New energy station operation and maintenance decision method, device, equipment and medium

Abstract

The present application relates to the field of new energy power generation technology, and discloses a new energy station operation and maintenance decision method; the present application obtains meteorological data of a target new energy station, equipment operation data of each target device and position relationship data; based on the spatio-temporal correlation characteristics of each target device, the meteorological data, equipment operation data and position relationship data are fused for spatio-temporal fusion, and the real-time state of each target device is determined based on the spatio-temporal fusion characteristics obtained by fusion; based on the real-time state and the preset operation and maintenance decision knowledge graph, a first operation and maintenance decision result is determined; the first operation and maintenance decision result is physically verified to obtain a physical verification result; the real-time state, the first operation and maintenance decision result and the physical verification result are safety verified to obtain a safety verification result; based on the physical verification result and the safety verification result, the first operation and maintenance decision result is verified to obtain a second operation and maintenance decision result as the operation and maintenance decision result of the new energy station, which has the advantage of high fault recognition accuracy.

Description

Technical Field

[0001] This invention relates to the field of new energy power generation technology, and specifically to a method for operation and maintenance decision-making for new energy power plants. Background Technology

[0002] Traditional new energy operation and maintenance methods mostly rely on manual inspections; manual inspections have limited coverage, slow response to station faults, and lack the ability to predict equipment health.

[0003] In addition, related technologies also rely on traditional data analysis based on production data such as power generation and electricity output. Production data can only provide a single-scale reference for operation and maintenance decisions. When making decisions, factors of other scales, such as production data and meteorological data, are not integrated and considered, resulting in a lack of accuracy in operation and maintenance decisions. Summary of the Invention

[0004] In view of this, the present invention provides a method, apparatus, equipment and medium for operation and maintenance decision-making of new energy power stations, so as to solve the problems existing in the operation and maintenance methods of new energy in related technologies.

[0005] In a first aspect, the present invention provides an operation and maintenance decision-making method for a new energy power station. The method includes: acquiring meteorological data of the target new energy power station, equipment operation data of each target device, and location relationship data; performing spatiotemporal fusion based on the spatiotemporal correlation characteristics of each target device, fusing the meteorological data, equipment operation data, and location relationship data, and determining the real-time status of each target device based on the fused spatiotemporal fusion characteristics; determining a first operation and maintenance decision result based on the real-time status and a preset operation and maintenance decision knowledge graph; performing physical verification on the first operation and maintenance decision result to obtain a physical verification result; obtaining the physical verification result based on a first verification relationship between the physical simulation result of the first operation and maintenance decision result and physical constraints during the physical verification process; performing security verification on the real-time status, the first operation and maintenance decision result, and the physical verification result to obtain a security verification result; and verifying the first operation and maintenance decision result based on the physical verification result and the security verification result to obtain a second operation and maintenance decision result as the operation and maintenance decision result of the new energy power station.

[0006] As an exemplary embodiment, spatiotemporal fusion is performed by fusing meteorological data, equipment operation data, and location relationship data based on the spatiotemporal correlation features of each target device, including: constructing an equipment health feature vector based on the equipment operation data; constructing an environmental feature vector based on the meteorological data; constructing spatial topological features based on the location features and correlation features of the target devices; dividing the equipment health feature vector and environmental feature vector according to a preset time window length to obtain multiple time series; and concatenating the time series to obtain multiple spatiotemporal fusion features.

[0007] As an exemplary embodiment, the step of splicing the time series sequences to obtain multiple spatiotemporal fusion features includes: acquiring the timeliness and accuracy features of each time series sequence; determining the feature fusion weight of each time series sequence based on the timeliness and / or accuracy features; and fusing the time series sequences based on the feature fusion weight to obtain the spatiotemporal fusion features.

[0008] As an exemplary embodiment, the step of physically verifying the first operation and maintenance decision result to obtain the verification result includes: performing digital twin simulation based on the first operation and maintenance decision result to obtain digital twin simulation result; extracting the simulation equipment operating parameters of each target device from the digital twin simulation result; obtaining a preset security boundary library for security judgment of the simulation equipment operating parameters; and performing physical constraint verification on the simulation equipment operating parameters based on the preset security boundary library to obtain the physical verification result.

[0009] As an exemplary embodiment, determining the first operation and maintenance decision result based on the real-time state and the preset operation and maintenance decision knowledge graph includes: matching the real-time state with preset device nodes, preset fault nodes, and preset feature nodes contained in the preset operation and maintenance decision knowledge graph to obtain a matching result; if the real-time state matches the preset device nodes, preset fault nodes, and preset feature nodes contained in the preset operation and maintenance decision knowledge graph, determining a decision node in the preset operation and maintenance decision knowledge graph based on the matched device node, matched fault node, and matched feature node; and determining the first operation and maintenance decision result based on the decision node.

[0010] As an exemplary embodiment, the operation and maintenance decision-making method for the new energy power station further includes: if the real-time state does not match the preset equipment nodes, preset fault nodes, and preset feature nodes contained in the preset operation and maintenance decision knowledge graph, calculating the actual feature similarity between the features in the real-time state and the preset feature nodes; if the actual feature similarity is less than the preset feature similarity, creating new fault nodes, new equipment nodes, and new feature nodes and their corresponding edge features in the preset operation and maintenance decision knowledge graph based on the real-time state; generating initial decision rules as new decision node features based on the new fault nodes, new equipment nodes, and new feature nodes; and updating the preset operation and maintenance decision knowledge graph based on the initial decision rules generated as new decision node features and the edge features.

[0011] As an exemplary embodiment, the operation and maintenance decision-making method for the new energy power station further includes: obtaining the actual fault type, the actual fault confidence level, and the actual fault data from the second operation and maintenance decision result; and updating the preset operation and maintenance decision knowledge graph based on the actual fault type, the actual fault confidence level, and the actual fault data.

[0012] Secondly, the present invention provides an operation and maintenance decision-making device for a new energy power station. The device includes: an acquisition module for acquiring meteorological data of the target new energy power station, equipment operation data of each target device, and location relationship data; a real-time status determination module for performing spatiotemporal fusion of the meteorological data, equipment operation data, and location relationship data based on the spatiotemporal correlation characteristics of each target device, and determining the real-time status of each target device based on the fused spatiotemporal fusion characteristics; a first operation and maintenance decision module for determining a first operation and maintenance decision result based on the real-time status and a preset operation and maintenance decision knowledge graph; a physical verification module for performing physical verification on the first operation and maintenance decision result to obtain a physical verification result; during the physical verification process, the physical verification result is obtained based on a first verification relationship between the physical simulation result of the first operation and maintenance decision result and the physical constraint conditions; a security verification module for performing security verification on the real-time status, the first operation and maintenance decision result, and the physical verification result to obtain a security verification result; and an operation and maintenance decision result determination module for verifying the first operation and maintenance decision result based on the physical verification result and the security verification result to obtain a second operation and maintenance decision result as the operation and maintenance decision result of the new energy power station.

