Power transmission and transformation equipment fault early warning system based on online monitoring

Abstract

The invention discloses a power transmission and transformation equipment fault early warning system based on online monitoring, and the system comprises the following modules: a multi-parameter sensor module which is used for collecting data in real time; the wireless sensor network module is used for transmitting the data to the data processing center for processing; the data fusion module is used for generating a comprehensive health score of the equipment by adopting a multi-modal data fusion algorithm based on a self-evolution nested network; the trend prediction module introduces an adaptive residual module to carry out trend prediction based on a nonlinear time sequence analysis technology of a generative adversarial adaptive residual network; the early warning threshold value adjusting module is used for dynamically adjusting an early warning threshold value by adopting a self-adaptive evolutionary algorithm of meta reinforcement learning; the maintenance suggestion generation module is used for generating maintenance suggestions; and the power grid dispatching linkage module is used for automatically linking a power grid dispatching system and detecting and adjusting an equipment load. According to the invention, a multi-parameter fusion and generative adversarial network method is adopted, and an adaptive residual module is introduced, so that accurate early warning of faults of power transmission and transformation equipment is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of power system monitoring and early warning, and in particular to a power transmission and transformation equipment fault early warning system based on online monitoring. Background Art

[0002] As the scale of the power system continues to expand, the safety and stability of power transmission and transformation equipment, as the core component of power grid operation, have become particularly important. However, since power transmission and transformation equipment has been operating under high voltage, high load and complex environment for a long time, the occurrence of failures is unpredictable. Traditional power transmission and transformation equipment maintenance methods mostly rely on regular maintenance and manual inspections, which not only requires a huge workload, but also may lead to equipment hidden dangers that cannot be discovered in time, thus causing serious power accidents. Therefore, how to use intelligent means to monitor and warn of power transmission and transformation equipment faults has become an important issue in the power industry.

[0003] Existing fault monitoring technologies for power transmission and transformation equipment mainly rely on the monitoring of a single or a small number of parameters, such as temperature, current, voltage and other physical quantities. Anomaly detection methods based on these parameters usually use preset thresholds. When a parameter exceeds the threshold, the system will issue an alarm. There are multiple problems with this single parameter monitoring method. First, equipment failure is often the result of multiple factors, and the change of a single parameter may not be sufficient to fully reflect the true status of the equipment; second, the fixed warning threshold cannot adapt to changes in the equipment operating environment, such as seasonal climate change or load fluctuations, which can easily lead to false alarms or missed alarms; third, due to the complexity of equipment operation and the huge amount of data, relying on a single parameter cannot effectively capture the nonlinear change trend of equipment failures and cannot achieve accurate fault warnings.

[0004] In order to improve the accuracy of fault warning, multi-parameter fusion analysis methods have emerged in recent years, attempting to fuse data collected by multiple sensors and generate more accurate equipment health status through data fusion algorithms. However, most of the existing data fusion technologies are based on simple weighted averages or logical reasoning. Although these methods have improved the detection accuracy to a certain extent, they are still very limited for data under complex working conditions because they cannot dynamically and adaptively adjust the fusion weights. In addition, most of the existing fusion methods are based on static models and cannot be optimized in real time according to changes in equipment status. This makes it easy for the system's early warning capability to decline when the equipment operating environment and load conditions change.

[0005] Another challenge is that most existing fault prediction models rely on linear or simple machine learning algorithms, such as support vector machines (SVMs) and decision trees. These methods are effective in dealing with simple fault modes, but they are clearly insufficient when faced with more complex fault modes. In particular, the operating status and fault trends of power transmission and transformation equipment usually exhibit nonlinear, time-varying, and multidimensional characteristics, and traditional linear models cannot capture the dynamic changes in the health status of equipment. In addition, these methods cannot adaptively optimize and update the prediction model, making it difficult to meet the personalized needs of different equipment.

[0006] In addition, existing technologies also have significant limitations in generating maintenance recommendations. Traditional methods usually rely on historical data and experience for simple inferences. When equipment fails, the system can only give fixed recommendations, such as arranging maintenance or replacing parts. However, such experience-based recommendations are often not targeted, and it is difficult to provide accurate maintenance plans based on the specific failure mode and operating status of the equipment. This not only reduces the efficiency of fault handling, but may even lead to increased maintenance costs.

[0007] For the linkage problem of the power grid dispatching system, the existing technology is usually limited to manual operation. When the equipment warning level is high, the operator needs to manually adjust the load distribution of the power grid equipment, start redundant equipment, etc. This method that relies on human judgment and operation is not only inefficient, but also prone to delays, and cannot respond to the fault in the first time, thus affecting the stability of the power system.

[0008] In summary, the existing technologies have many deficiencies in fault monitoring, early warning and maintenance of power transmission and transformation equipment: first, the single parameter monitoring method cannot fully reflect the operating status of the equipment; second, the existing data fusion method cannot dynamically adjust the weight and is difficult to cope with complex working conditions; third, the traditional fault prediction model is difficult to capture the nonlinear changes in the health status of the equipment and cannot achieve accurate trend prediction; finally, the maintenance recommendations lack pertinence, and the linkage response efficiency of the power grid dispatching system is low. These technical defects lead to the instability of the fault early warning system in practical applications and increase the risk of equipment operation.

[0009] Therefore, how to provide a power transmission and transformation equipment fault warning system based on online monitoring is a problem that technical personnel in this field need to solve urgently. Summary of the invention

[0010] One purpose of the present invention is to propose a power transmission and transformation equipment fault warning system based on online monitoring. The present invention adopts a multi-parameter fusion technology combined with a self-evolving nested network and a method of generating adversarial adaptive residual networks. By real-time collection of the operating status data of the power transmission and transformation equipment, and multi-modal data fusion and fault trend prediction, and further combined with the adaptive evolutionary algorithm of meta-reinforcement learning, the dynamic adjustment of the fault warning threshold is realized. The system can not only accurately predict the equipment failure trend, but also generate intelligent maintenance suggestions according to different warning levels, and link with the power grid dispatching system, improving the real-time and accuracy of fault warning and the efficiency of system emergency response, and has the advantages of high safety, precise maintenance and strong linkage.

