Page 47 - Market Analysis Report of Optical Communications Field in China & Global market 2025
P. 47
The system consists of two parts: one is the automatic Figure 8. Big data analytics-based quality management model for
conveying and precision evaporation system of raw optical fiber preform manufacturing
material gas, which realizes the automatic control of the
whole process from storage, conveying, evaporation to use;
the other is the industrial robots and automated distribution
system, and 28 sets of industrial robots and the system have
been applied to FiberHome digital factories in a large scale.
Relying on “vision + laser” navigation technology, they can
accurately complete the tasks of material distribution and
semi-finished product transfer, which greatly reduces the
labor intensity of manual picking. Through the application
of intelligent logistics system, FiberHome has achieved a
20% increase in optical fiber preform production capacity
and a 14% reduction in labor costs.
3.3 Intelligent Quality Control (2) Optimal design cycle iteration based on risk analysis
Quality control intelligence is at the core of FiberHome's and experimental design verification: Based on the optimal
smart manufacturing system, primarily enabled through process parameter combination package, the system
data acquisition and analysis to monitor, trace, and optimizes the characteristics of manufacturing elements,
optimize quality throughout the entire production product process characteristics and quality characteristics
process. To address quality management challenges in sequentially through regression analysis and experimental
the complex manufacturing conditions of optical fiber verification, so as to continuously improve the product
preforms, FiberHome has established a three-cycle quality of optical fiber preforms.
iterative quality- management model based on big data (3) Internal “V” cycle iteration based on process capability
analysis. This model is supported by a data collection and AI self-learning: The big data platform monitors the
platform, MES(Manufacturing Execution System), big data stability and controllability of each element in real time,
analysis system, data decision-making system, and expert and combines with the AI model to continuously iterate the
decision-making system, as shown in Fig. 8 and Fig. 9. 5M1E elements to realize the accurate matching of process
Through dynamic big data analysis and iterative process parameters, and the implementation model is shown in Fig. 9.
optimization covering the entire production process and all
key elements, FiberHome has achieved industry-leading Figure 9. Process capability analysis-based internal optimization
advancements in optical fiber preform manufacturing. iteration model
Figure 7. Traditional quality model for fiber preform production
The three-cycle iterative quality management model based The big data analysis system employs the random forest
on big data analysis includes: algorithm to preform regression analyze for predicting
(1) External “8” cycle iteration of big data traceability and product characteristic indicators. During model training,
verification: FiberHome has realized dynamic monitoring the random forest algorithm can autonomously construct
of the whole production process through the establishment an ensemble of decision trees, as shown in Fig. 10,
of a digital industrial big data platform, which can trace and evaluates feature importance by comparing model
the quality difference of optical fiber preforms and optical accuracy changes before and after feature perturbation.
fibers and screen the optimal process parameter combination This approach enables targeted analysis of key production
package. factors.
Through the big data platform and AI analysis model,
48 FiberHome has implemented precise management
of production elements, iteratively optimized optical
fiber preform quality, and resolved critical quality
45