[0013] Thirdly, the present invention provides a computer device, including: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to perform the operation and maintenance decision-making method for new energy power stations as described in the first aspect or any corresponding embodiment.

[0014] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions, which are used to cause a computer to execute the operation and maintenance decision-making method for new energy power stations according to the first aspect or any corresponding embodiment described above.

[0015] This invention provides an operation and maintenance decision-making method for a new energy power station. The method includes: acquiring meteorological data of the target new energy power station, equipment operation data of each target device, and location relationship data; fusing the meteorological data, equipment operation data, and location relationship data based on the spatiotemporal correlation characteristics of each target device to perform spatiotemporal fusion; determining the real-time status of each target device based on the fused spatiotemporal fusion characteristics; determining a first operation and maintenance decision result based on the real-time status and a preset operation and maintenance decision knowledge graph; performing physical verification on the first operation and maintenance decision result to obtain a physical verification result; and during the physical verification process, comparing the physical simulation result of the first operation and maintenance decision result with the first physical constraint conditions... The physical verification result is obtained by verifying the relationship; a security verification is performed on the real-time status, the first operation and maintenance decision result, and the physical verification result to obtain a security verification result; the first operation and maintenance decision result is verified based on the physical verification result and the security verification result to obtain a second operation and maintenance decision result as the operation and maintenance decision result of the new energy power station; after obtaining the first operation and maintenance decision result, the present invention further verifies the first operation and maintenance decision result, and finally obtains the second operation and maintenance decision result based on the verification result; compared with the new energy operation and maintenance methods in related technologies, the second operation and maintenance decision result obtained by considering multi-dimensional data fusion and verifying the physical model and rationality has the advantage of high fault identification accuracy. Attached Figure Description

[0016] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating the operation and maintenance decision-making method for new energy power stations according to an embodiment of the present invention;

[0018] Figure 2 This is a structural block diagram of the operation and maintenance decision-making device for a new energy power station according to an embodiment of the present invention;

[0019] Figure 3 This is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Most new energy operation and maintenance methods in related technologies rely on manual inspections. Manual inspections have limited coverage, slow response to station faults, and lack equipment health prediction capabilities. Furthermore, these technologies rely on traditional data analysis based on production data such as power generation and output. When conducting data analysis, operation and maintenance decisions are typically made based on the relationship between production data and preset thresholds, resulting in high false alarm rates and an inability to identify hidden faults. On the other hand, when comparing production data with preset thresholds, the equipment operating status is often determined based on the relationship between production data of a single type and the corresponding preset threshold (e.g., relying solely on SCADA data while ignoring meteorological influences). This data siloing leads to biased decision-making. Simultaneously, the resulting operation and maintenance decisions lack safety verification, easily triggering secondary faults.

[0022] In view of this, according to an embodiment of the present invention, an embodiment of an operation and maintenance decision-making method for new energy power stations is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0023] For example, new energy power stations may include photovoltaic power stations, wind power stations, etc.

[0024] This embodiment provides a method for operation and maintenance decision-making for new energy power stations. Figure 1 This is a flowchart of the operation and maintenance decision-making method for new energy power stations according to an embodiment of the present invention, such as... Figure 1 As shown, the process includes the following steps:

[0025] Step S101: Obtain meteorological data, equipment operation data and location relationship data of the target new energy power station, and other relevant data.

[0026] In this embodiment, the equipment operation data can be obtained by collecting SCADA data and sensor data from the target new energy power station; wherein, SCADA data may include current, voltage, operating temperature, real-time power, power curve, etc.; sensor data may include vibration noise, etc.

[0027] In this embodiment, the target device may include a wind turbine, a photovoltaic string, etc.

[0028] In this embodiment, meteorological data of the target new energy power station can be obtained by integrating meteorological APIs, and further obtained by spatiotemporal matching of meteorological data; wherein, meteorological data may include irradiance, wind speed, temperature and humidity, precipitation, etc.

[0029] In this embodiment, the location relationship data includes geographical location information between individual target devices and physical connection relationships between multiple target devices; wherein, physical connection relationships may include the series-parallel connection relationship of photovoltaic modules and / or the layout of wind turbine clusters, etc.

[0030] In one embodiment, the acquired meteorological data and equipment operation data can be data within a historical preset duration; wherein, the preset duration is greater than 15 minutes.

[0031] Step S102: Based on the spatiotemporal correlation characteristics of each target device, the meteorological data, the device operation data, and the location relationship data are fused to perform spatiotemporal fusion, and the real-time status of each target device is determined based on the spatiotemporal fusion characteristics obtained by fusion.

[0032] After obtaining meteorological data, equipment operation data, and location relationship data, the meteorological data, equipment operation data, and location relationship data are fused based on the spatiotemporal correlation characteristics of each target device to obtain spatiotemporal fusion features; further, the real-time status of each target device is determined based on the fused spatiotemporal fusion features.

[0033] For example, before fusing the data, feature vectors corresponding to each data point can be constructed in advance.

[0034] In one embodiment, after obtaining device operating data, a device health feature vector X is constructed based on the numerical features of the device operating data. s =[x s1 s x2 , ..., x sn ].

[0035] In one embodiment, after obtaining meteorological data, an environmental feature vector X is constructed based on the numerical characteristics of the meteorological data. w =[w1,w2,...,w m ].

[0036] For example, after obtaining the environmental feature vector, the geographical location of the site is matched using a spatiotemporal interpolation algorithm.

[0037] In one embodiment, after obtaining the location relationship data, a spatial network of station equipment is constructed based on the location data and GIS map in the location relationship data; and an adjacency matrix A is obtained based on the physical connection relationship between the target equipment.

[0038] Furthermore, after obtaining the equipment health feature vector and environmental feature vector, these vectors are unified to the UTC time coordinate system and the site geographic coordinate system. The data for each dimension is then sliced ​​according to a preset time window to obtain a time series. Where T represents the number of time steps.

[0039] Furthermore, the device health feature vector and the environmental feature vector are concatenated to obtain spatiotemporal fusion features.

[0040] Furthermore, when determining the real-time state of each target device based on the spatiotemporal fusion features obtained by fusion, the real-time state of each target device can be determined based on the numerical features and / or spatiotemporal change features of the spatiotemporal fusion features.

