A state analysis and early warning method and system based on online monitoring data of a catenary

Abstract

The application provides a state analysis and early warning method and system based on online monitoring data of a catenary, and belongs to the technical field of catenary monitoring. The method comprises the following steps: acquiring electrical data, environmental data and mechanical data of a target catenary; preprocessing and fusing the electrical data and the mechanical data, inputting the data into a preset catenary state judgment model, and correcting the data through the environmental data to obtain a final health state; judging the health state of the catenary based on the final health state to obtain a judgment result; if the judgment result is in an attention state, first early warning information is generated; if the judgment result is in a normal state, a state trend of the final health state is predicted to obtain predicted health state information of a next period, and it is judged whether the state is in an attention state or an abnormal state; if the state is in the attention state, second early warning information is generated; and if the state is in the abnormal state, third early warning information is generated.

Description

Technical Field

[0001] This application relates to the field of overhead contact line monitoring technology, and in particular to a method and system for status analysis and early warning based on online monitoring data of overhead contact lines. Background Technology

[0002] The overhead contact system is the only source of power for trains, and its operational stability determines the safety and efficiency of railway transportation. However, the overhead contact system is exposed to the elements, has numerous points of contact, a long length, and no backup equipment. It is subjected to pantograph friction and vibration, dynamic train loads, and is also exposed to complex environmental factors such as temperature, icing, wind speed, and corrosive gases. Failures are prone to numerous and sudden causes, and can easily lead to train stoppages, delays, or even safety accidents. Existing monitoring relies on manual inspections and single-threshold alarms, resulting in data silos, poor real-time performance, and delayed hazard identification, making it difficult to meet the maintenance needs of large-scale high-speed rail operations. With the acceleration and capacity expansion of electrified railways, the operating conditions of the overhead contact system are becoming more demanding, significantly increasing the requirements for the accuracy of condition monitoring and the timeliness of early warnings. Current technologies lack deep integration of multi-source data and dynamic correction of environmental factors, making it impossible to achieve a comprehensive assessment of health status and trend prediction.

[0003] Therefore, there is an urgent need for a status analysis and early warning method and system based on online monitoring data of the overhead contact system to provide technical support for proactive operation and maintenance. Summary of the Invention

[0004] To address the aforementioned technical problems, this application provides a method and system for status analysis and early warning based on online monitoring data of overhead contact lines.

[0005] A first aspect of this application provides a status analysis and early warning method based on online monitoring data of overhead contact lines, comprising: Electrical, environmental, and mechanical data of the target contact network are acquired through sensors installed on the contact network. The electrical data and mechanical data are preprocessed and fused, and then input into a preset contact network status judgment model to obtain an initial health status. The initial health status is then corrected based on the environmental data to obtain a final health status. The health status of the overhead contact system is assessed based on the final health status, resulting in a judgment result; the judgment result includes a normal status, a warning status, and an abnormal status. If the judgment result is in an abnormal state, an alarm message is generated; if the judgment result is in an attentive state, a first warning message is generated. If the judgment result is in a normal state, then a state trend prediction is performed on the final health state to obtain the predicted health state information for the next preset period; it is then determined whether the predicted health state information is in the attention state or the abnormal state. If the predicted health state information is in the attention state, then a second warning message is generated; if the predicted health state information is in the abnormal state, then a third warning message is generated. The alarm information, the first warning information, the second warning information, or the third warning information are identified based on different preset levels and then output to the operation and maintenance monitoring platform, mobile terminal, or audible and visual alarm device.

[0006] A second aspect of this application provides a status analysis and early warning system based on online monitoring data of overhead contact lines, comprising: The data acquisition module is used to acquire electrical, environmental, and mechanical data of the target contact network through sensors installed on the contact network; The status assessment module is used to preprocess and fuse the electrical data and the mechanical data, and input them into a preset contact network status judgment model to obtain an initial health status. Based on the environmental data, the initial health status is corrected to obtain a final health status. The status grading module is used to determine the health status of the overhead contact system based on the final health status, and obtain a judgment result; the judgment result includes normal status, alert status, and abnormal status; An instant response module is used to generate an alarm message if the judgment result is in an abnormal state, and to generate a first warning message if the judgment result is in an attentive state. The trend warning module is used to predict the final health status trend if the judgment result is in a normal state, and obtain the predicted health status information for the next preset period; determine whether the predicted health status information is in the attention state or the abnormal state; if the predicted health status information is in the attention state, generate a second warning message; if the predicted health status information is in the abnormal state, generate a third warning message. The information output module is used to identify the alarm information, the first warning information, the second warning information, or the third warning information based on different preset levels, and then output them to the operation and maintenance monitoring platform, mobile terminal, or audible and visual alarm device.

[0007] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the above-described state analysis and early warning method based on online monitoring data of the overhead contact line.

[0008] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described state analysis and early warning method based on online monitoring data of the overhead contact line.

[0009] The beneficial effects of the status analysis and early warning method and system based on online monitoring data of the overhead contact system provided in this application are as follows: This application uses sensor-acquired electrical, mechanical, and environmental data of the overhead contact system to break through the limitations of data fragmentation in traditional monitoring and achieve comprehensive perception of equipment operating status. Based on data preprocessing and fusion technology and dynamic correction of the initial health status by environmental factors, the accuracy of health assessment is improved, effectively avoiding the one-sidedness of single data assessment and misjudgment caused by environmental interference. Furthermore, by dividing the status into three levels—normal, alert, and abnormal—a two-layer early warning mechanism is constructed, combining current status response and future trend prediction. This mechanism not only quickly generates alarm information for immediate anomalies but also predicts trends in normal states, identifying potential hazards in advance and issuing graded warnings, effectively improving the initiative of operation and maintenance. Alarm and early warning information of different levels is accurately pushed to multiple terminals, achieving optimized allocation of operation and maintenance resources, reducing ineffective inspections, and lowering operation and maintenance costs. Attached Figure Description

[0010] Figure 1 A flowchart illustrating a status analysis and early warning method based on online monitoring data of overhead contact lines provided in an embodiment of this application; Figure 2 This application provides a structural block diagram of a status analysis and early warning system based on online monitoring data of overhead contact lines, according to an embodiment of the present application. Figure 3 This is a schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0011] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0012] To make the purpose, technical solution, and advantages of this application clearer, the following will be described in conjunction with the appendix. Figure 1 - Appendix Figure 3 The following is an explanation using specific examples.

[0013] Please refer to Figure 1 , Figure 1This is a flowchart illustrating a status analysis and early warning method based on online monitoring data of overhead contact lines, provided in an embodiment of this application. The method includes: S101: Acquire electrical, environmental, and mechanical data of the target contact network through sensors installed on the contact network.

[0014] In this embodiment, multiple types of sensors are rationally deployed in components such as the catenary, contact wire, cantilever arm, and positioner of the overhead contact system, as well as along the line. Among them, electrical sensors include current sensors, voltage sensors, insulation monitoring sensors, and arc detection sensors, which collect electrical data such as traction current, contact voltage, insulation resistance value, and frequency and intensity of arc occurrence during the operation of the overhead contact system in real time.

[0015] Mechanical sensors include tension sensors, displacement sensors, vibration sensors, and wear detection sensors. They collect mechanical data such as the tension value of the contact wire and the catenary, the displacement of the cantilever arm and the positioner, the dynamic vibration frequency and amplitude of the contact network, and the wear thickness of the contact wire, which indicate the stress state, geometric deformation, and component wear of the contact network structure.

[0016] Environmental sensors collect data such as ambient temperature, wind speed, wind direction, ice thickness, dust concentration, humidity, and corrosive gas concentration.

[0017] All sensors are equipped with real-time data transmission capabilities, and the collected multi-source data is synchronously uploaded to the data processing center via wireless communication modules or wired transmission links.

[0018] S102: Preprocess and fuse electrical and mechanical data, input them into a preset contact network status judgment model to obtain an initial health status, and correct the initial health status based on environmental data to obtain the final health status.

[0019] In this embodiment, the Kalman filter algorithm is used to smooth and denoise the current and voltage fluctuation noise in the electrical data and the vibration interference signal in the mechanical data to eliminate random errors. Missing data caused by sensor failure or transmission interruption is completed by interpolation based on time series. At the same time, the amplitude, frequency and other parameters of the electrical data and the tension, displacement and other indicators of the mechanical data are uniformly converted to a standardized numerical range to eliminate the influence of dimensional differences.

[0020] A feature-layer fusion strategy is employed to fuse preprocessed electrical and mechanical data. Specifically, features such as harmonic distortion rate and insulation resistance change rate are first extracted from the preprocessed electrical data, while features such as tension fluctuation coefficient, vibration dominant frequency, and wear accumulation are extracted from the mechanical data to construct a high-dimensional feature matrix. Principal component analysis is then used to reduce the dimensionality of the high-dimensional features, preserving key information while reducing computational complexity. Subsequently, a weighted fusion algorithm is used to assign dynamic weights to the electrical and mechanical features (adaptively adjusting the weight ratio based on the importance of the two types of data under different operating scenarios), and fusion is performed based on these dynamic weights to obtain the fused feature data.

[0021] The fused feature data is input into a preset catenary condition judgment model to obtain the initial health status.

[0022] The overhead contact line condition assessment model adopts a multi-level deep fusion network architecture, which is divided into three layers. The first layer is the feature receiving layer, which receives the electrical-mechanical fusion features after dimensionality reduction by principal component analysis and dynamic weighted fusion, and completes feature dimension mapping and shallow feature secondary mining. The second layer is the deep feature mining layer, which sets up multiple hidden layers and mines the correlation and coupling relationship between multi-source data according to the nonlinear activation function to capture the implicit correlation patterns of electrical degradation, mechanical deformation, and component wear. The third layer is the condition classification and regression output layer, which, combined with the health level classification standard, completes the initial health status determination.

[0023] The input to the overhead contact line condition assessment model is a multi-dimensional fusion feature that has undergone preprocessing, denoising and completion, normalization, and feature layer fusion. This includes two main categories: electrical and mechanical features. The electrical dimension includes harmonic distortion rate, insulation resistance change rate, arc characteristics, and voltage and current fluctuation characteristics. The mechanical dimension includes quantitative indicators after dimensionality reduction and fusion, such as tension fluctuation coefficient, vibration dominant frequency, structural displacement and deformation characteristics, and cumulative wear of the contact wire. The initial health status of the overhead contact line output by the model does not consider external interference. Before training, the model divides the fully labeled samples into a training set (80%) and a test set (20%). The training set is used for iterative learning of model weights, and the test set is used for generalization capability verification and performance evaluation. Core training hyperparameters are uniformly configured: the initial learning rate is set to 0.001, and a dynamic learning rate decay strategy is used to prevent oscillations during later training; the batch size is set to 32, the number of iterations is 150, and an early stopping mechanism is added to avoid overfitting; the Adam optimizer is used, and the loss function is a combination of mean squared error and classification cross-entropy, suitable for both health status regression and hierarchical assessment tasks. At the same time, a regularization coefficient is introduced to suppress model overfitting and retain key effective features.

[0024] The training process for the overhead contact line condition assessment model is as follows: First, the labeled training set fused feature data is loaded and input into the model to complete forward propagation and calculate the predicted state value. The error between the predicted value and the actual labeled value is calculated using a loss function, and backpropagation is performed to update the network weights and dynamic fusion weight coefficients layer by layer. The model is trained iteratively, and the results are validated using a test set after each round to continuously optimize the model parameters. After training, the optimized overhead contact line condition assessment model achieves a comprehensive discrimination accuracy of over 96.8% on the test set. This effectively adapts to the complex operating environment of electrified railways, ensuring the accuracy and stability of the initial health status assessment and providing a reliable foundation for subsequent environmental correction, fatigue assessment, and graded early warning.

