Full-life-cycle risk assessment and multi-dimensional toughness optimization method for intelligent micro-grid in uncertain environment

Abstract

The invention discloses a full-life-cycle risk assessment and multi-dimensional toughness optimization method for an intelligent micro-grid in an uncertain environment. The method can effectively solve the problems that in existing intelligent micro-grid operation and maintenance research, a power equipment layer and a data layer are not tightly coupled, real fault data are scarce and unbalanced, the generalization ability of a risk assessment model is limited, toughness assessment still needs to be refined and the like. A short-term fault diagnosis model based on combination of transfer learning and a support vector machine (TL-SVM), a fault diagnosis model based on multi-dimensional ensemble learning and a long-term fault diagnosis model based on a convolutional neural network of a double-end attention mechanism and a bidirectional gating circulation unit (CNN-BiGRU-Bi-Attention) are respectively provided, so that risk dynamic changes are accurately captured; a risk interval prediction model based on a Bootstrap method in an uncertain environment is provided; on the basis, a risk informed importance concept is introduced to evaluate the maintenance priority of the intelligent micro-grid equipment; meanwhile, a multi-dimensional toughness evaluation method is provided based on the perspective of a power supply chain, and toughness optimization is carried out on the intelligent micro-grid by utilizing the risk informed importance degree. The method provides powerful support for safe and stable operation and toughness improvement of the intelligent micro-grid, and is an innovation in risk assessment and toughness optimization construction methods of the intelligent micro-grid.

Description

1. Technical Field

[0001] The present invention relates to the field of health management of smart microgrids, and in particular to a full life cycle risk assessment and multi-dimensional resilience optimization method for smart microgrids under uncertain environments. 2. Background Technology

[0002] With the severe challenges of global climate change, the increasing demand for energy and the sharp increase in power load demand, building a supply and demand system based on renewable energy such as wind energy, solar energy and hydropower has become an important development direction of the global energy revolution. In this context, microgrids, as an emerging power structure, have the characteristics of high integration, intelligent control, flexibility and scalability, and can effectively solve the problem of renewable energy consumption. Its importance is becoming increasingly prominent. The widespread application of microgrids is also accompanied by challenges that cannot be ignored, especially the problem that they are easily restricted by external factors such as weather conditions and geographical location. In a complex and changeable natural environment, problems such as the reliability of power supply, delays in information processing and large-scale power outages may have a serious impact on the operation of microgrids.

[0003] The booming development of Internet of Things (IoT) technology has provided new solutions for the operation and monitoring of smart microgrids. IoT technology has realized real-time monitoring, data collection, intelligent analysis and remote control of each operation link of microgrids through sensors, smart devices, cloud computing and big data analysis, significantly improving the transparency and controllability of microgrid systems. This not only helps to timely discover and deal with potential faults, but also optimizes energy configuration and scheduling strategies based on real-time data, and improves the reliability and stability of power supply. However, in the practice of smart microgrids driven by IoT technology, there are still many technical difficulties and challenges. On the one hand, the loose coupling between the power equipment layer and the data layer limits the effective transmission and processing of data, affecting the overall performance and efficiency of the system. On the other hand, the generalization ability of the risk assessment model is limited, and it is difficult to accurately predict and respond to various uncertain factors, such as extreme weather and equipment failure. In addition, the refinement of resilience assessment and the perfection of risk-based resilience optimization methods also directly affect the adaptability and recovery ability of microgrids in complex and changing environments. Therefore, it is imperative to invent a full life cycle risk assessment and multi-dimensional resilience optimization method for smart microgrids under uncertain environments. III. Summary of the invention

[0004] In view of the above situation, in order to solve the defects of the prior art, the purpose of the present invention is to provide a full life cycle risk assessment and multi-dimensional resilience optimization method for smart microgrids under uncertain environments, which can effectively solve the problems existing in the existing smart microgrid health management research, such as the loose coupling between the power equipment layer and the data layer, the limited generalization ability of the risk assessment model, the resilience assessment that needs to be refined, and the risk-based resilience optimization method that needs to be improved.