[0041] In this embodiment, after obtaining the spatiotemporal fusion features, the real-time status of each target device is determined based on the numerical features and / or spatiotemporal change features of the spatiotemporal fusion features.

[0042] Since the spatiotemporal fusion feature is obtained by fusing the device health feature vector and the environment feature vector, in one embodiment, the spatiotemporal fusion feature is based on the fused value and the fusion direction that characterizes the fusion of the device health feature vector and the environment feature vector.

[0043] In this embodiment, the real-time state of each target device can be determined solely based on the numerical characteristics of the spatiotemporal fusion features; in another embodiment, the real-time state of the target device can be obtained by inputting the spatiotemporal fusion features into a pre-trained real-time state output model.

[0044] In one embodiment, the real-time status output model can be obtained by training a training set consisting of historical meteorological data, historical equipment operation data, location relationship data, and their corresponding historical states. During the model training process, the historical meteorological data, historical equipment operation data, and location relationship data are used as input features, and the historical states corresponding to their time series are used as labels. The preset model parameters are continuously adjusted so that the model can learn to fuse the meteorological data, the equipment operation data, and the location relationship data based on the spatiotemporal correlation features of each target device to obtain the spatiotemporal fusion feature fusion relationship, as well as the correspondence between the spatiotemporal fusion feature and the historical state.

[0045] For example, the real-time status may include a first status characterizing normal operation of the device and a second status characterizing a problem with the device.

[0046] For example, when the real-time state is the second state, other data of the device in the second state can be collected to verify the second state.

[0047] In one embodiment, infrared thermal images of the target device are additionally acquired from the UAV, and image processing algorithms are used to extract the feature vector of the hot spot region.

[0048] Furthermore, the feature vectors of the hot spot region are unified to the UTC time coordinate system and the site geographic coordinate system through the spatiotemporal alignment module.

[0049] Finally, the hot spot region feature vector, equipment health feature vector, and environmental feature vector are concatenated to obtain the spatiotemporal fusion feature.

[0050] Step S103: Determine the first operation and maintenance decision result based on the real-time status and the preset operation and maintenance decision knowledge graph.

[0051] In this embodiment, the first operation and maintenance decision result is first determined based on a preset operation and maintenance decision knowledge graph.

[0052] In one embodiment, the preset operation and maintenance decision knowledge graph is constructed based on nodes and edges.

[0053] For example, for nodes in a pre-defined operation and maintenance decision knowledge graph, the node types may include device nodes, fault nodes, decision nodes, and feature nodes; device nodes are used to refer to each target device, fault nodes are used to refer to fault types or specific faults, decision nodes are used to refer to specific decisions corresponding to fault types or specific faults, and feature nodes are used to refer to specific features corresponding to fault types or specific faults.

[0054] For example, the device node includes at least the device node corresponding to the photovoltaic device or the device node corresponding to the wind turbine device; for example, the device node may include "photovoltaic string-N" or "wind turbine blade-M" (M and N are positive integers).

[0055] For example, fault nodes may include "abnormal temperature", "hot spot effect", "abnormal rotation speed", "bearing wear", etc.

[0056] For example, decision nodes may include "disconnecting photovoltaic strings" or "shutting down wind turbines for maintenance".

[0057] For example, the feature nodes may include "sudden temperature change", "abnormal infrared imaging", "speed decrease", and "sudden speed drop".

[0058] For example, the edges of the preset operation and maintenance decision knowledge graph are used to represent the association between nodes; for example, the edges connecting the device node and the fault node can be used to characterize the probability of the fault occurring for the device; the edges connecting the feature node and the fault node can be used to characterize the confidence level of the fault corresponding to the feature of the device; and the edges connecting the fault node and the decision node can be used to characterize the execution effect of the execution strategy corresponding to the fault.

[0059] For example, after obtaining the real-time status, the target device corresponding to the real-time status and the features contained in the real-time status are matched with the nodes of the preset operation and maintenance decision knowledge graph. Furthermore, based on the matching results, the first operation and maintenance decision result is determined according to the decision node corresponding to the node of the preset operation and maintenance decision knowledge graph.

[0060] The first operation and maintenance decision typically includes operations such as disconnecting wind turbines or photovoltaic strings. After disconnecting wind turbines or photovoltaic strings, the target equipment of the new energy power station may be affected by the disconnected wind turbines or photovoltaic strings, and the equipment status may fluctuate, which may cause one or more pieces of equipment to change from a normal state to an abnormal state. To solve this problem, in this invention, after obtaining the first operation and maintenance decision result, the first operation and maintenance decision result is further verified.

[0061] Step S104: Perform physical verification on the first operation and maintenance decision result to obtain the physical verification result; during the physical verification process, obtain the physical verification result based on the first verification relationship between the physical simulation result and the physical constraint conditions of the first operation and maintenance decision result.

[0062] In this embodiment, the first operation and maintenance decision result is physically verified to confirm its rationality.

[0063] For example, for the first operation and maintenance decision result, a physical simulation model of the target device can be pre-built, and further constraints can be imposed on each physical simulation model based on the first operation and maintenance decision result, thereby driving the physical simulation model to perform simulation based on the constraints, and finally obtaining the actual performance of each target device based on physical constraints under the first operation and maintenance decision result.

[0064] Furthermore, after obtaining the actual performance of each target device based on physical constraints under the first operation and maintenance decision result, the physical verification result is obtained according to the first verification relationship of the physical constraint conditions corresponding to the first operation and maintenance decision result.

[0065] For example, the first verification relationship can be predetermined based on the target device; specifically, for the first operation and maintenance decision result of "cutting off the photovoltaic string", the first verification relationship can be to verify the electrical safety requirements or thermal safety requirements of the photovoltaic string; wherein, electrical safety may include the voltage fluctuation range and / or current fluctuation range, etc., and thermal safety may include the temperature of the photovoltaic module, etc.; for the first operation and maintenance decision result of "shutting down the wind turbine for maintenance", the first verification relationship can be the mechanical safety requirements of the wind turbine; wherein, mechanical safety may include the blade stress threshold, the upper limit of bearing temperature, etc.

[0066] Step S105: Perform security verification on the real-time status, the first operation and maintenance decision result, and the physical verification result to obtain the security verification result.

[0067] In the above embodiments, the real-time state determined based on spatiotemporal fusion features is obtained through inference or prediction. Therefore, there may be judgment errors in the determination of the real-time state. On this basis, the first operation and maintenance decision result determined based on the real-time state and the subsequent physical verification result may also have deviations. To solve this problem, in this invention, the real-time state, the first operation and maintenance decision result and the physical verification result are further subjected to security verification to obtain a security verification result.

[0068] In this invention, reinforcement learning is used to achieve security verification of real-time status, first operation and maintenance decision results, and physical verification results.