[0025] This embodiment obtains the comprehensive environmental force vector based on wind speed, wind direction, and icing thickness data, and corrects the initial health state to obtain first corrected health state information. Based on dust concentration, humidity, temperature, and corrosive gas concentration data, it determines the comprehensive corrosion-abrasion dose and corrects the first corrected health state information to obtain second corrected health state information. It monitors the temperature data change trajectory and identifies whether it crosses a phase transition critical point, which includes freezing point and dew point temperatures. If a phase transition critical point is crossed, it calculates the thermal stress of the temperature gradient based on temperature data from multiple monitoring points along the path, and corrects the mechanical relaxation and tension components in the second corrected health state information based on this thermal stress to obtain the final health state. The environmental data includes temperature, wind speed, wind direction, icing thickness, dust concentration, humidity, and corrosive gas concentration data.

[0026] S103: The health status of the overhead contact line is judged based on the final health status, and the judgment result is obtained; the judgment result includes normal status, attention status and abnormal status.

[0027] In this embodiment, multi-dimensional features are extracted from the final health state to obtain electrical health feature vectors, environmental health feature vectors, and mechanical health feature vectors. The electrical health feature vectors are matched with the electrical feature baseline distribution under historical normal conditions to calculate the electrical deviation. The environmental health feature vectors are input into a preset environment-equipment coupling influence model to obtain the environmental stress coefficient. The mechanical health feature vectors are compared with a preset standard health manifold to calculate the mechanical manifold deviation distance. The electrical deviation, the environmental stress coefficient, and the mechanical manifold deviation distance are input into a multi-source heterogeneous feature fusion discriminator to output a comprehensive health state discrimination index. Based on the comprehensive health state discrimination index and the historical health state evolution trajectory, the relative position and deviation rate of the current health state on the evolution trajectory are calculated. Based on the relative position and the deviation rate, the health state of the contact network is determined to obtain a judgment result, which includes an abnormal state, a state of attention, and a normal state.

[0028] S104: If the judgment result is in an abnormal state, an alarm message is generated; if the judgment result is in an alert state, a first warning message is generated.

[0029] In this embodiment, when the overall health assessment result of the overhead contact system is abnormal, it indicates a risk of power outages, component damage, or even traffic safety accidents. At this time, a high-priority alarm message is generated, identifying and marking the abnormal section, fault type, out-of-range parameters, and risk level. This provides maintenance personnel with fault location information and compels emergency investigation and repair work. When the overhead contact system is assessed as being in a "watchful" state, it indicates a deterioration trend in the overall health of the system, with potential defects, performance degradation, or escalating environmental stress in some components—early-stage hidden dangers. For this type of medium-risk condition, a first warning message is generated specifically, and the parameter fluctuation patterns, deterioration characteristics, and influencing factors are recorded.

[0030] S105: If the judgment result is in a normal state, then perform state trend prediction on the final health state to obtain the preset predicted health state information for the next cycle; determine whether the predicted health state information is in a state of attention or an abnormal state. If the predicted health state information is in a state of attention, then generate a second warning message; if the predicted health state information is in an abnormal state, then generate a third warning message.

[0031] In this embodiment, if the health status of the overhead contact system is determined to be normal, it means that the overall operation of the contact system is stable, with no immediate faults or obvious safety hazards. However, the contact system is exposed to a complex outdoor environment and is affected by factors such as load fluctuations, continuous environmental erosion, and material fatigue aging. Its health status will gradually deteriorate over time, and potential risks cannot be predicted by relying solely on real-time static assessment.

[0032] Therefore, based on historical multi-source monitoring data of the overhead contact system, continuous time-series health status sequences, and time-series prediction algorithms, a status trend extrapolation is conducted. Using the final health status as a benchmark, a time-series prediction model is constructed based on influencing factors such as the variation patterns of electrical parameters, the fatigue evolution characteristics of mechanical components, and environmental effects. This model dynamically extrapolates the operational performance of the overhead contact system in the next monitoring cycle and outputs predicted health status information. This time-series prediction model integrates historical operating patterns, environmental change cycles, and the aging characteristics of the equipment itself, and can objectively simulate the natural decline process and fluctuation trend of the overhead contact system's health. Quantitative predictions of the future cycle's health level are then provided.

[0033] The time-series prediction model takes the final health status as input and outputs the predicted health status for the next cycle. This model employs a lightweight time-series deep learning architecture with a three-layer structure. The first layer is the time-series embedding layer, which encodes multivariate time-series inputs and maps time-series features to uncover short-term fluctuation patterns. The second layer is the time-series feature extraction layer, incorporating time-series convolution and recurrent memory units to capture the time-series correlation features of the aging trend of the overhead contact system's health status, the periodic impact of the environment, and the slow deterioration of mechanical fatigue. The third layer is the trend regression output layer, which completes the fitting of health status scores and level mapping, achieving accurate prediction of future states for a single cycle. The dataset for the time-series prediction model is constructed based on accumulated overhead contact system time-series monitoring sample datasets, divided into an 80% training set and a 20% test set. The training set is used for learning time-series patterns, and the test set verifies the model's generalization and prediction capabilities. Core training parameter configuration: The initial learning rate is set to 0.0008, and a step-wise learning rate decay is used; the batch size is set to 16, the maximum number of iterations is 120, and an early stopping mechanism is configured to suppress overfitting; the AdamW optimizer is selected, and the loss function is the mean absolute error (MAE), which is suitable for time-series continuous value prediction tasks; at the same time, the time-series sliding window length is set, and historical data of a fixed period is used as the input sequence to ensure time-series correlation.

[0034] The training phase of the time-series prediction model is based on historical continuous health status time-series segments. Samples are input into the model through a sliding window, and forward propagation calculates the predicted health value. The prediction bias is calculated using a loss function, and backpropagation iteratively updates the network weights, repeatedly learning the temporal evolution patterns of equipment natural aging, environmental erosion, and load fluctuations. After each iteration, a test set is used for time-series simulation verification, and the optimal model weights are selected and saved. The model training convergence is good, with a health status trend prediction accuracy of up to 97.2% on the test set and small score prediction errors. It can accurately identify the slow deterioration trend of the overhead contact system, predict the next cycle's attention and abnormal states in advance, and provide reliable model support for the accurate triggering of secondary and tertiary early warnings, enabling proactive prediction and preventative maintenance of overhead contact system hazards.

[0035] In this embodiment, if the predicted health status information indicates that the health status of the overhead contact system will gradually fall into the attention range in the next cycle, it means that the overall performance of the overhead contact system will continue to decline, and electrical parameters, mechanical stress, or environmental impact indicators will gradually deviate from the optimal operating range, showing a trend of slow accumulation and gradual deterioration of hidden dangers. A second early warning message is then generated to prompt maintenance personnel to pay attention to the operational changes of this section of the overhead contact system in advance and increase the frequency of inspections.

[0036] If the predicted health status information indicates that the future health status will enter an abnormal state, it means that the degradation process of the overhead contact line equipment is accelerating, and serious problems such as component damage, unstable power supply, and structural deformation will subsequently occur. At this time, a third early warning message is generated, clearly indicating the risk level, the expected degradation time, and the main influencing factors. The third early warning message is used to remind the operation and maintenance department to formulate maintenance plans in advance and carry out preventive maintenance and rectification.

[0037] S106: After identifying the alarm information, first warning information, second warning information or third warning information based on different preset levels, output them to the operation and maintenance monitoring platform, mobile terminal or audible and visual alarm device.

[0038] In this embodiment, the relative position is used as the normalized arc length distance between the projection point of the current health status comprehensive discrimination index on the historical health status evolution trajectory and the trajectory starting point; the deviation rate is used as the ratio between the actual descent slope of the current health status comprehensive discrimination index and the standard descent slope of the corresponding projection point on the historical health status evolution trajectory; and the orthogonal distance between the relative position of the current health status on the evolution trajectory and the corresponding position on the historical health status evolution trajectory is calculated to obtain the trajectory deviation degree; when the normalized arc length distance is less than a preset first distance threshold, the ratio is less than a preset first rate threshold, and the trajectory deviation degree is less than a preset deviation threshold, it is determined to be a normal state; when the normalized arc length distance is greater than or equal to the first distance threshold, or the ratio is greater than or equal to the first rate threshold, or the trajectory deviation degree is greater than or equal to the deviation threshold, it is determined to be a state of attention or an abnormal state, wherein the state of attention and the abnormal state are further distinguished based on whether the ratio is greater than a preset second rate threshold or whether the trajectory deviation degree is greater than a preset severe trajectory deviation threshold.

[0039] As can be seen from the above, this application, through the acquisition of comprehensive electrical, mechanical, and environmental data of the overhead contact system by sensors, breaks through the limitations of fragmented data in traditional monitoring, achieving a comprehensive perception of the equipment's operating status. Based on data preprocessing and fusion technologies and dynamic correction of the initial health state by environmental factors, the accuracy of health assessment is improved, effectively avoiding the one-sidedness of single-data assessments and misjudgments caused by environmental interference. Furthermore, by dividing the status into three levels—normal, alert, and abnormal—a two-layer early warning mechanism is constructed, combining current status response and future trend prediction. This mechanism not only rapidly generates alarm information for immediate anomalies but also predicts trends in normal states, identifying potential hazards in advance and issuing graded warnings, effectively improving the initiative of operation and maintenance. Alarm and early warning information of different levels is accurately pushed to multiple terminals, optimizing the allocation of operation and maintenance resources, reducing ineffective inspections, and lowering operation and maintenance costs.

[0040] In one embodiment of this application, the initial health state is corrected based on environmental data to obtain the final health state, including: Environmental data includes temperature, wind speed, wind direction, ice thickness, dust concentration, humidity, and corrosive gas concentration. Based on wind speed data, wind direction data, and ice thickness data, the comprehensive environmental force vector is obtained, and the initial health state is corrected to obtain the first corrected health state information. Based on dust concentration data, humidity data, temperature data, and corrosive gas concentration data, the comprehensive corrosion-abrasion dose is determined, and the first corrected health status information is corrected to obtain the second corrected health status information. The temperature data change trajectory is monitored, and it is identified whether the phase transition critical point is crossed. The phase transition critical point includes the freezing point and dew point temperature. If there is a critical point that crosses the phase transition, the thermal stress of the temperature gradient is calculated based on the temperature data of multiple monitoring points along the line, and the mechanical relaxation and tension components in the second corrected health state information are corrected based on the thermal stress to obtain the final health state.

[0041] In this embodiment, the initial health status is corrected based on environmental data, which includes temperature, wind speed, wind direction, ice thickness, dust concentration, humidity, and corrosive gas concentration data.

[0042] In this embodiment, firstly, based on wind speed data, wind direction data, and icing thickness data, an environmental comprehensive force vector is obtained, and the initial health state is corrected to obtain first corrected health state information. Specifically, based on wind speed and wind direction data, an environmental wind vector is determined; the icing thickness data is converted into an additional mass load on the contact wire to obtain an icing load; the force components of the environmental wind vector and the icing load in the three spatial dimensions of the contact wire are calculated and vector-superimposed to obtain an environmental comprehensive force vector; the magnitude of the environmental comprehensive force vector is used as the correction intensity, and the angle between the direction of the environmental comprehensive force vector and the designed bearing direction of the contact wire is used as the direction correction coefficient; the product of the correction intensity and the direction correction coefficient is used exponentially to correct the initial health state to obtain first corrected health state information.

[0043] Secondly, based on the mechanical correction, a secondary correction is further carried out according to environmental factors such as atmospheric corrosion and abrasion. Specifically, the product of humidity data and corrosive gas concentration data is used as the basic corrosion rate, and temperature data is used as the reaction rate acceleration factor. The Arrhenius formula is used for correction to obtain the target corrosion rate. The target corrosion rate is then integrated over a preset N consecutive time windows to obtain the cumulative corrosive damage. Dust concentration data is used as the abrasion acceleration factor. Based on the abrasion acceleration factor and the cumulative corrosive damage, the comprehensive corrosion-abrasion dose is determined. The comprehensive corrosion-abrasion dose is then used as a continuous cumulative function to correct the first corrected health status information to obtain the second corrected health status information.

