Optical storage and charging integrated system based on distributed soft bus

Abstract

The invention relates to the technical field of intelligent power distribution, in particular to an optical storage and charging integrated system based on a distributed soft bus, which comprises hardware equipment: a photovoltaic module, a wind driven generator, an energy storage module, a charging pile, acquisition equipment, a photovoltaic controller, a DCDC module and a mains supply / thermoelectric generator; communication and control: a distributed soft bus, a time sensitive network (TSN) and an intelligent control unit (EMS); according to the scheme, photovoltaic, wind power, energy storage, charging piles and other hardware equipment are connected through a distributed soft bus, real-time transmission of control instructions is achieved through a TSN, data are analyzed through intelligent algorithms such as a charging and discharging model and a multi-task learning model, an optimization strategy is generated, multi-energy cooperative control (including photovoltaic direct supply and energy recovery in the case of faults) is executed through an EMS, and the energy efficiency is improved. A user can set priorities through a terminal to participate in scheduling, and finally model parameters are corrected based on feedback data to form a light storage and charging integrated system of data acquisition, prediction optimization, intelligent control, user interaction and closed-loop evolution.

Description

Technical Field

[0001] The present invention relates to the field of intelligent power distribution technology, and in particular to a distributed soft bus-based integrated photovoltaic storage and charging system. Background Art

[0002] Traditional integrated photovoltaic, energy storage and charging systems have significant technical shortcomings in the coordinated management of multi-source energy. First, the interconnection of equipment relies on a hard-wired architecture. PV panels, energy storage modules, charging piles and other equipment are difficult to plug and play due to the heterogeneity of communication protocols. When adding thermoelectric generators and new energy storage units, the hardware and control logic need to be reconstructed, and the system scalability and compatibility are insufficient. Secondly, multi-source data collection is scattered and lacks a unified data center. Data such as power generation power, energy storage status, and load demand are not integrated with a dynamic weight distribution mechanism. Only simple filtering pre-processing is used, which makes it impossible to analyze the coupling relationship between environmental factors (such as light and temperature) and energy production / consumption, resulting in limited accuracy of the prediction model.

[0003] At the energy dispatch level, traditional methods employ fixed-threshold switching strategies (e.g., passively connecting to the mains when the energy storage capacity falls below a threshold). These methods fail to integrate multi-task learning prediction and artificial bee colony algorithm optimization, making it impossible to achieve multi-dimensional coordinated scheduling of power generation, energy storage status, and user needs. This often leads to the abandonment of renewable energy or overloaded charging stations. Furthermore, in terms of fault handling, there is a lack of real-time fault detection and redundant switching mechanisms based on a soft bus. Energy storage module failures require manual intervention, which can easily cause charging interruptions.

[0004] In addition, users cannot participate in energy strategy adjustments through the terminal, the system lacks demand-side response capabilities, and it is difficult to optimize scheduling based on personalized needs such as electricity price periods and charging priorities. There is significant room for improvement in overall energy efficiency and user experience.

[0005] Therefore, there is an urgent need for an integrated photovoltaic storage and charging system based on a distributed soft bus to meet the needs of distributed energy systems for efficient collaboration, real-time response and dynamic optimization. Summary of the Invention

[0006] In response to the shortcomings of the existing technology, the present invention provides an integrated photovoltaic storage and charging system based on a distributed soft bus. The distributed soft bus realizes plug-and-play interconnection of devices, dynamic processing of multi-source data and intelligent prediction optimization. Combined with time-sensitive network scheduling, fault self-healing, user-participated energy scheduling and closed-loop feedback mechanism, it improves the compatibility, real-time performance, energy efficiency and reliability of the photovoltaic storage and charging system.

[0007] To achieve the above objectives, one of the present inventions is to provide an integrated optical storage and charging system based on a distributed soft bus, comprising the following modules:

[0008] Distributed soft bus module: Establishes plug-and-play communication connections between photovoltaic panels, wind turbines, energy storage modules, charging piles, intelligent control units (EMS), and mains / thermoelectric generators. It serves as a data hub to collect device status data in real time and issue control commands. The soft bus supports OPC UA / MQTT protocols and time-sensitive network (TSN) scheduling, ensuring microsecond-level transmission of control commands.

[0009] Multi-source data processing module: This module collects power generation data, energy storage status data, load data, and environmental data through a soft bus. After noise removal through a smoothing filter algorithm, it generates standardized characteristic data based on the charge-discharge model and the load model. It also generates a dynamic weight allocation matrix and stores it in the photovoltaic storage and charging data center.

[0010] Intelligent prediction and strategy optimization module: This module inputs pre-processed data into a multi-task learning model and combines it with a decision tree correction algorithm to predict power generation and energy storage status. It then uses an artificial bee colony algorithm to optimize the multi-objective decision model, generating a control strategy that includes energy switching thresholds and energy storage charge and discharge power, and reserves a user priority parameter interface.

[0011] Multi-energy collaborative control module: Based on the control strategy, the module prioritizes photovoltaic / wind power during the day and switches between energy storage and mains power at night through TSN. When a storage fault is detected, the EMS disconnects the faulty battery through the soft bus and activates the photovoltaic direct supply link, while also recovering redundant power from the charging pile to the energy storage module.

[0012] User interaction and closed-loop optimization module: The soft bus pushes data to the user terminal. After the user sets the charging priority, the instruction is transmitted to the EMS to adjust the energy storage strategy; the EMS updates the data center based on the feedback data from the charging pile, and corrects the decision tree model parameters through historical data to achieve load balancing and energy optimization.

