Coordinated control method, device, medium, product and multi-source heterogeneous energy system

Abstract

The coordination control method, device, medium, product and multi-source heterogeneous energy system provided by the embodiments of the present application, the multi-source heterogeneous energy system comprises a power grid side, a user side load, at least one energy storage end and at least one energy production end. The method comprises: acquiring electrical parameters corresponding to a common connection point between the power grid side and the user side and a branch where at least one load of the user side is located; determining a current load state of the user side based on the electrical parameters; acquiring load power prediction information of the user side based on the current load state of the user side and a past load state of the user side; and dynamically adjusting working parameters of the load, the energy storage end and the energy production end based on the load power prediction information, production energy prediction information of the energy production end and a state of charge of the energy storage end. Through load classification prediction and combined multi-source state collaborative adjustment, precise anti-backflow control is realized.

Description

Technical Field

[0001] This application relates to the field of power system control technology, and in particular to a coordinated control method, device, medium, product, and multi-source heterogeneous energy system. Background Technology

[0002] With the large-scale integration of distributed photovoltaic (PV) power into microgrids, backflow prevention has become a critical requirement for grid-connected operation. Traditional control methods rely heavily on passive power limiting at the point of common coupling (PCC) and real-time feedback adjustment. This makes it difficult to cope with rapid fluctuations in PV power and load, and it is prone to instantaneous backflow. The control accuracy and response speed are insufficient to meet the requirements for stable operation.

[0003] Existing load identification and prediction methods do not classify and model continuous, intermittent, and fluctuating loads, but only use a uniform prediction method, resulting in insufficient accuracy in load power prediction. This makes it impossible to provide accurate forward-looking data support for multi-source coordinated control, and the rationality of regulation is poor.

[0004] The current system does not combine load classification prediction, new energy output prediction and energy storage status for joint optimization, lacks intelligent collaborative control mechanism, and is difficult to improve energy utilization while ensuring backflow prevention. It has problems such as single control target and poor overall operation economy. Summary of the Invention

[0005] This application provides a coordinated control method, device, medium, product, and multi-source heterogeneous energy system, which achieves precise anti-backflow control by classifying and predicting three types of loads and combining multi-source state coordinated adjustment.

[0006] In a first aspect, embodiments of this application provide a coordinated control method for a multi-source heterogeneous energy system, wherein the multi-source heterogeneous energy system includes a grid side, a user-side load, at least one energy storage terminal, and at least one energy production terminal;

[0007] The method includes:

[0008] Obtain the common connection point between the power grid side and the user side, and the electrical parameters corresponding to at least one branch where the load is located on the user side;

[0009] Based on the electrical parameters, determine the current load status on the user side;

[0010] The current load status and past load status of the user side are input into the pre-trained time series prediction model. The load includes at least continuous load, intermittent load and fluctuating load.

[0011] By performing trend fitting based on the current and past temporal characteristics of the user side, the power fluctuation information corresponding to the continuous load is obtained as the load power prediction information.

[0012] Based on the time characteristics and environmental parameters corresponding to the current time-series characteristics of the user side, and the time characteristics and environmental parameters corresponding to the past time-series characteristics of the user side, the start and stop times and operating power corresponding to the intermittent load are obtained as the load power prediction information.

[0013] Based on the time period characteristics corresponding to the current time-series characteristics on the user side and the time period characteristics corresponding to the past time-series characteristics, the power demand corresponding to the fluctuating load is obtained as the load power prediction information.

[0014] Based on the load power prediction information, the energy production prediction information of the energy production end, and the state of charge of the energy storage end, the operating parameters of the load, the energy storage end, and the energy production end are dynamically adjusted to achieve anti-backflow control of the multi-source heterogeneous energy system.

[0015] In one possible implementation, the common connection point and at least one load branch on the user side are equipped with a non-intrusive acquisition device.

[0016] The acquisition of the common connection point between the power grid side and the user side, and the electrical parameters corresponding to at least one of the branches where the load is located on the user side, includes:

[0017] The non-intrusive acquisition device is used to acquire the electrical parameters of the common connection point between the power grid side and the user side, and the branch where at least one of the loads is located on the user side.

[0018] In one possible implementation, determining the current load state on the user side based on the electrical parameters includes:

[0019] Extract the harmonic and transient impulse characteristics from the electrical parameters;

[0020] The harmonic features and transient impact features are input into a pre-trained extraction model to extract nonlinear load features, and the load state is determined based on the extracted nonlinear load features.

[0021] In one possible implementation, after dynamically adjusting the operating parameters of the load, the energy storage terminal, and the energy production terminal, the method further includes:

[0022] Using the non-intrusive acquisition device, electrical parameters corresponding to the common connection point between the power grid side and the user side, and at least one branch of the load on the user side are obtained to determine the actual load status of the user side at the current moment.

[0023] If the deviation between the actual load status on the user side and the load power prediction information reaches a preset threshold, the time-series prediction model is adjusted based on the actual load status on the user side at the current moment and the corresponding load power prediction information.

[0024] In one possible implementation, dynamically adjusting the operating parameters of the load, the energy storage terminal, and the energy production terminal based on the load power prediction information, the energy production prediction information of the energy production terminal, and the state of charge of the energy storage terminal includes:

[0025] Under the constraints of reverse power on the user side, absorption rate at the energy production end, and charging and discharging loss at the energy storage end, a target optimization function is constructed based on the load power prediction information, the production energy prediction information, and the state of charge.

[0026] Based on the objective optimization function, a control strategy is generated to dynamically adjust the operating parameters of the load, the energy storage terminal, and the energy production terminal.

[0027] Secondly, embodiments of this application provide a coordinated control device for a multi-source heterogeneous energy system, wherein the multi-source heterogeneous energy system includes a grid side, a user-side load, at least one energy storage terminal, and at least one energy production terminal.

[0028] The device includes:

[0029] The first acquisition module is used to acquire the common connection point between the power grid side and the user side, and the electrical parameters corresponding to at least one branch where the load is located on the user side;

[0030] The determination module is used to determine the current load status on the user side based on the electrical parameters;

[0031] The input module is used to input the current load status of the user side and the past load status of the user side into the pre-trained time series prediction model. The load includes at least continuous load, intermittent load and fluctuating load.

[0032] The second acquisition module is used to perform trend fitting based on the current temporal characteristics of the user side and the past temporal characteristics of the user side to obtain the power fluctuation information corresponding to the continuous load, as load power prediction information.

[0033] The second acquisition module is used to acquire the start and stop times and operating power corresponding to the intermittent load based on the time characteristics and environmental parameters corresponding to the current time-series characteristics of the user side and the time characteristics and environmental parameters corresponding to the past time-series characteristics of the user side, as the load power prediction information.

[0034] The second acquisition module is used to acquire the power demand corresponding to the fluctuating load based on the time period characteristics corresponding to the current time-series characteristics of the user side and the time period characteristics corresponding to the past time-series characteristics, as the load power prediction information.

