How does RFID enter the inventory count fit?
Industry News

How does RFID enter the inventory count fit?

PublishedShanghai Zejie Smart Technology Co., Ltd.

Core conclusion: Mark and data entry completed at receipt, RFID-based bulk read at inventory count, key to connection is continuous data flow

The RFID is used to enter the inventory, and the most easily peddled is not the device itself, but the data from the receipt and inventory are not linked. Our experience is that: when you mark the electronic label at the receiving point, SKU (the smallest stock unit) is written, data is entered into WMS (storage management system); when you reach the count, whether you walk through the shelf with a mobile handheld or set a fixed reading machine on the portal, you can read it directly, without having to enter it twice. After this process runs in line, the single SKU count time can be reduced from dozens of seconds to one or two seconds, and the whole library difference rate can be contained within one thousand.

Why would RFID link the library to the count?

The biggest drawback of traditional barcode counts is that you must aim at and scan each barcode individually. A warehouse with 10,000 boxes takes two or three days to count with a handheld barcode scanner, and it is easy to miss or misread items. RFID (radio-frequency identification) tags operate in the UHF (ultra-high frequency, 860–960 MHz) band, typically read from 3–8 m away, and can be read hundreds at a time without individual alignment. That turns receiving and stocktaking from two separate tasks into a continuous data flow.

We actually had access to a warehouse where the car parts were made, and before getting on line with RFID, we could not start entering the system until the next morning after receiving the goods, and the count was set for half a day. Now when the fixed reader at the gate is in receipt, the label information automatically enters WMS; when the inventory is taken, the staff moves the mobile handheld around the shelf, reading all of them in thousands of SKUs within half an hour. What's behind this is that two links share the same label code and data format.

Receipt chain: mark, write and verify three steps

The first step is to mark the position.

The UHF label is sensitive to metals and liquids, and if it is attached directly to the surface of an iron box or a barrel, reading distance falls from 3-5 metres to less than 1 metre. It is common to put it on the side of an outer container, 10 centimetres above the top; if the metal is in the position (e.g., the shelf is made of steel), you can add a bubble sheet between the label and the metal, which is more stable at 5 mm thickness. We have tested that the reading distance after the gasket can be recovered from 0.5 metres to about 3 m.

Step two is tag data writing.

The EPC region (Electronic Production Code, Electronic Product Code) has three messages to write before the label enters the warehouse: Vendor code, Commodity code, Serial number. It is recommended that the EPC code length be fixed at 96 bits, distributed by a structure such as "3-bit manufacturer + 4-bit class + 5-bit batch + 8-bit flow" and facilitate subsequent WMS resolution. Writing with a desktop launcher is simple, speeds 2-3 units per second, and can be done with a printmarker, so that you can print side-coding code and save steps for individual writing.

Step three is the acceptance check.

After the truck has reached the platform, it is first read in bulk through the RFID doorbar Zreader and compares with the purchase documents. This link is to be checked for the location of the door-barlock: 1.5-2 metres between two antennas, which are at the middle height (normal 1.2-1.5 m), and the unreadable area is often the top and bottom, where a super-sky blinding is added to the doorbar. The unreadable boxes are to be individually manually cleared, so that they do not "like to have been cleaned but not actually swept."

Inventory: two methods of moving handheld and fixed channel

Move handheld point for bulk slots

In shelving areas, high-bay storage, and mezzanine racks where fixed antennas are impractical, use mobile RFID handhelds, which look like tablets with a handle, and scan as you move along the racks. Staff should walk at a steady pace and pause 3–5 seconds at each row so the built-in antenna can pick up the tags. Handheld readers typically capture 50–100 tags per second, allowing an initial count of a location with 10,000 tags within two hours.

Fixed access points fit the dense view of access to the library

If the warehouse is in and out of a large number of traffic, it can be equipped with a set of orientation antennas on each shelf alley to read this row. The system automatically triggers reading in the order in which the staff are not required to move at the time of the count. Shanghai-based Selectory Technologies has cross-checked data from fixed channels that are read over and over and over again with buried pressure sensors (the heavy induction devices installed on the shelves), and the difference will be further reduced.

Discrepancies are closed.

Regardless of how the readings were collected, compare the results with the inventory recorded in the WMS. If the discrepancy exceeds the threshold, such as a quantity difference of more than 0.5% or more than five units for a single SKU, the system automatically freezes the storage location and triggers a recount. Recounts use both manual and RFID verification. If necessary, open the outer carton to check the smallest package units. Without a process for resolving discrepancies, they will accumulate until system quantities bear little resemblance to physical stock.

Three key points of the two links

Uniform coding rules are a prerequisite

The tag code that you write when you receive the goods must be identical to the format read at the time of the inventory, otherwise WMS cannot resolve or match the error. Once the code rules are set, they cannot be changed halfway through every link in the warehouse -- receipt, up, move, count, exit -- and follow this set of rules. It is recommended that the code rules be archived as documents, and the first lesson for new staff training is to speak about it.

Data interfaces need to be accessed.

The original data collected by RFID readers is usually transmitted to the middle (RFID intermediate) via TCP/IP (local area network transfer protocol) or serial (a hardware direct-link communication interface), which aggregates, filters, and removes weights from multiple readers, then is pushed from the middle to WMS. Once this interface is delayed or dropped, the receipt of the data and the count data are disconnected. It is common practice to store local primary data for at least 7 days in the middle, and WMS can be re-set from the middle when it comes to a problem, so that the data is not lost.

The abnormal label needs a special process.

The actual run, about 10% of the labels are read out or wrong, often for reasons such as: tabs are blocked, sticker position is biased, tag itself corrupted, reader power set inappropriate. These anomalies cannot simply "jump" out, creating an anomaly pool (the account where the problem label is specifically recorded), each recording contains: tag EPC code, unusual type, discovery time, processor, handle result. At the end of the month, statistics on the distribution of unusual types are being made to improve the situation with regard to high-prevalence problems, such as adjusting the label position and using anti-metal labels.

A few pits easy to step on.

The first mistake is doing too much at the start and trying to switch every storage location to RFID at once. Pilot one category for one to two months to establish the workflow before expanding. The second is setting reader power too high on the assumption that longer range is always better. Excessive power can cause stray reads from adjacent shelves and interference between tags. Generally start at 20–26 dBm (decibel-milliwatts, a measure of RF transmit power) and tune on site with antenna gain in mind. The third mistake is neglecting training. Minor issues such as skewed tag placement or holding the handheld incorrectly can hurt results when warehouse staff are unfamiliar with RFID. Provide at least four hours of training, including a practical assessment.

Another common error is that it looks at one-time input costs. The RFID labels are now in the range of a few to two dollars (see if they are resistant to metals, whether they are printed) and the readers are from several thousand to tens of thousands of dollars, counting the single-pool modifications that are invested in hundreds of thousands to hundreds of thousands of dollars. But the gains are mainly in terms of human savings and accuracy, which alone, more than hardware costs, will lead to "too expensive" conclusions, with a total of 3-5 years.

Proposals for action

If your warehouse receives more than 500 boxes per day, SKU is over 2000 and RFID is usually the better input output. It is suggested to start with these three actions: first set a uniform code and fall into WMS field; select a group of goods with a steady volume for a small pilot of 2,000 labels; and then put the reader, medium, WMS tri- interface file combo and then start work. After these three steps, the technical risks of the entire retrofit are basically manageable.