A building photovoltaic storage and charging integrated energy management system

Abstract

The present invention relates to the field of data processing technology, and in particular to a building photovoltaic, energy storage and charging integrated energy management system, which is used to solve the problems of data processing delay and resource coordination mismatch caused by insufficient fusion of multi-source heterogeneous data in photovoltaic power generation, energy storage and grid connection. The system includes a data acquisition module, a protocol conversion module, a data fusion module, a scheduling optimization module, an instruction generation module and a feedback module. The protocol conversion module performs Modbus, CAN bus and IEC 61850 protocol parsing on photovoltaic output time series data, energy storage charging and discharging status data and grid scheduling instructions, respectively, to generate standardized data streams. The safety verification module intercepts over-limit instructions and triggers a limiting strategy. Through protocol unification, dynamic fusion and closed-loop control, the system achieves an increase in photovoltaic absorption rate, a reduction in energy storage cycle loss and an optimization of grid frequency regulation deviation, significantly improving the efficiency and economy of the coordinated operation of photovoltaic, energy storage and charging.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a building photovoltaic storage and charging integrated energy management system. Background Art

[0002] In scenarios where renewable energy generation and energy storage systems operate in tandem, multidimensional data on photovoltaic power generation output volatility, the dynamic response of energy storage charging and discharging, and load demand exhibit temporal differences and coupling characteristics. Existing energy management systems are limited in their ability to coordinate the real-time scheduling of heterogeneous data from multiple sources. The lack of a unified data fusion mechanism for the transmission protocol, sampling cycle, and control objectives of the discrete operational data generated by photovoltaic generation, energy storage, and substation grid connection makes it difficult for the system to dynamically adapt to the real-time matching of grid dispatch instructions with local operating states. This leads to data processing delays and resource coordination mismatches, impacting the overall operational efficiency and economic viability of the photovoltaic, energy storage, and charging system. Summary of the Invention

[0003] In response to the shortcomings of existing technologies, the present invention provides a building integrated photovoltaic, energy storage and charging energy management system to solve the problem that existing energy management systems lack a unified fusion mechanism for multi-source heterogeneous data in photovoltaic power generation, energy storage and grid connection, resulting in data processing delays and resource coordination mismatches, thereby reducing the operating efficiency and economy of the photovoltaic, energy storage and charging system.

[0004] In order to solve the above technical problems, the specific technical solutions of the present invention are as follows:

[0005] The present invention provides a building photovoltaic storage and charging integrated energy management system, comprising:

[0006] A data acquisition module is used to acquire multi-source heterogeneous data, including photovoltaic output time series data, energy storage charge and discharge status data, charging load demand data, and grid dispatch instruction data;

[0007] A protocol conversion module is used to perform protocol parsing on the multi-source heterogeneous data received from the data acquisition module, parsing the photovoltaic output time series data through the Modbus protocol, parsing the energy storage charge and discharge status data through the CAN bus protocol, and parsing the grid dispatch instruction data through the IEC 61850 protocol, to generate a standardized data stream in a unified data format;

[0008] a data fusion module, configured to perform a sliding window interception on the standardized data stream received from the protocol conversion module, extract the fluctuation characteristic vectors in the photovoltaic output time series data and the dynamic response vectors in the energy storage charge and discharge state data, input the fluctuation characteristic vectors and the dynamic response vectors into a dynamic weighted model constructed based on the photovoltaic output fluctuation rate and the energy storage response delay, perform a time-domain convolution operation, and generate a multi-dimensional coupling state matrix;

[0009] a dispatch optimization module, configured to construct a multi-objective function with photovoltaic absorption rate, energy storage cycle loss, and grid frequency deviation as optimization targets based on the multi-dimensional coupling state matrix received from the data fusion module and the grid dispatch instruction data received from the data acquisition module, and to generate a real-time dispatch strategy using an improved particle swarm algorithm with inertia weight dynamically adjusted according to photovoltaic output fluctuation rate;

[0010] An instruction generation module is used to convert the charge and discharge power values in the real-time scheduling strategy output by the scheduling optimization module into PWM modulation signals, and distribute the PWM modulation signals to the inverter in the photovoltaic power generation unit, the BMS system in the energy storage unit, and the charging pile controller in the charging facility unit via the OPC UA protocol;

[0011] A feedback module is used to monitor the actual output data of the photovoltaic power generation unit, the real-time charge status data of the energy storage unit, and the power consumption data of the charging facility unit, calculate the deviation between the actual output data and the predicted value in the real-time scheduling strategy, generate a feedback correction coefficient, and input the feedback correction coefficient into the scheduling optimization module to trigger the iterative optimization of the multi-objective function and adjust the real-time scheduling strategy.

[0012] Furthermore, in the building photovoltaic storage and charging integrated energy management system of the present invention, the protocol conversion module is further used to:

[0013] Performing Modbus protocol parsing on the photovoltaic output time series data received from the data acquisition module to extract photovoltaic power generation power and voltage fluctuation characteristics;

[0014] Performing CAN bus protocol analysis on the energy storage charge and discharge status data received from the data acquisition module to extract the state of charge and charge and discharge rate;

[0015] Performing IEC 61850 protocol parsing on the power grid dispatch instruction data received from the data acquisition module to extract the grid-connected power threshold and frequency modulation instruction parameters;

[0016] The photovoltaic power generation power, voltage fluctuation characteristics, state of charge, charge and discharge rate, grid-connected power threshold and frequency modulation instruction parameters are output as analysis results to the data fusion module.

[0017] Furthermore, in the building solar-storage-charging integrated energy management system of the present invention, the data fusion module includes:

[0018] a feature extraction unit, configured to perform sliding window interception on the standardized data stream received from the protocol conversion module, and extract the fluctuation feature vectors in the photovoltaic output time series data and the dynamic response vectors in the energy storage charge and discharge state data;

[0019] The coupling analysis unit is used to input the fluctuation characteristic vector and the dynamic response vector into a dynamic weighted model constructed based on the photovoltaic output fluctuation rate and the energy storage response delay to generate the multi-dimensional coupling state matrix.

[0020] Furthermore, the building photovoltaic storage and charging integrated energy management system of the present invention, the coupling analysis unit further includes:

[0021] The photovoltaic output fluctuation rate is calculated by the variance of the fluctuation characteristic vector;

[0022] The energy storage response delay is calculated by using the time delay parameter in the dynamic response vector;

[0023] The dynamic weighted model dynamically adjusts the weight coefficient according to the ratio of the photovoltaic output fluctuation rate to the energy storage response delay, performs weighted superposition on the fluctuation characteristic vector and the dynamic response vector, and generates a coupling characteristic tensor;

[0024] The coupling characteristic tensor is matrix-processed to obtain the multi-dimensional coupling state matrix.

[0025] Furthermore, in the building solar-storage-charging integrated energy management system of the present invention, the scheduling optimization module is further configured to:

[0026] generating a charge and discharge power allocation plan based on the photovoltaic output prediction value and the energy storage remaining capacity in the multi-dimensional coupling state matrix received from the data fusion module, wherein the photovoltaic output prediction value is calculated by regression analysis of the fluctuation eigenvector;

[0027] Combined with the grid-connected power threshold in the grid dispatch instruction data received from the data acquisition module, the charging and discharging power allocation scheme is subjected to constrained optimization, and the constrained optimization is based on the multi-objective function to generate the real-time dispatch strategy.

[0028] Furthermore, the building photovoltaic storage and charging integrated energy management system of the present invention further includes:

[0029] Based on the photovoltaic output prediction value in the multi-dimensional coupling state matrix received from the data fusion module and the weight coefficient output from the dynamic weighting model, a multi-objective function is constructed with the goal of maximizing the photovoltaic absorption rate, minimizing the energy storage cycle loss, and minimizing the grid frequency deviation;

[0030] An improved particle swarm algorithm is used to solve the multi-objective function, wherein the inertia weight of the improved particle swarm algorithm is dynamically adjusted according to the photovoltaic output fluctuation rate to generate a Pareto optimal solution set;

[0031] The real-time scheduling strategy is selected from the Pareto optimal solution set according to the priority weight in the power grid scheduling instruction data received from the data acquisition module.

[0032] Furthermore, in the building solar-storage-charging integrated energy management system of the present invention, the instruction generation module includes:

[0033] an instruction encoding unit, configured to convert the charge and discharge power values in the real-time scheduling strategy output by the scheduling optimization module into PWM modulation signals based on the Pareto optimal solution set generated by the scheduling optimization module;

[0034] The instruction distribution unit is used to synchronously send the PWM modulation signal to the BMS system in the energy storage unit and the charging pile controller in the charging facility unit through the OPC UA protocol.

[0035] Furthermore, in the building solar-storage-charging integrated energy management system of the present invention, the feedback module is further configured to:

[0036] Collecting actual output data of the photovoltaic power generation unit, real-time state of charge data of the energy storage unit, and power consumption data of the charging facility unit;

[0037] Calculating a power deviation rate based on a deviation between the actual output data and a predicted value in the real-time scheduling strategy generated by the scheduling optimization module;

[0038] generating a feedback correction coefficient based on the power deviation rate and the energy storage response delay obtained from the data fusion module;

[0039] The feedback correction coefficient is input into the scheduling optimization module to trigger the parameter update of the multi-objective function.

[0040] Furthermore, the building photovoltaic storage and charging integrated energy management system of the present invention further includes:

[0041] Performing a sliding average filter on the deviation between the actual output data and the predicted value, wherein the filter window length is set according to the sampling period of the photovoltaic output time series data;

[0042] Calculating a power compensation amount based on the filtered deviation and the rate of change of the energy storage charge state extracted from the dynamic response vector of the data fusion module;

[0043] The feedback correction coefficient is generated based on the power compensation amount.

[0044] Furthermore, the building photovoltaic storage and charging integrated energy management system of the present invention further includes: a safety verification module for performing overload protection verification and protocol compliance verification on the control instructions distributed by the instruction generation module;

[0045] When the safety verification module detects that the power value in the control instruction exceeds the rated capacity of the equipment acquired by the data acquisition module, it triggers the scheduling optimization module to regenerate the limiting scheduling strategy based on the multi-objective function.

[0046] Beneficial effects of the present invention:

[0047] The beneficial effect of the present invention lies in effectively addressing data heterogeneity and timing discrepancies across photovoltaic, energy storage, and power grid segments through protocol unification and dynamic fusion of multi-source heterogeneous data. The protocol conversion module utilizes Modbus, CAN bus, and IEC 61850 protocols to parse photovoltaic output, energy storage status, and grid dispatch instructions, generating standardized data streams that eliminate differences in transmission protocols and sampling periods, thereby improving data processing efficiency. The data fusion module, through sliding window interception and a dynamic weighting model, calculates weight coefficients based on photovoltaic fluctuations and energy storage response delays to generate a multidimensional coupling state matrix. This quantitatively characterizes the photovoltaic-storage synergy state and provides high-precision input for dispatch optimization. The dispatch optimization module constructs a multi-objective function and employs an improved particle swarm algorithm to generate a Pareto-optimal solution set. This dynamically generates real-time strategies based on grid priorities to optimize the balance between photovoltaic absorption rate, energy storage loss, and frequency regulation deviation. The feedback module iteratively adjusts optimization parameters using deviation correction coefficients, and the safety verification module constrains equipment operating boundaries, forming a closed-loop control link, ultimately improving system resource coordination efficiency and economic efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solution of the present invention, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, for ordinary technicians in this field, other drawings can be obtained based on the drawings without paying any creative labor.