[0013] The second aspect of the present invention is to provide a method for integrating light storage and charging based on a distributed soft bus, which is used to implement an integrated light storage and charging system based on a distributed soft bus, including the following steps:

[0014] S100: Establishing photovoltaic modules through a distributed soft bus, and then connecting various data collection devices through the distributed soft bus, using the distributed soft bus as a data center to collect status data of each device in real time and issue control instructions;

[0015] S200: The status data of each device collected through the soft bus is analyzed and generated into standardized charge-discharge characteristic data and load characteristic data based on the charge-discharge model and the load model. A dynamic weight distribution matrix is ​​also generated and stored in the solar storage and charging data center;

[0016] S300: Input the pre-processed power generation performance index and energy storage status index into the multi-task learning model, then dynamically adjust the model parameters in combination with the decision tree correction algorithm to predict photovoltaic / wind power and energy storage charging and discharging status. Subsequently, the dynamic weight allocation matrix is ​​input into the multi-objective decision model optimized by the artificial bee colony algorithm to generate a target control strategy including multi-energy switching thresholds and energy storage charging and discharging power, and reserve a user priority parameter interface;

[0017] S400: Based on the target control strategy, the control instruction set is distributed through the time-sensitive network. When an energy storage module fault is detected, the EMS disconnects the faulty battery through the soft bus control, activates the photovoltaic controller and DCDC module to directly supply power, and simultaneously feeds the redundant power of the charging pile back to the energy storage module through the soft bus for recycling;

[0018] S500: Data is pushed to the user terminal via the soft bus. The user sets the charging priority through the terminal. The soft bus transmits the demand instructions to the EMS to adjust the energy storage discharge strategy.

[0019] S600: The EMS receives feedback data from charging piles, updates the solar-storage-charging data center through the soft bus, and modifies the decision tree model parameters based on historical data distribution and environmental factors to achieve load balancing and energy optimization.

[0020] The third aspect of the present invention is to provide a computer device and a readable storage medium, characterized in that the computer device includes: a processor and a memory, the memory storing at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to realize an integrated optical storage and charging system based on a distributed soft bus.

[0021] Compared with the existing technology, the present invention provides an integrated photovoltaic storage and charging system based on a distributed soft bus, which has the following beneficial effects:

[0022] 1. We establish plug-and-play communication connections for photovoltaic panels, energy storage modules, charging piles and other devices through a soft bus. It serves as a data hub to collect status data and issue commands in real time, using standardized interfaces to abstract heterogeneous devices into a unified data model.

[0023] 2. This solution filters and removes noise from the original data, generates standardized characteristic data based on the charge and discharge model and load model, introduces a dynamic weight matrix to adaptively adjust the data fusion strategy, and stores it in the data center;

[0024] 3. This solution uses a multi-task learning model combined with a decision tree correction algorithm to predict power generation and energy storage status. It optimizes the multi-objective decision model using an artificial bee colony algorithm to generate a control strategy that includes energy switching thresholds and charge and discharge power, and reserves a user priority interface.

[0025] 4. This solution uses TSN to achieve microsecond-level transmission of control instructions, giving priority to renewable energy during the day and switching between energy storage and mains power at night. When a storage fault is detected, the faulty battery is automatically disconnected, the photovoltaic direct supply link is activated, and redundant power from the charging pile is recovered to the energy storage module.

[0026] 5. This solution pushes data to the user terminal through the soft bus. The user sets the charging priority and transmits instructions to the EMS to adjust the energy storage strategy to achieve "source-storage-load" coordination. At the same time, the EMS updates the data center based on the feedback from the charging pile execution, corrects the decision tree model parameters based on historical data, and dynamically optimizes load balancing and energy distribution.

[0027] This solution uses a distributed soft bus to achieve plug-and-play interconnection of devices, dynamic processing of multi-source data, and intelligent prediction and optimization. Combined with time-sensitive network scheduling, fault self-healing, user-participated energy scheduling, and a closed-loop feedback mechanism, it improves the compatibility, real-time performance, energy efficiency, and reliability of the photovoltaic storage and charging system, and realizes the coordinated optimization and continuous iteration of "source-storage-load". BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0029] Figure 1 This is the architecture diagram of the optical storage and charging system of the distributed soft bus of the present invention;

[0030] Figure 2 This is a timing diagram of the core process of the optical storage and charging method of the present invention;

[0031] Figure 3 The data processing and weight modeling flow chart of the present invention;

[0032] Figure 4 This is a flowchart of user interaction and closed-loop optimization of the present invention;

[0033] Figure 5 This is a redundancy switching flow chart of the present invention. DETAILED DESCRIPTION

[0034] In order to make the objectives, technical solutions and advantages of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0035] Traditional systems are unable to meet the needs of distributed energy systems for efficient collaboration, real-time response and dynamic optimization due to their rigid interconnection architecture, extensive data processing, lack of intelligent scheduling and insufficient user interaction.

[0036] This solution proposes an integrated photovoltaic storage and charging system based on a distributed soft bus.

