In 2025, the production capacity of TOPCon/HJT batteries will exceed 93%, the efficiency of perovskite tandem cells will exceed 35%, and the requirements for battery string testing in the photovoltaic industry have upgraded from "qualified screening" to "precise characterization" - traditional equipment faces four major technical bottlenecks: "multiple parameters not synchronized,片面 defect identification， difficulty in adapting to new types of batteries， and rigid sorting logic". Moreover， imported high-end equipment not only costs over 1.5 million yuan， but its core algorithms and customized services are also "technology chokepoints".
Beijing Saifan Photovoltaic has been deeply engaged in photovoltaic testing for 22 years， supported by 28 core patents， and has launched the SS-S6000 series automatic battery string testing and sorting machine. With "multi-modal AI defect detection engine， FPGA millisecond-level synchronization， digital twin sorting， and all-round adaptability technology"， it has pushed the detection accuracy， speed， and adaptability to new industry peaks， becoming the "technical benchmark" for component mass production improvement and new battery research and development.
1. Core Technology One： Multi-modal AI Defect Detection Engine - 0.01mm-level Defects "No死角" Identification
Traditional detection relies only on EL single-modal data， with a leakage rate of up to 5%-8%. However， Saifan has innovatively adopted "EL + PL + infrared thermal imaging" multi-modal fusion technology， combined with an independently trained defect feature deep learning engine， to achieve the leap from "surface screening" to "essential diagnosis"：
Technical Principle： Integrating 16 million-pixel high-resolution EL cameras， photoluminescence (PL) detection modules， and infrared thermal imaging units， through a million-level defect sample library (covering 25 types of perovskite defects and 38 types of crystalline silicon defects) for training， the algorithm can automatically correlate "electrical performance anomalies - microscopic structural defects - thermal distribution anomalies" three-dimensional data， forming a defect diagnosis loop；
Industry Breakthrough： Can accurately identify 18 types of defects such as 0.01mm ultra-fine cracks， false welding (welding resistance > 5mΩ)， broken gates (area > 0.01mm²)， black spots， etc.， with a leakage rate of ≤0.05%， an identification accuracy of 99.8%， which is 3 times higher than that of industry mainstream equipment， completely filtering out surface interference signals such as dust and scratches；
Efficiency Innovation： Multi-modal data synchronous collection， single-string detection takes only 8 seconds， which is 4 times more efficient than traditional single-modal equipment， and supports parallel detection of multiple strings， processing 1200 strings per hour.
2. Core Technology Two： FPGA-driven Millisecond-level Synchronous Acquisition - 8 Core Parameters Captured in 120ms
In response to the problem of low efficiency and data synchronization caused by "single parameter testing" in traditional equipment， Saifan adopts a FPGA + PCIE 3.0 high-speed transmission architecture to achieve "instantaneous synchronization" of multiple parameters：
Technical Principle： Equipped with Xilinx Virtex7 series FPGA chips， integrating 4 pieces of 1.25GSa/s high-speed ADCs， through its independently developed "path delay calibration algorithm"， eliminating time differences (≤52ps) in multi-channel acquisition， data is transmitted at a 5.12GB/s bandwidth through the PCIE 3.0 interface to the SSD storage array；
Industry Breakthrough： Synchronously complete the acquisition of 8 core electrical performance parameters (open-circuit voltage， short-circuit current， fill factor， conversion efficiency， etc.) within 120 milliseconds， with a test accuracy of ±0.3%， far exceeding the IEC 60904-9 standard requirements， and data synchronization is 2 times higher than that of imported equipment； Applicable scenarios: Voltage measurement range 0-1500V, current 0-30A. Perfectly compatible with large-sized battery strings of 182mm/210mm and high-power products of 500W and above. It meets the low-decay and high-consistency detection requirements for n-type batteries.
Three. Core Technology Three: Digital Twin Sorting Algorithm - 16-level Customized "Precise Classification"
Traditional sorting only relies on a single efficiency parameter, resulting in poor component consistency (performance deviation > 1%). Saifan introduces the digital twin sorting technology to achieve "multi-dimensional - traceable - adaptive" classification:
Technical principle: Based on the digital twin model of the battery string, the 6 core indicators (efficiency, fill factor, defect grade, attenuation trend) are quantified as sorting dimensions. Through the "multi-dimensional clustering algorithm", the optimal sorting scheme is automatically generated, supporting 16-level customized sorting rules;
Industry breakthrough: The performance deviation of components in the same batch can be controlled within ±0.5%, with a 50% improvement in consistency compared to the industry average, effectively reducing the power loss after parallel connection of battery strings (from 3% to 0.5%);
Intelligent linkage: Supports seamless connection with the MES system, real-time upload of detection data, and can reverse-trace the production process parameters of each battery string (such as welding temperature, string welding pressure), providing data support for process optimization, achieving a "detection - production - optimization" closed loop.
Four. Core Technology Four: Cross-technology Route All-Adaptation Technology - "Full Compatibility" from crystalline silicon to perovskite
Facing multiple technology routes such as TOPCon, HJT, perovskite, and stack, Saifan adopts hardware modularization + algorithm adaptive design to break the industry limitation of "one device adapting to one type of battery":
Technical principle: The hardware adopts replaceable test probe modules (flexible / rigid options available), and the software has multiple detection algorithm libraries, which can automatically adjust the test voltage (0-100V), current (0-20A), and light irradiation intensity (0.1-1.2S0) according to the type of battery;
Industry breakthrough: For calcium phosphate batteries with flexible and low-stability characteristics, a "low-damage testing mode" (test current ≤ 0.1C) is adopted to avoid performance attenuation during testing; for stack batteries, the spectral response matching algorithm is optimized to accurately capture the differences in photovoltaic conversion efficiency of different sub-batteries;
Compliance guarantee: The detection data complies with domestic and international standards such as IEC 60904 and GB/T 37655, and is calibrated by the national photovoltaic metrology station, which can be directly used for TÜV, UL, etc. international certifications.
Technology implementation: Verification of three scenarios - "Hardcore strength"
Production line improvement: A leading TOPCon component factory adopted 10 SS-S6000s, reducing the single-string detection efficiency from 30 seconds to 3 seconds, increasing the daily production capacity from 5,000 strings to 20,000 strings, reducing the defect rate from 0.8% to 0.05%, and saving over 8 million yuan annually;
Research and iteration acceleration: The Chinese Academy of Sciences' photovoltaic center successfully captured the evolution law of interface defects of perovskite stack batteries using its multi-modal detection function, reducing the experimental cycle by 30%, and related achievements were published in "Advanced Materials";
Third-party certification acceptance: Beijing Jianxing and German Rheinland TÜV accepted it as a benchmark device, and the detection data was directly included in the certification system. An enterprise, based on the equipment's detection report, shortened the EU certification cycle by 40%.
Brand technology endorsement: Domestic substitution, breaking overseas monopoly Patent protection: 28 core patents cover key aspects such as AI algorithms, synchronous collection, and sorting logic, and have participated in formulating the industry standard "Technical Requirements for Photovoltaic Cell String Detection and Sorting Equipment";
Cost-performance advantage: The equipment price is only 50% of that of imported similar products. Core components (FPGA, camera, detection module) offer a 3-year warranty, with a service life exceeding 8 years, and the annual maintenance cost is only 1/5 of that of imported equipment;
Customization capability: Supports rapid customization based on customer production line layout, battery type, and sorting requirements, with a customization cycle of only 45 days, which is 60% shorter than that of imported equipment.
Link Microblog