[0049] Figure 1 This is a system architecture diagram of a building photovoltaic storage and charging integrated energy management system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the specific embodiments of the present invention and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. The technical solutions provided by each embodiment of the present invention are described in detail below in conjunction with the drawings. In order to better understand the purpose of the present invention, the present invention is further described in detail below.

[0051] See also Figure 1The present invention provides a building photovoltaic storage and charging integrated energy management system, comprising:

[0052] A data acquisition module is used to acquire multi-source heterogeneous data, including photovoltaic output time series data, energy storage charge and discharge status data, charging load demand data, and grid dispatch instruction data;

[0053] The data acquisition module uses distributed sensors and communication interfaces to collect real-time data on the PV unit's light intensity, module temperature, and inverter output parameters, generating time-series data on PV output. This data, including instantaneous active power, reactive power, and voltage, is transmitted to the protocol conversion module via the RS-485 bus. The sampling frequency is adaptively adjusted based on the rate of change in light intensity. In high-intensity scenarios, a high-frequency sampling rate of 1 second is used to capture rapid fluctuations.

[0054] The energy storage unit uses a battery management system (BMS) to collect information about the battery pack's individual voltage, total voltage, charge / discharge current, and temperature. This information, combined with relay switch status, generates charge / discharge status data. The BMS utilizes multi-channel parallel data collection technology, updating the battery's state of health (SOH) and state of charge (SOC) parameters every 100 milliseconds. This data is transmitted via the CAN bus, using a frame format that complies with the SAE J1939 standard. During parsing, redundant frames are filtered using identifiers to reduce communication overhead.

[0055] Charging facility units collect charging power requirements, user reservation information, and real-time load profiles through charging pile controllers. The controllers interact with the central management system using the Modbus TCP protocol, synchronizing the charging pile's operating status every five seconds, including connection status, charging mode, and fault codes. Combined with user reservation periods, they generate dynamic load forecast curves, which are aggregated into charging load demand data.

[0056] Grid dispatch command data is received through a communication gateway, including grid-connected power limits, frequency regulation signals, and market electricity price information issued by the upper-level energy management system. The gateway parses dispatch commands using the IEC 104 protocol, extracting grid-connected active power thresholds, frequency regulation deadbands, and economic dispatch coefficients. Command data is synchronized with locally collected data through a timestamp alignment mechanism, eliminating timing deviations caused by communication delays. After data collection is complete from each data source, the module performs abnormal data cleaning, discarding anomalous sampling points that exceed the device's rated range. Missing data is supplemented through interpolation to form a complete multi-source heterogeneous dataset.

[0057] A protocol conversion module is used to perform protocol parsing on the multi-source heterogeneous data received from the data acquisition module, parsing the photovoltaic output time series data through the Modbus protocol, parsing the energy storage charge and discharge status data through the CAN bus protocol, and parsing the grid dispatch instruction data through the IEC 61850 protocol, to generate a standardized data stream in a unified data format;

[0058] After receiving the PV output time-series data from the data acquisition module, the protocol conversion module decodes the data frame using the Modbus protocol parsing mechanism. The data frame structure locates the instantaneous values of the inverter's output active power, reactive power, and voltage based on a register address mapping table. Register addresses 0x0001 to 0x0003 correspond to the high and low bits of active power and the checksum field, respectively. During the parsing process, the CRC checksum of the data frame is verified. If the check fails, a retransmission request mechanism is triggered, invalid data packets are discarded, and an exception log is recorded. Voltage fluctuation characteristics are calculated by calculating the peak-to-peak value of consecutive voltage samples within a sliding window. The window length matches the PV output sampling frequency to reflect the short-term fluctuation amplitude characteristics.

[0059] For energy storage charge and discharge status data, the protocol conversion module parses the multiplexed signals within the data domain based on the CAN bus protocol. The battery state of charge (SOC) is obtained by parsing the standard data frame corresponding to CAN ID 0x18FEF100. The data bytes are converted to actual values according to the scaling factors and offsets defined in the SAE J1939 protocol. For example, bytes 3-4 represent the SOC percentage. The charge and discharge rate is determined by parsing the raw data from the battery pack current sensor and calculating the current difference between adjacent sampling points based on the timestamp reported by the BMS to generate the current change gradient per unit time. The parsing process adapts to the CAN bus's non-destructive arbitration mechanism, prioritizing high-priority data frames to avoid bus conflicts while filtering out invalid frames and noise interference data.

[0060] When grid dispatch command data is parsed using the IEC 61850 protocol, the protocol conversion module invokes the MMS (Manufacturing Message Specification) service to access the DataSet in the logical device model. The grid-connected power threshold is extracted from the Class A data attributes of the MMXU logical node, parsing the floating-point value and unit identifier. For example, the VolMag attribute represents the voltage amplitude, and the HzMag represents the frequency reference value. Frequency regulation command parameters are obtained by parsing the FrequencySetpoint (frequency setpoint) and DeadBand (deadband range) attributes in the FSCH logical node to obtain the frequency regulation target value and allowable deviation range. During the parsing process, the communication service mapping relationship described in the SCL (System Configuration Language) file is dynamically loaded to convert abstract data types into engineering values recognizable by the device. For example, enumerated state variables in the logical node are mapped to binary control signals.

[0061] After parsing is complete, the protocol conversion module performs data formatting on photovoltaic power generation, voltage fluctuation characteristics, state of charge, charge and discharge rate, grid-connected power threshold, and frequency modulation instruction parameters. The unified data format uses a JSON-LD structured representation, with each parameter appended with a source protocol type tag, acquisition timestamp, and data confidence score tag. Voltage fluctuation characteristics and charge and discharge rates are converted to per-unit values through normalization, for example, by dividing the peak-to-peak voltage by the rated voltage value to eliminate dimensional differences. The formatted data stream is transmitted to the data fusion module via the message queue middleware. The queue adopts a priority scheduling strategy, with grid dispatch instruction data marked as high priority to ensure real-time performance. Photovoltaic and energy storage data are arranged in timestamp order to avoid fusion errors caused by timing misalignment.

[0062] a data fusion module, configured to perform a sliding window interception on the standardized data stream received from the protocol conversion module, extract the fluctuation characteristic vectors in the photovoltaic output time series data and the dynamic response vectors in the energy storage charge and discharge state data, input the fluctuation characteristic vectors and the dynamic response vectors into a dynamic weighted model constructed based on the photovoltaic output fluctuation rate and the energy storage response delay, perform a time-domain convolution operation, and generate a multi-dimensional coupling state matrix;

[0063] After receiving the standardized data stream output by the protocol conversion module, the data fusion module dynamically segments the data using a sliding window interception technique. The window length is adaptively adjusted based on the fluctuation period of the PV output time series data: when the fluctuation period is less than 10 seconds, the window length is set to 30 seconds to capture high-frequency fluctuations; when the fluctuation period exceeds 60 seconds, the window length is extended to 5 minutes to analyze trend changes. The window sliding step size is synchronized with the data sampling interval to avoid redundant calculations caused by data overlap while maintaining the continuity of the time series.

[0064] The feature extraction unit performs a discrete wavelet transform on the PV output time series data within the window, decomposing high-frequency detail components and low-frequency approximation components. The fundamental amplitude is calculated by averaging the low-frequency components, the harmonic distortion rate is determined by analyzing the energy proportion of the high-frequency components, and the fluctuation frequency is calculated by counting the number of fluctuations per unit time using a zero-crossing detection algorithm, together forming a fluctuation feature vector. The energy storage charge and discharge state data is processed within the same time window, and the dynamic response vector is generated by calculating the instantaneous rate of change of the state of charge (SOC), the rise time of the charge and discharge current, and the temperature gradient. The SOC rate of change is calculated using the differential method to calculate the capacity difference between adjacent timestamps, and the current rise time is obtained by measuring the delay after the step response trigger.

[0065] The dynamic weighting model dynamically assigns weight coefficients based on the ratio of PV output volatility to energy storage response delay. Volatility is calculated as the variance of the fundamental amplitude in the fluctuation eigenvector, reflecting the degree of power instability. Response delay is assessed as the sliding average of the current rise time in the dynamic response vector, characterizing the efficiency of energy storage regulation. When volatility exceeds 15%, the model increases the PV feature weight to 0.7 to enhance fluctuation suppression. When response delay exceeds 500 milliseconds, the energy storage feature weight is increased to 0.6 to optimize regulation compensation. The weighted eigenvectors are fused along the feature axis using tensor concatenation technology to form a three-dimensional coupled feature tensor containing time series, feature categories, and weight dimensions.

[0066] During the matrix processing phase, the three-dimensional tensor is sliced and expanded along the time axis, with each time slice corresponding to a two-dimensional feature matrix. The rows of the matrix represent the parameter relationships between the PV fluctuation amplitude, harmonic distortion rate, energy storage SOC change rate, and current response time, while the columns record the evolution path of the same parameter in different time windows. Consecutive time slices are stacked in the order of acquisition to form a multidimensional coupled state matrix with a three-dimensional structure of rows, columns, and time. This matrix provides the scheduling optimization module with structured input data that includes device interaction characteristics and time evolution patterns, supporting the coordinated optimization of PV absorption rate, energy storage loss, and frequency regulation deviation in the multi-objective optimization model.

[0067] a dispatch optimization module, configured to construct a multi-objective function with photovoltaic absorption rate, energy storage cycle loss, and grid frequency deviation as optimization targets based on the multi-dimensional coupling state matrix received from the data fusion module and the grid dispatch instruction data received from the data acquisition module, and to generate a real-time dispatch strategy using an improved particle swarm algorithm with inertia weight dynamically adjusted according to photovoltaic output fluctuation rate;

[0068] After receiving the multidimensional coupling state matrix transmitted by the data fusion module, the dispatch optimization module extracts the PV output forecast and energy storage remaining capacity parameters from the matrix's row and column dimensions. The PV output forecast is generated through time series regression analysis of the fluctuation characteristic vector. An adaptive sliding window mechanism is used to match the PV fluctuation cycle. A linear model is trained based on historical fluctuation characteristics to predict output trends in future time windows. The energy storage remaining capacity is calculated by combining the real-time state of charge value in the dynamic response vector with the battery aging coefficient and temperature attenuation factor. After correcting for capacity estimation errors, available energy reserve data is generated.

[0069] In constructing the multi-objective function, the PV absorption rate target is defined as maximizing the matching of PV output with local load and energy storage charging power. This matching is calculated by the phase consistency and amplitude overlap between the output curve and the demand curve. The energy storage cycle loss target is achieved by minimizing the weighted sum of the charge and discharge depth and the current change rate. The charge and discharge depth is calculated based on the amplitude of the state of charge change, and the loss coefficient is corrected in combination with battery temperature parameters. The grid frequency deviation target is achieved by constraining the root mean square error between the actual output frequency of the system and the dispatch command frequency. Frequency deviation data is acquired in real time from the grid-side synchronous acquisition device and dynamically updated to the optimization model.