[0037] Solve the problem that existing systems are difficult to meet the requirements of distributed energy systems for efficient coordination, real-time response and dynamic optimization;

[0038] Collect status data and issue instructions in real time, using standardized interfaces to abstract heterogeneous devices into a unified data model;

[0039] Generate standardized characteristic data based on charge and discharge models and load models, introduce dynamic weight matrix to adaptively adjust data fusion strategy, and store it in the data center;

[0040] Use a multi-task learning model combined with a decision tree correction algorithm to predict power generation and energy storage status;

[0041] The artificial bee colony algorithm is used to optimize the multi-objective decision-making model, generate a control strategy including energy switching thresholds and charge and discharge power, and reserve a user priority interface;

[0042] When a storage fault is detected, the faulty battery is automatically disconnected, the photovoltaic direct supply link is started, and the redundant power of the charging pile is recovered to the energy storage module.

[0043] This solution uses a distributed soft bus to achieve plug-and-play interconnection of devices, dynamic processing of multi-source data, and intelligent prediction and optimization. It combines time-sensitive network scheduling, fault self-healing, user-participated energy scheduling, and a closed-loop feedback mechanism to improve the compatibility, real-time performance, energy efficiency, and reliability of the solar-storage-charging system.

[0044] Example 1, as Figure 1-Figure 5 As shown, an integrated optical storage and charging method based on a distributed soft bus provided in an embodiment of the present application is exemplified.

[0045] S100: Establishes plug-and-play connections for photovoltaic, wind power, energy storage, charging piles, and other devices through a distributed soft bus. It uses the OPC UA / MQTT protocol to enable heterogeneous device communication and relies on time-sensitive networking (TSN) to ensure microsecond-level transmission of control commands. The soft bus serves as a data hub, frequently collecting device status data (such as photovoltaic power and energy storage SOC) and issuing commands such as charge and discharge start and stop, building the underlying "neural network" of system operation, reducing device access time to less than 5 minutes.

[0046] S200: After collecting raw data through the soft bus, it uses Kalman filtering or the 3σ principle to remove noise. Based on the charge-discharge model (such as the Thevenin equivalent circuit) and the load model (such as LSTM), it generates standardized characteristic data. It also introduces an attention mechanism to generate a dynamic weight matrix (for example, the photovoltaic power weight is increased to 45% when the sunlight is strong). The data is stored in the photovoltaic storage and charging data center, providing high-quality input with priority for the prediction model.

[0047] S300: The pre-processed power generation and energy storage indicators are fed into a bidirectional LSTM multi-task learning model. The model then dynamically adjusts parameters using a decision tree correction algorithm to predict power generation and energy storage status for the next 15 minutes (with an error rate of ≤7%). The artificial bee colony algorithm is used to optimize the multi-objective decision model, generating strategies that include energy switching thresholds (e.g., photovoltaic power supply priority when power is >65%) and charge and discharge power, with a user priority parameter interface reserved (e.g., valley power priority).

[0048] S400: Utilizing TSN (latency ≤ 100μs), it prioritizes photovoltaic / wind power during the day and switches between energy storage and utility power at night. When a voltage deviation of a single energy storage unit is detected to be greater than 15% or a temperature difference greater than 5°C, it disconnects the faulty battery within 500ms, activates the photovoltaic direct supply link (efficiency ≥ 96%), and simultaneously recovers redundant power from the charging pile (recovery rate ≥ 94%), ensuring charging interruption time is less than 100ms.

[0049] S500: The soft bus pushes data such as energy storage capacity and real-time electricity prices to the terminal. Users can select modes such as "Economy" and "Green" or customize priorities (such as charging during off-peak hours). After the instructions are transmitted to the EMS via the soft bus, the energy storage strategy is dynamically adjusted. For example, in "Green Mode," renewable energy is prioritized and charging is delayed (up to 120 minutes) when insufficient energy is available, reducing user costs by 15%-20%.

[0050] S600: EMS receives feedback data from charging piles (such as charging efficiency and energy storage loss), updates the data center every hour and corrects the decision tree model parameters. It adjusts the weights of environmental factors according to the season every year (such as increasing the impact factor of sunshine duration in winter). Through closed-loop optimization, the system's long-term energy efficiency is improved by ≥8%, and the load balancing rate is increased to more than 90%, achieving self-evolution capabilities.

[0051] The step S100 is used to establish device interconnection and data hub via a distributed soft bus, including:

[0052] S110:

[0053] S1101: Utilizing an industrial-grade distributed soft bus architecture, the core protocol layer supports dual protocol stacks: OPC UA (for real-time data exchange) and MQTT-SN (for low-power device access). The physical layer deploys an edge computing gateway (such as Huawei EdgeCube), with a built-in protocol conversion engine to normalize data formats across different devices (for example, converting the Modbus RTU protocol of a photovoltaic inverter to JSON format).

[0054] When a new device is connected, a device description file (DDF) must be uploaded, which contains the following metadata:

[0055] {

[0056] "deviceType":"photovoltaic_panel",

[0057] "communicationProtocol":"ModbusTCP",

[0058] "dataPoints":[

[0059] {"name":"output_power","unit":"kW","samplingRate":"100ms"},

[0060] {"name":"temperature","unit":"℃","samplingRate":"1s"}

[0061] ],

[0062] "controlCommands":["start_charging","adjust_power"]

[0063] }

[0064] The soft bus automatically generates device driver instances through DDF, enabling access within minutes;

[0065] S120:

[0066] S1201: Building a three-layer device network of "energy production-storage-consumption":

[0067] Production layer:

[0068] PV panels (single string power 500W) are connected to the soft bus through a DC combiner box, and each string is equipped with an intelligent collector (supporting IV curve scanning);

[0069] The wind turbine (10kW model) is connected via the converter's built-in ModbusTCP interface, uploading data such as wind speed, active power, and power factor in real time.

