Comprehensive energy system model prediction control method and system

Abstract

The invention discloses an integrated energy system model prediction control method and system, and the method comprises the steps: constructing an integrated energy system physical model, and obtaining the operation state data of an integrated energy system according to the integrated energy system physical model; constructing a neural network model, performing digital processing on the running state data of the integrated energy system, and training the neural network model by adopting the processed running state data; and according to the trained neural network model, predicting a dynamic response result of the integrated energy system in finite time, and solving an optimal control sequence to realize optimal scheduling of the integrated energy system. According to the comprehensive energy system model prediction control strategy based on the physical information neural network, the optimal scheduling of the comprehensive energy system can be realized under the conditions of complex system mechanism, deep coupling, system parameter time varying and model mismatching.

Description

Technical Field

[0001] The present invention relates to the field of integrated energy technology, and in particular to an integrated energy system model predictive control method and system. Background Art

[0002] With the development of clean energy, new energy sources such as wind power and photovoltaics have penetrated the power system on a large scale. On the other hand, as the economy and society move towards a high-quality development stage, the proportion of electricity consumption in the tertiary industry and residents continues to rise. The rapid development of multi-energy systems and flexible loads has become one of the important reasons for the widening of the peak-to-valley difference of the power grid and the prominence of peak loads, which has brought huge pressure to the stable operation of the power grid and the guarantee of electricity consumption for people's livelihood, and put forward higher requirements for the allocation of various types of flexible and available resources and the absorption capacity of clean energy. Multi-type flexible resources have the characteristics of inconsistent characteristics, complex behavior patterns and strong randomness, which makes it difficult to accurately characterize their characteristics. At the same time, the number of multi-type flexible resources is huge and the model structure is complex, which significantly increases the calculation and communication pressure of the overall control of the system. The existing integrated energy system scheduling method usually adopts the rolling optimization solution method, which solves the optimal control strategy by inputting the current system operation status and combining the integrated energy system model. However, the operating conditions of the integrated energy system are complex and changeable, the traditional mechanism model cannot accurately describe the dynamic process of the system, and there is a problem of model parameter mismatch during operation. The final scheduling strategy does not make full use of system resources, and the operating efficiency of the integrated energy system still has a lot of room for improvement. During the operation of integrated energy, sensors and measurement systems of various types of components will collect and save a large amount of historical operation data, which can effectively characterize the system operation status and dynamic response process. Therefore, how to establish a more accurate system response model and construct an optimal scheduling strategy based on the existing system operation historical data and combined with the system operation mechanism is crucial to ensure the safe and efficient operation of the integrated system. Summary of the invention

[0003] The present invention provides a model predictive control method and system for an integrated energy system, which utilizes massive historical operation data of the integrated energy system in combination with a system mechanism model, constructs a dynamic response model of a complex system through a physical information neural network, and realizes efficient operation of the integrated energy system through a model predictive control method.

[0004] In order to achieve the above object, an embodiment of the present invention provides a comprehensive energy system model predictive control method, comprising: Constructing a physical model of an integrated energy system, and obtaining the operating status data of the integrated energy system according to the physical model of the integrated energy system; wherein the physical model of the integrated energy system includes a physical model of the power system, a physical model of the thermal system, and a model of an electric boiler; Constructing a neural network model, digitally processing the operating status data of the integrated energy system, and using the processed operating status data to train the neural network model; According to the trained neural network model, the dynamic response results of the integrated energy system within a certain period of time are predicted, and the optimal control sequence of the integrated energy system is solved according to the dynamic response results to achieve the optimal scheduling of the integrated energy system.

[0005] As an improvement of the above solution, the step of constructing a physical model of the integrated energy system and obtaining the operating status data of the integrated energy system according to the physical model of the integrated energy system includes: Constructing a physical model of the integrated energy system, and simulating changes in specific parameters of the integrated energy system based on the physical model of the integrated energy system to obtain operating status data of the integrated energy system when the specific parameters change; The expression of the power system physical model is: , , The expression of the physical model of the thermal system is: , The expression of the electric boiler model is: , In the formula, For the The injected active power of each node; For the said The effective value of the AC voltage of each node; is the number of nodes in the power system network; For the The effective value of the AC voltage of each node; For the said nodes and the The conductance of the line between the nodes; For the said nodes and the The voltage phase angle difference between the nodes; For the said nodes and the The susceptance of the line between the nodes; For the said The injected reactive power of each node; and They are Moment Pipeline The starting and ending temperatures; and They are Moment described pipeline The starting and ending temperatures; For the pipeline water flow; is the time difference; is the density of water; For the pipeline The cross-sectional area of For the pipeline Length; For the pipeline The heat transfer coefficient of is the specific heat capacity of water; is the ambient temperature; It is a collection of network nodes of thermal system; For the The calorific value of an electric boiler; For the said The heating efficiency of an electric boiler; For the said The electrical power of an electric boiler.

[0006] As an improvement of the above solution, the neural network model is constructed, the operation status data of the integrated energy system is digitally processed, and the processed operation status data is used to train the neural network model, including: Constructing a neural network model to digitally process the operating status data of the integrated energy system to obtain processed operating status data; splicing the processed running status data to obtain training data of the neural network model; Using the training data to train the neural network model; The expression of the training data is: , , In the formula, and They are the input and output of the neural network model training respectively; and They are Moment and The AC voltage amplitude in the power system at the moment; and Respectively Moment and The phase angle in the power system at the moment; and Respectively Moment and The node active power at the time; and Respectively Moment and The node reactive power at the time; and Respectively Moment and The node temperature value in the thermal system at the moment; and Respectively Moment and Pipe water flow in the thermal system at all times; for The vector of control variables of the integrated energy system at the moment; The transpose symbol.

