bandiyah Sri Aprillia, Kharisma Bani Adam

    Purpose: The students of MA Islahul Amanah are the next generation of the nation who must take their steps to reach the future. Most of them have constraints in terms of costs to continue on to higher education. Thus, a motivation or direction is needed for MA Ishlahul Amanah students about entrepreneurship. Entrepreneurship is the right alternative choice to overcome unemployment. This is to encourage them to be more independent after completing their school education. At least they are not a burden to their parents, but can help their family's economy Methodology/approach: The methods used are: (1) condition survey; (2) team conciliation to discuss the activities carried out; (3) preparation to determine the implementation time and equipment to be prepared; (4) implementing activities by conducting online entrepreneurship training; (5) evaluation of activities to measure the achievement of activities with a questionnaire. Results/findings: Participants receive tips on building start-ups, namely innovating business ideas, creating business models, raising capital, inviting like-minded people to maintain service/product quality and "don't give up". Apart from that, they also gain knowledge and experience related to what the speaker has experienced because the material provided is also based on their own experience. This activity was also well achieved based on the results of the questionnaire which achieved an average percentage of agreeing is 77.33% and the percentage strongly agreeing is 20%. Conclusion: Community Service Activities with activities in the form of entrepreneurship training to increase interest in entrepreneurship for MA Ishlahul Amanah students run smoothly and goals are achieved. This can be seen from the results of the questionnaire, the participants chose to agree and strongly agree with the program being carried out because it met the needs of the participants.


  • Open Access

    Forecasting of Poverty Data Using Seasonal ARIMA Modeling in West Java Province

    by Desri Kristina Silalahi

    The government continues to carry out poverty reduction strategies in Indonesia, especially in West Java Province. West Java Province is a province that has the most populous population in Indonesia. This will affect the level of welfare and the amount of poverty. The strategy undertaken is inseparable from accurate poverty data and is available from year to year. Even from the available data, the government can forecast the number of poor people in the coming years. Seasonal Autoregressive Integrated Moving Average (SARIMA) method is one of forecasting methods. SARIMA is the development of the ARIMA model which has a seasonal effect. Based on the results of the study, that poverty data forecasting in the province of West Java using the SARIMA method obtained SARIMA model (0,1,1) (1,1,1)4. This model is the best model for forecasting data with an R-Squared value of 98%, Mean Square Error is 7.705.5800.000 and Mean Absolute Percentage Error IS 2,81%. It’s means this SARIMA model is very good in predicting poverty data in West Java Province.


  • Open Access

    Design of Automatic Switch System of Residential Load From Solar Cell and Power Plant Resources using Neural Network

    by D K Silalahi, B S Aprilia, W Priharti, K Kumillayly, S Saidah

    Residential loads are electronic equipment that is often used at home. Residential loads are supplied from two sources, they are solar panel and the State Power plant. Selecting the supply load source, so an automatic switch system is needed from the load. Optimizing residential load resources and avoiding overloading, load balancing techniques are needed. The switch system uses a relay, determined by the solar panel power output. This research discusses the design of automatic load switch systems for residential loads from solar panels and State Power plant using Artificial Neural Networks (ANN). ANN control system that is arranged using ANN Backpropagation consists of 4 inputs, four hidden layers, each consisting of 4 neurons and one neuron in the output layer. In this research, to determine the network that has been formed to provide changes of load. The results of testing that the parameters used to get the smallest error rate in the process of setting an automatic power load switch is best to use a number of repetitions of 2000 times with an error percentage 5.3%. Artificial Neural Networks can experience convergent failure or not close to output because the initial guess is not good. Initial experiments with actual solar panel data with as many as 98 data produced non-convergent output values. The solution to improving the initial experiment is to simplify learning data 40 data produces convergent output.


  • Open Access

    Desain Sistem On-Grid Energi Terbarukan Skala Rumah Tangga Menggunakan Perangkat Lunak HOMER

    by Bandiyah Sri Aprillia, Desri Kristina Silalahi, Muhammad Agung Foury Rigoursyah

    Bandung is located on the island of Java with Latitude -7.5 and longitude 107.5. This location has average solar radiation of 4.89 KWh / m2per day so suitable for solar energy utilization. Solar energy can be converted into electrical energy through photovoltaic system so that it can be applied to residential load. The design of solar power plants for residential load requires accurate calculations. In this research, HOMER was used to evaluate the design of renewable energy on grid systems at residential load 900 VA in Bandung. The research method used arecollecting solar radiation data, equipment specifications and 900 VA electrical load studies. Based on the results of the design on grid photovoltaic system using HOMER, the configuration of 6 PV 2040 Wp units, 4 unit batteries, and 2 kW inverter is the optimum configuration. The system configuration can provide 45.5% of the total daily loadconsumption with the lowest total Net Present Cost (NPC) is IDR 75,300,000.00. The On Grid system that has been designed will achieve an annual profit IDR 1,806,884.00 compared to conventional electricity usage (PLN) and only takes 7 years to recover investment costs.


  • Open Access

    Penyuluhan Penggunaan Listrik dari Sumber Energi Surya Di Pesantren Al Mukarramah Kabupaten Bandung

    by Wahmisari Priharti, Ekki Kurniawan, Desri Kristina Silalahi

    Al Mukarramah Boarding School is located in Bandung Regency, West Java Province. The total land area owned by this boarding school is 3000 m2 (three thousand square meters) consisting of several spaces and buildings. PLN is the main electricity source for lighting in this schools with a total power of 4400 VA and costs an average of Rp. 3,000,000.00 per month. This cost is certainly quite burdensome since not all santri are well-off to pay education coaching contributions (SPP) as a source of monthly operational costs. Therefore, efforts to reduce the costs borne by this school through alternative electricity sources need to be done. This was carried out in Community Service (PKM) through the installation of a simple Solar Power Plant. Counseling on how the PLTS works and the maintenance technique of PLTS is also provided so that PLTS can work well and can last a long time. Abstrak. Pesantren Al Mukarramah terletak di Kabupaten Bandung, Provinsi Jawa Barat. Luas tanah yang dimiliki pesantren ini sekitar 3000 m2 (tiga ribu meter persegi) yang terdiri dari beberapa ruang dan bangunan. Sumber listrik utama bagi penerangan di pesantren saat ini berasal dari PLN dengan total daya 4400 VA dan menghabiskan biaya rata-rata Rp 3.000.000,00 per bulan. Biaya sebesar ini tentunya cukup memberatkan sebab tidak semua santri berkecukupan untuk membayar sumbangan pembinaan pendidikan (SPP) sebagai sumber biaya operasional bulanan. Oleh karena itu, upaya untuk meringankan biaya yang ditanggung pesantren melalui sumber energi listrik alternatif perlu dilakukan. Hal ini dilaksanakan dalam Pengabdian Kepada Masyarakat (PKM) melalui pemasangan Pembangkit Listrik Tenaga Surya sederhana. Penyuluhan tentang cara kerja dan teknik pemeliharaan PLTS ini juga diberikan agar PLTS dapat bekerja dengan baik dan bertahan lama.