What is the RFID inventory solution? Design and implementation elements
A complete RFID inventory programme consisting of three-piece cores:Hardware Layer(labels, readers, antennas, handheld terminals etc. collection devices)Software Layer(intermediate, inventory application, data docking module), andProcess and Isolation Layer(Compatibility with ERP/WMS, inventory of workflows, abnormal mechanisms). It applies on the assumption that there are not many items of a variety, that the cargo is relatively fixed and that the firm's reconciliation accuracy requires a higher level of error than that of the manual inventory.
Hardware selection determines the upper limit of all work left
The most easily underestimated in the RFID inventory is the hardware selection. Once the label and the reader are laid, the cost of later replacement is much higher than the software overlap.
Frequency is the first decision. Low frequency (LF) and high frequency (HF) suit short-range, low-volume reads, such as tool cabinets and archive boxes. Ultra-high frequency (UHF) is the usual choice for inventory counts in warehousing, retail, and manufacturing. Range can run from 1 m to more than 10 m, with hundreds of tags read at a time. Most inventory applications ultimately use UHF.
The label is sealed in a form that matches the label. Metal surfaces are labelled as anti-metallic, liquid-rich environments are designed with special materials or antennas, and outdoor tolerance requires high-level engineering shell selection. When we took stock of plant fixed assets, we were forced to return to work because the general label was not read on the metal cabinet; then the problem was solved by replacing it with an anti-metallic payment with suction material.
The deployment of the reader depends on the mode of operation. Fixed-sized reading and writing machines, with antenna arrays, are suitable for a gateway-to-exit library, door-to-door or ceiling-to-top antenna form a scanning "door", goods are read as soon as they are pushed; handheld terminals or cartage type readers are suitable for inside-of-ware inspection in warehouses, pallet-by-clip points; tunnel-based readers are used to enter and exit water lines or large items. Budget allows for the most flexible combination of fixed + handheld, fixed point-to-place processing of daily access libraries, hand-held blind areas and drawers.
The software level is not a system.
Many people think that the inventory software is "sweeping out the number" and that actually the middle (connecting to a reader and operating system, an intermediate program for data filtering and aggregation) is the most powerful part of the RFID inventory.
The middle is the sort of information that can be used to organize the mass raw data reported in the reader. In one second, thousands of labels are coming in, with double reading, misreading, redundancy, heavy work, filtering, connecting positions, binding documents. The middle is not well done, and the count results seen at the front end are a bunch of random numbers.
The inventory application itself contains several basic modules: the inventory task management (by region, batch, position generation), real-time data display (cleaned, unscanning, abnormal tag real-time statistics), differential processing (auto-generated disk surplus and loss sheets), historical traceability (who, at what time, with what device). These functions are not complex, but it is essential to get through to existing ERP or WMS (storage management systems).
There are three types of integration: intermediate library table interconnection (data exchange with database view, simplest), API interfaces (real time synchronization, most flexible) and file exchange (producer import/export of a file by cycle, suitable for low-frequency scenarios). Most manufacturing clients we've contacted eventually choose the API interconnection because real time and accuracy are more manageable.
It's a four-step walk. Every step has its pit.
The first step is on-site survey and programme design. It requires a clear inventory of the object, frequency, precision, operational motion lines, drawing out a reader deployment map, marking possible blind areas. This step is often compressed, but 80% of the later problems in the actual project are due to poor surveys.
The second step is laboratory testing and pilot validation. Run through processes in small ranges (a shelf, a line), validate label selections, reader layouts, software logic. Pilot time suggests at least two weeks to cover normal operations, miscapsulation, abnormality.
The third step is to provide priority coverage of high-frequency access, high-value items and then to the full volume.
The fourth step is to optimize peacekeeping. The RFID system is not on line, the labels are worn out, the batteries are run down and out of power tags, and regular inspections and retracing are required; the reading machine antenna is subject to vibrations, temperature differentials may be skewed, and read rates are repeated every quarter.
The scheme is worthless. Look at three numbers.
The first one is...Time-consuming count. A manual count in a 5,000 m² warehouse takes six people two to three days. Using handheld RFID readers, two people can cover the same area in half a day. The difference is clear.
The second one is...Actual variance rate. Manual inventory counts typically show a 2%–5% discrepancy rate. RFID combined with standardized processes can bring this below 0.5%. For high-value items, every percentage point of improvement shortens payback.
The third one is...Data availability. The manual inventory is regular and static; the RFID allows for real-time physical access to each item, which directly facilitates procurement planning, security stock, and turnover rate analysis. This part of the value is not intuitive but tends to be most significant over the long term.
Some lessons for preparing the RFID inventory.
Do not be confused by the parameter "reading a distance of 10 metres". Laboratory data do not represent an on-site effect, environmental factors eat more than half of their performance and budgets are redundant.
Do not judge tag costs by unit price alone. Ordinary tags cost RMB 1 or 2, but industrial-grade tags for metal surfaces, high temperatures, or outdoor use can cost RMB 5 or 10. Break down the total by application rather than buying the same tag for everything.
The accuracy of the inventory is a process, not an equipment problem. Regulates the position of the paste, harmonizes the location code and identifies the abnormal processes, which are much more influenced by these "soft" inputs than by hardware upgrades.
Pilots are done before the big ones. The factory is built as soon as they get up, and eight out of ten projects will be back in between. The pilot project is contained within 15% of total inputs, running around again.
The RFID inventory is not a system that can be used for hardware, but a system of selection, deployment, process to delivery. When you're doing projects like Shanghai-based, you can first take the client's inventory, your operations are frequent, your precision requires in order to be fully consumed, and then give a phased implementation proposal, avoiding overloading, and making the system really useful. If you're evaluating a similar program, you can go from high frequency, high value, more dynamic objects, piloting and replicating, which is the most price-effective path.

Ms. Lei
Mr. Pang