[0011] The power transmission and transformation equipment fault early warning system based on online monitoring according to an embodiment of the present invention includes the following modules:

[0012] Multi-parameter sensor module, used to collect real-time operating status data of power transmission and transformation equipment;

[0013] A wireless sensor network module is used to transmit the collected operating status data to a data processing center for processing;

[0014] The data fusion module uses a multimodal data fusion algorithm based on a self-evolving nested network and combines it with an adaptive weight distribution model to fuse the data of each sensor modality and generate a comprehensive health score for the device;

[0015] The trend prediction module is based on the nonlinear time series analysis technology of the generative adversarial adaptive residual network, and introduces an adaptive residual module to predict the trend of the comprehensive health score and historical operation data of the equipment;

[0016] The warning threshold adjustment module uses an adaptive evolutionary algorithm of meta-reinforcement learning to dynamically adjust the warning threshold based on the historical data and real-time operation status of the equipment;

[0017] The maintenance suggestion generation module is used to generate maintenance suggestions based on the fault risk index and warning level when the system detects an abnormality;

[0018] The power grid dispatch linkage module is used to automatically link the power grid dispatch system when the equipment warning level reaches orange or red, detect and adjust the equipment load, and distribute the load to the backup equipment or start the redundant equipment when necessary.

[0019] According to an embodiment of the present invention, a method for early warning of power transmission and transformation equipment failure based on online monitoring comprises the following steps:

[0020] S1. Collecting the operating status data of the power transmission and transformation equipment in real time through multi-parameter sensors, wherein the operating status data includes temperature, vibration, current, voltage and partial discharge information;

[0021] S2, transmitting the collected operation status data to the data processing center for processing through the wireless sensor network;

[0022] S3, adopts a multimodal data fusion algorithm based on self-evolving nested networks and combines it with an adaptive weight distribution model to perform data fusion and generate a comprehensive health score for the device;

[0023] S4. Based on the nonlinear time series analysis technology of generative adversarial adaptive residual network, an adaptive residual module is introduced into the generator to predict the trend of the comprehensive health score and historical operation data of the equipment, monitor the changes of equipment health index, and identify fault trends;

[0024] S5. Based on the historical data and real-time operation status of the equipment, an adaptive evolutionary algorithm based on meta-reinforcement learning is used to dynamically adjust the warning threshold;

[0025] S6. When the system detects an abnormality, a maintenance suggestion is generated according to the warning level, wherein the warning level includes yellow warning, orange warning and red warning;

[0026] S7. When the equipment warning level reaches orange or red, the system will work together with the power grid dispatching system to automatically detect and adjust the equipment load, and distribute the load to the backup equipment or start the redundant equipment when necessary.

[0027] Optionally, the S3 specifically includes:

[0028] S31, the operating status data of the power transmission and transformation equipment collected by multi-parameter sensors, for each sensor mode D i Preprocessing is performed separately, where i = 1, 2, ..., n, n represents the number of sensor types, and the preprocessing includes data denoising, outlier removal, normalization and time window sliding operations to obtain a standardized data set D′ i ;

[0029] S32, establish a self-evolving nested network structure, for each sensor modality D′ i Construct the initial nested network N i , the weights of the nested network are initialized to a random normal distribution, and the nested network model N is defined i for:

[0030] F i =N(D′ i θ i ,L i );

[0031] Among them, θ i Represents a nested network N i The parameter set, L iRepresents the number of layers of the nested network, and initially for each modal data D′ i The network structure with the same number of layers is used for processing, and the network output is F i ;

[0032] S33, using an adaptive weight distribution model to weight the contribution of each modal output, the weight coefficient w i Dynamic update through multi-layer adaptive adjustment formula, the weight adjustment mechanism is based on the multi-modal nonlinear dependencies of sensor data;

[0033] Calculate the initial contribution of mode i at time t

[0034]

[0035] in, represents the output of mode i at time t, represents the output of mode j at time t, n represents the total number of modes, L total Represents the total loss function of multimodal fusion:

[0036]

[0037] in, represents the weight coefficient of mode i at time t, y true represents the real device health status value, λ represents the regularization coefficient, ∥θ i ∥ 2 Represents the L2 norm of the network parameters;

[0038] By adding the time decay term, the weight update value of the modality is calculated:

[0039]

[0040] in, represents the weight update value of mode i at time t, and ρ represents the time attenuation coefficient;

[0041] Add a nonlinear self-regulating function to perform a nonlinear transformation on the weights:

[0042]

[0043] Among them, exp represents the exponential function, κ represents the control factor, and μ represents the dynamic offset;

[0044] S34, for the nested network N i Perform self-evolution processing. The evolution process is based on the combination of genetic algorithm and adaptive neural network structure search algorithm to achieve dynamic optimization of network structure, including increasing or decreasing the number of network layers, changing the activation function type and adjusting the number of neurons.

[0045] S35. After each round of evolution, the fitness function of the network structure is calculated. The fitness function Fitness(N i ) is defined as:

[0046]

[0047] Among them, γ represents the weight balance coefficient, L i Indicates the network depth;

[0048] S36. Dynamically adjust the nested network structure during the evolution process and select the optimal network structure based on the fitness results The optimal fusion result obtained by each round of iterative calculation

[0049] S37. Repeat steps S33-S36 until the preset convergence condition is reached, the optimal fusion of the self-evolving nested network is completed, and finally a comprehensive health score of the device is generated.

