Micro-grid energy management system based on digital twinning

Abstract

A micro-grid energy management system based on digital twinning includes a physical micro-grid layer, a virtual space modeling layer, an energy management layer, an intelligent control layer, and a human-computer interaction layer. The communication between the layers is carried out by using the industrial Internet of Things technology based on the MQTT protocol, the virtual micro-grid system with the same life cycle as the entity is established in a data-driven manner, the energy management layer formulates a preliminary scheme through the entity system operation data, adjusts the scheme in combination with the feedback information of the virtual micro-grid system executing the preliminary scheme, applies the final scheme to the micro-grid system, and the operator can also issue control instructions to the virtual micro-grid system and the micro-grid system based on the publish / subscribe mechanism through the human-computer interaction interface. The application perfects the function of the energy management system, formulates a more accurate scheduling optimization scheme, effectively improves the energy utilization rate, realizes more human-computer interaction, and improves the safety and reliability of information transmission between the layers and the systems.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of micro-grid energy management, and in particular to a micro-grid energy management system operation method, management method and inter-level communication based on digital twinning. BACKGROUND

[0002] With the depletion of fossil fuels and the increasing environmental pollution, renewable energy distributed generation technology has attracted more and more attention from countries due to its low pollution emission and resource saving. However, distributed generation has the problems of randomness and intermittency, and a large number of distributed power sources connected will affect the power quality of the distribution network. Micro-grid is a small-scale power generation and distribution system composed of distributed power sources, energy storage systems and loads, and has the distinct feature of strong flexibility. As an effective supplement to the large power grid, it can solve the impact of large-scale distributed power sources on the power grid, and can achieve peak load shifting and frequency modulation through reasonable control of loads and energy storage systems. The role of the micro-grid energy management system is to generate effective energy scheduling strategies, improve the utilization rate of renewable energy through two-way interaction of the demand side and the power generation side, and realize the optimal allocation of resources in different time periods, thereby improving the economy and reliability of micro-grid operation while ensuring power quality.

[0003] The current micro-grid energy management system has the following problems: first, it focuses on monitoring, and most of the information is operating information such as voltage and frequency, with less collection of equipment information, and cannot timely respond to the health status of the equipment; second, it lacks overall simulation analysis of the operation of the micro-grid system; third, the information transmission within the micro-grid is relatively closed, the connection between systems is not strong, and the security of transmission is poor; fourth, the human-computer interaction interface in the current energy management system often only stays at the display level, and does not realize real-time control of the micro-grid system operation by operators, the human-computer interaction function is realized less, and the optimal allocation of resources is not accurate enough. SUMMARY

[0004] In view of the problems of the current micro-grid system mentioned in the background art, such as less human-computer interaction, inaccurate analysis of equipment health, poor security of inter-level communication, inaccurate scheduling and operation strategy, etc., the present application proposes a micro-grid energy management system based on digital twinning.

[0005] A micro-grid energy management system based on digital twinning, characterized in that the energy management system comprises a physical micro-grid layer, a virtual space modeling layer, an energy management layer, an intelligent control layer, and a human-computer interaction layer.

[0006] The entity micro-grid layer includes a micro-grid system and a data acquisition and transmission unit, wherein the micro-grid system includes a distributed power source, a rectifier, an inverter, a diesel generator, an energy storage system, and a load. The load is divided into a key load and a controllable load, and the controllable load and an electric vehicle together form a virtual energy storage system. The data acquisition and transmission unit includes intelligent sensors, cameras, electronic tags in equipment for data acquisition, and FPGA, intelligent gateways, optical fibers and other devices for data processing and transmission.

[0007] Further, the entity micro-grid layer collects various information such as operation information and equipment information through intelligent sensors, collects audio, video and other information through cameras, collects information such as the environment around the distributed power source and the health status of the equipment through the electronic tags installed in the distributed power source, and collects various information of the micro-grid system based on the above collection technologies. The collected information is mostly analog information, and the conversion from analog to digital is completed by using the A / D chip PCF8591 chip on the FPGA board with the model EP4CE10F17C8. The converted information is stored in the FIFO module for temporary storage, and then the data is serially transmitted to the intelligent gateway based on the UART protocol through the RS232 serial port. Based on the MQTT protocol, the information of the entity micro-grid layer is transmitted through wired networks such as optical fibers to the system modeling layer and the energy management layer.

[0008] The virtual space modeling layer includes a cloud server and a client. The cloud server can access a large amount of sensor data, and its functions include information processing and modeling. The information processing receives various information transmitted from the entity micro-grid layer and analyzes it, communicates with the energy management layer based on the cloud server, and transmits various historical operation information to the virtual space modeling layer based on the TCP / IP protocol. The virtual space modeling layer can transmit virtual system operation information. The modeling refers to the cloud server using cloud computing, machine learning and other means based on a large amount of historical data and real-time data to establish a virtual micro-grid system in a 1:1 ratio with the micro-grid system through data-driven methods. The virtual micro-grid system evolves in real time by receiving data from the entity object, thereby maintaining consistency with the entity throughout its life cycle, mapping the micro-grid system to the virtual space, completing 1:1 modeling and realizing the same life cycle as the entity, and truly realizing the mapping from the entity to the virtual.

