Distributed energy storage safety supervision method, system and cloud platform based on big data

Abstract

The present invention provides a distributed energy storage safety supervision method, system and cloud platform based on big data, belonging to the field of energy storage safety technology. The method includes collecting first-level data of the PCS layer and second-level data of the BMS layer based on the distributed energy storage system, and performing data fusion on the first-level data and the second-level data; performing cloud dynamic risk assessment on the distributed energy storage system according to the data fusion result, using the risk assessment result to call the matching strategy, performing multi-system linkage based on the matching strategy, and obtaining the linkage effect; performing multi-dimensional evaluation on the linkage effect, optimizing and adjusting the optimization strategy, and at the same time performing safety warning according to the linkage effect, and triggering corresponding protection measures to form safety supervision, realizing active safety protection of the energy storage system, improving the response speed, reducing the failure rate, and comprehensively improving the safety and intelligent level.

Description

Technical Field

[0001] The present invention relates to the technical field of energy storage safety, and particularly to a distributed energy storage safety supervision method, system and cloud platform based on big data. Background Art

[0002] With the increasing proportion of intermittent renewable energy sources such as wind power and photovoltaic power, distributed energy storage systems (such as battery energy storage) have become key facilities for power grid frequency regulation and peak shaving and valley filling. In the prior art, the data of BMS and PCS are separated, resulting in misjudgment of risks, and static strategies cannot adapt to changes such as battery attenuation and grid fluctuations.

[0003] Therefore, the present invention provides a distributed energy storage safety supervision method, system and cloud platform based on big data. Summary of the Invention

[0004] The distributed energy storage safety supervision method, system and cloud platform based on big data provided by the present invention construct a panoramic state of the system through multi-source data fusion, dynamically evaluate short-term / long-term risks in the cloud and generate optimization strategies, which are verified by digital twin simulation and then issued for execution; the linkage effect is fed back to the strategy iteration through multi-dimensional evaluation, and combined with dynamic threshold alarm to form a perception-decision-execution-optimization closed loop, realizing active safety protection of the energy storage system, improving the response speed, reducing the failure rate, and comprehensively improving the safety and intelligent level.

[0005] The distributed energy storage safety supervision method based on big data provided by the present invention includes:

[0006] Step 1: Collect the first-level data of the PCS layer and the second-level data of the BMS layer based on the distributed energy storage system, and perform data fusion on the first-level data and the second-level data;

[0007] Step 2: Perform cloud dynamic risk assessment on the distributed energy storage system according to the data fusion result, call the matching strategy using the risk assessment result, perform multi-system linkage based on the matching strategy, and obtain the linkage effect;

[0008] Step 3: Perform multi-dimensional evaluation on the linkage effect, optimize and adjust the optimization strategy, and at the same time perform safety alarm according to the linkage effect, and trigger corresponding protection measures to form safety supervision.

[0009] The distributed energy storage safety supervision method based on big data provided by the present invention determines the data acquisition layer based on the technical framework of the distributed energy storage system, collects the first-level data of the PCS layer and the second-level data of the BMS layer, and performs data processing on the first-level data and the second-level data, including:

[0010] Design the first data item regarding the PCS layer based on the distributed energy storage system, determine the first acquisition unit and the first communication method according to the first data item, and then acquire the first-level data;

[0011] Design the second data item regarding the BMS layer based on the distributed energy storage system, determine the second acquisition unit and the second communication method according to the second data item, and then acquire the second-level data;

[0012] Perform cross-level data fusion on the first-level data and the second-level data.

[0013] The distributed energy storage safety supervision method based on big data provided by the present invention performs cross-level data fusion on the first-level data and the second-level data, including:

[0014] Perform the first scenario label association on the first-level data to obtain the first association result, and perform the second scenario label association on the second-level data to obtain the second association result;

[0015] Determine the conflict type of the distributed energy storage system according to the first association result and the second association result, and determine the dynamic conflict monitoring algorithm based on the conflict type;

[0016] Perform data coupling on the first-level data and the second-level data based on the first association result and the second association result to obtain a coupling result;

[0017] Extract the first feature vector of the first-level data, extract the second feature vector of the second-level data, match the corresponding fusion method from the feature-fusion method table, and perform data fusion on the first feature vector and the second feature vector using the fusion method based on the coupling result and the dynamic conflict monitoring algorithm.

[0018] The distributed energy storage safety supervision method based on big data provided by the present invention performs cloud dynamic risk assessment on the distributed energy storage system according to the data fusion result, calls the matching strategy using the risk assessment result, performs multi-system linkage based on the matching strategy, and obtains the linkage effect, including:

[0019] The cloud platform receives the data fusion result, uses the risk assessment model to evaluate the short-term risk and long-term risk of the distributed energy storage system, and obtains the risk assessment result;

[0020] Use the risk assessment result to match the prefabricated strategy from the rule library. If the match fails, call the reinforcement learning model to generate a temporary strategy, and combine the prefabricated strategy and the temporary strategy to obtain the matching strategy;

[0021] Convert the matching strategy into a standard instruction, split the standard instruction, perform multi-system linkage according to the splitting result, and set a fault tolerance mechanism at the same time;

[0022] Build a digital twin model based on the data fusion result, use the digital twin model to simulate and verify the matching strategy and the multi-system linkage process, and determine the linkage effect according to the verification result.

[0023] The distributed energy storage safety supervision method based on big data provided by the present invention sets a fault tolerance mechanism, including:

[0024] Formulate unit fault tolerance measures according to the unit structure existing in the distributed energy storage system, formulate multi-level linkage fault tolerance measures according to multi-system linkage, and set a fault tolerance mechanism by synthesizing the unit fault tolerance measures and the multi-level linkage fault tolerance measures.