[0050] Optionally, the S4 specifically includes:

[0051] S41. Comprehensive health score data H for the equipment t and historical operation data D t Perform multidimensional nonlinear time series modeling and define the historical comprehensive health score series and operating data series as:

[0052] H t ={H t ,H t-1 ,…,H t-k},D t ={D t ,D t-1 ,…,D t-k};

[0053] Where k represents the length of the time window, and the generator G generates the health score at the future moment, and the generated output is:

[0054]

[0055] in, represents the generated prediction score, θ G Represents the initial parameters of the generator;

[0056] S42, calculating the error between the generator output and the actual health score by introducing an adaptive residual module and performing feedback correction;

[0057] S43, after correction by the adaptive residual module, the output of the generator is updated to:

[0058]

[0059] in, represents the updated generator output, α t Represents the dynamic adjustment coefficient, R t Represents the residual at the current time t, R t-i represents the residual at time ti, exp represents the exponential function, k represents the total step length, W (i) represents the weight matrix of the i-th step;

[0060] S44, the generator and the discriminator are trained adversarially through an adaptive residual module, and the discriminator D is used to distinguish the true health score from the predicted value generated by the generator:

[0061]

[0062] Where m represents the number of samples processed;

[0063] S45. Through multiple adjustments and corrections of the adaptive residual module, the final output prediction of the generator is obtained and the fault trend is identified.

[0064] Optionally, the adaptive residual module correction in S42 specifically includes:

[0065] S421, define the residual R at the current moment t for:

[0066] R t =H t -G(H t-1 ,D t-1 θ G );

[0067] Among them, G(H t-1 ,D t-1 θ G ) represents the predicted output of the generator at time t-1;

[0068] S422, when the residual | R t When the threshold δ1 is exceeded, the generator will adjust the network depth according to the residual size:

[0069]

[0070] Among them, L new represents the adjusted network depth, L represents the current network depth, β represents the adjustment factor, Denotes the generator loss function L G About parameter θ G The gradient of represents the gradient of the residual;

[0071] S423, when the residual δ2<|R t When |≤δ1, the adaptive residual module adjusts the number of neurons in each layer N l Optimize network width:

[0072]

[0073] in, Represents the number of neurons in the adjusted layer l, N l represents the number of neurons in the current layer l, γ1 represents the learning rate, represents the weight matrix of the i-th step of the l-th layer, ∈ represents the smoothing term, and k represents the total step length;

[0074] S424, when the residual δ2<|R t When |≤δ1, the generator introduces cross-layer connections to improve the network’s expressive power:

[0075]

[0076] Among them, y l represents the output of layer l, y l+k represents the output of the l+kth layer, and λ1 represents the adjustment coefficient of the cross-layer connection.

[0077] This cross-layer connection improves the ability to identify complex nonlinear fault trends by directly transmitting low-level information to high-level layers.

[0078] Optionally, the S5 specifically includes:

[0079] S51. Build historical operation data of the equipment D t and comprehensive health score H t The state space of t for:

[0080] S t ={D t ,H t ,…,D t-k ,H t-k};

[0081] Among them, t represents the current time point, and k represents the time window length of historical data;

[0082] S52, according to the state space S t , define the action space A of the device t , the action space includes operations for dynamically adjusting the warning threshold:

[0083]

[0084] in, represents the i-th action, representing different warning threshold adjustment strategies;

[0085] S53, using the meta-reinforcement learning algorithm to train the adaptive evolution strategy, including the design of the reward function and strategy update, the reward function Q t To measure the effect of each action:

[0086]

[0087] in, Indicates the real health score H t and predictive health scores The loss between, λ2 represents the regularization coefficient, ∥θ t ∥ 2 represents the L2 regularization term of the parameter;

[0088] S54, based on the meta-learning mechanism of reinforcement learning, the strategy update rule is defined as: according to the current state S t and action a t , update the policy function π(a t |S t ), select the optimal action a according to the current state at each time step t :

[0089]

[0090] Among them, τ represents the learning rate, Q t Indicates the actual reward, represents the estimated reward, represents the gradient of parameter π;

[0091] S55. The model adaptively adjusts the warning threshold according to the historical operation data and current health status of the equipment. The formula for threshold adjustment is:

[0092]

[0093] Among them, θ t+1 represents the warning threshold at the next time t+1, θ t represents the current warning threshold at time t, ξ represents the adjustment step size, and ζ represents the discount factor

[0094] S56. When the operating status of the equipment changes significantly, the model updates the early warning model through the accumulated reward information.

[0095] Optionally, generating maintenance suggestions according to the warning level in S6 specifically includes:

[0096] Yellow warning: reminds the operation and maintenance personnel to pay special attention to the equipment during the next routine inspection and check the key components of the equipment;

[0097] Orange warning: The system generates more detailed fault speculation and maintenance suggestions, including analyzing historical failure modes, recommending replacement or repair of specific parts, and scheduling maintenance in advance;

[0098] Red warning: The system recommends immediate shutdown or emergency maintenance, and conducts a detailed analysis of possible failure modes, such as electrical short circuits, component wear, or insulation damage, prompting maintenance personnel to take emergency response measures.

[0099] The beneficial effects of the present invention are:

[0100] First, the system realizes real-time monitoring of power transmission and transformation equipment through multi-parameter sensor modules, and collects a variety of operating status data including temperature, vibration, current, voltage and partial discharge. This multi-parameter monitoring method makes up for the shortcomings of traditional single-parameter monitoring methods, and can more comprehensively reflect the operating status of equipment, especially in complex working conditions where equipment operates in multiple dimensions, which can significantly improve the accuracy and comprehensiveness of monitoring.

[0101] Secondly, the system uses a multimodal data fusion algorithm based on a self-evolving nested network, and combines it with an adaptive weight allocation model to fuse and analyze various sensor data to generate a comprehensive health score. This dynamic weight allocation mechanism enables the data fusion process to adapt to changes in the equipment operating environment, adjust the contribution of each modal data in real time, and avoid the false positives and negatives caused by fixed weights in existing technologies. In addition, the nested network can continuously optimize its own structure according to the equipment operating status, further improving the accuracy and robustness of data fusion.

[0102] In terms of fault trend prediction, the system introduces a generative adversarial adaptive residual network, which corrects the output and adjusts the architecture of the generator through the adaptive residual module, making the equipment fault prediction more accurate. Compared with the traditional linear model that cannot cope with the complex nonlinear changes of the equipment, the generative adversarial network of the present invention can capture the dynamic change trend of the health status of the equipment, continuously optimize the generator output through the residual feedback mechanism, greatly improve the accuracy and timeliness of fault prediction, and significantly reduce the risk of fault occurrence.