[0009] The energy management layer includes a cloud server which communicates with the cloud server of the virtual space modeling layer based on TCP / IP, and the clients communicate with each other based on the MQTT-based publish / subscribe mechanism. The energy management layer and the virtual space modeling layer both include cloud servers, and the two servers perform their respective functions, enhance the computing power, improve the speed, and can better realize the establishment of the twin micro-grid system. The cloud server includes a real-time information database, a historical information database, a historical fault information database, a virtual system operation information database, a monitoring and early warning module, a prediction module, a scheduling module, an optimization module, and a service publishing module.

[0010] The real-time information database stores various types of information received and analyzed by the cloud server for the operation of the micro-grid system. The information in the real-time database is stored in the historical database at a set time interval. The historical database includes data transmitted from the real-time database and historical data of previous operations of the micro-grid system. Based on the information in the historical database, the range of various operating data and parameters for the normal operation of the micro-grid system can be obtained, and the cloud server will interact with the virtual space modeling layer to publish the historical database information. The historical fault information database stores various types of data changes based on fault reasons after an accident occurs in the micro-grid system. The virtual system operation information database includes various types of information for the operation of the virtual micro-grid system transmitted by the virtual micro-grid layer based on TCP / IP. The monitoring module compares the information in the real-time database with the range of various data and parameters for the normal operation of the micro-grid based on the historical database, and compares whether the real-time data is within the normal range. If there is data exceeding the normal range, the fault or safety hazard can be classified based on the fault information in the historical fault information database, based on K-neighbor method and neural network, to realize fault diagnosis and early warning. The energy management layer prediction module includes load prediction, electricity price prediction, and power generation prediction.

[0011] The load prediction can achieve long-term and short-term load prediction using the load information in the historical database and the real-time database. First, the data features are extracted, and different features are normalized. Due to the regional characteristics of the load, the processed load information is clustered using the k-means clustering method. Based on the clustering results, a time series model is constructed based on the historical sequence data and real-time sequence data of each type of load, and the neural network is trained. Based on the neural network, short-term load prediction can be achieved. When the load fluctuation is large, expert experience can be combined to correct the prediction and enhance the accuracy of the prediction. Using the same method, support vector machines can achieve medium and long-term load prediction.

[0012] The electricity price prediction, the prediction of short-term electricity price is not only related to historical electricity price and load value, but also related to external factors such as weather. Therefore, the electricity price information in the historical database and the predicted load information, and the weather information communicated and transmitted by the weather station are combined to make the prediction, and the prediction model of the short-term electricity price is E(t) = O(t) + B(t) + W(t) + S(t) + R(t), wherein E(t) represents the electricity price at time t, O(t) represents the use of raw materials, B(t) represents the basic load component, W(t) represents the weather change, and S(t) represents a special event. Based on the model, the real-time electricity price is predicted by using an LSTM (Long Short Term Memory network).

[0013] The power generation prediction includes power generation prediction of each distributed power source, wherein the power generation prediction of the distributed power source combines mechanical performance, illumination intensity, temperature, wind speed and other weather information to predict the power generation of each distributed power source, trains a neural network by using weather information and power generation information, and realizes the prediction of power generation of each distributed power source based on the neural network. The overall power generation prediction is the sum of the power generation predictions of each distributed power source.

[0014] The scheduling module of the energy management layer comprehensively considers the relationship between the distributed power output and the load, combines the results obtained by the prediction module, and controls the controllable load, the diesel generator and the energy storage system to ensure the reliability of power supply. In the grid-connected operation, when the distributed power output is greater than the load demand power, the excess power can charge the storage battery, considering the upper limit and service life, the virtual energy storage system composed of the controllable load can absorb the excess power, and at the same time, the virtual energy storage system can sell power to the grid during the peak electricity price period. When the distributed power output is insufficient to provide load power, when the power of the storage battery is not lower than the lower limit, the storage battery can be discharged, and at the same time, the virtual energy storage system can reduce power consumption under the condition of not affecting normal use, so that the key load power supply is guaranteed, and the diesel generator can be controlled to generate power to ensure reliable power supply. When the distribution network fails, the connection with the distribution network is disconnected, so that the microgrid operates in an island operation mode. The difference between this mode and the above-mentioned grid-connected mode is that the interaction with the distribution network is cut off, which is a self-sufficient mode. The preliminary scheduling scheme is determined in this way. The feedback information of the virtual microgrid executing the preliminary scheduling scheme is combined to correct the preliminary scheduling scheme, and the scheduling scheme is determined again.

