顺便说下，我为什么觉得写的好，因为其实在你做技术的过程中，设计的过程中，真正复杂的、难以解决的问题，归根结底都是那些基础的数学与力学知识，这些系统的知识、深入的知识必须高强度学习才能掌握。像“六轴机器人的传动结构”、“多少个动作的非标装配机”，这些找机器多看看，多拆拆，都会慢慢有办法解决。
我唯一想补充的一点，就是还要掌握材料知识，“材料”和“数学”，是机械工程的精华。
我写的如下这些建议，希望你不要泄气或厌恶。如果让我重过一次，我会把它给我自己看。如果现在招人，我会招这样的学生。
第一，Solidworks/ProE/AutoCAD/Rhino/Blender/CATIAand GD&T这些不是拿到学位工程师的技能要求，成为工程师不是一个画图员，就像在简历上说你会MSoffice一样，花点时间你能轻松学会它。现在，下面这些是你在大学四年中需要熟练掌握的。
第二，我们在这说的是成为一个工程师，是那种可以实实在在建造火箭和微型发动机的。我不反对分数制，不是很在乎它，因此这里我不讨论如何得最高分。
- 阅读维基百科.
- 编程：从Matlab/Python开始，接着C++。举个例子，要达到用这些语言可以自己编写一个图像引擎。为什么？因为这能让你把矢量、阵列、变换图形化，并通往高维代数。要确保你能理解和应用Runge-Kutta的算法，这样才算学好。不要只是用windows，也要领略一下Linux或Mac的风采。要能理解batch/shell脚本语言的原理，并能把利用重要的开源脚本搭建自己的脚本。如果你在一年级或二年级什么事也没干，确保一定要精通这些。
- 线性代数和微分方程：现在大部分机械工程的大纲都要求尽早学这门课，但很少有机械工程师能真正理解，它们是机械工程的根本，再怎么强调都不过分。很多机械专业教授都不理解线性代数的重要性，把它教砸，去听计算机、数学专业老师开的课。或去Youtube听GilbertStrang 的课。把它和编程结合，进行数值仿真。不要等编程学完，再学它们。
- 统计学：学两遍，第一年学，高年级再学。这是唯一一个任何专业都非常重要的一门课。
- 工程数学：空间变换、傅里叶分析、复变函数、位势理论、偏微分方程组、插值/曲线拟合、优化理论。结合编程技能，实践它们。如果认为有些没用跳过去，都是错误的。好的工程师每天都用它们。
- 动力学/高等动力学：听物理系的力学课，机械的教授总是用代数的方法对待力学，对概念解释不够好。你的目标是能独立建立复杂机械的FBDs(应该指自由体受力图)，能写出经典的随时间变化系统的自治/非自治、线性/非线性微分方程，熟悉指标记法，张量和算子空间，你的编程经验可以帮助你。
- 静力学/固体力学：精通铁木辛柯的弹性理论，即使花去你的余生。如掌握了第2点，你应该能知道SFDs和BMDs的无效和莫尔圆概念。要尝试把简单的例子图形化，其实这并不简单。使用你的编程技能去解ODEs方程（常微分方程）的数值解。
- 振动理论：如果你熟练掌握了第2点，这会比较轻松。振动理论主要是研究二阶、齐次/非齐次、自治/非自治、变参/非变参常微分方程。如你掌握了第5点，你会知道怎么计算响应、地震扰动、减震、旋转机械等。掌握第6点，可以解决板、梁的振动问题。同时掌握2和4，应能够解多自由度系统，掌握模态分析方法。在这里还要学习耦合的SHO/QHO概念。
- 热力学/流体力学：我不适合对这部分内容发表意见，但它们在本科阶段并不难，并且主要是应用微分方程和连续介质力学。
爱好者和数学家也设计机器，但机械工程师不从零开始。我们遵循工业标准，组合并匹配已有组件，用已有的算法来创造新东西，例如运动链、连杆综合和设计。确保读过齿轮、机械学、4连杆机构、凸轮、间歇传动轮。有可能，工程师创造这些机构并不在行，一个技师或工人会做得更好，但你能用固体力学知识去设计好零件，承受极大的冲击力。
忽略掉’制造’、‘产品工程’课程，因为在学校教这些，毫无价值。你不可能在教室里精通制造，你也不可能在学校学会设计一个好机器。那些公理设计原理、产品生命周期管理、甘特图、头脑风暴都是胡说八道。没人真的那样做，那样做的人，不是工程师。如果你想了解制造，粗读一下RobThompson的《面向设计的制造方法》，去跟车间的人交谈，去看youtube上的how it’s made。想去了解产品设计过程，去看Kickstarter。
不要浪费时间在概述或介绍类课程上，不要参加不感兴趣主题的讨论课。应该参加承诺展示你数学、方法或酷视频的讨论课。要时刻关注这样的案例研究：清楚详细的展示如何利用数学或实验对系统进行建模或实现。避免‘设计’研讨（通常来自Wharton、Sloan 或Kellog商学院），这些看着美好，但毫无用处。
参加所有实验课，只要你负担得起。在实习车间，有空余时间去看别人如何工作。使用哪里设备，直到弄坏，你已经为这买了单。尽可能犯错，但不要在车间那里打闹。
下面谈谈如何成为专业的机械工程师
- 读ISO/ASME/ASTM/ASTC/ASMI 这些标准文件，那才能告诉你理论如何满足实践，如果你的大学没有，投诉他们！跪求、借、偷，用任何方法。想要知道事情如果做的，去读标准，不是在网站或论坛。
- Take/Audit courses on electromagnetism, digital electronics, electrical theory, VLSI/Silicon based designs, electrical machinery. You should be able to design your own motor driver/filter/power regulator/multivibrator circuits and implement them on PCBs. Start dipping into embedded microcontrollers here. This is where you C++ experience should start paying off.