[0103] In addition, the system uses an adaptive evolutionary algorithm based on meta-reinforcement learning, which can dynamically adjust the warning threshold according to the historical operation data and real-time status of the equipment. This technology avoids the problem of inflexible warning caused by fixed thresholds in existing technologies. Through the reinforcement learning mechanism, the system can adaptively optimize the warning strategy according to different equipment and working conditions, improve the accuracy and adaptability of fault warning, thereby effectively reducing the false alarm rate and missed alarm rate, and enhancing the warning reliability of the system.

[0104] In terms of fault maintenance, the present invention generates corresponding maintenance suggestions based on the fault risk index and warning level. The system can automatically generate yellow, orange and red warnings, corresponding to different levels of fault handling strategies. Different from the prior art that relies on experience to generate fixed suggestions, the present invention provides a more detailed and accurate maintenance strategy based on multi-dimensional data analysis. Operation and maintenance personnel can perform targeted maintenance based on specific suggestions, thereby reducing fault handling time and improving maintenance efficiency.

[0105] Finally, the system realizes linkage with the power grid dispatching system. When the equipment warning level reaches orange or red, the system can automatically detect and adjust the load of the equipment, and distribute the load to the backup equipment or start the redundant equipment when necessary to ensure the stability of the power grid operation. This intelligent linkage mechanism greatly improves the emergency response capability of the system, effectively controls the risk in the early stage of the fault, prevents the fault from further deteriorating, and ensures the stability and safety of the entire power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0106] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0107] Figure 1 This is a schematic diagram of the structure of the power transmission and transformation equipment fault early warning system based on online monitoring proposed by the present invention;

[0108] Figure 2 This is an overall flow chart of the power transmission and transformation equipment fault early warning method based on online monitoring proposed by the present invention. DETAILED DESCRIPTION

[0109] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0110] refer to Figure 1 , the power transmission and transformation equipment fault early warning system based on online monitoring includes the following modules:

[0111] Multi-parameter sensor module, used to collect real-time operating status data of power transmission and transformation equipment;

[0112] A wireless sensor network module is used to transmit the collected operating status data to a data processing center for processing;

[0113] The data fusion module uses a multimodal data fusion algorithm based on a self-evolving nested network and combines it with an adaptive weight distribution model to fuse the data of each sensor modality and generate a comprehensive health score for the device;

[0114] The trend prediction module is based on the nonlinear time series analysis technology of the generative adversarial adaptive residual network, and introduces an adaptive residual module to predict the trend of the comprehensive health score and historical operation data of the equipment;

[0115] The warning threshold adjustment module uses an adaptive evolutionary algorithm of meta-reinforcement learning to dynamically adjust the warning threshold based on the historical data and real-time operation status of the equipment;

[0116] The maintenance suggestion generation module is used to generate maintenance suggestions based on the fault risk index and warning level when the system detects an abnormality;

[0117] The power grid dispatch linkage module is used to automatically link the power grid dispatch system when the equipment warning level reaches orange or red, detect and adjust the equipment load, and distribute the load to the backup equipment or start the redundant equipment when necessary.

[0118] refer to Figure 2 , a power transmission and transformation equipment fault early warning method based on online monitoring includes the following steps:

[0119] S1. Collecting the operating status data of the power transmission and transformation equipment in real time through multi-parameter sensors, wherein the operating status data includes temperature, vibration, current, voltage and partial discharge information;

[0120] S2, transmitting the collected operation status data to the data processing center for processing through the wireless sensor network;

[0121] S3, adopts a multimodal data fusion algorithm based on self-evolving nested networks and combines it with an adaptive weight distribution model to perform data fusion and generate a comprehensive health score for the device;

[0122] S4. Based on the nonlinear time series analysis technology of generative adversarial adaptive residual network, an adaptive residual module is introduced into the generator to predict the trend of the comprehensive health score and historical operation data of the equipment, monitor the changes of equipment health index, and identify fault trends;

[0123] S5. Based on the historical data and real-time operation status of the equipment, an adaptive evolutionary algorithm based on meta-reinforcement learning is used to dynamically adjust the warning threshold;

[0124] S6. When the system detects an abnormality, a maintenance suggestion is generated according to the warning level, wherein the warning level includes yellow warning, orange warning and red warning;

[0125] S7. When the equipment warning level reaches orange or red, the system will work together with the power grid dispatching system to automatically detect and adjust the equipment load, and distribute the load to the backup equipment or start the redundant equipment when necessary.

[0126] In this implementation, S3 specifically includes:

[0127] S31, the operating status data of the power transmission and transformation equipment collected by multi-parameter sensors, for each sensor mode D i Preprocessing is performed separately, where i = 1, 2, ..., n, n represents the number of sensor types, and the preprocessing includes data denoising, outlier removal, normalization and time window sliding operations to obtain a standardized data set D′ i ;

[0128] S32, establish a self-evolving nested network structure, for each sensor modality D′ i Construct the initial nested network N i , the weights of the nested network are initialized to a random normal distribution, and the nested network model N is defined i for:

[0129] F i =N(D′ i θ i ,L i );

[0130] Among them, θ i Represents a nested network N i The parameter set, L i Represents the number of layers of the nested network, and initially for each modal data D′ i The network structure with the same number of layers is used for processing, and the network output is F i ;

[0131] S33, using the adaptive weight distribution model to weight the contribution of each modal output, the weight coefficient w i Dynamic update through multi-layer adaptive adjustment formula, the weight adjustment mechanism is based on the multi-modal nonlinear dependencies of sensor data;

[0132] Calculate the initial contribution of mode i at time t

[0133]

[0134] in, represents the output of mode i at time t, represents the output of mode j at time t, n represents the total number of modes, L total Represents the total loss function of multimodal fusion:

[0135]

[0136] in, represents the weight coefficient of mode i at time t, y true represents the real device health status value, λ represents the regularization coefficient, ∥θi ∥ 2 Represents the L2 norm of the network parameters;

[0137] By adding the time decay term, the weight update value of the modality is calculated:

[0138]

[0139] in, represents the weight update value of mode i at time t, and ρ represents the time attenuation coefficient;

[0140] Add a nonlinear self-regulating function to perform a nonlinear transformation on the weights:

[0141]

[0142] Among them, exp represents the exponential function, κ represents the control factor, and μ represents the dynamic offset;

[0143] S34, for the nested network N i Perform self-evolution processing. The evolution process is based on the combination of genetic algorithm and adaptive neural network structure search algorithm to achieve dynamic optimization of network structure, including increasing or decreasing the number of network layers, changing the activation function type and adjusting the number of neurons.

