On June 27, the large model helps energy renewal - the energy model forum was held at the Beijing National Convention Center, and the Antelope energy model was released.
Based on the general capabilities of the iFLYTEK Xinghuo model, the large model combines the actual needs of the energy industry scenario to create five core capabilities: energy content generation, energy knowledge Q&A, energy understanding and calculation, energy task planning, and energy multimodality, covering six "large model + energy" scenario applications, including equipment operation and inspection, power questioning, power marketing customer service, auxiliary power trading, new energy power forecasting, and safety production.
Experts and scholars discuss the direction of new energy development
Shi Yubo, Chairman of the China Energy Research Society, emphasized in his speech that large-scale model technology plays a key role in the sustainable development of the energy industry, which can effectively optimize management, accelerate the research and development of new energy, and provide a scientific basis for policy formulation. He further pointed out that at present, it is necessary to deeply explore the deep integration of large model technology and the energy industry, focusing on important links such as data integration, model accuracy improvement and talent training, and iFLYTEK and Antelope provide new solutions for the application scenarios of the energy industry.
"In today's world, the energy landscape is undergoing tremendous changes. The rise of artificial intelligence technology and the deep integration with the energy field are leading a new wave of energy revolution. Ling Wen, academician of the Chinese Academy of Engineering, spoke highly of the application and exploration of iFLYTEK and Antelope in the energy field, and emphasized the importance of large model technology to improve energy efficiency, optimize resource allocation and promote the transformation of the energy industry in the context of global energy transformation and climate change.
At the meeting, the Antelope energy model renewal plan was launched. Sun Zhengyun, vice chairman and secretary general of China Energy Research Society, and Wang Rui, vice president of Antelope, signed a strategic cooperation agreement on behalf of both parties. The two sides will give full play to their respective roles as high-end think tanks in the energy field and their outstanding advantages in the field of artificial intelligence, and jointly promote the rapid formation of new qualitative productivity with energy data and artificial intelligence as elements.
The Antelope energy model helps energy renewal
2024 is the first year of large-scale model application, and the base capability of general large-scale models is the key to accelerating the implementation of large-scale model industry. How to open up new routes and use large-scale model technology to achieve a wider range of industry applications has become a key proposition for 2024 and future development. On June 27, iFLYTEK Xinghuo large model V4.0 and related landing applications were released, comprehensively improving the seven core capabilities of the large model base, not only ranking first in 8 international mainstream test sets, leading the domestic large model, but also surpassing GPT-4 Turbo in text generation, language understanding, knowledge question and answer, logical reasoning, mathematical ability, etc.
Large models are an important force to promote the current technological progress, especially in the energy field, which has a wide range of application scenarios. Based on the general capabilities of the iFLYTEK Xinghuo model, the Antelope energy model is based on the core technology, combined with the actual needs of the energy industry scenario, with five core capabilities: energy content generation, energy knowledge Q&A, energy understanding and calculation, energy task planning, and energy multi-modality, and has three major characteristics: cloud-edge-end collaboration, independent controllability, and data and model security, covering six major industry scenarios: wind, solar, water, fire, nuclear, and storage, providing rich landing solutions for promoting energy renewal in the context of green, low-carbon and high-quality energy development.
6 "large model + energy" scenario applications
Provide a wealth of industrial landing solutions for energy renewal
Xu Jiajia, President of Antelope Industrial Internet Co., Ltd., introduced the six "large model + energy" scenario applications of Antelope Energy, including equipment operation and inspection, power questioning, power marketing customer service, auxiliary power trading, new energy power prediction, and safety production, providing rich landing solutions for energy renewal application scenarios.
The "equipment operation and inspection" system is the "digital operation brain" of the power station. In the field of photovoltaics, equipment operation and inspection is an important and complex work, and the Antelope equipment operation and inspection model is equipped with a new generation of iFLYTEK Xinghuo 4.0 engine, which not only has the ability of accurate accident research and judgment, but also has the industry's leading intelligent decision-making generation ability, from diversified spatial information to the historical map that runs through the whole life of the power station, and builds the digital space-time brain of the power station.
For example, the station manager of the power supply company needs to log in to different systems to obtain business-related information before dealing with door-to-door problems, and the "power question" uses the antelope energy model to access the data of different business systems, so that the station manager can obtain data anytime and anywhere through natural language questions and answers, and it is estimated that the data acquisition time can be saved by 60%.
"Power Marketing Customer Service" uses large language model + knowledge base + agent technology to provide enterprise end users with automated customer service robots for consultation and business handling, and provides independent service capabilities such as click consultation, package recommendation, bill interpretation, and contract inquiry for power purchase customers. It is expected that the proportion of one-time user problem resolution will increase from 30% to 70%, the manual workload will be reduced by 40%, and the personalized customer service coverage will be 100%.
At present, the market scope of electricity trading is large, there are many trading entities, and the trading rules of different provinces are different, so the time series prediction ability of the energy model is used to predict the power demand, power generation and electricity price of the market. In "Auxiliary Power Trading", the energy model can realize the dynamic matching of power supply and demand, and make quotation suggestions for power trading based on the power generation forecast and demand forecast of the power plant.
Due to the intermittency, output fluctuations, regional differences, climate differences and other characteristics of new energy power generation, it brings uncertainty to the stability of the overall power grid. "New Energy Power Forecasting" can bring more accurate equipment power prediction results through comprehensive modeling and prediction of equipment, comprehensive environment and historical power data.
"Safe production" proposes a new solution paradigm for the field of "AI + safety" production. Taking the power generation industry as an example, the safety requirements are high, and a comprehensive inspection of power safety elements must be carried out on whether the safety helmets and tooling worn by the personnel entering the operation meet the safety specifications, and whether the professional equipment carried by them meets the safety qualifications.
The implementation of AI in the energy industry requires easy-to-use kits and platforms. Based on the Antelope energy model, energy enterprises can co-create and share on a one-stop AI development and application platform, realize the unified management of large and small models, and empower production, operation, management and other scenario applications through knowledge craftsmen, data craftsmen and process craftsmen. For different resource scenarios, Antelope accelerates the implementation of AI scenarios through the collaboration of large and small models, providing a variety of solutions for the energy industry.
Promote the two-way empowerment of AI and energy
Help China Energy Group run at the speed of "artificial intelligence +".
At the forum, experts from the Artificial Intelligence Innovation Center of China Energy Group Digital Intelligence Technology Company shared with the participants the country's first innovative application benchmark of "artificial intelligence + energy". Through the deep integration of the artificial intelligence platform and business, we will build an artificial intelligence platform for the energy industry that integrates the national computing power and the industry-leading training and reasoning of large and small models, build a group artificial intelligence platform base, support 8 major business segments and 5 types of core AI capabilities, empower various business scenarios in the energy industry, and promote the intelligent development of energy production, management and operation.
At present, iFLYTEK and Antelope have carried out relevant cooperation with China Energy Group and PetroChina. "Work together to build an energy model and promote the two-way empowerment of AI and energy." Xu Jiajia said that he looks forward to building a sustainable development ecology of the energy industry through "large model + energy" and jointly promoting the high-quality development of the energy industry. (Reporter Wang Qiao, intern Liu Yuning)