PERANCANGAN PROSES IMPLEMENTASI MACHINE LEARNING DALAM MAINTENANCE MANAGEMENT UNTUK MENCEGAH DERATING

Derating menjadi masalah yang sering kali muncul pada generator diesel di pembangkit listrik. Derating dapat menyebabkan penurunan kinerja dan produksi listrik pada generator. Kapasitas produksi listrik yang menurun nantinya akan menjadi masalah jika melihat kondisi konsumsi listrik nasional yang terus meningkat dari tahun ke tahun. Derating sering kali terjadi karena suhu yang tinggi pada charge air. Pencegahan derating dapat dilakukan dengan mengimplementasikan strategi maintenance yang mampu memprediksi derating dan mengakomodasi perubahan suhu charge air sebagai penyebab terjadinya derating. Penelitian ini akan memperlihatkan rancangan implementasi strategi maintenance berbasis data untuk memprediksi terjadinya derating dengan menggunakan pendekatan machine learning. Rancangan proses ini akan memberikan gambaran proses seperti apa yang dapat digunakan untuk mencegah derating sehingga membantu menjaga performa generator. Tahapan implementasi machine learning dilakukan dengan mengimpelementasikan proses knowledge discovery from data pada proses yang ada dalam maintenance management. Evaluasi terkait proses maintenance management dan machine learning menunjukkan bahwa machine learning dapat diimplementasikan pada tahap controlling. Klasifikasi kondisi generator juga didasarkan pada trend kondisi suhu charge air sehingga prediksi kondisi generator terkait derating tidak dipengaruhi oleh perubahan suhu yang bersifat cepat dan sementara. Penjabaran proses yang ada menunjukkan bahwa implementasi machine learning dalam maintenance management ini mungkin untuk dilakukan. Abstract[Designing The Implementation Process of Machine Learning in Maintenance Management to Avoid Derating] Derating is problem that often arises in power plant. Derating force generator to work below its optimum performance and resulting low production rate of electricity. Declining of electricity production capacity can be problem when we see condition of national electricity consumption in Indonesia which continues to increase year over year. Derating often occurs due to high temperatures in charge air. Derating prevention can be done by implementing maintenance strategy that is able to predict derating and accommodate changes in charge air temperature. This study designs processes of implementing data-based maintenance strategy to predict occurrence of derating using machine learning approach. Process design will provide overview of what kind of process can be used to avoid derating so that it helps maintain generator performance. Machine learning implementation can be done by implementing process of knowledge discovery from data in existing maintenance management process. Evaluations related to maintenance management and machine learning processes show that machine learning can be implemented at controlling stage. Classification of generator conditions is based on trend of charge air temperature so that prediction of generator conditions will not be affected by temporary changes in temperature. Process overview concludes that it is possible to implement machine learning in maintenance management.Keywords: decision tree; derating; machine learning; maintenance management

Three Dimensional Analysis and Comparison of Human Anterior Teeth and Experimentally Created Bitemark Depressions

A system was proposed to scan dental models to record three-dimensional features seen in the anterior teeth to create a database of dental profiles. Dental casts were randomly selected to create indentations in cowhide leather. Reid Bite Reader was used to measure the bite forces generated by Reynolds Controlled Bite Force Generator to make the teeth impressions. Using the Immersion MicroScribe® 3D, information from the 53 bitemark depressions and 62 sets of dental casts were transferred to an Excel Spreadsheet. Software was developed to perform the 3D comparison using metric and pattern analysis. Statistic analysis showed 100% success when comparing both arches together of the dental casts with the bitemarks or other dental casts.

Cadherin preserves cohesion across involuting tissues during C. elegans neurulation

The internalization of the central nervous system, termed neurulation in vertebrates, is a critical step in embryogenesis. Open questions remain regarding how force propels coordinated tissue movement during the process, and little is known as to how internalization happens in invertebrates. We show that in C. elegans morphogenesis, apical constriction in the retracting pharynx drives involution of the adjacent neuroectoderm. HMR-1/cadherin mediates this process via inter-tissue attachment, as well as cohesion within the neuroectoderm. Our results demonstrate that HMR-1 is capable of mediating embryo-wide reorganization driven by a centrally located force generator, and indicate a non-canonical use of cadherin on the basal side of an epithelium that may apply to vertebrate neurulation. Additionally, we highlight shared morphology and gene expression in tissues driving involution, which suggests that neuroectoderm involution in C. elegans is potentially homologous with vertebrate neurulation and thus may help elucidate the evolutionary origin of the brain.

Stoichiometric interactions explain spindle dynamics and scaling across 100 million years of nematode evolution

The spindle shows remarkable diversity, and changes in an integrated fashion, as cells vary over evolution. Here, we provide a mechanistic explanation for variations in the first mitotic spindle in nematodes. We used a combination of quantitative genetics and biophysics to rule out broad classes of models of the regulation of spindle length and dynamics, and to establish the importance of a balance of cortical pulling forces acting in different directions. These experiments led us to construct a model of cortical pulling forces in which the stoichiometric interactions of microtubules and force generators (each force generator can bind only one microtubule), is key to explaining the dynamics of spindle positioning and elongation, and spindle final length and scaling with cell size. This model accounts for variations in all the spindle traits we studied here, both within species and across nematode species spanning over 100 million years of evolution.