Solar solid-state battery collaborative energy management system and method based on AI optimization

Abstract

The invention relates to the technical field of energy management, and discloses a solar solid-state battery collaborative energy management system and method based on AI optimization. The system comprises a photovoltaic-battery cooperative controller which is used for executing an energy scheduling algorithm, an edge calculation unit, a chip carrying a horizon J5, a digital twinning module and supporting MATLAB / Simulink real-time simulation, wherein the calculation power is larger than or equal to 128 TOPS, and the simulation error is smaller than or equal to 1%.

Description

Technical Field

[0001] The present invention relates to the technical field of energy management, and specifically to a solar solid-state battery collaborative energy management system and method optimized based on AI. Background Art

[0002] Defects of the Existing Technology

[0003] Energy collaboration bottleneck: The energy conversion efficiency of the photovoltaic-battery system ≤ 75%, and the utilization rate of photovoltaic power generation fluctuates by ± 30% under dynamic conditions.

[0004] Shortcomings of solid-state batteries: The interfacial impedance of the lithium metal negative electrode is too high (> 10 Ω·cm at room temperature 2 ), and the cycle life is affected by the SOC estimation error (when the error > 5%, the attenuation accelerates by 40%).

[0005] Limitations of AI applications: Existing algorithms cannot achieve a photovoltaic power prediction accuracy > 90%, and the multi-scenario adaptability is poor (for example, the efficiency drops by > 40% in extremely cold environments). Summary of the Invention

[0006] Technical Problems to be Solved

[0007] In view of the deficiencies of the existing technology, the present invention provides a solar solid-state battery collaborative energy management system and method optimized based on AI.

[0008] Technical Solution

[0009] To achieve the above object, the present invention provides the following technical solution:

[0010] A solar solid-state battery collaborative energy management system and method optimized based on AI, comprising, characterized in that, a) a photovoltaic-battery collaborative controller for executing an energy scheduling algorithm;

[0011] b) an edge computing unit equipped with a Horizon J5 chip, with a computing power ≥ 128 TOPS;

[0012] c) a digital twin module that supports MATLAB / Simulink real-time simulation, with a simulation error ≤ 1%.

[0013] As a further solution of the present invention, the photovoltaic module is a perovskite / silicon tandem battery, with a photoelectric conversion efficiency ≥ 32.1% and passing the IEC 61215:2021 certification.

[0014] As a further solution of the present invention, the solid-state battery uses a sulfide electrolyte and a lithium metal negative electrode, with an energy density ≥ 450 Wh / kg and an interfacial impedance ≤ 4 Ω·cm 2 .

[0015] As a further solution of the present invention, the edge layer communication protocol is 5G NR V2X, with a time delay ≤ 50 ms, supporting MQTT+DDS hybrid communication, and a reliability ≥ 99.99%.

[0016] As a further solution of the present invention, the AI platform is a TensorFlow Edge TPU cluster, and the model inference delay ≤ 200 ms.

[0017] As a further solution of the present invention, a method for reducing the interfacial impedance of a lithium metal anode optimized by AI is characterized in that:

[0018] a) Regulating the growth rate of the SEI film through reinforcement learning to reduce the interfacial impedance by ≥ 60%;

[0019] b) The detection sensitivity of lithium dendrites reaches the 1-μm level, based on piezoelectric sensing technology.

[0020] As a further solution of the present invention, a multimodal photovoltaic power prediction method is characterized in that:

[0021] a) The input features include the solar altitude angle, cloud cover, temperature, and historical power generation data;

[0022] b) The output 24-hour power prediction error ≤ 5%.

[0023] As a further solution of the present invention, a method for energy management based on federated learning is characterized in that:

[0024] a) The number of federated learning rounds ≥ 100 rounds, and the global model aggregation interval ≤ 30 min;

[0025] b) The privacy budget satisfies ε = 8, δ = 10-5.

[0026] Beneficial effects

[0027] Compared with the prior art, the present invention provides a solar solid-state battery collaborative energy management system and method optimized by AI, having the following beneficial effects:

[0028] Architectural innovation

[0029] The first "photovoltaic-battery-vehicle network" three-level collaborative architecture is created, and the energy flow loss is reduced by 22%.

