How do RFID take an inventory of assets land? On-site operations synchronize with data
Industry News

How do RFID take an inventory of assets land? On-site operations synchronize with data

PublishedShanghai Zejie Smart Technology Co., Ltd.

A Warehouse at 3 A.M.: The First RFID Stocktake

Last winter we worked on an inventory overhaul at an automotive parts factory. The warehouse held nearly 30,000 pallets of parts, each with a barcode label. On inventory day, four workers with handheld terminals worked from 8 a.m. to 3 a.m. the next morning, and the captured records still had to be checked one by one. The next day the warehouse supervisor showed us the report: “More than 300 barcodes couldn't be read. Either they were worn or the labels were in the wrong place. Is there a way to do this without working all night?”

And then we re-checked the inventory with RFID, which basically puts a small chip on each item that can "auto-response" -- and the reader doesn't have to be singled out, and dozens of pieces can be identified -- and the same size of inventory, from entry to data uploading ERP. And the next thing you know, just four hours, and then we follow this project, how to get the RFIDs in the asset scene, how to synchronize the data, and make it clear.

First, look at the pain: where is the slow bar-coded count?

Many asset managers first come to RFID and ask, "How much better is this thing than a sweep?" Rather than a parameter, we should see where the card is.

The traditional bar-coded discs have three indissociable time points: one is a piece-by-case, short reading and writing distance, and the bar code is slightly shielded or worn; two is a move or box to read back, lower label; and three is that data are all aggregated manually, misread, missed, reread after a very high cost.

A customer who made the power equipment had given us a set of data: about 12,000 items in the vault, with a planned inventory cycle of two weeks, which was often dragged to more than three weeks for repeated clean-up; the first test run after RFID has been modified, completed full reading in four hours, with an identification rate of 96.8 per cent, and the remaining 3.2 per cent concentrated on areas where metal surfaces were attached -- that is exactly what RFID needs to design.

So, the first step of RFID's landing is not to rush into the purchase of equipment, but to figure out where the current process is stuck: are there multiple types of asset that label it that cause reading disruption? Are there more metals that cause signal decay or are large in storage areas and people move in a very high number of places? The problem is different, the options are quite different.

II. The four key elements of the field operation

1. Label selection and attachment position

The RFID labels are split by frequency, UHF, low frequency, and the majority of asset counts are based on UHF (UHF) because of distance from reading, and mass recognition is strong. But metal assets reflect and shield signals, normal paper labels cannot be read directly on the cabinet or server, and need to be "anti-metal label" - an internal suction layer envelope.

The position attached is described as follows: Priority is given to locations that are not easily hit or frictioned, such as the side of the cabinet, the bottom of the equipment, the urchin within the tray; the channel that moves up to allow the label to move on a regular basis against the operator; and as uniform as possible in the direction of the batch, so as to avoid going down at a moment. We have encountered a room cabinet, which is well read inside the front door, which falls below 40 per cent because it is covered by the metal floor.

2. Readers and antenna deployment

There are two main ways of doing this:

In a data centre room project we've done, 460 cabinets are set in four rows, with a channel space of 1.2 metres. If read from the top of a single antenna, it will be blocked by the next-door cabinet; after changing to "top antenna + end-direction antenna" combination, the single-discretion pass rate rises to 98.5 per cent.

3. Inventory operational rhythm

It is not the equipment that is most easily used to step on a pit.

  1. Predisposal phase:The 24-hour freeze of access was unnecessary, leaving only the inventory team at the scene.
  2. Initial roll-out phase:Readers follow the planning path and record all the marked EPC codes read (electronic product code, equivalent to the only ID number for each asset).
  3. Rewind phase:The initial results are compared to the asset desk accounts in the system and the discrepancies are located at specific locations for manual review.

The heart of the rhythm is "Fast start, double drive" instead of a full-scale high-intensity scan -- mechanically repeated scanning will make missing entries inadvertently overlooked.

4. Treatment of unusual assets

No RFID solution can guarantee a 100% identification rate. There are three common types of exception:

Data synchronization: "Live the business meaning" to read

Being able to read a tag does not make the data usable. An RFID reader returns an EPC string such as 'E280-1170-6000-0200-A1B2-C34D.' That means nothing to the business system without three layers of conversion.

First tier: map of the EPC to asset number

Maintain a relationship table for the EPC code -- asset number -- asset name -- department. The map relationship is completed once on the line, and all operations follow it. The map table is consistent with the ERP or EAM master data, and is synchronized every day as necessary.

Second level: Map of physical position to warehouse location

The inventory results are "What assets are in the vault" not enough, so answer "what's in which vault?" It is suggested that a bit antenna or geolabel should be placed at the warehouse level to position the library roughly by RSSI signal strength (receiving signals intensity indicators, which can be roughly judged closer to which antenna); high-precision scenarios superbity coding with UWB or visual recognition. This part of our cooperation with Hypetech, they suggest that RFID's location position location accuracy be kept at the level of the bank, without having to pursue centimetres, cost and complexity.

Third level: offline data interface with online systems

The warehouse network environment is often unstable, especially in large plant areas, underground storage. Readers must have enough local caches, and industrial handheld terminals must support at least 100,000 records stored offline. Once the network is restored, it is sent back to the intermediate library (usually MQTT or HTTP interface), then pushed from the intermediate to ERP or WMS (storage management system).

Reverted package fields suggest that they be fixed, at least: A round count batch number, collection time, EPC code, reader number, signal strength, library-id number. These five fields are not possible, otherwise data "clip neck" will be found when subsequent trend analysis is done.

IV. VALANCES AND RECOMMENDATIONS

Back to the first gas-packing plant. After three months of retrofitting, the inventory cycle was reduced from about two weeks to half a day; the difference in the accounts was reduced from around 2.4 per cent to less than 0.3 per cent; and the number of people counted down from four to two. The strongest feedback on the factory side was not how much time it saved, but "the bill is right" -- the hundreds of "disappeared" bar codes that were the most painful for the monthly reconciliation, and now the system will tell you which equipment labels need to be reposted.

If you are considering the inventory of assets of the landed RFID, three specific recommendations will be made:

  1. Do a local pilot first.Run a three-month pilot in one warehouse zone or with one asset category. Get tagging, reading, and data transmission working end to end before deciding whether to roll it out more broadly.
  2. Sets the data format ahead.The rules for the EPC mapping, offline cache format and back-over interface fields were established at the outset to avoid later fighting over several systems.
  3. Don't pursue 100% recognition.Acceptance of the anomaly rate within 3 per cent and its use through a double-disk mechanism is much less costly than hard recognition.

The RFID inventory is not a procurement, but rather a project. Its value is not in the reader itself, but in a closed link linking on-site operations and data governance - reading is only the starting point, and business implications are realized.