[0044] Subsequently, the trajectory of temperature data changes is monitored, and it is identified whether the temperature data crosses critical phase transition points such as freezing point and dew point. For example, when the ambient temperature changes abruptly and exceeds the critical temperature range, phenomena such as air condensation, icing, and material contraction and expansion will occur in a concentrated manner, causing sudden changes in stress within the contact network. At this time, the thermal stress of the temperature gradient is calculated based on the temperature data of multiple monitoring points along the line, and the mechanical relaxation and tension components in the second corrected health state information are corrected based on the thermal stress to obtain the final health state. Specifically, when the absolute value of the temperature gradient exceeds the preset temperature threshold range, the temperature difference between each monitoring point and the preset reference point is calculated, and the thermal expansion strain is calculated based on the linear expansion coefficient of the contact wire; the thermal expansion strain is converted into thermal stress and superimposed on the mechanical relaxation and tension components to be corrected; when the absolute value of the temperature gradient is less than or equal to the temperature threshold, the second corrected health state information is taken as the final health state.

[0045] As can be seen from the above, this embodiment integrates multiple environmental data such as temperature, wind conditions, icing, dust, humidity, and corrosive gases to calibrate the health status of the overhead contact system from three dimensions: mechanical external force, corrosion and abrasion, and temperature stress. The first correction is achieved by constructing a combined wind and ice force vector; the second correction is achieved by quantifying the cumulative damage from corrosion and abrasion; and finally, based on phase transition critical conditions such as freezing point and dew point, thermal stress is calculated through the line temperature gradient to correct the key mechanical components of the contact system sag and tension. This multi-dimensional, progressive environmental coupling correction method effectively reduces the assessment bias caused by the external environment and significantly improves the authenticity and accuracy of the contact system health assessment.

[0046] In one embodiment of this application, based on wind speed data, wind direction data, and icing thickness data, an environmental combined force vector is obtained, and the initial health state is corrected to obtain first corrected health state information, including: Determine the environmental wind vector based on wind speed and wind direction data; The icing thickness data is converted into an additional mass load on the overhead contact line to obtain the icing load; The force components of the environmental wind vector and the icing load on the three spatial dimensions of the catenary are calculated and superimposed to obtain the comprehensive environmental force vector. The magnitude of the combined environmental force vector is used as the correction strength, and the angle between the direction of the combined environmental force vector and the design bearing direction of the contact wire is used as the direction correction coefficient. The product of the correction intensity and the direction correction coefficient is used to correct the initial health state in an exponential form to obtain the first corrected health state information.

[0047] In this embodiment, firstly, a spatial reference coordinate system is established based on the overhead contact line route. Using due north and the line axis as reference standards, the wind direction angle is converted into a spatial azimuth angle. Combined with the magnitude of wind speed data, the magnitude and direction of wind force are defined, thus synthesizing the environmental wind vector. Using icing thickness as the sole variable, a correlation calculation is performed using a standardized ice density constant uniformly defined in the industry. For example, the icing load per unit area is calculated by multiplying the icing thickness by the ice density. Then, according to the average attachment rule across the entire line, the surface load is equivalently converted into an additional mass load per unit length. The average attachment rule is formulated based on historical operating data and field monitoring experience, and is obtained by statistically analyzing icing under various operating conditions based on accumulated icing observation records and historical environmental monitoring sample data from different climatic zones.

[0048] Secondly, using a three-dimensional spatial coordinate system as a reference, the force components of the environmental wind vector and the icing load in the axial, lateral, and vertical dimensions of the contact network are decomposed. Through vector decomposition and superposition calculation, the synergistic effect of the wind dynamic load and the icing static load is integrated to obtain the environmental comprehensive force vector, which represents the composite external force on the contact network under the coupling effect of wind and ice.

[0049] Finally, the magnitude of the environmental combined force vector is extracted as the correction intensity for the external force influence. Based on the angle between the actual direction of the environmental combined force vector and the pre-designed bearing direction of the contact network, a direction correction coefficient is determined. This coefficient represents the additional loss caused by an unreasonable force angle. Multiplying the correction intensity by the direction correction coefficient yields the environmental external force combined influence factor. An exponential correction algorithm is then used to attenuate and correct the initial health state, resulting in first corrected health state information that closely matches the actual stress conditions. The pre-designed bearing direction is set according to the overall structural layout of the contact network and the line erection standards. Specifically, the contact network line axis is used as the base axis. Based on the fixed installation methods of the cantilever arms, positioners, contact wires, and catenary cables, the structural stress characteristics, and engineering design parameters, standard lateral, vertical, and axial force bearing directions are defined. The optimal direction of action for the equipment to withstand external forces under normal conditions is determined according to the contact network structural mechanics design requirements.

[0050] The exponential correction algorithm takes the comprehensive environmental force influence factor as input, and the catenary structure tolerance threshold and environmental force safety benchmark value as auxiliary inputs. The algorithm outputs a health status correction coefficient, used to scale and adjust the initial health score, achieving a quantitative mapping of external force risk. The larger the input comprehensive environmental force influence factor value, the stronger the combined wind and ice stress, and the greater the attenuation of the correction coefficient. The algorithm uses a negative exponential function to construct the computational structure, with parameters pre-calibrated based on catenary design specifications and historical operating data. Specifically, a fixed base value is set based on the material mechanical attenuation law, the attenuation rate parameter is configured differently according to the catenary material and span structure of different sections, and the external force critical threshold parameter is uniformly defined in conjunction with the line's ultimate bearing capacity. The negative exponential function is: λ=e -α·x x is the comprehensive external force influence factor, α is the preset attenuation coefficient (fixed calibration parameter), and λ is the health status correction coefficient.

[0051] As can be seen from the above, this embodiment constructs an environmental wind vector based on wind speed, wind direction, and icing thickness data, calculates the additional mass load due to icing, and completes the component decomposition and vector superposition of the two types of external forces in three-dimensional space to synthesize a comprehensive environmental force vector. Simultaneously, based on the dual indicators of force magnitude and force angle, and combined with an exponential correction method, the initial health state is quantitatively corrected. This objectively represents the mechanical damage to the contact network structure caused by the coupled loads of strong winds and icing, effectively reducing the assessment error caused by complex meteorological environments and further improving the comprehensiveness and accuracy of the contact network health status assessment.

[0052] In one embodiment of this application, a comprehensive corrosion-erosion dose is determined based on dust concentration data, temperature data, humidity data, and corrosive gas concentration data, and the first corrected health status information is corrected to obtain second corrected health status information, including: The product of humidity data and corrosive gas concentration data is used as the base corrosion rate, and temperature data is used as the reaction rate acceleration factor. The Arrhenius formula is used for correction to obtain the target corrosion rate. The cumulative corrosive damage is obtained by integrating the target corrosion rate over N consecutive time windows. Dust concentration data is used as an erosion accelerator. Based on the erosion accelerator and the cumulative corrosive damage, the comprehensive corrosion-erosion dose is determined. The first corrected health status information is corrected by applying the comprehensive corrosion-wearing dose as a continuous cumulative function to obtain the second corrected health status information.

[0053] In this embodiment, firstly, humidity data and corrosive gas concentration data are used as basic parameters. Multiplying these two data yields the basic corrosion rate, which represents the natural corrosion rate of the contact wire metal components under the combined effects of a humid environment and corrosive media. Temperature alters the activity of chemical reactions; therefore, ambient temperature is used as an accelerating factor for the reaction rate. The basic corrosion rate is dynamically corrected according to the variation law of the Arrhenius equation, simulating the actual mechanism by which temperature increases accelerate metal oxidation, rusting, and chemical corrosion. Finally, a target corrosion rate that closely matches the on-site environmental conditions is obtained. The Arrhenius equation is k(T) = k0·exp(-E a / RT), k(T) is the target corrosion rate (varying with temperature), k0 is the basic corrosion rate, E a R is the activation energy of the corrosion reaction (an inherent value of the material, a fixed constant determined in advance through testing based on the material characteristics of the contact wire, steel structure, and alloy connectors), T is the ambient thermodynamic temperature.

[0054] Secondly, to reflect the long-term and cumulative nature of corrosion damage, N consecutive equal time windows are selected to continuously collect the real-time target corrosion rate at each time period; by performing continuous time-domain integration on the target corrosion rate within multiple time windows, the cumulative corrosion damage amount that accumulates over time is obtained.

[0055] The preset time window is a standardized base time window with a fixed duration, defined uniformly based on the changing patterns of corrosion in the contact network environment, and set in minutes or hours. Taking into account regional climate fluctuation characteristics, sensor data acquisition cycles, and the rate of change in corrosion reactions, a fixed duration adapted to the fluctuation patterns of environmental parameters is selected as a single time window. For example, the base time granularity is defined based on the inherent acquisition cycle of the sensors, with a fixed data upload interval as the smallest unit, ensuring that a single time window includes the entire set of valid monitoring data. Furthermore, the window duration is matched according to the differences in regional climate fluctuation characteristics. For sections with frequent monsoons, diurnal temperature variations, and alternating sunny and cloudy conditions, a first duration window is selected, capturing short-term fluctuations in temperature, humidity, and corrosive gases. For sections with stable climate and gradual changes in environmental parameters, a second duration window is selected. The second duration window is longer than the first duration window, used to reduce unnecessary duplicate calculations. Finally, the final calibration was completed by aligning with the rate of change in metal corrosion reactions. Since electrochemical corrosion and atmospheric oxidation are slow, gradual reactions with no instantaneous changes in parameters, a fixed timeframe was selected based on the hysteresis characteristics of the corrosion physicochemical response, balancing sensitivity to environmental fluctuations with computational efficiency. This ensures the corrosion state within a single time window is continuous and representative. N represents the number of time windows for continuous statistics, determined differently based on the intensity of environmental corrosion and the required assessment accuracy. Specifically, a moderate value was used for ordinary areas with stable atmospheric composition to balance computational load and assessment effectiveness; for special areas such as chemical industrial zones and coastal areas, the value of N was appropriately increased to extend the statistical period and fully accumulate corrosion damage data. Simultaneously, simulation verification was performed using historical corrosion test data to determine the optimal value range.

[0056] Finally, based on the cumulative corrosion damage, atmospheric dust concentration is used as the erosion accelerator. Dust particles continuously cause friction and wear on components such as contact lines and connectors, further accelerating equipment aging and damage. Therefore, a comprehensive corrosion-erosion dose is obtained by coupling the erosion accelerator with the cumulative corrosion damage (integrating the dual destructive effects of chemical corrosion and physical wear). Then, a correction method using a continuous cumulative function is adopted to continuously apply the superimposed losses of corrosion and erosion to the first corrected health status information, completing the second correction, and the second corrected health status information.

[0057] As can be seen from the above, this embodiment constructs a basic corrosion rate by coupling humidity data and corrosive gas concentration data, and completes rate correction based on the temperature-dependent Arrhenius mechanism, thus improving the accuracy of the accelerating effect of reduction temperature on chemical corrosion reactions. Furthermore, by calculating the dual destructive effects of cumulative corrosion damage and physical wear through multi-time-window integration, a comprehensive corrosion-wearing dose is obtained. Then, the health status is iteratively corrected using a continuous cumulative function, achieving a dynamic quantitative assessment of environmental aging damage. This effectively compensates for the shortcomings of evaluating a single environmental parameter and significantly improves the scientific rigor and accuracy of secondary correction of the health status.