[0007] As an improvement of the above scheme, the method predicts the dynamic response result of the integrated energy system within a certain period of time according to the trained neural network model, and solves the optimal control sequence of the integrated energy system according to the dynamic response result to achieve the optimal scheduling of the integrated energy system, including: Constructing the objective function and constraints of the integrated energy system; According to the objective function and constraint conditions and the dynamic response results of the integrated energy system within a certain period of time predicted by the trained neural network model, the optimal control sequence of the integrated energy system is solved to achieve the optimal scheduling of the integrated energy system.

[0008] As an improvement of the above solution, the expression of the objective function is: , The constraint condition is expressed as: , , , , , In the formula, is the objective function of the integrated energy system; is the model predictive control step size; For the The weight of state variables during step-by-step prediction; For the said The actual system operation trajectory during the step prediction; For the said The expected system operation trajectory during the step prediction; For the said The weight of the control variables in the first-step forecast; For the said The actual control input trajectory during step prediction; For the said The expected control input trajectory during step prediction; is the neural network model function; For the said The input of the neural network model during the step prediction; and The upper and lower limits of node voltage in the power system; For the The effective value of the AC voltage of each node; and are the upper and lower limits of active power of nodes in the power system; For the said The injected active power of each node; and The upper and lower limits of reactive power of nodes in the power system; For the said The injected reactive power of each node; and are the upper and lower limits of node temperature in the thermal system; For the said The temperature of a node.

[0009] In order to achieve the above object, an embodiment of the present invention provides a comprehensive energy system model predictive control system, comprising: A state data acquisition module, used to construct a physical model of the integrated energy system, and to acquire the operation state data of the integrated energy system according to the physical model of the integrated energy system; wherein the physical model of the integrated energy system includes a physical model of the power system, a physical model of the thermal system and a model of the electric boiler; A network model training module is used to construct a neural network model, digitally process the operating status data of the integrated energy system, and use the processed operating status data to train the neural network model; The control strategy prediction module is used to predict the dynamic response results of the integrated energy system within a certain period of time based on the trained neural network model, and solve the optimal control sequence of the integrated energy system based on the dynamic response results to achieve optimal scheduling of the integrated energy system.

[0010] As an improvement of the above solution, the status data acquisition module includes: Constructing a physical model of the integrated energy system, and simulating changes in specific parameters of the integrated energy system based on the physical model of the integrated energy system to obtain operating status data of the integrated energy system when the specific parameters change; The expression of the power system physical model is: , , The expression of the physical model of the thermal system is: , The expression of the electric boiler model is: , In the formula, For the The injected active power of each node; For the said The effective value of the AC voltage of each node; is the number of nodes in the power system network; For the The effective value of the AC voltage of each node; For the said nodes and the The conductance of the line between the nodes; For the said nodes and the The voltage phase angle difference between the nodes; For the said nodes and the The susceptance of the line between the nodes; For the said The injected reactive power of each node; and They are Moment Pipeline The starting and ending temperatures; and They are Moment described pipeline The starting and ending temperatures; For the pipeline water flow; is the time difference; is the density of water; For the pipeline The cross-sectional area of For the pipeline Length; For the pipeline The heat transfer coefficient of is the specific heat capacity of water; is the ambient temperature; It is a collection of network nodes of thermal system; For the The calorific value of an electric boiler; For the said The heating efficiency of an electric boiler; For the said The electrical power of an electric boiler.

[0011] As an improvement of the above solution, the network model training module includes: Constructing a neural network model to digitally process the operating status data of the integrated energy system to obtain processed operating status data; splicing the processed running status data to obtain training data of the neural network model; Using the training data to train the neural network model; The expression of the training data is: , , In the formula, and They are the input and output of the neural network model training respectively; and They are Moment and The AC voltage amplitude in the power system at the moment; and Respectively Moment and The phase angle in the power system at the moment; and Respectively Moment and The node active power at the time; and Respectively Moment and The node reactive power at the time; and Respectively Moment and The node temperature value in the thermal system at the moment; and Respectively Moment and Pipe water flow in the thermal system at all times; for The vector of control variables of the integrated energy system at the moment; The transpose symbol.

[0012] As an improvement of the above solution, the control strategy prediction module includes: Constructing the objective function and constraints of the integrated energy system; According to the objective function and constraint conditions and the dynamic response results of the integrated energy system within a certain period of time predicted by the trained neural network model, the optimal control sequence of the integrated energy system is solved to achieve the optimal scheduling of the integrated energy system.

[0013] As an improvement of the above solution, the expression of the objective function is: , The constraint condition is expressed as: , , , , , In the formula, is the objective function of the integrated energy system; is the model predictive control step size; For the The weight of state variables during step-by-step prediction; For the said The actual system operation trajectory during the step prediction; For the said The expected system operation trajectory during the step prediction; For the said The weight of the control variables in the first-step forecast; For the said The actual control input trajectory during step prediction; For the said The expected control input trajectory during step prediction; is the neural network model function; For the said The input of the neural network model during the step prediction; and The upper and lower limits of node voltage in the power system; For the The effective value of the AC voltage of each node; and are the upper and lower limits of active power of nodes in the power system; For the said The injected active power of each node; and The upper and lower limits of reactive power of nodes in the power system; For the said The injected reactive power of each node; and are the upper and lower limits of node temperature in the thermal system; For the said The temperature of a node.