[0030] The photovoltaic output power attenuation ≤ 15% in an extremely cold environment (-40 °C), and the system availability ≥ 98.7%.

[0031] Algorithm innovation

[0032] The multi-physical field coupling model improves the electrochemical-thermal-mechanical simulation accuracy to the micron level.

[0033] Reinforcement learning SOC scheduling extends the battery cycle life by 30%.

[0034] Engineering innovation

[0035] The self-healing solid electrolyte uses AI to predict crack generation and trigger in-situ repair, with a repair efficiency > 95%.

[0036] The fault degradation mechanism ensures that the local decision-making delay < 100ms when communication is interrupted.

[0037] Performance improvement

[0038] The daily average power generation increases by ≥ 27% (12.5 kWh compared to 9.8 kWh), and the power consumption per 100 kilometers decreases by ≥ 22% (12.1 kWh compared to 15.6 kWh).

[0039] The cruising range attenuation in extremely cold environments ≤ 12% (compared to 28% in the control group), and the system stability is significantly improved. Brief description of the drawings

[0040] Figure 1 It is the system architecture diagram of a solar solid-state battery collaborative energy management system and method based on AI optimization proposed by the present invention;

[0041] Figure 2 It is the algorithm flowchart of a solar solid-state battery collaborative energy management system and method based on AI optimization proposed by the present invention;

[0042] Figure 3 It is the comparative diagram of the example data of a solar solid-state battery collaborative energy management system and method based on AI optimization proposed by the present invention. Detailed implementation manners

[0043] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the following further details the present invention through examples and in conjunction with the accompanying drawings. It should be understood that the specific examples described herein are only used to explain the present invention and are not used to limit the present invention.

[0044] The serial numbers assigned to components in this text itself, such as "first", "second", etc., are only used to distinguish the described objects and do not have any sequential or technical meaning. And the "connection" and "coupling" mentioned in the present invention, unless otherwise specified, both include direct and indirect connection (coupling). In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present invention.

[0045] In the present invention, unless otherwise clearly specified and limited, the first feature being "on" or "under" the second feature may be that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. Moreover, the first feature being "above", "over" and "on top of" the second feature may be that the first feature is directly above or obliquely above the second feature, or merely indicates that the first feature has a higher horizontal height than the second feature. The first feature being "under", "beneath" and "underneath" the second feature may be that the first feature is directly below or obliquely below the second feature, or merely indicates that the first feature has a lower horizontal height than the second feature.

[0046] Refer to Figures 1 - 3 , a solar solid-state battery collaborative energy management system and method based on AI optimization, including, characterized in that, a) a photovoltaic-battery collaborative controller for executing an energy scheduling algorithm;

[0047] b) an edge computing unit equipped with a Horizon J5 chip with a computing power ≥ 128 TOPS;

[0048] c) a digital twin module supporting MATLAB / Simulink real-time simulation with a simulation error ≤ 1%.

[0049] Specifically, the photovoltaic module is a perovskite / silicon tandem battery with a photoelectric conversion efficiency ≥ 32.1% and passing the IEC61215:2021 certification.

[0050] Specifically, the solid-state battery uses a sulfide electrolyte and a lithium metal negative electrode with an energy density ≥ 450 Wh / kg and an interfacial impedance ≤ 4 Ω·cm 2 .

[0051] Specifically, the edge layer communication protocol is 5G NR V2X with a delay ≤ 50 ms, supporting MQTT+DDS hybrid communication with a reliability ≥ 99.99%.

[0052] Specifically, the AI platform is a TensorFlow Edge TPU cluster, and the model inference latency ≤ 200 ms.

[0053] Specifically, a method for reducing the interfacial impedance of a lithium metal anode optimized by AI, characterized in that:

[0054] a) Regulate the SEI film growth rate through reinforcement learning to reduce the interfacial impedance by ≥ 60%;

[0055] b) The detection sensitivity of lithium dendrites reaches the 1-μm level, based on piezoelectric sensing technology.

