A hybrid energy network strategy application system based on multi-energy complementary energy hub

By adopting a hybrid energy network strategy application system with deep neural networks in the energy Internet, and using LSTM neural networks for system evaluation and parameter optimization, the real-time monitoring and optimization of multi-energy systems is solved, and more efficient energy management and system performance is achieved promote.

CN113516268BActive Publication Date: 2024-11-19CHINA ELECTRIC POWER RESEARCH INSTITUTE CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202010278063.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-04-10
Publication Date
2024-11-19
Estimated Expiration
2040-04-10

AI Technical Summary

Technical Problem

There is a lack of scientific evaluation and monitoring methods for the systematic nature of multi-energy comprehensive systems in the energy Internet, and it is difficult to carry out real-time operation management and system parameter optimization under the conditions of diverse forms, complex scenarios and multi-objective constraints of multi-energy complementary systems.

Method used

Adopt a hybrid energy network strategy application system based on deep neural networks, and use long and short memory neural networks (LSTMs) to realize a two-way deep two-way long and short memory neural network, and retain the "past" and "future" information through memory cells to enhance the prediction capabilities of the neural network. And optimize parameter distribution through multi-layer architecture to achieve real-time online multi-objective optimization.

Benefits of technology

Effectively manage and optimize energy system parameters, improve the real-time operation capability and multi-objective optimization capability of the integrated energy system, and improve the flexibility and applicability of the system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113516268B_ABST
    Figure CN113516268B_ABST
Patent Text Reader

Abstract

The present invention relates to a hybrid energy network strategy application system based on a multi-energy complementary energy hub, comprising: a network construction unit, a data acquisition unit, an interface transmission unit, a data storage unit, a hybrid energy management unit, a comprehensive effect evaluation unit, a strategy application unit, and an effectiveness detection unit. The standardized distribution of parameters of the integrated energy system is achieved through a multi-layer architecture to better utilize the parameters, reduce the number of memories, increase the nonlinear processing of input data and parameters of the energy integrated system, and thus facilitate the integrated energy system to apply better strategies in the real-time operation process.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of energy internet, and in particular to a hybrid energy network strategy application system based on a multi-energy complementary energy hub. Background Art

[0002] With the rapid development of the global economy, the energy crisis and environmental crisis brought about by the large-scale development and utilization of fossil energy have become more prominent. The industrial civilization based on the traditional use of fossil energy has gradually fallen into trouble, threatening the sustainable development of mankind. Therefore, the call for developing low-carbon cities, building distributed energy systems in urban areas, and carrying out energy planning according to local conditions has become increasingly strong.

[0003] At present, many places have used the complementarity between various energy sources to achieve multi-energy complementarity, with the purpose of achieving the integration, coordination and linkage of multiple energy sources, thereby promoting the comprehensive utilization rate of energy and the absorption capacity of renewable energy. The energy hubs in the energy Internet are diverse in form and complex in scenarios. There are different forms of energy hub systems such as cold, heat and electricity combined supply systems, solar biogas power supply hybrid systems, and different application scenarios such as parks, smart ecological cities, and living communities. How to effectively monitor and evaluate the carbon emissions and energy efficiency of the multi-energy complementary system is one of the key and difficult issues of the multi-energy complementary system of the energy Internet.

[0004] In view of the diverse forms, complex scenarios, and multi-objective constraints of the multi-energy complementary hub system of the energy Internet, the current methods for solving this difficult problem are mainly divided into two categories: one is the basic physical modeling method, and the other is the data-driven method. The accuracy of the physical model method mainly depends on the domain knowledge and modeling accuracy of the actual energy hub system. However, since the comprehensive utilization efficiency of the energy hub and the consumption of renewable energy are also affected by factors such as the climate environment of the application area, the scale of use, and the user demand response, although this method has certain advantages in the early static planning, it is not universal enough in the practical application of the project, and there are shortcomings such as long dynamic adjustment of system parameters and insufficient dynamic adjustment with environmental changes. The data-driven method mainly uses the analysis of real-time collected data during the actual operation of the system to dynamically plan and dynamically adjust the parameters of the multi-energy complementary energy hub system. It has high flexibility and versatility, so it has strong applicability in project practice. With the development of evolutionary technology, modern heuristic algorithms such as genetic algorithms, ant colony algorithms, particle swarm optimization algorithms, and artificial neural network algorithms have aroused people's interest in research and have shown great application potential.