[0025] The distributed energy storage safety supervision method based on big data provided by the present invention builds a digital twin model based on the data fusion result, uses the digital twin model to simulate and verify the matching strategy and the multi-system linkage process, and determines the linkage effect according to the verification result, including:

[0026] Overlay the simulation verification curve and the actual execution curve, and calculate the fitting result;

[0027] Decompose the matching strategy into strategy actions that can be mapped to the digital twin model, perform causal inference on the strategy actions and the existing linkages, and calculate the relevant results;

[0028] Quantify the strategy actions and the existing linkages into a directed graph based on the relevant results and the fitting result, analyze the directed graph, and determine the critical path, invalid actions, positive feedback loops and negative feedback loops;

[0029] Perform the first optimization of the matching strategy based on the critical path and invalid actions, perform the second optimization of the matching strategy according to the positive feedback loop and negative feedback loop, and determine the optimized strategy by synthesizing the first optimization and the second optimization;

[0030] Use the digital twin model to synchronously simulate and verify the matching strategy and the optimized strategy to obtain the linkage effect.

[0031] The distributed energy storage safety supervision method based on big data provided by the present invention performs multi-dimensional evaluation on the linkage effect, optimizes and adjusts the optimized strategy, and at the same time issues a safety warning according to the linkage effect and triggers corresponding protection measures to form safety supervision, including:

[0032] Determine the evaluation dimension of the linkage effect from the effect-evaluation table in combination with the data fusion result, obtain the historical linkage effect to optimize the dynamic threshold, and determine the multi-dimensional evaluation result of the linkage effect based on the evaluation dimension and the threshold optimization;

[0033] Determine the cause of the linkage effect based on the multi-dimensional evaluation result, alarm the failed linkage, verify the execution device based on the cause, and conduct safety supervision on the execution device.

[0034] The distributed energy storage safety supervision system based on big data provided by the present invention includes:

[0035] A fusion module: collecting the first-level data of the PCS layer and the second-level data of the BMS layer based on the distributed energy storage system, and performing data fusion on the first-level data and the second-level data;

[0036] A linkage module: performing a cloud dynamic risk assessment on the distributed energy storage system according to the data fusion result, calling a matching strategy using the risk assessment result, performing multi-system linkage based on the matching strategy, and obtaining a linkage effect;

[0037] A supervision module: performing multi-dimensional evaluation on the linkage effect, optimizing and adjusting the optimization strategy, simultaneously performing safety alarm according to the linkage effect, and triggering corresponding protection measures to form safety supervision.

[0038] The distributed energy storage safety supervision cloud platform based on big data provided by the present invention, the cloud platform includes a distributed energy storage system, a digital twin model:

[0039] The distributed energy storage system includes a PCS layer and a BMS layer;

[0040] The digital twin model includes the simulation of the linkage process and the simulation of the risks existing during the operation of the distributed energy storage system.

[0041] Compared with the prior art, the beneficial effects of the present application are as follows: constructing a panoramic state of the system through multi-source data fusion, dynamically evaluating short-term / long-term risks in the cloud and generating an optimization strategy, which is verified by digital twin simulation and then sent down for execution; the linkage effect is fed back to the strategy iteration through multi-dimensional evaluation, combined with dynamic threshold alarm to form a perception-decision-execution-optimization closed loop, realizing the active safety protection of the energy storage system, improving the response speed, reducing the failure rate, and comprehensively improving the safety and intelligent level.

[0042] Other features and advantages of the present invention will be described in the following specification, and part of them will become obvious from the specification or be understood by implementing the present invention. The purpose and other advantages of the present invention can be realized and obtained through the structure specifically pointed out in the written specification and the drawings.

[0043] The technical solution of the present invention will be further described in detail below through the drawings and embodiments. Description of the Drawings

[0044] The accompanying drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention, and do not constitute a limitation to the present invention. In the accompanying drawings:

[0045] Figure 1 is a schematic flowchart of a big data-based distributed energy storage safety supervision method provided by an embodiment of the present invention;

[0046] Figure 2 is a schematic structural diagram of a big data-based distributed energy storage safety supervision system provided by an embodiment of the present invention;

[0047] Figure 3 is a schematic structural diagram of a big data-based distributed energy storage safety supervision cloud platform provided by an embodiment of the present invention;

[0048] Figure 4 is a schematic technical structural diagram of an energy storage system provided by an embodiment of the present invention. Detailed Embodiments

[0049] The following describes the preferred embodiments of the present invention with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0050] Embodiment 1: The embodiment of the present invention provides a big data-based distributed energy storage safety supervision method, as Figure 1 shown, including:

[0051] Step 1: Collect the first-level data of the PCS layer and the second-level data of the BMS layer based on the distributed energy storage system, and perform data fusion on the first-level data and the second-level data;

[0052] Step 2: Perform cloud dynamic risk assessment on the distributed energy storage system according to the data fusion result, use the risk assessment result to call the matching strategy, perform multi-system linkage based on the matching strategy, and obtain the linkage effect;

[0053] Step 3: Perform multi-dimensional evaluation on the linkage effect, optimize and adjust the optimization strategy, and at the same time perform safety warning according to the linkage effect, and trigger corresponding protection measures to form safety supervision.

[0054] In this embodiment, the big data-based distributed energy storage safety supervision cloud platform is characterized in that the cloud platform includes a distributed energy storage system and a digital twin model. The distributed energy storage system includes a PCS layer and a BMS layer; the digital twin model includes the simulation of the linkage process and the simulation of the risks existing during the operation of the distributed energy storage system.