[0069] Specifically, a state space and an action space are constructed based on reinforcement learning theory. For example, for real-time state security verification, the state space of the corresponding verification model may include spatiotemporal fusion features and the real-time state features, preceding action features, and preceding reward features of the target device; the action space may include real-time state actions. During training, the network is updated and the objective function is optimized using reinforcement learning algorithms based on historical spatiotemporal fusion features and historical real-time state features. During the update and optimization process, the model is guided by the reward function to maximize the reliability of the real-time state (updating the model network and optimizing the objective parameters, so that the model assigns relatively accurate spatiotemporal action features and real-time state features). Positive rewards are given, and the reward value is determined based on the accuracy of spatiotemporal action features and real-time state features; where the reward value is positively correlated with the accuracy; negative rewards are given for relatively inaccurate spatiotemporal action features and real-time state features. At the same time, the action cost is minimized, so that the accuracy of the model's state reliability judgment on the validation set meets the preset accuracy, and the target security verification model is obtained. In the application process, the real-time state and spatiotemporal fusion features are input into the first target security verification model, and the reward for the real-time state is obtained based on the first target security verification model. The correspondence between the real-time state and the spatiotemporal fusion features is determined based on the reward, and the real-time state is verified based on the first target security verification model.

[0070] For example, for the security verification of the first operation and maintenance decision result, when constructing the model, the state space of the corresponding verification model may include the real-time state characteristics of the target device, the characteristics of the first operation and maintenance decision result, the characteristics of the preceding actions, and the characteristics of the preceding rewards. The action space may include the actions of the first operation and maintenance decision result. The training of this model can refer to the content disclosed in the above embodiments. During the update and optimization process, the model is guided by a reward function to maximize the reliability of the first operation and maintenance decision result (update the model network and optimize the target parameters so that the model assigns positive rewards to relatively matching real-time state characteristics and the first operation and maintenance decision result, and determines the reward value according to the accuracy of the real-time state characteristics and the first operation and maintenance decision result; wherein, the reward value is positively correlated with the accuracy; and assigns negative rewards to relatively inaccurate spatiotemporal action characteristics and the first operation and maintenance decision result) while minimizing the action cost, thus obtaining the second target security verification model. In the application process, the real-time state and the first operation and maintenance decision result are input into the second target security verification model, and the reward of the real-time state is obtained based on the second target security verification model, thereby determining the correspondence between the real-time state and the first operation and maintenance decision result according to the reward.

[0071] For example, for the security verification of physical verification results, when constructing the model, the state space may include the first operation and maintenance decision result features of the target device, the physical verification result features, and the preceding action features and preceding reward features; the action space may include the physical verification result actions; the training of this type of model can refer to the content disclosed in the above embodiments. During the update and optimization process, the model is guided by a reward function to maximize the reliability of the physical verification results (update the model network and optimize the target parameters so that the model assigns higher positive rewards to the corresponding first operation and maintenance decision results and physical verification results, and assigns lower positive rewards to the relatively uncorresponding first operation and maintenance decision results and physical verification result actions) while minimizing the action cost, thus obtaining the target security verification model; during the application process, the first operation and maintenance decision results and physical verification results are input into the first target security verification model, and the reward of the physical verification result is obtained based on the first target security verification model, thereby determining the correspondence between the first operation and maintenance decision results and the physical verification results according to the reward.

[0072] In one embodiment, physical verification and security verification can be performed simultaneously to achieve dual-loop verification of the first operation and maintenance decision result. Specifically, physical verification is used as the inner loop verification, and security verification is used as the outer loop verification to verify the first operation and maintenance decision result. During the inner loop verification process, the physical verification result is obtained based on the first verification relationship between the physical simulation result and the physical constraints of the first operation and maintenance decision result. During the outer loop verification process, the security verification result is obtained based on the second verification relationship between the actual reward corresponding to the first operation and maintenance decision result and the corresponding safe feasible domain.

[0073] Step S106: Verify the first operation and maintenance decision result based on the physical verification result and the security verification result to obtain the second operation and maintenance decision result as the operation and maintenance decision result of the new energy power station.

[0074] After obtaining the physical verification results and the security verification results, if both the physical verification results and the security verification results indicate that the first operation and maintenance decision result can meet the verification requirements, then the first operation and maintenance decision result shall be used as the operation and maintenance decision result of the new energy power station.

[0075] If the first maintenance decision result fails to meet the verification requirements, the first maintenance decision result will be rolled back to the previous maintenance result.

[0076] For example, regarding the physical verification results, if the actual performance of each target device based on physical constraints under the first operation and maintenance decision result meets the electrical safety requirements, thermal safety requirements, and / or mechanical safety requirements, the first operation and maintenance decision result is confirmed to be consistent with the physical verification results; otherwise, if the actual performance of each target device based on physical constraints under the first operation and maintenance decision result does not meet any one of the electrical safety requirements, thermal safety requirements, or mechanical safety requirements, the first operation and maintenance decision result is confirmed to be inconsistent with the physical verification results.

[0077] For example, regarding the security verification result, if the first operation and maintenance decision result meets the corresponding verification reward, the first operation and maintenance decision result is confirmed to meet the security verification result; otherwise, if the first operation and maintenance decision result does not meet the corresponding verification reward, the first operation and maintenance decision result is confirmed to not meet the security verification result.

[0078] This embodiment provides an operation and maintenance decision-making method for a new energy power station. The method includes: acquiring meteorological data of the target new energy power station, equipment operation data of each target device, and location relationship data; fusing the meteorological data, equipment operation data, and location relationship data based on the spatiotemporal correlation characteristics of each target device to perform spatiotemporal fusion; determining the real-time status of each target device based on the fused spatiotemporal fusion characteristics; determining a first operation and maintenance decision result based on the real-time status and a preset operation and maintenance decision knowledge graph; performing physical verification on the first operation and maintenance decision result to obtain a physical verification result; and during the physical verification process, comparing the physical simulation result of the first operation and maintenance decision result with the first physical constraint condition... The physical verification result is obtained by verifying the relationship; a security verification is performed on the real-time status, the first operation and maintenance decision result, and the physical verification result to obtain a security verification result; the first operation and maintenance decision result is verified based on the physical verification result and the security verification result to obtain a second operation and maintenance decision result as the operation and maintenance decision result of the new energy power station; after obtaining the first operation and maintenance decision result, the present invention further verifies the first operation and maintenance decision result, and finally obtains the second operation and maintenance decision result based on the verification result; compared with the new energy operation and maintenance methods in related technologies, the second operation and maintenance decision result obtained by considering multi-dimensional data fusion and verifying the physical model and rationality has the advantage of high fault identification accuracy.