[0070] Energy storage layer:

[0071] The lithium battery pack (200kWh, 48 cells in series) is connected via the CAN bus of the BMS (battery management system) to collect 32 parameters such as cell voltage, SOC, and SOH;

[0072] Reserved flow battery interface, using standardized RS-485 communication protocol, supporting plug-and-play;

[0073] Consumer level:

[0074] The DC charging pile (120kW) complies with the GB / T22239-2019 standard, is connected via an Ethernet interface, and supports the OCPP1.6 protocol to report charging status in real time;

[0075] The intelligent control unit (EMS) is equipped with an ARM Cortex-A72 processor and runs a real-time operating system (QNX), responsible for global strategy generation and instruction scheduling;

[0076] S130:

[0077] S1301: Key parameters such as energy storage voltage and charging pile current are collected at a 100ms cycle, and the latest 1000 data items are stored in a ring buffer for real-time control.

[0078] Parameters such as ambient temperature, humidity, and light intensity are collected at a 1s cycle and stored in a time series database (InfluxDB) for trend analysis;

[0079] The control instructions are encrypted using the AES-128 algorithm, and the data packet format is as follows:

[0080] [Command header (2B)] [Device ID (4B)] [Command code (1B)] [Parameter length (1B)] [Encryption parameter (NB)] [CRC check (2B)]

[0081] For example, the instruction code for adjusting the energy storage discharge power is 0x03, the parameters include the target power (kW, 4-byte floating point number), and the transmission delay is ≤50ms;

[0082] This step lays the foundation for subsequent data processing and collaborative control. It eliminates heterogeneous device protocols through a unified interconnection architecture, forming the "neural network" for system operation.

[0083] The S200 is used for multi-source data preprocessing and dynamic weight matrix generation, specifically:

[0084] S210:

[0085] S2101: Photovoltaic power data: The Kalman filter algorithm is used to remove high-frequency noise caused by cloud cover. The state equation is:

[0086] P k =P k-1 +w k ,z k =P k +v k ;

[0087] Among them, w k is the process noise (covariance matrix Q = 0.01), v k is the observation noise (covariance matrix R = 0.05);

[0088] S2102: Single cell voltage: Use the 3σ principle to eliminate outliers. The calculation formula is:

[0089] Threshold = μ ± 3σ, μ = mean, σ = standard deviation;

[0090] Voltage data exceeding the threshold is considered a fault point and triggers an early warning;

[0091] S2103: Charging power demand: Use the sliding average method (window size 15 minutes) to smooth out glitches and eliminate instantaneous fluctuations caused by users plugging and unplugging charging piles;

[0092] S220:

[0093] S2201: Calculate the battery internal resistance R based on the Thevenin equivalent circuit model s With polarization voltage U p :

[0094]

[0095] The charging process voltage curve is fitted by the least square method to obtain R s And C parameters, and then evaluate SOH (health):

[0096] (R s0 is the initial internal resistance, R smax is the scrap threshold internal resistance);

[0097] An LSTM neural network is used to predict charging load. The model structure is as follows: input layer (historical 24-hour power data) → two LSTM layers (128 neurons each) → fully connected layer → output layer (power forecast for the next hour). The loss function is the root mean square error (RMSE).

[0098] S230:

[0099] S2301: Calculate the correlation between each feature and energy decision-making through the mutual information algorithm, for example:

[0100]

[0101] The higher the mutual information value between light intensity and photovoltaic power supply ratio, the greater the weight;

[0102] S2302: Generate a weight matrix W based on the current environment parameters every 15 minutes. An example is as follows (dimension: number of features × 1):

[0103]

[0104] S2303: Ensure that the sum of weights is 1 through the Softmax function:

[0105] On this basis, a dynamic weight allocation matrix is ​​introduced to automatically adjust the fusion strategy based on the impact of data on system decisions (such as the weight of light intensity on photovoltaic power supply). The processed data and the matrix are stored in the photovoltaic storage and charging data center to provide high-quality input for the prediction model.

[0106] The S300 is used for predictive modeling, strategy optimization, and user interface reservation, including:

[0107] S310:

[0108] S3101: The dimension is (None, 10), including photovoltaic efficiency, wind power fluctuation rate, energy storage SOC, SOH, light intensity, temperature, current electricity price, charging pile load rate, user priority, and timestamp;

[0109] Two fully connected layers (256 neurons each, ReLU activation function) to extract common features;

[0110] Output photovoltaic power, wind power (values ​​for the next 15 minutes), loss function MSE;

[0111] Output charge and discharge power upper limit, remaining number of cycles, and loss function MAE;

[0112] Adopting dynamic task weighting method, the back propagation gradient ratio is automatically adjusted according to the loss value of each task;

[0113] S320:

[0114] S3201: Use information gain to select split attributes. The formula is:

[0115]

[0116] Prioritize attributes with the largest GainRatio (such as light intensity and energy storage SOC);

[0117] S3202: Use the pessimistic pruning (PEP) algorithm to estimate the generalization error of the subtree based on the training set error, and delete branches whose contribution is lower than a threshold (such as 0.1);

[0118] S3203: Update the tree structure every hour based on the latest data increments, and trigger a re-split when the number of samples in a leaf node is less than 50;