[0055] In this embodiment, the distributed energy storage system refers to an energy management system composed of multiple decentralized energy storage units, such asFigure 4 As shown, the battery unit, i.e., the battery pack, uses lithium iron phosphate batteries, with a single cell capacity of 314 Ah. Each unit consists of 16 cells connected in series, storing 1004.8 Wh of electricity. The battery management system BMS has a three - level architecture, namely the BMU sub - control unit, the BCU main - control unit, and the BAU general - control unit. BMU: Each battery unit integrates a BMU sub - control unit, which collects the voltage and temperature of individual batteries and uploads them to the BCU. At the same time, it receives the fan on / off commands sent by the BCU. BCU: Collects the total voltage on the battery side, the total voltage on the PCS side, the charge - discharge current, and estimates the SOC, etc. Controls the closing or opening of the main positive, main negative, and pre - charge contactors. Receives battery data from the BMU via CAN. Exchanges information with the BAU via CAN. Performs battery protection logic judgment. BAU: Collects the working position status of each switch and the position status of the rotary switch. Exchanges information with the BCU via CAN. Communicates with the PCS via CAN to upload battery information. Controls the PCS working mode via 485 communication with the PCS and controls the PCS operation through a time strategy. Uploads data information to the LCD via 485 and receives LCD setting information. Reads the meter data via 485 communication with the metering meter. Exchanges data with the data terminal via 232 and simultaneously receives the time - synchronization information from the background. The energy management system EMS and the general - control unit BAU of the battery management BMS are integrated, reducing the technical architecture links and saving costs. The charge - discharge strategy is set through the touch - screen display unit and sent to the general - control unit BAU / EMS. The general - control unit executes according to the strategy and sends corresponding commands to the PCS.

[0056] In this embodiment, the PCS layer is the power conversion system layer. For example, a 500 - kW bidirectional converter of a certain brand is responsible for AC - DC power conversion and grid interaction.

[0057] In this embodiment, the BMS layer is the battery management system layer. For example, an intelligent management system for a 24 - series lithium - battery pack.

[0058] In this embodiment, cross - layer data fusion establishes the semantic association between PCS and BMS data through scenario - tag association to identify system - level conflict types. Based on the conflict characteristics, a dynamic monitoring algorithm is selected to achieve data coupling and feature - vector fusion. For example, fusing the real - time 500 - kW discharge demand of the PCS and the battery - pack SOC = 80% data reported by the BMS, calculating the optimal discharge power threshold of 400 kW through an optimization algorithm. For example, achieving micro - second - level time alignment using the Beidou satellite clock, and then realizing deep data coupling through feature - vector extraction and deep - learning models, and finally outputting the system - level optimal control strategy.

[0059] In this embodiment, the first-level data are the real-time operating parameters of the power conversion system, reflecting the grid interaction and energy conversion status, and the second-level data are the core battery parameters monitored by the battery management system, reflecting the cell health and safety status.

[0060] In this embodiment, the data fusion result is a standardized data set integrated across systems. For example, the conclusion of the dischargeable power of 400 kW is generated by fusing the 500 kW discharge instruction of the PCS and the 80% SOC data reported by the BMS.

[0061] In this embodiment, the risk assessment result is a quantitative output of the system safety status. For example, the short-term risk shows that the probability of overheating of battery pack No. 3 within 10 minutes is 72%, and the long-term risk predicts that the battery pack capacity will decay to 82% after 6 months.

[0062] In this embodiment, the matching strategy is the final determined execution plan, which may be a direct call of the prefabricated strategy or a combination of the prefabricated strategy and the temporary strategy.

[0063] In this embodiment, the input of the digital twin model is the simulation initial conditions including real-time data (current battery temperature) and policy parameters (the load reduction amplitude to be executed); the output of the digital twin model has predictive linkage. For example, it shows that after executing the current policy, the temperature will drop to the safe range within 8 minutes.

[0064] In this embodiment, the multi-system linkage is the collaborative execution across subsystems. Typically, it is a joint action such as simultaneously adjusting the PCS power, the BMS protection threshold, and the air-conditioning refrigeration power.

[0065] In this embodiment, the linkage effect is the actual system response after the execution of the strategy. For example, the load reduction instruction causes the battery temperature to drop by 8°C within 30 minutes without affecting the grid stability.

[0066] In this embodiment, the cloud platform drives the risk assessment model through fused data, analyzes short-term and long-term risks in real time; generates the optimal strategy based on the rule base and reinforcement learning, disassembles it into multi-system collaborative instructions after being verified by the digital twin simulation, and embeds a fault tolerance mechanism to achieve closed-loop control of assessment - decision - verification - execution.

[0067] In this embodiment, a dynamic assessment system is constructed by fusing real-time data and historical effects, and the linkage performance is quantified based on multi-dimensional indicators; the threshold self-optimization algorithm is used to identify anomalies, locate the root cause of faults, and trigger accurate alarms.

[0068] In this embodiment, the multi-dimensional assessment is a multi-angle quantitative analysis system for the linkage effect of the energy storage system. By fusing data characteristics and dynamic threshold optimization, it comprehensively evaluates the effectiveness of the linkage strategy from multiple orthogonal dimensions.

[0069] In this embodiment, the protection measures include a fire extinguishing system. An aerosol fire extinguishing system is installed in the energy storage cabinet, and an aerosol fire extinguishing system is integrated inside each point-out unit; for charge and discharge protection, the battery management system BMS adopts a three-level protection strategy for overcharging and over-discharging protection of the battery. The three-level strategies for too high or too low voltage of a single battery are: the first-level alarm, the second-level restricts the magnitude of the charge and discharge current, and the third-level prohibits charge and discharge; for temperature protection, the battery management system BMS collects the temperature of the voltage of a single battery cell and also adopts a three-level protection strategy; for the ambient temperature, the energy storage device uses an air-cooling technology and selects an intelligent air conditioner to adjust the temperature inside the energy storage cabinet; for smoke alarm: a smoke detector is provided at the top of the energy storage product. This device is one of the highest-level alarm devices. Once triggered, it will immediately stop all states of the energy storage device and notify the mobile phone of the on-duty management personnel through a short message; for water ingress alarm, a water ingress detector is provided at the bottom of the energy storage product. This device is one of the highest-level alarm devices. Once triggered, it will immediately stop all states of the energy storage device and notify the mobile phone of the on-duty management personnel through a short message.