[0005] However, in the process of building a multi-energy integrated system in the Energy Internet, there is still a lack of systematic scientific evaluation and monitoring methods for the multi-energy integrated system. This method can target the diverse forms, complex scenarios, and multi-objective constraints of the multi-energy complementary system in the Energy Internet, enable the integrated energy system to be more effectively managed during real-time operation, facilitate the optimization of system parameter configuration, and improve the real-time online multi-objective optimization of the integrated energy system. Summary of the invention

[0006] From the perspective of deep neural networks, the present invention utilizes the characteristics of strong versatility and wide expansion scale of neural networks to propose a hybrid energy network strategy application system based on a multi-energy complementary energy hub. It is based on the deep bidirectional long short-term memory neural network based on the long short-term memory neural network designed by the present invention. It is an important branch of the development of recursive neural networks. The ingenious design of memory units, input gates, forget gates, and output gates alleviates the gradient explosion and gradient mass phenomena in the RNN neural network. At the same time, it has relatively strong generalization ability and long-term memory ability, and can accurately realize the dynamic comprehensive management evaluation of the energy Internet multi-energy complementary hub and the online dynamic system parameter optimization configuration.

[0007] A hybrid energy network strategy application system based on a multi-energy complementary energy hub, characterized by comprising:

[0008] Network construction unit, used to build a multi-energy complementary energy hub micro-energy network;

[0009] A data collection unit, used to collect various required data information, wherein the various data include energy input data and load output data of the multi-energy complementary energy hub micro energy network;

[0010] An interface transmission unit, used to configure interface types corresponding to different data collection and transmission;

[0011] A data storage unit, used to receive and store the collected energy input data and load output data, and pre-process the data;

[0012] Hybrid energy management unit, used to build a neural network model of micro energy network and predict the allocation coefficients of various energy inputs corresponding to load demand;

[0013] A comprehensive energy efficiency evaluation unit, which is used to perform calculations and judgment analysis based on the distribution coefficients of the various energy inputs and the collected data information, predict the comprehensive energy efficiency level of the micro energy network, and generate an evaluation analysis report;

[0014] A strategy application unit, used to formulate a hybrid energy input control strategy according to the evaluation and analysis report, and apply the strategy to the multi-energy complementary energy hub micro-energy network;

[0015] The actual effect detection unit is used to detect the actual effect of the hybrid energy input control strategy.

[0016] The present invention aims to solve the problem of lack of systematic scientific evaluation and monitoring methods for multi-energy integrated systems in the construction of energy Internet multi-energy integrated systems. In view of the diverse forms, complex scenarios, and multi-objective constraints of the multi-energy complementary system of the energy Internet, a data-driven method is used to expand the memory characteristics of the LSTM neural network (long short-term memory neural network) to realize a bidirectional long short-term memory neural network. The memory cells in the LSTM are used to retain the information of the "past" and "future" at the same time, and the prediction ability of the neural network in the system monitoring and evaluation process is strengthened. On the basis of the bidirectional long short-term memory neural network, a deep architecture of the bidirectional long short-term memory neural network is realized, and the standardized distribution of parameters of the integrated energy system is realized through a multi-layer architecture, so as to better utilize the parameters, reduce the number of memories, and increase the nonlinear processing of the input data and parameters of the energy integrated system, so as to facilitate the integrated energy system to be more effectively managed in the real-time operation process, facilitate the optimization of the system parameter configuration, and improve the real-time online multi-objective optimization of the integrated energy system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Embodiments of the invention are illustrated by way of example and not limitation in the accompanying drawings in which like reference numerals refer to like elements and in which:

[0018] Figure 1 It is a system framework diagram of the present invention;

[0019] Figure 2 It is the deep bidirectional long short-term memory neural network model of the present invention. DETAILED DESCRIPTION

[0020] In the following description, for the purpose of illustration, in order to enable a thorough understanding of the embodiments of the present invention, many specific details are set forth. However, it is apparent to those skilled in the art that the present invention can be implemented without these specific details. The following detailed description includes references to the accompanying drawings, which constitute a part of the detailed description. The accompanying drawings show legends according to exemplary embodiments. These exemplary embodiments (which are also referred to herein as "examples") are described in sufficient detail to enable those skilled in the art to implement the present subject matter. The embodiments may be combined, other embodiments may be utilized, or structural and logical changes may be made without departing from the scope of the claims. Therefore, the following detailed description should not be construed as limiting, and the scope is determined by the appended claims and their equivalents.