[0079] As an exemplary embodiment, spatiotemporal fusion is performed by fusing meteorological data, equipment operation data, and location relationship data based on the spatiotemporal correlation features of each target device, including: constructing an equipment health feature vector based on the equipment operation data; constructing an environmental feature vector based on the meteorological data; constructing spatial topological features based on the location features and correlation features of the target devices; dividing the equipment health feature vector and environmental feature vector according to a preset time window length to obtain multiple time series; and concatenating the time series to obtain multiple spatiotemporal fusion features.

[0080] In this embodiment, the spatiotemporal fusion feature is obtained by fusing feature vectors constructed from the collected data; the method for constructing the feature vectors can be referred to the content disclosed in the above embodiments, and will not be repeated here.

[0081] After obtaining the time series, a fully connected layer can be used to splice the time series to obtain spatiotemporal fusion features.

[0082] In one embodiment, spatiotemporal fusion is performed considering the timeliness and accuracy of each data point in multi-source data. Specifically, as an exemplary embodiment, the step of splicing the time series sequences to obtain multiple spatiotemporal fusion features includes: acquiring the timeliness and accuracy features of each time series sequence; determining the feature fusion weight of each time series sequence based on the timeliness and / or accuracy features; and fusing the time series sequences based on the feature fusion weight to obtain the spatiotemporal fusion features.

[0083] For example, the timeliness characteristic can be determined based on the time difference between the data source collection time and the current time; specifically, the timeliness factor can be calculated using equation (1) as the timeliness characteristic:

[0084]

[0085] In formula (1), freshness i Δt represents the timeliness factor. i表示 The time difference T between the collection time of data source i and the current time. max To allow the maximum delay.

[0086] For example, T max Take 30 minutes.

[0087] For example, for the accuracy feature, a data source error matrix is ​​constructed based on the historical data corresponding to data source i; specifically, the accuracy feature can be calculated using equation (2):

[0088]

[0089] In equation (2), reliabilityi MAE represents the accuracy characteristics of data source i. i This represents the mean absolute error determined based on historical data, while max_MAE represents the maximum absolute error.

[0090] For example, the feature fusion weights of each time series can be determined individually based on the timeliness characteristics; specifically, the fusion weights corresponding to each time series are positively correlated with the timeliness factor, and the sum of the weights of the fusion weights corresponding to each time series is 1.

[0091] For example, the feature fusion weights of each time series can be determined individually based on the accuracy characteristics; specifically, the fusion weights corresponding to each time series are positively correlated with the accuracy factor, and the sum of the weights of the fusion weights corresponding to each time series is 1.

[0092] For example, the feature fusion weights of each time series can be jointly determined based on the timeliness and accuracy characteristics, and the time series can be further fused based on the feature fusion weights to obtain the spatiotemporal fusion features.

[0093] In one embodiment, for the fusion weight ω corresponding to data source i i， The fusion weights can be obtained using equation (3):

[0094]

[0095] Where, ω i This represents the fusion weight corresponding to data source i, where α and β are the weight determination parameters.

[0096] In one embodiment, α and β are obtained through gradient descent training; specifically, during the training process, the mean square error between the real-time state corresponding to the fused data and the actual state of the device is used as the loss function, see equation (4):

[0097]

[0098] Where ωi represents the fusion weight corresponding to data source i, Xk represents the real-time state corresponding to data source i, and Yk represents the actual state of the target device.

[0099] Furthermore, after weight allocation, feature fusion is achieved through a fully connected layer.

[0100] For example, for UAV infrared thermal images additionally acquired by the target device, image processing algorithms are used to extract feature vectors of the hot spot region.

[0101] Furthermore, the feature vectors of the hot spot region are spatiotemporally aligned to unify them to the UTC time coordinate system and the site geographic coordinate system.

[0102] Furthermore, Equation (5) can be used to achieve feature fusion of various features to obtain spatiotemporal fused features:

[0103] X fuscd =ReLU(W·[X s ;X w ;X img ]+b) (5)

[0104] Among them, X fused Represents the spatiotemporal fusion feature, ReLU represents the activation function, W represents the fusion weight, and b represents a constant. W and b can be determined through training. Xs represents the device health feature vector, Xw represents the environmental feature vector, Ximg represents the hot spot region feature vector, and ":" indicates feature concatenation.

[0105] As an exemplary embodiment, the step of physically verifying the first operation and maintenance decision result to obtain the verification result includes: performing digital twin simulation based on the first operation and maintenance decision result to obtain digital twin simulation result; extracting the simulation equipment operating parameters of each target device from the digital twin simulation result; obtaining a preset security boundary library for security judgment of the simulation equipment operating parameters; and performing physical constraint verification on the simulation equipment operating parameters based on the preset security boundary library to obtain the physical verification result.

[0106] In this embodiment, the feasibility of the first operation and maintenance decision result is verified by physical verification. For example, digital twin simulation is first performed based on the real-time status of each of the target devices to obtain the digital twin simulation result. Specifically, for each target device, a pre-built target device physical model is driven based on the real-time status of each target device to perform digital twin simulation and obtain the digital twin simulation result.

[0107] Specifically, for photovoltaic strings, the current-voltage characteristic equation is established based on the single diode model as shown in equation (6):

[0108]

[0109] Among them, I ph I0 is the photogenerated current, I0 is the reverse saturation current s The series resistor is V, where n is the diode ideality factor. T This is thermal voltage.

[0110] For wind turbine units, an aerodynamic power model is established based on blade element momentum theory as shown in equation (7):

[0111]

[0112] Where P is the fan power, ρ is the air density, R is the blade radius, v is the wind speed, and C is the air density. p λ is the power coefficient, λ is the tip speed ratio, and θ is the blade pitch angle.

[0113] Furthermore, based on the real-time state of the target equipment, the values ​​of each parameter in the current-voltage characteristic equation and / or aerodynamic power model are determined, and finally, the digital twin simulation results of the target equipment are obtained.

[0114] Simultaneously, a preset safety boundary library for making safety judgments on the operating parameters of the simulation device is obtained; for example, the preset safety boundary library can be the device's safety threshold matrix S, S = [s1, s2, ..., s2]. p It includes at least one of electrical safety conditions, mechanical safety conditions, and thermal safety conditions.

[0115] For example, electrical safety conditions may be that the simulated voltage and simulated current in the digital twin simulation results are within ±15% of the rated values.

[0116] For example, mechanical safety conditions may include: the actual blade stress in the digital twin simulation results is within a preset blade stress threshold; wherein the preset blade stress threshold can be determined by the ultimate stress obtained by stress testing the blade in a laboratory.