- Signal processing - Audio/image/Power signals - Master the topic of discrete Fourier transforms/spectral densities and how they are used and calculated. Figure out how digital sampling and digital filters work and how filters and masks get designed. Move on to z-transforms and recursive filters. Your statistics background starts to become useful here. At least figure out how to manipulate images using pixel-array math.
- Control systems - THIS ties up everything. And THIS was the topic that really got you into ME. You didn't join ME to make bridges or prepare CAD layouts for GE ovens or tractor engines or boiler chambers for plants or be a grease monkey. You joined ME to make structures that move, intelligently. If you have done things right so far, this is where you will get to have fun. It ties together your dynamics and linear algebra first, then programming, signal processing and statistics next, finally you implement it all using your electronics/embedded skills.
- Instrumentation – People have equipment that costs between a thousand dollars to over several million. You need to learn how to use them, AND how to construct them. You will find that making equipment is always cheaper than buying a turnkey system from a manufacturer. So companies prefer to design/assemble their own systems. This should segue into design of experiments/statistical validation. Your goal should be to know how to hook up the hydraulic pressure gauge in an EMD F51PHI locomotive cab suspended 10 ft up in a shed to an office in Minnesota.
If you want to get into finite elements, you can’t do that in undergrad. All you will learn is to push buttons. Most engineers only think they understand FEA – they actually don’t. It takes practice, study and experience. The pretty pictures don’t mean much by themselves. So I will say go to grad school or intern with a practicing consultant.
That should about cover your basics and get you a good job. But if you want to get a great job, you will need professional degrees or exhibit skills in some of the following. So, on to specialization:
- Fracture/fatigue/materials on the nanoscale.
- MEMS – Look up Sandia National Labs/MEMS. Biggest opportunity for MEs since all companies are moving from RnD to ramping up production right about now. Micromachining and processing technologies research is active as well. MOEMS was hot, sensors are sizzling, actuators not so much, lab-on-chip was meandering about, last I checked. Significant effort underway on determining lifetime/reliability as well. People were excited about energy harvesting, but that seems to be toned down now. Lot’s of material science opportunities.
