A distributed energy system operation and control method based on multi-energy storage collaboration

Abstract

A distributed energy system operation and control method with multi-element energy storage coordination is proposed. The method takes into account the optimized scheduling strategy for the coordinated control of multi-energy load curves by power-energy storage, heat storage, and cold storage technologies, and constructs a generalized technical model of energy equipment and system operation constraints. A preprocessing mechanism and weight determination model for energy system operation data under a given historical window are designed. A binary dynamic regression-based energy system scheduling sequence for coordinating the operation of electric and thermal units is proposed, and an economic operation planning model is constructed. By applying the model to the historical window, a heuristic algorithm is used to search and determine the optimal parameters of key control nodes to guide future time window scheduling, completing the scheduling closed loop. The method hyperparameters, such as historical and future window horizons, time nodes, and weights, are determined based on multiple scheduling closed-loop behaviors, completing the method closed loop. This method does not require a commercial solver and directly mines the laws of economic operation of distributed systems from historical operations to determine the switching and output plans of various system devices in future time series, facilitating engineering applications.

Description

Technical Field

[0001] The present invention relates to the technical field of energy system regulation and control, and in particular to an operation regulation and optimization strategy for a distributed power supply. Background Art

[0002] Developing distributed energy systems with a high proportion of integrated renewable energy enables distributed consumption of renewable energy and coordinates the flexible storage and conversion of multiple energy sources to address the intermittent and fluctuating nature of renewable energy and loads, thereby improving the cleanliness, reliability, affordability, and energy efficiency of energy supply. However, the various devices that comprise distributed systems have varying technical characteristics and complex coupling relationships with each other, and the systems often operate under uncertain source-load fluctuations. This poses significant challenges to system energy management, and energy management strategies directly determine the actual achievable system performance.

[0003] Most existing distributed energy system management strategies lack a holistic view of source-load-storage synergy. Although some strategies consider the development of forecasting technologies for uncertain factors such as sources and loads to achieve forward-looking system control, data forecasts are unreliable, and forecast errors themselves interfere with system operation. The contradiction of supply and demand mismatch in uncertain operating environments is becoming increasingly prominent. Given that time series data has a high correlation in temporal distribution and that energy storage devices can coordinate the temporal correlation operation of the system, it is of great practical value to establish a data mining mechanism based on the active and coordinated energy storage and release of multi-element energy storage systems to dynamically obtain energy storage system operation plans that maximize distributed system performance. This, in turn, determines the overall system operation plan. This can avoid the influence of uncertain factors, achieve coordinated operation that maximizes system resource utilization, and improve the system's adaptability to renewable energy.

[0004] Chinese patent application number CN 202311526205.9, entitled "A global optimization method and system for energy-saving operation of a distributed energy system", discloses a technical solution, which includes the following steps: comprehensively considering the active loss and reactive loss of the distributed energy system, constructing a first objective function with network loss optimization and a constraint condition of real-time power flow balance; based on the first objective function, statically reconstructing the distributed energy system to obtain the optimal switching state of the distributed energy system; taking the maximization of the active output of the distributed energy system as the second objective function, based on the optimal switching state of the distributed energy system, calculating the optimal output composition of the distributed energy system; taking the minimization of the reactive output of the distributed energy system as the third objective function, based on the optimal switching state and output composition of the distributed energy system, calculating the optimal installation quantity and operating capacity of the distributed energy system reactive compensation equipment. This patent can realize the economical operation of distributed energy and reduce network operation losses. However, the patent document uses a particle swarm algorithm to statically reconstruct the distributed energy system and optimize the switching state of the distributed power system to obtain the optimal state of the system switch and ensure the overall stability of the system. However, it does not take into account the historical cyclical operation laws of the system, nor does it consider the coordinated regulation between source, load and storage, and fully utilize the ability of the multi-energy storage system to adjust the timing of multiple energy loads and coordinate the operation between source and load, thereby making its regulation biased. Summary of the Invention

[0005] To address the deficiencies of the prior art, the present invention provides a method for operating and controlling a distributed energy system based on the synergy of multiple energy storage systems. The method directly mines the economic and technical operating laws of the system from historical operating data, determines future system switches and output configurations, and improves the system's adaptability to its integrated variable renewable energy sources and multi-energy loads, as well as the economic efficiency of its operation, while fully tapping the synergy potential and technical advantages of energy storage.