[0035] The adjustment module is used to dynamically adjust the operating parameters of the load, the energy storage end, and the energy production end based on the load power prediction information, the energy production prediction information of the energy production end, and the state of charge of the energy storage end, so as to realize the anti-backflow control of the multi-source heterogeneous energy system.

[0036] In one possible implementation, the common connection point and at least one load branch on the user side are equipped with a non-intrusive acquisition device.

[0037] The first acquisition module is also used to acquire electrical parameters corresponding to the common connection point between the power grid side and the user side, and at least one branch of the load on the user side, using the non-intrusive acquisition device.

[0038] In one possible implementation, the device further includes: an extraction module;

[0039] The extraction module is used to extract harmonic features and transient impact features from the electrical parameters;

[0040] The determination module is further configured to input the harmonic features and the transient impact features into a pre-trained extraction model to extract nonlinear load features, and determine the load state based on the extracted nonlinear load features.

[0041] In one possible implementation, the first acquisition module is used to acquire the electrical parameters corresponding to the common connection point between the power grid side and the user side, and at least one branch of the load on the user side, using the non-intrusive acquisition device, so as to determine the actual load status of the user side at the current moment.

[0042] The adjustment module is also used to adjust the parameters of the time-series prediction model based on the actual load status of the user side and the corresponding load power prediction information when the deviation between the actual load status on the user side and the load power prediction information reaches a preset threshold.

[0043] In one possible implementation, the adjustment module is further configured to construct a target optimization function based on the load power prediction information, the production energy prediction information, and the state of charge, under the constraints of the reverse power on the user side, the absorption rate of the energy production end, and the charging and discharging loss of the energy storage end.

[0044] The adjustment module is also used to generate a control strategy based on the target optimization function to dynamically adjust the operating parameters of the load, the energy storage terminal, and the energy production terminal.

[0045] Thirdly, embodiments of this application provide a multi-source heterogeneous energy system, including a grid side, a user-side load, at least one energy storage terminal, and at least one energy production terminal;

[0046] The multi-source heterogeneous energy system is configured to employ the coordinated control method described in any one of the first aspects and / or various possible implementations of the first aspect above, to obtain the common connection point between the grid side and the user side, and the electrical parameters corresponding to at least one branch where the load is located on the user side; based on the electrical parameters, to determine the current load state on the user side; to input the current load state and the past load state on the user side into a pre-trained time-series prediction model, wherein the load includes at least continuous load, intermittent load, and fluctuating load; and to obtain the power fluctuation information corresponding to the continuous load as the load power prediction information by performing trend fitting based on the current time-series characteristics and the past time-series characteristics on the user side. Based on the current time-series characteristics and environmental parameters corresponding to the user's current time-series characteristics, and the time-series characteristics and environmental parameters corresponding to the user's past time-series characteristics, the start-stop times and operating power corresponding to the intermittent load are obtained as the load power prediction information. Based on the time period characteristics corresponding to the current time-series characteristics of the user and the time period characteristics corresponding to the past time-series characteristics, the power demand corresponding to the fluctuating load is obtained as the load power prediction information. Based on the load power prediction information, the energy production prediction information of the energy production end, and the state of charge of the energy storage end, the operating parameters of the load, the energy storage end, and the energy production end are dynamically adjusted to achieve the anti-reverse flow control of the multi-source heterogeneous energy system.

[0047] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0048] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0049] The coordinated control method, device, medium, product, and multi-source heterogeneous energy system provided in this application determine the current load state by collecting electrical parameters of the common connection point and load branches. The current and historical load states are input into a time-series prediction model. For continuous loads, trend fitting is used to obtain power fluctuations. For intermittent loads, start-up, shutdown, and power are predicted by combining time and environmental parameters. For fluctuating loads, power demand is predicted based on time period characteristics, resulting in refined load power prediction information. Based on this prediction information, energy production prediction information, and energy storage charge status, the operating parameters of the load, energy storage, and energy production ends are dynamically adjusted to achieve anti-reverse flow control of the multi-source heterogeneous energy system, thereby improving load prediction accuracy, realizing forward-looking coordinated regulation, and improving local energy consumption efficiency and system operation stability while ensuring grid connection safety. Attached Figure Description

[0050] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0051] Figure 1 A schematic diagram of the architecture of the coordination and control method provided in this application;

[0052] Figure 2 A schematic diagram illustrating the load decomposition of various loads included in a multi-source heterogeneous energy system at different time periods in one embodiment of this application;

[0053] Figure 3 Flowchart of the coordination and control method provided in this application Figure 1 ;

[0054] Figure 4 Flowchart of the coordination and control method provided in this application Figure 2 ;

[0055] Figure 5 A schematic diagram of the coordination control device provided in this application;

[0056] Figure 6 A schematic diagram of the structure of the electronic device provided in this application.

[0057] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0058] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0059] The widespread adoption of distributed photovoltaics has driven the upgrading of microgrid anti-backflow requirements. Existing anti-backflow solutions are divided into three categories: EMS optimized scheduling, inverter power limiting, and external dedicated controllers. However, they are all mainly passive responses, relying on PCC real-time monitoring, which makes it difficult to cope with the dynamic changes in photovoltaic output and load, and easily leads to instantaneous backflow and energy waste.

[0060] Traditional anti-backflow technology has several core defects. It cannot achieve fine identification of individual loads by only collecting total load power. Large load prediction deviations lead to a lack of accurate scheduling support. Furthermore, it does not combine load behavior differences to formulate strategies, resulting in poor multi-energy coordination. In addition, intrusive monitoring has high deployment costs, weak scalability, and insufficient ability to quickly correct load changes.

[0061] This application provides a coordinated control method for multi-source heterogeneous energy systems. It acquires electrical parameters through a non-intrusive acquisition device and innovatively extracts harmonic and transient impact characteristics, enabling refined and differentiated identification of user-side load states (continuous, intermittent, fluctuating, and other types of loads). Based on this, and combined with historical data, it performs classified power prediction for different load behavior patterns, thereby obtaining high-precision load power prediction information. Furthermore, with reverse current prevention as a rigid constraint, it integrates energy production-side predictions and storage-side states to construct a multi-objective optimization function and generate a coordinated control strategy, dynamically adjusting the operating parameters of each unit within the system.

[0062] Figure 1 This is a schematic diagram of the architecture of a coordination control method provided in an embodiment of this application. Figure 1As shown, a coordinated control method for multi-source heterogeneous energy systems can be applied to this architecture, which includes a NILM monitoring unit, a load behavior prediction unit, a central coordination control unit, an execution unit, and a monitoring feedback unit. The NILM monitoring unit can collect and analyze electrical data in real time. Through feature extraction and pattern recognition algorithms, it can accurately identify and label the behavior characteristics of a single load type, generating NILM identification data and transmitting it to the load behavior prediction unit. The load behavior prediction unit can predict the power demand of the load within the next 15 minutes based on an LSTM-Transformer hybrid model, with the prediction error controlled within 8%, outputting high-precision load prediction results. The central coordination control unit can receive data from the above two units, combine it with a preset anti-reverse current threshold, construct a multi-objective optimization model, comprehensively consider photovoltaic output, energy storage status, and load demand, and generate coordinated control commands. The central coordination control unit can then issue commands to the execution unit, where the photovoltaic inverter, energy storage converter, and dispatchable load jointly perform power regulation.

