Railway four-electricity intelligent construction monitoring system based on multi-source data fusion

Abstract

The invention relates to the technical field of railway engineering construction, in particular to a railway four-electricity intelligent construction monitoring system based on multi-source data fusion, which comprises a division evaluation acquisition unit, a parameter calculation and mapping unit, a performance pre-judgment and strategy adjustment unit and a static parameter adjustment unit. Collecting static parameters of each measuring point, inputting a contact suspension dynamic performance prediction model, outputting average elasticity and maximum elasticity difference of a unit section, comparing with a standard to judge whether the dynamic performance reaches the standard or not, if not, generating a contact line target static parameter optimization scheme according to a model sensitivity analysis result, and performing closed-loop adjustment until the dynamic performance reaches the standard. The system realizes static and dynamic parameter association pre-judgment and accurate adjustment, is suitable for railway contact network construction monitoring, and improves the construction quality and efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of railway engineering construction, in particular to a railway four-electricity intelligent construction monitoring system based on multi-source data fusion. BACKGROUND

[0002] Railway engineering construction is an important technology. In the construction of high-speed railways and ordinary speed railways, the catenary is a core component of train power reception, and its dynamic performance directly determines the safety and stability of train operation. Good dynamic performance can avoid excessive wear and tear of the pantograph and the contact wire, reduce off-line sparks, and ensure continuous and stable current collection of the train. Precise construction monitoring can detect catenary hidden dangers in advance and reduce the cost of later operation and maintenance.

[0003] However, the existing railway catenary construction monitoring technology has the core problems of dynamic performance prediction lag and blind static parameter adjustment, which are caused by three actual limitations: first, traditional monitoring mainly collects static parameters of the contact wire, and does not establish a correlation model between static parameters and dynamic performance, so it is impossible to predict in advance whether the dynamic performance of the catenary meets the standard through static data. Problems can only be found through dynamic tests after construction is completed, resulting in high rework costs. Second, the adjustment of static parameters lacks pertinence. When the dynamic performance is not up to standard, only manual experience is used to adjust the parameters, and the influence weight of each parameter on the dynamic performance is not quantified, so repeated adjustment may still not meet the standard. Third, the parameter monitoring dimension is incomplete. Some systems do not synchronously collect the contact wire tension, positioner slope mechanics and geometric parameters, so static data cannot fully reflect the state of the contact suspension system, further affecting the accuracy of dynamic performance prediction. These problems will have a chain effect: dynamic prediction lag makes it difficult to discover construction hidden dangers in time, which may cause current collection failure during train test operation; blind static adjustment prolongs the construction period and increases the cost of manpower and materials; and incomplete parameter monitoring reduces the reliability of prediction and easily leads to misjudgment or omission. Ultimately, the existing technology cannot meet the needs of railway catenary construction, and an intelligent monitoring scheme that can fuse multi-source static data, establish static and dynamic correlation, and realize precise parameter optimization is urgently needed. To solve this technical problem, we provide a railway four-electricity intelligent construction monitoring system based on multi-source data fusion. SUMMARY

[0004] The purpose of the present application is to provide a railway four-electricity intelligent construction monitoring system based on multi-source data fusion to solve the problems raised in the background.

[0005] 1. Because traditional monitoring does not have a static and dynamic parameter correlation model, the dynamic prediction lags, so in this case, the parameter calculation and mapping unit inputs static parameters into the contact suspension dynamic performance prediction model, and the output unit outputs the average elastic dynamic index of the section, which can predict the dynamic performance in advance and avoid rework.

[0006] 2. Since static parameter adjustments lack weighting and are often arbitrary, this case study utilizes a static parameter adjustment unit to generate a target static parameter optimization scheme based on the sensitivity analysis results of the prediction model and performs closed-loop adjustments. This allows for precise parameter adjustments and shortens the construction cycle.

