Technological Innovation
数字化介绍
Transfar Synthetic Material’s digital transformation journey has progressed in sync with the company's development, going through different stages of standardization, automation, digitalization, and intelligence over the years, gradually forming a 1+2+N intelligent operational architecture.
From an operational management perspective, sales plans, production plans, procurement plans, energy plans, equipment maintenance plans, R&D plans, capital plans, and human resource plans are all managed through an online closed-loop budget management platform. All procurement is unified through a centralized procurement platform, all payments are processed through a financial shared services platform, the entire lifecycle of supplier admission, evaluation, and exit is completed in the SRM system, customer profiling, opportunity management, and credit management are handled in the CRM system, and accounting processing and financial accounting are completed in SAP ERP.
From a production process management perspective, real-time control of the entire production process is achieved through MES, comprehensively mastering information on unit production, energy management, quality control, and process operations. This enables full-process material data from raw material entry, unit production, to product shipment, establishing a plant-wide material movement model, making the entire production site completely transparent. Through PID loop tuning, AAS advanced alarm management, and APC advanced control, the unit's automatic control rate exceeds 90%, and control stability remains above 98%.
AI+全场景应用
Starting from 2024, the company began exploring AI+ full-scenario applications and formulated an AI transformation strategy to complete the strategic leap from "Advanced Materials Manufacturer" to "AI+ Global Advanced Materials High-tech Company" within five years. The advancement of the AI+ strategy can be divided into three phases, progressing from point to line to surface: starting with efficiency improvement (single-agent construction), continuously achieving business integration (multi-agent collaboration), and ultimately realizing model innovation (industrial brain).
After several years of effort, the company has made progress in AI transformation, with an organization-wide AI transformation atmosphere established, and AI+ scenarios being continuously and deeply promoted.
In terms of AI applications in manufacturing scenarios, we mainly deploy industrial time-series large models and build multiple AI agents focused on process control for production operations, gradually achieving self-perception, self-decision-making, self-execution, and self-optimization across all manufacturing scenarios. This enhances Transfar Synthetic Material's overall intelligent manufacturing level, ensuring safety, reducing energy consumption, and improving personnel efficiency.
Production operation AI agents are built on over 10 years of Transfar Synthetic Material’s operational data. Through pre-training and fine-tuning, we established an industrial brain powered by TPT, upon which we developed multiple agents: process warning, energy optimization, quality prediction, safety management, alarm optimization, autonomous monitoring, control optimization, and operation optimization.
Quality full-process intelligent control mainly achieves online volatile content detection, real-time Mooney viscosity prediction, automatic rubber block sampling, intelligent sample delivery vehicle transportation, automatic sample preparation, AI appearance defect detection, collaborative robot rubber picking and digging, and AI agent automatic generation of quality reports.
Predictive maintenance for all equipment starts with key rotating equipment, followed by static equipment, instruments, etc. Key rotating equipment has already achieved online temperature and vibration monitoring and predictive maintenance, and online diagnosis and predictive maintenance for instrument valves are currently underway.
In-plant logistics intelligence automatically identifies vehicle position and posture through AI, measures internal compartment dimensions, and generates loading plans, achieving iron box loading and paper bag loading for different vehicle types. The entire circulation of empty containers is completed by AGV carts, and our automated storage and retrieval system enables unmanned warehouse operations.
In office management scenarios, based on DingTalk's native AI platform and Qianwen large model, we achieve intelligent meeting management, intelligent project management, intelligent operation management, and knowledge base management, significantly improving collaboration efficiency.
In terms of R&D scenarios, the transition is made from 1.0 manual experience to 2.0 AI-driven intelligent decision-making. An R&D design and experimentation platform is established to integrate the advantages of data, AI models, and automated equipment, thereby creating a new R&D model that is efficient, flexible, and market-competitive.