[0021] like Figure 1 As shown, the present invention proposes a hybrid energy network strategy application system based on a multi-energy complementary energy hub, including:

[0022] Network construction unit, used to build a multi-energy complementary energy hub micro-energy network;

[0023] A data collection unit, used to collect various required data information, wherein the various data include energy input data and load output data of the multi-energy complementary energy hub micro energy network;

[0024] An interface transmission unit, used to configure interface types corresponding to different data collection and transmission;

[0025] A data storage unit, used to receive and store the collected energy input data and load output data, and pre-process the data;

[0026] Hybrid energy management unit, used to build a neural network model of micro energy network and predict the allocation coefficients of various energy inputs corresponding to load demand;

[0027] A comprehensive energy efficiency evaluation unit, which is used to perform calculations and judgment analysis based on the distribution coefficients of the various energy inputs and the collected data information, predict the comprehensive energy efficiency level of the micro energy network, and generate an evaluation analysis report;

[0028] A strategy application unit, used to formulate a hybrid energy input control strategy according to the evaluation and analysis report, and apply the strategy to the multi-energy complementary energy hub micro-energy network;

[0029] The actual effect detection unit is used to detect the actual effect of the hybrid energy input control strategy.

[0030] Preferably, the multi-energy complementary energy hub micro-energy network constructed by the network construction unit is a solar-biogas micro-energy network composed of solar thermal collection, photovoltaic power generation, cogeneration units and biogas tanks.

[0031] Preferably, the data storage unit preprocesses the data, specifically including: prejudging the rationality of the data according to a threshold value, and discarding the data if abnormal, otherwise retaining the data.

[0032] Preferably, the energy input data collected by the data collection unit includes load demand variables, environmental parameters, and device node data of the micro energy network;

[0033] The environmental parameters include lighting conditions, average ambient temperature, photovoltaic panel power temperature, and biogas concentration per unit volume; the device node data is collected by sensors, including system energy input, system configuration parameter quantity, energy carbon emissions, and energy transfer efficiency.

[0034] Preferably, the load output data collected by the data collection unit includes the power load output, cooling load output, heating load output and gas load output of the micro energy network.

[0035] Preferably, the hybrid energy management unit is used to construct a micro-energy network neural network model to predict the distribution coefficients of various energy inputs corresponding to load demand, specifically including:

[0036] A sample preparation module, used to form a training data sample of a neural network model of multi-input and multi-output of the micro energy network according to the energy input data and the load output data;

[0037] A training construction module is used to determine the energy input distribution coefficients of the micro energy network, and to construct a micro energy network neural network model that conforms to the energy input and load output relationship of the micro energy network environment in combination with the neural network model training data samples;

[0038] The prediction output module is used to use the constructed micro-energy network neural network model to predict the output of the neural network model, that is, the distribution coefficient of various energy inputs, by inputting load demand; wherein the energy input data is used as the input of the neural network model, and the output of the neural network model is the distribution coefficient of the load demand corresponding to various energy inputs.

[0039] Preferably, the training construction module is used to determine the energy input distribution coefficients of the micro energy network, and to construct a micro energy network neural network model model that conforms to the relationship between energy input and load output of the micro energy network environment in combination with the neural network model training data sample, specifically including:

[0040] Basic building blocks for building basic deep bidirectional long short-term memory neural network models:

[0041] Multi-layer bidirectional long short-term memory neural network models are stacked to form a deep bidirectional long short-term memory neural network model, and the basic formula of the deep bidirectional long short-term memory neural network model is obtained:

[0042]

[0043]

[0044]

[0045]

[0046]

[0047]

[0048] in, g, They represent the output values ​​of the input gate, forget gate, current input unit transition state and output gate at time t in the first layer of the neural network respectively;

[0049] Respectively represent the weight matrices of the input gate, forget gate, current input unit transition state, and output gate in the first layer of the neural network;

[0050] They represent the bias items of the input gate, forget gate, current input unit state and output gate in the first layer of the neural network respectively;

[0051] They represent the states of the current hidden layer and the previous hidden layer in the first layer of the neural network respectively;

[0052] They represent the states of the previous layer and the current layer of the cell in the first layer of the neural network respectively;

[0053] In a deep bidirectional long short-term memory neural network, the first layer takes feature data as input, and the input of each other layer is the output of the previous layer;

[0054] The characteristic data include grid output power, solar energy output power and biogas output power;

[0055] In the formula, σ is the activation function, is the feature data input to the neural network layer l at time t;