[0063] Understandably, the execution unit is responsible for receiving instructions from the central coordination and control unit and executing specific power regulation operations. These operations may include: photovoltaic inverter control, such as adjusting the output power of the photovoltaic grid-connected inverter to limit excess photovoltaic power from flowing back to the grid; energy storage converter control, to control the charging and discharging power of the energy storage system, charging when photovoltaic output is in surplus and discharging during peak load periods to smooth out power fluctuations; and dispatchable load control, such as time-series scheduling of delayable loads like water heaters and air conditioners to optimize load curves and improve local consumption capacity.

[0064] Optionally, the monitoring and feedback unit can also collect the operating data and status of each unit at the common connection point (PCC) in real time, and send the status feedback data back to the central coordination and control unit to correct the multi-objective optimization model, update the anti-backflow threshold and adjust the control commands to achieve closed-loop control.

[0065] Specifically, the scheme achieves refined load identification through NILM, performs short-term load forecasting by combining the LSTM-Transformer hybrid model, generates forward-looking control commands by the central coordination and control unit, and finally completes closed-loop control through the execution unit and monitoring feedback unit. This can solve the problems of lag and blindness in traditional anti-reverse flow technology, so as to achieve 100% local consumption of photovoltaic power in microgrids.

[0066] Figure 2 This is a schematic diagram illustrating the load decomposition of various loads in a multi-source heterogeneous energy system at different time periods, provided as an embodiment of this application. For example... Figure 2The figure shows a comparison of the total load power curve and the power changes over time for the four typical load types (base load, air conditioning load, charging pile load, and machine tool load) decomposed by the NILM system. The horizontal axis represents time (0–60 minutes), and the vertical axis represents power (kW). The total load power curve (P_total) is composed of the superposition of the base load, air conditioning load, charging pile load, and machine tool load, exhibiting multi-segment stepped fluctuations, corresponding to the switching behavior of different loads. The base load (lighting / server) has a stable power of approximately 20kW, representing a constant load that operates continuously throughout the entire period and constitutes the system's power base. The air conditioning load operates stably during the 0-20min and 40-60min periods, with a power of approximately 15kW; it shuts down during the 20–40min period, with a power of 0. The charging pile load is put into operation during the 15-45min period, with a power of approximately 25kW, and is 0 during the remaining periods, representing a typical intermittent high-power load. The machine tool load operates during the 32-40min period, with a power of approximately 30kW, and is 0 during the remaining periods, representing a short-term high-power industrial load.

[0067] Understandably, the process begins by collecting total electrical data at the PCC using the NILM monitoring unit. A load decomposition algorithm is then employed to break down the total load into four categories: basic load, air conditioning load, charging pile load, and machine tool load. The start / stop times, power amplitudes, and operating patterns of each load are then marked, forming a standardized load characteristic dataset. Subsequently, the NILM-identified data is input into an LSTM-Transformer hybrid model to predict the power demand of each load over the next 15 minutes. For example, the model predicts the basic load will maintain a constant power of 20kW; the air conditioning will be restarted after 40 minutes with a power of approximately 15kW; the charging pile will be deactivated after 45 minutes with a power reduction to 0; and the machine tool will be deactivated after 40 minutes with a power reduction to 0. The prediction error is controlled within 8%.

[0068] Based on load forecasting results and preset anti-reverse current thresholds (e.g., PCC reverse current power ≤ 0kW), the central coordination and control unit constructs a multi-objective optimization model, such as objective 1: satisfying the zero reverse current constraint; objective 2: maximizing local photovoltaic power consumption; and objective 3: minimizing energy storage losses and the impact on load comfort. Taking the 30–40 minute period as an example, the peak total load is approximately 90kW (20+15+25+30). If the photovoltaic output is 80kW, the energy storage system is instructed to discharge at 10kW to supplement the photovoltaic output gap and avoid purchasing electricity from the grid. If the photovoltaic output is 100kW, the photovoltaic inverter is instructed to limit its output to 90kW, while the energy storage is charged at 10kW to consume the remaining photovoltaic power, achieving zero reverse current.

[0069] After the instructions are determined, the photovoltaic inverter control adjusts the output power according to the optimization results to avoid backflow of excess photovoltaic power; the energy storage converter control charges when there is surplus photovoltaic power and discharges during peak load periods to smooth out power fluctuations; the dispatchable load control delays the charging pile's operation by 10 minutes if the photovoltaic output is severely insufficient, avoiding the high-power operation of the machine tool and reducing the total load peak. Simultaneously, the monitoring and feedback unit collects the PCC power flow direction and the operating status of each unit in real time. If instantaneous backflow is detected or the prediction deviation exceeds the threshold, the status is fed back to the central coordination and control unit to dynamically correct the optimization model and control instructions, achieving closed-loop control.

[0070] Specifically, for example, in a microgrid system of an industrial park, the NILM monitoring unit of this application collects total current and voltage signals in real time at the main incoming line of the distribution cabinet. Through an edge computing gateway, an LSTM-Transformer hybrid model is run to achieve second-level identification and decomposition of loads such as lighting, air conditioning, electric vehicle charging stations, and CNC machine tools within the park. The central coordination and control unit, combining real-time photovoltaic output (assuming 40kW) with load forecasting results, automatically reduces the output power of the photovoltaic inverter when it detects that the total load is lower than the photovoltaic output due to the simultaneous operation of charging stations and machine tools, thus preventing backfeeding to the grid.

[0071] Figure 3 A flowchart illustrating a coordination control method provided in this application. Figure 1 .like Figure 3 As shown in the embodiment of this application, a coordinated control method for a multi-source heterogeneous energy system is provided. The multi-source heterogeneous energy system includes a grid side, a user-side load, at least one energy storage terminal, and at least one energy production terminal. The method includes:

[0072] S301. Obtain the electrical parameters corresponding to the common connection point between the power grid side and the user side, and the branch where at least one load on the user side is located.

[0073] In this system, non-intrusive data acquisition devices are installed at the common connection point between the grid and the user side, and at least one load branch on the user side. In a multi-source heterogeneous energy system, the grid side is the public power grid, serving as the system's backup power source and standard grid connection interface; examples include public distribution networks and transformers. User-side loads are various electrical devices in commercial and industrial parks, industrial parks, or residential scenarios, including continuous, intermittent, and fluctuating loads, such as production line equipment, air conditioners, and charging piles. The energy storage end is the user-side energy storage battery system, which may include energy storage battery packs and energy storage converters (PCS) or battery energy storage systems (BESS), used to smooth power fluctuations and absorb surplus energy. The energy production end is the user-side self-consumption distributed generation equipment, such as distributed photovoltaic modules in industrial parks, small wind turbines, and micro gas turbines, belonging to the user-side local power generation units. Furthermore, non-intrusive data acquisition devices are deployed at the common connection point (PCC) between the grid and the user side, and at least one load branch, to collect electrical parameters without wiring or disconnecting the load.