[0070] The improved particle swarm optimization algorithm's inertia weight adjustment mechanism is dynamically linked to the PV output volatility. The volatility is calculated using the variance of the PV characteristic parameters in the multidimensional coupled state matrix. For every 5% increase in volatility, the inertia weight is linearly decreased by 0.1 to narrow the global search range and accelerate algorithm convergence. As volatility decreases, the weight is gradually restored to enhance global optimization under stable conditions. During the algorithm's initialization phase, a non-dominated sorting strategy is used to generate the initial Pareto solution set. During the iteration process, congestion distance calculation is used to maintain the diversity of the solution set distribution and avoid clustering of local optimal solutions. The Pareto solution set is selected based on the economic dispatch coefficient, frequency regulation urgency, and market electricity price fluctuation parameters embedded in the grid dispatch instructions. A weighted score table is generated for each optimization item in the objective function. The solution with the highest score is output as the real-time dispatch strategy.

[0071] After the real-time scheduling strategy is generated, the module performs a secondary verification process, comparing the power allocation value in the strategy with the real-time state of charge of the energy storage unit and the actual sampled PV output. If the remaining energy storage capacity is detected to be lower than the strategy requirement or the PV forecast value deviates from the actual value by more than 10%, the strategy backtracking mechanism is triggered, re-calling the historical coupling state matrix to iteratively generate a revised scheduling strategy. Before the revised strategy is distributed through the instruction generation module, the dynamic weight coefficients in the multi-dimensional coupling state matrix are synchronously updated, forming a closed-loop optimization link to maintain the system's scheduling stability and adaptability under dynamic changes in multi-source data.

[0072] An instruction generation module is used to convert the charge and discharge power values in the real-time scheduling strategy output by the scheduling optimization module into PWM modulation signals, and distribute the PWM modulation signals to the inverter in the photovoltaic power generation unit, the BMS system in the energy storage unit, and the charging pile controller in the charging facility unit via the OPC UA protocol;

[0073] After the command generation module receives the real-time scheduling policy output by the scheduling optimization module, the command encoding unit normalizes the charge and discharge power values. This normalization process converts the power allocation value into a per-unit value in the 0-1 range based on the rated output power of the PV inverter, the maximum charge and discharge rate of the energy storage battery pack, and the load capacity of the charging station. This per-unit value is mapped to a duty cycle parameter using a pulse width modulation (PWM) encoder. The duty cycle is linearly related to the target power value. For example, a 2MW energy storage charge and discharge power corresponds to a 100% duty cycle. This generates a modulated signal waveform compatible with the inverter, BMS, and charging station controller interfaces. During the encoding process, a slope limit is imposed on the amplitude of sudden changes in the duty cycle, for example, a maximum rate of change of no more than 5% per second, to prevent power command jumps from causing equipment overload or voltage fluctuations.

[0074] The instruction distribution unit establishes a communication link with the execution device via the OPC UA protocol, defining a unified information model. The PWM modulated signal is encapsulated as OPC UA node attributes. The node contains the signal amplitude, frequency, effective timestamp, and target device address identifier. For example, the inverter node address is marked as "PV_Inverter_01," and the energy storage BMS node is marked as "ESS_BMS_01." The distribution process adopts a publish-subscribe model. Grid dispatch instruction data is marked as high priority, and independent transmission time slots are allocated through the Time-Sensitive Networking (TSN) extension function to ensure the real-time performance of critical instructions. Transmission latency is controlled within 50 milliseconds, and periodic network clock calibration eliminates clock drift between devices, maintaining synchronized execution of instructions on multiple devices.

[0075] Protocol compliance verification is performed before signal transmission to verify whether the node namespace conforms to the OPC UA specification and whether the data field matches the communication interface definition of the target device. For example, it verifies whether the PWM signal frequency of the charging pile controller is within the allowable range of 1kHz-10kHz and whether the duty cycle is within the valid range of 0-100%. Signal nodes that fail the verification trigger an alarm log and notify the scheduling optimization module to regenerate the strategy and discard abnormal instructions. After the transmission is completed, the module receives the status confirmation message returned by the execution device. The message contains the command execution result and real-time operating parameters, such as the actual charge and discharge current feedback from the BMS and the output power reported by the inverter, forming a closed-loop feedback link to verify the effectiveness of the command execution.

[0076] An exception handling mechanism is embedded in the command generation process. If the charge and discharge power reported by the energy storage BMS deviates by more than 5% from the command, the module immediately triggers the command rollback process, restoring the power value to the last valid state and sending an interrupt request to the scheduling optimization module. If the charging pile controller detects a mismatch between the charging demand and the power command, such as a user terminating charging prematurely, it initiates local priority arbitration, dynamically adjusts power allocation based on real-time load, and reports the corrected load parameters to the command generation module. This supports the system's dynamic restructuring of the scheduling strategy and maintains the robustness of multi-device collaborative control.

[0077] A feedback module is used to monitor the actual output data of the photovoltaic power generation unit, the real-time charge status data of the energy storage unit, and the power consumption data of the charging facility unit, calculate the deviation between the actual output data and the predicted value in the real-time scheduling strategy, generate a feedback correction coefficient, and input the feedback correction coefficient into the scheduling optimization module to trigger the iterative optimization of the multi-objective function and adjust the real-time scheduling strategy.

[0078] The feedback module periodically collects actual output data from photovoltaic power generation units through distributed sensors, including active power, reactive power, and instantaneous voltage output by the inverter. This data is transmitted via the Modbus protocol and timestamped to the millisecond level. Real-time state-of-charge (SOC) data for energy storage units is extracted from the CAN bus communication frames of the BMS. During parsing, invalid frames are filtered out and converted to actual values based on the SAE J1939 protocol. Capacity estimation errors are corrected based on battery temperature parameters. Power consumption data for charging facility units is obtained through the Modbus TCP interface of the charging pile controller. This data is synchronized with user reservation information and real-time load curves to generate a dynamic load data set.

[0079] Power deviation is calculated using a root mean square error (RMSE) algorithm within a sliding window, with the window length matching the scheduling optimization module's forecast period. The algorithm calculates the point-by-point difference between the actual output data and the predicted value, sums the squares, averages them, and takes the square root to generate a scalar indicator reflecting the overall deviation level. A sliding average filter is introduced during the calculation process, and the window length is dynamically adjusted based on the PV output sampling frequency: a short 5-second window is used to filter out transient noise in high-frequency sampling scenarios, while a long 30-second window is used in low-frequency scenarios to preserve trend deviation characteristics. The filtered deviation data is combined with the energy storage response delay parameter, extracted from the dynamic response vector, to characterize the system's efficiency in compensating for historical deviations.

[0080] The feedback correction coefficient is calculated by multiplying the deviation rate by the time delay, where the deviation rate is normalized to a per-unit value in the range of 0-1. When the energy storage response delay exceeds 500 milliseconds, the correction coefficient is proportionally increased to enhance compensation strength. When the deviation rate is persistently high, a dynamic attenuation factor is introduced, with the factor value increasing with the number of consecutive deviation cycles to suppress overcorrection caused by short-term disturbances. After the correction coefficient is input into the scheduling optimization module, it triggers an iterative update of the weight parameters of the multi-objective function: if the deviation is primarily caused by PV output fluctuations, the weight coefficient of the PV absorption rate target is increased; if the deviation is due to lag in energy storage response, the optimization weight of the energy storage cycle loss target is increased.

[0081] The closed-loop control mechanism builds an experience database by mapping historical correction coefficients to policy execution results. The optimization module prioritizes weighted combinations from similar scenarios in subsequent iterations to accelerate convergence. If the deviation rate fails to drop below the set threshold after three consecutive corrections, the module triggers an abnormality warning process, notifying the data fusion module to recalibrate the parameters of the feature extraction model. This involves adjusting the sliding window length or the weight allocation logic of the dynamic weighting model, optimizing state representation accuracy at the source of the data and maintaining the system's adaptability under complex operating conditions.

[0082] The data acquisition module uses distributed sensors and communication interfaces to collect real-time data on the photovoltaic power generation unit's light intensity, module temperature, and inverter output parameters, generating photovoltaic output time series data. The energy storage unit uses the battery management system (BMS) to collect battery pack voltage, current, and temperature information, and combines this with the charge and discharge switch status to generate energy storage charge and discharge status data. The charging facility unit uses the charging pile controller to collect charging power requirements, user reservation information, and real-time load curves, which are aggregated into charging load demand data. Grid dispatch command data receives grid-connected power limits, frequency modulation signals, and market electricity price information issued by the upper-level dispatch system via a communication gateway. Data from each data source is aligned using a timestamp synchronization mechanism to eliminate timing misalignment caused by differences in sampling periods.

[0083] The protocol conversion module uses the Modbus protocol to parse photovoltaic output time-series data, extracting the inverter's output active power, reactive power, and voltage fluctuation characteristics. Energy storage charge and discharge status data is parsed via the CAN bus protocol, extracting battery state-of-charge (SOC), charge and discharge current rates, and state-of-health (SOH) parameters. Grid dispatch command data is parsed via the IEC 61850 protocol, extracting grid-connected power thresholds, frequency modulation command amplitudes, and frequency response requirements. During the parsing process, the protocol conversion module performs integrity checks on data frames, discarding abnormal or failed packets. It then performs standardized mapping of data fields across different protocols to generate a standardized data stream with uniform timestamps, data tags, and value ranges.

[0084] The data fusion module applies a sliding window capture technique to the standardized data stream, with the window length dynamically adjusted based on the fluctuation period of the PV output data. The feature extraction unit performs Fourier transform and wavelet decomposition on the PV output time series data within the window, extracting the fundamental and harmonic component characteristics and constructing a fluctuation feature vector reflecting the amplitude and frequency of the output fluctuation. The dynamic response vector is calculated using the SOC change rate, charge and discharge current response time, and temperature gradient from the energy storage charge and discharge status data. The coupling analysis unit inputs the fluctuation feature vector and the dynamic response vector into a dynamic weighted model. This model dynamically assigns weight coefficients based on the calculated variance of the PV output fluctuation rate and the inverse of the energy storage response delay, generating a coupling feature tensor through weighted superposition. Matrix processing expands the tensor along the time dimension, forming a multidimensional coupling state matrix containing spatiotemporal correlation features.

[0085] The dispatch optimization module constructs a multi-objective function based on the predicted PV output and remaining energy storage capacity in the multi-dimensional coupling state matrix, combined with the grid-connected power threshold in the grid dispatch instruction. The PV absorption rate objective is defined by maximizing the matching degree between PV output and load demand and energy storage charging power; the energy storage cycle loss objective is defined by minimizing the charge and discharge depth and current change rate; and the grid frequency deviation objective is defined by minimizing the frequency deviation between actual output and frequency regulation instructions. The inertia weight of the improved particle swarm algorithm adjusts linearly with the PV output volatility. When volatility is high, the global search weight is reduced to enhance local optimization capabilities. After the algorithm solves and generates a Pareto optimal solution set, the comprehensive optimal solution is selected according to the grid priority weights as the real-time dispatch strategy.