[0056] The forward propagation building block is used to build a forward propagation multi-layer long short-term memory neural network model:

[0057]

[0058]

[0059]

[0060]

[0061]

[0062]

[0063] Backpropagation building block for building backpropagation multi-layer long short-term memory neural network models:

[0064]

[0065]

[0066]

[0067]

[0068]

[0069]

[0070] The arrow → indicates the forward propagation of the multi-layer long short-term memory neural network to obtain the output value The arrow ← indicates the backward propagation of the multi-layer long short-term memory neural network to obtain the output value

[0071] Combined building block for forward propagating the output values ​​of the multi-layer long short-term memory neural network model and backward propagating the output values ​​of the multi-layer long short-term memory neural network model and Combine and construct the final deep bidirectional long short-term memory neural network model, that is, the micro-energy network neural network model, and obtain the output result:

[0072]

[0073] The deep bidirectional long short-term memory neural network model outputs s t is the distribution coefficient of various energy inputs corresponding to the load demand. The deep bidirectional long short-term memory neural network model outputs s t Yes, the load demand can be trained by inputting characteristic data and historical data of load demand to obtain the neural network. The value of the weight, the distribution coefficient of each type of energy power input corresponding to the value of the weight W.

[0074] The deep bidirectional long short-term memory neural network model was built using Google's Tensorflow, with the dropout value set to 0.5, the batch_size of the training set to 32, the batch-size of the test set to 64, the dimension of the hidden state in the forward and backward directions of Bi-LSTM to 100, the training learning rate to 0.001, and the Adam (adaptive moment estimation) optimizer. Before each round of training, all training data were shuffled.

[0075] Through simulation and comparative tests of CNN, LSTM, and Deep Bi-LSTM neural networks, the recall rate, accuracy rate, and F1 value are used to evaluate the model. The experimental results are compared as follows:

[0076] Accuracy (%) Recall rate (%) F value (%) CNN 86.93 85.38 86.79 LSTM 87.28 86.12 86.63 Deep Bi-LSTM 92.43 89.67 90.89

[0077] Through experimental comparison results, it can be found that the method based on the Deep Bi-LSTM deep bidirectional long short-term memory neural network model is better than the CNN and LSTM methods as a whole, because the deep bidirectional long short-term memory neural network model can obtain features from the sequence forward and backward, so that more comprehensive sequence knowledge information can be obtained. At the same time, combined with the deep neural network, a deep bidirectional long short-term memory neural network model is constructed, which enables the model to better obtain sequence features.

[0078] The present invention solves the problem of lack of systematic scientific evaluation and monitoring methods for multi-energy integrated systems in the process of building a multi-energy integrated system of the energy Internet. Aiming at the characteristics of the multi-energy complementary system of the energy Internet, such as diverse forms, complex scenarios, and multi-objective constraints, the present invention uses a data-driven method to expand the memory characteristics of the LSTM neural network (long short-term memory neural network) and realize a bidirectional long short-term memory neural network. The memory cells in the LSTM are used to retain the information of the "past" and "future" at the same time, and the prediction ability of the neural network in the process of system monitoring and evaluation is enhanced. On the basis of the bidirectional long short-term memory neural network, the deep architecture of the bidirectional long short-term memory neural network is realized, and the standardized distribution of parameters of the integrated energy system is realized through a multi-layer architecture, so as to better utilize the parameters, reduce the number of memories, and increase the nonlinear processing of the input data and parameters of the energy integrated system, so as to facilitate the integrated energy system to be more effectively monitored during real-time operation, facilitate the optimization of the system parameter configuration, and improve the real-time online multi-objective optimization of the integrated energy system.

[0079] Although the present invention has been described with reference to specific exemplary embodiments, it is apparent that various modifications and changes may be made to these embodiments without departing from the broader spirit and scope of the inventive subject matter. Therefore, the present specification and drawings should be regarded as illustrative rather than restrictive.