[0015] The optimization module of the energy management layer is optimization of the operation scheme on the basis of the scheduling module, and the micro-grid operation optimization planning is performed on the premise of ensuring power supply reliability, with the economic efficiency, minimum power purchase from the distribution network and minimum carbon emission as the targets, the twin micro-grid system model is established based on the virtual space modeling layer, the optimization targets are determined as the multi-objective functions of meeting the economic efficiency, minimum peak power purchase and minimum carbon emission, and the intelligent algorithm is used for optimization, and the improved particle swarm algorithm is used for optimization, so as to determine the preliminary optimized operation scheme.

[0016] The economic efficiency target function is The total cost of the micro-grid operation is N, the total scheduling section number is c, the price of diesel oil is f, the oil consumption of the diesel generator is g, the power purchase price of the micro-grid from the distribution network is P g The power purchase amount is S, the price of selling electric energy to the micro-grid is P S The sold electric amount is gP g sP S Only exists in the grid-connected operation, and is zero in the island operation. M is the number of distributed energy, a is the power generation cost of the distributed power supply, P a is the power generation power;

[0017] The carbon emission target function is The total carbon emission amount is x, and the carbon emission coefficient of the diesel generator is x. The virtual micro-grid layer subscribes to the preliminary operation scheme and is applied to the virtual micro-grid system, the virtual micro-grid layer publishes various information of the virtual micro-grid system running after the optimized scheme is executed, the energy management layer receives the feedback information of the micro-grid executing the operation scheme based on the TCP / IP protocol, adjusts the preliminary optimized operation scheme based on the information of the virtual micro-grid system executing the preliminary optimized scheme fed back in the virtual micro-grid database, and the adjusted operation scheme is the final optimized scheme.

[0018] The service publishing module of the energy management layer is based on the subscription / publishing mechanism of MQTT, the energy management layer performs operation processing and analysis on the transmitted data based on the cloud server, and various services can be published to the client through the cloud server, wherein the client of the virtual micro-grid layer can subscribe to the preliminary scheduling scheme and the preliminary optimized operation scheme, and control the virtual micro-grid system according to the scheme, the client of the intelligent control layer subscribes to the final optimized operation scheme to control the micro-grid system, and the client of the man-machine interaction layer can subscribe to various information to check the running state of the micro-grid.

[0019] The human-computer interaction layer includes a client device, and an operator can understand various parameters of the micro-grid operation by subscribing to various information of the energy management layer. Based on the application software, the operator can realize visual processing of the data, and the operator can also issue control instructions to the entity and virtual micro-grid system based on the publish / subscribe mechanism according to the experience of the operator, so that the interaction between the human and the computer is better realized.

[0020] The intelligent control layer includes a controller and a grid-connected switch, and the micro-grid system is controlled by subscribing to the scheduling scheme and the operation optimization scheme of the energy management layer. The control mode includes feedback of the virtual micro-grid system operation information, and a closed-loop feedback of virtual control and actual control is realized. The control includes control of the inverters and converters of the distributed power supply, and the start / stop of the distributed power supply and the diesel generator, the adjustment of the output power, the charging / discharging of the energy storage system and the like can be realized. Meanwhile, the grid-connected switch is controlled to realize the operation mode instruction issued by the operator. In the grid-connected operation mode, PQ control is adopted for the inverters to meet the requirements of frequency and quality, and in the island operation mode, droop control is adopted to ensure the unity of the micro-grid frequency and voltage.

[0021] A micro-grid energy management system operation method based on digital twinning, characterized in that it comprises the following steps:

[0022] Step 1: Real-time collection of various information of the micro-grid system by intelligent sensors, cameras and radio frequency technology based on tags;

[0023] Step 2: The analog quantity is converted by the A / D conversion chip with the model PCF8591 on the FPGA with the model EP4CE10F17C8, the converted digital quantity is temporarily stored in the FIFO module of the FPGA, and then the converted digital quantity information is serially transmitted to the intelligent gateway based on the UART protocol through the RS232 serial port on board, and the information is transmitted to the cloud server of the energy management layer and the virtual modeling layer through the optical fiber wired network technology based on the MQTT protocol;

[0024] Step 3: The cloud server of the energy management layer analyzes and stores the collected information, and transmits various historical information of the micro-grid to the virtual space modeling layer based on the TCP / IP protocol. Meanwhile, the cloud server of the energy management layer monitors various data in real time based on the change range of various data in the historical database during normal operation, compares the parameters exceeding the range with the information of the historical fault database, and determines the type of fault based on the K- nearest neighbor method.