[0058] In one embodiment of this application, a status analysis and early warning method based on online monitoring data of overhead contact lines further includes: Obtain the timestamps of tension or acceleration response signals from at least three adjacent monitoring points along the contact wire, and calculate the apparent velocity vector of the mechanical wave or stress wave along the contact wire; When the section where two adjacent monitoring points are located is determined to be in a state of attention, and the magnitude of the combined environmental force vector is greater than the preset interpolation trigger threshold, the mechanical relaxation and tension component data of at least one virtual node between the two monitoring points are reconstructed by using the difference scheme of the one-dimensional unsteady heat conduction equation based on the real-time temperature data and tension data of the two monitoring points. The mechanical sag and tension component data of the virtual nodes are used as additional health status monitoring points and fused with the mechanical health feature vector.

[0059] In this embodiment, firstly, at least three adjacent monitoring points are evenly distributed along the entire contact wire line. The tension signal and structural acceleration response signal of each point are simultaneously collected, and the timestamps corresponding to each set of signals are recorded. Based on the temporal differences of the signals from different monitoring points, the propagation laws of mechanical waves and stress waves in the online contact wire medium are integrated. Through correlation calculations between propagation time and point spacing, the propagation rates of mechanical waves and stress waves are obtained. The propagation direction is determined based on the order of propagation of mechanical waves and stress waves. A coordinate system is established with the longitudinal direction of the contact wire line as the reference. The propagation rates and spatial directions of the corresponding waves are integrated to obtain the corresponding apparent velocity vector. The apparent velocity vector provides dynamic basic parameters for judging the overall structural vibration and stress transmission state of the contact wire.

[0060] Secondly, the operational status of the line sections between adjacent monitoring points is determined in real time. If the section where the adjacent monitoring point is located is determined to be in a state of alert, and the magnitude of the current environmental combined force vector exceeds the preset interpolation trigger threshold, a high-precision data reconstruction operation is initiated. Specifically, the high-precision data reconstruction operation uses the temperature data and wire tension signal of the two monitoring points as boundary conditions. Based on the one-dimensional unsteady-state heat conduction equation, iterative calculations are performed using a finite difference scheme to deduce and reconstruct the mechanical relaxation and tension component values ​​of several virtual nodes between the two monitoring points. This high-precision data reconstruction operation is used to fill the data gaps caused by insufficient monitoring points in long intervals. Among them, the one-dimensional unsteady-state heat conduction equation is the fundamental physical equation describing the dynamic conduction and change of temperature over time along a single direction of the contact wire. It can characterize the heat transfer law of the conductor under alternating environmental temperatures. Combined with differential iterative solutions, the reconstruction of the temperature and mechanical parameters of virtual nodes in the interval can be realized. Its standard mathematical model is as follows: B represents the contact wire temperature, t represents time, x represents the longitudinal coordinate along the contact wire (one-dimensional), and ρ, c, and λ represent the thermal conductivity, density, and specific heat capacity of the conductor (inherent material parameters). The preset interpolation trigger threshold is determined based on the statistical results of the comprehensive environmental force vector under normal operating conditions of the contact wire section. Specifically, the mean value of the comprehensive environmental force vector and the upper limit of the normal fluctuation range are used as the basic benchmark value. Then, combined with the contact wire structure bearing capacity margin, the mechanical tolerance limit of the warning section, and the critical value of the wind and ice coupled external force, a safety correction margin is added to complete the numerical calibration. At the same time, it is adjusted according to the different climatic conditions, wire materials, and span structure differences of different sections, and finally fixed to form an independently configurable interpolation trigger threshold.

[0061] Finally, the key mechanical data of virtual node sag and tension are incorporated into the data processing system as additional health status monitoring points. The virtual node monitoring data is deeply fused and linked with the mechanical health feature vectors extracted from actual measurement points to enrich the dimensions of the catenary's overall status perception and compensate for the shortcomings of limited physical monitoring point deployment and insufficient interval coverage.

[0062] As can be seen from the above, this embodiment effectively perceives the structural force propagation characteristics by collecting the tension, acceleration signals, and timestamps of multiple sets of adjacent monitoring points and calculating the apparent velocity vectors of the contact wire mechanical wave and stress wave. For line sections determined to be in a state of attention and where the magnitude of the combined environmental force vector exceeds the interpolation trigger threshold, the sag and tension key data of the virtual nodes in the section are reconstructed based on the measured temperature and tension data at both ends, combined with the differential iterative algorithm of the one-dimensional unsteady heat conduction equation. This supplements the coverage blind spots of physical monitoring points, and the fusion analysis of virtual node data and mechanical health feature vectors effectively improves the continuity, precision, and accuracy of the structural status assessment of long sections of the contact wire.

[0063] In one embodiment of this application, before calculating the thermal stress of the temperature gradient based on temperature data from multiple monitoring points along the line, and correcting the mechanical relaxation and tension components in the second corrected health state information based on the thermal stress to obtain the final health state, the method further includes: The temperature gradient along the overhead contact line is calculated based on the temperature data differences from multiple monitoring points along the line and the distance between the monitoring points. Based on the temperature gradient, the linear expansion coefficient of the contact wire, the elastic modulus, and the original tension state, the additional thermal stress caused by uneven temperature distribution is calculated using thermoelastic mechanics formulas; where the original tension state is the design rated tension value of the contact wire. The second corrected health state information is corrected based on the additional thermal stress, and the components related to mechanical relaxation and tension are extracted from the second corrected health state information as the reference value of the mechanical state to be corrected.

[0064] In this embodiment, firstly, temperature data is synchronously collected from multiple monitoring nodes deployed along the overhead contact line to obtain the temperature difference between adjacent monitoring points and the actual spacing between them. Through quantitative calculation of the temperature difference and interval distance, and by solving segment by segment, the temperature gradient is obtained. The temperature gradient represents the quantified unevenness of the temperature distribution along the entire overhead contact line, characterizing the differences in the heating environment of different sections, providing accurate basic temperature parameters for subsequent thermal stress calculations, and also representing the temperature variation characteristics of the line caused by external temperature fluctuations and local environmental differences. Specifically, in the process of solving the overhead contact line temperature gradient, the quantitative calculation uses the measured temperature difference between adjacent monitoring points and the distance between monitoring points as known quantifiable parameters, substituting them into a fixed mathematical formula to perform standardized numerical calculations, transforming the macroscopic phenomenon of uneven temperature distribution into a quantifiable temperature gradient index. The fixed mathematical formula is ▽M=ΔL / ΔM, where ▽M is the longitudinal temperature gradient of the line, ΔM is the temperature difference between two adjacent monitoring points, and ΔL is the distance between adjacent monitoring points.

[0065] Secondly, using the temperature gradient as an input parameter and based on the inherent linear expansion coefficient and elastic modulus of the contact wire, the rated tension value of the contact network design is introduced as the original tension reference state. Coupled calculations are carried out according to the classical thermoelasticity theory formula to quantitatively deduce the conductor expansion and contraction deformation and internal force changes caused by uneven temperature distribution, accurately solve for the additional thermal stress generated by non-uniform temperature change, and clarify the additional thermal stress on the internal force of the contact wire caused by temperature difference.

[0066] Finally, the additional thermal stress is used as a correction factor to perform a secondary correction on the second-corrected health state information that has undergone corrosion and wear correction, eliminating the evaluation bias caused by uneven temperature distribution. At the same time, from the updated health state information, the two core components of mechanical relaxation and tension are selectively screened and extracted, and uniformly calibrated as the mechanical state benchmark values ​​to be corrected, providing standardized mechanical benchmark data for subsequent state assessment, section parameter reconstruction, and health feature fusion.

[0067] As can be seen from the above, this embodiment calculates the temperature gradient based on the temperature difference at multiple points along the line and the distance between measuring points. It then uses thermoelasticity formulas to quantitatively calculate the additional thermal stress caused by uneven temperature distribution, based on the contact wire's linear expansion coefficient, elastic modulus, and rated original tension. Using this additional thermal stress as a correction basis, the second corrected health state information is optimized and adjusted to compensate for structural internal forces and deformation deviations caused by the non-uniform temperature field. Simultaneously, key components of mechanical relaxation and tension are extracted as mechanical state benchmark values, achieving quantitative control of the influence of temperature stress. This effectively balances the thermal expansion characteristics of materials with the line design conditions, reduces evaluation errors caused by temperature differences, and enhances the scientific rigor of the contact wire mechanical state analysis.

[0068] In one embodiment of this application, component data related to mechanical sag and tension are extracted from the second corrected health status information as a reference value for the mechanical status to be corrected, including: The second corrected health status information is vector decomposed to separate the tension component along the contact line axis, the lateral sag component perpendicular to the contact line direction, and the vertical lift component. The tension component is mapped to the contact wire design tension reference range, and the tension deviation rate is obtained by comparing the current tension value with the rated tension value. The lateral sag component and the vertical uplift component are geometrically synthesized to obtain the comprehensive sag vector, and the sag ratio is calculated based on the span length as the sag reference value. Anomalies in the tension deviation rate and sag reference value are eliminated and smoothed using a sliding time window to obtain the tension reference value and sag reference value. The tension reference value and the sag reference value are used as the mechanical state reference values.

[0069] In this embodiment, firstly, based on the second corrected health state information, the decoupling and separation of multiple coupled state quantities are achieved through professional vector decomposition techniques. Specifically, the second corrected health state information is decomposed according to the spatial force and deformation characteristics of the contact line to obtain the tension component along the contact line axis, the lateral sag component perpendicular to the contact line direction, and the vertical lift component.

[0070] Secondly, the axial tension component is benchmarked against the pre-set design tension reference range of the contact wire, and the difference between the real-time tension and the rated design tension is compared to quantitatively calculate the tension deviation rate. The tension deviation rate indicates the degree to which the conductor's stress deviates from the standard operating conditions. Simultaneously, the lateral and vertical deformation components are integrated, and a comprehensive sag vector is obtained through geometric vector synthesis. This vector is then converted into a sag ratio index based on the line span length, and the sag ratio index is used as the sag reference value. The sag reference value is a unified quantitative benchmark index for the degree of contact wire slack, abnormal deformation, and excessive lateral and vertical offset in the contact wire, used to determine whether the conductor deformation is within the allowable range specified in the standards.

[0071] Finally, anomaly removal and smoothing filtering are performed on the tension deviation rate and sag reference values ​​using a sliding time window algorithm. This filters out random errors caused by environmental disturbances, instantaneous impacts, and sensor noise, reducing data fluctuation interference. After noise reduction optimization, stable and reliable tension and sag reference values ​​are output, which serve as unified mechanical state reference values.

[0072] In this embodiment, the input to the sliding time window algorithm is the original sequence data of tension deviation rate, lateral sag component, vertical uplift component, and real-time sag ratio under continuous time sequence. The algorithm is divided into multiple execution layers: data loading, window sliding traversal, anomaly detection and removal, smoothing filtering calculation, and result output. First, a local time sequence data segment is truncated with a fixed window length, and then anomaly detection and removal are carried out segment by segment. Subsequently, the data is corrected by mean filtering or weighted smoothing, eliminating random noise and abrupt interference step by step, and finally outputting a regular and continuous steady-state index sequence. The parameters in the algorithm include sliding window duration, window sliding step size, anomaly threshold coefficient, and smoothing weighting coefficient. The parameters are uniformly set according to the fluctuation characteristics of the catenary environment and the sampling frequency: the window duration is calibrated according to the deformation and tension change cycle, the sliding step size matches the data acquisition interval, the anomaly threshold is defined according to the normal fluctuation range of the line, and the smoothing coefficient is adaptively set according to the evaluation accuracy requirements.

[0073] As can be seen from the above, this embodiment effectively decouples multi-dimensional coupled parameters such as axial tension, lateral sag, and vertical lift of the contact line by performing vector decomposition on the second corrected health status information, achieving independent analysis of mechanical and deformation indices. Furthermore, by calculating the tension deviation rate against the design reference interval and solving for the sag ratio through geometric synthesis and span parameters, the standardized quantitative conversion of key mechanical indicators is completed. The use of a sliding time window algorithm for anomaly removal and smoothing filtering effectively suppresses data distortion caused by sensor noise and instantaneous operating condition disturbances, optimizing the stability of time-series data.

