结论：噪声和振动往往是结构问题的外在信号

噪声与振动在多数轮椅标准里不是强制考察项，所以常被当成体验层面的小事。但从工程角度看，它们经常是结构问题的早期外在表现：某处配合变松、某个部件共振、某个轴承开始磨损，最先被察觉到的往往是声音和抖动。

所以把它们纳入评价，除了改善体验，还有一个附加价值：提前发现潜在的可靠性问题。

来源分类

类型 典型来源 关联的潜在问题
结构振动 车架共振、部件松动 连接件疲劳、螺栓松脱
传动噪声 齿轮啮合、皮带、链条 装配精度、润滑状态
电机噪声 电磁噪声、轴承噪声 电机选型、轴承寿命
滚动噪声 轮胎与地面 轮胎材质、轮子对中
机构异响 折叠、调节机构 配合间隙、磨损
路面激励 颠簸传递到座椅 悬挂设计、坐垫吸振

第五行在售后反馈里占比不低。 异响本身不一定影响功能，但使用者会把它当作质量问题，而且它往往确实预示着某处配合已经变化。

评价方法怎么选

噪声和振动的评价有不同的深度，按需要选择：

主观评价。 由若干评价者按统一场景试用并打分。成本低，贴近真实感受，但重复性差、无法定位原因。

声压级测量。 在规定位置测整体噪声水平。有客观数值，可以做产品间比较，但不能区分噪声来源。

频谱分析。 测量噪声或振动的频率成分。能定位来源——齿轮啮合频率、电机转频、结构固有频率各有特征，这是排查问题的有效手段。

振动传递测量。 在座椅界面测量传递到使用者的振动。这与乘坐舒适直接相关。

建议的组合是：主观评价发现问题，频谱分析定位原因。 只做主观评价，知道有问题不知道在哪；只做客观测量，可能测了一堆数据但与实际体验对不上。

把主观感受转成可测指标

使用者反馈往往是模糊的：「跑起来有点吵」「过坎震得厉害」「转弯时有异响」。要处理这些反馈，需要把它们转成可测的东西。

做法是先复现，再定位，最后量化：

复现是指找到反馈描述的具体工况——多快的速度、什么路面、什么动作。这一步做不到，后面都无从谈起。

定位是在复现的工况下测量，用频谱找出主要成分对应哪个部件。

量化是定出一个可以判定的指标：在该工况下某频段的幅值不超过某值。有了这个指标，改进效果才能验证。

与结构问题的关联

几种典型的对应关系值得记住：

特定速度下才出现的振动，通常指向共振——该速度下的激励频率与某部件的固有频率接近。解决方向是改变固有频率（改刚度或质量）或者避开该转速。

随使用时间逐渐出现的异响，通常指向磨损或松动。这类问题会持续恶化，应当追查具体部位。

只在特定动作时出现的响声，指向该动作涉及的机构。

噪声随载荷增大而明显增大，可能是结构变形导致配合改变。

设计与验证的衔接

如果产品把安静平稳作为卖点，那么这项指标应当进设计输入，而不是等做出来再评价。具体做法包括：在设计阶段做模态分析，避开常见激励频率；选型时把噪声指标纳入部件评价；在样机阶段做一次基线测量，作为后续改进的对照。

基线数据的价值容易被低估。 没有基线，后面改了之后无法判断是真的改善还是感觉上的差异。

耐久之后的复测

噪声与振动会随使用变化，所以耐久试验之后复测一次很有意义。对比前后的频谱，能看出哪个部件的状态变化最明显。

这项数据对确定维护周期有直接帮助：如果某个频率成分在耐久后显著增强，对应部件就是维护重点。

测量条件的控制

噪声测量对环境要求较高。背景噪声、反射面、测点位置都会影响结果，所以测量条件必须固定并记录。

实际操作中，如果条件无法达到理想状态（比如没有专门的声学环境），可以采用相对比较的方式：同一条件下比较改进前后，或者比较不同型号。相对比较对环境要求低得多，而多数改进工作需要的正是相对比较。