[0074] Understandably, non-intrusive data acquisition devices are smart meters or non-intrusive load monitoring (NILM) acquisition terminals deployed at the microgrid's point of common coupling (PCC) and critical load branches. These devices do not require disassembling the load's internal wiring; they acquire electrical parameters seamlessly simply by collecting electrical signals at the grid's inlet or the front end of the load branch. The electrical parameters corresponding to the PCC include: three-phase voltage amplitude, current amplitude, active power, reactive power, power factor, frequency, and power flow direction, used to characterize the power interaction between the user side and the public grid. The electrical parameters corresponding to at least one load branch on the user side include: branch voltage, current, instantaneous power, harmonic components, and event characteristics (such as power surge amplitude and rising / falling edge slope), used to provide raw data support for subsequent load type identification and behavioral feature extraction.

[0075] Specifically, in the deployment of non-intrusive data acquisition devices, smart meters or NILM (Non-Intrusive Energy Meters) are installed at the point of common connection (PCC) and key load branches (such as air conditioners and charging piles). For example, in a microgrid in an industrial or commercial park, three-phase voltage / current sensors are deployed at the PCC to collect real-time power interaction data between the grid side and the user side; single-phase sensors are deployed at the front end of key load branches (such as air conditioners and charging piles) to extract branch-level electrical signals. This deployment method ensures that the collected data covers the full load status of the user side, providing complete input for subsequent load identification and prediction.

[0076] S302. Determine the current load status on the user side based on electrical parameters.

[0077] The load status may include, for example, the operating / stopping status of each load, the current active / reactive power, the load combination composition, and the total load power.

[0078] Understandably, load status identification is achieved through the Non-Intrusive Load Breakdown (NILM) algorithm. For example, power surge events in the PCC and each load branch are detected, and key features such as the time of event occurrence, power change amplitude, and current waveform characteristics are extracted. The extracted features are then matched with a preset load feature library (such as electrical feature templates for typical loads like lighting, servers, air conditioners, charging piles, and machine tools) to complete load type identification. Combined with the switching status, real-time operating power, and operating time of each load, the current load status on the user side is determined, including: the operating / stopping status of each load, current active / reactive power, load combination composition, and total load power.

[0079] Specifically, steady-state characteristics (such as power factor and harmonic content) and transient characteristics (such as peak starting current) can be extracted from electrical parameters. If the acquisition point is located at the PCC, non-intrusive load monitoring technology can be used to decompose the total current signal into the power signals of each sub-load, and then assign a status label of "running", "standby", "fault" or "offline" to each identified load.

[0080] S303. Input the current load status of the user side and the past load status of the user side into the pre-trained time series prediction model.

[0081] The load includes at least continuous loads, intermittent loads, and fluctuating loads. Continuous loads (such as servers and lighting) can be predicted for their stable power fluctuations by trend fitting based on current and past time-series characteristics; intermittent loads (such as air conditioners and water heaters) can be predicted for their future start-up and shutdown times and corresponding operating power by combining time characteristics (such as current time and day of the week) and environmental parameters (such as outdoor temperature); fluctuating loads (such as charging piles and machine tools) can be predicted for their random power demand based on time-period characteristics (such as commuting peak hours).

[0082] Load power prediction information may include the following dimensions:

[0083] Time dimension: Equally spaced time series points within the next 15 minutes (or other preset durations, such as 5 minutes or 30 minutes), for example, a power prediction sequence of 16 time points with a step size of 1 minute.

[0084] Power Dimension: Total load power forecast, which is the sum of the active power of all loads on the user side at various points in the future, used to characterize the overall electricity demand; Type-specific load power forecast, which is the forecast power broken down by load type, such as the future power sequence of basic load (lighting / server), air conditioning load, charging pile load, machine tool load, etc.; Critical load event forecast, which is to predict the switching events of high-power, intermittent loads, such as event-level forecasts such as "charging pile will be taken out of operation at 45 minutes" and "machine tool will be put into operation at 32 minutes".

[0085] Confidence level dimension: the confidence interval or error range of the predicted power values ​​for each time point and each type of load, such as "total load prediction error ≤ 8%" and "charging pile power prediction error ≤ 5%", used to evaluate the reliability of the prediction results.

[0086] Constraint Dimension: Constraint information generated by combining load operating characteristics, such as "the minimum operating power of the air conditioning load is 10kW" and "charging piles cannot be started after 22:00 at night", provides boundary conditions for subsequent optimization control.

[0087] Understandably, by determining the current load status on the user side and the past load status stored in the system, this data is transformed into time series data, and periodic, trend, and burst features are extracted. Subsequently, the extracted features are input into a pre-trained time series prediction model.

[0088] Optionally, a hybrid LSTM-Transformer model can be used. LSTM (Long Short-Term Memory) is responsible for capturing long-term cyclical patterns of the load (such as fixed daily start times), while Transformer is responsible for handling complex nonlinear dependencies (such as random equipment start-ups and shutdowns).

[0089] S304. Trend fitting is performed based on the current time-series characteristics of the user side and the past time-series characteristics of the user side to obtain the power fluctuation information corresponding to the continuous load, which is used as the load power prediction information.

[0090] S305. Based on the time characteristics and environmental parameters corresponding to the current time-series characteristics of the user side, and the time characteristics and environmental parameters corresponding to the past time-series characteristics of the user side, obtain the start and stop times and operating power corresponding to the intermittent load as load power prediction information.

[0091] S306. Based on the time period characteristics corresponding to the current time-series characteristics of the user side and the time period characteristics corresponding to the past time-series characteristics, obtain the power demand corresponding to the fluctuating load as load power prediction information.

[0092] Among them, the time-series prediction model is a dedicated time-series prediction model built by the load behavior prediction unit based on the load ledger and historical data identified by NILM.

[0093] For continuous loads, trend fitting is performed based on the current and past time-series characteristics of the user side to obtain the power fluctuation information of the continuous load in the future prediction period, and this power fluctuation information is used as part of the load power prediction information. For intermittent loads, the start and stop times and corresponding operating power of the intermittent loads are predicted by combining the time characteristics and environmental parameters corresponding to the current and past time-series characteristics of the user side, and these start and stop times and operating power are used as part of the load power prediction information. For fluctuating loads, the power demand of the fluctuating loads is obtained based on the time period characteristics and user behavior habits corresponding to the current and past time-series characteristics of the user side, and this power demand is used as part of the load power prediction information.

[0094] Based on historical user charging habits (such as average charging time and common charging periods) and current time characteristics, the power demand curve for future periods is predicted. After inputting the current load status and past load status into the model, the model is set to a prediction period of 15 minutes. The system performs rolling optimization according to the set time step, that is, data is re-collected and the prediction results are updated at fixed intervals (such as 1 minute or 5 minutes) to correct prediction deviations. When the predicted power at a certain future moment changes drastically compared to the current moment, and the change is greater than or equal to 20%, it is judged as a power surge event, and the system automatically issues a warning signal, indicating that there may be high-power equipment switching or abnormal operating conditions.