[0007] To achieve the above objectives, a railway electrical, signaling, and communication intelligent construction monitoring system based on multi-source data fusion is provided, including: The catenary line to be adjusted is divided into several continuous evaluation unit segments. Within each evaluation unit segment, several representative measurement points are selected. Using a catenary geometric parameter measuring instrument and a tension meter, the static parameters at each measurement point are measured and recorded. The parameter calculation and mapping unit inputs static parameters into the preset contact suspension dynamic performance prediction model and predicts and outputs the key dynamic performance prediction indicators of the evaluation unit segment, including the average elasticity of the unit segment and the maximum elasticity difference of the unit segment. The performance prediction and strategy adjustment unit compares the average elasticity of the unit segment with the preset standard elasticity range, and at the same time compares the maximum elasticity difference of the unit segment with the preset maximum allowable elasticity difference to determine whether the dynamic performance prediction of the evaluation unit segment meets the standard. If it does not meet the standard, it enters the static parameter adjustment unit. For evaluation unit segments determined to be unqualified in dynamic performance prediction, the static parameter adjustment unit generates an optimized adjustment scheme for the static height and static pull-out value of the contact wire target at specific measurement points within the evaluation unit segment based on the sensitivity analysis results output by the contact suspension dynamic performance prediction model. The static geometric parameters of the specified measurement points within the evaluation unit segment are adjusted according to the optimized adjustment scheme until the evaluation unit segment is determined to be qualified in dynamic performance prediction.

[0008] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. By using the parameter calculation and mapping unit, static parameters are input into the dynamic performance prediction model of the contact suspension to output dynamic indicators, thereby predicting the dynamic performance of the contact network in advance, solving the problem of lagging dynamic prediction in traditional monitoring, avoiding rework after construction, reducing costs and shortening the cycle.

[0009] 2. Based on the static parameter adjustment unit, an optimization scheme is generated and adjusted in a closed loop according to the sensitivity analysis results of the prediction model, so as to achieve precise optimization of static parameters, solve the problem of blind adjustment by manual adjustment, reduce repeated adjustments, and improve construction efficiency.

[0010] 3. By dividing the evaluation and acquisition units, static parameters are collected from multiple dimensions, covering the geometric and mechanical characteristics of the contact suspension. This solves the problem of incomplete parameter monitoring dimensions, provides comprehensive data for dynamic performance prediction, and improves the reliability of prediction. Attached Figure Description

[0011] Figure 1The overall block diagram of the present application.

[0012] The meanings of the various reference numerals in the figures are as follows: 1, division evaluation acquisition unit; 2, parameter calculation and mapping unit; 3, performance prediction and strategy adjustment unit; 4, static parameter adjustment unit. DETAILED DESCRIPTION

[0013] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of the present application.

[0014] The present application provides a railway four-electricity intelligent construction monitoring system based on multi-source data fusion, please refer to Figure 1 as shown, comprising: The division evaluation acquisition unit 1 divides the contact line to be adjusted into a plurality of continuous evaluation unit segments, and in each evaluation unit segment, a plurality of representative measurement points are selected, and a contact line geometry parameter measuring instrument and a tension meter are used to measure and record the static parameters at each measurement point. The static parameters include: The static height of the contact wire at the measurement point, reflecting the vertical position reference value of the wire, the static pull-out value, reflecting the horizontal offset reference value of the wire, the real-time collected contact wire tension, representing the mechanical stress state of the suspension system, the length of the adjacent dropper node, indicating the geometric relationship between the suspension points and the slope of the positioner at the positioning point, describing the spatial angle between the support structure and the wire, wherein the contact wire tension is measured by dynamic calibration of the tension meter, the length of the dropper and the slope of the positioner are synchronously collected by high-precision laser range finder and inclinometer, and the static parameter set covers the spatial geometry and mechanical properties of the contact suspension.

