RFID technology analysis and application scenario
When discussing RFID applications, it is easier to name asset management, storage logistics, production backsliding, and so on, without indicating which step the system is taking. A real technical analysis should tear a scene down to a verifiable task: what to identify, at what point, what to do, and how much error is allowed.
Asking business questions, choosing how to identify them.
“Quick Know What's In A Carton” “Preve to Who You Hand a Tool” “Find the Goods Out”. They have three tasks. They have different reading and writing ranges, and they confirm that actions and anomalies are being handled differently.
RFID is good at identifying objects by wireless means, supporting a reduction in the number of scans on a case-by-case basis with appropriate configuration. It does not automatically understand the attribution of the object, the direction to which it is going, the business authorization or the intent on the ground. Wireless reading should be seen as input evidence, and then the process will determine what it means.
Three levels, two different questions.
| Level | Issues to be addressed | Common miscalculation |
|---|---|---|
| Tabs and Wireless Read | Whether the designated object can be identified in the target environment | I'm gonna finish my job when I read the number. |
| On-site incident recognition | Which post, direction and mission is this observation for? | Make a serial reading of the neighborhood as real movement. |
| Operations systems processing | Whether to allow a change of inventory, person responsible or status | We'll change the official account as soon as we get the data. |
These three dimensions also correspond to three types of debugging. Increased launch power may improve some of the errors, but it does not solve the misdeductive deductions in stockpiles after mission cancellation; or re-interfaceing will not remedy the always mislabelling of labels. The project should first locate where failure occurs and avoid dealing with different problems over and over again with the same means.
Asset inventory: focus on unidentified objects rather than scanned quantities
Asset counts must compare reads against the list of assets expected on site. Reading the same tag 100 times still confirms only one asset, and reading a tag in the next room does not mean the current room's stock matches the records. A useful interface separates confirmed assets, expected but not found assets, unexpected locations, and unknown tags.
If the management objective is to track responsibility, then you should also separate inventory observation from access, transfer, and end-of-life records, otherwise a single count may inadvertently change liability.
Logistics and production: frequent reading, but incident need to be restrained
The box may be read over and over again while you are at the door. The system needs to recognize a valid pass, consider a return, walk in parallel and read in a series of adjacent positions. Original reading records can be saved for searching, and formal entry or process completion events should be generated by clear rules, avoiding an increase in the number of numbers per receipt.
Production also requires a retrospective verification of the relationship between the item identification and the assembly, batch, worksheet. When the label is placed on a reusable load, the package number cannot be permanently equal to one of the items. Replacement, de-separating and re-entry binding rules are missing, and automatic collection only results in an unexplainable resume more quickly.
Check if the project is worth continuing with four errors
It is recommended that the statistics be omitted, misread, mislinked and event delayed. The missing is not confirmed; the error is not related to the object entering the result; the error is a genuine numbering but mismatching task; the delay may render the correct information useless. They are different in cost and processing, and it is inappropriate to press a vague "accuracy rate."
Pilots should include failed processing. If the network is disconnected, how to continue working, how to change the label when it is broken, and how to reverse a job before it is extended. A system that has a normal process quickly, but no place for an abnormal process, may have a higher long-term running cost than the original manual method.
The maturity of the RFID programme ultimately depends on whether it makes a particular business judgement more reliable. If the package-by-case sweep is fast enough, if the number of items is limited and manual confirmation cannot be omitted, it may be more appropriate to optimize the bar code process; if the main cost comes from a large number of duplicate identifications and reconciliations, RFID will have better chances of working.

Ms. Lei
Mr. Pang