[0025] Step 4: The cloud server of the virtual space modeling layer parses the information transmitted by the entity micro-grid layer, and combines various historical data information transmitted by the energy management layer. Based on real-time and historical data information, the virtual micro-grid system modeling is completed by data driving. The established virtual micro-grid system maintains a 1:1 ratio with the original system and has the same life cycle as the micro-grid system;

[0026] Step 5: The energy management layer realizes the prediction of load, the prediction of electricity price and the prediction of load based on the cloud server;

[0027] Step 6: The energy management layer determines the initial scheduling scheme based on the cloud server based on power supply reliability, and provides the initial scheduling scheme theme service;

[0028] Step 7: The client of the virtual space modeling layer controls the virtual micro-grid system to run according to the initial scheduling scheme by subscribing to the initial scheduling scheme;

[0029] Step 8: The running information of the virtual micro-grid system after executing the scheduling scheme is transmitted to the cloud server of the virtual space modeling layer through the gateway, and the cloud server of the virtual space modeling layer transmits the running information after executing the preliminary scheduling to the cloud server of the energy management layer based on TCP / IP protocol;

[0030] Step 9: The cloud server of the energy management layer adjusts the preliminary scheduling scheme based on the information fed back after the virtual micro-grid system executes the preliminary scheduling scheme;

[0031] Step 10: The client of the virtual space modeling layer modifies the scheduling instruction according to the subscription scheduling scheme theme service;

[0032] Step 11: The cloud server of the energy management layer determines the preliminary operation optimization scheme based on the scheduling scheme, and publishes the preliminary operation optimization scheme theme service;

[0033] Step 12: The client of the virtual space modeling layer controls the virtual micro-grid system to adjust according to the initial optimization scheme by subscribing to the initial optimization scheme;

[0034] Step 13: The running information of the virtual micro-grid system after executing the operation optimization scheme is transmitted to the cloud server of the virtual space modeling layer through the gateway, and the cloud server of the virtual space modeling layer transmits the running information after executing the preliminary scheduling to the cloud server of the energy management layer based on TCP / IP protocol;

[0035] Step 14: The cloud server of the energy management layer adjusts the preliminary operation optimization scheme based on the information fed back after the virtual micro-grid system executes the preliminary operation optimization scheme;

[0036] Step 15: The client of the intelligent control layer controls the controller by subscribing to the running optimization scheme, so as to realize the control of the micro-grid system to run according to the publishing scheme;

[0037] Step 16: The man-machine interaction layer can master various running information of the micro-grid by subscribing to various services published by the energy management layer, and the visual processing of the data can be completed by designing and referencing software through programming and other means;

[0038] Step 17: In the client of the man-machine interaction layer, the operator can publish the dispatching instruction to the server of the energy management layer according to the observed various running information and experience, and the client of the virtual space modeling layer and the intelligent control layer can realize the control of the entity and the virtual micro-grid system by subscribing to the control instruction published by the client of the man-machine interaction layer.

[0039] Compared with the prior art, the present application has the following advantages:

[0040] (1) By combining the feedback of the virtual micro-grid system running information, the dispatching and running optimization scheme can be more accurate.

[0041] (2) The publishing / subscription mechanism based on the MQTT protocol can make the information transmission more secure.

[0042] (3) By means of implanting "tags" to the distributed power supply, the radio frequency technology is used for information collection, so that the equipment information and surrounding information can be better obtained.

[0043] (4) The man-machine interaction does not stop at the display level, and good man-machine interaction can be realized. BRIEF DESCRIPTION OF DRAWINGS

[0044] The embodiments of the present application will be further described in detail below with reference to the drawings.

[0045] Figure 1 It is a kind of micro-grid energy management system plane structure diagram based on digital twinning;

[0046] Figure 2 It is a kind of micro-grid energy management system layer communication structure diagram based on digital twinning;

[0047] Figure 3 It is a kind of micro-grid energy management system running method flow chart based on digital twinning.

[0048] Figure 1 It is a kind of micro-grid energy management system plane structure diagram based on digital twinning provided by the present application, Figure 1The micro-grid system includes a micro-grid system, a virtual micro-grid system, a power distribution network, an energy management system, and a central controller, wherein the micro-grid system specifically includes a photovoltaic module connected to a micro-grid main network through a solar panel, a D / A converter, and a switch, a fan module connected to the micro-grid main network through an A / D converter, a D / A converter, and a switch, an energy storage system connected to the micro-grid main network through a D / A converter and a switch, a diesel generator connected to the micro-grid main network through a switch, a load connected to the micro-grid main network through a switch, and the micro-grid main network connected to a 0.4kv / 10kv power distribution network through a switch. Figure 1 The virtual micro-grid system is composed of the same components as the micro-grid system, and the energy management layer is connected to the central controller, the physical micro-grid layer, and the virtual micro-grid layer.

[0049] Figure 2 The present application provides a micro-grid energy management system interlayer communication structure based on digital twinning, Figure 2 The micro-grid system includes a physical micro-grid layer, a virtual space modeling layer, an energy management layer, a human-computer interaction layer, an intelligent control layer, and communication between the layers, wherein the communication between the layers includes communication between the physical micro-grid layer and the energy management layer and the virtual space modeling layer, communication between the energy management layer and the virtual space modeling layer, communication between the energy management layer and the human-computer interaction layer, communication between the intelligent control layer and the energy management layer, communication between the client of the human-computer interaction layer and the client of the intelligent control layer, and communication between the client of the virtual space modeling layer.