[0080] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0081] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0082] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0083] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0084] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A hybrid energy network strategy application system based on a multi-energy complementary energy hub, characterized in that: include: Network construction unit, used to build a multi-energy complementary energy hub micro-energy network; A data collection unit, used to collect various required data information, wherein the various data include energy input data and load output data of the multi-energy complementary energy hub micro energy network; An interface transmission unit, used to configure interface types corresponding to different data collection and transmission; A data storage unit, used to receive and store the collected energy input data and load output data, and pre-process the data; Hybrid energy management unit, used to build a neural network model of micro energy network and predict the allocation coefficients of various energy inputs corresponding to load demand; A comprehensive energy efficiency evaluation unit, which is used to perform calculations and judgment analysis based on the distribution coefficients of the various energy inputs and the collected data information, predict the comprehensive energy efficiency level of the micro energy network, and generate an evaluation analysis report; A strategy application unit, used to formulate a hybrid energy input control strategy according to the evaluation and analysis report, and apply the strategy to the multi-energy complementary energy hub micro-energy network; An actual effect detection unit, used to detect the actual effect of the hybrid energy input control strategy; Among them, the training construction module is used to determine the energy input distribution coefficients of the micro energy network, and to build a micro energy network neural network model that conforms to the relationship between energy input and load output of the micro energy network environment in combination with the neural network model training data samples, specifically including: Basic building blocks for building basic deep bidirectional long short-term memory neural network models: Multi-layer bidirectional long short-term memory neural network models are stacked to form a deep bidirectional long short-term memory neural network model, and the basic formula of the deep bidirectional long short-term memory neural network model is obtained: in, g, They represent the output values ​​of the input gate, forget gate, current input unit transition state and output gate at time t in the lth layer of the neural network respectively; Respectively represent the weight matrices of the input gate, forget gate, current input unit transition state, and output gate in the lth layer of the neural network; They represent the bias items of the input gate, forget gate, current input unit state and output gate in the lth layer of the neural network respectively; They represent the states of the current hidden layer and the previous hidden layer in the lth layer of the neural network respectively; They represent the states of the previous layer and the current layer of the cell in the lth layer of the neural network respectively; In a deep bidirectional long short-term memory neural network, the first layer takes feature data as input, and the input of each other layer is the output of the previous layer; The characteristic data include grid output power, solar energy output power and biogas output power; In the formula, σ is the activation function, is the feature data input to the lth layer of the neural network at time t; The forward propagation building block is used to build a forward propagation multi-layer long short-term memory neural network model: Backpropagation building module, used to build a backpropagation multi-layer long short-term memory neural network model: The arrow → indicates the forward propagation of the multi-layer long short-term memory neural network to obtain the output value The arrow ← indicates the backward propagation of the multi-layer long short-term memory neural network to obtain the output value Combined building block for forward propagating the output values ​​of the multi-layer long short-term memory neural network model and backward propagating the output values ​​of the multi-layer long short-term memory neural network model and Combine and construct the final deep bidirectional long short-term memory neural network model, that is, the micro-energy network neural network model, and obtain the output result: The deep bidirectional long short-term memory neural network model outputs s t It is the allocation coefficient of various energy inputs corresponding to load demand.

2. The system of claim 1, wherein: The multi-energy complementary energy hub micro-energy network constructed by the network construction unit is a solar-biogas micro-energy network composed of solar thermal collection, photovoltaic power generation, cogeneration units and biogas tanks.

3. The system of claim 1, wherein: The data storage unit pre-processes the data, specifically including: pre-judging the rationality of the data according to a threshold value, and removing the data if abnormal, otherwise retaining the data.

4. The system of claim 1, wherein: The energy input data collected by the data collection unit includes load demand variables, environmental parameters, and device node data of the micro energy network; The environmental parameters include lighting conditions, average ambient temperature, photovoltaic panel power temperature, and biogas concentration per unit volume; the device node data is collected by sensors, including system energy input, system configuration parameter quantity, energy carbon emissions, and energy transfer efficiency.

5. The system of claim 1, wherein: The load output data collected by the data collection unit includes the power load output, cooling load output, heating load output, and gas load output of the micro energy network.

6. The system according to any one of claims 1 to 5, wherein: The hybrid energy management unit is used to construct a micro-energy network neural network model to predict the distribution coefficients of various energy inputs corresponding to load demand, specifically including: A sample preparation module, used to form a training data sample of a neural network model of multi-input and multi-output of the micro energy network according to the energy input data and the load output data; A training construction module is used to determine the energy input distribution coefficients of the micro energy network, and to construct a micro energy network neural network model that conforms to the energy input and load output relationship of the micro energy network environment in combination with the neural network model training data samples; The prediction output module is used to use the constructed micro-energy network neural network model to predict the output of the neural network model, that is, the distribution coefficient of various energy inputs, by inputting load demand; wherein the energy input data is used as the input of the neural network model, and the output of the neural network model is the distribution coefficient of the load demand corresponding to various energy inputs.