[0095] S307. Based on load power prediction information, energy production prediction information from the energy production end, and the state of charge of the energy storage end, dynamically adjust the operating parameters of the load, energy storage end, and energy production end to achieve anti-backflow control of multi-source heterogeneous energy systems.

[0096] Among them, with the optimization objectives of preventing backflow, high absorption, and low loss, the load power prediction information, energy production end output prediction information, and energy storage end state of charge (SOC) are used as inputs. Under the constraint that the reverse power is not greater than the preset ratio of the rated power of the point of common coupling, a multi-objective optimization function is constructed and solved to generate a coordinated control strategy.

[0097] Understandably, the controller first performs a fusion analysis of the three types of core input information to identify potential supply and demand contradictions, such as supply and demand balance verification, which involves superimposing "load power prediction information" and "production energy prediction information" to calculate the net load gap or surplus; and energy storage capacity assessment, which involves reading the state of charge (SOC) of the energy storage system. If the SOC is lower than a set threshold (e.g., 20%), its discharge capacity is limited; if the SOC is too high (e.g., 90%), its charging capacity is limited to prevent overcharging.

[0098] Optionally, an improved multi-objective optimization algorithm (such as the improved Whale Algorithm IWOA or Model Predictive Control MPC) can be used to solve the problem. The objective function can be set as follows: economic efficiency, minimizing the user-side electricity cost (prioritizing the use of low-cost photovoltaic / energy storage power supply and taking advantage of peak-valley electricity price differences for arbitrage); or stability, smoothing out power fluctuations, eliminating line overload risks, and maintaining voltage / frequency stability; or equipment lifespan, optimizing energy storage charging and discharging strategies, reducing the number of deep charge and discharge cycles, and extending battery life.

[0099] Dynamic adjustment of operating parameters can specifically include:

[0100] Regulating energy production: Limiting or increasing the output of photovoltaic inverters and wind turbine converters to prioritize matching the power generation to the predicted load demand.

[0101] Regulating the energy storage end: Controlling the energy storage converter to perform charging or discharging, charging to consume surplus electricity when there is a photovoltaic surplus, and discharging to supplement the power gap when the load is at its peak;

[0102] Adjusting user-side loads: Sequential start-up and shutdown of dispatchable loads, power adjustment, and optimization of power consumption curves to improve local consumption levels.

[0103] Specifically, this embodiment uses a microgrid in an industrial and commercial park as the actual application scenario. The park is equipped with a 100kW distributed photovoltaic system as the energy production end and a 150kWh energy storage system as the energy storage end. User-side loads include lighting, air conditioning, charging piles, CNC machine tools, etc. The system is connected to the public grid through PCC points, requiring strict adherence to anti-reverse current, self-consumption, and no grid connection of surplus power. Non-intrusive data acquisition devices deployed at PCC points and load branches can be used to collect electrical parameters such as voltage, current, and power in real time. Based on the collected electrical parameters, the type, start / stop status, and real-time power of each load on the user side are identified to determine the current load status. Combining the current load status with historical load operation data, a time-series prediction model is used to obtain future load power prediction information. Based on the load power prediction information, photovoltaic output prediction information, and energy storage state of charge (SOC), under the premise of meeting anti-reverse current constraints, the load operation sequence, energy storage charging and discharging power, and photovoltaic output power are dynamically adjusted to achieve forward-looking anti-reverse current and multi-energy collaborative optimization.

[0104] This application provides a coordinated control method for multi-source heterogeneous energy systems. It determines the current load state by collecting electrical parameters from the common junction point and load branches, inputs the current and historical load states into a time-series prediction model, uses trend fitting to obtain power fluctuations for continuous loads, predicts start-up and shutdown and power for intermittent loads based on time and environmental parameters, and predicts power demand for fluctuating loads based on time-period characteristics, thus obtaining refined load power prediction information. Based on this prediction information, energy production prediction information, and energy storage state of charge, the method dynamically adjusts the operating parameters of the load, energy storage, and energy production ends to achieve anti-reverse flow control of the multi-source heterogeneous energy system. This improves load prediction accuracy, enables forward-looking coordinated regulation, and enhances local energy consumption efficiency and system operational stability while ensuring grid connection safety.

[0105] Figure 4 A flowchart illustrating a coordination control method provided in this application. Figure 2 .like Figure 4 As shown, this embodiment is... Figure 3 Based on the embodiments, a coordinated control method for multi-source heterogeneous energy systems is described in detail. The method includes:

[0106] S401. A non-intrusive acquisition device is used to acquire the electrical parameters corresponding to the common connection point between the power grid side and the user side, and the branch where at least one load on the user side is located.

[0107] Step S401 is similar to step S301 above, and will not be described again in this application.

[0108] S402. Extract harmonic and transient impact characteristics from electrical parameters.

[0109] Among them, harmonic characteristics include total harmonic distortion rate of current, harmonic content rate of each order, and harmonic phase; transient impact characteristics include current abrupt change, rate of change, and transient duration.

[0110] Specifically, wavelet transform algorithms can be used to decompose the collected electrical parameters into multiple scales to extract time-frequency domain information of transient impact characteristics; fast Fourier transform can be used to perform spectral analysis on steady-state electrical parameters to extract harmonic characteristics. The extracted harmonic and transient impact characteristics are used as inputs for non-intrusive load identification to distinguish different types of electrical equipment.

[0111] S403. Input the harmonic features and transient impact features into the pre-trained extraction model to extract nonlinear load features, and determine the load state based on the extracted nonlinear load features.

[0112] Nonlinear load characteristics include the rectification characteristics of the equipment, the current waveform distortion mode, and the switching frequency fingerprint. The extraction model can be a nonlinear feature extractor built on a deep belief network (DBN) or a convolutional neural network (CNN), which reduces and reconstructs the harmonic and transient features of the input through multi-layer nonlinear transformation, thereby separating the feature vector that represents the essential attributes of the load.

[0113] Understandably, the obtained harmonic and transient impact features are input into the model. Local electrical features are extracted through convolutional layers, and the load time-series variation is modeled through long short-term memory layers. Then, through an attention mechanism for weighting, the specific features of nonlinear loads such as inverter-driven motors and high-power rectifier equipment are accurately extracted. Based on the above nonlinear load features, matching and discrimination are performed in conjunction with the load feature library to determine the current load status on the user side, including: load type, start / stop status, current operating power, load switching events, and behavior pattern markers, thus completing the refined determination of the load status at the single device level.

[0114] Specifically, in the nonlinear load feature extraction process, the input features of the improved CNN-LSTM model include: current harmonic features: extracting the content and phase of each harmonic through Fast Fourier Transform; voltage transient features: extracting the amplitude and duration of voltage fluctuations through Wavelet Transform; and power factor features: calculating the ratio of active power to apparent power through steady-state power calculation. After processing by the CNN-LSTM model, the essential attributes of the load (such as the switching frequency fingerprint of the inverter drive motor) are separated, significantly improving the single-device-level recognition accuracy. For example, when a specific interharmonic spectrum and current impulse sequence are detected, the corresponding load is determined to be in the startup phase; when the harmonic content is stable within a specific range, the load is determined to be in normal operating condition.