[0070] The working principle and beneficial effects of the above technical solution are: By fusing multi-source data to construct a panoramic state of the system, the cloud dynamically evaluates short-term / long-term risks and generates optimization strategies, which are issued for execution after being verified by digital twin simulation; the linkage effect is fed back to the strategy iteration through multi-dimensional evaluation, and combined with dynamic threshold alarm to form a perception-decision-execution-optimization closed loop, realizing the active safety protection of the energy storage system, improving the response speed, reducing the failure rate, and comprehensively improving the safety and intelligent level.

[0071] Embodiment 2: The embodiment of the present invention provides a distributed energy storage safety supervision method based on big data. Based on the technical framework of the distributed energy storage system, the data acquisition layer is determined, and the first-level data of the PCS layer and the second-level data of the BMS layer are collected, and data processing is performed on the first-level data and the second-level data, including:

[0072] Design the first data item regarding the PCS layer based on the distributed energy storage system, determine the first acquisition unit and the first communication method according to the first data item, and then collect the first-level data;

[0073] Design the second data item regarding the BMS layer based on the distributed energy storage system, determine the second acquisition unit and the second communication method according to the second data item, and then collect the second-level data;

[0074] Perform cross-level data fusion on the first-level data and the second-level data.

[0075] In this embodiment, the first data item refers to the key operating parameters of the PCS layer, such as data of the AC side voltage, output power, efficiency, and harmonic distortion rate, etc.

[0076] In this embodiment, the first acquisition unit is a hardware device for acquiring these PCS data. For example, a Hall sensor is used in conjunction with an FPGA high-speed acquisition card to achieve 10 kHz sampling.

[0077] In this embodiment, the first communication method is the PCS layer data transmission protocol. For example, the Modbus TCP protocol is used to overlay a 5G wireless redundant channel to ensure that the transmission delay is less than 5 ms.

[0078] In this embodiment, the second data item refers to the key parameters monitored by the BMS, including data such as single-cell voltage, temperature, SOC, and SOH.

[0079] In this embodiment, the second acquisition unit is a dedicated device for acquiring BMS data. For example, a TI BQ76952 analog front-end chip is used in conjunction with a CAN bus communication module.

[0080] In this embodiment, the second communication method is the BMS layer data transmission protocol. For example, a communication architecture with the CAN bus as the main and RS-485 as the backup is adopted, and the transmission rate is 250 kbps.

[0081] The working principle and beneficial effects of the above technical solution are as follows: By structurally defining the key data items of the PCS layer and the BMS layer, matching dedicated acquisition units and communication protocols, efficient acquisition of multi-source heterogeneous data is achieved; Cross-system data fusion is completed based on spatio-temporal alignment and feature coupling, a unified state perception system is constructed, covering all-link parameters of charging and discharging, monitoring blind spots are eliminated, the data integrity is strong, and the modular design supports the rapid access of new data items, which helps to improve scalability.

[0082] Embodiment 3: The embodiment of the present invention provides a distributed energy storage safety supervision method based on big data, which performs cross-layer data fusion on the first-level data and the second-level data, including:

[0083] Perform the first scenario label association on the first-level data to obtain the first association result, and perform the second scenario label association on the second-level data to obtain the second association result;

[0084] Determine the conflict type of the distributed energy storage system according to the first association result and the second association result, and determine the dynamic conflict monitoring algorithm based on the conflict type;

[0085] Perform data coupling on the first-level data and the second-level data based on the first association result and the second association result to obtain the coupling result;

[0086] Extract the first feature vector of the first-level data, extract the second feature vector of the second-level data, match the corresponding fusion method from the feature-fusion method table, and perform data fusion on the first feature vector and the second feature vector using the fusion method based on the coupling result and the dynamic conflict monitoring algorithm.

[0087] In this embodiment, the first scenario label association refers to the process of attaching operating condition labels to PCS layer data (such as power, voltage, etc.). For example, attach an overload risk label to data where the output power continuously exceeds 90% of the rated value.

[0088] In this embodiment, the first association result is the structured data formed after label association. For example, the PCS data set marked as overload risk contains dimensions such as timestamp, power value, and duration.

[0089] In this embodiment, the second scenario label association is the process of classifying scenarios for BMS layer data (such as temperature, SOC, etc.). For example, mark data where the battery temperature exceeds 45°C and SOC > 95% as a precursor to thermal runaway.

[0090] In this embodiment, the second association result is the labeled output of BMS data. For example, the thermal runaway precursor data set in the above example contains features such as temperature curve and voltage fluctuation.

[0091] In this embodiment, data coupling is the process of establishing a physical association between two types of data. For example, align the overload time points of PCS with the temperature data of BMS at the same moment for time alignment and logical matching.

[0092] In this embodiment, the coupling result is the intermediate product after data association. For example, discover the timing association rule that the BMS temperature rises 2°C 5 minutes after PCS overload.

[0093] In this embodiment, the first feature vector is a set of features extracted from PCS data. For example, a vector composed of [power fluctuation amplitude, efficiency degradation rate, harmonic distortion rate].

[0094] In this embodiment, the second feature vector is a set of features extracted from BMS data. For example, a vector composed of [highest temperature, SOC decline rate, voltage dispersion].

[0095] In this embodiment, the feature-fusion method table is a predefined fusion method matching rule library. For example: when the conflict type is power-temperature conflict, use the weighted fusion algorithm; when the conflict type is SOC-efficiency conflict, use the neural network fusion.

[0096] In this embodiment, the fusion method is a specific integration algorithm. For example, use the entropy weight method to determine the weight ratio for power and temperature data.

