In recent years, the global installation of ESS (energy storage systems) has been on a steady upward trend. However, safety incidents have emerged as a persistent obstacle, impeding the healthy economic growth of the new energy industry. Data from the International Energy Agency (IEA) indicates that the number of global energy storage incidents has increased by 47% annually over the past three years. Notably, 82% of these incidents were fires triggered by thermal runaway. Amidst this challenging situation, industry players are mobilizing to explore a variety of solutions, ranging from optimizing intrinsic safety and early diagnosis to upgrading fire protection systems. Nevertheless, the crucial question of transitioning from “passive defense” to “active management” remains unanswered.
Why does the safety defense of ESS frequently falter? An in – depth analysis of incidents reveals that critical issues such as the gradual degradation of battery performance, the dispersion of parameters among individual cells, and the accumulation of latent failure factors often develop undetected within the monitoring blind spots of conventional battery management systems (BMS). So, how can we break free from the constraints of passive battery safety management? The ancient Chinese wisdom of “superior physicians prevent disease before its onset” from The Yellow Emperor's Canon of Medicine offers valuable insights. Merely extinguishing fires is a reactive measure; the ultimate solution lies in establishing preventive mechanisms at the source. This concept has inspired Hoenergy's research and development philosophy.
Innovation in Practice:
The Plug – and – Play AI – Powered Battery Safety Diagnostic and Prognostic Device
At the 2025 Energy Storage International Summit (ESIE), Hoenergy “Plug – and – Play AI – Powered Battery Safety Diagnostic and Prognostic Device” was honored with the Top 10 Prize in the International Energy Storage Innovation Competition. As an edge – AI – terminal innovation of BMS 2.0, this product successfully overcomes the limitations of conventional BMS passive monitoring approach. It constructs a multi – layer protection system that includes cell – level anomaly warnings and system – level situational awareness, marking a significant technological advancement in ESS safety management. Its innovative features have received high acclaim from international authoritative organizations.
Wen Qingwu, the R&D Director of BMS at Hoenergy, disclosed in an interview that this product integrates multi – physics coupling modeling with high – dimensional data feature extraction techniques. By deploying cloud – edge collaborative lifespan prediction algorithms and adaptive active balancing control strategies, it can analyze latent failure factors such as internal polarization characteristics and lithium plating tendencies in real – time. This effectively mitigates cell parameter dispersion and extends the battery cycle life by approximately 8%.
Technical Sophistication: Cloud – Edge Collaborative Protection
Safety serves as the bedrock of ESS and the focal point of Hoenergy's innovation efforts. Hoenergy has been a pioneer in integrating the TCM philosophy of “preventive treatment” into ESS safety R&D. Leveraging its extensive expertise in digital energy storage technologies accumulated over the years, Hoenergy introduced BMS 2.0. This system creates a safety – closed loop encompassing “dynamic monitoring, intelligent diagnosis, and active intervention” to tackle the technical challenges of precisely identifying microscopic hazards under complex operating conditions. Its core operational logic can be dissected into three key technological components:
- Comprehensive Perception and Real – Time Monitoring: Hoenergy's BMS 2.0 is equipped with a 7M (voltage, current, temperature, impedance, pressure, gas, particulate) multi – dimensional sensing system. This system can capture cell – level microscopic state parameters at a millisecond – level frequency, enabling comprehensive monitoring that spans from electrochemical characteristics to mechanical deformation. By utilizing the edge – side heterogeneous computing architecture and high – frequency ripple injection technology, the system can conduct real – time polarization analysis and accurately identify latent failure factors that are undetectable by traditional BMS.
- Intelligent Diagnosis and Prediction Network: On the edge side, the system employs an “end – edge AI double – layer fault diagnosis algorithm.” Through two methods, namely fault scoring thresholds and clustering, this double – layer fault diagnosis algorithm model can accurately identify various fault types, including micro – short circuits, low capacity issues, and connection line faults. This provides precise guidance for operation and maintenance tasks. Meanwhile, the digital twin model established in the cloud combines mechanism analysis with neural network algorithms, achieving a significant leap in three key prediction capabilities:
（1）Thermal Runaway Early Warning: A cell – level thermal runaway detection model is developed using deep learning based on an unsupervised algorithm. This model can issue temperature – based early warnings within minutes. The early warning signal directly triggers a three – level protection mechanism. Additionally, the model has cloud – based self – learning capabilities. By leveraging the accumulation of data in the abnormal database, the accuracy of the early warning model is continuously optimized and promptly synchronized to local devices.
（2）Precise Life Estimation: A single – cell – level state – of – health prediction model is established through the LSTM deep learning algorithm in combination with feature engineering. By comprehensively analyzing multiple parameters such as temperature, internal resistance, and voltage, the prediction error of the state – of – health (SOH) for a single cell is maintained within 1.5%.