[0006] The purpose of the present invention is achieved through the following technical solutions:

[0007] A method for operating and controlling a distributed energy system with multi-energy storage coordination includes the following steps:

[0008] S1: Construct a generalized technical model of energy equipment and system operation constraints;

[0009] S2: Collect historical operation data of the energy system and design a pre-processing mechanism and a historical scenario weight distribution model;

[0010] S3: Propose an energy system scheduling sequence and model for coordinating the operation of electric heating units based on the operation instructions of electric cooling / heating equipment, dynamically adjusting at the midpoint between the upper and lower bounds to avoid energy redundancy;

[0011] S4: Build a distributed system economic operation planning model and determine the key parameters of key control nodes;

[0012] S5: Optimize the algorithm for historical scenarios based on the scheduling sequence and model to determine the optimal parameters for control;

[0013] S6: Determine the system switching and output plan for the future window based on the scheduling sequence and optimal parameters, completing the scheduling closed loop;

[0014] S7: Based on the operating data of multiple scheduling closed loops obtained by rolling execution of S2 to S6, a scheduling model hyperparameter search method is constructed. The hyperparameters include parameters such as the window matrix field of view, the number of window inclusion cycles, the time resolution combination, and the window weight distribution, completing the method closed loop.

[0015] In the above-mentioned multi-energy storage coordinated distributed energy system operation control method, in step S1, the generalized technical model includes a power link, a multi-energy conversion link, and a multi-energy storage link, wherein:

[0016] 1) Power link

[0017] The power links are divided into cogeneration and non-cogeneration. Typical cogeneration types include gas turbines or internal combustion engines and their heat recovery components, PVT; non-cogeneration types include photovoltaics, wind turbines, solar collectors, boilers and other equipment that only produces electricity or heat. The model is:

[0018]

[0019]

[0020] Where: OUP and INP represent the energy output and input power of the equipment, respectively; the subscript k / m represents the cogeneration / non-cogeneration equipment; K and M are the sets of corresponding equipment in the distributed system; and the superscript t indicates time. is the power generation / heating efficiency of the equipment under rated conditions, Fitting model for the performance coefficient of power generation / heating equipment, in the factor set FS affecting equipment performance m The performance coefficient is obtained under the input of γ; k / m Is a Boolean variable that identifies the start and stop of the system. are the upper and lower limits of the equipment's output, and Provide up and down climbing constraints for equipment output;

[0021] 2) Multi-energy conversion link

[0022] Typical equipment in the conversion link includes an electrothermal driven heat pump to convert multiple energy flows from the power link and adjust energy distribution. Its model is as follows:

[0023]

[0024]

[0025]

[0026] Where: O is the set of multi-energy conversion link equipment, and the subscripts etc / eth / htc / … respectively represent the working conditions of electricity-to-heat / electricity-to-cooling / heat-to-cooling / …;

[0027] 3) Multi-energy storage link

[0028] Typical equipment in the multi-energy storage link includes power storage, energy storage, heat storage, and cold storage technologies. The energy storage mentioned above includes chemical batteries, compressed air energy storage, etc. The same thermal storage technology can achieve heat storage or cold storage by adjusting the medium and its operating temperature, and can be switched in a planned manner. The model is as follows:

[0029]

[0030]

[0031]

[0032] Where: L is the collection of energy storage devices, STH / SE is the heat storage / capacity storage of energy storage devices, DIS / CHA represents the discharge / charge power, and the subscripts pis / eis identify power-type / energy-type storage devices. It is a Boolean variable, indicating that the energy storage device has only one function at the same time, and ΔT is the duration or step size.

[0033] In addition, the system's operating constraints include the cooling-heating-electricity balance, which is expressed as follows:

[0034]

[0035] Where: γ o A Boolean variable indicating that the energy from the electric heat conversion device cannot drive the heat-to-cooling device to provide cooling.