[0005] A full life cycle risk assessment and multi-dimensional resilience optimization method for smart microgrids under uncertainty environment includes the following steps:

[0006] S100: Use electrical betweenness and communication betweenness to analyze the failure mechanisms of power equipment and communication equipment in smart microgrids respectively; among them, physical layer failures mainly involve failures of power equipment in the power supply chain, including power generation equipment failures, transmission line failures, distribution equipment failures, and energy storage system failures, etc. Data layer failures mainly include failures of communication links in smart microgrids, including failures of perception layer communication equipment, edge cloud layer communication equipment, and central cloud layer communication equipment; on this basis, analyze the coordinated failures of the physical layer and data layer of smart microgrids;

[0007] S200: Use VGG16 and support vector machine (SVM) combined model to fuse multi-source heterogeneous data; considering the nonlinear separability and multimodal characteristics of multi-source heterogeneous data, in view of VGG16's excellent ability in capturing the deep characteristics of the operating status of power equipment, combined with the advantage of SVM's efficient classification in feature space, the VGG16-SVM model is used to process multi-source heterogeneous data, that is, convert multi-source heterogeneous data into distributed data format and upload it to the cloud for processing, so as to improve data processing efficiency;

[0008] S300: Use the fault diagnosis model considering the whole life cycle to perform short-term, medium-term and long-term fault diagnosis on the smart microgrid respectively; among them, a short-term fault diagnosis model based on transfer learning and SVM, a medium-term fault diagnosis model based on multi-dimensional ensemble learning, and a long-term fault diagnosis model based on convolutional neural network and bidirectional gated recurrent unit (CNN-BiGRU-Bi-Attention) based on dual-head attention mechanism are constructed to accurately capture the dynamic changes of risks;

[0009] S400: A two-stage interval prediction method combining convolutional neural bidirectional long short-term memory network and Bootstrap method is used to predict the risk interval;

[0010] S500: Based on the perspective of the power supply chain, a multi-dimensional resilience assessment model for smart microgrids is constructed. The resilience of smart microgrids is comprehensively assessed from three dimensions: power supply chain, time, and power supply capacity, facilitating rapid fault location and rapid execution of maintenance decisions.

[0011] S600: Use risk-informed importance to determine maintenance priorities for smart microgrid equipment, making maintenance activities more targeted and efficient by comprehensively considering factors such as reliability, risk, and cost.

[0012] In step S100, the electrical intermediate value is expressed as follows:

[0013]

[0014] Where e(a,b) represents the electrical between branches (a,b), V g Represents the set of power generation nodes, V L represents the set of load nodes, W i is the actual power generation of the power generation node, W j is the actual load of the load node, I ij (a, b) represents the current on branch (a, b) when (i, h) injects a unit current element;

[0015] The communication betweenness is expressed as follows:

[0016]

[0017] Where l(v) represents the data layer node v(v∈V D ) load capacity, δ ij represents the number of shortest paths from node i to node j in the data layer, δ ij (v) represents the number of shortest paths from node i to node j passing through node v in the data layer.

[0018] The process of coordinated failure of the physical layer and the data layer is as follows: when a power node in the power network suddenly fails and stops operating, the power grid's power flow distribution will be severely impacted, which is likely to lead to a crisis of power flow exceeding the limit. At this critical juncture, the control center in the data layer will quickly intervene and send emergency instructions to each relevant node through an efficient data channel, requiring immediate adjustment of power generation, implementation of load reduction and other response measures to quickly restore the supply and demand balance of the power grid, reduce overload pressure, and strive to restore the power grid to a stable operating state in the shortest time. If the failure of the power node lasts too long, it may trigger a chain reaction, affecting those perception nodes in the data layer that also rely on power supply. The failure of these nodes will greatly weaken the real-time perception capability of the power grid status, thereby increasing the uncertainty and risk of microgrid operation. When the key nodes in the data layer fail for some reason, the power nodes closely associated with them will lose effective control instructions, which may cause these power nodes to fail to operate as expected, thereby triggering cascading failures.