[0056] Specifically, a multimodal photovoltaic power prediction method, characterized in that:

[0057] a) The input features include the solar altitude angle, cloud cover, temperature, and historical power generation data;

[0058] b) The output 24-hour power prediction error ≤ 5%.

[0059] Specifically, an energy management method based on federated learning, characterized in that:

[0060] a) The number of federated learning rounds ≥ 100 rounds, and the global model aggregation interval ≤ 30 min;

[0061] b) The privacy budget satisfies ε = 8, δ = 10-5.

[0062] Furthermore, 1. The beneficial effects of the photovoltaic-battery co-controller and the edge computing unit

[0063] Through the combination of the photovoltaic-battery co-controller and the edge computing unit, the present invention solves the problem of insufficient coordination between the photovoltaic and battery systems in the prior art. First of all, the co-controller can optimize the input of photovoltaic energy and the charge-discharge strategy of the solid-state battery in real time to ensure the efficient distribution of the energy flow. The edge computing unit is equipped with a Horizon J5 chip (INT8 computing power of 128 TOPS), which has powerful real-time computing capabilities and can complete complex energy scheduling tasks within milliseconds. This architecture not only improves the response speed of the system but also significantly reduces the latency and bandwidth requirements of cloud communication.

[0064] In practical applications, the system improves the utilization rate of photovoltaic power generation to over 95% through a multimodal photovoltaic power prediction algorithm, solving the problem of fluctuations in the utilization rate of photovoltaic power generation under dynamic working conditions. At the same time, the co-controller can dynamically adjust the charging strategy according to the real-time state of the battery (such as SOC, temperature, etc.) to extend the battery cycle life. In addition, the local processing ability of the edge computing unit ensures that the system can still operate stably during communication interruptions, improving the reliability and availability of the system.

[0065] In extreme environments (such as extremely cold or high temperatures), the collaborative controller combines a self-heating coating and a preheating strategy to further optimize the performance of photovoltaic panels and batteries. For example, in an extremely cold environment of -40°C, the system availability can still reach 98.7%, far higher than 83% of the traditional solution. This collaborative architecture provides technical support for the stable operation of new energy vehicles in complex environments and lays a foundation for energy management in the vehicle-to-grid (V2G) scenario.

[0066] 2. Beneficial effects of perovskite / silicon tandem cells

[0067] The present invention uses a perovskite / silicon tandem cell as the core of the photovoltaic module, with a photoelectric conversion efficiency of 32.1% and passing the IEC 61215:2021 certification, ensuring the reliability and durability of the module. Compared with traditional single-crystalline silicon cells (efficiency about 22%), the perovskite / silicon tandem cell can generate higher power generation in the same area, significantly improving the vehicle's self-power generation ability.

[0068] In practical applications, this cell has a wider spectral absorption range, especially performs well under low-light and scattered light conditions, and can achieve higher power generation efficiency in complex lighting environments such as urban roads and cloudy days. For example, in the urban road scenario, the daily power generation of the present invention reaches 12.5 kWh, while the traditional solution is only 9.8 kWh, with a 27% increase in power generation. In addition, the perovskite / silicon tandem cell has excellent low-temperature performance, and the output power attenuation at extremely cold temperatures (-40°C) is only 15%, far lower than 28% of the traditional solution. This performance advantage significantly improves the vehicle's endurance in extremely cold regions and provides technical support for the global deployment of new energy vehicles.

[0069] 3. Beneficial effects of sulfide electrolyte and lithium metal anode

[0070] The solid-state battery of the present invention uses a sulfide electrolyte and a lithium metal anode, with an energy density of 450 Wh / kg and the interfacial impedance reduced to 4 Ω·cm 2 below. Compared with traditional liquid batteries (energy density about 250 Wh / kg), this solid-state battery has significant advantages in terms of energy density and safety. The sulfide electrolyte has high ionic conductivity and good mechanical properties, which can effectively inhibit the growth of lithium dendrites and prevent internal short circuits in the battery.