[0050] The communication between the physical micro-grid layer and the energy management layer and the virtual space modeling layer in the micro-grid system is realized by smart sensors, cameras, and analog information collected based on radio frequency technology, and the analog quantity is converted by an A / D conversion chip with model PCF8591 on an FPGA with model EP4CE10F17C8, the converted digital quantity is temporarily stored in the FIFO module of the FPGA, and then the converted digital quantity information is serially transmitted to the smart gateway based on the UART protocol through the on-board RS232 serial port, and then the information is transmitted to the cloud server of the energy management layer and the virtual modeling layer based on the MQTT protocol through optical fiber wired network technology, the cloud server of the energy management layer and the virtual space modeling layer can analyze and process the transmitted information.

[0051] The communication between the energy management layer and the virtual space modeling layer includes the communication between the cloud server of the energy management layer and the cloud server of the virtual space modeling layer, and the communication between the cloud server of the energy management layer and the client of the virtual space modeling layer. The communication between the cloud server of the energy management layer and the cloud server of the virtual space modeling layer is based on the TCP / IP protocol to realize the data transmission between the servers, so as to realize the transmission of the historical database information in the energy management layer to the virtual space modeling layer for the modeling of the virtual micro-grid system. The virtual space modeling layer can also feed back the virtual micro-grid operation information after the preliminary operation scheme is executed to the energy management layer based on the TCP / IP protocol to help generate the final scheduling and operation scheme. The cloud server of the energy management layer can generate the preliminary scheduling and optimization scheme and other services through calculation, and the virtual space modeling layer can control the virtual micro-grid system to operate according to the operation scheme by subscribing to the preliminary operation scheme.

[0052] The virtual space modeling layer and the energy management layer both include cloud servers, but actually only the cloud server of the energy management layer can realize the above functions. However, due to the large amount of calculation, the time required is also relatively long. The addition of the server in the virtual space modeling layer can help share the important modeling task. The two servers perform their respective functions, accelerate the calculation speed, and can realize the same life cycle as the micro-grid system.

[0053] The communication between the energy management layer and the human-computer interaction layer, the cloud server of the energy management layer can provide various services, and the client of the human-computer interaction layer can subscribe to various services such as operation parameters, device conditions, scheduling scheme, optimization operation scheme, etc. The human-computer interaction layer can perform visual processing on the data based on an application program.

[0054] The interaction between the energy management layer and the intelligent control layer is also based on the subscription / publishing mechanism. The server of the energy management layer can publish the final scheduling scheme and optimization scheme service, and the intelligent control layer can subscribe to the final scheduling scheme and optimization scheme to guide the operation of the micro-grid system.

[0055] The communication between the human-computer interaction layer and the virtual micro-grid layer and the intelligent control layer is also based on the publishing / subscription mechanism. The human-computer interaction layer can realize the comprehensive understanding of the micro-grid system by subscribing to the services of the energy management layer. The human-computer interaction layer can publish control instructions to the cloud server of the energy management layer. The client of the virtual space modeling layer and the intelligent control layer subscribes to the information published by the human-computer interaction layer. Based on the cloud server of the energy management layer, the indirect communication between the human-computer interaction layer and the virtual micro-grid layer and the intelligent control layer can be realized. The operator can control the virtual space modeling layer and the intelligent control layer,

[0056] The communication service quality of the energy management layer and each layer client adopts QOS2 service quality, so that the accuracy of information transmission can be guaranteed.

[0057] Figure 3 is a micro-grid energy management system operation method flow chart provided by the present application, combining Figure 1 the plane structure diagram and Figure 2 the communication between each level Figure 3 The micro-grid energy management system operation method is described, and the specific steps include:

[0058] Step 1: Real-time collection of various types of information of the micro-grid system by intelligent sensors, cameras and tag-based radio frequency technology;

[0059] Step 2: The analog quantity collected by the A / D conversion chip with model PCF8591 on the FPGA board with model EP4CE10F17C8 is converted, and the converted digital quantity is temporarily stored in the FIFO module of the FPGA. The converted digital quantity information is serially transmitted to the intelligent gateway based on the UART protocol through the RS232 serial port on board, and the information is transmitted to the cloud server of the energy management layer and the virtual modeling layer through optical fiber and other wired network technologies based on the MQTT protocol;

[0060] Step 3: The cloud server of the energy management layer analyzes and stores the collected information, and transmits various types of historical information of the micro-grid to the virtual space modeling layer based on the TCP / IP protocol. The cloud server of the energy management layer monitors various types of data in real time based on the change range of various types of data in the historical database during normal operation. When the parameter exceeds the range, compare with the historical fault database information, and determine the type of fault based on the K- nearest neighbor method;