[0019] In step S200, the multi-source heterogeneous data fusion method is: first, heterogeneous data such as sensors and images of the smart microgrid operation from different sources are collected; second, the pre-trained VGG16 model is loaded and the fully connected layer in the VGG16 model is removed, and then the feature vector output by VGG16 is used as the input feature of the SVM; finally, the multi-source heterogeneous data is converted into a distributed data format and uploaded to the cloud for processing of the multi-source heterogeneous data, so as to improve the data processing efficiency.

[0020] In step S300, the loss function of the short-term fault diagnosis model based on the combination of transfer learning and SVM is expressed as follows:

[0021]

[0022] Among them, L(h) represents the total loss function, D(h(x i ),y i ) represents the predicted value h(x) under the current environment i ) and the label value y in the real environment i The error between It represents the predicted value under the historical environment Tag values ​​in historical context The error between them, η is the loss balance parameter.

[0023] In step S300, the mid-term fault diagnosis model based on multi-dimensional ensemble learning adopts the following expression:

[0024]

[0025] Among them, n α is the number of samples of the αth microgrid device, Mα is the number of base learners of the αth microgrid device, is the weight of the mth base learner. When the αth microgrid device fails, on the contrary, T * Indicates the threshold at which a fault occurs.

[0026] In step S300, the long-term fault diagnosis model based on CNN-BiGRU-Bi-Attention is as follows: first, CNN is used to extract high-dimensional feature vectors in sensor data, and BiGRU is used to capture long-term temporal dependencies in high-dimensional feature vectors to enhance the temporal modeling capability of the model; secondly, the Bi-Attention mechanism is introduced to simultaneously apply attention to different dimensions of the time series to assign different weight values ​​to the hidden states output by the BiGRU layer; finally, the final fault diagnosis result of the smart microgrid device is output through the fully connected layer.

[0027] In step S400, the proposed two-stage interval prediction method combining the convolutional neural bidirectional long short-term memory network and the Bootstrap method is: first, the advantages of CNN in capturing local features and the advantages of BiLSTM in capturing long-term dependencies in the sequence are used to perform point predictions on the covariates; secondly, the Bootstrap method is used to analyze the prediction errors of the covariate data and calculate the distribution boundary values ​​of the prediction errors under a given confidence interval.

[0028] In step S500, the proposed smart microgrid resilience index R S Take the following expression:

[0029]

[0030] Among them, T is the duration of the study, N t is the number of power consumption nodes at time t, Whether the effective load value of a certain power consumption node meets the specified load value at a specific research time, if it does, Otherwise, it is 0. The proposed resilience index uses the load supply rate as a dynamic indicator to reflect the power supply capacity of the smart microgrid.

[0031] In step S600, the proposed risk awareness importance is expressed as follows:

[0032]

[0033] in, is the risk-informed importance of the kth device at time t, C(t) is the expected cost of maintaining the smart microgrid system, F(t) is the risk function of the smart microgrid system, and F k (t) is the risk function of the kth device at time t.

[0034] The present invention proposes a full life cycle risk assessment and multi-dimensional resilience optimization method for smart microgrids from a data-driven perspective. Through failure mechanism analysis, full life cycle risk assessment, multi-dimensional resilience assessment and resilience optimization method based on risk-informed importance, it provides strong support for the safe and stable operation and resilience improvement of smart microgrids, introduces a new decision-making framework for the health management and predictive maintenance of smart microgrids, and is an innovation in the construction method of risk assessment and resilience optimization model of smart microgrids. IV. Description of the drawings

[0035] Figure 1 The figure is a schematic diagram of the process of constructing the method of the present invention.

[0036] Figure 2 A schematic diagram of a cascading fault propagation process provided by an embodiment of the present invention.

[0037] Figure 3 A schematic diagram of the comparison results of the confusion matrix of the proposed short-term fault diagnosis model and the SVM model provided in an embodiment of the present invention.