[0071] With the AI-optimized lithium metal interface regulation algorithm, the interface impedance is further reduced by 60%, significantly improving the charge-discharge efficiency and cycle life of the battery. Experimental data shows that the cycle life of the battery of the present invention reaches more than 4000 times, while the traditional solution is only about 2000 times. In addition, the system controls the SOC estimation error (≤2%), avoiding the problems of overcharging or over-discharging of the battery caused by estimation errors, and further extending the battery life. In extreme environments (such as high temperature or low temperature), the thermal stability of the solid-state battery is better than that of the liquid battery, ensuring the safety and reliability of the system.

[0072] Beneficial effects of the 4.5G NR V2X communication protocol

[0073] The present invention adopts the 5G NR V2X communication protocol, with a latency lower than 50ms, supporting MQTT+DDS hybrid communication, and a reliability reaching 99.99%. Compared with the traditional 4G communication protocol (latency > 100ms), 5G NR V2X significantly improves the real-time performance and response speed of the system, providing technical support for vehicle-to-grid interaction (V2G) and vehicle-to-vehicle communication (V2V).

[0074] In practical applications, this communication protocol can achieve real-time interaction between vehicles and the power grid and optimize the energy distribution strategy. For example, in the urban road scenario, the system collaborates with the power grid through 5G communication, reducing the power consumption per 100 kilometers to 12.1kWh, while the traditional solution is 15.6kWh, with a 22% reduction in power consumption. In addition, the MQTT+DDS hybrid communication protocol ensures high reliability and low latency of data transmission, supporting the stable operation of vehicles in high-speed driving or complex traffic scenarios. When the communication is interrupted, the system ensures a latency lower than 100ms through the local decision-making ability of the edge computing unit, further enhancing the robustness of the system.

[0075] Beneficial effects of the TensorFlow Edge TPU cluster

[0076] The AI platform of the present invention adopts the TensorFlow Edge TPU cluster, with a model inference latency lower than 200ms, supporting federated learning and real-time simulation. Compared with the traditional cloud AI platform (inference latency > 500ms), this architecture significantly improves the real-time performance and scalability of the system.

[0077] In practical applications, the Edge TPU cluster reduces the latency and bandwidth requirements of cloud communication through local inference, while enhancing data privacy. For example, after 100 rounds of global model aggregation through federated learning, the photovoltaic power prediction accuracy of the system reaches over 95%, while that of the traditional solution is only about 70%. In addition, the digital twin module combines MATLAB / Simulink real-time simulation to control the error of electrochemistry-thermal-mechanical simulation within 1%, providing a high-precision virtual test environment for the optimal design of the system. This architecture not only improves the intelligence level of the system but also provides technical support for multi-scenario adaptation and large-scale deployment.

[0078] 6. Beneficial Effects of the Lithium Metal Interface Optimization Algorithm

[0079] The present invention regulates the growth rate of the SEI film through reinforcement learning, reducing the interfacial impedance by 60%. At the same time, the piezoelectric sensing technology is used to achieve a lithium dendrite detection sensitivity of 1 μm level. This algorithm significantly improves the safety and cycle life of solid-state batteries.

[0080] In practical applications, the optimized growth of the SEI film can effectively inhibit the penetration of lithium dendrites and prevent internal short circuits in the battery. Experimental data shows that the battery cycle life of the present invention reaches more than 4000 times, while that of the traditional solution is only about 2000 times. In addition, the 1 μm level lithium dendrite detection sensitivity enables the system to identify potential short circuit risks at an early stage and adjust the charging strategy in a timely manner, further improving the safety of the battery. In extreme environments (such as high temperature or low temperature), this algorithm combined with self-healing electrolyte technology ensures the stability and reliability of the battery.

[0081] 7. Beneficial Effects of the Multimodal Photovoltaic Power Prediction Algorithm

[0082] The multimodal photovoltaic power prediction algorithm of the present invention controls the 24-hour power prediction error within 5% by fusing features such as solar altitude angle, cloud cover, temperature, and historical power generation data. Compared with traditional prediction models (error > 15%), this algorithm significantly improves the prediction accuracy and provides a reliable basis for the energy scheduling of the system.