绝对数值的测量则需要更严格的条件，如果要对外公布噪声指标，应当在合适的声学环境中测量。

使用者感受的个体差异

对噪声和振动的敏感度因人而异，年龄、听力状况、使用时长都有影响。所以主观评价应当有足够的样本量，而不是一两个人试用就下结论。

评价者的构成也要贴近目标使用者。 由年轻工程师做主观评价，结论未必适用于主要使用人群。

改进措施的验证

针对噪声振动做的改进，效果要用同样的方法验证。常见的改进包括：增加阻尼材料、调整结构刚度改变固有频率、提高装配精度、更换部件、增加隔振元件。

验证时注意两点：用改进前的基线数据做对照，条件保持一致；同时确认改进没有带来新问题，比如加了阻尼材料后散热变差、增加隔振后操控响应变迟钝。

与产品定位的关系

噪声振动指标的目标值应当与产品定位匹配。面向医院和养老机构的产品，夜间使用较多，对噪声更敏感；户外型产品的使用环境本身噪声就大，优先级可以低一些。

把资源投在使用者真正在意的地方，比全面追求指标更有效。这需要先了解目标使用者的真实关注点，而不是按工程师的偏好设定。

我们的做法

做这类评价时，我们会先与委托方确认关注的工况和使用者反馈的具体描述，再设计测量方案。没有明确工况就开测，容易测出一堆无关数据。

对于有异响投诉的产品，建议做耐久前后的对比测量，通过频谱变化定位磨损部位。这比拆机逐个排查效率高。

有需要可以把产品结构和反馈描述发过来一起定方案，或者直接联系：132 4819 8029。检测能力见服务介绍，产品分类见手动轮椅检测与电动轮椅检测，案例见案例。

English version

Conclusion. Noise and vibration are not mandatory items in most wheelchair standards and are often treated as matters of comfort. From an engineering view they are frequently the outward sign of a structural problem: a joint loosening, a component resonating, a bearing beginning to wear. Sound and shake are usually what people notice first. Including them in evaluation therefore does more than improve the experience; it surfaces reliability problems early.

Sources. Structural vibration comes from frame resonance or loose components and relates to joint fatigue and bolt loosening. Transmission noise comes from gear meshing, belts or chains and relates to assembly accuracy and lubrication. Motor noise comes from electromagnetic effects and bearings and relates to motor selection and bearing life. Rolling noise comes from tyre and surface interaction and relates to tyre compound and wheel alignment. Mechanism noise comes from folding and adjustment mechanisms and relates to clearance and wear. Road excitation transmitted to the seat relates to suspension design and cushion damping.

Mechanism noise features heavily in field feedback. It may not affect function, but users read it as a quality defect, and it usually does indicate that a fit has changed.

Choosing a method. Subjective evaluation, with several assessors scoring under a common scenario, is cheap and close to real perception but poorly repeatable and cannot localise a cause. Sound pressure measurement at defined positions gives an objective figure suitable for comparison but does not separate sources. Spectral analysis identifies frequency content and can localise a source, since gear mesh, motor rotation and structural natural frequencies each have characteristic signatures. Vibration transmission measured at the seat interface relates directly to ride comfort. The useful combination is subjective evaluation to find the problem and spectral analysis to locate it. Subjective work alone tells you something is wrong without saying where; objective measurement alone may generate data that does not correspond to what users actually notice.

Turning subjective reports into measurable indicators. Feedback arrives vague: noisy when moving, harsh over bumps, a click when turning. Reproduce, localise, then quantify. Reproduction means identifying the specific condition described, the speed, the surface, the action; without it nothing further is possible. Localisation means measuring under that condition and using the spectrum to match content to a component. Quantification means defining a judgeable indicator, such as an amplitude limit in a frequency band under that condition, so that improvements can be verified.

Links to structural problems. Vibration appearing only at a particular speed generally indicates resonance, where the excitation frequency approaches a component's natural frequency; the remedy is to shift the natural frequency through stiffness or mass, or to avoid that operating point. Noise that develops gradually with use indicates wear or loosening, will continue to worsen, and should be traced to a specific location. Noise occurring only during a particular action points to the mechanism involved. Noise rising sharply with load may indicate that structural deflection is changing a fit.

Connecting design and verification. Where quiet, smooth operation is a selling point, it belongs in the design inputs rather than being assessed after the fact. Perform modal analysis during design to avoid common excitation frequencies, include noise in component selection criteria, and take a baseline measurement at prototype stage as a reference for later work. The value of a baseline is easily underestimated: without one, later changes cannot be distinguished from perceived improvement.

Retesting after endurance. Noise and vibration change with use, so a repeat measurement after endurance testing is informative. Comparing spectra shows which component's condition has changed most, which supports setting maintenance intervals: a frequency component that grows markedly identifies the part to prioritise.

How we handle it. We agree the conditions of interest and the specific wording of user feedback before designing the measurement, because testing without a defined condition tends to produce a great deal of irrelevant data. For products with noise complaints we suggest before-and-after measurement around endurance, locating the worn part through spectral change, which is faster than dismantling and inspecting component by component.

Send us the structure and the feedback descriptions and we will design the approach. Phone or WeChat: +86 132 4819 8029.