（3）System – Level Safety Situation Awareness: The system can track parameters such as the voltage balance of the battery pack and the temperature gradient in real – time. Based on this data, it dynamically generates safety – boundary control strategies.
- Dynamic Balancing and Active Self – Healing: Throughout the battery's life cycle, the “health disparities” among individual battery cells can trigger a chain reaction of performance degradation, similar to the domino effect. Hoenergy adopts an interventional active balancing approach to achieve three – dimensional collaborative optimization of capacity, internal resistance, and temperature dispersion. By continuously monitoring the microscopic parameters of battery cells, the system compensates for differences through energy transfer methods, increasing the discharge depth of the energy storage system by approximately 2%.
Moreover, the system can monitor the risk of lithium plating and perform active repair operations. When the system detects that the amount of lithium plating exceeds the safety threshold or when the number of charge – discharge cycles reaches 500, the repair module is immediately activated. It emits specific high – frequency micro – current pulses to stimulate the redistribution of ions within the battery cell, effectively suppressing the growth of lithium dendrites. Experimental data shows that for battery cells that have undergone this repair treatment, the risk of lithium plating is reduced by approximately 70%, truly embodying the safety protection concept of “preventing diseases before they occur.”
Closed – Loop Management: Intelligent Evolution from Data Collection to Decision – Making
As an advanced artificial intelligence platform, DeepSeek has broad application prospects in areas such as energy storage safety and intelligent operation and maintenance. By integrating DeepSeek's large – model deep reasoning and knowledge graph technology across the cloud – edge – end architecture, Hoenergy can conduct component – level fault diagnosis for abnormal equipment conditions and generate intelligent and proactive operation and maintenance strategies. This significantly enhances the intelligence level of energy storage power station operation and maintenance work and effectively ensures the safe operation of energy storage power stations throughout their entire life cycle.
The integration of DeepSeek has revolutionized the human – machine collaboration paradigm. Its reasoning engine can analyze the battery operation data stream in real – time and automatically generate visual safety diagnosis reports that cover multiple dimensions, including operation information, consistency diagnosis, performance diagnosis, and potential risk diagnosis and prediction. Additionally, AI technology can dynamically optimize the diagnostic logic through a cognitive model formed by continuously learning from battery failure cases. When a new type of fault mode emerges, the system can autonomously update the knowledge graph and generate countermeasures within 24 hours, driving the shift of the energy storage system's operation and maintenance mode from “experience – driven” to “knowledge – driven.”
When discussing the safety cost strategy, Wen Qingwu emphasized: “The essence of our technology lies in upgrading battery management from ‘passive response' to ‘active regulation.' Through the precise division of labor among the three – level computing power of ‘cloud – edge – end,' we ensure that every battery cell remains in its optimal health state. This is the fundamental logic for maximizing the value of the energy storage system over its entire life cycle.” This technological architecture breaks through the traditional safety governance framework, transforming the industry – mainstream passive protection method of relying on hardware accumulation into an active defense system based on intelligent prediction. It also marks a paradigm shift from cost – investment – based models to asset – value – appreciation – oriented models.
Industry Leadership: Hoenergy's Safety Vision
In an era where the energy storage industry is rapidly moving towards scale and intelligence, safety is not only the bottom – line requirement but also the core proposition that determines the industry's future. As Feng Anhua, the Chairman of Hoenergy, put it, safety is not a cost but a leverage point for asset value appreciation. When every battery cell becomes a reliable profit – generating carrier, safety ceases to be a constraint on industry development and instead becomes a driving force for the energy storage industry's leapfrog evolution.
As a forerunner in adopting the 3S (BMS – PCS – EMS) full – stack self – research strategy in the energy storage industry, Hoenergy has ingrained the safety philosophy of “predictive prevention with integrated protection” into its technological foundation. Leveraging its in – house capabilities that cover everything from core components (BMS, PCS, EMS) to system integration, Hoenergy has established an All – Domain Safety Framework that encompasses R&D, product design, manufacturing, and application scenarios. This framework enables real – time cell – level tracking, module – level dynamic balancing, and system – level coordinated protection. By optimizing energy efficiency and whole – life – cycle value management, it achieves a dynamic balance between safety assurance and operational economics.
Fundamentally, the key to breakthroughs in energy storage safety lies in the deep resonance between technological innovation and industrial thinking. Only by using innovative technologies to preemptively identify latent risks and reconstructing safety defenses from a systems – thinking perspective can we establish effective barriers to prevent the spread of microscopic failure factors into macroscopic risks. This shift from post – incident response to source – level prevention, through the dialectical unity of “prevention” and “remediation,” not only unlocks the code for transcending the cycle of safety issues but also represents the inevitable path for the industry to transition from scale expansion to quality – based leapfrogging.