[0061] Step 4: The cloud server of the virtual space modeling layer analyzes the information transmitted from the entity micro-grid layer, and combines various types of historical data information transmitted by the energy management layer. Based on real-time and historical data information, data-driven virtual micro-grid system modeling is completed, and the established virtual micro-grid system maintains a 1:1 ratio with the original system and has the same life cycle as the micro-grid system;

[0062] Step 5: The energy management layer realizes the prediction of load, the prediction of electricity price and the prediction of load based on the cloud server;

[0063] The load prediction can realize long and short term prediction of the load by using the load information in the historical database and the real-time database. Firstly, the data characteristics are extracted, and the different characteristics are normalized. Due to the regional characteristics of the load, the processed load information is clustered by using the k-means clustering method. According to the clustering result, the historical sequence data and the real-time sequence data of each type of load are used to construct a time series model, and the neural network is trained. Based on the neural network, the short-term prediction of the load can be realized. When the load fluctuation is large, the prediction can be corrected combined with the expert experience, so as to enhance the accuracy of the prediction. The support vector machine can be used to realize the medium and long term prediction of the load by using the same method.

[0064] The electricity price prediction, the short-term electricity price prediction is not only related to the historical electricity price and the load value, but also related to the external factors such as weather. Therefore, the electricity price information in the historical database and the predicted load information, and the meteorological information communicated and transmitted by the meteorological station are combined to predict the short-term electricity price prediction model E(t) = O(t) + B(t) + W(t) + S(t) + R(t), wherein E(t) represents the electricity price at time t, O(t) represents the use of raw materials, B(t) represents the basic load component, W(t) represents the weather change, S(t) represents the special event. Based on the model, the real-time electricity price is predicted by using LSTM (Long Short Term Memory Network).

[0065] The power generation prediction includes the power generation prediction of each distributed power supply. The power generation prediction of the distributed power supply combines the mechanical performance, the light intensity, the temperature, the wind speed and other weather information to predict the power generation of each distributed power supply. The neural network is trained by using the meteorological information and the power generation information, and the power generation of each distributed power supply is predicted based on the neural network. The overall power generation prediction is the sum of the power generation prediction of each distributed power supply.

[0066] Step 6: The energy management layer determines the initial scheduling scheme based on the cloud server, and publishes the initial scheduling scheme related topic service;

[0067] The initial scheduling scheme is to ensure the reliability of power supply by controlling controllable load, diesel generator and energy storage system considering the relationship between distributed power output and load. The implementation is combined with the prediction module. When the distributed power output is greater than the load demand, the excess power can charge the battery. Considering the upper limit and service life, the virtual energy storage system composed of controllable load can absorb excess power, and at the same time, it can sell power to the grid during peak hours. When the distributed power output is insufficient to provide power for the load, the battery can be discharged when the battery power is not lower than the lower limit, and the virtual energy storage system can reduce power consumption without affecting normal use, so that the key load power supply is guaranteed, and the diesel generator can also be controlled to generate electricity to ensure reliable power supply. When the distribution network fails, disconnect the connection with the distribution network, and make the microgrid run in island mode. The difference between this mode and the above grid-connected mode is that the interaction with the distribution network is cut off, which is a self-sufficient way;

[0068] Step 7: The client of the virtual space modeling layer subscribes to the initial scheduling scheme, and controls the virtual microgrid system to run according to the initial scheduling scheme;

[0069] Step 8: The running information of the virtual microgrid system after executing the scheduling scheme is transmitted to the cloud server of the virtual space modeling layer through the gateway, and the cloud server of the virtual space modeling layer transmits the running information after executing the preliminary scheduling to the cloud server of the energy management layer based on TCP / IP protocol.

[0070] Step 9: The cloud server of the energy management layer adjusts the preliminary scheduling scheme based on the information fed back after the virtual microgrid system executes the preliminary scheduling scheme;

[0071] Step 10: The client of the virtual space modeling layer modifies the scheduling instruction according to the subscribed scheduling scheme topic service;

[0072] Step 11: The cloud server of the energy management layer determines the preliminary operation optimization scheme based on the scheduling scheme, and publishes the preliminary operation optimization scheme topic service;

[0073] The preliminary economic optimization operation scheme is to optimize the operation scheme based on the scheduling scheme. On the premise of ensuring the reliability of power supply, the microgrid operation optimization planning is optimized with the economic, minimum power purchase from the distribution network and minimum carbon emission as the target. Based on the virtual space modeling layer, the twin microgrid system model is established, the constraint conditions such as generator constraint, power balance constraint and battery constraint are combined, the optimization target is determined as the multi-objective function of meeting economy, minimum peak power purchase and minimum carbon emission, and the intelligent algorithm is used for optimization. Here, the improved particle swarm algorithm is used for optimization, and the preliminary optimization operation scheme is determined.