[0086] The command generation module converts the charge and discharge power allocation values in the real-time scheduling strategy into a modulation signal with an adjustable duty cycle using a pulse-width modulation (PWM) encoder. During the encoding process, the charge and discharge power values are normalized according to the device's rated capacity, generating PWM waveform parameters compatible with the inverter, BMS system, and charging pile controller interface. The command distribution unit establishes a bidirectional communication link via the OPC UA protocol, synchronously sending modulation signals to each execution unit and receiving device status confirmation messages, ensuring real-time and reliable command transmission.

[0087] The feedback module periodically acquires actual output data from photovoltaic power generation units, real-time SOC data from energy storage units, and actual power consumption data from charging facilities through the data acquisition module. The power deviation rate is calculated by the root mean square error (RMS) between the actual output and the predicted value. This power deviation factor is then factored into the effect of energy storage response delay on the propagation of the deviation to generate a feedback correction coefficient. This correction coefficient, when input into the scheduling optimization module, triggers the iterative update of the weight parameters in the multi-objective function, dynamically adjusting the optimization direction of subsequent scheduling strategies and forming a closed-loop control mechanism.

[0088] The safety verification module performs overload protection checks on control commands issued by the command generation module. It compares the difference between the power allocation value and the rated capacity of the equipment. If the difference exceeds the safety threshold, the limiting strategy is triggered. Protocol compliance verification is based on the OPC UA communication specification, verifying the integrity of the command fields and the validity of the data format. If the verification fails, the safety verification module sends an interrupt signal to the scheduling optimization module, requesting the regeneration of a scheduling strategy that complies with the equipment's operating boundaries. This prevents equipment damage or system failures caused by incorrect commands.

[0089] Specifically, in the building photovoltaic storage and charging integrated energy management system of the present invention, the protocol conversion module is further used to:

[0090] Performing Modbus protocol parsing on the photovoltaic output time series data received from the data acquisition module to extract photovoltaic power generation power and voltage fluctuation characteristics;

[0091] Performing CAN bus protocol analysis on the energy storage charge and discharge status data received from the data acquisition module to extract the state of charge and charge and discharge rate;

[0092] Performing IEC 61850 protocol parsing on the power grid dispatch instruction data received from the data acquisition module to extract the grid-connected power threshold and frequency modulation instruction parameters;

[0093] The photovoltaic power generation power, voltage fluctuation characteristics, state of charge, charge and discharge rate, grid-connected power threshold and frequency modulation instruction parameters are output as analysis results to the data fusion module.

[0094] After receiving the PV output time-series data from the data acquisition module, the protocol conversion module decodes the data frame using the Modbus protocol parsing mechanism. The data frame structure includes a register address mapping table, which is used to locate the instantaneous values of the inverter's output active power, reactive power, and voltage. During the parsing process, the protocol conversion module verifies the data frame's CRC checksum. If the check fails, a retransmission request is triggered and invalid data packets are discarded. Voltage fluctuation characteristics are calculated by calculating the peak-to-peak value and rate of change of voltage samples within a sliding window, reflecting the short-term fluctuation characteristics of PV output.

[0095] For energy storage charge and discharge status data, the protocol conversion module parses the multiplexed signals within the data domain based on the CAN bus protocol. The battery state of charge (SOC) is obtained by parsing the standard data frame corresponding to the CAN ID. The data format adheres to the scaling factors and offsets defined in the SAE J1939 protocol. The charge and discharge rate is determined by parsing the raw data from the battery pack current sensor and, combined with the timestamp reported by the BMS, calculating the current gradient per unit time to generate dynamic response parameters. During the parsing process, the module adapts to the CAN bus arbitration mechanism to avoid data conflicts and delays in parsing.

[0096] When grid dispatch command data is parsed using the IEC 61850 protocol, the protocol conversion module invokes the MMS (Manufacturing Message Specification) service to access the logical device model. The grid-connected power threshold extracts the Class A data attributes of the MMXU logical node from the DataSet and parses their floating-point values and unit identifiers. Frequency regulation command parameters are obtained by parsing the FrequencySetpoint and DeadBand attributes of the FSCH logical node to obtain the frequency regulation target and allowable deviation range. During the parsing process, the module dynamically loads the communication service mappings described in the SCL (System Configuration Language) file to achieve precise conversion of protocol semantics.

[0097] After parsing, the protocol conversion module formats the data for photovoltaic power generation, voltage fluctuation characteristics, state of charge, charge and discharge rate, grid-connected power threshold, and frequency modulation command parameters. The unified data format uses a JSON-LD structured representation, with each parameter appended with the data source protocol type, timestamp, and data confidence level. Voltage fluctuation characteristics and charge and discharge rate parameters are normalized to per-unit values to eliminate dimensional differences. The formatted data stream is transmitted to the data fusion module via a message queue. The queue utilizes a priority scheduling strategy to ensure that grid dispatch command data is transmitted first, maintaining real-time system response.

[0098] Specifically, the building photovoltaic storage and charging integrated energy management system of the present invention includes:

[0099] a feature extraction unit, configured to perform sliding window interception on the standardized data stream received from the protocol conversion module, and extract the fluctuation feature vectors in the photovoltaic output time series data and the dynamic response vectors in the energy storage charge and discharge state data;

[0100] The coupling analysis unit is used to input the fluctuation characteristic vector and the dynamic response vector into a dynamic weighted model constructed based on the photovoltaic output fluctuation rate and the energy storage response delay to generate the multi-dimensional coupling state matrix.

[0101] After the data fusion module receives the standardized data stream output by the protocol conversion module, the feature extraction unit dynamically sets the sliding window length based on the fluctuation period of the PV output data. The window length is inversely proportional to the PV output sampling frequency. In high-frequency fluctuation scenarios, short windows are used to capture rapidly changing characteristics, while in low-frequency scenarios, long windows are used to enhance the stability of trend analysis. The window sliding step size is set to an integer multiple of the sampling interval to avoid redundant calculations caused by data overlap. For the PV output time series data within the window, the feature extraction unit decomposes the high-frequency detail component and the low-frequency approximate component using a discrete wavelet transform, extracting the fundamental amplitude, harmonic distortion rate, and fluctuation frequency as the core parameters of the fluctuation feature vector.

[0102] Energy storage charge and discharge status data is processed synchronously within the same time window. A dynamic response vector is generated by calculating the instantaneous rate of change of the state of charge (SOC), the rise time of the charge and discharge current, and the temperature gradient. The SOC rate of change is calculated using a differential method to calculate the capacity difference between adjacent timestamps, and the charge and discharge current response time is measured by measuring the delay after the step signal triggers. The feature extraction unit aligns these parameters in the time domain to eliminate timing deviations caused by communication delays, generating a dynamic response vector that strictly matches the timestamp of the fluctuation feature vector.

[0103] The coupling analysis unit inputs the fluctuation eigenvector and dynamic response vector into a dynamic weighted model. The PV output fluctuation rate is calculated as the variance of the fundamental amplitude in the fluctuation eigenvector, reflecting the degree of power instability. The energy storage response delay is assessed as the sliding average of the current rise time in the dynamic response vector, characterizing the system's dynamic regulation capability. The dynamic weighted model assigns weight coefficients based on the inverse ratio of fluctuation rate to response delay, increasing the weight of the PV feature with increasing fluctuation rate and the weight of the energy storage feature with increasing response delay. The weighted eigenvectors are fused using tensor splicing technology to form a coupled feature tensor that incorporates spatiotemporal correlations.

[0104] During the matrix processing phase, the coupling feature tensor is expanded along the time dimension, with each time slice corresponding to a two-dimensional feature matrix. The rows of the matrix represent the characteristic parameter categories of photovoltaic and energy storage, while the columns represent the evolution sequence of the same parameter in different time windows. A multidimensional coupling state matrix is constructed by stacking a three-dimensional structure of consecutive time slices. The row, column, and depth dimensions represent the feature type, time sequence, and device coupling relationship, respectively. This matrix provides input data for the scheduling optimization module, including spatiotemporal evolution patterns and device interaction characteristics, supporting the accurate solution of multi-objective optimization models.

[0105] Specifically, the building photovoltaic storage and charging integrated energy management system of the present invention, the coupling analysis unit also includes:

[0106] The photovoltaic output fluctuation rate is calculated by the variance of the fluctuation characteristic vector;

[0107] The energy storage response delay is calculated by using the time delay parameter in the dynamic response vector;

[0108] The dynamic weighted model dynamically adjusts the weight coefficient according to the ratio of the photovoltaic output fluctuation rate to the energy storage response delay, performs weighted superposition on the fluctuation characteristic vector and the dynamic response vector, and generates a coupling characteristic tensor;

[0109] The coupling characteristic tensor is matrix-processed to obtain the multi-dimensional coupling state matrix.

[0110] After receiving the fluctuation eigenvector and dynamic response vector from the feature extraction unit, the coupled analysis unit calculates the variance of the PV output fluctuation rate using the time series data of the fundamental component of the fluctuation eigenvector. This variance is calculated by measuring the discreteness of the sampled values within a sliding window. The window size matches the analysis period set by the feature extraction unit and reflects the stability of the PV output in the short-term domain. The calculated volatility results are stored as a continuous scalar sequence by timestamp, which is used to quantify the driving force of the PV output's dynamic characteristics on system regulation requirements.

[0111] The energy storage response delay is calculated using the time delay parameter in the dynamic response vector, extracted from the energy storage charge and discharge current step response curve. After the current step is triggered, the system records the time difference required for the current value to reach 90% of the target value from the initial state, which is used as the delay measurement value for a single response. The delay measurement results of multiple consecutive time windows are processed through a moving average filter to generate a delay sequence that reflects the dynamic performance of the energy storage system. The delay sequence is aligned with the photovoltaic fluctuation rate sequence through timestamps to form a correlation mapping of the dynamic characteristics between devices.

[0112] The dynamic weighting model dynamically adjusts the weight coefficient based on the ratio of PV output fluctuation to energy storage response delay. This ratio is calculated as the quotient of the absolute value of the fluctuation and the normalized delay, which is linearly scaled by the maximum allowable response delay threshold. When fluctuation increases or delay decreases, the model increases the weight coefficient of the fluctuation eigenvector, enhancing the dominant role of PV output fluctuations in the coupling analysis. When delay increases or fluctuation decreases, the model increases the weight coefficient of the dynamic response vector, strengthening the compensatory effect of energy storage regulation. The weight coefficient is updated frame by frame within a time window, generating a dynamic weighting parameter table that matches the real-time system status.

[0113] During the weighted superposition process, the fluctuation feature vector and the dynamic response vector are each multiplied by their corresponding weight coefficients to form a weighted feature subspace. The features of each dimension within the subspace are fused along the feature axis using tensor splicing technology to generate a coupled feature tensor that encompasses the PV fluctuation characteristics and the energy storage response characteristics. This tensor structure retains the time series and feature category dimensions of the original vectors while adding a weight dimension to record the dynamic adjustment process, thereby constructing a three-dimensional feature representation space.