[0015] After the division evaluation acquisition unit completes the collection and preprocessing of the static parameters of each measurement point, a core associated carrier of static parameters and dynamic performance, a contact suspension dynamic performance prediction model, needs to be built through the parameter calculation and mapping unit. This model is the key to solving the problem of isolated static data and lagging dynamic prediction in traditional monitoring. It not only integrates the achievements of the multi-dimensional static parameter collection described above, but also converts abstract static data into dynamic indicators reflecting the actual operating state of the contact line, providing quantitative basis for subsequent performance prediction and accurate adjustment, and ensuring that the construction monitoring is transferred from post-detection to pre-prediction. The specific implementation is as follows: The parameter calculation and mapping unit 2 inputs the static parameters into the preset contact suspension dynamic performance prediction model. The preset here means that the model needs to be trained and verified before system deployment, rather than being constructed in real time. The core function of the model is to receive the standardized static parameter set output by the division evaluation collection unit, and output the dynamic performance prediction result through the built-in calculation logic. Before input, the static parameters need to be format adapted to ensure that the parameter dimension, unit and input features during model training are completely consistent, avoiding prediction failure due to format deviation. At the same time, the data checking module is used to eliminate abnormal values beyond the reasonable range of engineering, so as to input pure and reliable basic data to the model. The establishment method of the contact suspension dynamic performance prediction model is as follows: The establishment of the model follows the logic of data-driven-model training-engineering verification, relying on massive historical data and engineering experience to ensure that the model prediction results not only conform to the algorithm logic, but also fit the actual operation rules of the catenary. The specific steps are as follows: Based on the massive samples in the historical maintenance database, the historical maintenance database integrates the catenary construction and operation data of different types of railway lines in different climate environments in China in the past 10 years, containing complete samples of 5000+ catenaries. Each sample is labeled with basic information such as line mileage pile number, construction time and environmental temperature, ensuring the diversity and representativeness of the samples and avoiding the model adapting to only a single scenario. Extreme abnormal working conditions need to be excluded during sample screening, and data under normal construction and operation conditions need to be retained to provide high-quality basis for model training. The measured data of static height, static pull-out value, contact wire tension, dropper length and positioner slope at each measuring point are extracted as input features, and the extraction process needs to go through multiple rounds of data preprocessing to ensure feature effectiveness: For static height (reflecting the vertical position reference value of the conductor, unit: mm) and static pull-out value (reflecting the horizontal offset reference value of the conductor, unit: mm), 3σ criterion is used to remove abnormal values (data with static height deviation of more than ±50mm and pull-out value deviation of more than ±30mm) that exceed the design specification range, and then moving average filtering (window size: 5 measuring points) is used to smooth the measurement noise. For contact wire tension (representing the mechanical stress state of the suspension system, unit: kN), it needs to be compensated and corrected combined with the environmental temperature at the time of collection, the correction formula is corrected tension = measured tension + 0.02 × (design reference temperature - collection temperature) (0.02 is the tension temperature compensation coefficient, and the design reference temperature is usually 20℃), which eliminates the tension measurement deviation caused by temperature change; For dropper length (indicating the geometric relationship between suspension points, unit: mm) and positioner slope (describing the spatial angle between support structure and conductor, unit: °), through the original collection data of high-precision laser range finder and inclinometer, after removing the zero drift error (zero point calibration once every 100 measuring points), the standardized parameters are formed.Finally, each measurement point generates a 5-dimensional static parameter vector (static height, static pull-out value, contact line tension, dropper length, positioner slope), and the parameter vectors of all measurement points in a unit section are arranged in order of mileage to form an input feature matrix with dimensions of measurement point number x 5, which serves as the input basis for model training. The elastic uniformity index and offline rate index obtained in the dynamic test are simultaneously associated as training labels. The dynamic test refers to the performance test of the standard pantograph simulating the actual running state of the train (speed covering 80%-120% of the designed maximum speed of the line, such as 280-350 km / h for high-speed railways) after the completion of the overhead contact system. During the test, the contact pressure data is collected by the three-dimensional force sensor installed on the pantograph head, and the current data is collected by the overhead current sensor. The elastic uniformity index is calculated according to the coefficient of variation of the contact pressure in the section, and the formula is elastic uniformity = contact pressure standard deviation / contact pressure mean x 100%. The smaller the coefficient of variation, the more uniform the elastic distribution of the overhead contact system under dynamic load, and the more stable the contact between the pantograph and the contact wire. The offline rate index is calculated according to the proportion of the duration of the interruption of the current between the pantograph and the contact wire, and the formula is offline rate = offline duration / total test duration x 100%. The lower the offline rate, the better the current collection stability. When associating, the line mileage pile number (such as K100+200 to K100+500) is used to ensure that the static parameter collection section and the dynamic test section are completely consistent, and the time interval between the two is not more than 24 hours (to avoid natural changes in the overhead contact system parameters caused by wind, rain, and temperature changes). Finally, a complete sample pair of the input feature matrix-elastic uniformity-offline rate is formed as the label data for model training, ensuring that the model learning goal is completely aligned with the actual engineering requirements. A deep neural network architecture is used, which is designed for