[0036] In the above-mentioned multi-energy storage coordinated distributed energy system operation control method, in step S2, the data preprocessing mechanism and historical scenario weight distribution include:

[0037] S2-1) The data preprocessing mechanism is to construct a historical window matrix, continuously monitor and collect high-time resolution data of the system's historical operation, the data of which is several seconds or tens of seconds, and construct three matrices with dual time scales:

[0038] S2-2. Assign weights to each period in the historical window: As the time interval between two periods increases, the correlation between the historical operation period and the future operation period becomes weaker, and the reference value for the scheduling plan of the future operation window becomes weaker. Therefore, the following model is constructed to assign weights:

[0039]

[0040] Where: is the weight quota of the i-th historical scheduling period (corresponding to the i-th row of the matrix in step S2-1), and w i is the final weight value, and θ is the control parameter of the weight quota model. By adjusting θ, the allocation ratio of the weight quota in Z historical scheduling cycles can be controlled.

[0041] The above-mentioned multi-energy storage coordinated distributed energy system operation and control method, in the said step S3, proposes an energy system scheduling sequence for coordinating the operation of electric heating units based on the operation instructions of the electric-to-cold / heat equipment, dynamic adjustment at the midpoint between the upper and lower bounds, and avoiding energy redundancy through binary dynamic regression. The sequence includes two parts corresponding to the dual time scale in step S2, namely, any time step with low time resolution and any random fluctuation event with high time resolution are processed as the subject respectively.

[0042] In the above-mentioned multi-energy storage coordinated distributed energy system operation control method, in step S4, the economic operation planning model constructed includes an objective function, a constraint function, and a decision variable set composed of key control parameters, which are listed as follows:

[0043] 1) Planning model:

[0044] The optimization model of distributed energy system is expressed as:

[0045]

[0046] fee∈[fuel, electricity purchase and sale, equipment operation and maintenance, emissions, …](29)

[0047] Where: the operating costs of the system include the purchase of fuel generated during the scheduling cycle, the purchase and sale of electricity from the power grid, the operation and maintenance of equipment operation, and emissions such as carbon emissions. Penalty costs will also be considered to adjust the scheduling direction: such as wind and solar power curtailment and energy efficiency. The constraints of the model include the equation part h(DEC) = 0 and the inequality part g(DEC) < 0, which are mainly the equipment technology model and system energy flow balance model constructed in step S1. Where DEC is the decision variable set.

[0048] 2) Decision variables

[0049] Referring to step S3, the decision variables mainly include the critical startup threshold of the power generation unit in a scheduling cycle, the charging and discharging plan of the energy storage and cold and hot energy storage equipment at each time step in the cycle, and the three key state points of the power storage equipment at each time step in the cycle:

[0050]

[0051] In step S5, the operation planning model constructed in step S4 is applied to the three matrices under the dual time scale constructed in step S2, and the values ​​of a set of key control parameters (Equation (30)) are determined by a heuristic search algorithm such as genetic or particle swarm. After the system sequentially controls the operation of each scheduling period system in the historical window according to the set of control parameters and the strategy described in step S3, the weighted cumulative sum of the operating costs of each scheduling period is minimized as shown in Equation (28).

[0052] In the aforementioned multi-element energy storage coordinated distributed energy system operation control method, in step S6, the optimal control parameters obtained in step S5 are combined with the strategy described in step S4 to guide system operation in a future scheduling cycle, which can be considered as i = 0. Next, the features and methods described in steps S2 to S6 are repeated to guide system operation in the next future scheduling cycle, cyclically rolling to form a closed scheduling loop.

[0053] In the above-mentioned multi-energy storage coordinated distributed energy system operation and control method, in step S7, when the number of scheduling closed-loop scrolls reaches a threshold, a combination of factors not limited to the step configuration of the scheduling cycle, the number of scheduling cycles that constitute the historical window, the parameters of the weight model, and the scheduling time scale are selected as decision variables. Combined with the optimization of steps S2 to S6, a further layer of search optimization is performed with the goal of minimizing the sum of the system operating costs of the i=0 row in each scroll, and the hyperparameters of the control method are updated in time to adapt to the latest operating environment of the system, forming a method closed loop.

[0054] Beneficial effects

[0055] Compared with the prior art methods, the beneficial effects and advantages of the present invention are:

[0056] 1. This invention establishes a source-load-storage coordinated control framework, taking the active storage and release of energy by the multi-element energy storage system as its starting point. It fully leverages the multi-element energy storage system's ability to sequentially adjust multiple energy loads and coordinate source-load operation, improving the utilization rate of each system's equipment and its adaptability to renewable energy. It can guide the formulation of distributed energy system operation plans.