[0038] Figure 4 A schematic diagram of confusion matrix results of a mid-term fault diagnosis model for a wind turbine and a photovoltaic generator provided in an embodiment of the present invention.

[0039] Figure 5 Schematic diagram of different mid-term fault diagnosis methods provided by embodiments of the present invention.

[0040] Figure 6 A schematic diagram of confusion matrix results of a long-term fault diagnosis model for a wind turbine and a photovoltaic generator provided in an embodiment of the present invention.

[0041] Figure 7 A schematic diagram of a risk probability prediction interval for a risk generator provided in an embodiment of the present invention.

[0042] Figure 8 Schematic diagram of six typical power supply chain topologies provided by embodiments of the present invention.

[0043] Fig. 9 A schematic diagram of the multi-dimensional resilience evaluation results of six power supply chains provided in an embodiment of the present invention.

[0044] Fig.10 A schematic diagram of cost increment results of different maintenance sequences provided by an embodiment of the present invention. V. Specific implementation methods

[0045] The specific implementation of the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. Please note that although specific embodiments are given here as reference, the essence of the present invention is not limited to these embodiments, but can be adjusted and expanded in a variety of ways according to actual needs. The main purpose of these examples is to allow readers to have a deeper understanding of the present invention, thereby promoting the comprehensive understanding and flexible application of the technical solutions of the present invention by technicians.

[0046] like Figure 1 As shown, the present invention, when implemented, includes the following steps:

[0047] S100: Use electrical betweenness and communication betweenness to analyze the failure mechanisms of power equipment and communication equipment in smart microgrids respectively; among them, physical layer failures mainly involve failures of power equipment in the power supply chain, including power generation equipment failures, transmission line failures, distribution equipment failures, and energy storage system failures, etc. Data layer failures mainly include failures of communication links in smart microgrids, including failures of perception layer communication equipment, edge cloud layer communication equipment, and central cloud layer communication equipment; on this basis, analyze the coordinated failures of the physical layer and data layer of smart microgrids;

[0048] Hebi City, Henan Province is located in the northern part of Henan Province, with geographical coordinates of 35°26′00″~36°02′54″N, 113°05′23″~114°45′12″E, and has a warm temperate semi-humid monsoon climate. Hebi City is famous for its abundant wind energy resources and an average annual sunshine duration of 1787.2 to 2566.7 hours, which creates unique conditions for solar and wind power generation. In addition, the city took the lead in implementing the country's first commercial microgrid project that complies with IEC technical specifications, making it a representative sample for studying microgrid technology. Taking an isolated wind-solar microgrid system in Hebi City, Henan Province as an example, an IoT-enabled smart microgrid system architecture is established. The smart microgrid system architecture consists of two core parts: the physical power network layer and the data communication layer. The physical power network layer contains six power nodes (such as wind turbines, photovoltaic panels, energy storage batteries, load points, etc.), which are interconnected through intelligent power conversion and distribution equipment to form a self-sufficient energy system. The data communication layer contains six data communication nodes (IoT sensors, data collectors, edge computing devices, etc.), which are responsible for collecting key parameters in the power network in real time and transmitting them to the central control unit through wireless networks or wired networks. If the communication nodes C2 and C3 of the data layer are attacked and fail, the real-time power parameter (such as current, voltage, power) data collected based on IoT technology, combined with formulas (1) and (2), the propagation process of its cascade failure is obtained as follows: Figure 2 shown.

[0049] S200: Use the VGG16-SVM model to fuse multi-source heterogeneous data; considering the nonlinear separability and multimodal characteristics of multi-source heterogeneous data, in view of the excellent ability of VGG16 in capturing the deep characteristics of the operating status of power equipment, combined with the advantage of SVM in efficient classification in the feature space, the VGG16-SVM model is used to process multi-source heterogeneous data, that is, convert the multi-source heterogeneous data into a distributed data format and upload it to the cloud for processing, so as to improve the data processing efficiency;

[0050] In order to improve the data processing efficiency, the microgrid image data needs to be converted into a distributed data format and uploaded to the cloud for processing. After obtaining the operation images of different types of smart microgrid devices, the data image size is adjusted to the same as the network size. At the same time, the SVM classifier is fitted with the features of the extracted convolutional neural network (pool5) layer. The confusion matrix of the VGG16-SVM and SVM models on the test set is shown in Table 1.