[0097] In this embodiment, the dynamic conflict monitoring algorithm is an algorithm for real-time identification of system contradictions. For example, when it is detected that the required power of the PCS > the available power of the BMS, a power supply shortage conflict is triggered, and the sliding window algorithm is used to continuously monitor the evolution trend of the conflict.

[0098] In this embodiment, the difference between data fusion and data coupling is that data coupling is to establish an association relationship between data (such as time alignment). For example, align the moment of power mutation of the PCS with the BMS temperature recording time; data fusion is to deeply integrate the associated data. For example, generate a system health index by using the power and temperature data through Kalman filtering.

[0099] In this embodiment, the role of the dynamic conflict monitoring algorithm in data fusion is that the dynamic conflict monitoring algorithm is the intelligent arbitration center of data fusion, and the fusion quality is guaranteed through multi-dimensional technologies.

[0100] The working principle and beneficial effects of the above technical solution are: semantic association of PCS and BMS data is established through scenario label association to identify system-level conflict types; a dynamic monitoring algorithm is selected based on conflict characteristics to achieve data coupling and feature vector fusion, and finally an adaptive fusion method is used to generate a unified data representation, improving the conflict recognition rate and enhancing data consistency.

[0101] Embodiment 4: The embodiment of the present invention provides a distributed energy storage safety supervision method based on big data. According to the data fusion result, a cloud dynamic risk assessment is carried out on the distributed energy storage system, and the matching strategy is called using the risk assessment result, and multi-system linkage is carried out based on the matching strategy, and the linkage effect is obtained, including:

[0102] The cloud platform receives the data fusion result, and uses a risk assessment model to evaluate the short-term risk and long-term risk of the distributed energy storage system to obtain a risk assessment result;

[0103] Use the risk assessment result to match the prefabricated strategy from the rule library. If the match fails, call the reinforcement learning model to generate a temporary strategy, and combine the prefabricated strategy and the temporary strategy to obtain a matching strategy;

[0104] Convert the matching strategy into a standard instruction, split the standard instruction, and perform multi-system linkage according to the split result, and set a fault tolerance mechanism at the same time;

[0105] Build a digital twin model based on the data fusion result, use the digital twin model to simulate and verify the matching strategy and the multi-system linkage process, and determine the linkage effect according to the verification result.

[0106] In this embodiment, the cloud platform is the energy storage system management center deployed on the cloud server. For example, a certain energy storage cloud platform uses Alibaba Cloud ECS instances to process data from multiple energy storage sites in real time.

[0107] In this embodiment, short-term risks are sudden threats that need to be addressed immediately, such as immediate risks like early warnings of battery thermal runaway and sudden changes in grid frequency.

[0108] In this embodiment, long-term risks are systematic risks that develop gradually, such as risks that require periodic monitoring like the attenuation of battery pack capacity and the aging of insulation materials.

[0109] In this embodiment, the rule base is a set of predefined policies, including hundreds of policy rules such as forcibly switching to the charging mode when the SOC < 20%.

[0110] In this embodiment, the prefabricated policy is a standardized response plan stored in the rule base, such as a preset process of starting the liquid cooling system + reducing the load by 50% when the detected temperature > 50°C.

[0111] In this embodiment, the reinforcement learning model is an adaptive policy generation algorithm, such as a model trained based on the DQN framework, which autonomously generates frequency modulation policies when encountering new grid fluctuation patterns.

[0112] In this embodiment, the temporary policy is a dynamic solution for unforeseen overload scenarios, such as a battery scheduling plan for preferentially ensuring the power supply of the ICU generated during a certain sudden power outage.

[0113] In this embodiment, the fault tolerance mechanism is a fault response plan system, including dual guarantees of hardware redundancy (spare PCS module) and software fault tolerance (instruction retry mechanism).

[0114] In this embodiment, the standard instruction is a standardized control command that the device can execute. For example, device PCS-07 sets the output power to 300 kW ± 5%; instruction splitting is decomposing complex policies into device-level operations, such as decomposing the system load reduction into sub-instructions like PCS power reduction + BMS current limiting + air conditioner frequency increase; the splitting result is the operation sequence after instruction decomposition, including metadata such as the execution device ID, parameter values, and timing requirements.

[0115] In this embodiment, simulation verification is to test the feasibility of policies in a virtual environment, such as verifying whether the load reduction policy will cause the grid voltage to exceed the limit.

[0116] In this embodiment, the verification result is a quantitative index output by the simulation, including evaluation data such as success rate, expected energy consumption, and device loss.

[0117] The working principle and beneficial effects of the above technical solution are as follows: The cloud platform analyzes short-term and long-term risks in real time by integrating a data-driven risk assessment model; generates an optimal strategy based on a rule base and reinforcement learning, disassembles it into multi-system collaborative instructions after verification by digital twin simulation, and embeds a fault tolerance mechanism to achieve closed-loop control of assessment - decision - verification - execution, with comprehensive risk coverage and high policy reliability.

[0118] Embodiment 5: The embodiment of the present invention provides a distributed energy storage safety supervision method based on big data, and sets a fault tolerance mechanism, including:

[0119] Formulate unit fault tolerance measures according to the unit structure existing in the distributed energy storage system, formulate multi-level linkage fault tolerance measures according to multi-system linkage, and set the fault tolerance mechanism by integrating the unit fault tolerance measures and multi-level linkage fault tolerance measures.

[0120] In this embodiment, the unit structure is the smallest functional module that can operate independently in the distributed energy storage system, and has complete energy storage and conversion capabilities. For example, a single battery energy storage cabinet (including battery pack, BMS, cooling system), a PCS converter module (including IGBT unit, control board, filter circuit), a photovoltaic inverter and its supporting DC / DC converter.

[0121] In this embodiment, the unit fault tolerance measure is a fault isolation and self-recovery solution designed for a single functional module to ensure that local faults do not affect the basic functions of the unit. For example, at the battery module level: when a single battery is short-circuited, the BMS triggers a fuse to cut off the faulty circuit, and at the same time enables a standby battery pack (such as a certain brand of energy storage cabinet configured with N+1 redundant batteries); when the temperature sensor fails, switch to adjacent sensor data + AI prediction compensation (error < ±1°C).