[0057] 2. The present invention relies on a constructed generalized distributed energy system model to extract the periodic operation rules of system equipment from the historical operation data of the system in the recent several cycles, and uses them for system scheduling in a cycle in the near future. It also timely nests the outer layer optimization to determine the hyperparameters of the method. Through the dual closed loop of system operation and control methods, the proposed method has wide adaptability and strong and flexible system management capabilities.

[0058] The present invention is further described below in conjunction with the accompanying drawings and examples. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 Schematic diagram of the operation and control process of multi-energy storage coordination. DETAILED DESCRIPTION

[0060] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0061] The present invention provides an operation control and optimization strategy based on the collaboration of multiple energy storage systems, which helps to extract the periodic operation laws of system equipment from historical operation data and use them for system scheduling in the near future. It also timely adjusts the window matrix field of view, the number of window periods, the time resolution combination, the window weight distribution and other hyperparameters of the scheduling model to form a double closed loop of system operation and control methods.

[0062] The present invention comprises the steps of:

[0063] S1: Construct a generalized technical model of energy equipment and system operation constraints;

[0064] S2: Collect historical operation data of the energy system and design a pre-processing mechanism and a historical scenario weight distribution model;

[0065] S3: Propose an energy system scheduling sequence and model for coordinating the operation of electric heating units based on the operation instructions of electric cooling / heating equipment, dynamically adjusting at the midpoint between the upper and lower bounds to avoid energy redundancy;

[0066] S4: Build a distributed system economic operation planning model and determine the key parameters of key control nodes;

[0067] S5: Optimize the algorithm for historical scenarios based on the scheduling sequence and model to determine the optimal parameters for control;

[0068] S6: Determine the system switching and output plan for the future window based on the scheduling sequence and optimal parameters, completing the scheduling closed loop;

[0069] S7: Based on the operating data of multiple scheduling closed loops obtained by rolling execution of S2 to S6, a scheduling model hyperparameter search method is constructed. The hyperparameters include parameters such as the window matrix field of view, the number of window inclusion cycles, the time resolution combination, and the window weight distribution, completing the method closed loop.

[0070] In step S1, the generalized technical model includes a power link, a multi-energy conversion link, and a multi-energy storage link:

[0071] In step S1, the generalized technical model includes a power link, a multi-energy conversion link, and a multi-energy storage link, which are described as follows:

[0072] 1) Power link

[0073] The power links are divided into cogeneration and non-cogeneration. Typical cogeneration types include gas turbines or internal combustion engines and their heat recovery components, PVT; non-cogeneration types include photovoltaics, wind turbines, solar collectors, boilers and other equipment that only produces electricity or heat. The model is:

[0074]

[0075]

[0076] Where: OUP and INP represent the energy output and input power of the equipment, respectively; the subscript k / m represents the cogeneration / non-cogeneration equipment; K and M are the sets of corresponding equipment in the distributed system; and the superscript t indicates time. is the power generation / heating efficiency of the equipment under rated conditions, Fitting model for the performance coefficient of power generation / heating equipment, in the factor set FS affecting equipment performance m The performance coefficient is obtained under the input of γ; k / m Is a Boolean variable that identifies the start and stop of the system. are the upper and lower limits of the equipment's output, and Provide up and down climbing constraints for equipment output;

[0077] 2) Multi-energy conversion link

[0078] Typical equipment in the conversion link includes an electrothermal driven heat pump to convert multiple energy flows from the power link and adjust energy distribution. Its model is as follows:

[0079]

[0080]

[0081] Where: O is the set of multi-energy conversion link equipment, and the subscripts etc / eth / htc / … respectively represent the working conditions of electricity-to-heat / electricity-to-cooling / heat-to-cooling / …;

[0082] 3) Multi-energy storage link

[0083] Typical equipment in the multi-energy storage link includes power storage, energy storage, heat storage, and cold storage technologies. The energy storage mentioned above includes chemical batteries, compressed air energy storage, etc. The same thermal storage technology can achieve heat storage or cold storage by adjusting the medium and its operating temperature, and can be switched in a planned manner. The model is as follows:

[0084]

[0085]

[0086] Where: L is the collection of energy storage devices, STH / SE is the heat storage / capacity storage of energy storage devices, DIS / CHA represents the discharge / charge power, and the subscripts pis / eis identify power-type / energy-type storage devices. It is a Boolean variable, indicating that the energy storage device has only one function at the same time, and ΔT is the duration or step size.