[0117] For example, thermal safety may include: the actual bearing temperature in the digital twin simulation results is less than a bearing temperature threshold; wherein the bearing temperature threshold can be determined by a limit temperature obtained by testing the blade in a laboratory.

[0118] In one embodiment, the bearing temperature threshold can be 90°C.

[0119] For example, thermal safety may also include: the photovoltaic panel temperature in the digital twin simulation results is less than a photovoltaic panel temperature threshold; wherein the photovoltaic panel temperature threshold can be determined by the limit temperature obtained by stress testing the blades in a laboratory.

[0120] In one embodiment, the photovoltaic panel temperature threshold can be 85°C.

[0121] Furthermore, the operating parameters of the simulation equipment are physically constrained and verified based on a preset safety boundary library to obtain the physical verification results.

[0122] As an exemplary embodiment, determining the first operation and maintenance decision result based on the real-time state and the preset operation and maintenance decision knowledge graph includes: matching the real-time state with preset device nodes, preset fault nodes, and preset feature nodes contained in the preset operation and maintenance decision knowledge graph to obtain a matching result; if the real-time state matches the preset device nodes, preset fault nodes, and preset feature nodes contained in the preset operation and maintenance decision knowledge graph, determining a decision node in the preset operation and maintenance decision knowledge graph based on the matched device node, matched fault node, and matched feature node; and determining the first operation and maintenance decision result based on the decision node.

[0123] In this embodiment, the first operation and maintenance decision result is determined based on the feature matching degree between the real-time status and the preset device nodes, preset fault nodes and preset feature nodes contained in the preset operation and maintenance decision knowledge graph.

[0124] For example, the feature matching degree can be obtained by calculating the similarity between the first feature vector contained in the real-time status and the second feature vector corresponding to the preset feature node contained in the operation and maintenance decision result.

[0125] In one embodiment, the similarity between the first feature vector and the second feature vector can be calculated based on cosine similarity using equation (8):

[0126]

[0127] In equation (8), sim is the similarity score, f is the first vector, and f old Let ||f|| be the second vector, and ||f|| represent the magnitude of the first vector. old || represents the modulus of the second vector.

[0128] Furthermore, when the similarity is greater than a preset similarity, it is confirmed that the real-time state matches the preset device nodes, preset fault nodes, and preset feature nodes contained in the preset operation and maintenance decision knowledge graph. Based on the matched device nodes, matched fault nodes, and matched feature nodes, a decision node is determined in the preset operation and maintenance decision knowledge graph. The first operation and maintenance decision result is determined based on the decision node.

[0129] As an exemplary embodiment, the operation and maintenance decision-making method for the new energy power station further includes: if the real-time state does not match the preset equipment nodes, preset fault nodes, and preset feature nodes contained in the preset operation and maintenance decision knowledge graph, calculating the actual feature similarity between the features in the real-time state and the preset feature nodes; if the actual feature similarity is less than the preset feature similarity, creating new fault nodes, new equipment nodes, and new feature nodes and their corresponding edge features in the preset operation and maintenance decision knowledge graph based on the real-time state; generating initial decision rules as new decision node features based on the new fault nodes, new equipment nodes, and new feature nodes; and updating the preset operation and maintenance decision knowledge graph based on the initial decision rules generated as new decision node features and the edge features.

[0130] In this embodiment, if the real-time status does not match the preset device nodes, preset fault nodes, and preset feature nodes contained in the preset operation and maintenance decision knowledge graph, the preset operation and maintenance decision knowledge graph is updated based on the real-time status.

[0131] For example, if the calculated feature matching degree is less than 0.95, it is considered that the real-time state does not match the preset device nodes, preset fault nodes and preset feature nodes contained in the preset operation and maintenance decision knowledge graph. At this time, the preset operation and maintenance decision knowledge graph is updated based on the real-time state.

[0132] In one embodiment, when the feature matching degree is less than 0.95 but greater than 0.9, the first feature vector contained in the real-time state is taken as a new feature node, and the preset device node, preset fault node, and preset decision node corresponding to the second feature vector in the preset operation and maintenance decision knowledge graph whose feature matching degree with the first feature vector is in the range of (0.9, 0.95) are taken as new device nodes, preset fault nodes, and preset decision nodes, and are combined with the new feature node to add as new faults to the preset operation and maintenance decision knowledge graph to update the preset operation and maintenance decision knowledge graph.

[0133] In one embodiment, when the feature matching degree is less than 0.9, a new device node, a new fault node, and a new decision node corresponding to the real-time state are generated based on the real-time state.

[0134] For example, new device nodes, new fault nodes, and new decision nodes can be generated based on real-time status through a pre-built expert system, and edge features can be determined based on the decision execution results in the expert system.

[0135] For example, the generation of device nodes, new fault nodes, and new decision nodes based on real-time status can be achieved through reinforcement learning methods as described in the above embodiments. Specifically, the real-time status is input into a pre-trained reinforcement learning model, and the third operation and maintenance decision result and / or physical verification result corresponding to the real-time status are obtained based on the reinforcement learning model. The target third operation and maintenance decision result with an actual reward value greater than the preset reward value is selected as the new decision node.

[0136] Simultaneously, the edge features are determined based on the new fault node, the new device node, and the new feature node.

[0137] In one embodiment, the determination of edge features based on new device nodes, new fault nodes, and new decision nodes can be achieved through reinforcement learning methods as described in the above embodiments. Specifically, the real-time state is input into a pre-trained reinforcement learning model, and the actual reward value of the third operation and maintenance decision result and / or physical verification result corresponding to the real-time state is obtained based on the reinforcement learning model. Furthermore, the fault confidence is determined based on the interval in which the actual reward value is located. The fault confidence is used as an edge feature.

[0138] Create new fault nodes, new device nodes, and new feature nodes and their corresponding edge features in the preset operation and maintenance decision knowledge graph; generate initial decision rules as features of new decision nodes based on the new fault nodes, new device nodes, and new feature nodes; obtain decision execution effect data based on the new fault nodes, new device nodes, new feature nodes, and new decision nodes; determine the edge features based on the decision execution effect data.

[0139] For example, the performance data can be determined by the confidence level of the third-party operation and maintenance decision results.

[0140] Finally, based on the new fault nodes, new device nodes, and new feature nodes, initial decision rules are generated as new decision node features and edge features to update the preset operation and maintenance decision knowledge graph.

[0141] In this invention, the preset operation and maintenance decision knowledge graph is also updated based on the second operation and maintenance decision result. Specifically, as an exemplary embodiment, the operation and maintenance decision method for the new energy power station further includes: obtaining the actual fault type, the actual fault confidence level, and the actual fault data from the second operation and maintenance decision result; and updating the preset operation and maintenance decision knowledge graph based on the actual fault type, the actual fault confidence level, and the actual fault data.

