GN Y-1 #5011 in 1927 (before wreck and rebuild). Motor–generator supplied direct current fed to 6 traction motors. The Great Northern Railway 's class Y-1 comprised eight electric locomotives with AAR 1-C+C-1 wheel arrangements.

The GN numbered the units 5010–5017 and classified them Y-1. After being involved in a wreck at Tonga, Washington in July 1945, the 5011 was rebuilt with a streamlined appearance and EMD F-unit cabs; the GN reclassified it as Y-1a. In 1956, the GN dieselized operations through the Cascade Tunnel.

This is a complete and indexed 1947 Great Northern publication on the operation of GN's electric locomotives from Skykomish to Wenatchee loaned to me courtesy of Mr. Dave Sprau. For reference, this book refers to the Y-1 and Y-1a classes of …

They used motor-generator sets to rectify the alternating current line voltage into direct current for their traction motors. The GN numbered the units 5010–5017 and classified them Y-1.

Original Y-1 configuration. The head light is mounted on the top center with the bell behind it. In the thirties Y-1's were modified by adding a busbar and moving the bell and headlight. Intermediate Y-1 configuration. The Busbar, known to crews as a stinger, has been added to the roof in the former location of the headlight and bell.

Great Northern used 2 different paint schemes for its electric locomotives. All electrics were delivered in a Pullman Green livery except for the two W-1 class locomotives which arrived new in the Empire Builder scheme of orange and green separated by imitation gold stripes.

GN 5017 Y-1 class electric locomotive on GN Train #6, The Cascadian at Skykomish, WA. Philip B. Korst, Jr. photo. GN 5018 and 5019 W-1 class electric locomotives at Skykomish, WA.

In the late 1920s, Great Northern received eight Alco-GE box cab electrics, classed Y1. They were used over the heavily tunneled and steeply graded Cascade Mountains in Washington, where heavy freight had to be muscled to the coast.

2020年4月28日 · GN算法，即Girvan-Newman 社区划分算法，它属于分裂的层次聚类算法，基本思想是不断删除网络中“介值”最大的边，直到所有边都被删除。目标函数值取得最大值的时候，以当前图的连通分量为社区进行划分。

2024年2月27日 · 在训练 神经网络 时，往往需要 标准化（Normalization）输入数据，使得网络的训练 更加快速和有效，然而 SGD 等学习算法会在训练中不断改变网络的参数，隐含层的激活值的分布会因此发生变化，而这一种变化就称为 内协变量偏移（Internal Covariate Shift，ICS）。 为了减轻ICS问题，Nomalization固定激活 函数 的输入变量的均值和方差（Nomalization后接激活函数），使得网络的训练更快。 其计算方式 大白话 如下： 将输入的数据，按照一定维度进行统 …