[0087] In addition, the system's operating constraints include the cooling-heating-electricity balance, which is expressed as follows:

[0088]

[0089] Where: γ o A Boolean variable indicating that the energy from the electric heat conversion device cannot drive the heat-to-cooling device to provide cooling.

[0090] In step S2, the data preprocessing mechanism and historical scene weight distribution include:

[0091] S2-1) The data preprocessing mechanism is to construct a historical window matrix, continuously monitor and collect high-time resolution data of the system's historical operation, the data of which is several seconds or tens of seconds, and construct the following three matrices with dual time scales:

[0092]

[0093]

[0094]

[0095] Where: the superscript renew represents the energy output of renewable energy sources such as wind and solar that are difficult to control. and are the net value matrices of electricity and heating / cooling loads at low time resolutions (e.g., a few minutes, more than ten minutes, or tens of minutes), respectively. The number of columns H represents the period of the historical window, and the number of rows Z represents the number of periods that constitute the historical window. is the nth monitoring data covered by the i-th time step in the j-th cycle, and In order to monitor the mean value of the net value data matrix, this operation filters out random fluctuations to a certain extent and retains the time series coupling characteristics. For net cooling and heating loads with large inertia, high-frequency quantities can no longer be retained, and the operation plan can be specified based on the low-frequency matrix alone. For electric energy that requires instantaneous balance, a high-frequency random fluctuation matrix with a corresponding high time resolution (such as seconds or tens of seconds) is constructed.

[0096] S2-2) Assign weights to each period in the historical window: As the time interval between two periods increases, the correlation between the historical operation period and the future operation period becomes weaker, and the reference value for the scheduling plan of the future operation window becomes weaker. Therefore, the following model is constructed to assign weights:

[0097]

[0098] Where: is the weight quota of the i-th historical scheduling period (corresponding to the i-th row of the matrix in step S2-1), and w i is the final weight value, and θ is the control parameter of the weighted quota model. By adjusting θ, we can control the proportion of weighted quotas allocated over Z historical scheduling cycles. We can also adjust the trend and magnitude of weighted quota changes as the number of historical cycles increases, for example, to reduce the quota quickly first and then slowly, or vice versa.

[0099] In step S3, an energy system scheduling sequence is proposed to coordinate the operation of electric heating units based on the operation instructions of the electric cooling / heating equipment and the dynamic adjustment at the midpoint between the upper and lower bounds to avoid energy redundancy. The sequence includes two parts corresponding to the dual time scale in step S2, which are detailed as follows:

[0100] 1) Any time step with low temporal resolution:

[0101] Step S3-1: Set the upper and lower critical states of the power storage at time step t and target state If the state of this type of device at time step t Reaching the upper critical state Then To determine the discharge power position for the discharge target If you touch the Nether The same logic is used to determine the charging power position Otherwise, the charge and discharge power occupancy is 0. The occupancy means that within the maximum charge and discharge power range of the device, Adjust to The charge and discharge power to be performed.

[0102] Step S3-2: Superimpose the charge and discharge occupancy of the power storage device and the charge and discharge plans of other energy storage devices on the net electric and thermal load to adjust the net load curve:

[0103]

[0104] Step S3-3: According to the technical sequence of electricity to cooling and heat to cooling, the energy and heat required to meet the cooling load are allocated to the multi-energy conversion equipment in turn, and the energy and heat required to meet the cooling load are added to the equipment. and

[0105] Step S3-4: According to the technical sequence of cogeneration power equipment (initial value is set to 0), electricity-to-heat equipment, and heat-only power equipment, the power required for heating is determined and added to

[0106] Step S3-5: Setting the critical grid electricity price for starting the cogeneration / non-cogeneration power generation equipment, This is just a preference, and other similar thresholds can also be set. Determine the heat output of the cogeneration power equipment according to the technical sequence of cogeneration / non-cogeneration power equipment (the former is preferred in the cogeneration scenario, otherwise the latter), power grid, etc. In addition, when the electricity price is lower than The power equipment is restricted to the off state when If it is less than 0, the discharge of energy storage equipment will be weakened and / or the charging power will be increased, and the future discharge plan of such equipment will be changed in conjunction with it, and the adjustment amount will be evenly distributed to the remaining time steps of the scheduling period.