[0051] Table 1 Confusion matrix of VGG16-SVM and SVM models

[0052]

[0053] S300: Use the fault diagnosis model that considers the entire life cycle to perform short-term, medium-term and long-term fault diagnosis on the smart microgrid. Among them, a short-term fault diagnosis model based on transfer learning and TL-SVM, a medium-term fault diagnosis model based on multi-dimensional ensemble learning, and a long-term fault diagnosis model based on CNN-BiGRU-Bi-Attention are constructed to accurately capture the dynamic changes of risks.

[0054] In short-term fault diagnosis, considering the challenges of small samples and class imbalance faced by microgrid equipment in the early stage, the transfer learning method is combined with SVM to improve the accuracy of fault diagnosis. Taking the 96 data obtained by a wind turbine on January 1, 2022 as an example, the wind direction, temperature, humidity, air pressure, wind speed, power, etc. are used as input to obtain the confusion matrix of the proposed short-term fault diagnosis model and SVM model on the validation set as shown below: Figure 3 shown.

[0055] In mid-term fault diagnosis, in order to ensure the generalization ability of the fault diagnosis model, the faults are diagnosed from two dimensions: the accuracy of fault diagnosis and the differences between base learners. Taking 2978 data obtained from a wind turbine and a photovoltaic generator in January 2022 as an example, four methods including SVM, multi-layer perceptron (MLP), KNN, and random forest (RF) are selected as the base learners of ensemble learning. Based on formula (4), the confusion matrix of wind turbines and photovoltaic generators on the validation set is obtained as follows: Figure 4 The comparison results of the proposed method with the four base learner methods are shown in Figure 5 shown.

[0056] In long-term fault diagnosis, the main consideration is the long-term patterns and trends in the data, so as to make accurate fault diagnosis. Taking 34,397 data of wind turbines and photovoltaic generators from January to September 2021 as an example, the confusion matrix of wind turbines and photovoltaic generators on the validation set is obtained as follows: Figure 6 shown.

[0057] S400: A two-stage interval prediction method combining convolutional neural bidirectional long short-term memory network and Bootstrap method is used to predict the risk interval;

[0058] Taking the wind speed and power of wind turbines as an example, the operating data from January to August 2022 is regarded as the training set, and the prediction interval of the risk probability of some data (58 in total) on September 1, 2022 is obtained as follows: Figure 7 shown.

[0059] S500: Based on the perspective of the power supply chain, a multi-dimensional resilience assessment model for smart microgrids is constructed. The resilience of smart microgrids is comprehensively assessed from three dimensions: power supply chain, time, and power supply capacity, facilitating rapid fault location and rapid execution of maintenance decisions.

[0060] Based on the results of risk assessment and prediction, in order to quickly locate the faulty equipment in the microgrid, six typical power supply chains are selected for analysis. Their topological structures are as follows: Figure 8 As shown, the multi-dimensional resilience assessment image on September 1, 2022 is obtained as follows Fig. 9 shown.

[0061] S600: Use risk-informed importance to determine maintenance priorities for smart microgrid equipment, making maintenance activities more targeted and efficient by comprehensively considering factors such as reliability, risk, and cost.

[0062] Based on status assessment, combined Fig. 9, it can be found that the substation node 19 located on the 6th power supply chain fails, resulting in the interruption of power transmission. Similarly, it can be found that the distribution node 21 on the 2nd power supply chain fails. Set the tolerance parameter of the electrical intermediate number of the physical layer node to 1.2, the tolerance parameter of the communication intermediate number of the data layer to 1.2, and the adjustable parameter of the load distribution to 1. Based on the failure mechanism analysis, it can be seen that the transmission node 19 of the 6th power supply chain will have a cascading failure, and the adjacent communication nodes will fail due to lack of power supply. At this time, the transmission node 15 fails. Similarly, the communication node and substation node 17 connected to the distribution node 21 all fail. The corresponding risk probabilities and maintenance cost values ​​of the four nodes at this time are shown in Table 2.