[0074] wherein the economic objective function is is the total cost of microgrid operation, N is the total number of dispatch segments, c is the price of diesel, f is the oil consumption of diesel generator, g is the electricity purchase price of microgrid from the distribution network, P g is the electricity purchase amount, S is the electricity sale price to the microgrid, P S is the electricity sale amount. Wherein gP g , sP S only exists in grid-connected operation, and is zero in island operation. M is the number of distributed energy sources, a is the generation cost of distributed power sources, wherein the renewable energy sources such as light and wind energy are excluded, P a is the power generation;

[0075] wherein the carbon emission objective function is is the total carbon emission, x is the carbon emission coefficient of diesel generator;

[0076] Step 12: the client of the virtual space modeling layer controls the virtual microgrid system to adjust according to the initial optimization scheme by subscribing to the initial operation optimization scheme;

[0077] Step 13: the operation information of the virtual microgrid system after executing the operation optimization scheme is transmitted to the cloud server of the virtual space modeling layer through the gateway, and the cloud server of the virtual space modeling layer transmits the operation information after executing the preliminary dispatch to the cloud server of the energy management layer based on the TCP / IP protocol;

[0078] Step 14: the cloud server of the energy management layer adjusts the preliminary operation optimization scheme based on the information fed back after the virtual microgrid system executes the preliminary operation optimization scheme.

[0079] Step 15: the client of the intelligent control layer controls the controller by subscribing to the operation optimization scheme, so as to control the microgrid system to operate according to the published scheme;

[0080] Step 16: the man-machine interaction layer can master various operation information of the microgrid by subscribing to various services published by the energy management layer, and the visual processing of data can be completed by designing and referencing software through programming and other means;

[0081] Step 17: in the client of the man-machine interaction layer, the operator can publish dispatch instructions to the server of the energy management layer according to the observed various operation information and own experience, and the clients of the virtual space modeling layer and the intelligent control layer can realize the control of the entity and the virtual microgrid system by subscribing to the control instructions published by the client of the man-machine interaction layer.