[0119] S330:

[0120] S3301: 50 employed bees / 50 observer bees;

[0121] X=[T pv_grid ,T es_grid ,P es_charge_max ,P es_discharge_max ];

[0122] (PV switching threshold, energy storage switching threshold, charging power upper limit, discharging power upper limit);

[0123]

[0124] α=0.5,β=0.3,γ=0.2, corresponding to renewable energy utilization rate, utility power cost savings, and energy storage remaining life, respectively);

[0125] The fitness value changes less than 0.1% for 50 consecutive generations or the maximum number of iterations reaches 200 generations;

[0126] S340:

[0127] S3401: Define the user requirement parameter structure as follows:

[0128] typedefstruct{

[0129] uint8_tpriority_level; / / Priority level (1-5, 1 is the highest)

[0130] floatgrid_price_threshold; / / Off-peak electricity price threshold (yuan / kWh)

[0131] uint8_tenergy_preference; / / Energy preference (0=mains power priority, 1=storage priority, 2=photovoltaic priority)

[0132] uint16_tcharge_deadline; / / Charging deadline (minutes, relative to current time)}UserDemand;

[0133] After receiving user requirements, the EMS embeds parameters into the control strategy through the soft bus. For example, when energy_preference = 2, the photovoltaic power supply threshold is forced to drop from 70% to 50%, giving priority to the use of photovoltaic energy.

[0134] This step converts data value into an executable scheduling plan through a closed loop of "data-model-strategy".

[0135] Step S400 is used for time-sensitive network scheduling, fault handling, and energy recovery, and includes:

[0136] S410:

[0137] S4101: Implements gated list (GL) scheduling based on the IEEE802.1Qbv standard, assigning priorities to different types of data:

[0138] Data Type Priority Transmission cycle Gating status (open time ratio) Control instructions 7 1ms 90％ Real-time status data 5 10ms 60％ Historical data synchronization 3 100ms 30％

[0139] When the photovoltaic power P pv ≥P load +P es_charge When executing: P pv_to_load =P load ,P pv_to_es =P pv -P load ;

[0140] (All remaining photovoltaic power is stored in the energy storage module);

[0141] If the energy storage SOC is less than 30% and the current electricity price is off-peak (less than 0.3 yuan / kWh), then:

[0142]

[0143] Calculate the required charging power to ensure that the target SOC is charged to 70% during the off-peak period);

[0144] S420:

[0145] S4201: Single cell voltage deviation from mean > 15% or temperature difference > 5°C;

[0146] Internal resistance growth rate during charge and discharge > 10% / month

[0147] S4202: Redundancy switching process: Figure 5 As shown;

[0148] S4203: From fault detection to switching completion < 500ms, ensuring that the charging pile output interruption time is < 100ms;

[0149] S430:

[0150] S4301: When the charging pile output power is less than 20% of the rated power for 5 minutes, it is determined to be standby loss and the redundant power is calculated:

[0151] P recover =P input -P output -P loss ,(P loss is the fixed loss of the converter, about 50W);

[0152] S4302: Send instructions to the bidirectional DC-DC controller via the soft bus to adjust the duty cycle so that the redundant power flows into the battery pack through the energy storage BMS, with a conversion efficiency of >94%;

[0153] S4303: When the charging pile is idle at night, the recovered redundant power can enable the energy storage module to store an additional approximately 5kWh of electricity per day, increasing the annual energy saving rate by 3%-5%;

[0154] This step is the core of strategy execution, achieving seamless energy switching and fault self-healing through the collaboration of hardware and algorithms.

[0155] Step S500 is used for user terminal interaction and demand response, and includes:

[0156] S510:

[0157] S5101: Dynamic curve: Displays the photovoltaic power, energy storage SOC, and charging pile load rate in the past hour;

[0158] Status indicator light: green (normal operation), yellow (low energy storage), red (device failure);

[0159] Real-time electricity prices: Synchronize grid API data and mark peak and valley periods (peak electricity: 8:00-22:00, valley electricity: 22:00-8:00);

[0160] S5102: Automatically select off-peak charging time, target electricity price < 0.5 yuan / kWh;

[0161] Green mode: Only uses photovoltaic / wind power and waits in queue when energy storage is insufficient;

[0162] Fast mode: Ignoring electricity price and energy type, charging at maximum current (additional service fee required)

[0163] S5103: Supports querying charging capacity, electricity cost, and carbon emission reduction by day / week / month, and generates PDF reports;

[0164] S520:

[0165] S5201: Convert the user-selected "green mode" to JSON format:

[0166] {

[0167] "command":"set_strategy",

[0168] "strategy_id":"green_mode",

[0169] "parameters":{

[0170] "energy_source":"renewable_only",

[0171] "max_wait_time":120 / / Maximum waiting time (minutes)

[0172] }

[0173] }

[0174] S5202: The user command has a higher priority than the default policy, but a lower priority than the emergency fault handling command (such as energy storage overload protection).

[0175] S5203: After receiving the instruction, the MS re-runs the artificial bee colony algorithm and generates new control parameters under the user constraints (such as using only renewable energy). For example:

[0176] Increase the photovoltaic power supply threshold to 40% (originally 70%) to maximize the use of photovoltaic energy;

[0177] When the energy storage SOC is less than 10% and there is no sunlight, a push notification will be sent to the user: "There is currently no renewable energy available. Do you want to switch to the mains?"

