In the eyes of Pang Haitian, CEO of the national-level specialized and innovative “little giant” Jiangxing Intelligent, technology is not an “ivory tower”. Only by transforming cutting-edge algorithm codes from the laboratory into tangible “carbon reduction codes” in industrial scenarios can traditional fields like power inspection achieve a “green revolution”.
As the wave of green energy development intertwines with the wave of digital intelligence, a 32-year-old PhD is becoming a pioneer in this trend. He is Pang Haitian, CEO of Jiangxing Intelligent, a national-level specialized and innovative “little giant”. This young leader has a keen vision, accurately identifying edge computing as the “key to victory” in energy transformation, leading his team to launch a “technological raid” that has opened up innovative breakthroughs in traditional industries.
Currently, the green energy wave is unstoppable, transitioning from a “substitute” to a “main force”. As of the first quarter of this year, the total installed capacity of wind and solar power in the country has surpassed that of thermal power for the first time, with wind and solar power accounting for over 20% of total electricity consumption. This is not just a numerical leap, but a deep coupling of the “dual carbon” goals and market vitality. Among them, the role of private enterprises cannot be ignored, and young scientific and technological talents are the vanguards of innovation. Innovators like Pang Haitian have a broader stage for development.
In Pang Haitian’s view, technology is not an “ivory tower”. Only by transforming cutting-edge algorithm codes from the laboratory into tangible “carbon reduction codes” in industrial scenarios can traditional fields like power inspection achieve a “green revolution”. He has also been recognized for his outstanding contributions by being selected for the “Science and Technology Innovation China Youth Entrepreneurship List” by the China Association for Science and Technology, writing a chapter of struggle that embodies “technology towards green, youth with responsibility”.
Making Intelligent Inspections Truly Intelligent
During his studies, based on his knowledge of computer science, Pang Haitian recognized the integration of new generation information technology and energy development. Studying in Canada further deepened his understanding that new generation information technology is key to promoting green development in various countries. The idea of “empowering China’s power industry through edge computing and artificial intelligence technology” gradually took shape. In 2018, he gave up opportunities abroad and returned to China to co-found Jiangxing Intelligent with several co-founders from his laboratory, seeking breakthroughs at the intersection of energy and digitalization.
Under the “dual carbon” goals, there are vast application scenarios for intelligent power applications. The challenge for high-tech enterprises is how to find small entry points in complex large scenarios. For Pang Haitian, who does not come from an energy background, this is not an easy task. “Not only do we need to delve into the field of computer science, but we also need to expand cross-disciplinary knowledge, continuously enrich ourselves on-site, understand the operational processes of each link, and engage in dialogue with industry experts. Many of our orders and technological achievements have emerged from these exchanges and collisions.”
“During the construction of an intelligent inspection system at a certain ultra-high voltage converter station, we encountered a tricky problem with low algorithm recognition accuracy.” Pang Haitian recalls an unforgettable work experience. This ultra-high voltage converter station is a national key project, responsible for ensuring the safe and stable operation of the “West-to-East Power Transmission” and the power grid. The station is equipped with many world-leading electrical devices, but due to the large number and area, traditional manual inspections cannot cover everything, necessitating new technology to assist in achieving intelligent inspections.
At that time, Pang Haitian and the project team went deep into the site, and after analyzing the root causes, they found that the problems mainly concentrated on three aspects: first, the insufficient number of defect samples made it difficult to support algorithm training; second, the frequent updates and iterations of equipment meant that previously developed algorithms could not be directly adapted to new devices; third, the diversity of equipment suppliers meant that even the same model produced by the same manufacturer at different times had design and process differences. The same type of equipment in the station even involved two to three manufacturers, resulting in different fault points for the equipment, making the three-dimensional models and experimental data from previous inspections unusable. These factors combined led to a significant reduction in the actual application effectiveness of the intelligent inspection system.
“We ultimately decided to adopt a phased implementation plan, collaborating with the owner and major equipment manufacturers to integrate multi-source data such as equipment drawings and maintenance reports to build an industry knowledge base, alleviating the data shortage issue. We used a composite large model of ‘small model + Jiangxing source question big model’, where the small model handles basic recognition tasks with high accuracy, while the large model, with its powerful multi-modal understanding capability, integrates multi-dimensional information such as equipment images, operational sounds, and environmental parameters, reducing the algorithm’s dependence on sample size, lowering training costs, and significantly enhancing inference capabilities.” Thus, the new solution developed by Pang Haitian and the Jiangxing Intelligent team solved the inspection problem.
Machine Dogs Grow an “Edge Brain”
The entrepreneurial journey has not been smooth sailing. For Pang Haitian, who has deep expertise in computer technology and has crossed into the “energy + edge computing” integration innovation, this journey is filled with challenges. Facing these ups and downs, Pang Haitian said, “Failure is a compulsory course for innovation.” He admits that every setback contains valuable experience, and as long as one can accurately learn from it, failure can be transformed into a stepping stone for progress.