[0082] The above merely describes a preferred implementation method of the specific operation method of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1.A method for operating a digital-twin-based microgrid energy management system, the method comprising: Comprise the following steps; Step 1: Real-time acquisition of various types of information of the micro-grid system by intelligent sensors, cameras and tag-based radio frequency technology; Step 2: The collected various types of information are converted by the A / D conversion chip with model PCF8591 on board the FPGA with model EP4CE10F17C8, the converted digital quantity is temporarily stored in the FIFO module of the FPGA, and then the converted digital quantity information is serially transmitted to the intelligent gateway based on UART protocol through the RS232 serial port, and then the information is transmitted to the cloud server of the energy management layer and the virtual modeling layer through optical fiber wired network technology based on MQTT protocol; Step 3: The cloud server of the energy management layer parses and stores the collected information, transmits various types of historical information of the micro-grid to the virtual space modeling layer based on TCP / IP protocol, and simultaneously monitors various types of data in real time based on the change range of various types of data in the historical database during normal operation, compares the parameters exceeding the range with the information of the historical fault database, and determines the type of fault based on the K-neighbor method; Step 4: The cloud server of the virtual space modeling layer parses the information transmitted from the entity micro-grid layer, combines various types of historical data information transmitted from the energy management layer, and uses data driving to complete virtual micro-grid system modeling based on real-time and historical data information, the established virtual micro-grid system maintains a 1:1 ratio with the original system and has the same life cycle as the micro-grid system; Step 5: The energy management layer realizes the prediction of load, the prediction of electricity price and the prediction of power generation based on the cloud server; The load prediction can realize long-term and short-term prediction of load by using load information in the historical database and the real-time database, first extracting data features, normalizing different features, clustering the processed load information by using k-means clustering method due to the regional characteristics of load, according to the clustering results, using data to construct a time series model, and training a neural network, based on the neural network, short-term prediction of load can be realized, when the load fluctuation is large, the prediction can be corrected combined with expert experience to enhance the accuracy of prediction, and the same method can be used to realize the prediction of medium and long-term load by using support vector machine; The electricity price prediction, the short-term electricity price prediction is not only related to historical electricity price and load value, but also related to external factors such as weather; Therefore, the electricity price information in the historical database and the predicted load information, and the weather information communicated and transmitted by the weather station are combined for prediction, the short-term electricity price prediction model is E(t)=O(t)+B(t)+W(t)+S(t)+R(t), wherein E(t) represents the electricity price at time t, O(t) represents the use of raw materials, B(t) represents the basic load component, W(t) represents the weather change, S(t) represents special events; Real-time electricity price prediction is realized based on the model by using LSTM (Long Short Term Memory Network). The power generation prediction includes power generation predictions of each distributed power source, wherein the power generation prediction of the distributed power source is combined with mechanical performance, light intensity, temperature, wind speed and other weather information to predict the power generation of each distributed power source, the neural network is trained by using the meteorological information and the power generation information, and the power generation of each distributed power source is predicted based on the neural network, and the overall power generation prediction is the sum of the power generation predictions of each distributed power source; Step 6: The energy management layer determines the initial scheduling scheme based on the cloud server and provides the initial scheduling scheme subject service; The initial scheduling scheme considers the relationship between the distributed power output and the load, controls the controllable load, the diesel generator and the energy storage system to ensure the reliability of power supply, and is realized in combination with the prediction module. When the distributed power output is greater than the load demand power in the grid-connected operation, the excess power can charge the storage battery, considering the upper limit and service life, the virtual energy storage system composed of the controllable load can absorb the excess power, and at the same time, the virtual energy storage system can sell power to the grid during the peak electricity price period. When the distributed power output is insufficient to provide load power, when the power of the storage battery is not lower than the lower limit, the storage battery can be discharged, and at the same time, the virtual energy storage system can reduce power consumption under the condition of not affecting normal use, so that the key load power supply is guaranteed, and the diesel generator can also be controlled to generate power to ensure reliable power supply. When the distribution network fails, the connection with the distribution network is disconnected, so that the microgrid operates in an island operation mode. The difference between this mode and the above-mentioned grid-connected mode is that the interaction with the distribution network is cut off, and it is a self-sufficient mode. Step 7: The client of the virtual space modeling layer subscribes to the initial scheduling scheme, and controls the virtual microgrid system to operate according to the initial scheduling scheme; Step 8: The running information of the virtual microgrid system after executing the scheduling scheme is transmitted to the cloud server of the virtual space modeling layer through the gateway, and the cloud server of the virtual space modeling layer transmits the running information after executing the preliminary scheduling to the cloud server of the energy management layer based on the TCP / IP protocol; Step 9: The cloud server of the energy management layer adjusts the preliminary scheduling scheme based on the information fed back after the virtual microgrid system executes the preliminary scheduling scheme; Step 10: The client of the virtual space modeling layer modifies the scheduling instruction according to the subscribed scheduling scheme subject service; Step 11: The cloud server of the energy management layer determines the preliminary operation optimization scheme based on the scheduling scheme, and publishes the preliminary operation optimization scheme subject service; The preliminary economic optimization operation scheme is an optimization of the operation scheme on the basis of the scheduling scheme. The result predicted by the prediction module is combined, that is, under the premise of ensuring power supply reliability, the microgrid operation optimization planning is optimized in terms of economy, minimum power purchase from the distribution network, and minimum carbon emission. Based on the twin microgrid system model established by the virtual space modeling layer, the optimization target is determined as the multi-objective function of meeting economy, minimum peak power purchase, and minimum carbon emission, etc. in combination with the constraints of generator, power balance, battery, etc. An intelligent algorithm is used for optimization. Here, the improved particle swarm algorithm is used for optimization, thereby determining the preliminary optimization operation scheme; Wherein, the economic objective function is is the total cost of micro-grid operation, N is the total number of scheduling segments, c is the price of diesel, f is the oil consumption of diesel generator, g is the purchase price of electricity from the distribution network, P g is the purchase of electricity, s is the price of selling electricity to the micro-grid, P S is the amount of electricity sold; wherein gP g , sP S Only exists in grid-connected operation, in island operation is zero; M is the number of distributed energy, a is the generation cost of distributed power supply, wherein, except for renewable energy such as light energy, wind energy, P a is the power generation; Wherein, the carbon emission target function is is the total carbon emissions, x is the diesel generator carbon emission coefficient; Step 12: The client of the virtual space modeling layer subscribes to the initial operation optimization scheme, and controls the virtual microgrid system to adjust according to the initial optimization scheme; Step 13: The running information of the virtual microgrid system after executing the operation optimization scheme is transmitted to the cloud server of the virtual space modeling layer through the gateway. The cloud server of the virtual space modeling layer transmits the running information after executing the preliminary scheduling to the cloud server of the energy management layer based on the TCP / IP protocol; Step 14: The cloud server of the energy management layer adjusts the preliminary operation optimization scheme based on the information fed back after the virtual microgrid system executes the preliminary operation optimization scheme; Step 15: The client of the intelligent control layer controls the controller by subscribing to the operation optimization scheme, thereby realizing the control of the microgrid system to run according to the published scheme; Step 16: The man-machine interaction layer can master various running information of the microgrid by subscribing to various services published by the energy management layer, and can complete the visual processing of data by designing and referencing software through programming; Step 17: In the client of the man-machine interaction layer, the operator can publish scheduling instructions to the server of the energy management layer according to the observed various running information and own experience. The clients of the virtual space modeling layer and the intelligent control layer realize the control of the entity and the virtual microgrid system by subscribing to the control instructions published by the client of the man-machine interaction layer.

Images