[0074] In one embodiment of this application, the thermal stress of the temperature gradient is calculated based on temperature data from multiple monitoring points along the line, and the mechanical relaxation and tension components in the second corrected health state information are corrected based on the thermal stress to obtain the final health state, including: When the absolute value of the temperature gradient exceeds the preset temperature threshold range, the temperature difference between each monitoring point and the preset reference point is calculated, and the thermal expansion strain is calculated based on the linear expansion coefficient of the contact line. The thermal expansion strain is converted into thermal stress, which is then superimposed on the mechanical relaxation and tension components to be corrected. When the absolute value of the temperature gradient is less than or equal to the temperature threshold, the second corrected health status information is taken as the final health status.

[0075] In this embodiment, firstly, a fixed temperature threshold is set using the absolute value of the temperature gradient as a discrimination threshold to determine the different operating conditions. Specifically, when the temperature gradient exceeds the preset temperature threshold range, it indicates that the temperature distribution along the contact wire is significantly different, and non-uniform temperature change will cause uneven expansion and contraction deformation of the conductor. At this time, using a unified reference point as a benchmark, the relative temperature difference of each monitoring point is calculated one by one, and the thermal expansion strain generated in different sections is calculated based on the inherent linear expansion coefficient of the contact wire material and the basic principle of thermal deformation. Specifically, based on the basic principle of material thermal deformation, a unified temperature reference point is first selected, and the actual temperature difference between each contact wire monitoring point and the reference point is measured. Then, based on the inherent linear expansion characteristic parameters of the contact wire material itself, the correlation law between temperature difference and the inherent thermal expansion properties of the material is used to quantitatively calculate the thermal expansion strain generated by temperature differences in each section of the line. The thermal expansion strain characterizes the degree of expansion and contraction deformation of the contact wire caused by uneven temperature distribution. The preset temperature threshold range is determined based on the historical temperature gradient monitoring data of the entire overhead contact line. Specifically, temperature gradient samples from monitoring points along the line are first collected under different seasons, day and night, sunny and cloudy conditions, and extreme weather conditions. The reasonable fluctuation range of the temperature gradient under normal operation is statistically analyzed and used as the basic range. Combining the linear expansion characteristics of the contact wire, the structural constraints of the span, and the allowable limit of thermal stress safety, the critical gradient value that will not induce uneven expansion and contraction deformation and additional thermal stress is determined. Then, combined with the environmental differences of the line section, the conductor material specifications, and the actual operation and maintenance experience, a reasonable buffer margin is reserved, and the upper and lower limit values ​​are uniformly set to form a fixed temperature threshold range. This range is used to distinguish between uniform temperature steady-state operating conditions and temperature difference abnormal operating conditions, thereby determining whether thermal stress correction calculations need to be carried out.

[0076] Secondly, based on the thermoelasticity correlation, the thermal expansion strain is steadily transformed into equivalent thermal stress, which represents the internal additional force generated under the constraint of temperature difference deformation. This thermal stress is then superimposed and fused with the previously extracted mechanical sag and tension components to complete the compensation and correction of mechanical and deformation parameters under temperature disturbances. This reduces the deviation in state assessment caused by non-uniform temperature environments, ensuring that the mechanical state data closely matches the actual stress and deformation conditions. The thermoelasticity correlation is the inherent mechanical constraint law between the thermal expansion strain caused by temperature changes and the internal thermal stress of the component. Specifically, when the contact wire undergoes expansion and contraction deformation due to temperature difference, it is restricted by its own structure, span constraints, and fixed tension at both ends, preventing free deformation and thus generating additional forces within the conductor. This thermoelasticity correlation, based on the elastic deformation characteristics of the material, transforms the temperature-induced deformation strain, and according to the elastic properties of the contact wire itself, into the corresponding equivalent thermal stress.

[0077] Finally, when the absolute value of the temperature gradient does not exceed the temperature threshold range, it indicates that the temperature distribution is uniform throughout the entire line, and no additional thermal strain or thermal stress will be generated. The influence of temperature factors on the mechanical state of the contact network can be ignored.

[0078] As can be seen from the above, this embodiment uses the line temperature gradient as the criterion to quantify the impact of uneven temperature on the contact network in layers, achieving differentiated correction of thermal effects. When the temperature gradient exceeds the limit, the thermal expansion strain is calculated by referring to the reference temperature difference and the thermal expansion characteristics of the contact wire material, and then converted into equivalent thermal stress according to the thermoelastic correlation. This stress is then superimposed and compensated to the mechanical relaxation and tension components, effectively reducing the deformation and internal force deviations caused by the non-uniform temperature field. For sections with uniform temperature distribution and weak thermal effects, the corrected second health state data is used to reduce redundant calculations.

[0079] In one embodiment of this application, after obtaining the final health status, the method further includes: If the rate of change of the contact wire tension exceeds a preset dynamic threshold, a buffer window of a preset time length is triggered. Within the buffer window, the transient response feature envelope of the tension signal is extracted. The transient response feature envelope includes the maximum peak-to-valley difference, the damping attenuation coefficient, and the local stress spectrum obtained based on the rainflow counting method. Based on the comprehensive corrosion-erosion dose and the preset fatigue notch coefficient correction function, the comprehensive corrosion-erosion dose is mapped to the increment of the fatigue notch sensitivity coefficient; The equivalent fatigue damage increment caused by a single train passing event is calculated based on the increment of the fatigue notch sensitivity coefficient, the transient response feature envelope, and the mechanical relaxation and tension components in the final healthy state. The equivalent fatigue damage increment is added to the total historical fatigue damage of the overhead contact line. When the total historical fatigue damage exceeds the preset fatigue life threshold, fatigue life warning information is generated.

[0080] In this embodiment, the catenary tension change rate is used as the dynamic identification criterion. When the tension fluctuation is severe and exceeds a set dynamic threshold, a buffer window of a preset time length is triggered. The set dynamic threshold is determined by statistically analyzing the upper limit of the normal fluctuation range of the tension change rate under normal and stable catenary operation and the fluctuation range of slight disturbances during normal train passage. The buffer window of the preset time length is a fixed-duration data acquisition interval that is activated after a sudden tension change that exceeds the dynamic threshold. It is used to completely capture the entire transient tension signal caused by train passage. This duration is preset based on a combination of the train passage speed, the catenary vibration attenuation period, and the signal sampling frequency.

[0081] Secondly, within the effective data range of the buffer window, the transient response feature envelope of the tension signal is extracted. The transient response feature envelope includes the peak-to-valley difference and the damping attenuation coefficient. The local stress spectrum is then statistically analyzed using the rainflow counting method. This local stress spectrum characterizes the dynamic stress fluctuations, vibration attenuation patterns, and cyclic load characteristics of the contact wire after being impacted by the train. The joint extraction of multi-dimensional transient features can represent the mechanical response under short-term impact loads.

[0082] Finally, based on the comprehensive corrosion-wearing dose and the preset fatigue notch coefficient correction function, the increment of the fatigue notch sensitivity coefficient is obtained. The increment of the fatigue notch sensitivity coefficient represents the weakening effect of environmental aging on the fatigue resistance of the structure. By integrating the increment of the fatigue sensitivity coefficient, transient characteristic envelope, and steady-state mechanical parameters, the equivalent fatigue damage increment caused by a single train passage is quantitatively calculated, and continuously accumulated to form the total historical cumulative damage. By comparing it with the preset fatigue life threshold in real time, a fatigue life warning is output when the total historical cumulative fatigue damage exceeds the preset fatigue life threshold.

[0083] As can be seen from the above, this embodiment, based on obtaining the final health status of the overhead contact system, dynamically triggers a buffer window according to the tension change rate to capture the transient mechanical response signal brought about by train passage. By extracting multi-dimensional transient features such as the tension peak-valley difference, damping attenuation coefficient, and rainflow stress spectrum, and correcting the fatigue notch sensitivity coefficient according to the corrosion and abrasion dosage, the weakening effect of environmental aging on the fatigue resistance of the structure is quantified. Furthermore, the incremental fatigue damage from a single train passage is quantitatively calculated and continuously accumulated, and compared with the fatigue life threshold. This achieves a full-cycle quantitative assessment of dynamic fatigue damage in the overhead contact system, taking into account the coupled effects of environmental degradation and train impact loads, thereby improving the safety and predictability of overhead contact system operation and maintenance.

[0084] In one embodiment of this application, the health status of the overhead contact line is determined based on the final health status, and a determination result is obtained, including: Multidimensional feature extraction is performed on the final health status to obtain electrical health feature vector, environmental health feature vector, and mechanical health feature vector; The electrical health feature vector is matched with the baseline distribution of electrical features under historical normal conditions to calculate the electrical deviation. The environmental health feature vector is input into the preset environment-equipment coupling effect model to obtain the environmental stress coefficient; The mechanical health feature vector is compared with the preset standard health manifold, and the deviation distance of the mechanical manifold is calculated. Electrical deviation, environmental stress coefficient, and mechanical manifold deviation distance are input into a multi-source heterogeneous feature fusion discriminator, which outputs a comprehensive health status discriminant index. Based on the comprehensive health status discrimination index and the historical health status evolution trajectory, the relative position and deviation rate of the current health status on the evolution trajectory are calculated. Based on relative position and deviation rate, the health status of the overhead contact line is determined, and the judgment results are obtained, including abnormal status, warning status and normal status.

[0085] In this embodiment, based on the final health status data of the overhead contact system, multi-dimensional decomposition and feature quantification extraction are carried out. Key indicators of three independent dimensions—electrical, environmental, and mechanical—are separated and screened from the final health status, and corresponding electrical health feature vectors, environmental health feature vectors, and mechanical health feature vectors are constructed respectively. Each type of vector is composed of a normalized combination of multiple monitoring indicators under the corresponding dimension, unifying the data units and value ranges, and eliminating interference from differences in indicator magnitudes.

[0086] Specifically, the electrical dimension employs a similarity matching algorithm to perform distance or similarity calculations between real-time electrical feature vectors and historical electrical baseline distribution sets under normal operating conditions. The degree of deviation is then determined using quantification methods such as cosine similarity or Euclidean distance. The numerical values ​​are then used to calculate the electrical deviation degree, where a larger electrical deviation degree indicates a more severe deviation from the normal operating range. The environmental dimension substitutes environmental health feature vectors into a pre-trained environment-equipment coupling influence model. Based on the internal correlation mapping relationships within the model, the combined effect of external factors such as temperature, humidity, corrosion, and weather on the equipment is quantified to obtain the environmental stress coefficient. The mechanical dimension, based on manifold learning theory, maps mechanical health feature vectors to a pre-defined standard health manifold space. The spatial geometric distance between the real-time vector and the standard manifold surface is calculated and used as the mechanical manifold deviation distance. The pre-defined standard health manifold space uses the mechanical health feature vectors of the overhead contact system during normal operation as the basic sample. Through manifold learning algorithms, dimensionality reduction and high-dimensional feature topology reconstruction are performed to fit the normal feature distribution manifold surface when the equipment is free of defects and abnormal deterioration. The spatial boundary is defined according to the standard stress parameters of the overhead contact system, the standard deformation range, and the normal vibration and wear boundary constraints. Finally, a benchmark manifold space that can characterize the distribution law of the ideal health state in the mechanical dimension is constructed.

[0087] The environment-equipment coupling impact model employs a lightweight, fully connected coupled mapping network, comprising a three-layer architecture: an input mapping layer, which receives normalized environmental feature vectors, performs dimensional transformation and shallow feature purification, and mines the basic correlations of single-class environmental indicators; a coupling correlation hidden layer, which uses a double-layer hidden neuron structure with a nonlinear activation function to mine the cross-coupling effects of multiple environmental factors such as temperature and humidity coupling, wind and ice coupling, and corrosion and aging, and learns the intrinsic correlation between the environment and the performance degradation of the overhead contact line equipment; and a coefficient regression output layer, which constrains the output range through a normalized activation function, fits a continuous environmental stress coefficient, and achieves accurate quantitative regression of the degree of environmental stress. The input data for this environment-equipment coupling impact model consists of environmental health feature vectors, including normalized environmental monitoring time-series features such as temperature, humidity, wind speed, wind direction, ice thickness, dust concentration, and corrosive gas concentration along the line. The output data is the environmental stress coefficient, a quantified value in the range of 0-1, where a larger environmental stress coefficient indicates a stronger comprehensive deterioration and stress effect of external factors such as icing on the overhead contact line equipment.