[0144] S35. After each round of evolution, the fitness function of the network structure is calculated. The fitness function Fitness(N i ) is defined as:

[0145]

[0146] Among them, γ represents the weight balance coefficient, L i Indicates the network depth;

[0147] S36. Dynamically adjust the nested network structure during the evolution process and select the optimal network structure based on the fitness results The optimal fusion result obtained by each round of iterative calculation

[0148] S37. Repeat steps S33-S36 until the preset convergence condition is reached, the optimal fusion of the self-evolving nested network is completed, and finally a comprehensive health score of the device is generated.

[0149] In this implementation, S4 specifically includes:

[0150] S41. Comprehensive health score data H for the equipment t and historical operation data D t Perform multidimensional nonlinear time series modeling and define the historical comprehensive health score series and operating data series as:

[0151] H t ={H t ,H t-1 ,…,H t-k},D t ={D t ,D t-1 ,…,D t-k};

[0152] Where k represents the length of the time window, and the generator G generates the health score at the future moment, and the generated output is:

[0153]

[0154] in, represents the generated prediction score, θ G Represents the initial parameters of the generator;

[0155] S42, calculating the error between the generator output and the actual health score by introducing an adaptive residual module and performing feedback correction;

[0156] S43, after correction by the adaptive residual module, the output of the generator is updated to:

[0157]

[0158] in, represents the updated generator output, α t Represents the dynamic adjustment coefficient, R t Represents the residual at the current time t, R t-i represents the residual at time ti, exp represents the exponential function, k represents the total step length, W (i) represents the weight matrix of the i-th step;

[0159] S44, the generator and the discriminator are trained adversarially through an adaptive residual module, and the discriminator D is used to distinguish the true health score from the predicted value generated by the generator:

[0160]

[0161] Where m represents the number of samples processed;

[0162] S45. Through multiple adjustments and corrections of the adaptive residual module, the final output prediction of the generator is obtained and the fault trend is identified.

[0163] In this implementation manner, the adaptive residual module correction in S42 specifically includes:

[0164] S421, define the residual R at the current moment t for:

[0165] R t =H t -G(H t-1 ,D t-1 θ G );

[0166] Among them, G(H t-1 ,D t-1 θ G ) represents the predicted output of the generator at time t-1;

[0167] S422, when the residual | R t When the threshold δ1 is exceeded, the generator will adjust the network depth according to the residual size:

[0168]

[0169] Among them, L new represents the adjusted network depth, L represents the current network depth, β represents the adjustment factor, Denotes the generator loss function L G About parameter θ G The gradient of represents the gradient of the residual;

[0170] S423, when the residual δ2<|R t When |≤δ1, the adaptive residual module adjusts the number of neurons in each layer N l Optimize network width:

[0171]

[0172] in, Represents the number of neurons in the adjusted layer l, N l represents the number of neurons in the current layer l, γ1 represents the learning rate, represents the weight matrix of the i-th step of the l-th layer, ∈ represents the smoothing term, and k represents the total step length;

[0173] S424, when the residual δ2<|R t When |≤δ1, the generator introduces cross-layer connections to improve the network’s expressive power:

[0174]

[0175] Among them, y l represents the output of layer l, y l+k represents the output of the l+kth layer, and λ1 represents the adjustment coefficient of the cross-layer connection.

[0176] This cross-layer connection improves the ability to identify complex nonlinear fault trends by directly transmitting low-level information to high-level layers.

[0177] In this implementation manner, S5 specifically includes:

[0178] S51. Build historical operation data of the equipment D t and comprehensive health score H t The state space of t for:

[0179] S t ={D t ,H t ,…,D t-k ,H t-k};

[0180] Among them, t represents the current time point, and k represents the time window length of historical data;

[0181] S52, according to the state space S t , define the action space A of the device t , the action space includes operations for dynamically adjusting the warning threshold:

[0182]

[0183] in, represents the i-th action, representing different warning threshold adjustment strategies;

[0184] S53, using the meta-reinforcement learning algorithm to train the adaptive evolution strategy, including the design of the reward function and strategy update, the reward function Q t To measure the effect of each action:

[0185]

[0186] in, Indicates the real health score H t and predictive health scores The loss between, λ2 represents the regularization coefficient, ∥θ t ∥ 2 represents the L2 regularization term of the parameter;

[0187] S54, based on the meta-learning mechanism of reinforcement learning, the strategy update rule is defined as: according to the current state S t and action a t , update the policy function π(a t |S t ), select the optimal action a according to the current state at each time step t :

[0188]

[0189] Among them, τ represents the learning rate, Q t Indicates the actual reward, represents the estimated reward, represents the gradient of parameter π;

[0190] S55. The model adaptively adjusts the warning threshold according to the historical operation data and current health status of the equipment. The formula for threshold adjustment is:

[0191]

[0192] Among them, θ t+1 represents the warning threshold at the next time t+1, θ t represents the current warning threshold at time t, ξ represents the adjustment step size, and ζ represents the discount factor

[0193] S56. When the operating status of the equipment changes significantly, the model updates the early warning model through the accumulated reward information.