[0083] In practical applications, the high-precision power prediction enables the system to plan the charging strategy in advance and optimize the energy distribution. For example, in the urban road scenario, the system improves the daily power generation to 12.5 kWh through accurate prediction of photovoltaic power generation, while the traditional solution is only 9.8 kWh. In addition, this algorithm combined with federated learning technology can achieve global optimization in the multi-vehicle collaboration scenario, further improving the overall efficiency of the system. In extremely cold environments, the algorithm combined with self-heating coating technology ensures the power generation performance of photovoltaic panels at low temperatures and provides technical support for the stable operation of vehicles.

[0084] 8. Beneficial Effects of the Federated Learning Energy Management Method

[0085] Through the federated learning method, the present invention realizes the collaborative optimization of energy management for multiple vehicles while protecting data privacy. Through 100 rounds of global model aggregation, the privacy budget meets ε = 8 and δ = 10-5, ensuring the security of data and the accuracy of the model.

[0086] In practical applications, the federated learning method avoids the direct transmission of sensitive data through distributed training and improves the generalization ability of the model at the same time. For example, in the urban road scenario, the system optimizes the energy scheduling strategy through federated learning, reducing the electricity consumption per 100 kilometers to 12.1 kWh, while the traditional solution is 15.6 kWh. In addition, the combination of federated learning and reinforcement learning algorithms can dynamically adjust the energy allocation strategy in multiple scenarios, improving the overall efficiency of the system. In large-scale deployment scenarios, this method significantly reduces the data transmission and storage costs, providing technical support for the global deployment of intelligent connected vehicles.

[0087] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0088] The above-described embodiments merely represent several implementation manners of the present invention, and their descriptions are relatively specific and detailed, but should not be construed as limiting the scope of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the appended claims.

Claims

1. A solar solid-state battery collaborative energy management system and method optimized based on AI, comprising, characterized in that, a) A photovoltaic-battery collaborative controller for executing an energy scheduling algorithm; b) An edge computing unit equipped with a Horizon J5 chip, with computing power ≥ 128 TOPS; c) A digital twin module that supports MATLAB / Simulink real-time simulation, with a simulation error ≤ 1%.

2. The collaborative energy management system and method of a solar solid-state battery based on AI optimization according to claim 1, wherein The photovoltaic module is a perovskite / silicon tandem cell, with a photoelectric conversion efficiency ≥ 32.1% and is certified by IEC61215:2021.

3. A collaborative energy management system and method for a solar solid-state battery optimized based on AI according to claim 1, characterized in that, The solid-state battery uses a sulfide electrolyte and a lithium metal anode, with an energy density ≥ 450 Wh / kg and an interfacial impedance ≤ 4 Ω·cm 2 .

4. A collaborative energy management system and method for a solar solid-state battery optimized based on AI according to claim 1, characterized in that, The edge layer communication protocol is 5G NR V2X, with a latency ≤ 50 ms, supports MQTT+DDS hybrid communication, and the reliability ≥ 99.99%.

5. A collaborative energy management system and method for a solar solid-state battery optimized based on AI according to claim 1, characterized in that, The AI platform is a TensorFlow Edge TPU cluster, with a model inference latency ≤ 200 ms.

6. A method for reducing the interfacial impedance of a lithium metal anode optimized based on AI according to claim 1, characterized in that: a) Regulating the growth rate of the SEI film through reinforcement learning to reduce the interfacial impedance by ≥ 60%; b) The detection sensitivity of lithium dendrites reaches the 1-μm level, based on piezoelectric sensing technology.

7. A multi-modal photovoltaic power prediction method according to claim 1, characterized in that: a) The input features include the solar altitude angle, cloud cover, temperature, and historical power generation data; b) The output 24-hour power prediction error ≤ 5%.

8. A method for energy management based on federated learning according to claim 1, characterized in that: a) The number of federated learning rounds ≥ 100 rounds, and the global model aggregation interval ≤ 30 min; b) The privacy budget satisfies ε = 8, δ = 10-5.

Images