[0088] The dataset for the environment-equipment coupling impact model is divided into an 80% training set and a 20% test set. The training set is used for learning coupling patterns, while the test set is used for generalization capability verification and model optimization. The training parameters are as follows: the initial learning rate is set to 0.001, using an adaptive learning rate decay strategy; the batch size is set to 32, the maximum number of iterations is 100, and L2 regularization and early stopping mechanisms are added to prevent overfitting; the Adam optimizer is used, and the mean squared error (MSE) loss function is adopted to suit continuous coefficient regression tasks; the number of neurons in the hidden layer is reasonably configured based on the environmental feature dimensions to ensure the ability to extract coupling features while controlling the computational load.

[0089] Three heterogeneous indicators—electrical deviation, environmental stress coefficient, and mechanical manifold deviation distance—are input into a multi-source heterogeneous feature fusion discriminator. Through weighted fusion, feature coupling mapping, and a comprehensive scoring mechanism, single-dimensional defect information is fused, and the coupled influence relationships are analyzed to obtain a comprehensive health status discrimination index. Then, based on the historical health status evolution trajectory formed by the operation of the overhead contact system, the comprehensive discrimination index is used as the current coordinate to calculate the real-time relative position of the current state on the evolution trajectory. Simultaneously, the state deviation rate is obtained by the ratio of the change in the time-series index to the time interval.

[0090] Based on relative position and deviation rate, the health status of the overhead contact line is determined, and the judgment results are obtained, including abnormal status, warning status, and normal status.

[0091] As can be seen from the above, this embodiment decomposes the final health status of the overhead contact system into three categories of health feature vectors: electrical, environmental, and mechanical. It then calculates the electrical deviation through similarity matching, solves the environmental stress coefficient based on a coupling model, and obtains the mechanical deviation distance based on manifold comparison, thereby improving the quantitative accuracy of the anomaly degree in each dimension. A comprehensive health index is then generated by a multi-source heterogeneous fusion discriminator, and the relative position of the current state and the rate of deterioration deviation are calculated based on historical evolution trajectories, taking into account both static deviation degree and dynamic change trends. Finally, the system determines the three levels of status: normal, attentive, and abnormal, effectively integrating multi-source heterogeneous data and further enhancing the comprehensiveness, scientific rigor, and fault prediction capabilities of the overhead contact system health assessment.

[0092] In one embodiment of this application, the health status of the overhead contact line is determined based on relative position and deviation rate, and a determination result is obtained, including: The relative position is used as the normalized arc length distance between the projection point of the current comprehensive health status index on the historical health status evolution trajectory and the trajectory starting point; The deviation rate is used as the ratio between the actual descent slope of the current health status comprehensive discrimination index and the standard descent slope of the projection point corresponding to the historical health status evolution trajectory; and the orthogonal distance between the current health status relative position on the evolution trajectory and the corresponding position on the historical health status evolution trajectory is calculated to obtain the trajectory deviation degree. When the normalized arc length is less than the preset first distance threshold, the ratio is less than the preset first speed threshold, and the trajectory deviation is less than the preset deviation threshold, it is determined to be in a normal state. When the normalized arc length is greater than or equal to the first distance threshold, or the ratio is greater than or equal to the first rate threshold, or the trajectory deviation is greater than or equal to the deviation threshold, it is determined to be a state of attention or an abnormal state. The state of attention and the abnormal state are further distinguished based on whether the ratio is greater than the preset second rate threshold or whether the trajectory deviation is greater than the preset severe trajectory deviation threshold.

[0093] In this embodiment, firstly, a complete historical health status evolution trajectory is constructed based on continuous health status data of the overhead contact system. The progress of the current health index along the overall evolution path is represented by a normalized arc length, quantifying the cumulative process of equipment aging and performance degradation to eliminate evaluation errors caused by differences in dimensionality.

[0094] Secondly, based on two core correction indicators—deviation rate and trajectory deviation—refined discrimination is achieved from two dimensions: the speed of change and the degree of deviation. Specifically, the deviation rate is defined as the ratio of the actual decay slope to the standard evolution slope, which measures the rate at which the current state deteriorates; the trajectory deviation is calculated using orthogonal distance. For example, using the projection point of the current health status comprehensive discrimination index on the historical evolution trajectory as a benchmark, the standard trajectory curve of normal health evolution at that location is selected, and the perpendicular orthogonal straight-line distance from the real-time health status sample point to the corresponding standard trajectory line segment is calculated. This perpendicular distance is normalized and used as the trajectory deviation. The larger the trajectory deviation distance value, the more obvious the deviation of the current health operation state from the normal decay evolution law, which can accurately quantify the trajectory deviation differences caused by sudden anomalies, local damage, and external disturbances.

[0095] Finally, a multi-level threshold system is established to create a graded judgment mechanism, which uses joint logical judgment based on three indicators: arc length distance, deviation rate, and trajectory deviation degree. Specifically, the catenary equipment is judged to be in a stable and controllable normal operating state only when the normalized arc length distance, rate ratio, and trajectory deviation degree are all simultaneously less than the preset first distance threshold, preset first rate threshold, and preset deviation threshold, respectively. If any one of the three indicators is greater than or equal to the corresponding threshold limit, it deviates from the normal state range and is uniformly classified into the risk control category, judged as a state of attention or an abnormal state, and a secondary classification is carried out, including: further delineating and defining the over-limit conditions based on the preset second rate threshold and the preset severe trajectory deviation threshold. If the ratio is greater than the preset second rate threshold or the trajectory deviation degree is greater than the preset severe trajectory deviation threshold, the health status of the catenary is classified as an abnormal state; if the ratio is less than the second rate threshold or the trajectory deviation degree is less than the severe trajectory deviation threshold, the health status of the catenary is classified as a state of attention. The preset first distance threshold is based on the upper limit of the normalized arc length fluctuation of the health status evolution trajectory within the normal service life of the overhead contact system. It uses the distribution range of arc lengths of historical normal samples without faults or significant aging, along with the equipment's design service life and the reasonable process of natural aging, to define a critical value, used to distinguish between the normal healthy degradation range and the range of cumulative deterioration exceeding the standard. The preset first rate threshold is based on the stable ratio range of the standard decline slope under normal slow aging of the overhead contact system. It uses the maximum rate fluctuation of normal evolution as a benchmark, and after removing extreme and accidental data, calibrates a critical ratio to identify whether the rate of health status deterioration exceeds the allowable normal decay rate. The preset deviation threshold is set based on the steady-state range of the orthogonal distance from the health status points of the summarized normal operating samples to the standard evolution trajectory, determining the statistical upper limit of trajectory deviation under normal conditions, used to determine whether the current state deviates from the standard health evolution law. The preset second rate threshold is based on the first rate threshold. It is determined by statistically analyzing slope ratio samples under various potential hazard conditions, considering the deterioration rate characteristics of accelerated degradation of overhead contact line equipment, hidden damage to components, and fatigue wear. A critical value faster than normal degradation is selected for this setting. The second rate threshold is greater than the first rate threshold and serves as the rate classification boundary distinguishing between noticeable and abnormal states. The severe trajectory deviation threshold is based on the deviation threshold. For conditions such as sudden failures, structural anomalies, and severe environmental stress that severely deviate from normal evolution, orthogonal distance samples under the offset are statistically analyzed to define a severe deviation threshold. The severe trajectory deviation threshold is greater than the first rate threshold and is used to distinguish between general trajectory deviation and severe abnormal deviation, achieving accurate classification of abnormality levels.

[0096] As can be seen from the above, this embodiment quantifies the overall degradation progress of the health status by normalizing the arc length, characterizes the deviation rate of the status decline by using the slope ratio, and calculates the trajectory deviation degree based on the orthogonal distance to depict the health evolution characteristics of the overhead contact system. By employing a multi-condition joint judgment and hierarchical threshold grading mechanism, the limitations of a single evaluation index are effectively avoided. This approach can identify both routine, gradual aging of equipment and timely capture sudden abnormal deviations and rapid deterioration problems, further improving the comprehensiveness, accuracy, and hierarchy of health status identification.

[0097] In one embodiment of this application, alarm information, first warning information, second warning information, or third warning information are identified based on preset different levels and then output to an operation and maintenance monitoring platform, a mobile terminal, or an audible and visual alarm device, including: The target level is determined based on the type of alarm information, first warning information, second warning information, or third warning information and the corresponding severity of health status; Alarm information corresponds to emergency alarm level. The first warning information corresponds to the first-level warning level, the second warning information corresponds to the second-level warning level, and the first warning information corresponds to the first-level warning level. Based on the target level, the corresponding information is labeled with color, sound frequency and priority value; the output target combination and output order are determined according to the priority value. Based on the output order, the information with the additional identifier will be output to one or more of the following: the operation and maintenance monitoring platform, the mobile terminal, or the audible and visual alarm device.

[0098] In this embodiment, firstly, based on the specific types of alarm information and the severity of the real-time health status of the overhead contact system, risk levels are precisely classified and matched with corresponding target handling levels. Different warning and alarm information are bound to fixed level standards, with emergency alarms, Level 1 warnings, and Level 2 warnings differentiated step-by-step, forming a level correspondence. Specifically, alarm information corresponds to the emergency alarm level, the first warning information corresponds to the Level 1 warning level, the second warning information corresponds to the Level 2 warning level, and the first warning information corresponds to the Level 1 warning level.

[0099] Secondly, for the identified target levels, uniformly configure differentiated visual and auditory identifiers and priority values. Through unique color coding, different frequency sound prompts, and quantified priority parameters, the urgency of various alarms is distinguished, allowing operations and maintenance personnel to intuitively identify the level of risk. Simultaneously, output sorting rules are set based on priority values, ensuring that high-level urgent information is pushed first, preventing information congestion and the omission of critical alarms.

[0100] Finally, based on the output order, the information with the additional identifier is output to one or more of the operation and maintenance monitoring platform, mobile terminal, or audible and visual alarm device.

[0101] As can be seen from the above, this embodiment classifies standardized target risk levels based on alarm and multi-level early warning information types and the severity of health status, achieving standardized hierarchical management of risk levels. By matching each level of information with a unique color, sound frequency, and priority value, the severity and urgency of alarms are distinguished, and the output order is reasonably sorted according to priority, ensuring that high-risk information is pushed first. Simultaneously, it supports multi-terminal collaborative distribution, synchronously pushing identifiable alarms to the monitoring platform, mobile terminals, and audible and visual alarm devices, constructing a multi-channel linkage early warning mechanism. This effectively improves alarm identification efficiency and information push logic.

[0102] In one embodiment of this application, based on the output order, the information with the additional identifier is output to one or more of the following: an operation and maintenance monitoring platform, a mobile terminal, or an audible and visual alarm device, including: When the target level is an emergency alarm level, the alarm information with the attached label will be output to the operation and maintenance monitoring platform, mobile terminal and audible and visual alarm device, and the audible and visual alarm device will be triggered. When the target level is a Level 1 warning, the first warning information with an additional label will be output to the operation and maintenance monitoring platform and mobile terminal, and the triggering of the audible and visual alarm device will be prohibited. When the target level is a Level II warning, the second warning information with the additional label will only be output to the operation and maintenance monitoring platform and displayed in the form of pop-up windows and lists; When the target level is a Level 3 warning level, the third warning information with the attached label will only be output to the operation and maintenance monitoring platform and displayed in list form; In the output order, the output is processed sequentially based on the emergency alarm level, the first-level warning level, and the second-level warning level. When multiple different levels of information exist at the same time, the information corresponding to the emergency alarm level is processed first.