[0194] In this implementation manner, generating maintenance suggestions according to the warning level in S6 specifically includes:

[0195] Yellow warning: reminds the operation and maintenance personnel to pay special attention to the equipment during the next routine inspection and check the key components of the equipment;

[0196] Orange warning: The system generates more detailed fault speculation and maintenance suggestions, including analyzing historical failure modes, recommending replacement or repair of specific parts, and scheduling maintenance in advance;

[0197] Red warning: The system recommends immediate shutdown or emergency maintenance, and conducts a detailed analysis of possible failure modes, such as electrical short circuits, component wear, or insulation damage, prompting maintenance personnel to take emergency response measures.

[0198] Embodiment 1:

[0199] In order to verify the feasibility of the present invention in implementation, the present invention is applied to a high-voltage transmission and transformation substation in the North China Power Grid. Since the power transmission and transformation equipment is operated in a high-voltage and high-load environment for a long time, the occurrence of equipment failure poses a great threat to the stability and security of the power grid. In the past, since most of the equipment in the substation adopted the traditional single parameter monitoring system, only the physical parameters such as temperature, current or voltage of the equipment were monitored. Once these parameters exceeded the preset fixed threshold, the system would sound an alarm. However, this method has a high false alarm rate, especially in the case of temperature changes and load fluctuations, the false alarm frequency is high, resulting in a large amount of invalid equipment maintenance work. At the same time, because the traditional method fails to fully consider the multi-dimensional operating status of the equipment, many potential equipment failures have not been discovered in time, resulting in the high frequency of sudden equipment failures in the past few years, which has brought huge hidden dangers to the safety of the power grid.

[0200] In order to solve these problems, North China Power Grid introduced a power transmission and transformation equipment fault warning system based on online monitoring. The system uses a multi-parameter sensor module to collect various operating status data of power transmission and transformation equipment in real time, such as temperature, vibration, current, voltage, partial discharge, etc. These data are transmitted to the data processing center through a wireless sensor network. The system uses a multi-modal data fusion algorithm based on a self-evolving nested network to comprehensively process various sensor data to generate a comprehensive health score for the equipment.

[0201] In a certain application, the temperature data of a main transformer continued to be higher than the normal range during the peak load period in summer, but because the traditional system used a fixed warning threshold, it was unable to determine whether this anomaly was a real potential fault. The new system not only collects temperature data, but also combines multi-dimensional data such as vibration, partial discharge, and current, and performs data fusion through a self-evolving nested network to generate a comprehensive health score for the device. The results show that the comprehensive health score only dropped slightly and did not meet the system's warning standard. The system dynamically adjusted the weight of the temperature parameter based on the fused health score to avoid false alarms caused by single data anomalies. Finally, through the trend prediction module, the device was judged to be operating normally and did not need to be shut down for maintenance, which greatly saved maintenance costs.

[0202] In another actual application, the switchgear of a certain substation showed large fluctuations in vibration and abnormal current in the multi-dimensional monitoring data. The traditional system only monitors the current parameters and fails to issue a timely warning at the early stage of the abnormality, resulting in unexpected fault tripping of the switchgear during use, which has a great impact on the operation of the power grid. After the introduction of the new system, the system identified the fault trend of the equipment in advance by generating a trend prediction technology of adversarial adaptive residual network before the fault occurred. The adaptive residual module conducted an in-depth analysis of the comprehensive health score and historical operation data of the equipment. The generated trend prediction data showed that the rate of decline of the health index of the equipment was significantly accelerated. The system issued an orange warning and recommended that the operation and maintenance personnel conduct further inspections of the equipment.

[0203] Based on the maintenance suggestions generated by the system, the operation and maintenance personnel quickly inspected the switchgear and found that there was a potential tripping risk due to wear of mechanical parts caused by long-term high-load operation. Based on the maintenance suggestions provided by the system, the operation and maintenance personnel replaced the relevant parts in advance and successfully avoided a possible equipment failure tripping event. This example proves that the prediction accuracy of the new system has been greatly improved, and it can identify equipment failure risks in advance, providing a sufficient time window for power grid dispatching and maintenance.

[0204] In addition, in another 500kV substation located in Hebei Province, the new system was effectively linked with the grid dispatching system. During a high-load operation, the partial discharge data of the main transformer equipment showed abnormal fluctuations. The system predicted that the health score of the equipment would continue to decline through trend analysis of the adaptive residual network, and the system immediately issued a red warning. The system automatically linked with the grid dispatching center, first reducing the load of the equipment, and then automatically started the backup redundant equipment to ensure the stable operation of the grid. In the end, the equipment was further detected and repaired during the load reduction operation, and did not affect the operation of the grid.

[0205] According to data provided by the substation, the system conducted a total of 256 equipment status assessments from June 2019 to June 2020, of which 78 equipment anomalies were detected, and the system generated 52 yellow warnings, 19 orange warnings, and 7 red warnings. In the orange and red warning events, the system linked the power grid dispatching system and started 5 redundant equipment and 13 load transfer operations, all of which effectively avoided sudden failures of power grid equipment. Compared with the previous traditional early warning system, the system's false alarm rate was reduced by about 60%, and no power outages caused by sudden equipment failures occurred.

[0206] Through these practical application scenarios, it can be seen that the power transmission and transformation equipment fault warning system based on online monitoring of the present invention can significantly improve the accuracy and timeliness of fault warnings when dealing with complex power equipment operating environments, reduce false alarms and missed alarms, improve the efficiency of equipment maintenance, and reduce the risks and costs of power system operation. Significant beneficial effects have been achieved in equipment failure prediction, warning threshold adjustment, maintenance suggestion generation, and power grid dispatch linkage, greatly improving the safety and stability of the power system.

[0207] Table 1 Comparative analysis of power transmission and transformation equipment status based on traditional and intelligent fault warning systems

[0208]

[0209]

[0210] Through the comparative analysis in Table 1, the traditional power transmission and transformation equipment fault warning system and the newly introduced intelligent monitoring system show significant differences in equipment status monitoring and fault warning. First, in terms of the number of equipment status evaluations, the new system evaluated the equipment status 256 times in one year, an increase of 28% compared to the 200 times of the traditional system. This improvement is due to the new system's ability to monitor more accurately and automatically, reducing the need for human intervention and significantly improving the frequency and efficiency of equipment status monitoring.

