结论：抽检要能发现什么问题，取决于抽多少、测什么
采购方在收货时做验收抽检，目的是确认交付的产品与合同约定一致、且质量可接受。
但抽检的能力是有限的： 抽一台测几项，只能发现比较明显的问题，发现不了低比例的缺陷，也发现不了长期使用后才暴露的问题。
所以抽检设计的关键是：明确这次抽检要发现什么问题，据此确定抽多少、测什么。 目的不明确的抽检，往往变成走过场——抽一台看看外观，签字收货。
抽检的目的
验收抽检通常服务于几个目的：
确认货物与合同一致。 型号、配置、数量、外观。
确认基本功能正常。 产品能用。
确认关键参数符合。 与招标或合同约定的参数。
发现批次性问题。 整批都存在的缺陷。
形成质量记录。 作为后续追责的依据。
「发现批次性问题」是抽检最能发挥作用的地方。 如果某个缺陷是批次性的（工艺问题、材料问题、设计问题），抽检很容易发现；如果是偶发的个别缺陷，小样本抽检发现的概率就很低。
抽样方案的考虑
批量大小。 交付的数量。
可接受的质量水平。 允许多大比例的不合格。
风险的严重程度。 缺陷可能造成的后果。
抽检的成本。 包括破坏性检验的损失。
历史质量记录。 供应商以往的表现可以作为调整依据。
建议的做法是： 按批量和可接受质量水平确定基本的样本量；对安全相关的项目适当增加；对历史表现好的供应商可以适当放宽，表现差的收紧。
「按历史表现调整」是有效的做法，它既减少了对可靠供应商的重复检验成本，也对质量不稳定的供应商形成压力。
项目的选择
验收阶段能做的项目有限，选择的原则：
| 项目类型 | 是否适合验收阶段 |
|---|---|
| 外观与标识 | 适合，成本低，每台可做 |
| 尺寸与质量 | 适合，可快速测量 |
| 基本功能 | 适合，操作验证 |
| 配置核对 | 适合，对照清单 |
| 静态强度 | 破坏性，抽样少量 |
| 疲劳耐久 | 周期长，通常不在验收阶段 |
| 阻燃 | 破坏性，取样检测 |
| 电气安全 | 部分项目可现场做 |
建议的组合是：全数做外观、标识、配置核对和基本功能；少量抽样做需要设备的项目；破坏性项目按风险决定是否做。
「全数做基本项目」的成本很低但效果明显。 外观缺陷、配置错误、功能异常，这些问题全数检查就能发现，而它们恰恰是交付中最常出现的问题。
现场可做的检查
不需要专业设备就能做的检查：
外观。 划伤、变形、装配间隙、涂层缺陷。
标识。 铭牌信息、警示标识、载重标识是否齐全清晰。
配置。 随机附件、备件、说明书是否齐全。
功能。 折叠展开、制动、调节机构、电动功能。
装配。 紧固件是否紧固、活动部件是否顺畅。
尺寸。 关键尺寸的简易测量。
稳定性。 简易的倾斜检查。
这些检查加起来每台几分钟，但能发现交付问题中的大部分。
与检测报告的配合
验收抽检不能替代检测报告，两者的作用不同：
检测报告。 证明产品的设计和型式符合要求，覆盖面广但针对的是型式样品。
验收抽检。 确认实际交付的批次与报告中的产品一致、质量可接受。
两者结合才完整。 只看报告不做抽检，无法确认交付的就是报告中的产品；只做抽检不看报告，无法确认产品的设计和性能符合要求。
建议在验收时核对： 交付产品的型号、铭牌信息、外观特征是否与报告中描述的一致。
不合格的处理
判定规则。 事先约定：抽检发现几个不合格算整批不合格。
复检规则。 是否允许加倍抽样复检。
处理方式。 退货、换货、返修、让步接收。
让步接收的条件。 什么情况下可以让步、需要什么审批。
记录与追溯。 不合格的记录、处理过程、最终结论。
「让步接收」要谨慎。 安全相关的不合格不应让步；非安全相关的轻微缺陷可以在约定条件下让步，但要有记录。
长期的质量跟踪
单次抽检的信息有限，长期的跟踪更有价值：
记录每批的抽检结果。 形成历史数据。
趋势分析。 质量是稳定、改善还是劣化。
与使用中的问题关联。 抽检发现的问题与后续的故障是否相关。
反馈给供应商。 让供应商知道问题并改进。
调整抽检方案。 根据历史表现调整样本量和项目。
「趋势分析」能发现单次抽检看不出的问题。 每批都合格但某个参数在逐步劣化，说明供应商的工艺在漂移。
与合同条款的配合
验收抽检的效力来自合同约定，相关条款应当明确：
抽检的权利。 采购方有权抽检。
抽检的方式。 样本量、项目、方法。
费用的承担。 检测费用由谁承担，不合格时是否由供方承担。
破坏性检验的处理。 破坏的样品如何计价。
判定的效力。 抽检结论的约束力。
争议的解决。 双方对结论有分歧时的处理，比如由第三方复检。
「破坏性检验的样品处理」建议事先约定，否则实际操作中会有分歧——做强度试验的样品已经损坏，这台算不算交付数量。
验收记录的内容
建议记录：交付批次信息；抽样方式与样本标识；检验项目与结果；判定结论；不合格的处理；以及双方签认。这份记录是质量追溯和责任认定的依据。
我们的做法
如果采购方需要在验收阶段做超出现场能力的检测，我们可以按约定的抽样方案承接——抽取样品做实验室检测，出具报告作为验收依据。
项目选择上，我们建议优先做能反映批次质量的项目，比如关键尺寸、静态强度、以及与合同参数对应的性能项 ——这些项目的结果能直接支撑验收判定。
如果你在设计验收抽检方案，或者需要验收阶段的检测支持，可以把产品和合同参数发过来一起讨论，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，标准信息见标准查询。
English version
Conclusion. Acceptance sampling at delivery aims to confirm that what arrived matches the contract and is of acceptable quality. Its power is limited: testing a few items on a few characteristics reveals fairly obvious problems while missing low-rate defects and anything that only emerges after prolonged use. The key to designing it is therefore stating what this inspection is meant to find and setting sample size and test items accordingly. Inspection without a stated purpose tends to become a formality: look at one unit, check the appearance, sign for the delivery.