[0122] In this embodiment, the multi-level linkage fault tolerance measure is a fault response strategy for cross-unit collaboration to solve the problem of system-level functional chain breakage. For example, grid frequency modulation interruption recovery: when the main PCS fails, the dispatching system automatically transfers the load to the standby PCS, and at the same time the BMS adjusts the output priority of the battery pack (such as a certain microgrid system achieving seamless switching within 200ms); cooling system failure emergency: after detecting an air conditioner failure, link the BMS to limit the charge and discharge power to 50%, and at the same time start the standby air duct of the adjacent energy storage cabinet to share cooling (such as a certain project adopting a hybrid redundant design of air cooling + liquid cooling); communication interruption fault tolerance: when the cloud control is disconnected, the edge gateway automatically switches to the locally preset strategy (such as SOC balance mode), and caches data to be transmitted back after communication is restored.

[0123] The working principle and beneficial effects of the above technical solution are as follows: By hierarchically designing the fault tolerance mechanism, local fault isolation is achieved at the unit level, and global function maintenance is ensured at the system level. Combining real-time status monitoring and dynamically switching fault tolerance strategies, a multi-level protection system combining points and surfaces is formed to achieve precise isolation of faults.

[0124] Example 6: The embodiment of the present invention provides a distributed energy storage safety supervision method based on big data, constructs a digital twin model based on the data fusion result, uses the digital twin model to simulate and verify the matching strategy and the multi-system linkage process, and determines the linkage effect according to the verification result, including:

[0125] Overlay the simulation verification curve and the actual execution curve, and calculate the fitting result;

[0126] Decompose the matching strategy into strategy actions that can be mapped to the digital twin model, perform causal inference on the strategy actions and the existing linkages, and calculate the relevant results;

[0127] Quantify the strategy actions and the existing linkages into a directed graph based on the relevant results and the fitting result, analyze the directed graph, and determine the critical path, invalid actions, positive feedback loops, and negative feedback loops;

[0128] Perform the first optimization on the matching strategy based on the critical path and invalid actions, perform the second optimization on the matching strategy according to the positive feedback loops and negative feedback loops, and determine the optimized strategy by integrating the first optimization and the second optimization;

[0129] Use the digital twin model to synchronously simulate and verify the matching strategy and the optimized strategy, and obtain the linkage effect.

[0130] In this embodiment, , where represents the dynamic time warping distance function; represents the simulation verification curve; represents the actual execution curve; represents the simulation parameter value corresponding to the kth monitoring point; represents the actual execution parameter value corresponding to the kth monitoring point; N represents the total number of monitoring points; k represents the kth monitoring point; β represents the adjustment weight coefficient of the curve distribution form difference function and the dynamic time warping distance function; represents the energy normalization factor of the actual execution curve; represents the curve distribution form difference function; F represents the fitting result. If F < 0.7, it means that the simulation parameters are mismatched. For example, a certain energy storage station discovers that the BMS temperature sampling module is damaged through the decrease of the F value.

[0131] In this embodiment, causal inference is to analyze the causal relationship between strategy actions and system responses and exclude the influence of confounding factors. For example, use double machine learning to estimate the causal effect of a 10% PCS load reduction on a 3°C decrease in battery temperature, and perform a Granger causality test (causality is determined when p < 0.05).

[0132] In this embodiment, the policy action is an executable control instruction or parameter adjustment. For example, limit the charging current to 0.5C and start the 3rd air conditioner to cool at maximum wind speed.

[0133] In this embodiment, the relevant result is an index that quantifies the correlation between actions and effects. For example, the Pearson correlation coefficient r = 0.82 (linear correlation strength between ↑ liquid cooling power and ↓ temperature difference), and the information gain IG = 0.6 (contribution degree of the BMS balancing strategy to the SOC consistency).

[0134] In this embodiment, the directed graph represents the system dynamics using nodes (actions / effects) and edges (causal relationships).

[0135] In this embodiment, the ineffective action is an action that has no significant impact on the target. For example, the gain of switching to the backup sensor for temperature control is only 0.05.

[0136] In this embodiment, the positive feedback is a self-reinforcing cycle (such as ↑ cooling → ↓ temperature → allowing higher power ↑ → ↑ cooling demand); negative feedback: a self-inhibiting cycle (such as overload protection → ↓ power → ↓ revenue → reduced investment → ↑ equipment aging).

[0137] In this embodiment, the first optimization is to strengthen the critical path and eliminate ineffective actions. For example, increase the weight of the liquid cooling power (critical path action) and cancel the battery pack rotation strategy (ineffective action). The second optimization is to use the feedback loop to adjust the system stability. For example, for the positive feedback loop: set an upper limit on power growth to prevent runaway; for the negative feedback loop: introduce a compensation mechanism to break the vicious cycle.

[0138] In this embodiment, the optimization strategy is an integrated set of optimized control rules.

[0139] [[ID=2I]]In this embodiment, the existing linkage is the actual collaborative behavior between subsystems. For example, when the BMS limits the current, the PCS synchronously reduces the power and the air conditioner increases the cooling.

[0140] In this embodiment, the linkage effect is the change in the system state after the execution of the linkage strategy. For example, positive effect: the temperature drops by 8℃ within 30 minutes and there is no power grid fluctuation; negative effect: the frequent power adjustment causes a +2% loss in the battery cycle life.

[0141] The working principle and beneficial effects of the above technical solution are as follows: By comparing and analyzing the digital twin simulation with the actual operation data, the causal relationship between the policy action and the linkage effect is quantified; the critical path and feedback mechanism are identified using the directed graph, and the policy parameters and logic are optimized in stages. Finally, the optimal control scheme is output through twin verification, realizing the dynamic iteration and upgrade of the policy and improving the policy accuracy.