[0103] In this embodiment, a differentiated and refined multi-terminal distribution and push strategy was formulated for alarm information of different risk levels. Dedicated output channels were matched according to the hierarchical differences between emergency alarms and level 1, 2, and 3 warnings. Specifically, the highest-level emergency alarm information is covered across all channels, simultaneously pushed to the operation and maintenance monitoring platform and mobile terminals, and activating audible and visual alarms to maximize the alerting effect of emergency faults and ensure that major risks are detected as soon as possible.

[0104] Secondly, low- and medium-level alerts are pushed out in a tiered manner. Specifically, Level 1 alerts are pushed simultaneously to the management platform and mobile terminals, accommodating both remote monitoring and mobile viewing; Level 2 alerts are only displayed in a pop-up window and list on the platform, and Level 3 alerts are only silently recorded in the platform list. This approach reduces excessive disruption to maintenance personnel caused by minor alerts while ensuring complete retention of all abnormal status records, achieving a balanced approach to alarm control.

[0105] Finally, the system is configured with a fixed priority logic for outputting alarms, processing them in a queue according to the order of emergency alarm, first-level warning, second-level warning, and third-level warning. When multiple alarms occur concurrently, the system automatically prioritizes responding to the highest-level emergency information.

[0106] As can be seen from the above, this embodiment sets differentiated distribution channels and display formats for the four alarm levels, constructing a hierarchical alarm push mechanism. Emergency alarms provide full terminal coverage and activate strong audible and visual alerts to ensure rapid response to major faults; Level 1 warnings are simultaneously pushed to the platform and mobile terminals, accommodating remote control; Level 2 warnings are displayed in a pop-up window and list, while Level 3 warnings are only silently recorded in a list, reasonably controlling the scope of alarm disturbance. Simultaneously, alarm output priorities are set from high to low, prioritizing the processing of high-level alarms when multiple types of information occur concurrently. This achieves precise matching between risk severity and alert intensity, further reducing excessive interference from low-risk warnings, and standardizing alarm output order. This helps maintenance personnel handle alarms according to their urgency, improving the scientific nature of overhead contact line alarm management and maintenance response efficiency.

[0107] In one embodiment of this application, a status analysis and early warning method based on online monitoring data of overhead contact lines further includes: Obtain maintenance feedback data for alarm information, first warning information, second warning information, or third warning information. Maintenance feedback data includes on-site inspection results, maintenance records, equipment replacement records, and retested health status data. The operation and maintenance feedback data is time-aligned and matched with the corresponding judgment results and predicted health status information, and a closed-loop record of early warning-handling-verification is generated. When there is an inconsistency between the operation and maintenance feedback data and the judgment result or predicted health status information, it is marked as a prediction deviation sample. Based on the prediction deviation sample in the closed-loop record, the preset threshold set is adjusted. The first distance threshold, the first rate threshold, the deviation threshold, the second rate threshold, and the severe trajectory deviation threshold are used as the threshold set.

[0108] In this embodiment, comprehensive operation and maintenance feedback data corresponding to various alarms and multi-level early warnings are collected. This data includes on-site inspection results, maintenance logs, equipment replacement records, and post-repair health status data. Secondly, the various operation and maintenance feedback data are time-aligned and correlated with previous status judgment results and health status prediction information, connecting the entire process of hazard warning, manual handling, and on-site verification to form a standardized warning-handling-verification closed-loop record. This ensures that the entire lifecycle of each alarm event is traceable and verifiable.

[0109] Finally, the actual feedback data from the site is compared with the analysis and prediction results, and the discrepancies are marked as prediction deviation samples. Based on the deviation sample data in the closed-loop record, the preset threshold set is adjusted, wherein the first distance threshold, the first velocity threshold, the deviation threshold, the second velocity threshold, and the severe trajectory deviation threshold are used as the threshold set.

[0110] As can be seen from the above, this embodiment establishes a closed-loop record of early warning, handling, and verification by aligning and matching operation and maintenance feedback data with time, achieving full traceability of alarm events. By comparing actual on-site feedback with analysis and prediction results, prediction deviation samples are automatically marked, and various discrimination thresholds such as distance, rate, and trajectory deviation are dynamically optimized based on operation and maintenance data. This effectively reduces the problems of misjudgment and missed judgment of status, and further improves the accuracy of overhead contact line health assessment and early warning analysis.

[0111] In one embodiment of this application, adjusting a preset threshold set based on prediction deviation samples in closed-loop records includes: The prediction bias samples are classified according to the associated threshold type to obtain distance bias samples, rate bias samples and deviation bias samples respectively. Calculate the first mean and standard deviation of the deviation values ​​in the distance deviation sample. Based on the first mean and a preset adjustment step size coefficient, determine the first threshold increment. Update the first distance threshold by adding the first distance threshold and the first threshold increment, and use a preset multiple of the standard deviation as the upper confidence bound of the adjustment range. If the updated first distance threshold is greater than or equal to the second distance threshold, then set the first distance threshold to the second distance threshold minus a preset minimum value. Calculate the second mean of the deviation values ​​corresponding to the first rate threshold in the rate deviation sample, determine the second threshold increment based on the second mean and the adjustment step size coefficient, and complete the update by adding the first rate threshold to the second threshold increment; Calculate the third mean of the deviation values ​​corresponding to the second rate threshold in the rate deviation sample, determine the third threshold increment based on the third mean and the adjustment step size coefficient, complete the update by adding the second rate threshold to the third threshold increment, and keep the updated first rate threshold less than the second rate threshold. Calculate the fourth mean of the deviation values ​​corresponding to the deviation threshold in the deviation deviation sample, determine the fourth threshold increment based on the fourth mean and the adjustment step size coefficient, and complete the update by adding the deviation threshold to the fourth threshold increment; Calculate the fifth mean of the deviation values ​​in the deviation samples that correspond to the severe trajectory deviation threshold. Based on the fifth mean and the adjustment step size coefficient, determine the fifth threshold increment. Update the threshold by adding the severe trajectory deviation threshold to the fifth threshold increment. Keep the updated deviation threshold less than the severe trajectory deviation threshold.

[0112] In this embodiment, various prediction deviation samples are divided according to the discrimination threshold type associated with the occurrence of the deviation, and uniformly classified into three independent datasets: distance deviation samples, rate deviation samples, and deviation degree deviation samples.

[0113] Specifically, for the adaptive update of distance-related thresholds, a separate distance deviation sample set is extracted. The first mean and standard deviation of the deviation values ​​within each sample are calculated. The overall deviation level is quantified based on the mean and multiplied by a preset adjustment step size coefficient to obtain the first threshold increment. During the iterative update of the first distance threshold, an incremental stacking method is used to complete the basic adjustment. At the same time, a fixed multiple of the standard deviation is introduced as a confidence constraint upper limit for the adjustment range, limiting the magnitude of a single correction and preventing large jumps in the threshold. The fixed multiple of the standard deviation is a dimensionless constant (fixed coefficient, such as 1.5, 2.0, 2.5) preset by the system. It is a globally fixed configuration parameter that does not change with samples or operating conditions. It is used to limit the maximum allowable range of the first threshold increment to prevent individual abnormal deviation samples or outliers from causing a one-time large jump in the threshold. The specific value is preferentially set to 2.0 by default, and is differentiated between 1.5 and 2.5 based on the line operation stability and the intensity of environmental interference. The more stable the line operation and the lower the intensity of environmental interference, the smaller the value of the fixed multiple of the standard deviation. Meanwhile, a hierarchical boundary constraint logic is added. Once the updated first distance threshold approaches or exceeds the second distance threshold, it is automatically corrected to the second distance threshold minus a preset minimum value, strictly ensuring the hierarchical relationship of the distance-type dual thresholds and maintaining the logical rationality of the distance discrimination system. Specifically, based on historical threshold deviation data accumulated across the entire overhead contact line, the normal deviation fluctuation range and reasonable single correction range of each type of threshold under natural operating conditions are statistically analyzed to determine the value range of the step size coefficient. Then, through multiple sets of simulation iteration tests, the optimal fixed value is selected.

[0114] A hierarchical differentiated update process is implemented for the two-level rate thresholds. Rate deviation samples corresponding to the first and second rate thresholds are extracted separately, and the second and third mean values ​​of the deviation values ​​are calculated sequentially. The increments of the second and third thresholds are calculated respectively based on a unified adjustment step size coefficient. The synchronous iterative optimization of the two rate thresholds is completed by incremental superposition. After the update, the numerical relationship is forcibly verified to ensure that the adjusted first rate threshold is always less than the second rate threshold, thus solidifying the hierarchical architecture of the rate-based early warning system.

[0115] Simultaneous correction is performed on the two sets of core thresholds for the trajectory deviation dimension. Deviation samples corresponding to the normal deviation threshold and the severe trajectory deviation threshold are selected respectively. The fourth and fifth means of the deviation data are calculated, and the step size coefficients are matched and adjusted to solve for the fourth and fifth threshold increments in turn. The two deviation thresholds are adjusted by superposition and update. In addition, numerical constraints are added to ensure that the normal deviation threshold is always less than the severe trajectory deviation threshold after the update, thus fully preserving the classification boundary between mild and severe deviation anomalies.

[0116] As can be seen from the above, this embodiment classifies the predicted deviation samples according to the threshold type, realizes independent statistics and targeted correction of deviation data in different dimensions, and quantitatively calculates the threshold increment based on the mean of each type of deviation sample and a fixed adjustment step size coefficient, thus completing the adaptive iterative update of each level threshold. At the same time, the introduction of a standard deviation preset multiple constraint on the upper limit of single adjustment reduces large-scale threshold changes, and the addition of hierarchical size constraints between multiple sets of thresholds effectively ensures the logical hierarchy and numerical rationality of various discrimination thresholds such as distance, rate, and deviation.

[0117] Corresponding to the status analysis and early warning method based on online monitoring data of the overhead contact line in the above embodiment, Figure 2 This is a structural block diagram of a condition analysis and early warning system based on online monitoring data of overhead contact lines, provided as an embodiment of this application. For ease of explanation, only the parts relevant to the embodiment of this application are shown. References Figure 2 The status analysis and early warning system 20 based on online monitoring data of the overhead contact line includes: a data acquisition module 21, a status assessment module 22, a status classification module 23, an instant response module 24, a trend early warning module 25, and an information output module 26.

[0118] Among them, the data acquisition module 21 is used to acquire electrical data, environmental data and mechanical data of the target contact network through sensors installed on the contact network; The condition assessment module 22 is used to preprocess and fuse electrical and mechanical data, input them into a preset contact network condition judgment model to obtain an initial health state, and correct the initial health state based on environmental data to obtain a final health state. The status classification module 23 is used to determine the health status of the overhead contact system based on the final health status, and obtain the judgment result; the judgment result includes normal status, attention status, and abnormal status; The instant response module 24 is used to generate an alarm message if the judgment result is in an abnormal state, and to generate a first warning message if the judgment result is in an alert state. The trend warning module 25 is used to predict the final health status trend if the judgment result is in a normal state, and obtain the preset predicted health status information for the next period; to determine whether the predicted health status information is in an attentive state or an abnormal state; if the predicted health status information is in an attentive state, a second warning message is generated; if the predicted health status information is in an abnormal state, a third warning message is generated. The information output module 26 is used to identify alarm information, first warning information, second warning information or third warning information based on different preset levels, and then output them to the operation and maintenance monitoring platform, mobile terminal or audible and visual alarm device.