[0115] S404. Input the current load status of the user side and the past load status of the user side into the pre-trained time series prediction model.

[0116] Among them, temporal feature extraction involves the model first performing feature engineering on the input data to extract key temporal features, including but not limited to: trend features, i.e., the long-term growth or decline trend of load power; periodic features, such as daily cycles (e.g., morning and evening peak hours), weekly cycles (e.g., differences between weekdays and weekends), and seasonal cycles (e.g., summer air conditioning load); and lag features, such as power values ​​at specific times in the past (e.g., the previous hour, the same time yesterday) as a reference benchmark.

[0117] Optionally, the time series prediction model can preferably employ a deep learning architecture, such as a Long Short-Term Memory (LSTM) network, a Gated Recurrent Unit (GRU) network, or a Temporal Convolutional Network (TCN), to effectively capture long-term dependencies and nonlinear dynamic characteristics in the load data.

[0118] Specifically, the model learns the fluctuation patterns of historical load curves and combines them with the real-time change rate of the current load to calculate the power change trajectory within a specific future time window (such as the next 15 minutes to 1 hour), and generates load power prediction information on the user side. Specifically, this can be represented as a power prediction sequence containing multiple future time steps, along with corresponding confidence intervals.

[0119] The current load status is input into the model along with past user-side load status stored in the historical operation database. This fully explores the temporal variation patterns of the load under different time periods and operating conditions, extracting the temporal features implicit in the current and historical states. Based on these temporal features, the model performs inference calculations and outputs user-side load power prediction information for a preset future time period. This includes the total load power prediction value, the load power prediction values ​​for each category, load change warning information, and the prediction error range, forming a complete and highly accurate load prediction result.

[0120] S405. Trend fitting is performed based on the current time-series characteristics of the user side and the past time-series characteristics of the user side to obtain the power fluctuation information corresponding to the continuous load, which is used as the load power prediction information.

[0121] S406. Based on the time characteristics and environmental parameters corresponding to the current time-series characteristics of the user side, and the time characteristics and environmental parameters corresponding to the past time-series characteristics of the user side, obtain the start and stop times and operating power corresponding to the intermittent load as load power prediction information.

[0122] S407. Based on the time period characteristics corresponding to the current time-series characteristics of the user side and the time period characteristics corresponding to the past time-series characteristics, obtain the power demand corresponding to the fluctuating load as load power prediction information.

[0123] Steps S405-S407 are similar to steps S305-S307 above, and will not be repeated here.

[0124] S408. Under the constraints of reverse power on the user side, absorption rate at the energy production end, and charging and discharging losses at the energy storage end, construct an objective optimization function based on load power prediction information, production energy prediction information, and state of charge.

[0125] In this embodiment, the optimization directions are anti-reverse flow compliance, maximization of new energy consumption, and minimization of energy storage loss. A multi-objective optimization function is constructed, and the constraints may include, but are not limited to: the reverse power on the user side does not exceed a preset proportion of the rated output power of the common connection point (such as 3% in this embodiment); the output of the energy production end does not exceed its maximum power generation capacity; the state of charge (SOC) of the energy storage end is maintained in a safe operating range of 20%-85%; and the energy storage charging and discharging loss is controlled within a preset range.

[0126] Understandably, load power prediction information, energy production prediction information from the energy production end, and the current state of charge of the energy storage end are used as inputs to construct an objective optimization function under the above constraints, which is then used to solve for the optimal coordinated control strategy.

[0127] Specifically, the central coordination and control unit can perform multi-objective optimization scheduling calculations, integrating load forecast curves, photovoltaic output forecast data obtained based on numerical weather forecasts and historical data, energy storage SOC status that meets 20%-85% operating constraints, and preset anti-reverse current thresholds to construct optimization objective functions including: reverse power ≤ 3% of rated output power; maximizing photovoltaic absorption rate; and minimizing energy storage charging and discharging losses.

[0128] S409. Based on the objective optimization function, generate a control strategy to dynamically adjust the operating parameters of the load, energy storage end, and energy production end.

[0129] Among these measures, an improved particle swarm optimization algorithm can be used to solve the objective optimization function and generate a forward-looking control strategy. For example, photovoltaic output can be prioritized to match load demand, and surplus power can be prioritized to charge energy storage. If energy storage is close to full charge, photovoltaic output can be dynamically reduced or flexible loads can be scheduled to run in advance. If a sudden increase in load leads to a power gap, energy storage can be prioritized to discharge and supplement the power supply, so as to avoid the superposition of peak power consumption from the grid.

[0130] Understandably, based on the optimal control parameters obtained from the solution, the central coordination and control unit generates specific control commands and issues them to each execution unit to implement dynamic coordinated control. For example, adjusting the energy production side can control the output power of the photovoltaic inverter, so that the photovoltaic output prioritizes matching the predicted power demand of the load; adjusting the energy storage side can control the charging and discharging power according to the energy storage SOC state, with surplus electricity prioritized to charge the energy storage, and the energy storage being controlled to discharge and supplement when there is a load power shortage; adjusting the user-side load can optimize the timing of dispatchable flexible loads, and start them in advance when necessary to absorb surplus photovoltaic power.

[0131] Optionally, when the state of charge (SBC) of the energy storage is close to its upper limit (e.g., 85% full charge) and the photovoltaic system still has a surplus, an overcharge prevention strategy can be implemented. This includes dynamically reducing the inverter output power at the energy production end, or activating a demand response mechanism to schedule flexible loads (such as non-real-time industrial equipment and electric vehicle charging) to operate in advance to absorb excess energy. Furthermore, when the energy storage SBC is within a safe range but requires adjustment, the charging and discharging power can be optimized to ensure optimal overall system operating efficiency.

[0132] S410. Using a non-intrusive acquisition device, obtain the electrical parameters corresponding to the common connection point between the power grid side and the user side, and the branch where at least one load on the user side is located, in order to determine the actual load status of the user side at the current moment.

[0133] Among them, non-intrusive acquisition devices deployed at the point of common coupling (PCC) and load branches can collect electrical parameters such as three-phase voltage, current, active power, harmonics and transient characteristics in real time. Based on the improved CNN-LSTM load decomposition model, the above parameters are processed to restore and determine the actual switching status, actual operating power and load combination of each load on the user side at the current moment, so as to obtain the actual load status on the user side at the current moment.

[0134] S411. When the deviation between the actual load status on the user side and the load power prediction information reaches a preset threshold, the time series prediction model is adjusted based on the actual load status on the user side at the current moment and the corresponding load power prediction information.

[0135] Specifically, when there is a significant deviation between the actual load status on the user side and the predicted information, the underlying time series prediction model is corrected online to improve the accuracy of subsequent predictions.