[0142] Embodiment 7: The embodiment of the present invention provides a distributed energy storage safety supervision method based on big data, which conducts multi-dimensional evaluation on the linkage effect, optimizes and adjusts the optimization strategy, and simultaneously conducts safety warnings according to the linkage effect and triggers corresponding protection measures to form safety supervision, including:

[0143] Determine the evaluation dimensions of the linkage effect from the effect-evaluation table in combination with the data fusion result, obtain the historical linkage effect to optimize the dynamic threshold, and determine the multi-dimensional evaluation result of the linkage effect based on the evaluation dimension and threshold optimization;

[0144] Determine the cause of the linkage effect according to the multi-dimensional evaluation result, alarm the failed linkage, verify the execution device based on the cause, and conduct safety supervision on the execution device.

[0145] In this embodiment, the evaluation dimension is a set of multi-angle indicators for quantifying the linkage effect. For example, safety: the number of times of temperature exceeding the standard (such as the daily cumulative duration when >50°C); efficiency: command response delay (such as the average time from receiving to execution of PCS); economy: the cost of a single linkage (such as the amount converted from the consumption of fire extinguishing agent).

[0146] In this embodiment, the historical linkage effect is the recorded data of the execution of past similar strategies. For example, the average cooling rate (°C / min) of the load + strong cooling strategy in the past 30 days, and the historical value of the maximum discharge current allowed for the same type of battery pack when SOC = 20%.

[0147] In this embodiment, the dynamic threshold is a determination boundary that adapts to the system state, and the threshold optimization is a method of adjusting the threshold based on historical data.

[0148] In this embodiment, the multi-dimensional evaluation result is the comprehensive output of the scores of each dimension, and the alarm is a hierarchical notification of abnormal linkages. For example, level 1 alarm (text message + sound and light): The temperature linkage of battery cabinet No. 3 fails and the system has been forced to stop; level 2 alarm (system pop-up window): The response delay of PCS-05 exceeds 200 ms.

[0149] In this embodiment, the execution device is a physical device that directly executes control instructions in the distributed energy storage system, including PCS converters, PCS converters, PCS converters, air conditioning systems, circuit breakers / contactors, and standby power switchers.

[0150] The working principle and beneficial effects of the above technical solutions are: By integrating real-time data and historical effects to construct a dynamic evaluation system, quantifying the linkage performance based on multi-dimensional indicators; using the threshold self-optimization algorithm to identify abnormalities, locate the root cause of faults and trigger accurate alarms, and simultaneously verify the status of the execution device to form a closed-loop management of evaluation-diagnosis-disposal.

[0151] Embodiment 8: The embodiment of the present invention provides a distributed energy storage safety supervision system based on big data, as shown in Figure 2 and includes:

[0152] Fusion module: Based on the distributed energy storage system, collect the first-level data of the PCS layer and the second-level data of the BMS layer, and perform data fusion on the first-level data and the second-level data;

[0153] Linkage module: Perform cloud dynamic risk assessment on the distributed energy storage system according to the data fusion result, use the risk assessment result to call the matching strategy, perform multi-system linkage based on the matching strategy, and obtain the linkage effect;

[0154] Supervision module: Perform multi-dimensional evaluation on the linkage effect, optimize and adjust the optimization strategy, and at the same time perform safety warning according to the linkage effect, and trigger corresponding protection measures to form safety supervision.

[0155] The working principle and beneficial effects of the above technical solution are: By fusing multi-source data to construct the panoramic state of the system, dynamically evaluate short-term / long-term risks in the cloud and generate optimization strategies, which are verified by digital twin simulation and then issued for execution; The linkage effect is fed back to the strategy iteration through multi-dimensional evaluation, combined with dynamic threshold warning to form a perception-decision-execution-optimization closed loop, realizing the active safety protection of the energy storage system, improving the response speed, reducing the failure rate, and comprehensively improving the safety and intelligent level.

[0156] Embodiment 9: The embodiment of the present invention provides a distributed energy storage safety supervision cloud platform based on big data, as shown in Figure 3 and includes a distributed energy storage system, a digital twin model:

[0157] The distributed energy storage system includes a PCS layer and a BMS layer;

[0158] The digital twin model includes the simulation of the linkage process and the simulation of the risks existing during the operation of the distributed energy storage system.

[0159] The working principle and beneficial effects of the above technical solution are: The digital twin model synchronizes the data of the PCS layer and the BMS layer in real time, verifies the feasibility of the linkage strategy through multi-physical field coupling simulation, and predicts the system risks based on historical data and machine learning, forming a closed-loop management of data-driven-simulation verification-dynamic optimization.

[0160] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A distributed energy storage safety supervision method based on big data, characterized in that Including: Step 1: Collect the first-level data of the PCS layer and the second-level data of the BMS layer based on the distributed energy storage system, and perform data fusion on the first-level data and the second-level data; Step 2: Conduct cloud dynamic risk assessment on the distributed energy storage system according to the data fusion result, call the matching strategy using the risk assessment result, perform multi-system linkage based on the matching strategy, and obtain the linkage effect; Step 3: Conduct multi-dimensional evaluation on the linkage effect, optimize and adjust the optimization strategy, and at the same time issue security warnings according to the linkage effect and trigger corresponding protection measures to form security supervision; Among them, determine the data acquisition layer based on the technical framework of the distributed energy storage system, collect the first-level data of the PCS layer and the second-level data of the BMS layer, and perform data processing on the first-level data and the second-level data, including: Design the first data item regarding the PCS layer based on the distributed energy storage system, determine the first acquisition unit and the first communication method according to the first data item, and then collect the first-level data; Design the second data item regarding the BMS layer based on the distributed energy storage system, determine the second acquisition unit and the second communication method according to the second data item, and then collect the second-level data; Perform cross-level data fusion on the first-level data and the second-level data; Among them, performing cross-level data fusion on the first-level data and the second-level data includes: Perform the first scenario label association on the first-level data to obtain the first association result, perform the second scenario label association on the second-level data to obtain the second association result; Determine the conflict type of the distributed energy storage system according to the first association result and the second association result, and determine the dynamic conflict monitoring algorithm based on the conflict type; Perform data coupling on the first-level data and the second-level data based on the first association result and the second association result to obtain the coupling result; Extract the first feature vector of the first-level data, extract the second feature vector of the second-level data, match the corresponding fusion method from the feature-fusion method table, and perform data fusion on the first feature vector and the second feature vector using the fusion method based on the coupling result and the dynamic conflict monitoring algorithm.