[0107] Step S3-6: For the electric heat / cooling device, in a scenario with a relatively large heat load, such as winter, the heating operation is carried out, and conversely, the cooling operation is carried out. Energy redundancy is inevitable in steps 4.3 to 4.5. If there is heat redundancy, the operating power of the electric heat / cooling device determined in steps 4.3 and 4.4 is used as the upper limit, and the shutdown state is used as the lower limit. The half point of the upper and lower limits, i.e., the mean, is used to define the output power of the electric heat / cooling device, and steps 4.3, 4.5, and 4.4 are executed. If there is still heat redundancy, the half point is set as the upper limit. If only heat power is provided, the power supply is set to the upper limit. If the output of the device increases, the bisection point is set as the lower bound, the bisection point is recalculated, and steps 4.3, 4.5, and 4.4 are executed again; the iteration is repeated until convergence is achieved; the convergence condition is: the heat redundancy or the output increment of the heating power device is less than the threshold, or the electric heat / cooling device is shut down; if the device is shut down and there is heat redundancy, the energy discharge of the heat storage / cooling device at the current time step t is reduced and / or the charging power is increased, and the future energy discharge plan of this type of equipment is adjusted accordingly, and the adjustment amount is evenly distributed among the remaining time steps of the scheduling period (t~H).

[0108] Step S3-7: Through the above steps, the operation plan of each device in the distributed system can be obtained for renewable energy output, energy demand, and energy price scenarios at any time step, and the system enters the stage of smoothing out high-frequency random fluctuations in electricity.

[0109] 2) Any random fluctuation event with high time resolution:

[0110] Step S3-8: Based on the power occupancy of the power type storage device determined by the low time resolution scheduling, that is, the power type device state determined in step S3-1 is determined according to the nth fluctuation monitoring event at the tth time step, j=t; pull high or lower Limits of the power curve:

[0111]

[0112] Where: and is the maximum charge and discharge power of the power storage device under the current monitoring event, which can be calculated according to formulas (8)--(12);

[0113] Step S3-9: Raise the electric power within the limit of the power storage device to suppress the downward fluctuation of electric energy, or lower the electric power to suppress the upward fluctuation. The random fluctuation exceeding the limit is distributed to the energy storage device or the power grid in sequence.

[0114] In step S4, the constructed economic operation planning model includes three parts: objective function, constraint function and decision variable set consisting of key control parameters. The details are as follows:

[0115] 1) Planning model:

[0116] The optimization model of distributed energy system is expressed as:

[0117]

[0118] fee∈[fuel, electricity purchase and sale, equipment operation and maintenance, emissions, …](29)

[0119] Where: the operating costs of the system include the purchase of fuel generated during the scheduling cycle, the purchase and sale of electricity from the power grid, the operation and maintenance of equipment operation, and emissions such as carbon emissions. Penalty costs will also be considered to adjust the scheduling direction: such as wind and solar power curtailment and energy efficiency. The constraints of the model include the equation part h(DEC) = 0 and the inequality part g(DEC) < 0, which are mainly the equipment technology model and system energy flow balance model constructed in step S1. Where DEC is the decision variable set.

[0120] 2) Decision variables

[0121] Referring to step S3, the decision variables mainly include the critical startup threshold of the power generation unit in a scheduling cycle, the charging and discharging plan of the energy storage and cold and hot energy storage equipment at each time step in the cycle, and the three key state points of the power storage equipment at each time step in the cycle:

[0122]

[0123] In step S5, the operation planning model constructed in step S4 is applied to the three matrices under the dual time scale constructed in step S2, and the values ​​of a set of key control parameters (Equation (30)) are determined by a heuristic search algorithm such as genetic or particle swarm. After the system sequentially controls the operation of each scheduling period system in the historical window according to the set of control parameters and the strategy described in step S3, the weighted cumulative sum of the operating costs of each scheduling period is minimized as shown in Equation (28).

[0124] In step S6, the optimal control parameters obtained in step S5 are combined with the strategy described in step S4 to guide system operation in a future scheduling cycle, which can be regarded as i = 0. Then, the features and methods described in steps S2 to S6 are repeated to guide system operation in the next future scheduling cycle, and the cycle is repeated to form a closed scheduling loop.

