China's First! New Energy Storage AI Analysis Platform Put into Operation, New Energy Consumption Increased by 30%

According to CCTV News, China’s first domestically developed new energy storage artificial intelligence data analysis platform was officially put into operation yesterday.

The platform connects various types of new energy storage devices. Relying on AI autonomous learning and massive data analysis, it can remotely and in real-time identify device defects and potential hazards, automatically generate maintenance plans, and respond quickly.

Currently, the platform has integrated 8 new energy storage power stations across Guangdong, Yunnan, Hainan, and other regions, with over 2.3 million data collection points. After a year of trial operation, the failure rate of equipment at these 8 stations has decreased by 34%, new energy consumption has increased by approximately 30%, and system regulation capacity has significantly improved.

Technicians from the Southern Power Grid Energy Storage Maintenance and Testing Branch say that the platform now has the ability to perform intelligent analysis on over 100 large-scale energy storage stations, with high-quality data sets built for lithium and sodium batteries. Next, it will connect to new demonstration stations such as all-vanadium redox flow batteries.

What is the new energy storage artificial intelligence data analysis platform?

As the installed capacity of battery energy storage rapidly increases, the volume of data generated grows exponentially. Traditional cloud-centric computing methods face high latency and low efficiency, making it difficult to meet the needs of cell-level smart operation and maintenance.

Essentially, the new energy storage AI data analysis platform is like giving energy storage systems a “brain” that can learn and think. It is not just monitoring data but deeply mining data value to shift energy storage systems from passive execution to proactive prediction and intelligent optimization.

Its working principle can be summarized as a closed-loop system from data perception to intelligent decision-making—

• Terminal collects multi-dimensional data on batteries, equipment, and environment, which is pre-processed in real-time at the edge layer and uploaded to the cloud;
• Cloud analyzes data using AI models and digital twins, conducting battery health prediction, early fault warning, charge/discharge strategy optimization, and intelligent operation and maintenance diagnostics, enhancing accuracy by combining electrochemical mechanisms and data models;
• Using a collaborative architecture of end-edge-cloud, models are trained in the cloud, local risks are responded to in milliseconds at the edge, and terminal devices can autonomously protect themselves even when disconnected from the network, ensuring fast and reliable execution.

The platform ultimately forms a “perception—analysis—decision—execution” closed loop, issuing optimization commands and early warning information to the terminal execution layer, and continuously iterating models with new data, turning maintenance experience into a knowledge graph, and combining large models to assist human-machine collaborative decision-making. This aims to achieve safety early warning, extend system lifespan, reduce operation and maintenance costs, increase revenue, and ensure efficient and stable system operation.

Green electricity is expected to further support AI development

In March 2026, compute-electricity collaboration was first included in the government work report as a national-level new infrastructure project.

Liu Liehong, Director of the National Data Bureau, stated at the China Development High-Level Forum 2026 that the next step will be to vigorously promote the compute-electricity collaboration project with relevant departments, ensuring that more than 80% of new computing power infrastructure at key nodes uses green electricity, maximizing the support role of renewable energy.

Liu explained that compute-electricity collaboration involves deep integration of computing infrastructure and power systems through digital technology, intelligent algorithms, and information networks, promoting dynamic resource matching and optimized allocation—creating a virtuous cycle of “power-driven computing and computing-driven power.” Main initiatives include promoting direct green power supply, green power aggregation, increasing green power support for computing, recycling waste heat, and enhancing green low-carbon cycle benefits.

“Energy storage + AI” has become an excellent example of the implementation of compute-electricity collaboration scenarios.

Currently, several energy storage platforms are connected to Deepseek.

For example, Xinyuan Zhichu AIOPS2000 energy storage intelligent operation platform focuses on safety, economy, and reliability, launching three intelligent agents. With DeepSeek’s intelligent agent collaboration framework, adaptive learning mechanisms, and multi-modal perception algorithms, decision feedback delay is less than 2 seconds, and the device health scoring system has an accuracy of 98.2%.

Ronghe Yuan Storage’s “Ronghe Baize” system, deployed privately with DeepSeek, monitors over 20 million cells, processes terabytes of data daily, detects faults in milliseconds, and reduces operation costs by over 30%.

Exxon Excellence Storage’s “Senmi Smart Storage” AI-EMS uses DeepSeek as the decision core, with photovoltaic prediction errors below 3% and load prediction accuracy over 95%.

Tong Fei from Zheshang Securities believes that compute-electricity collaboration builds a bridge through two modes: “compute with electricity” and “electricity with compute”. “Electricity with compute” involves creating a “flexible power pool” (integrating generation, grid, load, and storage) to provide stable, green power support for highly volatile computing loads. Industry-wise, core sectors such as power equipment, smart dispatching software, energy storage, and project operation will see a reassessment of value.

Hua Fu Securities states that State Grid is encouraging pre-construction and early deployment of power infrastructure, including early layout of UHV transmission corridors, reserving substation capacity, and planning green power supporting construction schemes, to quickly meet new computing power demands. Based on widespread deployment of sensors and smart devices, digital hardware and software—especially AI’s powerful computing capabilities—can optimize power system operation, improve grid prediction of renewable energy generation and urban or regional electricity demand, enhancing the grid’s perception of generation and consumption, and increasing the accuracy of energy market transactions and virtual power plant aggregation.

From an investment perspective, AI’s ultimate application is in power, with markets mainly focusing on gas turbines, solar-storage, and power grids; while the end goal of power is AI, which mainly involves integrating digital intelligence and virtual power plants, combined with green electricity market trading, to close the wind-solar-storage network loop. Currently, virtual power plant digitalization is recommended to focus on companies like Weisheng Information, State Power Investment R&D, Sifang Co., Ltd., and Nanjing Power Grid Technology; green electricity supply should consider Tongli Tianqi, JinkoSolar, GCL Energy Science and Technology, and other green power operators, as well as Sifang Co., Ltd. and China Energy Nixin for green power equipment solutions.

(Source: Caixin)