[0142] In this embodiment, after obtaining the second operation and maintenance decision result, the actual fault type, actual fault confidence level, and actual fault data are acquired.

[0143] Furthermore, the preset operation and maintenance decision knowledge graph is updated based on the actual fault type, the actual fault confidence level, and the similarity between the actual fault data and the preset operation and maintenance decision knowledge graph, to obtain an updated preset operation and maintenance decision knowledge graph.

[0144] In one embodiment, the actual fault type and the actual fault data can be determined based on the second operation and maintenance decision result, and the confidence level of the actual fault can be obtained by acquiring the confidence level when the real-time status of the target device is obtained based on spatiotemporal fusion features.

[0145] Furthermore, the preset operation and maintenance decision knowledge graph is updated based on the actual fault type, the actual fault confidence level, and the similarity between the actual fault data and the preset operation and maintenance decision knowledge graph, to obtain an updated preset operation and maintenance decision knowledge graph.

[0146] Specifically, when the confidence level of the actual fault is greater than the preset confidence level, a similarity match is performed on the preset operation and maintenance decision knowledge graph based on the real-time status to obtain the actual similarity.

[0147] When the actual similarity is greater than the preset similarity, the preset operation and maintenance decision knowledge graph is updated based on the actual fault type, the actual fault confidence, and the similarity between the actual fault data and the preset operation and maintenance decision knowledge graph.

[0148] This embodiment provides an operation and maintenance decision-making device for a new energy power station, such as... Figure 2 As shown, it includes:

[0149] The acquisition module 501 is used to acquire meteorological data of the target new energy power station, equipment operation data of each target device, and location relationship data.

[0150] The real-time status determination module 502 is used to perform spatiotemporal fusion of the meteorological data, the equipment operation data and the location relationship data based on the spatiotemporal correlation characteristics of each of the target devices, and to determine the real-time status of each of the target devices based on the spatiotemporal fusion characteristics obtained by fusion.

[0151] The first operation and maintenance decision module 503 is used to determine the first operation and maintenance decision result based on the real-time status and the preset operation and maintenance decision knowledge graph.

[0152] The physical verification module 504 is used to perform physical verification on the first operation and maintenance decision result to obtain a physical verification result; during the physical verification process, the physical verification result is obtained based on the first verification relationship between the physical simulation result and the physical constraint condition of the first operation and maintenance decision result.

[0153] The security verification module 505 is used to perform security verification on the real-time status, the first operation and maintenance decision result and the physical verification result to obtain the security verification result.

[0154] The operation and maintenance decision determination module 506 is used to verify the first operation and maintenance decision result based on the physical verification result and the security verification result, and obtain a second operation and maintenance decision result as the operation and maintenance decision result of the new energy power station.

[0155] It should be noted that the examples and application scenarios implemented by the above modules and corresponding steps are the same, but are not limited to the content disclosed in the above embodiments.

[0156] It should be noted that the above modules, as part of the device, can be implemented in software or hardware, with the hardware environment including the network environment.

[0157] This invention also provides a computer device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; the memory is used to store computer programs; and the processor is used to execute the methods described in any of the above embodiments by running the computer programs stored in the memory.

[0158] Figure 3 This is a structural block diagram of an optional computer device according to an embodiment of this application, such as... Figure 3 As shown, it includes a processor 10, a communication interface 20, a memory 30, and a communication bus 40. The processor 10, communication interface 20, and memory 30 communicate with each other via the communication bus 40.

[0159] Memory 30 is used to store computer programs;

[0160] When the processor 10 executes the computer program stored in the memory 30, it implements the operation and maintenance decision-making method for new energy power stations as described in any of the above embodiments.

[0161] Optionally, in this embodiment, the communication bus can be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. This communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0162] The communication interface is used for communication between the aforementioned computer equipment and other devices.

[0163] The memory may include RAM, or non-volatile memory, such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0164] The processors mentioned above can be general-purpose processors, including but not limited to: CPU (Central Processing Unit), NP (Network Processor), etc.; they can also be DSP (Digital Signal Processor), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0165] Optionally, specific examples in this embodiment can refer to the examples described in the above embodiments, and will not be repeated here.

[0166] Those skilled in the art will understand that Figure 3 The structure shown is for illustrative purposes only. The device that implements any of the methods in the above embodiments can be a terminal device, such as a smartphone (e.g., an Android phone, an iOS phone), a tablet computer, a PDA, a mobile Internet device (MID), a PAD, or other terminal devices. Figure 3 This does not limit the structure of the aforementioned electronic device. For example, the terminal device may also include components that are more... Figure 3 The more or fewer components shown (such as network interfaces, display devices, etc.), or having the same Figure 3 The different configurations shown.

[0167] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, which may include: flash drive, ROM, RAM, disk or optical disk, etc.

[0168] As an exemplary embodiment, this application also provides a computer-readable storage medium storing a computer program, wherein the computer program is configured to execute the method steps of any one of the embodiments in this application at runtime.

[0169] Optionally, in this embodiment, the storage medium described above can be used to execute program code for the method steps of the embodiments of this application.

[0170] Optionally, in this embodiment, the storage medium may be located on at least one of the network devices in the network shown in the above embodiment.

[0171] Optionally, in this embodiment, the storage medium is configured to store methods for performing the above embodiments.

[0172] Optionally, specific examples in this embodiment can refer to the examples described in the above embodiments, and will not be repeated in this embodiment.

[0173] Optionally, in this embodiment, the storage medium may include, but is not limited to, various media capable of storing program code, such as USB flash drives, ROMs, RAMs, portable hard drives, magnetic disks, or optical disks.

[0174] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0175] If the integrated units in the above embodiments are implemented as software functional units and sold or used as independent products, they can be stored in the aforementioned computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause one or more computer devices (which may be personal computers, servers, or network devices, etc.) to execute all or part of the steps of the methods in the above embodiments.

[0176] In the several embodiments provided in this application, it should be understood that the disclosed client can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or the indirect coupling or communication connection of units or modules may be electrical or other forms.

[0177] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the solution provided in this embodiment, depending on actual needs.