[0136] Understandably, the central coordination and control unit obtains the actual load status on the user side at the current moment and the corresponding load power prediction information, and calculates the deviation between the two. The deviation can be expressed in the form of absolute error, root mean square error, or relative percentage error. A preset threshold is set (e.g., 5% or adaptively adjusted according to historical volatility). When a deviation greater than or equal to the preset threshold is detected, the model parameter tuning mechanism is triggered.

[0137] After triggering hyperparameter tuning, the system automatically extracts environmental and load features for the current moment to construct a new training sample set. These features include, but are not limited to: time features (hour, day of the week, and whether it is a holiday); meteorological features (real-time light intensity and temperature); and actual load features (current actual power curve and harmonic characteristics). The detected actual load state is used as a new label value, which, together with the extracted features, forms a correction sample used to supplement or replace the original training dataset. Subsequently, depending on the specific type of time series prediction model, the corresponding hyperparameter tuning strategy is used to correct the model parameters.

[0138] Specifically, hyperparameter tuning based on machine learning / deep learning models (such as LSTM and LightGBM) can employ online learning or incremental learning strategies. This involves fine-tuning the model using newly acquired samples. Then, the model's hyperparameters are adjusted. For example, when training an LSTM, if prediction lag is observed, the learning rate can be adjusted or the number of memory units increased. If overfitting is detected, an early stopping mechanism or an increased dropout rate can be introduced. Finally, optimization algorithms (such as particle swarm optimization) are used to re-search for the optimal parameter combination. Using the prediction error of new samples as the fitness function, iterative optimization is performed to find a new parameter set that minimizes the error.

[0139] It should be noted that the updated parameters are used to retrain or fine-tune the time-series prediction model. To ensure the reliability of the new model, the system can quickly validate it using the retained validation set. If the validation error metrics meet the requirements, the new model is officially deployed as the current prediction engine, replacing the old model; if it still does not meet the requirements, the parameter tuning failure is recorded, and the system can choose to roll back to the previous version of the model. At the same time, an alarm is issued, prompting manual intervention to check the data quality or model architecture.

[0140] In some optional embodiments, the execution unit performs actions according to control commands. The photovoltaic inverter adjusts its output power through pulse width modulation, the energy storage converter switches between charging and discharging modes, and the dispatchable load operates according to an optimized timing sequence. Simultaneously, the monitoring and feedback unit collects real-time data on power flow, voltage, current, and the operating status of each device at the point of common coupling (PCC) using non-intrusive acquisition devices. Based on this, it determines the actual load status on the user side at the current moment, including actual switching status, actual power, and actual load combination. If the deviation between the actual operating data and the predicted data exceeds 5%, the central coordination and control unit immediately corrects the prediction model parameters and control commands to ensure the system maintains backflow-free operation and simultaneously updates the load behavior database to continuously improve subsequent prediction accuracy.

[0141] This application provides a coordinated control method for multi-source heterogeneous energy systems. It non-intrusively collects and monitors the user-side load status in real time, comparing it with predicted information. When the deviation exceeds a threshold, it automatically triggers parameter tuning of the time-series prediction model, using actual data to correct model parameters and achieve dynamic optimization of the prediction model. Simultaneously, based on the optimized prediction results, an objective function is constructed under multiple constraints to generate a control strategy, dynamically adjusting the operating parameters of photovoltaics, energy storage, and load, forming a closed-loop optimization system of "monitoring-prediction-control-correction." By using NILM (Non-Intrusive Prediction Model) and load forecasting, it achieves a shift from passive response to proactive control, significantly improving the renewable energy absorption rate and system stability while ensuring compliance with backflow prevention regulations.

[0142] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0143] Based on the same inventive concept, this application also provides a coordination control device for implementing the coordination control method described above. The solution provided by this coordination control device is similar to the solution described in the coordination control method above. Therefore, the specific limitations in one or more device embodiments provided below can be found in the limitations of the coordination control method above, and will not be repeated here.

[0144] In one embodiment, such as Figure 5 As shown, a coordination control device 500 for a multi-source heterogeneous energy system is provided. The multi-source heterogeneous energy system includes a grid side, a user-side load, at least one energy storage terminal, and at least one energy production terminal. The coordination control device 500 includes:

[0145] The first acquisition module 501 is used to acquire the common connection point between the power grid side and the user side, and the electrical parameters corresponding to the branch where at least one load on the user side is located.

[0146] The determination module 502 is used to determine the current load status on the user side based on electrical parameters;

[0147] The input module 503 is used to input the current load status of the user side and the past load status of the user side into the pre-trained time series prediction model. The load includes at least continuous load, intermittent load and fluctuating load.

[0148] The second acquisition module 504 is used to perform trend fitting based on the current time-series characteristics of the user side and the past time-series characteristics of the user side to obtain the power fluctuation information corresponding to the continuous load, as the load power prediction information.

[0149] The second acquisition module 504 is used to acquire the start and stop times and operating power of intermittent loads based on the time characteristics and environmental parameters corresponding to the current time-series characteristics of the user side and the time characteristics and environmental parameters corresponding to the past time-series characteristics of the user side, as load power prediction information.

[0150] The second acquisition module 504 is used to acquire the power demand corresponding to the fluctuating load based on the time period characteristics corresponding to the current time-series characteristics of the user side and the time period characteristics corresponding to the past time-series characteristics, as load power prediction information.

[0151] The adjustment module 505 is used to dynamically adjust the operating parameters of the load, energy storage end, and energy production end based on load power prediction information, energy production prediction information of the energy production end, and state of charge of the energy storage end, so as to realize the anti-backflow control of the multi-source heterogeneous energy system.

[0152] In one possible implementation, a non-intrusive data acquisition device is provided at the common connection point and at least one load branch on the user side.

[0153] The first acquisition module 501 is also used to acquire electrical parameters corresponding to the common connection point between the power grid side and the user side, and the branch where at least one load on the user side is located, using a non-intrusive acquisition device.

[0154] In one possible implementation, the device further includes: an extraction module;

[0155] The extraction module is used to extract harmonic and transient impulse characteristics from electrical parameters;

[0156] The determination module 502 is also used to input harmonic features and transient impact features into a pre-trained extraction model to extract nonlinear load features and determine the load state based on the extracted nonlinear load features.

[0157] In one possible implementation, the second acquisition module 504 is further configured to input the current load status of the user side and the past load status of the user side into a pre-trained time series prediction model, so as to obtain the load power prediction information of the user side based on the time series characteristics corresponding to the current load status of the user side and the time series characteristics corresponding to the past load status of the user side.

[0158] In one possible implementation, the first acquisition module 501 is used to acquire electrical parameters corresponding to the common connection point between the power grid side and the user side, and the branch where at least one load on the user side is located, using a non-intrusive acquisition device, so as to determine the actual load status of the user side at the current moment.

[0159] The adjustment module 505 is also used to adjust the parameters of the time-series prediction model based on the actual load status of the user side and the corresponding load power prediction information when the deviation between the actual load status on the user side and the load power prediction information reaches a preset threshold.