[0014] Compared with the prior art, the embodiment of the present invention discloses a comprehensive energy system model prediction control method and system, which constructs a comprehensive energy system physical model, and obtains the comprehensive energy system operation status data according to the comprehensive energy system physical model; wherein the comprehensive energy system physical model includes a power system physical model, a thermal system physical model and an electric boiler model; constructs a neural network model, digitally processes the operation status data of the comprehensive energy system, and trains the neural network model using the processed operation status data; predicts the dynamic response results of the comprehensive energy system within a certain period of time according to the trained neural network model, and solves the optimal control sequence of the comprehensive energy system according to the dynamic response results to achieve the optimal scheduling of the comprehensive energy system. It can utilize massive comprehensive energy system historical operation data, combine the system mechanism model, construct a complex system dynamic response model through a neural network, and achieve efficient operation of the comprehensive energy system through a model prediction system control method; the comprehensive energy system model prediction control strategy based on the physical information neural network can achieve the optimal scheduling of the comprehensive energy system under the conditions of complex system mechanism, deep coupling, time-varying system parameters and model mismatch. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a flow chart of a comprehensive energy system model predictive control method provided by an embodiment of the present invention; Figure 2 is a schematic diagram of a heat network transmission structure provided by an embodiment of the present invention; Figure 3 is a schematic diagram of a neural network model provided by an embodiment of the present invention; Figure 4 A neural network model predictive control response curve diagram provided by an embodiment of the present invention; Figure 5 is a topological diagram of a comprehensive energy system provided by an embodiment of the present invention; Figure 6 It is a structural schematic diagram of a comprehensive energy system model predictive control system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] It should be noted that the terms "comprises" and "specifically" and any variations of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or inherent to these processes, methods, products or apparatuses.

[0018] See also Figure 1 , Figure 1 1 is a flow chart of a comprehensive energy system model predictive control method provided by an embodiment of the present invention, the comprehensive energy system model predictive control method comprising: S1, constructing a physical model of an integrated energy system, and obtaining operating status data of the integrated energy system according to the physical model of the integrated energy system; wherein the physical model of the integrated energy system includes a physical model of an electric power system, a physical model of a thermal system, and an electric boiler model; S2, constructing a neural network model, digitally processing the operating status data of the integrated energy system, and using the processed operating status data to train the neural network model; S3, predicting the dynamic response results of the integrated energy system within a certain period of time according to the trained neural network model, and solving the optimal control sequence of the integrated energy system according to the dynamic response results to achieve optimal scheduling of the integrated energy system.

[0019] Exemplarily, the model predictive control method of the integrated energy system described in the embodiment of the present invention is implemented by an integrated energy system management server. The integrated energy system management server can exchange information with the target user. The integrated energy system management server obtains the historical operation data of the integrated energy system (such as the power system, the thermal system and the thermal power unit) (such as the power system's flow constraint data, the thermal system's temperature influencing factors and the electric-heat interaction data, etc.), builds the physical model of the integrated energy system (such as the power system physical model, the thermal system physical model and the electric boiler model) according to the historical operation data, simulates the changes of specific parameters of the integrated energy system through the physical model of the integrated energy system, so as to obtain the operation status data of the integrated energy system when the specific parameters change, digitally processes the operation status data of the integrated energy system, and uses the processed operation status data to train the neural network model, so that the neural network model can characterize the operation status and dynamic response process of the integrated energy system, predicts the dynamic response results of the integrated energy system within a certain period of time according to the trained neural network model, and solves the optimal control sequence of the integrated energy system according to the dynamic response results to achieve the optimal scheduling of the integrated energy system. The embodiment of the present invention can utilize massive historical operation data of the integrated energy system, combine it with the system mechanism model, build a dynamic response model of the complex system through a neural network, and realize efficient operation of the integrated energy system through a model prediction system control method.

[0020] Specifically, the step S1 includes: Constructing a physical model of the integrated energy system, and simulating changes in specific parameters of the integrated energy system based on the physical model of the integrated energy system to obtain operating status data of the integrated energy system when the specific parameters change; The expression of the power system physical model is: , , The expression of the physical model of the thermal system is: , The expression of the electric boiler model is: , In the formula, For the The injected active power of each node; For the said The effective value of the AC voltage of each node; is the number of nodes in the power system network; For the The effective value of the AC voltage of each node; For the said nodes and the The conductance of the line between the nodes; For the said nodes and the The voltage phase angle difference between the nodes; For the said nodes and the The susceptance of the line between the nodes; For the said The injected reactive power of each node; and They are Moment Pipeline The starting and ending temperatures; and They are Moment described pipeline The starting and ending temperatures; For the pipeline water flow; is the time difference; is the density of water; For the pipeline The cross-sectional area of For the pipeline Length; For the pipeline The heat transfer coefficient of is the specific heat capacity of water; is the ambient temperature; It is a collection of network nodes of thermal system; For the The calorific value of an electric boiler; For the said The heating efficiency of an electric boiler; For the said The electrical power of an electric boiler.

