Adaptive energy management method for digital twin drive

Abstract

The invention discloses an adaptive energy management method for digital twin drive, and belongs to the technical field of energy management. Comprising an equipment digital model, an energy flow simulation model and a constraint condition model, the equipment digital model comprises equipment dynamic characteristic parameters and an aging attenuation model, and the energy flow simulation model comprises an energy flow simulation model and a constraint condition model; the constraint condition model comprises a safe operation constraint, a capacity constraint and an energy efficiency constraint; according to the method, real-time interaction between a physical system and a virtual twinborn body is realized by constructing a high-precision digital twinborn body; based on the simulation and prediction capability of the twinborn body, a self-adaptive strategy is generated in combination with a dynamic optimization algorithm, and continuous iterative optimization is performed through closed-loop feedback; real-time simulation and prediction of the digital twin enable an energy management strategy to cope with load fluctuation and equipment state change in advance, and the response speed is increased.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of energy management, and particularly relates to a digital twin driven adaptive energy management method. BACKGROUND

[0002] In the prior art, some energy management methods introduce real-time data acquisition and optimization algorithms, but lack the global dynamic simulation capability of the physical system, and cannot predict the energy flow trend in advance. At the same time, the generation of the optimization strategy is not deeply coupled with the actual state of the device, resulting in deviation in the strategy execution process.

[0003] The digital twin technology can realize dynamic mapping and simulation of the whole life cycle of the system by constructing a virtual mirror of the physical system, but its application in energy management currently stays at the state monitoring level. SUMMARY

[0004] The application aims at solving the above problems, and provides a digital twin driven adaptive energy management method.

[0005] In order to achieve the above purpose, the application adopts the following technical scheme: a digital twin driven adaptive energy management method, comprising the following steps: 1) Constructing a digital twin of the physical energy system: based on the device parameters, topological structure and energy conversion law of the physical energy system, a digital twin containing a device digital model, an energy flow simulation model and a constraint condition model is established, the device digital model contains device dynamic characteristic parameters and aging attenuation model, and the constraint condition model contains safety operation constraint, capacity constraint and energy efficiency constraint; 2) Establishing a real-time data interaction channel between the physical system and the digital twin: collecting real-time operation data of the physical energy system through a sensing network, the real-time operation data including device output power, energy consumption data, state parameters and environmental parameters, transmitting the real-time operation data to the digital twin, and realizing state synchronization between the digital twin and the physical system; 3) Multi-dimensional state evaluation and prediction based on the digital twin: the digital twin reconstructs the dynamic energy distribution state of the physical system through the energy flow simulation model according to the synchronized real-time operation data, evaluates the device health degree and energy efficiency level in combination with the device digital model, and outputs load demand prediction value, renewable energy output prediction value and device state trend prediction result in a future preset time period based on the prediction model trained based on historical data and real-time data; 4) Dynamically generating adaptive energy management strategy: the digital twin takes the system energy efficiency optimization, the lowest operation cost and the highest stability as the optimization goal, based on the evaluation results and the prediction results, calls the dynamic optimization algorithm to solve the energy distribution scheme, the device start-stop plan and the energy storage scheduling strategy, and generates the initial energy management strategy; 5) Execute the energy management strategy and feedback optimization: the generated initial energy management strategy is issued to the execution mechanism of the physical energy system, and after the physical system executes, the strategy execution effect data is collected through the sensor network, including the actual energy consumption deviation, the device response delay and the system stability index, and the effect data is fed back to the digital twin; 6) Iterative optimization of digital twin and energy management strategy: the digital twin corrects the parameters of the device digital model and the prediction model according to the feedback effect data, re-executes steps 3)-5), until the execution effect of the energy management strategy meets the preset optimization threshold, forming a closed-loop adaptive management mechanism.

[0006] As a further description of the above technical solutions: In step 1), the device digital model is constructed by fusing a physical mechanism model and a data-driven model, the physical mechanism model is established based on a device energy conversion formula, and the data-driven model uses a neural network algorithm to fit the actual operation deviation of the device.