the nonlinear correlation characteristics of the static and dynamic parameters of the overhead contact system. The architecture uses a fully connected network structure with 3 hidden layers: the number of input layer neurons matches the input feature dimension (5 neurons corresponding to 5 static parameters), ensuring that the original features can be completely input into the model; the first hidden layer has 64 neurons, which preliminarily extracts the basic features of a single parameter and the simple correlation between parameters through linear transformation; the second hidden layer has 32 neurons, which introduces the ReLU activation function to mine the nonlinear interaction features between parameters; the third hidden layer has 16 neurons, which further integrates high-order features; and the output layer has 2 neurons, which outputs the predicted values of the elastic uniformity index and the offline rate index, respectively.Meanwhile, Dropout layers (dropout rate = 0.2) are added between every two layers of the network to prevent overfitting by randomly discarding some neurons, and BatchNorm layers are added before the output layer to accelerate model training convergence and ensure that the architecture can capture complex correlations and have generalization ability in engineering scenarios. The network weights are iteratively optimized using the backpropagation algorithm, with the goal of minimizing the deviation between predicted values and true labels. The mean squared error (MSE) is used as the loss function, with the formula Loss = 1 / n x Σ (model predicted value - dynamic test true value) 2 (n is the number of samples in a single batch), which accurately quantifies the model prediction error. The Adam optimizer is selected (the initial learning rate is set to 1e-4, and it is reduced by 10% every 100 iterations), and the gradients of the loss function with respect to each layer's weights are calculated through backpropagation. The weights are updated in the direction of the gradient descent to balance optimization speed and accuracy. During training, samples are input in batches (batch size = 32), and the total training rounds are set to 500. At the same time, 20% of the samples are set aside as a validation set, and the validation set loss is calculated after each training round. If the validation loss does not decrease for 20 consecutive rounds (with a fluctuation range of ≤0.001), the early stopping mechanism is triggered to prevent overfitting to the training data. In addition, the signs and sizes of the weights corresponding to each parameter need to be monitored in real time during the training process. For example, an increase in the length of the suspension should correspond to an increase in the uniformity of elasticity (the weight sign is positive), and an increase in the contact wire tension should correspond to a decrease in elasticity (the weight sign is negative). If the weight sign contradicts the engineering logic, the data preprocessing stage needs to be reviewed to identify abnormal samples (such as measurement errors or distorted dynamic test data), and the model weight updates need to be ensured to comply with the mechanics of the overhead contact system. The end-to-end non-linear mapping relationship from the static parameter set to the dynamic performance indicator is constructed, which means that no manual feature engineering is required, and the model can automatically learn the complete correlation path from the static parameters to the dynamic performance. For example, when the contact wire tension of an evaluation unit segment increases from 28 kN to 30 kN and the length of the suspension decreases from 1200 mm to 1150 mm, the model can capture the dual effects of tension increase leading to increased rigidity of the suspension system and suspension shortening leading to reduced spacing between suspension points, resulting in a decrease in the uniformity of elasticity. The model outputs the predicted value of the overall uniformity of elasticity through a non-linear transformation, avoiding the limitations of traditional linear models that cannot depict the synergistic effects of multiple parameters. The effectiveness of the mapping relationship is verified by the coefficient of determination R². When the R² of the training set is ≥0.85 (indicating that the model has strong explanatory power for the training data) and the R² of the validation set is ≥0.8 (indicating that the model has reliable prediction ability for new data), the mapping relationship is determined to meet engineering requirements. Otherwise, additional samples (such as those from high-cold regions) or adjustments to the network architecture (such as increasing the number of hidden layer neurons) are needed until the requirements are met.Finally, a calculation model is generated that can predict the dynamic elastic distribution characteristics according to real-time static parameters. The model predicts the output of the key dynamic performance estimation indicators of the evaluation unit section, including the unit section average elasticity and the unit section maximum elasticity difference. After the model is deployed, the real-time static parameters input by the parameter calculation and mapping unit are received (after the same preprocessing process as the training data). The local elasticity prediction value of each measurement point in the evaluation unit section is output first. This value is based on the elasticity uniformity index output by the model, combined with the weight contribution degree of the static parameters of each measurement point. The formula is local elasticity = line design reference elasticity x (1 + elasticity uniformity x parameter contribution degree) (for example, if the high-speed railway design reference elasticity is 80 N / mm and the parameter contribution degree of a measurement point is 0.05, then local elasticity = 80 x (1 + 0.08 x 0.05) = 80.32 N / mm). The key dynamic indicators are calculated: the unit section average elasticity is the arithmetic average of the local elasticity of all measurement points, the formula is average elasticity = 1 / m x Σ (local elasticity of each measurement point) (m is the number of measurement points in the evaluation unit section), which reflects the overall level of the dynamic performance of the catenary. The unit section maximum elasticity difference is the maximum value of the absolute value of the local elasticity difference of all adjacent measurement points. For example, the local elasticity of adjacent measurement points is 79.5 N / mm and 85.2 N / mm, respectively, and the absolute value of the difference is 5.7 N / mm. If it is the maximum value of the whole section, it is the maximum elasticity difference, which reflects the local mutation risk of the elastic distribution of the catenary. These two indicators provide quantitative basis for the subsequent performance prediction and strategy adjustment unit, ensuring that the dynamic performance evaluation is transferred from qualitative judgment to quantitative analysis, and forming a complete logical closed loop with the static parameter collection in the previous section and the performance judgment in the subsequent section.