[0178] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0179] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0180] The above are merely preferred embodiments of this application. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for operation and maintenance decision-making of new energy power stations, characterized in that, The operation and maintenance decision-making methods for the new energy power stations include: Acquire meteorological data, equipment operation data, and location relationship data of the target new energy power station; Based on the spatiotemporal correlation characteristics of each target device, the meteorological data, the device operation data and the location relationship data are fused to perform spatiotemporal fusion, and the real-time status of each target device is determined based on the spatiotemporal fusion characteristics obtained by fusion. The first operation and maintenance decision result is determined based on the real-time status and the preset operation and maintenance decision knowledge graph. The first operation and maintenance decision result is physically verified to obtain a physical verification result. During the physical verification process, the physical verification result is obtained based on a first verification relationship between the physical simulation result and the physical constraints of the first operation and maintenance decision result. The physical verification of the first operation and maintenance decision result to obtain the verification result includes: performing digital twin simulation based on the first operation and maintenance decision result to obtain a digital twin simulation result; extracting the simulation equipment operating parameters of each target device from the digital twin simulation result; obtaining a preset security boundary library for security judgment of the simulation equipment operating parameters; and performing physical constraint verification on the simulation equipment operating parameters based on the preset security boundary library to obtain the physical verification result. Security verification results are obtained by performing security checks on the real-time state, the first operation and maintenance decision result, and the physical verification result using reinforcement learning; wherein... The security verification of the real-time state is used to determine the correspondence between the real-time state and the spatiotemporal fusion feature; The security verification of the first operation and maintenance decision result is used to determine the correspondence between the real-time status and the first operation and maintenance decision result; The security verification of the physical verification results is used to determine the correspondence between the first operation and maintenance decision result and the physical verification results; The first operation and maintenance decision result is verified based on the physical verification result and the security verification result to obtain the second operation and maintenance decision result as the operation and maintenance decision result of the new energy power station.

2. The operation and maintenance decision-making method for new energy power stations as described in claim 1, characterized in that, Spatiotemporal fusion is performed based on the spatiotemporal correlation characteristics of each target device, fusing the meteorological data, the device operation data, and the location relationship data, including: Construct a device health feature vector based on the device operation data; An environmental feature vector is constructed based on the meteorological data; Construct spatial topology features based on the location and association features of the target device; The device health feature vector and the environment feature vector are divided according to a preset time window length to obtain multiple time series sequences; The time series sequences are spliced ​​together to obtain multiple spatiotemporal fusion features.

3. The operation and maintenance decision-making method for new energy power stations as described in claim 2, characterized in that, The concatenation of the various time-series sequences yields multiple spatiotemporal fusion features, including: Obtain the timeliness and accuracy characteristics of each of the time series sequences; The feature fusion weights of each time series are determined based on the timeliness and / or accuracy characteristics. The spatiotemporal fusion feature is obtained by fusing the time series based on the feature fusion weights.

4. The operation and maintenance decision-making method for new energy power stations as described in claim 1, characterized in that, The determination of the first operation and maintenance decision result based on the real-time status and the preset operation and maintenance decision knowledge graph includes: The matching results are obtained by matching the real-time status with the preset device nodes, preset fault nodes and preset feature nodes contained in the preset operation and maintenance decision knowledge graph; If the real-time status matches the preset device nodes, preset fault nodes, and preset feature nodes contained in the preset operation and maintenance decision knowledge graph, a decision node is determined in the preset operation and maintenance decision knowledge graph based on the matched device nodes, matched fault nodes, and matched feature nodes. The first operation and maintenance decision result is determined based on the decision node.

5. The operation and maintenance decision-making method for new energy power stations as described in claim 4, characterized in that, The operation and maintenance decision-making method for the new energy power stations also includes: If the real-time status does not match the preset device nodes, preset fault nodes, and preset feature nodes contained in the preset operation and maintenance decision knowledge graph, calculate the similarity between the features in the real-time status and the actual features of the preset feature nodes; If the actual feature similarity is less than the preset feature similarity, a new fault node, a new device node, and a new feature node and their corresponding edge features are created in the preset operation and maintenance decision knowledge graph based on the real-time status. Initial decision rules are generated based on the new fault node, new device node, and new feature node as features of the new decision node; Based on the new fault nodes, new device nodes, and new feature nodes, initial decision rules are generated as new decision node features and edge features to update the preset operation and maintenance decision knowledge graph.

6. The operation and maintenance decision-making method for new energy power stations as described in claim 1, characterized in that, The operation and maintenance decision-making method for the new energy power stations also includes: In the second operation and maintenance decision result, the actual fault type, the actual fault confidence level, and the actual fault data are obtained; The preset operation and maintenance decision knowledge graph is updated based on the actual fault type, the actual fault confidence level, and the actual fault data.

7. A decision-making device for the operation and maintenance of a new energy power station, characterized in that, The operation and maintenance decision-making device for the new energy power station includes: The acquisition module is used to acquire meteorological data, equipment operation data and location relationship data of the target new energy power station; The real-time status determination module is used to perform spatiotemporal fusion based on the spatiotemporal correlation characteristics of each of the target devices, fusing the meteorological data, the device operation data and the location relationship data, and to determine the real-time status of each of the target devices based on the spatiotemporal fusion characteristics obtained by the fusion. The first operation and maintenance decision module is used to determine the first operation and maintenance decision result based on the real-time status and the preset operation and maintenance decision knowledge graph. The physical verification module is used to perform physical verification on the first operation and maintenance decision result to obtain a physical verification result. During the physical verification process, the physical verification result is obtained based on a first verification relationship between the physical simulation result and the physical constraints of the first operation and maintenance decision result. The physical verification module is also used to perform digital twin simulation based on the first operation and maintenance decision result to obtain a digital twin simulation result. From the digital twin simulation result, the simulation equipment operating parameters of each target device are extracted. A preset security boundary library for security judgment of the simulation equipment operating parameters is obtained. Based on the preset security boundary library, physical constraint verification is performed on the simulation equipment operating parameters to obtain the physical verification result. The security verification module is used to perform security verification on the real-time state, the first operation and maintenance decision result, and the physical verification result respectively through reinforcement learning, to obtain the security verification result; wherein, The security verification of the real-time state is used to determine the correspondence between the real-time state and the spatiotemporal fusion feature; The security verification of the first operation and maintenance decision result is used to determine the correspondence between the real-time status and the first operation and maintenance decision result; The security verification of the physical verification results is used to determine the correspondence between the first operation and maintenance decision result and the physical verification results; The operation and maintenance decision determination module is used to verify the first operation and maintenance decision result based on the physical verification result and the security verification result, and obtain a second operation and maintenance decision result as the operation and maintenance decision result of the new energy power station.

8. A computer device, characterized in that, include: The system includes a memory and a processor, which are interconnected. The memory stores computer instructions, and the processor executes the computer instructions to perform the operation and maintenance decision-making method for new energy power stations as described in any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions, which are used to cause the computer to execute the operation and maintenance decision-making method for new energy power stations as described in any one of claims 1 to 7.

Images