After distributed photovoltaic projects reach a certain scale, manufacturers often find that the real management challenge is not a single inverter, but the growing number of devices within a single site.

Photovoltaic inverters, energy storage PCS, BMS, electricity meters, temperature and humidity sensors, fire protection equipment and other devices may come from different manufacturers. Some use RS485, some use CAN, and others directly adopt Ethernet; their protocols, point lists and data formats are also completely different. For a single project, the problem can still be solved by engineers conducting on-site debugging for each device one by one. But once the scale expands to dozens of industrial parks and hundreds of grid-connected points, later maintenance will easily get out of control.

Therefore, distributed photovoltaic networking cannot be simply understood as "connecting devices to 4G". More importantly, a stable data and control entry should be established at the station side first.

1. First Solve the Problem of "Unifying All On-Site Devices"

For a typical photovoltaic-energy storage site, the architecture can be sorted out as:
Inverter / PCS / BMS / Electricity Meter / Environmental Equipment → EG628 → EMS or Remote Monitoring Platform

The Fanless Industrial PC EG628 is equipped with 2 RS485 ports, 1 RS232 port, 1 CAN port and 2 Ethernet interfaces, and supports 4G, WiFi and Ethernet networking. For data of different protocols such as Modbus, IEC104, IEC61850 and OPC UA, collection, sorting and conversion can be completed at the edge side first, and then uploaded uniformly in the format required by the upper-layer EMS.

The biggest change brought by this approach is that the platform does not need to adapt to each device manufacturer separately.

For example, in the same project, the inverter from Manufacturer A and the electricity meter from Manufacturer B both use Modbus, but their register definitions are completely different; the BMS outputs data through CAN. During project implementation, device templates and point list mapping are configured on the Fanless Industrial PC EG628 side in advance, so that the data finally uploaded to the platform can be unified into standard data such as "photovoltaic power, grid-connected power, SOC, PCS power, and alarm status".

When replacing an inverter of a certain brand later, the main work is to adjust the station-side device adaptation, rather than re-modifying the entire cloud platform.

2. After Energy Storage Is Connected, Networking Is No Longer Just "Viewing Data"

Pure photovoltaic monitoring mainly solves the visualization problem, but after adding energy storage, on-site scenarios begin to involve power coordination.

When photovoltaic power suddenly rises in the daytime, when will the energy storage start charging; when the park load increases, whether the energy storage needs to discharge; when the grid-connected point power exceeds the set range, how will the PCS respond — if these actions completely rely on the remote server, the local control will be affected once the public network fluctuates.

The Fanless Industrial PC EG628 runs Linux Ubuntu, supports C/C++ secondary development, and has built-in environments such as Node-RED and OpenPLC, so part of the project logic can be executed on site. Official documents show that it supports edge collection, calculation, active reporting and local PLC logic.

Therefore, a more reasonable design is not "the cloud controls everything", but:
The cloud is responsible for centralized monitoring and strategy management, while the station side is responsible for collection and linkage that require rapid response.

For example, adjusting the PCS target value according to the grid-connected point power, and maintaining the established operation strategy after the network is interrupted, can be designed to be executed locally according to project requirements.

However, safety functions such as fire protection interlock, BMS protection and PCS self-protection should still be undertaken by the original professional control system. Adding an edge computer should not change the original safety link.

3. On-Site Deployment, It Is Recommended to Carry Out in Three Steps
3.1 Do Not Install Devices Directly — Make a Communication Table First

Sort out the interface type, device address, baud rate, register, data unit and control permission of the inverter, PCS, BMS and electricity meter clearly. Many of the most difficult problems to troubleshoot in the later stage of distributed projects are caused by the lack of unified point lists in the early stage.

3.2 Then Build the Network

A small number of devices can be directly connected to the Fanless Industrial PC EG628; if a site has multiple Ethernet devices such as inverters and PCS, they should be converged uniformly through an industrial switch before being connected to the Fanless Industrial PC EG628. The Fanless Industrial PC EG628 only has two Ethernet interfaces, one Gigabit and one Fast Ethernet, so it cannot be directly used as a multi-port switch.

For projects with fixed networks, Ethernet is preferred, and 4G can be used as the networking or backup channel according to on-site conditions.

3.3 Finally Configure the Edge Logic

Complete data collection first, and then gradually add control strategies such as power limitation and energy storage linkage. Do not write all control logic at one time, otherwise it will be difficult to judge whether the problem lies in communication, point list or control algorithm when an on-site fault occurs.

4. Project Acceptance Should Not Only Focus on "Platform Online Status"

Before the official launch, it is best to actively create several abnormal scenarios.

Unplug the public network to check whether local collection and necessary control can continue; restore the network to check whether the data can be re-uploaded; simulate the communication interruption of the inverter to verify whether the system can correctly judge that the device is offline; then simulate the change of photovoltaic power to observe the actual response of the PCS.

If the project involves power regulation, the whole machine test should also be carried out in combination with the response time of the inverter, PCS and BMS. The Fanless Industrial PC EG628 supports millisecond-level data collection and local processing, but the final regulation speed depends on the entire control link, not the performance of a single edge computer.

The installation environment cannot be ignored either. The Fanless Industrial PC EG628 is specified with DC 9~36V power supply and an operating temperature range of -20℃~70℃. Heat dissipation space should still be reserved inside the cabinet, and wiring should be separated from high-interference power lines as much as possible.

For distributed photovoltaic energy storage manufacturers, the more sites they have, the less they can rely on "separate debugging for each project site".

A more practical approach is to make the Fanless Industrial PC EG628 a standardized station-side unit: connected to inverters, energy storage systems and electricity meters at the lower end, and connected to EMS or cloud platforms at the upper end, completing protocol unification, data processing and necessary local linkage in the middle.

In this way, when replicating new photovoltaic-energy storage sites later, the main adjustments needed are the device point list and project strategy, rather than building a complete communication system from scratch. For manufacturers that need to operate dozens or hundreds of distributed sites, this is the key to whether the networking architecture can be implemented on a large scale.

需要我帮你把这份译文中的‌工业边缘计算相关专业术语整理成中英对照表‌，方便后续技术文档统一使用吗？