MatriX Factory edge computing platform (Nano / Lite / Pro) | MatriBOX software

Software · under MatriBOX

Natively fusing IT + OT + AI — from device level to shop-floor data foundation

What MatriX Factory is

MatriX Factory (MXF) is an edge computing software platform developed by MatriBOX for its MX series edge intelligent controllers. It targets the device edge and the industrial shop floor, meeting the needs of shop-floor, production-line and single-machine deployments for high-volume real-time data acquisition, storage, analysis, modelling and AI inference. It can also be delivered as a modular, customisable and scalable AI / IT / OT platform.

The closed loop is control → data → cloud or local compute → strategy → control, which makes it easier to encapsulate process algorithms and decouple them from the hardware.

Value proposition — small, light, fast, smart

Small & light (Nano) — ultra-compact DIN-rail format for any control cabinet; lightweight container runtime, OTA updates, hardened industrial Linux, pre-installed software and a low resource footprint, with no complex system integration
Fast (Lite) — high-performance CPU, rich I/O, edge AI, few-shot model training and predictive maintenance; a drag-and-drop development environment shortens project cycles
Smart (Pro) — data-flow engine, low-code streaming, GPU acceleration and 300+ industrial protocols, with millisecond-level analysis and a high-availability architecture

Three software tiers and three hardware tiers can be combined freely to suit different scales and scenarios.

**MatriX Factory product set** — small, light, fast, smart; software and hardware deployed together

MatriX Factory tiers — **Three tiers by scenario** — Nano for single machines, Lite for lines, Pro for shop-floor level

Three tiers

Nano / Lite / Pro at a glance

Item	MXF Nano	MXF Lite	MXF Pro
Target	Single-machine intelligent gateway	Production line / flexible manufacturing island	Shop-floor digital foundation
Hardware	MX4000 / MX5000	MX6000 / NVIDIA Jetson Orin	MX8000 / MXF Cloud Data Center
Connected devices	5	20	50
Throughput	500 × 5 byte/s	500 × 20 byte/s	500 × 50 byte/s
Private deployment	No	Yes	Yes
Customisation	No	Yes	Yes

MXF Nano highlights

Zero programming — device connection and data acquisition in 5 minutes
20+ industrial protocol conversions, compatible with old and new equipment
Pre-processing at source cuts invalid data transmission by 95 percent
Breakpoint resume plus SM2/3/4 national cryptographic algorithms
Device health monitoring with early warnings; device and IT network isolation
Remote program download, diagnostics and maintenance, cutting travel for commissioning
Classified storage: real-time, cached, feature, historical, configuration and base data

MXF Lite highlights

Multi-sensor fusion and edge computing
Predictive maintenance and device health management
Protocol parsing and communication interfaces
SM2/3/4 cryptography for data security and real-time performance
Can act as an edge AI inference unit — MX6300 heterogeneous multi-core, or embedded NVIDIA Jetson Orin

MXF Pro highlights

Multi-node clustering with automatic failover and 99.99 percent availability
GPU / NPU acceleration supporting parallel multi-model inference
Cross-shop data sharing and an industrial application App Store
Builds a high-availability shop-floor data centre
Targets automotive and parts, energy and power, tobacco, oil and gas, smart mining and metallurgy

Capabilities

Data engine, HMI configuration and openness

Data engine

The engine packages hundreds of industrial protocol and function nodes into a data-flow, drag-and-drop programming model. It supports high-frequency acquisition, cleaning and encryption, storage, algorithm encapsulation, database access and two-way exchange with third-party platforms.

Low-code streaming framework with millisecond-level throughput
Drag-and-drop programming, component-based design, custom logic and visual conditional branching
Preset protocol nodes for Siemens, Beckhoff, Omron, Mitsubishi and many more
Node categories: trigger rules, data processing, industrial protocols, databases, API interfaces, notifications

MatriX Factory data engine — **Data engine node configuration** — drag-and-drop programming with visual node parameters

HMI configuration

Zero-code drag-and-drop configuration with a 300+ industrial component library
Millisecond-level data refresh with dynamic effects and alarm flashing
Responsive layout for large screens, PC, tablet and phone — design once, use everywhere
Rich chart types: industrial gauges and trend curves
Real-time and historical trend curves, groupable and comparable across multiple windows, with traceable records

The two dashboards below are configuration examples taken from the manufacturer's manual (a warehouse-management template and a visualisation-supervision template). They are not delivered project screenshots and do not represent existing customer cases.