[0125] In step S7, when the number of scheduling closed-loop scrolling times reaches a threshold, a combination of factors not limited to the step configuration of the scheduling cycle, the number of scheduling cycles constituting the historical window, the parameters of the weight model, and the scheduling time scale are selected as decision variables. Combined with the optimization of steps S2 to S6, a further layer of search optimization is performed with the goal of minimizing the sum of the system operating costs of the i=0 row in each scrolling, and the hyperparameters of the control method are updated in time to adapt to the latest operating environment of the system, thereby forming a method closed loop.

[0126] In summary, the operation control and optimization strategy based on the coordination of multiple energy storage provided by the present invention helps to extract the periodic operation rules of system equipment from the periodic historical operation data of the system and use them for system scheduling in the near future. It also embeds an optimization layer to timely adjust the hyperparameters of the method, forming a double closed loop of system operation and control methods, and improving the adaptability and matching of the methods.

Claims

1. A distributed energy system operation control method based on multi-energy storage collaboration, characterized in that: The steps include: S1: Construct a generalized technical model of energy equipment and system operation constraints, wherein the generalized technical model includes a power link, a multi-energy conversion link, and a multi-energy storage link; S2: Collect historical operation data of the energy system and design a pre-processing mechanism and a historical scenario weight distribution model; S3: Propose an energy system scheduling sequence and model for coordinating the operation of electric heating units based on the operation instructions of electric cooling / heating equipment, dynamically adjusting at the midpoint between the upper and lower bounds to avoid energy redundancy; S4: Build a distributed system economic operation planning model and determine the key parameters of key control nodes; S5: Optimize the algorithm for historical scenarios based on the scheduling sequence and model to determine the optimal parameters for control; S6: Determine the system switching and output plan for the future window based on the scheduling sequence and optimal parameters, completing the scheduling closed loop; S7: Based on the operating data of multiple scheduling closed loops obtained by rolling execution of S2 to S6, a scheduling model hyperparameter search method is constructed. The hyperparameters include parameters such as the window matrix field of view, the number of window inclusion cycles, the time resolution combination, and the window weight distribution, completing the method closed loop.

2. The distributed energy system operation control method based on multi-energy storage collaboration according to claim 1 is characterized in that: In step S1, the generalized technical model includes a power link, a multi-energy conversion link, and a multi-energy storage link, wherein: 1) Power link The power links are divided into cogeneration and non-cogeneration. Typical cogeneration types include gas turbines or internal combustion engines and their heat recovery components, PVT; non-cogeneration types include photovoltaics, wind turbines, solar collectors, boilers and other equipment that only produces electricity or heat. The model is: ； Where: OUP and INP Represent the energy output and input power of the equipment, respectively. k / m Indicates co-generation / non-co-generation equipment, while K and M are the sets of corresponding equipment in the distributed system. t Mark time; is the power generation / heating efficiency of the equipment under rated conditions, Fitting model for the performance coefficient of power generation / heating equipment, in the set of factors affecting equipment performance The performance coefficient is obtained under the input of; Is a Boolean variable that identifies the start and stop of the system. are the upper and lower limits of the equipment's output, and Provide up and down climbing constraints for equipment output; 2) Multi-energy conversion link Typical equipment in the conversion link includes an electrothermal driven heat pump to convert multiple energy flows from the power link and adjust energy distribution. Its model is as follows: ； Where: O is the collection of multi-energy conversion link equipment, subscript etc / eth / htc / respectively represent the operating conditions of electricity-to-heat / electricity-to-cooling / heat-to-cooling; 3) Multi-energy storage link Typical equipment in the multi-energy storage link includes power storage, energy storage, heat storage, and cold storage technologies. The energy storage mentioned above includes chemical batteries and compressed air energy storage. The same thermal storage technology can achieve heat storage or cold storage due to the adjustment of the medium and its operating temperature, and can be switched in a planned manner. The model is as follows: ； Where: L is the collection of energy storage devices, STH / SE The heat storage / electricity storage of energy storage equipment, DIS / CHA Indicates the discharge / charge power, subscript pis / eis Identify power / energy storage devices, It is a Boolean variable, indicating that the energy storage device has only one function at the same time. is the duration or step length; In addition, the system's operating constraints include the cooling-heating-electricity balance, which is expressed as follows: ； Where: A Boolean variable indicating that the energy from the electric heat conversion device cannot drive the heat-to-cooling device to provide cooling.