[0211] In terms of the number of times equipment anomalies were detected, the traditional system only detected 60 equipment anomalies, while the new system detected 78 times, an increase of 30%. This increase reflects that the new system can identify more potential failure risks by monitoring the operating status of equipment more comprehensively and finely through multimodal data fusion and trend prediction technology.

[0212] For the warning level, the new system has 52 yellow warnings, while the traditional system has 30, indicating that the new system can detect mild anomalies earlier, which helps to take preventive measures in advance to prevent small problems from turning into serious failures. At the same time, the number of orange and red warnings has also increased significantly, to 19 and 7 respectively, indicating that the new system can accurately identify moderate and severe anomalies of equipment and issue more targeted alarms, thereby improving the response speed and maintenance efficiency of operation and maintenance personnel.

[0213] In terms of redundant equipment startup and load transfer operations, the new system is significantly better than the traditional system, with the number of redundant equipment startups increased from 2 to 5 times, and the load transfer operations increased from 5 to 13 times. This reflects the intelligent linkage between the new system and the power grid dispatching system, which can quickly take effective dispatching measures when serious warnings are issued by the equipment, reduce the impact of equipment failures on power grid operations, and ensure power grid stability.

[0214] Finally, it is worth noting that the false alarm rate has dropped significantly, which is about 60% lower than that of the traditional system. This result shows that the multi-parameter fusion and adaptive warning threshold adjustment mechanism of the new system can effectively avoid false alarms and reduce ineffective maintenance work. In addition, within one year, the new system did not experience power outages due to equipment failure, while the traditional system experienced two power outages in the same period of time. This result fully demonstrates the superiority of the new system in preventing major failures.

[0215] In general, based on the comparative analysis of this table, it can be concluded that the new system is not only significantly superior to the traditional system in terms of comprehensiveness and accuracy of equipment status monitoring, but also effectively reduces the risk of equipment failure through intelligent early warning and dispatch linkage, and improves the overall operation stability and safety of power transmission and transformation equipment. This shows that the new system has strong application prospects and technical advantages in the field of early warning and maintenance of power transmission and transformation equipment.

[0216] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. The power transmission and transformation equipment fault early warning system based on online monitoring is characterized by: Includes the following modules: Multi-parameter sensor module, used to collect real-time operating status data of power transmission and transformation equipment; A wireless sensor network module is used to transmit the collected operating status data to a data processing center for processing; The data fusion module uses a multimodal data fusion algorithm based on a self-evolving nested network and combines it with an adaptive weight distribution model to fuse the data of each sensor modality and generate a comprehensive health score for the device; The trend prediction module is based on the nonlinear time series analysis technology of the generative adversarial adaptive residual network, and introduces an adaptive residual module to predict the trend of the comprehensive health score and historical operation data of the equipment; The warning threshold adjustment module uses an adaptive evolutionary algorithm of meta-reinforcement learning to dynamically adjust the warning threshold based on the historical data and real-time operation status of the equipment; The maintenance suggestion generation module is used to generate maintenance suggestions based on the fault risk index and warning level when the system detects an abnormality; The power grid dispatch linkage module is used to automatically link the power grid dispatch system when the equipment warning level reaches orange or red, detect and adjust the equipment load, and distribute the load to the backup equipment or start the redundant equipment when necessary.

2. According to the power transmission and transformation equipment fault early warning system based on online monitoring according to claim 1, a power transmission and transformation equipment fault early warning method based on online monitoring is implemented, characterized in that: The steps include: S1. Collecting the operating status data of the power transmission and transformation equipment in real time through multi-parameter sensors, wherein the operating status data includes temperature, vibration, current, voltage and partial discharge information; S2, transmitting the collected operation status data to the data processing center for processing through the wireless sensor network; S3, adopts a multimodal data fusion algorithm based on self-evolving nested networks and combines it with an adaptive weight distribution model to perform data fusion and generate a comprehensive health score for the device; S4. Based on the nonlinear time series analysis technology of generative adversarial adaptive residual network, an adaptive residual module is introduced into the generator to predict the trend of the comprehensive health score and historical operation data of the equipment, monitor the changes of equipment health index, and identify fault trends; S5. Based on the historical data and real-time operation status of the equipment, an adaptive evolutionary algorithm based on meta-reinforcement learning is used to dynamically adjust the warning threshold; S6. When the system detects an abnormality, a maintenance suggestion is generated according to the warning level, wherein the warning level includes yellow warning, orange warning and red warning; S7. When the equipment warning level reaches orange or red, the system will work together with the power grid dispatching system to automatically detect and adjust the equipment load, and distribute the load to the backup equipment or start the redundant equipment when necessary.

3. The method for early warning of power transmission and transformation equipment failure based on online monitoring according to claim 2 is characterized in that: The S3 specifically includes: S31, the operating status data of the power transmission and transformation equipment collected by multi-parameter sensors, for each sensor mode D i Preprocessing is performed separately, where i = 1, 2, ..., n, n represents the number of sensor types, and the preprocessing includes data denoising, outlier removal, normalization and time window sliding operations to obtain a standardized data set D′ i ; S32, establish a self-evolving nested network structure, for each sensor modality D′ i Construct the initial nested network N i , the weights of the nested network are initialized to a random normal distribution, and the nested network model N is defined i for: F i =N(D′ i ;θ i ,L i ); Among them, θ i Represents a nested network N i The parameter set, L i Represents the number of layers of the nested network, and initially for each modal data D′ i The network structure with the same number of layers is used for processing, and the network output is F i ; S33, using an adaptive weight distribution model to weight the contribution of each modal output, the weight coefficient w i Dynamic update through multi-layer adaptive adjustment formula, the weight adjustment mechanism is based on the multi-modal nonlinear dependencies of sensor data; Calculate the initial contribution of mode i at time t in, represents the output of mode i at time t, represents the output of mode j at time t, n represents the total number of modes, L total Represents the total loss function of multimodal fusion: in, represents the weight coefficient of mode i at time t, y true represents the real device health status value, λ represents the regularization coefficient, ∥θ i ∥ 2 Represents the L2 norm of the network parameters; By adding the time decay term, the weight update value of the modality is calculated: in, represents the weight update value of mode i at time t, and ρ represents the time attenuation coefficient; Add a nonlinear self-regulating function to perform a nonlinear transformation on the weights: Among them, exp represents the exponential function, κ represents the control factor, and μ represents the dynamic offset; S34, for the nested network N i Perform self-evolution processing. The evolution process is based on the combination of genetic algorithm and adaptive neural network structure search algorithm to achieve dynamic optimization of network structure, including increasing or decreasing the number of network layers, changing the activation function type and adjusting the number of neurons. S35. After each round of evolution, the fitness function of the network structure is calculated. The fitness function Fitness(N i ) is defined as: Among them, γ represents the weight balance coefficient, L i Indicates the network depth; S36. Dynamically adjust the nested network structure during the evolution process and select the optimal network structure based on the fitness results The optimal fusion result obtained by each round of iterative calculation S37. Repeat steps S33-S36 until the preset convergence condition is reached, the optimal fusion of the self-evolving nested network is completed, and finally a comprehensive health score of the device is generated.