“New generation information technology is the ‘accelerator’ for energy transformation,” Pang Haitian emphasized to reporters from the China Energy News. He explained using big data technology as an example: in energy production, real-time monitoring and analysis of equipment operation data can accurately grasp the operational status of equipment, predict failures in advance, and improve the safety and stability of energy production, reducing energy waste caused by equipment failures.
However, truly converting theory and ideas into practice is not easy.
Pang Haitian introduced that during the initial practice of monitoring the status of intelligent substation equipment, we faced dual challenges of insufficient computing resources and communication resources in the early stages of deploying embodied intelligent robots (such as machine dogs) for inspections. The multi-dimensional perception data (visual, vibration, infrared, etc.) collected by the machine dogs was limited by communication bandwidth, and real-time transmission to the backend required long link transmission, leading to a surge in the load on the central processing unit. At the same time, massive raw data directly flooded into the cloud without preprocessing, not only slowing down analysis efficiency but also putting the backend services under long-term high-pressure operation.
He further stated: “To address this pain point, we innovatively integrated an edge computing box at the machine dog terminal, constructing a new architecture of ‘edge-side intelligent preprocessing + edge algorithm sinking’, which filters, extracts features, and performs preliminary analysis on the collected data in real-time through lightweight customized algorithms deployed on the edge box, only uploading key alarm information and simplified feature vectors to the backend. For example, for six types of high-frequency monitoring scenarios such as abnormal equipment temperature and mechanical component deformation, edge algorithms can directly complete data verification, threshold comparison, and risk grading, triggering local audible and visual alarms in real-time, and simultaneously pushing structured warning reports to the operation and maintenance system.”
This technological innovation significantly optimized system performance. Calculated results show that data transmission volume was reduced by over 70%, backend service processing pressure decreased by 65%, and abnormal alarm response time was compressed from minutes to seconds.
“The pre-deployment of edge computing not only broke the communication bottleneck limiting real-time performance but also enabled local computing priority mode, allowing the robots to truly possess ‘on-site instant decision-making’ capabilities, transforming them into embodied intelligence and greatly enhancing the timeliness and accuracy of inspections,” Pang Haitian said.
Continuously Opening Up New Paths for Innovation
The breakthroughs of Jiangxing Intelligent extend far beyond power inspections. Relying on core technologies such as industry algorithms, edge-cloud collaboration, and smart IoT, the solutions they have developed have penetrated multiple cutting-edge fields including smart grids, virtual power plants, “dual carbon” management, and smart power generation, successfully attracting multiple rounds of capital interest, growing into a national-level specialized and innovative “little giant” enterprise. Even more noteworthy is that they have gained the trust of a number of heavyweight central and state-owned enterprise clients, including the State Grid, State Energy Group, China Huaneng, State Power Investment, Datang Group, China Electric Power Research Institute, and Jingneng Group. This not only reflects market recognition of their technology but also affirms their comprehensive strength.
Pang Haitian candidly stated that the speed of technological iteration in the new energy industry is beyond imagination, and “not advancing is retreating” is a hard rule. The achievements of Jiangxing Intelligent today are not the result of one person but the crystallization of the entire team’s joint efforts. “In team collaboration, I often play the role of ‘igniting the spark’,” Pang Haitian shared. “When facing difficulties, everyone brainstorms together to break the deadlock. This spirit of collaboration is crucial for project advancement and market expansion.” As a leader, he must grasp macro planning and direction, delve into the front lines to understand the real situation, and provide team members with ample trust and room for trial and error. “Last year, we hit a bottleneck in an intelligent inspection project at a new energy station, and technical barriers once dampened the team’s confidence. We immediately organized a ‘technical breakthrough meeting’ for all staff, and through several days of brainstorming and repeated algorithm model adjustments, we not only tackled the hard problem but also exceeded delivery on key indicators. This is the best proof of the team sharing pressure and results together.”
“As a young technical talent in the new era and also a navigator of an energy enterprise, the responsibility on my shoulders is heavy,” Pang Haitian expressed his expectations for the future. “I aspire to explore an innovative path of ‘technical breakthroughs—scene implementation—ecological co-construction’ to inject new momentum into the development of the power system in the new era.” He stated that in the future, he hopes to continuously play his role, focus on practical issues, drive sustainable development through technological innovation, and contribute to accelerating the construction of a modern industrial system and promoting the comprehensive green transformation of economic and social development.
Written by | Reporters Dong Zitong, Su Nan, Wang Lin
EndFeel free to share with your friends!Produced by | China Energy News (ID: cnenergy)Edited by | Yan Zhiqiang