[0114] During the matrix processing phase, the three-dimensional coupling feature tensor is sliced and expanded along the time axis. The rows of the two-dimensional matrix corresponding to each time slice represent the feature categories, and the columns represent the weighted eigenvalues. The matrices of consecutive time slices are stacked in chronological order to form a multidimensional coupling state matrix with a three-dimensional structure of rows, columns, and time. The rows of the matrix record the correlation between the characteristic parameters of photovoltaic and energy storage, the columns represent the evolution path of the same parameter in different time windows, and the time dimension reflects the dynamic evolution of the system state. This matrix provides the scheduling optimization module with structured input data that includes device interaction characteristics and time evolution laws, supporting the precise solution of multi-objective optimization models and the generation of real-time strategies.

[0115] Specifically, in the building photovoltaic storage and charging integrated energy management system of the present invention, the scheduling optimization module is further used to:

[0116] generating a charge and discharge power allocation plan based on the photovoltaic output prediction value and the energy storage remaining capacity in the multi-dimensional coupling state matrix received from the data fusion module, wherein the photovoltaic output prediction value is calculated by regression analysis of the fluctuation eigenvector;

[0117] Combined with the grid-connected power threshold in the grid dispatch instruction data received from the data acquisition module, the charging and discharging power allocation scheme is subjected to constrained optimization, and the constrained optimization is based on the multi-objective function to generate the real-time dispatch strategy.

[0118] After receiving the multidimensional coupling state matrix transmitted by the data fusion module, the dispatch optimization module generates a PV output forecast through time series regression analysis of the fluctuation characteristic vector. This regression analysis utilizes an adaptive sliding window mechanism, with the window length matching the PV output fluctuation period. A linear regression model is trained based on historical fluctuation characteristics to predict output trends within the future time window. The remaining energy storage capacity extracts the real-time state of charge (SOC) measurement from the dynamic response vector. The capacity estimation accuracy is corrected by combining the battery aging coefficient and temperature attenuation factor to generate the available energy reserve data for the energy storage system.

[0119] The charging and discharging power allocation plan is dynamically generated based on the predicted PV output and the remaining energy storage capacity. This plan generation process utilizes a hierarchical decision-making mechanism, prioritizing the matching of the predicted PV output with the charging load demand. The remaining power is allocated to energy storage charging or grid-connected feed. The upper limit of energy storage charging power is calculated based on the product of remaining capacity and the maximum charging rate, while the lower limit of discharge power is dynamically set based on the grid's frequency regulation requirements and load shortfall. During the allocation process, the module establishes a power balance equation for PV output, energy storage charging and discharging, and grid interaction power as a basic constraint.

[0120] The grid-connected power thresholds in the grid dispatch instruction data are embedded as hard constraints in the optimization model. These thresholds are obtained by parsing the ActivePowerLimit attribute in the grid dispatch instruction. The module then sets upper and lower grid-connected power limits in the charge and discharge power allocation scheme. During the constrained optimization phase, the power balance equation, grid-connected threshold limits, and equipment operating boundaries are integrated into a set of linear inequality constraints to construct a multi-objective function that includes PV absorption rate, energy storage cycle losses, and grid frequency deviation.

[0121] In the multi-objective function, the PV absorption rate target is defined by maximizing the matching between PV output and local loads and energy storage charging; the energy storage cycle loss target is achieved by minimizing the weighted sum of charge and discharge depth and current change rate; and the grid frequency deviation target is quantified by minimizing the absolute value of the frequency deviation between actual output and frequency regulation instructions. The improved particle swarm optimization algorithm's inertia weight is dynamically adjusted based on the PV output volatility. When volatility is high, the global search weight is reduced to enhance local optimization efficiency and prevent the algorithm from falling into suboptimal solutions.

[0122] During the solution process, an improved particle swarm optimization algorithm generates a Pareto-optimal solution set, each corresponding to a power allocation strategy. The module assigns a weighted score to the strategies in the solution set based on the priority weights embedded in the grid dispatch instructions. Priority weights are calculated by analyzing the economic dispatch coefficient and frequency regulation urgency parameter in the grid instructions. The strategy with the highest score is output as the real-time dispatch strategy. The strategy includes the PV inverter power setpoint, energy storage charging and discharging power instructions, and grid-connected power control parameters, forming an executable set of control instructions.

[0123] Before the real-time scheduling strategy is converted into device-level control signals by the instruction generation module, the module performs a secondary verification process. This process compares the power allocation value in the strategy with the real-time device status data. If insufficient remaining energy storage capacity is detected or the predicted PV output deviates from the actual sampled value by more than a threshold, the strategy backtracking mechanism is triggered, re-calling the historical coupling state matrix to iteratively generate a revised scheduling strategy, maintaining the robustness and adaptability of system operation.

[0124] Specifically, the building photovoltaic storage and charging integrated energy management system of the present invention also includes:

[0125] Based on the photovoltaic output prediction value in the multi-dimensional coupling state matrix received from the data fusion module and the weight coefficient output from the dynamic weighting model, a multi-objective function is constructed with the goal of maximizing the photovoltaic absorption rate, minimizing the energy storage cycle loss, and minimizing the grid frequency deviation;

[0126] An improved particle swarm algorithm is used to solve the multi-objective function. The inertia weight of the improved particle swarm algorithm is dynamically adjusted according to the photovoltaic output fluctuation rate to generate a Pareto optimal solution set. The inertia weight of the improved particle swarm algorithm is dynamically adjusted with the photovoltaic output fluctuation rate. Specifically, when the photovoltaic output fluctuation rate increases by 5%, the inertia weight linearly decreases by 0.1; when the fluctuation rate decreases, the inertia weight gradually returns to the initial value of 0.9.

[0127] The real-time scheduling strategy is selected from the Pareto optimal solution set according to the priority weight in the power grid scheduling instruction data received from the data acquisition module.

[0128] The dispatch optimization module extracts the weight coefficients between the PV output forecast and the output of the dynamic weighting model based on the multidimensional coupling state matrix provided by the data fusion module. The PV output forecast is generated by extrapolating the time series of the fluctuation eigenvectors. Exponential smoothing is used to correct for forecast deviations and improve short-term forecast accuracy. The weight coefficients are derived from the dynamic weighting model, recording the real-time coupling relationship between PV fluctuations and energy storage response delays. These weight coefficients serve as key adjustment parameters for the objective terms in the multi-objective function.

[0129] In constructing the multi-objective function, the PV absorption rate target is defined by maximizing the matching degree between PV output, energy storage charging power, and local load demand. This matching degree is calculated using the cosine similarity algorithm, which measures the phase and amplitude consistency between the output curve and the demand curve. The energy storage cycle loss target is achieved by minimizing the product of the number of charge and discharge cycles and the depth of a single cycle. The cycle depth is calculated by the amplitude of the state of charge change and the loss coefficient is corrected by combining the battery temperature attenuation factor. The grid frequency deviation target is quantified by minimizing the mean square error between the actual system output frequency and the dispatch command frequency. Frequency deviation data is acquired in real time from the grid-side synchronous acquisition device.

[0130] The improved particle swarm optimization algorithm's inertia weight adjustment mechanism is dynamically linked to the PV output volatility. The volatility is calculated using the variance of the PV characteristic parameters in the multidimensional coupled state matrix. As the volatility increases, the inertia weight decreases linearly, limiting the global search range of the particle swarm to accelerate convergence. As the volatility decreases, the inertia weight gradually recovers, enhancing the algorithm's global optimization capabilities under stable conditions. A non-dominated sorting strategy is implemented during the algorithm's initialization phase to generate an initial Pareto front solution set. During iterations, the crowding distance is calculated to maintain the diversity of the solution set's distribution.

[0131] The Pareto-optimal solution set is selected based on the priority weights embedded in the grid dispatch instructions. These priority weights are calculated by analyzing the economic dispatch coefficient, frequency response level, and market price fluctuation parameters contained in the grid instructions to generate a weighted score table for each optimization item in the objective function. During the solution set selection phase, each Pareto solution is scored using a multi-objective weighted approach, and the solution with the highest score is output as the real-time dispatch strategy. This strategy includes the power regulation instructions for the PV inverter, the charge and discharge power profiles for the energy storage system, and the grid-connected power control threshold, forming a complete set of execution parameters.

[0132] After the real-time scheduling strategy is generated, the module performs a robustness check on the strategy. This check compares the predicted charge and discharge power in the strategy with the real-time state of charge of the energy storage unit. If the remaining capacity is insufficient to support the strategy execution, the prediction model backtracking mechanism is triggered, regenerating the charge and discharge power allocation plan based on the historical coupling state matrix. Before the final strategy is distributed through the instruction generation module, the module synchronously updates the weight coefficients in the multi-dimensional coupling state matrix, forming a closed-loop optimization link to maintain the system's scheduling stability under dynamic changes in multi-source data.

[0133] Specifically, the building photovoltaic storage and charging integrated energy management system of the present invention includes:

[0134] an instruction encoding unit, configured to convert the charge and discharge power values in the real-time scheduling strategy output by the scheduling optimization module into PWM modulation signals based on the Pareto optimal solution set generated by the scheduling optimization module;

[0135] The instruction distribution unit is used to synchronously send the PWM modulation signal to the BMS system in the energy storage unit and the charging pile controller in the charging facility unit through the OPC UA protocol.

[0136] The dispatch optimization module extracts the weight coefficients between the PV output forecast and the output of the dynamic weighting model based on the multidimensional coupling state matrix provided by the data fusion module. The PV output forecast is generated by extrapolating the time series of the fluctuation eigenvectors. Exponential smoothing is used to correct for forecast deviations and improve short-term forecast accuracy. The weight coefficients are derived from the dynamic weighting model, recording the real-time coupling relationship between PV fluctuations and energy storage response delays. These weight coefficients serve as key adjustment parameters for the objective terms in the multi-objective function.

[0137] In constructing the multi-objective function, the PV absorption rate target is defined by maximizing the matching degree between PV output, energy storage charging power, and local load demand. This matching degree is calculated using the cosine similarity algorithm, which measures the phase and amplitude consistency between the output curve and the demand curve. The energy storage cycle loss target is achieved by minimizing the product of the number of charge and discharge cycles and the depth of a single cycle. The cycle depth is calculated by the amplitude of the state of charge change and the loss coefficient is corrected by combining the battery temperature attenuation factor. The grid frequency deviation target is quantified by minimizing the mean square error between the actual system output frequency and the dispatch command frequency. Frequency deviation data is acquired in real time from the grid-side synchronous acquisition device.

[0138] The improved particle swarm optimization algorithm's inertia weight adjustment mechanism is dynamically linked to the PV output volatility. The volatility is calculated using the variance of the PV characteristic parameters in the multidimensional coupled state matrix. As the volatility increases, the inertia weight decreases linearly, limiting the global search range of the particle swarm to accelerate convergence. As the volatility decreases, the inertia weight gradually recovers, enhancing the algorithm's global optimization capabilities under stable conditions. A non-dominated sorting strategy is implemented during the algorithm's initialization phase to generate an initial Pareto front solution set. During iterations, the crowding distance is calculated to maintain the diversity of the solution set's distribution.

[0139] The Pareto-optimal solution set is selected based on the priority weights embedded in the grid dispatch instructions. These priority weights are calculated by analyzing the economic dispatch coefficient, frequency response level, and market price fluctuation parameters contained in the grid instructions to generate a weighted score table for each optimization item in the objective function. During the solution set selection phase, each Pareto solution is scored using a multi-objective weighted approach, and the solution with the highest score is output as the real-time dispatch strategy. This strategy includes the power regulation instructions for the PV inverter, the charge and discharge power profiles for the energy storage system, and the grid-connected power control threshold, forming a complete set of execution parameters.