Purposes of acceptance inspection. Confirming that goods match the contract in model, configuration, quantity and appearance. Confirming that basic functions work. Confirming that key parameters conform to what was tendered or contracted. Detecting batch-wide problems present across the delivery. And creating a quality record as a basis for later recourse. Detecting batch-wide problems is where sampling works best: a defect arising from process, material or design affects the whole batch and is easily found, while a sporadic individual defect has a low probability of appearing in a small sample.
Designing the sampling plan. Consider batch size. Consider the acceptable quality level, meaning what proportion of nonconformity is tolerable. Consider severity, meaning the consequences a defect could cause. Consider the cost of inspection, including losses from destructive testing. And consider historical quality records, since a supplier's past performance justifies adjustment. In practice: set a base sample size from batch size and acceptable quality level; increase it for safety-related characteristics; and relax somewhat for suppliers with good history while tightening for those with poor. Adjusting by history works well, reducing duplicated inspection cost for reliable suppliers while applying pressure to inconsistent ones.
Choosing test items. What can be done at acceptance is limited. Appearance and marking suit the stage, costing little and feasible on every unit. Dimensions and mass suit it, being quick to measure. Basic function suits it, verified by operation. Configuration checking suits it, against a list. Static strength is destructive and suits a small sample. Fatigue durability takes too long and generally does not belong at acceptance. Flammability is destructive and requires sampling for laboratory testing. And some electrical safety items can be done on site. The recommended combination inspects appearance, marking, configuration and basic function on every unit; samples a small number for items needing equipment; and decides destructive items by risk. Inspecting basic items on every unit costs little and achieves a great deal: appearance defects, configuration errors and functional faults are exactly the problems that most often arise in delivery, and full inspection finds them.
Checks feasible on site. Appearance, covering scratches, distortion, assembly gaps and coating defects. Marking, covering whether nameplate information, warnings and load markings are present and legible. Configuration, covering whether accessories, spares and instructions are complete. Function, covering folding, braking, adjustment mechanisms and powered functions. Assembly, covering fastener tightness and smoothness of moving parts. Dimensions, through simple measurement of key values. And stability, through a simple tilt check. Together these take a few minutes per unit and catch most delivery problems.
Relationship to test reports. Acceptance inspection does not replace test reports; the two do different things. A test report demonstrates that the design and type meet requirements, covering broadly while addressing a type sample. Acceptance inspection confirms that the batch actually delivered matches that product and is of acceptable quality. Both are needed: reports alone cannot confirm that what arrived is what was tested, and inspection alone cannot confirm that design and performance meet requirements. At acceptance, check that model, nameplate information and physical characteristics of the delivered product match those described in the reports.
Handling nonconformity. Agree judgement rules in advance, stating how many nonconforming units reject the batch. Agree re-inspection rules, including whether double sampling is permitted. Agree handling, whether return, replacement, repair or concession. Agree conditions for concession, covering when it is permissible and what approval it needs. And record everything, covering the nonconformity, its handling and the conclusion. Treat concession cautiously: safety-related nonconformity should not be conceded, while minor non-safety defects may be under agreed conditions with a record.
Long-term quality tracking. A single inspection yields limited information, and tracking over time is more valuable. Record each batch's results to build history. Analyse trends to see whether quality is stable, improving or deteriorating. Relate inspection findings to problems in service. Feed back to suppliers so that they can improve. And adjust the sampling plan according to history. Trend analysis reveals what single inspections cannot: every batch passing while one parameter drifts steadily indicates that the supplier's process is moving.
How we handle it. Where a purchaser needs testing beyond what can be done on site at acceptance, we can work to an agreed sampling plan, drawing samples for laboratory testing and issuing reports as an acceptance basis. On item selection we suggest prioritising items reflecting batch quality, such as key dimensions, static strength and the performance items corresponding to contract parameters, since their results support acceptance decisions directly.
Send us the product and contract parameters and we will work through the plan. Phone or WeChat: +86 132 4819 8029.