- Microfluidics – These guys blow bubbles through microchannels! Look up lab-on-a-chip.
- Bioengineering – Tissue printing/engineering! There’s also research on mechanical characterization of bio-materials (bones/ligaments/RBCs)
- Medical devices/robotics – da Vinci/intuitive. Also swallowable robots and cameras. Lots of health monitoring devices and OR assistants.
- Robotics/control systems – Typically, you need to be core CS/EE for this. They are the ones doing most of this research. But you can create opportunities for yourself by choosing to focus on dynamic structure design or kinematics or something on that order. Look up Hod Lipson/Cornell or Red Whittaker/CMU or Marc Raibert/ex CMU/MIT leg labs or Rob Wood/Harvard for inspiration. Google and Amazon have raised this field’s profile over the last couple of years. Look up compliant mechanisms/robots, autonomous vehicles, haptics, telepresence, Raytheon XOS II,... Lot’s of bullshit in the name of ‘assistive robotics’ (that no one can or will want to afford or use, and medicare won’t support).
- Control systems/avionics – I worked on optimizing damage-resilient, real-time coolant distribution through nuclear subs, my ex-boss worked on guidance systems for the Pershing/Hera systems. This is a mature engineering field at the moment (not much RnD) but scope for new applications.
- Thermo research – They do crazy things with combustion, not my domain.
- Nonlinear dynamics – Applied theory, predicting weather(?!), galloping (hopf) systems, .. this field goes on till quantum cryptography and then some.
- Aerospace vehicles – SpaceX. Etc. Vibrations theory, dynamical systems and controls. Your vibrations theory needs to be strongly coupled.
- Infrastructure – Given Keystone or fracking, infrastructure is going to undergo another massive boom.
- Petroleum - …
- FEA – Meshing and geometry algorithms, data compression, rendering are being researched
- Energy – fuel cell research, the cryptozoology equivalent in ME They’ve been at it for a while, but it seems to be a funding generation ploy.
- Marine systems - …
- Theoretical systems – Lots of work on rule based machine learning based design synthesis, structural optimization (back in early 2000’s it was all about simulated annealing and genetic algos, now they call it machine learning), dynamic self modeling, multi-agent systems,
- MAV/Flight dynamics – Concentrated around rotorcraft/flapping wing architectures. Mostly experimental, some theoretical research going on.
- ICE research – Very avoid!
- Tribology - Nonlinear dynamics of rate state dependent friction generate P/S/Love/Rayleigh wave phenomena used to predict earthquakes. Studying hydrodynamic lubrication of journal bearings is a trifle boring compared to that. See Ruina's work at Brown.
I have written this like the "Survival guide for mechanical engineers on the journey to create astonishing engineering". This is written with North-American ADHD undergrads in mind. So I tend to be didactic, and, in the spirit of times, use hyperbole to signify importance (no selfies, however. Much disappoint.). I also abuse education professionals profusely - But that's only my personal experience – all the additional work I had to put in because courses were not designed right, or because a newly hired asst professor was in charge of a particular course that they had no experience in or because the lecturer, originally from Asia, had this distracting accent and circuitous description that just beat about the bush more than I could keep track of or maybe because most of the freshman/sophomore/introductory courses, specially non-core ME courses, are generally fanned out to temp staff/lecturers that generally don't know jackshit about how things are done or don’t care. So you see, personal failing on my part. That's my excuse for the abuse. And there's catharsis involved as well. So I apologize in advance.
I have a BS/AME USC, and MS/MAE, UC system, PhD/ME (and RI+LTI+ECE) CMU. I wasn't a great student during my BS; 2.7 GPA, almost dropped out to be a professional musician. GRE 1600/6.0 happened. I joined the master’s program because I was getting a fellowship & stipend. Programming happened. YouTube happened. OCW video content happened. I worked on projects with all or some of the following labs - LLNL/SNL/LL MIT/NRLMRY/NECSI/SFI through my PhD. For your reference: MS/PhD GPA 3.6/3.8. No money, at the time of graduation. Now making some.
/3 