[0140] After the real-time scheduling strategy is generated, the module performs a robustness check on the strategy. This check compares the predicted charge and discharge power in the strategy with the real-time state of charge of the energy storage unit. If the remaining capacity is insufficient to support the strategy execution, the prediction model backtracking mechanism is triggered, regenerating the charge and discharge power allocation plan based on the historical coupling state matrix. Before the final strategy is distributed through the instruction generation module, the module synchronously updates the weight coefficients in the multi-dimensional coupling state matrix, forming a closed-loop optimization link to maintain the system's scheduling stability under dynamic changes in multi-source data.

[0141] Specifically, in the building integrated solar-storage-charging energy management system of the present invention, the feedback module is further used to:

[0142] Collecting actual output data of the photovoltaic power generation unit, real-time state of charge data of the energy storage unit, and power consumption data of the charging facility unit;

[0143] Calculating a power deviation rate based on a deviation between the actual output data and a predicted value in the real-time scheduling strategy generated by the scheduling optimization module;

[0144] generating a feedback correction coefficient based on the power deviation rate and the energy storage response delay obtained from the data fusion module;

[0145] The feedback correction coefficient is input into the scheduling optimization module to trigger the parameter update of the multi-objective function.

[0146] The feedback module periodically acquires actual output data from photovoltaic power generation units through a distributed data acquisition interface. The data acquisition frequency aligns with the scheduling optimization module's prediction period. Actual output data, including the instantaneous values of active power, reactive power, and grid voltage output by the inverter, is matched with the predicted values in the real-time scheduling strategy through a timestamp alignment mechanism. Real-time state-of-charge (SOC) data for energy storage units is extracted from the CAN bus communication frames of the battery management system (BMS) and combined with battery temperature and state-of-health (SOH) parameters to correct capacity estimation errors. Power consumption data for charging facility units is generated by fusing the load curve reported by the charging pile controller with user reservation information, reflecting the difference between real-time charging demand and planned load.

[0147] The power deviation rate is calculated using a root mean square error (RMSE) algorithm within a sliding window. The window length is the same as the lookback period of the PV output forecast model. The algorithm calculates the point-by-point difference between the actual output data and the predicted value, accumulates the squares, takes the average, and then takes the square root to generate a scalar indicator reflecting the overall deviation level. A sliding average filter is introduced during the deviation calculation process to filter out transient abnormal fluctuations caused by sudden changes in illumination or equipment noise, while preserving the trend deviation characteristics. The filter window length is dynamically adjusted based on the sampling interval of the PV output data. In high-frequency sampling scenarios, a short window is used to reduce calculation latency.

[0148] The feedback correction coefficient is generated based on the synergistic effect of the power deviation rate and the energy storage response delay. The energy storage response delay is extracted from the dynamic response vector of the data fusion module and represents the system's efficiency in compensating for historical deviations. The correction coefficient is calculated by multiplying the deviation rate and the delay. As the deviation rate or delay increases, the coefficient value increases accordingly, enhancing the adjustment range of the optimization parameters. A dynamic attenuation factor is introduced into the calculation process, and its value increases with the cumulative number of consecutive deviation cycles to avoid overcorrection caused by short-term disturbances.

[0149] After the feedback correction coefficient is input into the dispatch optimization module, the parameter update process of the multi-objective function is triggered. The parameter update focuses on the weight allocation of the photovoltaic absorption rate, energy storage cycle loss, and grid frequency deviation in the objective function. The correction coefficient is converted into a weight adjustment through normalization. The weight adjustment direction is dynamically set according to the deviation type: if the power deviation is mainly caused by photovoltaic output fluctuations, the weight of the photovoltaic absorption rate target is increased; if the deviation is caused by energy storage response lag, the optimization weight of the energy storage cycle loss target is increased. The updated multi-objective function drives the improved particle swarm algorithm to regenerate the Pareto solution set and iteratively optimize the real-time dispatch strategy.

[0150] The closed-loop control mechanism is implemented through the interaction between the feedback module and the scheduling optimization module. After each parameter update, the module records the mapping between historical correction coefficients and the execution results of the scheduling strategy, forming an experience database. During subsequent corrections, the module prioritizes optimized parameter combinations from similar historical scenarios, accelerating convergence and improving the system's adaptability. If the deviation rate does not drop below the threshold after multiple consecutive corrections, the module triggers the abnormal warning process, notifying the data fusion module to recalibrate the feature extraction model and optimize the accuracy of system state representation at the source of the data.

[0151] Specifically, the building photovoltaic storage and charging integrated energy management system of the present invention also includes:

[0152] Performing a sliding average filter on the deviation between the actual output data and the predicted value, wherein the filter window length is set according to the sampling period of the photovoltaic output time series data;

[0153] Calculating a power compensation amount based on the filtered deviation and the rate of change of the energy storage charge state extracted from the dynamic response vector of the data fusion module;

[0154] The feedback correction coefficient is generated based on the power compensation amount.

[0155] The feedback module applies a sliding average filter to the deviation between actual output data and predicted values. The filter window length is dynamically adjusted based on the sampling period of the PV output time series data: when the sampling period is short, a smaller window length is used to reduce processing latency and accommodate high-frequency data updates; when the sampling period is long, the window length is increased to enhance the filtering effect and suppress periodic noise interference. The filtering algorithm uses an exponentially weighted moving average (EWMA), assigning a higher weight to recent deviation data, preserving trend-based deviation characteristics while smoothing out instantaneous fluctuations.

[0156] The filtered deviation is combined with the energy storage state of charge (SOC) rate of change to calculate the power compensation. The SOC rate of change is extracted from the dynamic response vector of the data fusion module and generated by differential calculation of SOC values within adjacent time windows, reflecting the real-time charging and discharging dynamics of the energy storage system. A dynamic coupling factor is introduced into the power compensation calculation, which is generated by multiplying the SOC rate of change by the filtered deviation. As the SOC rate of change increases, the coupling factor increases the weight of the compensation calculation, strengthening the energy storage regulation's ability to compensate for the deviation. As the SOC rate of change decreases, the weight decreases to avoid overcompensation.

[0157] The feedback correction factor is generated based on the power compensation. The compensation amount is normalized to a per-unit value in the 0-1 range. This per-unit value is multiplied by a preset baseline correction factor to generate a dynamically adjusted feedback correction factor. The baseline correction factor is set based on empirical evidence of deviation compensation effectiveness from historical operating data and is regularly updated using an offline training model. If the power compensation exceeds a preset threshold for multiple consecutive cycles, the module triggers an adaptive adjustment mechanism, dynamically increasing the baseline correction factor based on the real-time operating scenario, enhancing the system's ability to adapt to persistent deviations.

[0158] Once the correction coefficient is input into the dispatch optimization module, it triggers the dynamic relaxation or tightening of constraints within the multi-objective function. For example, as the correction coefficient increases, the module relaxes the constraints on the PV utilization rate target, allowing for higher fluctuations in energy storage charging and discharging power to compensate for output deviations. As the correction coefficient decreases, the constraint threshold for grid frequency regulation deviation is tightened, prioritizing grid interaction stability. The correction process associates the correction coefficient with the optimization function parameters via a parameter mapping table. This mapping table is dynamically updated based on the device's operating status and the priority of grid dispatch instructions, forming a closed-loop logical chain for parameter adjustment.

[0159] The exception handling mechanism is embedded in the feedback correction process. If the filtered deviation continues to exceed the device's rated capacity, the module sends a recalibration request to the data fusion module, triggering a reset of the model parameters in the feature extraction unit. Furthermore, an attenuation factor is introduced into the correction coefficient generation process. The factor value increases as the anomaly persists. This suppresses the cumulative effect of corrections caused by device failures or communication anomalies, maintaining system operational safety.

[0160] Specifically, the building photovoltaic storage and charging integrated energy management system of the present invention further includes: a safety verification module for performing overload protection verification and protocol compliance verification on the control instructions distributed by the instruction generation module;

[0161] When the safety verification module detects that the power value in the control instruction exceeds the rated capacity of the equipment acquired by the data acquisition module, it triggers the scheduling optimization module to regenerate the limiting scheduling strategy based on the multi-objective function.

[0162] After receiving the control command from the command generation module, the safety verification module begins the overload protection verification process by obtaining the rated capacity parameters of the energy storage unit, photovoltaic inverter, and charging pile from the data acquisition module. These rated capacity parameters include the maximum charge and discharge power of the battery pack, the upper limit of the inverter's output power, and the charging pile's load capacity threshold. Dynamic constraint boundaries are generated by integrating device nameplate information with real-time status data. The verification unit compares the power allocation value in the control command with the dynamic constraint boundaries of the corresponding device, identifying any exceeding-limit commands and marking the offending device address.

[0163] Protocol compliance verification validates the communication format and data semantics of control instructions. The verification unit parses the OPC UA node attributes embedded in the instruction, verifying that the node namespace is consistent with the target device's communication interface definition and that the data fields conform to the attribute specifications of the logical device model in the IEC 61850 protocol. For instructions using PWM modulated signals, the verification unit verifies that the signal frequency and duty cycle range are within the device driver's allowable operating range, preventing instruction execution failures due to protocol compatibility issues.

[0164] When a power value is detected that exceeds the rated capacity of a device, the safety verification module sends an interrupt signal to the scheduling optimization module, annotating the identifier of the device exceeding the limit and the offending power value. This interrupt signal triggers the scheduling optimization module's multi-objective function reconstruction process, adding device power limit constraints to the function. The constraint values are dynamically set based on the real-time status of the exceeding device. For example, when an energy storage unit is operating under high temperature conditions, the upper limit of charge and discharge power is dynamically adjusted downward based on the temperature attenuation coefficient, generating a limit scheduling strategy that meets the device's safety boundaries.

[0165] During the generation of the power-limiting scheduling strategy, the multi-objective function retains the existing targets for PV utilization, energy storage loss, and frequency regulation deviation, while adding a new device power limit target. The limit targets are implemented through a penalty function mechanism. Exceeding power limits are assigned penalty weights based on the degree of deviation, driving the improved particle swarm optimization algorithm to prioritize solutions that meet device safety constraints. After the algorithm solves, the module selects the strategy with the lowest penalty weight from the Pareto optimal solution set as the output power-limiting scheduling strategy.

[0166] The exception handling mechanism incorporates redundancy into the safety verification module. If the same device triggers overload protection multiple times in a row, the module initiates a device health assessment process, incorporating battery aging parameters, inverter operation logs, and charging station fault records from the BMS system to calculate the device derating factor. The derating factor is then updated to a multidimensional coupling state matrix via the data fusion module, driving subsequent scheduling strategies to automatically adapt to device performance degradation and avoid repeated overloads. Verification results and processing logs are synchronized to the monitoring platform via the event bus, enabling operations and maintenance personnel to trace and analyze abnormal scenarios.