[0063] Table 2 Risk probability values ​​and maintenance cost values ​​of four nodes

[0064]

[0065]

[0066] Combined with formula (6), based on the change of cost increment, a comprehensive comparison of the maintenance order of nodes is carried out. The relevant results are as follows: Fig.10 shown. Fig.10 The comparison results of the total expected cost between risk-informed importance and random recovery strategies are given, indicating that the proposed maintenance sequence based on risk-informed importance has a smaller cost increment.

[0067] Compared with the existing modeling methods, the beneficial effects brought by the present invention are: with the help of the full life cycle risk assessment and multi-dimensional resilience optimization method of the smart microgrid under the uncertain environment developed by the present invention, by integrating advanced technologies such as transfer learning, integrated learning, and deep learning, the risks of the smart microgrid at different life cycle stages are accurately captured and dynamically assessed. At the same time, the concept of risk-informed importance is introduced, the equipment maintenance strategy is optimized, and the multi-dimensional resilience assessment method is combined to significantly improve the risk resistance and resilience level of the smart microgrid in the face of uncertain environments, providing solid technical support for the safe, stable operation and sustainable development of the smart microgrid. The present invention has good promotion and application value, and provides new methods and technical support for the operation and maintenance management of smart microgrids.

[0068] It should be pointed out that the above is only a preferred embodiment of the present invention and does not limit the present invention in any form. Any technician familiar with the profession can make changes or modify the technical contents disclosed above into equivalent embodiments with equivalent changes without departing from the scope of the technical solution of the present invention, which all fall within the protection scope of the present invention.

Claims

1. A full life cycle risk assessment and multi-dimensional resilience optimization method for smart microgrids under uncertain environments. It is characterized in that The following steps are involved: S100: Use electrical betweenness and communication betweenness to analyze the failure mechanisms of power equipment and communication equipment in smart microgrids respectively; among them, physical layer failures mainly involve failures of power equipment in the power supply chain, including power generation equipment failures, transmission line failures, distribution equipment failures, and energy storage system failures, etc. Data layer failures mainly include failures of communication links in smart microgrids, including failures of perception layer communication equipment, edge cloud layer communication equipment, and central cloud layer communication equipment; on this basis, analyze the coordinated failures of the physical layer and data layer of smart microgrids; S200: Using VGG16 and support vector machine (SVM) combined model to fuse multi-source heterogeneous data; considering the nonlinear separability and multimodal characteristics of multi-source heterogeneous data, in view of VGG16's excellent ability in capturing the deep characteristics of the operating status of power equipment, combined with the advantage of SVM's efficient classification in feature space, the VGG16-SVM model is used to process multi-source heterogeneous data, that is, converting multi-source heterogeneous data into distributed data format and uploading it to the cloud for processing, so as to improve data processing efficiency; S300: Use the fault diagnosis model considering the whole life cycle to perform short-term, medium-term and long-term fault diagnosis on the smart microgrid respectively; among them, a short-term risk assessment model based on transfer learning and SVM, a medium-term risk assessment model based on multi-dimensional ensemble learning, and a long-term risk assessment model based on convolutional neural network and bidirectional gated recurrent unit (CNN-BiGRU-Bi-Attention) with dual-head attention mechanism are constructed to accurately capture the dynamic changes of risks; S400: A two-stage interval prediction method combining convolutional neural bidirectional long short-term memory network and Bootstrap method is used to predict the risk interval; S500: Based on the perspective of the power supply chain, a multi-dimensional resilience assessment model for smart microgrids is constructed. The resilience of smart microgrids is comprehensively assessed from three dimensions: power supply chain, time, and power supply capacity, facilitating rapid fault location and rapid execution of maintenance decisions. S600: Use risk-informed importance to determine maintenance priorities for smart microgrid equipment, making maintenance activities more targeted and efficient by comprehensively considering factors such as reliability, risk, and cost.