[0021] In the specific implementation, the historical operation data of the integrated energy system is obtained, and a physical model of the integrated energy system is constructed according to the historical operation data; wherein the historical operation data includes the flow constraint data of the power system, the temperature influencing factors of the thermal system and the electric-thermal interaction data; according to the physical model of the integrated energy system, the specific parameter changes of the integrated energy system are simulated to obtain the operation status data of the integrated energy system when the specific parameters change; More specifically, constructing a comprehensive energy system physical model based on the historical operation data includes: constructing a power system physical model according to the power flow constraint data; Constructing a physical model of the thermal system according to the temperature influencing factors; An electric boiler model is constructed according to the electric-heat interaction data.

[0022] For example, since the actual power system needs to meet the flow constraints during operation, it ensures the balance and reasonable distribution of power in the power system from a physical level. In the actual power system operation, the injected power of each node must meet the flow constraints, otherwise it will lead to system voltage instability, power imbalance and other problems. In severe cases, it may cause power system failures and affect the normal operation and power supply reliability of the power system. Therefore, to construct a physical model of the power system including new energy sources such as wind and light, it is necessary to obtain power data related to the flow constraints (such as power, voltage, phase angle, etc.). The expression of the physical model of the power system is: , , In the formula, For the The injected active power of each node; For the said The effective value of the AC voltage of each node; is the number of nodes in the power system network; For the The effective value of the AC voltage of each node; For the said nodes and the The conductance of the line between the nodes; For the said nodes and the The voltage phase angle difference between the nodes; For the said nodes and the The susceptance of the line between the nodes; For the said The injected reactive power of each node.

[0023] The above variables are expressed in the form of per unit values. The injected power of each node and It is expressed as follows: , , In the formula, For the The active power generated by each node includes wind power, photovoltaic power and grid injection into the Active power of each node; For the The active power absorbed by the electrical load of each node; For the The reactive power generated by each node includes wind power, photovoltaic power, reactive power compensation equipment and grid injection. Reactive power of each node; For the The reactive power absorbed by the electrical load of each node.

[0024] In the actual operation of the thermal system, when heat energy propagates in the water supply pipe, it is affected by temperature factors such as water flow, ambient temperature, and pipe cross section, and there will be a certain loss in the process of heat energy propagation. A thermal system physical model including heat source and cold and hot loads is constructed to express the dynamic process of heat energy propagation in the heat network pipe: , In the formula, and They are Moment Pipeline The starting and ending temperatures; and They are Moment described pipeline The starting and ending temperatures; For the pipeline water flow; is the time difference; is the density of water; For the pipeline The cross-sectional area of For the pipeline Length; For the pipeline The heat transfer coefficient of is the specific heat capacity of water; is the ambient temperature; A collection of network nodes for thermal systems.

[0025] In the heat network, the input heat energy of each node is equal to the output heat energy, such as Figure 2 As shown, Figure 2 is a schematic diagram of a heat network transmission structure provided by an embodiment of the present invention, Figure 2 Included in the node ,node ,node and nodes and the pipelines that connect the nodes, Flow direction node in the heat network pipeline A collection of nodes; Outflow node in the heat network pipeline The node set, the temperature balance and heat energy loss model of the heating network node is: , , , In the formula, For Node Flow to Node The water flow rate of the pipe, For Node Flow to Node The pipe inlet supply / return water temperature, For Node Flow to Node The water flow rate of the pipe, For Node Flow to Node The pipe outlet supply / return water temperature, is a natural constant, For Node Flow to Node The length of the pipeline; For Node Flow to Node The heat transfer coefficient of the pipe, For Node Flow to Node The cross-sectional area of ​​the pipe, For Node Flow to Node The pipe outlet supply / return water temperature, For Node Flow to Node The pipe inlet supply / return water temperature, For Node temperature.

[0026] The heat source output and heat load are expressed as the supply and return water temperature difference of the heat node, satisfying the following energy relationship: , , In the formula, and for Time Node Heat source output and heat load; for Time passing through the node water flow; and They are Nodes in the water supply and return pipes at all times temperature.

[0027] Use electric boilers to convert electrical energy into thermal energy to achieve interaction between electric and thermal networks. The energy conversion relationship is shown in the following formula: , In the formula, For the The calorific value of an electric boiler; For the said The heating efficiency of an electric boiler; For the said The electrical power of an electric boiler.

[0028] Electric boilers convert electrical energy into thermal energy for heating. They are both electrical loads and heat sources, thus satisfying the following requirements: .

[0029] Specifically, the step S2 includes: S21, constructing a neural network model to digitally process the operating status data of the integrated energy system to obtain processed operating status data; S22, splicing the processed running status data to obtain training data of the neural network model; S23, training the neural network model using the training data; The expression of the training data is: , , In the formula, and They are the input and output of the neural network model training respectively; and They are Moment and The AC voltage amplitude in the power system at the moment; and Respectively Moment and The phase angle in the power system at the moment; and Respectively Moment and The node active power at the time; and Respectively Moment and The node reactive power at the time; and Respectively Moment and The node temperature value in the thermal system at the moment; and Respectively Moment and Pipe water flow in the thermal system at all times; for The vector of control variables of the integrated energy system at the moment; The transpose symbol.

[0030] For example, the constructed recurrent neural network is as follows Figure 3 As shown, Figure 3 : is a schematic diagram of a neural network model provided by an embodiment of the present invention, and its working principle can be expressed as follows: , , In the formula, Indicates The status of the location; , , is the weight matrix; and is the bias vector; is the activation function, and They are Input and output of neural network model training at this moment.