[0178] S530:

[0179] S5301: Demand response rate (DRR) is used to evaluate user participation. The calculation formula is:

[0180]

[0181] P base is the baseline power when there is no user participation, P actual,i is the actual power after response, t i is the response duration);

[0182] In a typical scenario, user participation can reduce peak load by 15%-20%, increase off-peak load by 10%-15%, and reduce the peak-to-off-peak difference of the power grid by about 12%.

[0183] The above steps achieve coordinated optimization of the three aspects of “source-storage-load” and reduce users’ electricity costs by 15%-20%.

[0184] Step S600 is used for system closed-loop optimization and continuous iteration, and includes:

[0185] S610:

[0186] S6101: Actual charge capacity (kWh) = end charge capacity - start charge capacity;

[0187] Charging efficiency (%) = output energy / input energy × 100%;

[0188] Charging waiting time (minutes) = the time interval from when the system receives the request to when charging starts;

[0189] S6102: Depth of charge / discharge (DOD) = 1 - SOC_end / SOC_start;

[0190] Cycle life loss (%) = equivalent number of cycles consumed in this charge and discharge / rated number of cycles × 100%;

[0191] Temperature field uniformity (°C) = maximum monomer temperature - minimum monomer temperature;

[0192] Data cleaning steps:

[0193] Eliminate abnormal records with charging efficiency less than 80% (possibly due to equipment failure);

[0194] Extreme operating conditions with DOD > 90% are marked as "high loss events" to optimize charging and discharging strategies.

[0195] S620:

[0196] S6201: Uses the Flink stream processing engine to analyze the execution results of newly collected control instructions in real time. If the charging wait time exceeds 30 minutes, it immediately triggers a local policy adjustment.

[0197] At 2:00 AM every day, we clean all historical data and use the K-means algorithm to cluster and analyze user charging behavior patterns. For example:

[0198] fromsklearn.clusterimportKMeans

[0199] X = user characteristics such as charging time, priority, and charging capacity

[0200] kmeans=KMeans(n_clusters=3,random_state=0).fit(X);

[0201] The clustering results are used to optimize the input features of the load forecasting model;

[0202] S630:

[0203] S6301: Adjust the decision tree model parameters in four seasons each year:

[0204] Spring (March-May):

[0205] Added a "precipitation probability" split node. When precipitation is greater than 50%, the PV power prediction value is reduced by 20%-30%;

[0206] The energy storage SOC switching threshold has been increased from 30% to 40% to cope with unstable lighting conditions;

[0207] Summer (June-August):

[0208] A "high temperature warning" parameter (temperature > 35°C) is introduced to trigger the active cooling strategy of the energy storage module and reduce the charging power limit by 15%;

[0209] The proportion of users' demand for "fast mode" has increased, and the charging speed weight in the model has increased from 0.2 to 0.3;

[0210] Autumn (September-November):

[0211] Restore normal threshold settings, focusing on optimizing wind power forecasts (increased wind speed fluctuations);

[0212] Start balanced charging of the energy storage module and modify the SOH evaluation weight in the decision tree;

[0213] Winter (December-February):

[0214] A "sunshine duration" feature has been added to the photovoltaic power prediction model. When the threshold is less than 4 hours / day, the system switches to auxiliary power supply from the mains.

[0215] The proportion of users' demand for "economic mode" increased, and the weight of valley electricity increased from 0.3 to 0.4;

[0216] S640:

[0217] S6401: When the load rate of a charging pile is greater than 90% for three consecutive sampling periods (300ms), the following operations are performed:

[0218] Obtain the load rate of charging piles within 50 meters through the soft bus;

[0219] Power split calculation:

[0220] (P overload is the overload power, and m is the number of available adjacent charging piles);

[0221] S6402: Send a power reduction command to the overloaded charging pile and a power increase command to the adjacent charging piles to ensure that the total charging demand remains unchanged;

[0222] Through three years of operational data training, the system's overall energy efficiency has increased from an initial 78% to 92%, and the average user waiting time has been reduced from 12 minutes to less than 5 minutes.

[0223] This step uses machine learning and data-driven methods to enable the system to transition from "passive execution" to "active evolution," adapting to long-term operational challenges such as seasonal changes and equipment aging.

[0224] Experimental example: Energy efficiency verification experiment of distributed soft bus optical storage and charging system;

[0225] Experimental Objective: To verify the effectiveness of the distributed soft bus-based integrated photovoltaic, storage, and charging method in multi-source energy coordinated scheduling, user-participated optimization, and system closed-loop iteration, and to compare the energy efficiency differences of traditional hard-wired connection solutions.

