We respect your privacy. Please consent to all cookies when you visit our website. Refer to our "Privacy Policy" for more information on how we handle personal data.

g1_banner_pc.webp

R&D and Innovation

Electricity Price Forecasting

Multi-model fusion-based forecasting of electricity prices and renewable energy output provides data-driven price insights for trading decisions.

  • Multi-Model Parallel Forecasting

    DNN / LightGBM / Transformer

    g_icon1.svg
  • Multi-Type Forecasting

    Covering Day-Ahead and Intraday Electricity Prices and Market Load Forecasts

    g_icon2.svg
  • Renewable Energy Output Forecasting

    PV & Wind Power Output Analysis

    g_icon3.svg
  • Model Management Center

    Monitoring · Comparison · Switching

    g_icon4.svg
g_img4_transparent.png

Forecasting Models &
Model Management Capabilities

Core algorithms address diverse scenario requirements, with unified scheduling and monitoring through the Model Management Center

  • download.png
  • g1_img_02.png
  • g1_img_02-1.png
  • Applicable Scenario: Day-Ahead Electricity Price Forecasting

    DNN (Deep Neural Network Algorithm Model)

    Suitable for high-dimensional feature learning, capable of capturing nonlinear patterns in electricity price time series, and serving as the core foundational model for electricity price forecasting. Supports joint modeling of multi-dimensional features such as historical prices, load, and weather.

  • Applicable Scenario: Short-Term Power Forecasting

    LightGBM (Gradient Boosting Tree Model)

    Fast training and highly interpretable, ideal for short-term renewable energy output forecasting, widely used in multi-site comparisons and rapid iteration scenarios.

  • Applicable Scenario: Long-Horizon Output Forecasting

    Transformer (Attention Mechanism Model)

    Models long-term temporal dependencies via the attention mechanism, excels at capturing weather-driven output trends, suitable for intraday rolling forecasts and multi-step ahead predictions.

AI Intelligent Strategy

Based on electricity price and renewable energy output forecasts Generates interpretable charging, discharging, and bidding strategies through a trading agent

  • download.png
  • Frame 1321323348.jpg
  • download.png
  • download.svg

    MILP Strategy Optimization

    Charge/Discharge Time Windows & Power Allocation

  • g_icon7.svg

    Bidding Strategy Support

    Price Boundaries & Bidding Volume Recommendations

  • g_icon8.svg

    Strategy Review & Analysis

    Execution Comparison & Experience Accumulation

  • g_icon5.svg

    Risk Constraint Management

    Boundary Validation & Compliance Alerts

From Strategy Generation to Execution:
End-to-End Workflow Overview

A complete workflow from forecast data input to manual confirmation and execution, with explainable outputs and review checkpoints at every stage

  • download.svg
    1

    Forecast Data Input

    Receives electricity price forecast curves, renewable energy output forecasts, and weather data as inputs for strategy optimization.

  • download.svg
    2

    Constraint Configuration

    Loads market bidding rules, equipment capacity limits, SOC constraints, and compliance parameters.

  • download.svg
    3

    MILP Strategy Optimization

    The Trading Agent uses Mixed-Integer Linear Programming (MILP) to optimize charging / discharging strategies and bidding plans within defined constraints.

  • download.svg
    4

    Strategy Explanation Output

    Outputs charging/discharging plans and bidding recommendations, highlighting key assumptions, risk factors, and manual review priorities.

  • download.svg
    5

    Manual Review & Confirmation

    Traders review strategy recommendations and risk alerts before confirming execution; the system does not make decisions autonomously.

RelyEZ Fairy

AI Dialogue | Ask Questions in Natural Language and Instantly Receive Professional Insights Connects knowledge bases with analytical results to help traders review strategies and identify risks

download.png
  • download.svg

    Knowledge Base

    Energy storage regulations and historical operational data with precise, one-click retrieval

  • download.svg

    AI-Powered Dialogue

    Ask questions in natural language and instantly receive professional analysis and recommendations

  • download.svg

    Task Structuring

    Automatically decomposes complex tasks into actionable, structured task lists

Three Core Scenarios Covering Key
Energy Storage Operational Needs

Capability Overview: Transforms natural language queries into multi-step tasks and automatically selects the appropriate tools for execution

  • download.svg
    1

    Ask

    Traders can ask questions about markets, weather, pricing, or strategies in natural language without needing to understand the underlying system architecture.

  • download.svg
    2

    Retrieve

    The system automatically retrieves relevant context from policy and regulatory databases, weather databases, forecasting databases, and market databases.

  • download.svg
    3

    Analyze

    It leverages electricity price forecasting, renewable energy output forecasting, supply-demand analysis, and similar scenario identification capabilities for in-depth analysis.

  • download.svg
    4

    Explain

    The LLM interprets optimization results generated by the Trading Agent, explaining the rationale behind strategies, key constraints, and major risks.

  • download.svg
    5

    Refine

    It provides risk alerts, highlights priorities for manual review, and recommends strategy refinements, with final execution confirmed by the trader.

Immersion Liquid- Cooled Energy Storage Battery System

RelyEZ's self-developed immersion-cooled battery modules use optimized series-parallel configurations and a dual-path thermal management design combining immersion temperature equalization with cold-plate heat exchange. Direct contact between the coolant and heat sources improves cooling efficiency, minimizes temperature differences, and reduces thermal runaway propagation risks, enhancing safety and durability in high-density, large-scale energy storage systems.

download.jpg
  • download.svg

    Superior Temperature Uniformity

    Eliminates excessive temperature differences between the top and bottom of battery cells in conventional liquid-cooled solutions, achieving a system-level cell temperature difference of ≤3°C.

  • download.svg

    Intrinsic Safety

    The immersion coolant isolates oxygen to prevent combustion and rapidly absorbs heat to prevent heat accumulation and thermal propagation, enhancing overall system safety.

  • download.svg

    Extended Service Life

    By significantly improving operating temperature uniformity, the usable battery capacity and cycle life are enhanced.

  • download.svg

    Enhanced Efficiency

    Efficient heat dissipation reduces system cooling power demand; intelligent O&M precisely regulates system power consumption for optimized energy efficiency.

Iterative Upgrades of Immersion Liquid-Cooled
Energy Storage Solutions

  • Generation 1

    Shared Coolant Tank Solution for Multiple Battery Clusters

    The manufacturing process is extremely complex, making overall maintenance of the energy storage system highly challenging. Furthermore, high internal pressure significantly increases the risk of coolant leakage caused by poor sealing.

  • Generation 2

    Single Battery Cluster-Level Spray Cooling Solution

    Compared with the first-generation solution, efficiency is improved and temperature uniformity is optimized, but the original limitations remain.

  • Generation 3

    Module-Level Dual-Path Coupled Solution

    The simplified manufacturing process facilitates energy storage system maintenance, while lower internal pressure greatly reduces leakage risk. Reduced immersion coolant circulation energy consumption improves efficiency, and shorter heat transfer paths further enhance temperature uniformity.

download.jpg