[0160] In one possible implementation, the adjustment module 505 is further configured to construct a target optimization function based on load power prediction information, production energy prediction information, and state of charge, under the constraints of reverse power on the user side, absorption rate at the energy production end, and charging and discharging loss at the energy storage end.

[0161] The adjustment module 505 is also used to generate a control strategy based on the objective optimization function to dynamically adjust the operating parameters of the load, energy storage end, and energy production end.

[0162] Each module in the above-mentioned device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0163] Figure 6 A schematic diagram of the structure of the electronic device provided in this application. Figure 6 As shown, the electronic device 600 provided in this embodiment includes at least one processor 601 and a memory 602. Optionally, the device 600 further includes a communication component 603. The processor 601, memory 602, and communication component 603 are connected via a bus 604.

[0164] In a specific implementation, at least one processor 601 executes computer execution instructions stored in memory 602, causing at least one processor 601 to perform the above-described method.

[0165] The specific implementation process of processor 601 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0166] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0167] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0168] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0169] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0170] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0171] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0172] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0173] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0174] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0175] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0176] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0177] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0178] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A coordinated control method applied to a multi-source heterogeneous energy system, the method is applied to a distributed photovoltaic grid-connected scene, characterized in that, The multi-source heterogeneous energy system includes a grid side, a user-side load, at least one energy storage terminal, and at least one energy production terminal; The method includes: Obtain the common connection point between the power grid side and the user side, and the electrical parameters corresponding to at least one branch where the load is located on the user side; Harmonic and transient impact characteristics are extracted from the electrical parameters; the transient impact characteristics include current surge, rate of change, and transient duration. The harmonic features and transient impact features are input into a pre-trained extraction model to extract nonlinear load features, and the current load state on the user side is determined based on the extracted nonlinear load features. The current load status and past load status of the user side are input into the pre-trained time series prediction model. The load includes at least continuous load, intermittent load and fluctuating load. By performing trend fitting based on the current and past temporal characteristics of the user side, the power fluctuation information corresponding to the continuous load is obtained as the load power prediction information. Based on the time characteristics and environmental parameters corresponding to the current time-series characteristics of the user side, and the time characteristics and environmental parameters corresponding to the past time-series characteristics of the user side, the start and stop times and operating power corresponding to the intermittent load are obtained as the load power prediction information. Based on the time period characteristics corresponding to the current time-series characteristics on the user side and the time period characteristics corresponding to the past time-series characteristics, the power demand corresponding to the fluctuating load is obtained as the load power prediction information. Under the constraints of reverse power on the user side, absorption rate at the energy production end, and charging and discharging loss at the energy storage end, an objective optimization function is constructed based on the load power prediction information, the energy production prediction information at the energy production end, and the state of charge at the energy storage end. Based on the objective optimization function, a control strategy is generated to dynamically adjust the operating parameters of the load, the energy storage terminal, and the energy production terminal, so as to achieve anti-backflow control of the multi-source heterogeneous energy system.

2. The method of claim 1, wherein, The common connection point and at least one branch where the load is located are equipped with a non-intrusive data acquisition device; The acquisition of the common connection point between the power grid side and the user side, and the electrical parameters corresponding to at least one of the branches where the load is located on the user side, includes: The non-intrusive acquisition device is used to acquire the electrical parameters of the common connection point between the power grid side and the user side, and the branch where at least one of the loads is located on the user side.

3. The method of claim 1, wherein, The common connection point and at least one load branch on the user side are equipped with a non-intrusive data acquisition device. After dynamically adjusting the operating parameters of the load, the energy storage terminal, and the energy production terminal, the method further includes: Using the non-intrusive acquisition device, electrical parameters corresponding to the common connection point between the power grid side and the user side, and at least one branch of the load on the user side are obtained to determine the actual load status of the user side at the current moment. If the deviation between the actual load status on the user side and the load power prediction information reaches a preset threshold, the time-series prediction model is adjusted based on the actual load status on the user side at the current moment and the corresponding load power prediction information.

4. A coordination control device applied to a multi-source heterogeneous energy system, the device being used to implement the coordination control method applied to the multi-source heterogeneous energy system according to any one of claims 1-3, characterized in that, The multi-source heterogeneous energy system includes a grid side, a user-side load, at least one energy storage terminal, and at least one energy production terminal; The device includes: The first acquisition module is used to acquire the common connection point between the power grid side and the user side, and the electrical parameters corresponding to at least one branch where the load is located on the user side; The determination module is used to determine the current load status on the user side based on the electrical parameters; The input module is used to input the current load status of the user side and the past load status of the user side into the pre-trained time series prediction model. The load includes at least continuous load, intermittent load and fluctuating load. The second acquisition module is used to perform trend fitting based on the current temporal characteristics of the user side and the past temporal characteristics of the user side to obtain the power fluctuation information corresponding to the continuous load, as load power prediction information. The second acquisition module is used to acquire the start and stop times and operating power corresponding to the intermittent load based on the time characteristics and environmental parameters corresponding to the current time-series characteristics of the user side and the time characteristics and environmental parameters corresponding to the past time-series characteristics of the user side, as the load power prediction information. The second acquisition module is used to acquire the power demand corresponding to the fluctuating load based on the time period characteristics corresponding to the current time-series characteristics of the user side and the time period characteristics corresponding to the past time-series characteristics, as the load power prediction information. The adjustment module is used to dynamically adjust the operating parameters of the load, the energy storage end, and the energy production end based on the load power prediction information, the energy production prediction information of the energy production end, and the state of charge of the energy storage end, so as to realize the anti-backflow control of the multi-source heterogeneous energy system.

5. A multi-source heterogeneous energy system, characterized in that, This includes loads on the grid side and the user side, at least one energy storage terminal, and at least one energy production terminal; The multi-source heterogeneous energy system is configured to employ the coordinated control method as described in any one of claims 1-3 to obtain the common connection point between the grid side and the user side, and the electrical parameters corresponding to at least one branch where the load is located on the user side; based on the electrical parameters, determine the current load state on the user side; input the current load state and the past load state on the user side into a pre-trained time-series prediction model, wherein the load includes at least continuous load, intermittent load, and fluctuating load; perform trend fitting based on the current time-series characteristics and the past time-series characteristics on the user side to obtain the power fluctuation information corresponding to the continuous load, as load power prediction information; based on the current time-series characteristics on the user side... The start-stop times and operating power of the intermittent load are obtained based on the time characteristics and environmental parameters corresponding to the sequential characteristics, as well as the time characteristics and environmental parameters corresponding to the past sequential characteristics on the user side, and are used as the load power prediction information. Based on the time period characteristics corresponding to the current sequential characteristics on the user side and the time period characteristics corresponding to the past sequential characteristics, the power demand corresponding to the fluctuating load is obtained as the load power prediction information. Based on the load power prediction information, the energy production prediction information of the energy production end, and the state of charge of the energy storage end, the operating parameters of the load, the energy storage end, and the energy production end are dynamically adjusted to realize the anti-reverse flow control of the multi-source heterogeneous energy system.

6. A computer-readable storage medium having stored thereon a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 3.

7. A computer program product comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 3.

Images