2. According to the full life cycle risk assessment and multi-dimensional resilience optimization method of a smart microgrid under an uncertain environment as described in claim 1, It is characterized in that In step S100, considering that the electrical betweenness can reflect the contribution of nodes or edges in the smart microgrid to the propagation of the whole network power flow, the electrical betweenness e(a, b) between branches (a, b) is introduced to describe the fault propagation process of the power equipment, that is, Among them, V g Represents the set of power generation nodes, V L represents the set of load nodes, W i is the actual power generation of the power generation node, W j is the actual load of the load node, I ij (a,b) represents the current on branch (a,b) when (i,j) injects a unit current element; for communication equipment, considering that the capabilities of data packet generation and forwarding are relatively close, the communication betweenness is introduced to describe the fault propagation process of power equipment, that is, Where l(v) represents the data layer node v(v∈V D ) load capacity, δ ij represents the number of shortest paths from node i to node j in the data layer, δ ij (v) represents the number of shortest paths from node i to node j passing through node v in the data layer.

3. According to the full life cycle risk assessment and multi-dimensional resilience optimization method of a smart microgrid under an uncertain environment as described in claim 1, It is characterized in that In step S200, the VGG16-SVM model is used to fuse multi-source heterogeneous data. A key step is to remove the fully connected layer of the VGG16 model to avoid redundancy or overfitting that may be introduced by the fully connected layer to a specific classification task. At the same time, combined with the advantage of SVM in efficient classification in feature space, the feature vector output by VGG16 is used as the input of SVM, and the multi-source heterogeneous data is converted into a distributed data format and uploaded to the cloud for processing, so as to improve the data processing efficiency.

4. According to the full life cycle risk assessment and multi-dimensional resilience optimization method of a smart microgrid under an uncertain environment as described in claim 1, It is characterized in that In the step S300, a data-driven fault diagnosis model considering the entire life cycle is proposed. The model specifically includes short-term (1 day to 1 week), medium-term (1 week to 1 month), and long-term (more than 1 month) fault diagnosis models for smart microgrids. Among them, a short-term risk assessment model based on transfer learning and SVM, a medium-term risk assessment model based on multidimensional ensemble learning, and a long-term risk assessment model based on CNN-BiGRU-Bi-Attention are constructed to accurately capture the dynamic changes of risks.

5. According to the full life cycle risk assessment and multi-dimensional resilience optimization method of a smart microgrid under an uncertain environment as described in claim 1, It is characterized in that In step S400, a two-stage interval prediction method combining a convolutional neural bidirectional long short-term memory network and a Bootstrap method is used to perform interval prediction on the risk. First, the advantage of CNN in capturing local features and the advantage of BiLSTM in capturing long-term dependencies in the sequence are used to perform point prediction on the covariate; secondly, the Bootstrap method is used to analyze the prediction error of the covariate data and calculate the distribution boundary value of the prediction error under a given confidence interval.

6. According to the full life cycle risk assessment and multi-dimensional resilience optimization method of a smart microgrid under an uncertain environment as described in claim 1, It is characterized in that In step S500, based on the perspective of the power supply chain, a multi-dimensional resilience assessment model of the smart microgrid is constructed, and the load supply rate is used as a dynamic indicator to reflect the power supply capacity of the smart microgrid. The resilience index R is defined as S The specific formula is: Among them, T is the duration of the study, N t is the number of power consumption nodes at time t, Whether the effective load value of a certain power consumption node meets the specified load value at a specific research time, if it does, Otherwise 0.

7. According to the full life cycle risk assessment and multi-dimensional resilience optimization method of a smart microgrid under an uncertain environment as described in claim 1, It is characterized in that In step S600, the maintenance priority of the smart microgrid equipment is determined by using the risk awareness importance, that is, in, is the risk-informed importance of the kth device at time t, C(t) is the expected cost of maintaining the smart microgrid system, F(t) is the risk function of the smart microgrid system, and F k (t) is the risk function of the kth device at time t.

Images