2. The distributed energy storage safety supervision method based on big data according to claim 1, wherein Conduct cloud dynamic risk assessment on the distributed energy storage system according to the data fusion result, call the matching strategy using the risk assessment result, perform multi-system linkage based on the matching strategy, and obtain the linkage effect, including: The cloud platform receives the data fusion result, uses the risk assessment model to evaluate the short-term risk and long-term risk of the distributed energy storage system, and obtains the risk assessment result; Use the risk assessment result to match the prefabricated strategy from the rule library. If the match fails, call the reinforcement learning model to generate a temporary strategy, and combine the prefabricated strategy and the temporary strategy to obtain the matching strategy; Convert the matching strategy into a standard instruction, split the standard instruction, and perform multi-system linkage according to the split result, and at the same time set a fault tolerance mechanism; Build a digital twin model based on the data fusion result, use the digital twin model to simulate and verify the matching strategy and the multi-system linkage process, and determine the linkage effect according to the verification result.

3. The distributed energy storage safety supervision method based on big data according to claim 2, wherein Set up a fault tolerance mechanism, including: Formulate unit fault tolerance measures according to the unit structure existing in the distributed energy storage system, formulate multi-level linkage fault tolerance measures according to the multi-system linkage, and set up a fault tolerance mechanism by integrating the unit fault tolerance measures and the multi-level linkage fault tolerance measures.

4. The distributed energy storage safety supervision method based on big data according to claim 2, wherein, Build a digital twin model based on the data fusion result, use the digital twin model to simulate and verify the matching strategy and the multi-system linkage process, and determine the linkage effect according to the verification result, including: Overlay the simulation verification curve and the actual execution curve, and calculate the fitting result; Decompose the matching strategy into strategy actions that can be mapped to the digital twin model, conduct causal inference on the strategy actions and the existing linkages, and calculate the relevant results; Quantify the strategy actions and the existing linkages into a directed graph based on the relevant results and the fitting result, analyze the directed graph, and determine the critical path, invalid actions, positive feedback loops, and negative feedback loops; Conduct the first optimization of the matching strategy based on the critical path and invalid actions, conduct the second optimization of the matching strategy according to the positive feedback loops and negative feedback loops, and determine the optimized strategy by integrating the first optimization and the second optimization; Use the digital twin model to synchronously simulate and verify the matching strategy and the optimized strategy, and obtain the linkage effect.

5. The distributed energy storage safety supervision method based on big data according to claim 4, characterized in that Conduct a multi-dimensional evaluation of the linkage effect, optimize and adjust the optimized strategy, and at the same time issue a safety warning according to the linkage effect and trigger corresponding protection measures to form safety supervision, including: Determine the evaluation dimensions of the linkage effect from the effect-evaluation table in combination with the data fusion result, obtain the historical linkage effect to optimize the dynamic threshold, and determine the multi-dimensional evaluation result of the linkage effect based on the evaluation dimensions and the threshold optimization, where the dynamic threshold is a decision boundary that adapts to the system state; Determine the cause of the linkage effect according to the multi-dimensional evaluation result, issue a warning for the failed linkage, verify the execution device based on the cause, and conduct safety supervision on the execution device.

6. A distributed energy storage safety supervision system based on big data, characterized in that Including: Fusion module: Collect the first-level data of the PCS layer and the second-level data of the BMS layer based on the distributed energy storage system, and conduct data fusion on the first-level data and the second-level data; Linkage module: Conduct a cloud dynamic risk assessment on the distributed energy storage system according to the data fusion result, call the matching strategy using the risk assessment result, conduct multi-system linkage based on the matching strategy, and obtain the linkage effect; Supervision module: Conduct a multi-dimensional evaluation of the linkage effect, optimize and adjust the optimized strategy, and at the same time issue a safety warning according to the linkage effect and trigger corresponding protection measures to form safety supervision; Among them, the fusion module includes: First-level data determination unit: Design the first data item regarding the PCS layer based on the distributed energy storage system, determine the first collection unit and the first communication method according to the first data item, and then collect the first-level data; Second-level data determination unit: Design the second data item regarding the BMS layer based on the distributed energy storage system, determine the second acquisition unit and the second communication method according to the first data item, and then acquire the second-level data; Data fusion unit: Perform cross-level data fusion on the first-level data and the second-level data; Among them, the data fusion unit includes: Association result determination subunit: Perform the first scenario label association on the first-level data to obtain the first association result, and perform the second scenario label association on the second-level data to obtain the second association result; Conflict detection subunit: Determine the conflict type of the distributed energy storage system according to the first association result and the second association result, and determine the dynamic conflict monitoring algorithm based on the conflict type; Coupling subunit: Perform data coupling on the first-level data and the second-level data based on the first association result and the second association result to obtain a coupling result; Vector fusion subunit: Extract the first feature vector of the first-level data, extract the second feature vector of the second-level data, match the corresponding fusion method from the feature-fusion method table, and perform data fusion on the first feature vector and the second feature vector using the fusion method based on the coupling result and the dynamic conflict monitoring algorithm.

Images