[0226] Experimental environment:

[0227]

[0228] Experimental steps:

[0229] Step 1:

[0230] Connect to photovoltaic panels, energy storage modules, and charging piles through Huawei OceanConnect soft bus, and the equipment plug-and-play time is less than 5 minutes;

[0231] Configure data acquisition channels: photovoltaic power (100ms sampling), energy storage SOC (100ms sampling), charging pile current (100ms sampling);

[0232] Step 2:

[0233] Applying Kalman filtering to photovoltaic power data achieves a noise suppression rate of 92%;

[0234] Predict charging pile load peak based on LSTM model with an error rate of <8%;

[0235] Generate a dynamic weight matrix, and when the light intensity is greater than 800 Lux, the photovoltaic power weight will automatically increase to 45%;

[0236] Step 3:

[0237] The multi-task learning model predicts photovoltaic power with an error rate of less than 5% during the day and less than 7% at night;

[0238] After optimization by the artificial bee colony algorithm, the energy switching threshold is dynamically adjusted to: power supply is prioritized when photovoltaic power is greater than 65% (the fixed threshold for the traditional group is 70%);

[0239] Step 4:

[0240] Daytime test (9:00-15:00, sufficient light):

[0241] Photovoltaic power is given priority, and the remaining power is stored in energy storage. The energy storage charging power is dynamically adjusted to 5-10kW, and there is no abandoned light;

[0242] When the photovoltaic power exceeds the demand of the charging pile, the abandonment rate is 15%, and the energy storage charging power is fixed at 8kW (not optimized);

[0243] Nighttime fault simulation (22:00, abnormal voltage of energy storage module):

[0244] Soft bus group: disconnect the faulty battery within 500ms, start the photovoltaic direct supply, and the charging interruption time is less than 100ms;

[0245] Traditional group: Manual intervention required 15 minutes, and charging interruption time was >10 minutes;

[0246] Step 5:

[0247] Select the "off-peak electricity priority" mode through the soft bus terminal (off-peak electricity period 22:00-6:00, electricity price 0.3 yuan / kWh);

[0248] Delaying non-urgent charging to off-peak hours reduces charging costs by 17%;

[0249] Step 6:

[0250] After 72 hours of continuous operation, the renewable energy utilization rate increased from 82% to 89% through feedback data optimization strategies;

[0251] Key performance indicator comparison, experimental data table is as follows:

[0252]

[0253] The experimental data table of dynamic weight distribution effect (typical time) is as follows:

[0254]

[0255] This solution's integrated photovoltaic, storage, and charging approach based on a distributed soft bus is significantly superior to traditional hard-wired solutions in terms of device interconnection, data processing, intelligent scheduling, user interaction, and system optimization. It realizes the upgrade of the photovoltaic, storage, and charging system from "passive control" to "active intelligence" and has broad engineering application value.

[0256] Example 2: An integrated optical storage and charging system based on a distributed soft bus, comprising the following modules;

[0257] Distributed soft bus module: Establishes plug-and-play communication connections between photovoltaic panels, wind turbines, energy storage modules, charging piles, intelligent control units (EMS), and mains / thermoelectric generators. It serves as a data hub to collect device status data in real time and issue control commands. The soft bus supports OPC UA / MQTT protocols and time-sensitive network (TSN) scheduling, ensuring microsecond-level transmission of control commands.

[0258] Multi-source data processing module: This module collects power generation data, energy storage status data, load data, and environmental data through a soft bus. After noise removal through a smoothing filter algorithm, it generates standardized characteristic data based on the charge-discharge model and the load model. It also generates a dynamic weight allocation matrix and stores it in the photovoltaic storage and charging data center.

[0259] Intelligent prediction and strategy optimization module: This module inputs pre-processed data into a multi-task learning model and combines it with a decision tree correction algorithm to predict power generation and energy storage status. It then uses an artificial bee colony algorithm to optimize the multi-objective decision model, generating a control strategy that includes energy switching thresholds and energy storage charge and discharge power, and reserves a user priority parameter interface.

[0260] Multi-energy collaborative control module: Based on the control strategy, the module prioritizes photovoltaic / wind power during the day and switches between energy storage and mains power at night through TSN. When a storage fault is detected, the EMS disconnects the faulty battery through the soft bus and activates the photovoltaic direct supply link, while also recovering redundant power from the charging pile to the energy storage module.

[0261] User interaction and closed-loop optimization module: The soft bus pushes data to the user terminal. After the user sets the charging priority, the instruction is transmitted to the EMS to adjust the energy storage strategy; the EMS updates the data center based on the feedback data from the charging pile, and corrects the decision tree model parameters through historical data to achieve load balancing and energy optimization.

[0262] The above modules are implemented through a distributed soft bus-based integrated optical storage and charging method in the first embodiment.

[0263] Embodiment 3: A computer device and a readable storage medium, characterized in that the computer device includes: a processor and a memory, the memory storing at least one instruction, at least one program, a code set or an instruction set, the at least one instruction, the at least one program, the code set or the instruction set being loaded and executed by the processor to realize an integrated optical storage and charging system based on a distributed soft bus.

[0264] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An integrated optical storage and charging system based on a distributed soft bus, characterized in that: Includes the following modules: Distributed soft bus module: Establishes plug-and-play communication connections between photovoltaic panels, wind turbines, energy storage modules, charging piles, intelligent control units (EMS), and mains / thermoelectric generators. It serves as a data hub to collect device status data in real time and issue control commands. The soft bus supports OPC UA / MQTT protocols and time-sensitive network (TSN) scheduling, ensuring microsecond-level transmission of control commands. Multi-source data processing module: This module collects power generation data, energy storage status data, load data, and environmental data through a soft bus. After noise removal through a smoothing filter algorithm, it generates standardized characteristic data based on the charge-discharge model and the load model. It also generates a dynamic weight allocation matrix and stores it in the photovoltaic storage and charging data center. Intelligent prediction and strategy optimization module: This module inputs pre-processed data into a multi-task learning model and combines it with a decision tree correction algorithm to predict power generation and energy storage status. It then uses an artificial bee colony algorithm to optimize the multi-objective decision model, generating a control strategy that includes energy switching thresholds and energy storage charge and discharge power, and reserves a user priority parameter interface. Multi-energy collaborative control module: Based on the control strategy, the module prioritizes photovoltaic / wind power during the day and switches between energy storage and mains power at night through TSN. When a storage fault is detected, the EMS disconnects the faulty battery through the soft bus and activates the photovoltaic direct supply link, while also recovering redundant power from the charging pile to the energy storage module. User interaction and closed-loop optimization module: The soft bus pushes data to the user terminal. After the user sets the charging priority, the instruction is transmitted to the EMS to adjust the energy storage strategy; the EMS updates the data center based on the feedback data from the charging pile, and corrects the decision tree model parameters through historical data to achieve load balancing and energy optimization.