3. The distributed energy system operation control method based on multi-energy storage collaboration according to claim 1 is characterized in that: In step S2, the data preprocessing mechanism and historical scene weight distribution include: S2-1) The data preprocessing mechanism is to build a historical window matrix, continuously monitor and collect high-time resolution data of the system's historical operation, whose data is several seconds or tens of seconds, and build three matrices with dual time scales: S2-2. Assign weights to each period in the historical window: As the time interval between two periods increases, the correlation between the historical operation period and the future operation period becomes weaker, and the reference value for the scheduling plan of the future operation window becomes weaker. Therefore, the following model is constructed to assign weights: ； Where: For the i Historical scheduling period (corresponding to the matrix in step S2-1 i row), and is the final weight value, is the control parameter of the weight quota model, by adjusting , which can realize the control of the allocation ratio of weight quota in Z historical scheduling cycles.

4. The distributed energy system operation control method based on multi-energy storage collaboration according to claim 1 is characterized in that: In step S3, an energy system scheduling sequence is proposed for coordinating the operation of electric heating units based on the operation instructions of electric-to-cold / heat equipment, dynamic adjustment at the midpoint between the upper and lower bounds, and avoiding energy redundancy, including two parts corresponding to the dual time scale in step S2, namely, any time step with low time resolution and any random fluctuation event with high time resolution are processed as subjects respectively.

5. The distributed energy system operation control method based on multi-energy storage coordination according to claim 1 is characterized in that: In step S4, the economic operation planning model constructed includes an objective function, a constraint function, and a decision variable set consisting of key control parameters, which are listed as follows: 1) Planning model: The optimization model of distributed energy system is expressed as: ； Where: The operating costs of the system include fuel purchases generated during the dispatch cycle, electricity purchases and sales from the grid, equipment operation and maintenance, and adjustments to the dispatch direction based on carbon emission costs and penalty costs, including wind / solar curtailment and energy efficiency adjustments. The constraints of the model include the equation part , mainly refers to the equipment technology model and system energy flow balance model constructed in step S1, where, is the set of decision variables; 2) Decision variables Referring to step S3, the decision variables mainly include the critical startup threshold of the power generation unit in a scheduling cycle, the charging and discharging plan of the energy storage and cold and hot energy storage equipment at each time step in the cycle, and the three key state points of the power storage equipment at each time step in the cycle: ； Where: Critical grid electricity price thresholds for controlling the start-up of power generation units; Can represent energy density type electricity storage, heat storage, cold storage and other energy storage equipment l At time step 1~ H Charging / discharging power; Can represent power storage devices l At time step 1~ H There are three key state points: upper critical state, target state, and lower critical state; In step S5, the operation planning model constructed in step S4 is applied to the three matrices under the dual time scale constructed in step S2, and the values ​​of a set of key control parameters (Equation (30)) are determined by a heuristic search algorithm such as genetic or particle swarm. After the system sequentially controls the operation of each scheduling period system in the historical window according to the set of control parameters and the strategy in step S3, the weighted cumulative sum of the operating costs of each scheduling period is minimized, as shown in Formula (28).

6. The distributed energy system operation control method based on multi-energy storage coordination according to claim 1 is characterized in that: In step S6, the optimal control parameters obtained in step S5 are combined with the strategy in step S4 to guide the system operation in a future scheduling cycle, which can be regarded as i =0; then, repeat the features and methods described in steps S2 to S6 to guide the system operation in the next future scheduling cycle, and roll the cycle to form a scheduling closed loop.

7. The distributed energy system operation control method based on multi-energy storage coordination according to claim 1 is characterized in that: In step S7, when the number of scheduling closed-loop rolling times reaches the threshold, the combination of factors not limited to the step configuration of the scheduling cycle, the number of scheduling cycles constituting the historical window, the parameters of the weight model, and the scheduling time scale are selected as decision variables, and the optimization of steps S2 to S6 is combined to determine the optimal number of rolling times. i =0 rows, and conduct a further layer of search optimization with the goal of minimizing the sum of the system operating costs. The hyperparameters of the control method are updated in a timely manner to adapt to the latest operating environment of the system, forming a closed loop of the method.

Images