Openness and integration

Industrial protocols: Siemens S7, Modbus TCP / RTU, AB EtherNet/IP, Mitsubishi MC, OPC UA, ADS, Fins TCP
Databases: MySQL, Oracle, Sqlite, PostgreSQL, SqlServer, Access
Network: MQTT, RabbitMQ, Kafka, FTP
Notifications: Email, DingTalk, WeCom, Feishu, SMS
Edge AI: AIGC-assisted few-shot training models and common vision algorithms; built-in linear regression and DLinear models with inference executed on NPU / GPU
Three layers: IT systems (ERP / MES / PLM / BI / cloud) · OT devices (PLC / DCS / sensors / industrial equipment) · application ecosystem (MatriX Control / MatriX Factory / WSCAD)

HMI configuration example — **HMI example 1** — warehouse-management template

Supported NVIDIA JetPack SDK components

Package / component	Type	Function
Ubuntu 22.04	操作系统	提供基础系统环境（L4T），含引导加载程序、定制化 Linux 内核、NVIDIA 驱动程序、示例文件系统
CUDA Toolkit	SDK / 平台	在 NVIDIA GPU 上构建和运行加速应用的开发环境（编译器、数学库、开发工具）
cuDNN	库	高性能深度学习基元库，为深度神经网络框架提供优化过的核心操作实现
TensorRT	库 / 运行时	高性能深度学习推理优化器和运行时，用于部署低延迟、高吞吐量 AI 推理应用
DeepStream SDK	SDK / 库	构建基于 AI 的视频分析、多传感器处理应用的流媒体分析工具包
VisionWorks / VPI	库	跨硬件（PVA / GPU / CPU）的计算机视觉算法加速库
OpenCV	库	开源计算机视觉和图像处理库，JetPack 版本含针对 Jetson 的 GPU 与硬件加速优化
NVIDIA Container Runtime	软件 / 工具	支持 GPU 加速的容器运行时，简化容器化 GPU 应用的构建与部署
cuBLAS / cuFFT / cuSOLVER	库	CUDA 加速数学库：线性代数、傅里叶变换、求解器
PyTorch / TensorFlow	AI 框架	主流深度学习框架，在 JetPack 中预构建并针对 Jetson 优化
nvgstcapture / nvjpeg	工具 / 库	Jetson 专用工具与库，如摄像头捕获应用、JPEG 硬件加速编解码

Hardware and support

Supported hardware and service levels

Model	Specification	Devices	Throughput	Private deployment	Customisation
MX4000	四核 A55 2.0GHz；内存 4G；存储 32G（标配）/128G（MAX）；算力 1Tops	5	500 × 5 byte/s	不支持	不支持
MX5000	A76 四核 2.4G + 四核 A55；内存 16G；存储 64 / 128 / 256 / 512G	5	500 × 5 byte/s	不支持	不支持
MX6000	NPU 算力 6Tops；可部署于 NVIDIA Jetson Orin 嵌入式 AI 计算机	20	500 × 20 byte/s	支持	支持
MX8000	工厂级边缘集群	50	500 × 50 byte/s	支持	支持

MX4000 / 5000 / 6000 集成了符合 IEC 61131-3 的 MatriX Control。按设备每秒 100KB 数据计算，128G 存储约可保存 20 天数据。表中规格取自厂家手册，部分字段手册未逐一列出，选型前请与我们核对。

Service levels

Item	Standard	Advanced	Premium
响应时间	8 小时	4 小时	1 小时
远程支持	5×8	7×16	7×24
现场支持	72 小时	24 小时	24 小时
系统健康检查	每季度	每月	每月

Service capabilities: 私有化部署, 定制化开发, SLA 服务等级, 7×24 NOC 中心, 专家团队（工业自动化与 AI 领域）

Service flow

问题提交（多渠道提交工单、自动分配优先级）
远程诊断（专家远程连接、快速定位问题）
解决方案（远程实施或安排现场）
跟踪回访（问题解决后跟踪回访）

Resources: 产品手册, 技术白皮书, 视频教程, API 文档, 常见问题, 社区论坛

Manufacturer hotline 4000-322-300; support@matribox.com; www.matribox.com. This page consolidates MatriBOX product material; contractual service terms follow the manufacturer's agreement. We can also handle product selection and technical support directly.

MatriX Factory edge computing platform