[0167] The specific explanations of the various technical features in the technical solution of the present invention are as follows:

[0168] Multi-source heterogeneous data refers to photovoltaic output time-series data (such as voltage, current, and power) from photovoltaic power generation units, charge and discharge status data (such as state of charge, temperature, and charge and discharge rate) from energy storage units, load demand data (such as charging power and user reservation information) from charging facilities, and grid dispatch instruction data (such as grid-connected power thresholds and frequency modulation ranges). Due to the different source devices, this data varies in transmission protocols, sampling frequencies, and data formats, necessitating standardization through protocol conversion.

[0169] Modbus protocol parsing: An industrial communications protocol used to parse data frames output by photovoltaic inverters, extracting active power, reactive power, and instantaneous voltage values mapped to register addresses, and verifying data integrity through CRC. Parsing generates standardized fields containing timestamps and data tags, eliminating communication delays caused by protocol differences.

[0170] CAN bus protocol parsing: Targeting the communication protocol of the energy storage unit's battery management system (BMS), this system extracts the battery pack's state of charge (SOC), charge and discharge current rates, and state of health (SOH) by parsing the identifier (CAN ID) in the CAN data frame. The data format complies with the SAE J1939 standard, and the bus arbitration mechanism is adapted during parsing to avoid data conflicts.

[0171] Sliding Window Capture: This technology, used in the data fusion module, dynamically captures time series data. The window length is adaptively adjusted based on the PV output fluctuation period. For example, in high-frequency fluctuation scenarios, a short window (e.g., 30 seconds) is used to capture rapidly changing features, while in low-frequency scenarios, a long window (e.g., 5 minutes) is used to analyze trend changes. The window sliding step size is aligned with the data sampling interval to reduce redundant calculations.

[0172] Dynamic weighting model: Dynamically assigns weight coefficients based on the ratio of PV output fluctuation (calculated by the variance of the fluctuation eigenvector) to the energy storage response delay (calculated by the charge and discharge current step response time). When PV fluctuation increases, the weight of PV characteristics is increased to strengthen output fluctuation compensation; when energy storage response delay increases, the weight of energy storage characteristics is increased to optimize regulation efficiency.

[0173] Pareto optimal solution set: A set of non-dominated solutions generated by an improved particle swarm optimization algorithm in the dispatch optimization module. Each solution corresponds to a power allocation strategy that balances the objectives of PV utilization, energy storage loss, and frequency regulation deviation. When selecting solutions from this set, a weighted score is applied based on the priority of grid dispatch instructions (such as economic coefficient and frequency regulation urgency), and the optimal overall solution is selected as the real-time dispatch strategy.

[0174] OPC UA: An industrial communication protocol that encapsulates PWM modulation signals used in real-time scheduling strategies into standardized information model nodes and synchronously distributes them to energy storage BMSs, photovoltaic inverters, and charging station controllers via a publish-subscribe model. The protocol supports Time-Sensitive Networking (TSN) extensions, uses time slot allocation strategies to avoid communication conflicts, and keeps command transmission latency to milliseconds.

[0175] Feedback correction coefficient: This coefficient is generated by the feedback module based on the deviation between actual output data and the predicted value (noise is eliminated through a sliding average filter) and the energy storage response delay. This coefficient is converted through normalization to adjust the target weight in the optimization function. For example, if the deviation is primarily caused by PV fluctuations, the weight of the target consumption rate is increased, driving the iterative optimization algorithm.

[0176] Limit Scheduling Strategy: This strategy reconfiguration process is triggered when the safety verification module detects that the power command exceeds the rated capacity of the equipment. By introducing a penalty function into the multi-objective function, a penalty weight is assigned to the power value exceeding the limit. This drives an improved particle swarm algorithm to generate a power allocation plan that meets the equipment's safety boundaries. For example, it can dynamically lower the upper limit of energy storage charging and discharging power under high-temperature conditions.

[0177] Multidimensional coupling state matrix: This is a three-dimensional data structure output by the data fusion module. The rows represent characteristic parameters of PV and energy storage (such as fluctuation amplitude and response time), the columns record time series evolution, and the depth reflects device interactions. This matrix provides temporally and spatially correlated inputs for scheduling optimization, supporting the precise solution of multi-objective models.

[0178] The above technical features eliminate data heterogeneity through unified protocols, dynamically integrate and quantify device coupling relationships, and achieve resource coordination through closed-loop optimization. Ultimately, they solve the processing delays and scheduling mismatches caused by insufficient data fusion in existing systems, and improve the operating efficiency and economy of the photovoltaic storage and charging system.

[0179] Dynamic Weighted Model: Deployed in the data fusion module, it quantifies the dynamic coupling relationship between PV output and energy storage response. This model uses PV output fluctuation (calculated using the variance of the fluctuation eigenvector) and energy storage response delay (calculated based on the charge and discharge current step response time) as input parameters and dynamically assigns weight coefficients based on the ratio of fluctuation to delay. When PV fluctuation is high, the model increases the weight of PV characteristics to prioritize output fluctuation suppression. When energy storage response delay is long, the model increases the weight of energy storage characteristics to enhance regulation compensation. The output is a weighted fused coupling feature tensor, providing temporally and spatially correlated input data for scheduling optimization.

[0180] Improved Particle Swarm Optimization (PSO): Applied to the dispatch optimization module, it solves multi-objective optimization problems. This algorithm incorporates a dynamic inertia weight adjustment mechanism within the standard PSO framework: the weight changes with the PV output volatility. For every 5% increase in volatility, the inertia weight decreases linearly by 0.1. This narrows the search range and accelerates convergence in high-volatility scenarios. The algorithm generates a Pareto-optimal solution set through non-dominated sorting and then combines grid priority weights (such as frequency regulation urgency and economic dispatch coefficient) to select the optimal solution, achieving a balance between PV absorption rate, energy storage losses, and frequency regulation deviation.

[0181] Sliding Window Extraction Model: This model is used in the feature extraction phase of the data fusion module. It dynamically adjusts the window length based on the fluctuation period of the PV output time series data. For example, a short 30-second window is used to capture rapidly changing features in high-frequency fluctuations (period < 10 seconds), while a long 5-minute window is used to analyze trend changes in low-frequency scenarios (period > 60 seconds). Wavelet decomposition is used to extract fluctuation feature vectors, such as the fundamental amplitude and harmonic distortion rate, from the data within the window. These are then time-aligned with dynamic response vectors, such as the energy storage system's SOC change rate and response time, to eliminate timing deviations caused by communication delays.

[0182] Feedback Correction Model: Deployed in the feedback module, this model generates optimization parameter adjustments. This model uses a sliding average filter to process deviations between actual output and predicted values, filtering out transient noise (such as sudden changes in illumination) while preserving the trend-based deviation characteristics. A correction coefficient is calculated based on the energy storage response delay (extracted from the dynamic response vector). For example, for every 200 millisecond increase in delay, the correction coefficient increases by 0.2 to enhance compensation for energy storage regulation lag. This correction coefficient adjusts the weight distribution in the multi-objective function through normalized mapping, driving iterative optimization of the scheduling strategy.

[0183] Protocol parsing model: Includes Modbus, CAN bus, and IEC 61850 protocol parsers. The Modbus parser locates the register addresses of the PV inverter and extracts active power, reactive power, and voltage fluctuation characteristics. The CAN bus parser parses the energy storage BMS data frames based on the SAE J1939 standard to extract SOC, charge and discharge rates, and temperature parameters. The IEC 61850 parser accesses the logical device model through the MMS service to extract grid-connected power thresholds and frequency regulation command parameters. The parsed data is formatted using JSON-LD, with timestamps and protocol tags appended, forming a standardized data stream.

[0184] The limiting strategy generation model embeds a safety verification module for device overload protection. When a power command exceeds the device's rated capacity (e.g., the maximum charge / discharge power of the energy storage system), the model introduces a penalty function into the multi-objective function. For example, for every 10% excess power, the penalty weight increases by 0.5, forcing the improved particle swarm algorithm to generate a solution set that meets safety boundaries. Furthermore, the limiting threshold is dynamically adjusted based on the device's real-time status (e.g., battery temperature). For example, for every 5°C increase in temperature, the energy storage power limit decreases by 8% to mitigate the risk of thermal runaway.

[0185] The PWM modulation model, located in the instruction generation module, converts the power values in the scheduling policy into modulation signals executable by the device. The model normalizes the power values based on the device's rated capacity. For example, a maximum energy storage power of 2MW corresponds to a 100% duty cycle. This generates a PWM waveform compatible with inverters, BMSs, and charging station controllers. The signal is encapsulated as an information model node using the OPC UA protocol, defining the amplitude, frequency, and effective timestamp attributes. This supports simultaneous subscription and real-time control of multiple devices.

[0186] The above model eliminates data heterogeneity through unified protocols, dynamically integrates and quantifies the interaction status of devices, and realizes dynamic resource matching through closed-loop optimization. It ultimately solves the data collaboration problem in photovoltaic, energy storage and power grid links, and improves the overall efficiency and economy of the system.

[0187] The present invention solves the problems of data processing delay and resource coordination mismatch in existing systems through a protocol unification and dynamic fusion mechanism for multi-source heterogeneous data. The protocol conversion module performs Modbus, CAN bus and IEC 61850 protocol parsing on photovoltaic output timing data, energy storage charge and discharge status data and grid dispatch instructions, respectively, to generate a standardized data stream in a unified data format. During the protocol parsing process, the module extracts key parameters such as photovoltaic power generation power, voltage fluctuation characteristics, state of charge, charge and discharge rate, and eliminates protocol differences and sampling cycle inconsistencies through timestamp synchronization and data formatting, providing highly consistent input for subsequent data fusion and reducing transmission and processing delays caused by data heterogeneity.

[0188] The data fusion module, based on standardized data streams, employs sliding window extraction and a dynamic weighting model to implement spatiotemporal coupling analysis of multi-source data. The feature extraction unit dynamically extracts the PV output fluctuation characteristic vector and the energy storage dynamic response vector using a sliding window. The coupling analysis unit dynamically adjusts the weight coefficients based on the ratio of PV output fluctuation rate to energy storage response delay, generating a multidimensional coupling state matrix. This matrix quantifies the relationship between PV output fluctuation characteristics and energy storage regulation capabilities, providing input data for scheduling optimization that incorporates spatiotemporal evolution patterns and device interaction characteristics, addressing resource coordination mismatch issues caused by insufficient multi-source data fusion.

[0189] The closed-loop control chain achieves dynamic resource optimization through the collaboration of the dispatch optimization module and the feedback module. The dispatch optimization module constructs an optimization function with multiple objectives, including PV absorption rate, energy storage cycle loss, and grid frequency deviation. It uses an improved particle swarm algorithm with dynamic inertia weight adjustment to generate a Pareto optimal solution set, and selects a real-time dispatch strategy based on grid priority weights. The feedback module calculates the feedback correction coefficient based on the deviation between actual output data and predicted values, iteratively updating the optimization function parameters. Simultaneously, the safety verification module constrains equipment operating boundaries and triggers a limiting strategy, forming a closed-loop control process from data acquisition, fusion, optimization, to execution, improving the operating efficiency and economic efficiency of the photovoltaic storage and charging system.