2. The integrated optical storage and charging system based on a distributed soft bus according to claim 1 is characterized by: The distributed soft bus module adopts an industrial-grade Internet of Things protocol stack, supports the OPCUA protocol for real-time control instruction transmission, the MQTT protocol for low-power device data acquisition, and has a built-in protocol conversion engine to achieve normalized conversion of heterogeneous data formats such as ModbusRTU and CAN bus.

3. The integrated photovoltaic storage and charging system based on a distributed soft bus according to claim 1 is characterized in that: In the multi-source data processing module, the smoothing filter algorithm adopts Kalman filtering or 3σ principle for denoising, with a noise suppression rate of ≥90% for photovoltaic power data and an accuracy rate of eliminating abnormal values ​​of energy storage voltage data of ≥95%.

4. The integrated optical storage and charging system based on a distributed soft bus according to claim 1 is characterized by: In the intelligent prediction and strategy optimization module, the multi-task learning model uses a bidirectional LSTM network combined with a convolutional layer structure. The input layer contains 10-dimensional features such as photovoltaic efficiency, energy storage SOC, and light intensity. The output layer parallelly predicts the power generation power and energy storage charging and discharging status in the next 15 minutes, with a prediction error rate of ≤7%.

5. The integrated photovoltaic storage and charging system based on a distributed soft bus according to claim 1 is characterized by: In the user interaction and closed-loop optimization module, the user terminal supports at least three preset strategies, including "economic mode," "green mode," and "fast mode." "Green mode" prioritizes the use of renewable energy and delays charging for a maximum of 120 minutes when energy storage is insufficient.

6. The integrated optical storage and charging system based on a distributed soft bus according to claim 1 is characterized by: In the multi-energy collaborative control module, the time-sensitive network (TSN) adopts the IEEE802.1Qbv gated list mechanism to assign the highest priority queue to the control instructions, ensuring that the transmission delay is ≤100μs and the energy switching interruption time is ≤200ms.

7. The integrated photovoltaic storage and charging system based on a distributed soft bus according to claim 1 is characterized by: In the multi-energy collaborative control module, the energy storage module fault detection threshold is that the single cell voltage deviation from the mean is greater than 15% or the temperature difference is greater than 5°C, the fault switching time is ≤500ms, and the photovoltaic direct supply link conversion efficiency is ≥96%.

8. The integrated optical storage and charging system based on a distributed soft bus according to claim 1 is characterized by: In the closed-loop optimization module, the decision tree model is automatically updated every hour, and the weights of environmental factors are adjusted annually according to seasonal characteristics, so that the long-term energy efficiency of the system is improved by ≥8% and the user demand response rate is ≥90%.

9. A method for integrating optical storage and charging based on a distributed soft bus, for realizing an integrated optical storage and charging system based on a distributed soft bus as claimed in any one of claims 1 to 8, characterized in that: The following steps are involved: S100: Establishing photovoltaic modules through a distributed soft bus, and then connecting various data collection devices through the distributed soft bus, using the distributed soft bus as a data center to collect status data of each device in real time and issue control instructions; S200: The status data of each device collected through the soft bus is analyzed and generated into standardized charge-discharge characteristic data and load characteristic data based on the charge-discharge model and the load model. A dynamic weight distribution matrix is ​​also generated and stored in the solar storage and charging data center; S300: Input the pre-processed power generation performance index and energy storage status index into the multi-task learning model, then dynamically adjust the model parameters in combination with the decision tree correction algorithm to predict photovoltaic / wind power and energy storage charging and discharging status. Subsequently, the dynamic weight allocation matrix is ​​input into the multi-objective decision model optimized by the artificial bee colony algorithm to generate a target control strategy including multi-energy switching thresholds and energy storage charging and discharging power, and reserve a user priority parameter interface; S400: Based on the target control strategy, the control instruction set is distributed through the time-sensitive network. When an energy storage module fault is detected, the EMS disconnects the faulty battery through the soft bus control, activates the photovoltaic controller and DCDC module to directly supply power, and simultaneously feeds the redundant power of the charging pile back to the energy storage module through the soft bus for recycling; S500: Data is pushed to the user terminal via the soft bus. The user sets the charging priority through the terminal. The soft bus transmits the demand instructions to the EMS to adjust the energy storage discharge strategy. S600: The EMS receives feedback data from charging piles, updates the solar-storage-charging data center through the soft bus, and modifies the decision tree model parameters based on historical data distribution and environmental factors to achieve load balancing and energy optimization.

10. A computer device and a readable storage medium, characterized in that: The computer device includes: a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement an integrated optical storage and charging system based on a distributed soft bus as described in any one of claims 1 to 8.

Images