[0031] In the actual operation of the integrated energy system, sensors and measurement systems will measure and record a large amount of historical data, which needs to be digitized in batches. For the power system, the phase measurement system can measure the AC voltage amplitude and phase angle, the active power and reactive power injected into the node; for the thermal system, the sensor can measure the temperature value of the node and the water flow in the pipeline. Construct the following data input form: , , in, is the number of nodes in the power system; is the number of nodes in the thermal system, for The AC voltage amplitude in the power system at the moment, for The moment power system The AC voltage amplitude of each node is for The phase angle in the power system at time instant, for The moment power system The phase angle of each node, for The node active power at time for The moment power system The active power of each node, for The node reactive power at time, for The moment power system The reactive power of each node, for The node temperature value in the thermal system at the moment, for The moment in the thermal system The temperature of the node, for The pipe water flow in the thermal system at any moment, for The moment in the thermal system The pipe water flow of each node, for The control variable vector of the integrated energy system at time, for The moment power system The active power generated by the power generation equipment (such as wind power, photovoltaic power and power grid) at each node, for The moment power system The reactive power generated by the power generation equipment at each node, for Moment The active power of the thermal power station at each node, The transpose symbol.

[0032] The above data is concatenated as follows to obtain the input and output for deep learning training: , , In the formula, and They are the input and output of the neural network model training respectively; and They are Moment and The AC voltage amplitude in the power system at the moment; and Respectively Moment and The phase angle in the power system at the moment; and Respectively Moment and The node active power at the time; and Respectively Moment and The node reactive power at the time; and Respectively Moment and The node temperature value in the thermal system at the moment; and Respectively Moment and Pipe water flow in the thermal system at all times; for The vector of control variables of the integrated energy system at the moment; is the transposition symbol; For The sampling time scale is continuous sampling system status data points.

[0033] Physical information (processed operating status data) is used to guide neural network training. Combined with data-driven training methods, the accuracy and interpretability of neural network prediction results are improved. The loss function is designed as follows: In the formula, is the total loss function; is the data-driven loss function; Guide the loss function for physical information; for The neural network predicts the output at each moment; and They are the dynamic equation and the steady-state equilibrium equation respectively; is the weight of the loss function during the training process. As the training progresses, The value gradually increases. The recurrent neural network is trained according to the corresponding input and output, and the loss function involved is used as the objective function. The trainable parameters of the neural network are trained using the gradient descent method to minimize the mean square error.

[0034] Specifically, the step S3 includes: S31, constructing the objective function and constraint conditions of the integrated energy system; S32, solving the optimal control sequence of the integrated energy system according to the objective function and constraint conditions and the dynamic response results of the integrated energy system within a certain period of time predicted by the trained neural network model to achieve optimal scheduling of the integrated energy system.

[0035] More specifically, the objective function is expressed as: , The constraint condition is expressed as: , , , , , In the formula, is the objective function of the integrated energy system; is the model predictive control step size; For the The weight of state variables during step-by-step prediction; For the said The actual system operation trajectory during the step prediction; For the said The expected system operation trajectory during the step prediction; For the said The weight of the control variables in the first-step forecast; For the said The actual control input trajectory during step prediction; For the said The expected control input trajectory during step prediction; is the neural network model function; For the said The input of the neural network model during the step prediction; and The upper and lower limits of node voltage in the power system; For the The effective value of the AC voltage of each node; and are the upper and lower limits of active power of nodes in the power system; For the said The injected active power of each node; and The upper and lower limits of reactive power of nodes in the power system; For the said The injected reactive power of each node; and are the upper and lower limits of node temperature in the thermal system; For the said The temperature of a node.

[0036] Exemplarily, the objective function of the integrated energy system is constructed according to the trained neural network model, and the constraints of the integrated energy system are constructed according to the safe operation requirements of the integrated energy system. After determining the objective function and the constraints, the dynamic response results of the system in a limited future interval are predicted by the trained neural network model, and the optimal control sequence is solved using an optimization algorithm. That is, the objective function is used to weigh the deviation between system operation and control input trajectory, and the constraints ensure the safety and stability of the system, and the control strategy is determined by solving the optimal sequence. Commonly used optimization algorithms such as gradient descent method and genetic algorithm can find the control variable sequence that minimizes the objective function while satisfying the constraints, that is, the optimal control sequence. For example, the process of solving the optimal control sequence, such as Figure 4 As shown, Figure 4 A neural network model predictive control response curve diagram provided by an embodiment of the present invention, with a certain time interval as the sampling time, the current moment As the dividing line, the ideal state value (reference trajectory) is set according to the operation goals and requirements of the integrated energy system. is the model predictive control step length, based on the future control input Predict the system dynamic response results within a limited future interval (prediction output) This optimal control sequence is the integrated energy system scheduling strategy based on model predictive control. By implementing this strategy, the optimized scheduling of the integrated energy system can be achieved, and the economy, reliability and other performance of the system operation can be improved. The integrated energy system can comprehensively consider the deviation of the system operation status and control input within a limited period in the future to achieve the optimization of system performance.