4. The method for early warning of power transmission and transformation equipment failure based on online monitoring according to claim 2 is characterized in that: The S4 specifically includes: S41. Comprehensive health score data H for the equipment t and historical operation data D t Perform multidimensional nonlinear time series modeling and define the historical comprehensive health score series and operating data series as: H t ={H t ,H t-1 ,…,H t-k },D t ={D t ,D t-1 ,…,D t-k }; Where k represents the length of the time window, and the generator G generates the health score at the future moment, and the generated output is: in, represents the generated prediction score, θ G Represents the initial parameters of the generator; S42, calculating the error between the generator output and the actual health score by introducing an adaptive residual module and performing feedback correction; S43, after correction by the adaptive residual module, the output of the generator is updated to: in, represents the updated generator output, α t Represents the dynamic adjustment coefficient, R t Represents the residual at the current time t, R t-i represents the residual at time ti, exp represents the exponential function, k represents the total step length, W (i) represents the weight matrix of the i-th step; S44, the generator and the discriminator are trained adversarially through an adaptive residual module, and the discriminator D is used to distinguish the true health score from the predicted value generated by the generator: Where m represents the number of samples processed; S45. Through multiple adjustments and corrections of the adaptive residual module, the final output prediction of the generator is obtained and the fault trend is identified.

5. The method for early warning of power transmission and transformation equipment failure based on online monitoring according to claim 4 is characterized in that: The adaptive residual module correction in S42 specifically includes: S421, define the residual R at the current moment t for: R t =H t -G(H t-1 ,D t-1 ;θ G ); Among them, G(H t-1 ,D t-1 θ G ) represents the predicted output of the generator at time t-1; S422, when the residual | R t When the threshold δ1 is exceeded, the generator will adjust the network depth according to the residual size: Among them, L new represents the adjusted network depth, L represents the current network depth, β represents the adjustment factor, Denotes the generator loss function L G About parameter θ G The gradient of represents the gradient of the residual; S423, when the residual δ2<|R t When |≤δ1, the adaptive residual module adjusts the number of neurons in each layer N l Optimize network width: in, Represents the number of neurons in the lth layer after adjustment, N l represents the number of neurons in the current layer l, γ1 represents the learning rate, represents the weight matrix of the i-th step of the l-th layer, ∈ represents the smoothing term, and k represents the total step length; S424, when the residual δ2<|R t When |≤δ1, the generator introduces cross-layer connections to improve the network’s expressive power: Among them, y l represents the output of layer l, y l+k represents the output of the l+kth layer, and λ1 represents the adjustment coefficient of the cross-layer connection.

6. The method for early warning of power transmission and transformation equipment failure based on online monitoring according to claim 2 is characterized in that: The S5 specifically includes: S51. Build historical operation data of the equipment D t and comprehensive health score H t The state space of t for: S t ={D t ,H t ,…,D t-k ,H t-k }; Among them, t represents the current time point, and k represents the time window length of historical data; S52, according to the state space S t , define the action space A of the device t , the action space includes operations for dynamically adjusting the warning threshold: in, represents the i-th action, representing different warning threshold adjustment strategies; S53, using the meta-reinforcement learning algorithm to train the adaptive evolution strategy, including the design of the reward function and the strategy update, the reward function Q t To measure the effect of each action: in, Indicates the real health score H t and predictive health scores The loss between, λ2 represents the regularization coefficient, ∥θ t ∥ 2 represents the L2 regularization term of the parameter; S54, based on the meta-learning mechanism of reinforcement learning, the strategy update rule is defined as: according to the current state S t and action a t , update the policy function π(a t |S t ), select the optimal action a according to the current state at each time step t : Among them, τ represents the learning rate, Q t Indicates the actual reward, represents the estimated reward, represents the gradient of parameter π; S55. The model adaptively adjusts the warning threshold according to the historical operation data and current health status of the equipment. The formula for threshold adjustment is: Among them, θ t+1 represents the warning threshold at the next time t+1, θ t represents the current warning threshold at time t, ξ represents the adjustment step size, and ζ represents the discount factor S56. When the operating status of the equipment changes significantly, the model updates the early warning model through the accumulated reward information.

7. The method for early warning of power transmission and transformation equipment failure based on online monitoring according to claim 2 is characterized in that: Generating maintenance suggestions according to the warning level in S6 specifically includes: Yellow warning: reminds the operation and maintenance personnel to pay special attention to the equipment during the next routine inspection and check the key components of the equipment; Orange warning: The system generates more detailed fault speculation and maintenance suggestions, including analyzing historical failure modes, recommending replacement or repair of specific parts, and scheduling maintenance in advance; Red warning: The system recommends immediate shutdown or emergency maintenance, and conducts a detailed analysis of possible failure modes, such as electrical short circuits, component wear, or insulation damage, prompting maintenance personnel to take emergency response measures.

Images