[0190] In its implementation, the present invention addresses the challenge of coordinating heterogeneous multi-source data across photovoltaic power generation, energy storage, and grid connection in building photovoltaic-storage-charging scenarios, achieving efficient system operation through protocol unification, dynamic fusion, and closed-loop optimization mechanisms. The data acquisition module uses distributed sensors to obtain real-time information on the output power of photovoltaic modules, battery pack charge and discharge currents, and charging pile load requirements, and receives grid connection thresholds and frequency modulation instructions from the grid dispatching system. The sampling periods of each data source are aligned through a timestamp synchronization mechanism, eliminating time misalignment between photovoltaic output time series data and energy storage status data, providing time-consistent multi-source input for subsequent processing.

[0191] The protocol conversion module uses Modbus protocol to parse photovoltaic data, extracting inverter output active power, reactive power, and voltage fluctuation amplitude. Energy storage data is analyzed using the CAN bus protocol to analyze battery state of charge, charge and discharge rates, and temperature parameters. Grid dispatch instructions are analyzed based on the IEC 61850 protocol to identify grid-connected power limits and frequency ranges. The parsed data is normalized and converted into a standardized data stream in JSON-LD format, with a protocol type tag and timestamp appended. The data is then transmitted to the data fusion module via a priority message queue. This module utilizes a dynamic sliding window capture technique, with the window length adaptively adjusted based on the frequency of photovoltaic output fluctuations. When the PV output fluctuation period is less than 10 seconds, the window length is set to 30 seconds to capture rapid fluctuations; when the fluctuation period exceeds 60 seconds, the window length is extended to 5 minutes to analyze trend changes. The feature extraction unit performs wavelet decomposition on the photovoltaic data within the window, extracting the fundamental amplitude, harmonic distortion, and fluctuation frequency to construct a fluctuation feature vector. Simultaneously, the instantaneous rate of change of the energy storage state of charge and the charge and discharge current response time are calculated to form a dynamic response vector. The coupling analysis unit dynamically assigns weights based on the ratio of volatility variance to response delay. When PV output volatility exceeds 15%, the weights tilt toward PV characteristics to enhance fluctuation compensation. When energy storage response delay exceeds 500 milliseconds, the weights shift toward energy storage characteristics to optimize regulation efficiency. The weighted fusion feature tensor is expanded along the time dimension into a multidimensional coupling state matrix. The rows of the matrix characterize the correlation between the characteristic parameters of PV and energy storage, while the columns record the evolution of the time series, providing temporally and spatially correlated input data for scheduling optimization.

[0192] The dispatch optimization module constructs a multi-objective function based on the predicted PV output and remaining energy storage capacity in the multidimensional coupled state matrix, encompassing PV absorption rate, cycle losses, and frequency deviation. The PV absorption rate objective is achieved by maximizing the cosine similarity between the output curve and the local load; the energy storage loss objective is defined by minimizing the product of the charge / discharge depth and the temperature attenuation coefficient; and the frequency deviation objective is achieved by constraining the root mean square error (RMS) between the actual output frequency and the grid command. The improved particle swarm optimization algorithm dynamically adjusts its inertia weight based on PV volatility: for every 5% increase in volatility, the inertia weight decreases linearly by 0.1, limiting the global search range to accelerate convergence. The resulting Pareto optimal solution set is weighted by the grid's economic dispatch coefficient and frequency regulation urgency, and the optimal solution is selected as the real-time dispatch strategy. The command generation module converts the power values in the strategy into PWM signals with adjustable duty cycles, which are then simultaneously transmitted to the PV inverter, energy storage BMS, and charging pile controller via the OPC UA protocol. Command transmission latency is kept within 50 milliseconds to maintain real-time control. The feedback module periodically collects the deviation between the actual output and the strategy prediction value, generates a correction coefficient after sliding average filtering and combines it with the energy storage response delay, and dynamically adjusts the target weight in the optimization function; the safety verification module verifies the matching degree between the power instruction and the rated capacity of the equipment in real time. When an over-limit is detected, the limiting strategy reconstruction process is triggered, and the equipment operation boundary is constrained through the penalty function mechanism, forming a complete technical chain from data collection, fusion optimization to closed-loop control, ultimately improving the resource coordination efficiency and economy of the solar storage and charging system.

Claims

1. A building solar energy storage and charging integrated energy management system, characterized in that: include: A data acquisition module is used to acquire multi-source heterogeneous data, including photovoltaic output time series data, energy storage charge and discharge status data, charging load demand data, and grid dispatch instruction data; A protocol conversion module is used to perform protocol parsing on the multi-source heterogeneous data, parsing photovoltaic output time series data through the Modbus protocol, energy storage charge and discharge status data through the CAN bus protocol, and grid dispatch instruction data through the IEC61850 protocol, to generate a standardized data stream; a data fusion module for performing a sliding window interception on the standardized data stream, extracting the fluctuation characteristic vectors in the photovoltaic output time series data and the dynamic response vectors in the energy storage charge and discharge state data, inputting the fluctuation characteristic vectors and the dynamic response vectors into a dynamic weighted model constructed based on the photovoltaic output fluctuation rate and the energy storage response delay, performing a time-domain convolution operation, and generating a multi-dimensional coupling state matrix; a dispatch optimization module for constructing a multi-objective function with photovoltaic absorption rate, energy storage cycle loss, and grid frequency deviation as optimization targets based on the multi-dimensional coupling state matrix and the grid dispatch instruction data received from the data acquisition module, and using an improved particle swarm algorithm with inertia weight dynamically adjusted according to photovoltaic output fluctuation rate to solve and generate a real-time dispatch strategy; The instruction generation module is used to convert the charge and discharge power values in the real-time scheduling strategy into PWM modulation signals, and distribute the PWM modulation signals to the inverter in the photovoltaic power generation unit, the BMS system in the energy storage unit, and the charging pile controller in the charging facility unit through the OPCUA protocol; A feedback module is used to monitor the actual output data of the photovoltaic power generation unit, the real-time charge status data of the energy storage unit, and the power consumption data of the charging facility unit, calculate the deviation between the actual output data and the predicted value in the real-time scheduling strategy, generate a feedback correction coefficient, and input the feedback correction coefficient into the scheduling optimization module to trigger the iterative optimization of the multi-objective function and adjust the real-time scheduling strategy.

2. The building solar energy storage and charging integrated energy management system according to claim 1 is characterized in that: The protocol conversion module is also used for: Performing Modbus protocol parsing on the photovoltaic output time series data received from the data acquisition module to extract photovoltaic power generation power and voltage fluctuation characteristics; Performing CAN bus protocol analysis on the energy storage charge and discharge status data received from the data acquisition module to extract the state of charge and charge and discharge rate; Performing IEC 61850 protocol parsing on the power grid dispatch instruction data received from the data acquisition module to extract the grid-connected power threshold and frequency modulation instruction parameters; The photovoltaic power generation power, voltage fluctuation characteristics, state of charge, charge and discharge rate, grid-connected power threshold and frequency modulation instruction parameters are output as analysis results to the data fusion module.

3. The building solar energy storage and charging integrated energy management system according to claim 1 is characterized in that: The data fusion module includes: a feature extraction unit, configured to perform sliding window interception on the standardized data stream received from the protocol conversion module, and extract the fluctuation feature vectors in the photovoltaic output time series data and the dynamic response vectors in the energy storage charge and discharge state data; The coupling analysis unit is used to input the fluctuation characteristic vector and the dynamic response vector into a dynamic weighted model constructed based on the photovoltaic output fluctuation rate and the energy storage response delay to generate the multi-dimensional coupling state matrix.

4. The building solar-storage-charging integrated energy management system according to claim 3 is characterized in that: The coupling analysis unit further includes: The photovoltaic output fluctuation rate is calculated by the variance of the fluctuation characteristic vector; The energy storage response delay is calculated by using the time delay parameter in the dynamic response vector; The dynamic weighted model dynamically adjusts the weight coefficient according to the ratio of the photovoltaic output fluctuation rate to the energy storage response delay, performs weighted superposition on the fluctuation characteristic vector and the dynamic response vector, and generates a coupling characteristic tensor; The coupling characteristic tensor is matrix-processed to obtain the multi-dimensional coupling state matrix.

5. The building solar-storage-charging integrated energy management system according to claim 1, characterized in that: The scheduling optimization module is also used to: generating a charge and discharge power allocation plan based on the photovoltaic output prediction value and the energy storage remaining capacity in the multi-dimensional coupling state matrix received from the data fusion module, wherein the photovoltaic output prediction value is calculated by regression analysis of the fluctuation eigenvector; Combined with the grid-connected power threshold in the grid dispatch instruction data received from the data acquisition module, the charging and discharging power allocation scheme is subjected to constrained optimization, and the constrained optimization is based on the multi-objective function to generate the real-time dispatch strategy.

6. The building solar energy storage and charging integrated energy management system according to claim 5, characterized in that: Also includes: Based on the photovoltaic output prediction value in the multi-dimensional coupling state matrix received from the data fusion module and the weight coefficient output from the dynamic weighting model, a multi-objective function is constructed with the goal of maximizing the photovoltaic absorption rate, minimizing the energy storage cycle loss, and minimizing the grid frequency deviation; An improved particle swarm algorithm is used to solve the multi-objective function, wherein the inertia weight of the improved particle swarm algorithm is dynamically adjusted according to the photovoltaic output fluctuation rate to generate a Pareto optimal solution set; The real-time dispatching strategy is selected from the Pareto optimal solution set according to the priority weight in the power grid dispatching instruction data received from the data acquisition module.

7. The building solar energy storage and charging integrated energy management system according to claim 1 is characterized in that: The instruction generation module includes: An instruction encoding unit, configured to convert the charge and discharge power values in the real-time scheduling strategy output by the scheduling optimization module into PWM modulation signals based on a Pareto optimal solution set; The instruction distribution unit is used to synchronously send the PWM modulation signal to the BMS system in the energy storage unit and the charging pile controller in the charging facility unit through the OPC UA protocol.

8. The building solar-storage-charging integrated energy management system according to claim 1, characterized in that: The feedback module is further configured to: Collecting actual output data of the photovoltaic power generation unit, real-time state of charge data of the energy storage unit, and power consumption data of the charging facility unit; Calculating a power deviation rate based on a deviation between the actual output data and a predicted value in the real-time scheduling strategy generated by the scheduling optimization module; generating a feedback correction coefficient based on the power deviation rate and the energy storage response delay obtained from the data fusion module; The feedback correction coefficient is input into the scheduling optimization module to trigger the parameter update of the multi-objective function.

9. The building solar-storage-charging integrated energy management system according to claim 8, characterized in that: Also includes: Performing a sliding average filter on the deviation between the actual output data and the predicted value, wherein the filter window length is set according to the sampling period of the photovoltaic output time series data; Calculating a power compensation amount based on the filtered deviation and the rate of change of the energy storage charge state extracted from the dynamic response vector of the data fusion module; The feedback correction coefficient is generated based on the power compensation amount.

10. The building solar-storage-charging integrated energy management system according to claim 1, characterized in that: Also includes: A security verification module, configured to perform overload protection verification and protocol compliance verification on the control instructions distributed by the instruction generation module; When the safety verification module detects that the power value in the control instruction exceeds the rated capacity of the equipment acquired by the data acquisition module, it triggers the scheduling optimization module to regenerate the limiting scheduling strategy based on the multi-objective function.

Images