[0037] First, construct the objective function as follows: , During system operation, the following constraints need to be met: , , , , , In the formula, is the objective function of the integrated energy system; is the model predictive control step size; For the The weight of state variables during step-by-step prediction; For the said The actual system operation trajectory during the step prediction; For the said The expected system operation trajectory during the step prediction; For the said The weight of the control variables in the first-step forecast; For the said The actual control input trajectory during step prediction; For the said The expected control input trajectory during step prediction; is the neural network model function; For the said The input of the neural network model during the step prediction; and The upper and lower limits of node voltage in the power system; For the The effective value of the AC voltage of each node; and are the upper and lower limits of active power of nodes in the power system; For the said The injected active power of each node; and The upper and lower limits of reactive power of nodes in the power system; For the said The injected reactive power of each node; and are the upper and lower limits of node temperature in the thermal system; For the said The temperature of a node.

[0038] In a specific embodiment, Figure 5 As shown, Figure 5 It is a topological diagram of an integrated energy system provided by an embodiment of the present invention. An IEEE 33-node distribution network system (1-33 nodes) is coupled with a 6-node heat network system (1-6 nodes), 2 wind turbines (Wind1 and Wind2), and 2 photovoltaic units (Solar1 and Solar2) to build an integrated energy system example for implementation. Among them, the installed capacity of Wind1 is 2MW, the installed capacity of Wind2 is 1.2MW, and the installed capacity of Solar1 and Solar2 is 1MW. One node of the heat network system is connected to the 8 nodes of the distribution network system through a thermal power unit (electric boiler EB) to achieve thermal-electric coupling. The distribution network system adjusts the electric load in different time periods by pointing to transferable loads (TL1-TL5). The heat network contains 3 heat load nodes, 1 heat source node, and 2 hybrid nodes. The embodiment of the present invention constructs a physical model of the integrated energy system, simulates the changes in some system parameters, obtains the operating data of the integrated energy system when the parameters change, digitizes the operating data collected by the integrated energy system, and processes the 10 5A set of comprehensive energy system data sets are used as neural network training data and brought into the neural network model (physical information neural network model) for training, so that the neural network model can characterize the system operation status and dynamic response process. According to the construction of the optimization model, the system decision-making process is optimized based on the predicted dynamic response results of the system in the future time series, and the comprehensive energy system is optimized and dispatched, so that the comprehensive energy system operates economically optimally. The system's new energy consumption before and after the model predictive control is compared. After using the model predictive control, the future operation status of the system can be predicted and optimized. The new energy consumption and consumption rate are effectively improved through measures such as advance heating. After adopting this control strategy, due to the optimization of the heating process of the comprehensive energy system and the consideration of the slow dynamic characteristics of the heat network, the system's heating power consumption is also significantly reduced, achieving the optimal system economy.

[0039] The embodiment of the present invention discloses a model predictive control method for an integrated energy system, which constructs a physical model of the integrated energy system, and obtains the operating status data of the integrated energy system according to the physical model of the integrated energy system; wherein the physical model of the integrated energy system includes a physical model of the power system, a physical model of the thermal system, and an electric boiler model; constructs a neural network model, digitally processes the operating status data of the integrated energy system, and trains the neural network model using the processed operating status data; predicts the dynamic response results of the integrated energy system within a certain period of time according to the trained neural network model, and solves the optimal control sequence of the integrated energy system according to the dynamic response results, so as to realize the optimized scheduling of the integrated energy system. It is possible to utilize massive historical operating data of the integrated energy system, combine the system mechanism model, construct a complex system dynamic response model through a neural network, and realize the efficient operation of the integrated energy system through the model predictive system control method.

[0040] See also Figure 6 , Figure 6 1 is a schematic diagram of a structure of a comprehensive energy system model prediction control system 10 provided by an embodiment of the present invention. The comprehensive energy system model prediction control system 10 includes: The state data acquisition module 11 is used to construct a physical model of the integrated energy system and acquire the operation state data of the integrated energy system according to the physical model of the integrated energy system; wherein the physical model of the integrated energy system includes a physical model of the power system, a physical model of the thermal system and a model of the electric boiler; A network model training module 12 is used to construct a neural network model, digitally process the operating status data of the integrated energy system, and use the processed operating status data to train the neural network model; The control strategy prediction module 13 is used to predict the dynamic response results of the integrated energy system within a certain period of time according to the trained neural network model, and solve the optimal control sequence of the integrated energy system according to the dynamic response results to achieve optimal scheduling of the integrated energy system.

[0041] Specifically, the status data acquisition module 11 includes: Constructing a physical model of the integrated energy system, and simulating changes in specific parameters of the integrated energy system based on the physical model of the integrated energy system to obtain operating status data of the integrated energy system when the specific parameters change; The expression of the power system physical model is: , , The expression of the physical model of the thermal system is: , The expression of the electric boiler model is: , In the formula, For the The injected active power of each node; For the said The effective value of the AC voltage of each node; is the number of nodes in the power system network; For the The effective value of the AC voltage of each node; For the said nodes and the The conductance of the line between the nodes; For the said nodes and the The voltage phase angle difference between the nodes; For the said nodes and the The susceptance of the line between the nodes; For the said The injected reactive power of each node; and They are Moment Pipeline The starting and ending temperatures; and They are Moment described pipeline The starting and ending temperatures; For the pipeline water flow; is the time difference; is the density of water; For the pipeline The cross-sectional area of For the pipeline Length; For the pipeline The heat transfer coefficient of is the specific heat capacity of water; is the ambient temperature; It is a collection of network nodes of thermal system; For the The calorific value of an electric boiler; For the said The heating efficiency of an electric boiler; For the said The electrical power of an electric boiler.