[0007] As a further description of the above technical solutions: In step 3), the prediction model uses a fusion algorithm of long short-term memory network and gradient boosting tree, the time granularity of the load demand prediction value and the renewable energy output prediction value is 5-15 minutes, and the prediction time is 1-24 hours.

[0008] As a further description of the above technical solutions: In step 4), the dynamic optimization algorithm is an improved particle swarm optimization algorithm, the improved particle swarm optimization algorithm introduces an adaptive inertia weight, the inertia weight is dynamically adjusted according to the real-time load fluctuation degree of the system, when the load fluctuation amplitude exceeds the preset threshold, the inertia weight is increased to expand the search range, otherwise the inertia weight is reduced to improve the convergence speed.

[0009] As a further description of the above technical solutions: In step 6), the parameter correction uses a Kalman filter algorithm, the dynamic characteristic parameters in the device digital model are updated through the residual error of the effect data and the twin simulation results, so that the simulation error of the digital twin is controlled within 5%.

[0010] As described above, due to the adoption of the above technical solutions, the beneficial effects of the present application are: In the present application, by constructing a high-precision digital twin, real-time interaction between the physical system and the virtual twin is realized; based on the simulation and prediction ability of the twin, an adaptive strategy is generated by combining a dynamic optimization algorithm, and continuously iterated optimization is achieved through a closed-loop feedback; the real-time simulation and prediction of the digital twin enable the energy management strategy to respond to load fluctuations and equipment state changes in advance, improving response speed; the use of dynamic optimization algorithms and closed-loop feedback mechanisms enables the system to improve energy efficiency and reduce operating costs; the device digital model combines physical mechanisms and data-driven methods, providing reliable basis for strategy optimization. BRIEF DESCRIPTION OF DRAWINGS

[0011] Figure 1 A flowchart of a digital twin-driven adaptive energy management method. DETAILED DESCRIPTION

[0012] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0013] Embodiment: S01: Constructing a digital twin: The microgrid physical system includes photovoltaic components, energy storage batteries, diesel generators, and load devices. When constructing the digital twin: Device digital model: the photovoltaic component uses a physical mechanism model combined with a neural network model; the energy storage battery model includes a capacity attenuation curve and dynamic characteristics of charging and discharging efficiency; Energy flow simulation model: based on the node voltage method, a microgrid energy flow calculation model is constructed to simulate the transmission and distribution of power among devices; Constraint condition model: set the energy storage battery SOC, the number of diesel generator start-stop times, and the system frequency stability threshold; S02: Real-time data interaction: Real-time data is collected by sensors deployed on each device: photovoltaic output power, energy storage SOC, diesel generator fuel consumption, load power, ambient temperature, and light intensity, with a data sampling frequency of 1 minute / second. The data is transmitted to the digital twin platform through a 5G network, realizing the state synchronization of the twin and the physical system; S03: State evaluation and prediction: State evaluation: based on real-time data, the digital twin calculates the power loss of each node through the energy flow simulation model, and evaluates the energy storage battery health and diesel generator operating efficiency in combination with the device digital model; Prediction model: The fusion algorithm of LSTM (Long Short-Term Memory Network) and GBDT (Gradient Boosting Tree) is used to predict the load demand and photovoltaic output for the next 24 hours, and to control the prediction error. S04: Generate an adaptive strategy: With the objective of "minimizing the overall energy consumption cost (electricity purchase cost + diesel cost - grid connection revenue)", an improved particle swarm optimization algorithm is used to solve the problem. When the load fluctuation exceeds 10%, the inertia weight is adjusted from 0.5 to 0.8 to expand the search range; Generate energy storage charging and discharging plans, such as charging when photovoltaic output is excessive, discharging during peak load periods, diesel generator start-up and shutdown times, and interaction power with the main power grid; S05: Implementation and Feedback The optimization strategy is sent to the microgrid controller, and after execution, actual data is collected, such as the deviation between the actual charging and discharging power of energy storage and the plan, and the system frequency fluctuation value, and then fed back to the digital twin. S06: Iterative Optimization By employing the Kalman filter algorithm, the weight parameters of the photovoltaic module neural network model and the capacity decay coefficient of the energy storage battery are corrected based on feedback data, reducing the twin simulation error from the initial 12% to less than 5%. The S3-S5 algorithm is re-executed to iteratively optimize the strategy, ultimately reducing the daily operating cost of the microgrid by 12% and increasing the photovoltaic absorption rate to 98%.