[0119] See Figure 3 , Figure 3 This is a schematic block diagram of an electronic device provided according to an embodiment of this application. Figure 3 The electronic device 300 in this embodiment may include one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The processors 301, input devices 302, output devices 303, and memories 304 communicate with each other via a communication bus 305. The memories 304 store computer programs, including program instructions. The processors 301 execute the program instructions stored in the memories 304. Specifically, the processors 301 are configured to invoke the program instructions to perform the functions of the modules in the aforementioned device embodiments, for example... Figure 2 The functions of the data acquisition module 21, status assessment module 22, status classification module 23, real-time response module 24, trend early warning module 25, and information output module 26 are shown.

[0120] It should be understood that, in the embodiments of this application, the processor 301 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0121] Input device 302 may include a touchpad, a fingerprint sensor (for collecting the user's fingerprint information and fingerprint orientation information), a microphone, etc., and output device 303 may include a display (LCD, etc.), a speaker, etc.

[0122] The memory 304 may include read-only memory and random access memory, and provides instructions and data to the processor 301. A portion of the memory 304 may also include non-volatile random access memory. For example, the memory 304 may also store device type information.

[0123] In specific implementations, the processor 301, input device 302, and output device 303 described in the embodiments of this application can execute the implementation methods described in any embodiment of the status analysis and early warning method based on online monitoring data of the contact network provided in the embodiments of this application, or they can execute the implementation methods of the electronic devices described in the embodiments of this application, which will not be repeated here.

[0124] In another embodiment of this application, a computer-readable storage medium is provided. This computer-readable storage medium stores a computer program, which includes program instructions. When executed by a processor, the program instructions implement all or part of the processes in the methods described above. Alternatively, the computer program can instruct related hardware to complete the process. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include any entity or device capable of carrying computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0125] The computer-readable storage medium can be an internal storage unit of the electronic device in any of the foregoing embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device. Furthermore, the computer-readable storage medium can include both internal and external storage units of the electronic device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0126] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0127] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the electronic devices and units described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0128] In the several embodiments provided in this application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, 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. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces or units, or it may be an electrical, mechanical, or other form of connection.

[0129] 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 embodiments of this application, depending on actual needs.

[0130] 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.

[0131] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for status analysis and early warning based on online monitoring data of overhead contact lines, characterized in that, include: Electrical, environmental, and mechanical data of the target contact network are acquired through sensors installed on the contact network. The electrical data and mechanical data are preprocessed and fused, and then input into a preset contact network status judgment model to obtain an initial health status. The initial health status is then corrected based on the environmental data to obtain a final health status. The health status of the overhead contact system is assessed based on the final health status, resulting in a judgment result; the judgment result includes a normal status, a warning status, and an abnormal status. If the judgment result is in an abnormal state, an alarm message is generated; If the judgment result indicates an attentive state, then a first warning message is generated; If the judgment result is in a normal state, then the final health state is predicted to obtain the predicted health state information for the next cycle. Determine whether the predicted health status information is in the attention state or the abnormal state. If the predicted health status information is in the attention state, then generate a second warning message. If the predicted health status information is in an abnormal state, a third early warning message is generated; The alarm information, the first warning information, the second warning information, or the third warning information are identified based on different preset levels and then output to the operation and maintenance monitoring platform, mobile terminal, or audible and visual alarm device.

2. The method for status analysis and early warning based on online monitoring data of overhead contact lines according to claim 1, characterized in that, The step of correcting the initial health state based on the environmental data to obtain the final health state includes: The environmental data includes temperature, wind speed, wind direction, ice thickness, dust concentration, humidity, and corrosive gas concentration. Based on the wind speed data, the wind direction data, and the ice thickness data, an environmental comprehensive force vector is obtained, and the initial health state is corrected to obtain first corrected health state information. Based on the dust concentration data, humidity data, temperature data, and corrosive gas concentration data, a comprehensive corrosion-abrasion dose is determined, and the first corrected health status information is corrected to obtain the second corrected health status information. The trajectory of the temperature data change is monitored, and it is identified whether the phase transition critical point is crossed, wherein the phase transition critical point includes the freezing point and dew point temperature. If there is a critical point that crosses the phase transition, the thermal stress of the temperature gradient is calculated based on the temperature data of multiple monitoring points along the line, and the mechanical relaxation and tension components in the second corrected health state information are corrected based on the thermal stress to obtain the final health state.

3. The method for status analysis and early warning based on online monitoring data of overhead contact lines according to claim 2, characterized in that, Based on the wind speed data, wind direction data, and icing thickness data, an environmental comprehensive force vector is obtained, and the initial health state is corrected to obtain first corrected health state information, including: Based on the wind speed data and the wind direction data, determine the environmental wind vector; The ice thickness data is converted into an additional mass load on the overhead contact line to obtain the ice load; The force components of the environmental wind vector and the icing load in the three spatial dimensions of the contact wire are calculated and vector superimposed to obtain the comprehensive environmental force vector. The magnitude of the combined environmental force vector is used as the correction strength, and the angle between the direction of the combined environmental force vector and the design bearing direction of the contact wire is used as the direction correction coefficient. The product of the correction intensity and the direction correction coefficient is used to correct the initial health state in an exponential form to obtain the first corrected health state information.

4. The method for status analysis and early warning based on online monitoring data of overhead contact lines according to claim 3, characterized in that, Based on the dust concentration data, temperature data, humidity data, and corrosive gas concentration data, a comprehensive corrosion-abrasion dose is determined, and the first corrected health status information is corrected to obtain second corrected health status information, including: The product of the humidity data and the corrosive gas concentration data is used as the basic corrosion rate, and the temperature data is used as the reaction rate acceleration factor. The Arrhenius formula is used for correction to obtain the target corrosion rate. The cumulative corrosive damage is obtained by integrating the target corrosion rate over N consecutive time windows. The dust concentration data is used as an erosion acceleration factor. Based on the erosion acceleration factor and the cumulative corrosive damage, the comprehensive corrosion-erosion dose is determined. The first corrected health status information is corrected by applying the comprehensive corrosion-wearing dose as a continuous cumulative function to obtain the second corrected health status information.

5. The method for status analysis and early warning based on online monitoring data of overhead contact lines according to claim 2, characterized in that, Before calculating the thermal stress of the temperature gradient based on temperature data from multiple monitoring points along the line, and correcting the mechanical relaxation and tension components in the second corrected health state information based on the thermal stress to obtain the final health state, the process further includes: The temperature gradient along the overhead contact line is calculated based on the temperature data differences from multiple monitoring points along the line and the distance between the monitoring points. Based on the temperature gradient, the linear expansion coefficient of the contact wire, the elastic modulus, and the original tension state, the additional thermal stress caused by uneven temperature distribution is calculated using thermoelastic mechanics formulas; wherein, the original tension state is the design rated tension value of the contact wire. The second corrected health state information is corrected based on the additional thermal stress, and the components related to mechanical sag and tension are extracted from the second corrected health state information as the mechanical state reference value to be corrected.

6. The method for status analysis and early warning based on online monitoring data of overhead contact lines according to claim 5, characterized in that, The step of extracting component data related to mechanical sag and tension from the second corrected health status information as the mechanical status reference value to be corrected includes: The second corrected health status information is vector decomposed to separate the tension component along the contact line axis, the lateral sag component perpendicular to the contact line direction, and the vertical lift component. The tension component is mapped to the contact network design tension reference range, and the tension deviation rate is obtained by comparing the current tension value with the rated tension value. The lateral sag component and the vertical lift component are geometrically synthesized to obtain a comprehensive sag vector, and the sag ratio is calculated based on the span length as the sag reference value. Anomaly removal and smoothing filtering are performed on the tension deviation rate and the sag reference value using a sliding time window to obtain the tension reference value and the sag reference value. The tension reference value and the relaxation reference value are used as the mechanical state reference value.

7. The method for status analysis and early warning based on online monitoring data of overhead contact lines according to claim 6, characterized in that, The process of calculating the thermal stress based on temperature gradients from multiple monitoring points along the line, and then correcting the mechanical relaxation and tension components in the second corrected health state information based on the thermal stress to obtain the final health state, includes: When the absolute value of the temperature gradient exceeds the preset temperature threshold range, the temperature difference between each monitoring point and the preset reference point is calculated, and the thermal expansion strain is calculated based on the linear expansion coefficient of the contact line. The thermal expansion strain is converted into thermal stress and superimposed on the mechanical relaxation and tension components to be corrected. When the absolute value of the temperature gradient is less than or equal to the temperature threshold, the second corrected health status information is taken as the final health status.

8. The method for status analysis and early warning based on online monitoring data of overhead contact lines according to claim 1, characterized in that, The process of judging the health status of the overhead contact line based on the final health status, and obtaining the judgment result, includes: Multidimensional feature extraction is performed on the final health state to obtain electrical health feature vector, environmental health feature vector and mechanical health feature vector; The electrical health feature vector is matched with the electrical feature benchmark distribution under historical normal conditions to calculate the electrical deviation. The environmental health feature vector is input into a preset environment-equipment coupling influence model to obtain the environmental stress coefficient; The mechanical health feature vector is compared with a preset standard health manifold, and the deviation distance of the mechanical manifold is calculated. The electrical deviation, the environmental stress coefficient, and the mechanical manifold deviation distance are input into a multi-source heterogeneous feature fusion discriminator, which outputs a comprehensive health status discriminant index. Based on the comprehensive health status discrimination index and the historical health status evolution trajectory, the relative position and deviation rate of the current health status on the evolution trajectory are calculated. Based on the relative position and the deviation rate, the health status of the overhead contact line is determined, and a judgment result is obtained, wherein the judgment result includes abnormal status, alert status and normal status.

9. A method for status analysis and early warning based on online monitoring data of overhead contact lines according to claim 8, characterized in that, The process of determining the health status of the overhead contact system based on the relative position and the deviation rate, and obtaining the determination result, includes: The relative position is taken as the normalized arc length distance between the projection point of the current health status comprehensive discrimination index on the historical health status evolution trajectory and the trajectory starting point. The deviation rate is used as the ratio between the actual descent slope of the current health status comprehensive discrimination index and the standard descent slope of the projection point corresponding to the historical health status evolution trajectory; and the orthogonal distance between the current health status relative position on the evolution trajectory and the corresponding position on the historical health status evolution trajectory is calculated to obtain the trajectory deviation degree. When the normalized arc length is less than a preset first distance threshold, the ratio is less than a preset first speed threshold, and the trajectory deviation is less than a preset deviation threshold, it is determined to be in a normal state. When the normalized arc length is greater than or equal to the first distance threshold, or the ratio is greater than or equal to the first rate threshold, or the trajectory deviation is greater than or equal to the deviation threshold, it is determined to be a state of attention or an abnormal state. The state of attention and the abnormal state are further distinguished based on whether the ratio is greater than a preset second rate threshold or whether the trajectory deviation is greater than a preset severe trajectory deviation threshold.

10. A status analysis and early warning system based on online monitoring data of overhead contact lines, characterized in that, include: The data acquisition module is used to acquire electrical, environmental, and mechanical data of the target contact network through sensors installed on the contact network; The status assessment module is used to preprocess and fuse the electrical data and the mechanical data, and input them into a preset contact network status judgment model to obtain an initial health status. Based on the environmental data, the initial health status is corrected to obtain a final health status. The status grading module is used to determine the health status of the overhead contact system based on the final health status, and obtain a judgment result; the judgment result includes normal status, alert status, and abnormal status; An instant response module is used to generate an alarm message if the judgment result is in an abnormal state. If the judgment result indicates an attentive state, then a first warning message is generated; The trend warning module is used to predict the trend of the final health status if the judgment result is in a normal state, and obtain the predicted health status information for the next cycle. Determine whether the predicted health status information is in the attention state or the abnormal state. If the predicted health status information is in the attention state, then generate a second warning message. If the predicted health status information is in an abnormal state, a third early warning message is generated; The information output module is used to identify the alarm information, the first warning information, the second warning information, or the third warning information based on different preset levels, and then output them to the operation and maintenance monitoring platform, mobile terminal, or audible and visual alarm device.

Images