[0042] Specifically, the network model training module 12 includes: Constructing a neural network model to digitally process the operating status data of the integrated energy system to obtain processed operating status data; splicing the processed running status data to obtain training data of the neural network model; Using the training data to train the neural network model; The expression of the training data is: , , In the formula, and They are the input and output of the neural network model training respectively; and They are Moment and The AC voltage amplitude in the power system at the moment; and Respectively Moment and The phase angle in the power system at the moment; and Respectively Moment and The node active power at the time; and Respectively Moment and The node reactive power at the time; and Respectively Moment and The node temperature value in the thermal system at the moment; and Respectively Moment and Pipe water flow in the thermal system at all times; for The vector of control variables of the integrated energy system at the moment; The transpose symbol.

[0043] Specifically, the control strategy prediction module 13 includes: Constructing the objective function and constraints of the integrated energy system; According to the objective function and constraint conditions and the dynamic response results of the integrated energy system within a certain period of time predicted by the trained neural network model, the optimal control sequence of the integrated energy system is solved to achieve the optimal scheduling of the integrated energy system.

[0044] More specifically, the objective function is expressed as: , The constraint condition is expressed as: , , , , , In the formula, is the objective function of the integrated energy system; is the model predictive control step size; For the The weight of state variables during step-by-step prediction; For the said The actual system operation trajectory during the step prediction; For the said The expected system operation trajectory during the step prediction; For the said The weight of the control variables in the first-step forecast; For the said The actual control input trajectory during step prediction; For the said The expected control input trajectory during step prediction; is the neural network model function; For the said The input of the neural network model during the step prediction; and The upper and lower limits of node voltage in the power system; For the The effective value of the AC voltage of each node; and are the upper and lower limits of active power of nodes in the power system; For the said The injected active power of each node; and The upper and lower limits of reactive power of nodes in the power system; For the said The injected reactive power of each node; and are the upper and lower limits of node temperature in the thermal system; For the said The temperature of a node.

[0045] An integrated energy system model predictive control system 10 provided in an embodiment of the present invention can implement all processes of the integrated energy system model predictive control method of the above-mentioned embodiment. The functions of each module in the system and the technical effects achieved are respectively the same as the functions and technical effects achieved by the integrated energy system model predictive control method of the above-mentioned embodiment, and will not be repeated here.

[0046] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A comprehensive energy system model predictive control method, characterized in that: include: Constructing a physical model of an integrated energy system, and obtaining the operating status data of the integrated energy system according to the physical model of the integrated energy system; wherein the physical model of the integrated energy system includes a physical model of the power system, a physical model of the thermal system, and a model of an electric boiler; Constructing a neural network model, digitally processing the operating status data of the integrated energy system, and using the processed operating status data to train the neural network model; According to the trained neural network model, the dynamic response results of the integrated energy system within a certain period of time are predicted, and the optimal control sequence of the integrated energy system is solved according to the dynamic response results to achieve the optimal scheduling of the integrated energy system.

2. The integrated energy system model predictive control method according to claim 1, characterized in that: The step of constructing a physical model of the integrated energy system and obtaining the operating status data of the integrated energy system according to the physical model of the integrated energy system includes: Constructing a physical model of the integrated energy system, and simulating changes in specific parameters of the integrated energy system based on the physical model of the integrated energy system to obtain operating status data of the integrated energy system when the specific parameters change; The expression of the power system physical model is: ， ， The expression of the physical model of the thermal system is: ， The expression of the electric boiler model is: ， In the formula, For the The injected active power of each node; For the The effective value of the AC voltage of each node; is the number of nodes in the power system network; For the The effective value of the AC voltage of each node; For the nodes and the The conductance of the line between the nodes; For the nodes and the The voltage phase angle difference between the nodes; For the nodes and the The susceptance of the line between the nodes; For the The injected reactive power of each node; and They are Moment Pipeline The starting and ending temperatures; and They are Moment described pipeline The starting and ending temperatures; For the pipeline water flow; is the time difference; is the density of water; For the pipeline The cross-sectional area of For the pipeline Length; For the pipeline The heat transfer coefficient of is the specific heat capacity of water; is the ambient temperature; It is a collection of network nodes of thermal system; For the The calorific value of an electric boiler; For the The heating efficiency of an electric boiler; For the The electrical power of an electric boiler.

3. The comprehensive energy system model predictive control method according to claim 1, characterized in that: The step of constructing a neural network model, digitally processing the operating status data of the integrated energy system, and using the processed operating status data to train the neural network model includes: Constructing a neural network model to digitally process the operating status data of the integrated energy system to obtain processed operating status data; splicing the processed running status data to obtain training data of the neural network model; Using the training data to train the neural network model; The expression of the training data is: ， ， In the formula, and They are the input and output of the neural network model training respectively; and They are Moment and The AC voltage amplitude in the power system at the moment; and Respectively Moment and The phase angle in the power system at the moment; and Respectively Moment and The node active power at the time; and Respectively Moment and The node reactive power at the time; and Respectively Moment and The node temperature value in the thermal system at the moment; and Respectively Moment and Pipe water flow in the thermal system at all times; for The vector of control variables of the integrated energy system at the moment; The transpose symbol.