[0014] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A digital twin-driven adaptive energy management method, characterized in that: Includes the following steps: 1) Constructing a digital twin of a physical energy system: Based on the equipment parameters, topology and energy conversion laws of the physical energy system, establish a digital twin including a digital equipment model, an energy flow simulation model and a constraint model. The digital equipment model includes equipment dynamic characteristic parameters and an aging and decay model, and the constraint model includes safe operation constraints, capacity constraints and energy efficiency constraints. 2) Establish a real-time data interaction channel between the physical system and the digital twin: collect real-time operating data of the physical energy system through a sensor network. The real-time operating data includes the output power, energy consumption data, status parameters and environmental parameters of each device. Transmit the real-time operating data to the digital twin to achieve state synchronization between the twin and the physical system. 3) Multi-dimensional status assessment and prediction based on digital twins: The digital twin reconstructs the dynamic energy distribution of the physical system through an energy flow simulation model based on synchronized real-time operating data, assesses the health and energy efficiency of the equipment by combining the equipment digital model, and outputs the predicted load demand, renewable energy output and equipment status trend prediction results for a preset time period based on the prediction model trained on historical and real-time data. 4) Dynamically generate adaptive energy management strategy: The digital twin takes the optimal system energy efficiency, the lowest operating cost and the highest stability as the optimization goal. Based on the evaluation results and prediction results, it calls the dynamic optimization algorithm to solve the energy allocation scheme, equipment start-up and shutdown plan and energy storage scheduling strategy to generate the initial energy management strategy. 5) Execute energy management strategy and provide feedback optimization: The generated initial energy management strategy is sent to the execution mechanism of the physical energy system. After the physical system executes the strategy, it collects the strategy execution effect data through the sensor network. The effect data includes actual energy consumption deviation, equipment response delay and system stability indicators. The effect data is then fed back to the digital twin. 6) Iterative optimization of digital twin and energy management strategy: The digital twin corrects the parameters of the device digital model and prediction model based on the feedback effect data, and re-executes steps 3) to 5) until the execution effect of the energy management strategy meets the preset optimization threshold, forming a closed-loop adaptive management mechanism.

2. The digital twin-driven adaptive energy management method according to claim 1, characterized in that, In step 1), the digital model of the equipment is constructed by integrating the physical mechanism model and the data-driven model. The physical mechanism model is established based on the energy conversion formula of the equipment, and the data-driven model uses a neural network algorithm to fit the actual operating deviation of the equipment.

3. The digital twin-driven adaptive energy management method according to claim 1, characterized in that, In step 3), the prediction model adopts a fusion algorithm of long short-term memory network and gradient boosting tree. The time granularity of the load demand prediction value and the renewable energy output prediction value is 5-15 minutes, and the prediction duration is 1-24 hours.

4. The digital twin-driven adaptive energy management method according to claim 1, characterized in that, In step 4), the dynamic optimization algorithm is an improved particle swarm optimization algorithm. The improved particle swarm optimization algorithm introduces adaptive inertia weights. The inertia weights are dynamically adjusted according to the real-time load fluctuation of the system. When the load fluctuation exceeds a preset threshold, the inertia weights are increased to expand the search range, and vice versa to reduce the inertia weights to improve the convergence speed.

5. The digital twin-driven adaptive energy management method according to claim 1, characterized in that, In step 6), the parameter correction adopts the Kalman filter algorithm, and the dynamic characteristic parameters in the digital model of the device are updated by the residual between the effect data and the twin simulation results, so that the simulation error of the digital twin is controlled within 5%.

Images