[0016] After the dynamic performance prediction model of the contact suspension is trained (the weights are optimized by the back propagation algorithm to construct a nonlinear mapping between static parameters and dynamic indicators) and verified by engineering (the R² of the training set and the verification set is ≥0.85 and ≥0.8, respectively), the real-time static parameters obtained by the division evaluation and collection unit need to be converted into key dynamic indicators that can be directly used for performance judgment. This process is the core function landing link of the parameter calculation and mapping unit, which not only inherits the training results of the model, but also provides quantitative basis for the subsequent performance prediction and strategy adjustment unit, ensuring that the dynamic performance evaluation is transferred from an abstract model to specific indicators. The specific implementation is as follows: The technical means for predicting the output of key dynamic performance estimation indicators is as follows: This technical means is based on the logic of first obtaining a single-point local elasticity and then aggregating and calculating global indicators. Through step-by-step processing, accurate conversion from static parameters to dynamic indicators is achieved, avoiding the omission of local risks caused by directly outputting global indicators. The specific steps are as follows: In the first step, the static parameter set of all measurement points in the current evaluation unit section is input into the trained prediction model. The static parameter set of all measurement points in the current evaluation unit section needs to be preprocessed in the same way as in the model training stage to ensure that the data format and feature distribution are completely matched. First, the original static parameters of all measurement points in the evaluation collection unit are retrieved from the database, including static height, static pull-out value, contact wire tension, dropper length, and positioner slope. Then, the data cleaning module is used to remove abnormal values, such as static height deviation exceeding the design value ± 50 mm and tension fluctuation exceeding ± 2 kN. The effective data is standardized by scaling the mean and standard deviation of the training set. Finally, the preprocessed parameters are arranged in the order of the measurement point mileage (from the start to the end of the unit section), forming an input feature matrix with dimensions of measurement point quantity x 5. This matrix is passed through the input layer of the model to ensure that the model can capture the parameter variation pattern in the spatial order rather than a disordered parameter stack. The model outputs the local elastic prediction value of each measurement point in the evaluation unit section. The local elastic prediction value is the elastic response value of the contact suspension under dynamic load (such as the pantograph sliding at the design speed), with a unit of N / mm, which directly reflects the contact pressure stability between the contact wire and the pantograph. The closer the value is to the design reference elasticity of the line (such as 80-100 N / mm for high-speed railways), the more stable the contact pressure. During the model output process, the input feature matrix is first transformed by a 3-layer hidden layer (64→32→16 neurons) to explore the coupling relationship between static parameters (such as the cooperative effect of dropper length shortening and contact wire tension increase on elasticity). Then, the high-order features are mapped to the local elastic prediction value of each measurement point through the linear activation function of the output layer (since the elastic prediction is a regression task, there is no need for ReLU non-linear activation). At the same time, the model outputs the confidence of the prediction value based on the error distribution during training (such as a confidence of 95% indicating that the prediction value deviates from the true dynamic elasticity by ≤5%). If the confidence is less than 80%, a data re-sampling instruction is triggered to ensure the reliability of the output value. For example, in a certain evaluation unit section with 5 measurement points, the model outputs local elastic prediction values of 82 N / mm, 81 N / mm, 85 N / mm, 83 N / mm, and 82 N / mm, with a confidence of ≥90%. This indicates that the elastic prediction results of each point are stable and reliable.