4. The comprehensive energy system model predictive control method according to claim 1, characterized in that: The method of predicting the dynamic response results of the integrated energy system within a certain period of time according to the trained neural network model, and solving the optimal control sequence of the integrated energy system according to the dynamic response results to achieve the optimal scheduling of the integrated energy system includes: Constructing the objective function and constraints of the integrated energy system; According to the objective function and constraint conditions and the dynamic response results of the integrated energy system within a certain period of time predicted by the trained neural network model, the optimal control sequence of the integrated energy system is solved to achieve the optimal scheduling of the integrated energy system.

5. The integrated energy system model predictive control method according to claim 4, characterized in that: The expression of the objective function is: ， The constraint condition is expressed as: ， ， ， ， ， In the formula, is the objective function of the integrated energy system; is the model predictive control step size; For the The weight of state variables during step-by-step prediction; For the The actual system operation trajectory during the step prediction; For the The expected system operation trajectory during the step prediction; For the The weight of the control variables in the step-by-step forecast; For the The actual control input trajectory during step prediction; For the The expected control input trajectory during step prediction; is the neural network model function; For the The input of the neural network model during the step prediction; and The upper and lower limits of node voltage in the power system; For the The effective value of the AC voltage of each node; and are the upper and lower limits of active power of nodes in the power system; For the The injected active power of each node; and The upper and lower limits of reactive power of nodes in the power system; For the The injected reactive power of each node; and are the upper and lower limits of node temperature in the thermal system; For the The temperature of a node.

6. A comprehensive energy system model predictive control system, characterized in that: include: A state data acquisition module, used to construct a physical model of the integrated energy system, and to acquire the operation state data of the integrated energy system according to the physical model of the integrated energy system; wherein the physical model of the integrated energy system includes a physical model of the power system, a physical model of the thermal system and a model of the electric boiler; A network model training module is used to construct a neural network model, digitally process the operating status data of the integrated energy system, and use the processed operating status data to train the neural network model; The control strategy prediction module is used to predict the dynamic response results of the integrated energy system within a certain period of time based on the trained neural network model, and solve the optimal control sequence of the integrated energy system based on the dynamic response results to achieve optimal scheduling of the integrated energy system.

7. The integrated energy system model predictive control system according to claim 6, characterized in that: The state data acquisition module comprises: Constructing a physical model of the integrated energy system, and simulating changes in specific parameters of the integrated energy system based on the physical model of the integrated energy system to obtain operating status data of the integrated energy system when the specific parameters change; The expression of the power system physical model is: ， ， The expression of the physical model of the thermal system is: ， The expression of the electric boiler model is: ， In the formula, For the The injected active power of each node; For the said The effective value of the AC voltage of each node; is the number of nodes in the power system network; For the The effective value of the AC voltage of each node; For the said nodes and the The conductance of the line between the nodes; For the said nodes and the The voltage phase angle difference between the nodes; For the said nodes and the The susceptance of the line between the nodes; For the said The injected reactive power of each node; and They are Moment Pipeline The starting and ending temperatures; and They are Moment described pipeline The starting and ending temperatures; For the pipeline water flow; is the time difference; is the density of water; For the pipeline The cross-sectional area of For the pipeline Length; For the pipeline The heat transfer coefficient of is the specific heat capacity of water; is the ambient temperature; It is a collection of network nodes of thermal system; For the The calorific value of an electric boiler; For the said The heating efficiency of an electric boiler; For the said The electrical power of an electric boiler.

8. The integrated energy system model predictive control system according to claim 6, characterized in that: The network model training module includes: Constructing a neural network model to digitally process the operating status data of the integrated energy system to obtain processed operating status data; splicing the processed running status data to obtain training data of the neural network model; Using the training data to train the neural network model; The expression of the training data is: ， ， In the formula, and They are the input and output of the neural network model training respectively; and They are Moment and The AC voltage amplitude in the power system at the moment; and Respectively Moment and The phase angle in the power system at the moment; and Respectively Moment and The node active power at the time; and Respectively Moment and The node reactive power at the time; and Respectively Moment and The node temperature value in the thermal system at the moment; and Respectively Moment and Pipe water flow in the thermal system at all times; for The vector of control variables of the integrated energy system at the moment; The transpose symbol.

9. The integrated energy system model predictive control system according to claim 7, characterized in that: The control strategy prediction module comprises: Constructing the objective function and constraints of the integrated energy system; According to the objective function and constraint conditions and the dynamic response results of the integrated energy system within a certain period of time predicted by the trained neural network model, the optimal control sequence of the integrated energy system is solved to achieve the optimal scheduling of the integrated energy system.

10. The integrated energy system model predictive control system according to claim 9, characterized in that: The expression of the objective function is: ， The constraint condition is expressed as: ， ， ， ， ， In the formula, is the objective function of the integrated energy system; is the model predictive control step size; For the The weight of state variables during step-by-step prediction; For the The actual system operation trajectory during the step prediction; For the The expected system operation trajectory during the step prediction; For the The weight of the control variables in the step-by-step forecast; For the The actual control input trajectory during step prediction; For the The expected control input trajectory during step prediction; is the neural network model function; For the The input of the neural network model during the step prediction; and The upper and lower limits of node voltage in the power system; For the The effective value of the AC voltage of each node; and are the upper and lower limits of active power of nodes in the power system; For the The injected active power of each node; and The upper and lower limits of reactive power of nodes in the power system; For the The injected reactive power of each node; and are the upper and lower limits of node temperature in the thermal system; For the The temperature of a node.

Images