[0017] Second, after obtaining the local elastic prediction value of all measurement points, further aggregation calculation of key dynamic indicators is required, that is, based on the local elastic prediction value sequence of all measurement points, the arithmetic mean value is calculated as the average elasticity of the unit segment. The local elastic prediction value sequence is a set of prediction values arranged in the order of the mileage of the measurement points (such as the sequence of the above five measurement points [82, 81, 85, 83, 82]), and the arithmetic mean value is calculated because the contribution weight of each measurement point in the unit segment to the overall dynamic performance is consistent (all are representative measurement points covering different positions in the unit segment). The calculation process is unit segment average elasticity = (sum of local elastic prediction values of all measurement points) / number of measurement points. For example, the average value of the above sequence is (82+81+85+83+82) / 5=82.6N / mm. This indicator reflects the overall level of the dynamic performance of the evaluation unit segment. If the design reference elasticity of the line is 80-85N / mm, then 82.6N / mm is within the reasonable range, indicating that the overall elasticity meets the design expectation. If the average value is 75N / mm (lower than the lower limit), it means that the overall elasticity is too soft, which may lead to insufficient pantograph contact pressure. If it is 90N / mm (higher than the upper limit), the overall elasticity is too stiff, which may exacerbate pantograph wear. This result will be directly used as the core basis for the mean compliance judgment in the subsequent performance prediction, and at the same time, to capture the risk of local performance mutation, the difference of local elastic prediction value between all adjacent measurement points needs to be calculated. Adjacent measurement points refer to two measurement points that are adjacent in the order of mileage (such as K100+200 and K100+250, K100+250 and K100+300). The traversal process is performed in the order of mileage from the starting point to the ending point. For each group of adjacent points, the absolute difference of the local elastic prediction value is calculated. The formula is adjacent point elasticity difference = |current measurement point prediction value - previous measurement point prediction value|. For example, in the above sequence, the first group of adjacent points difference = |81-82|=1N / mm, the second group = |85-81|=4N / mm, the third group = |83-85|=2N / mm, and the fourth group = |82-83|=1N / mm. The difference between adjacent points is selected because the pantograph slides continuously through each measurement point on the catenary, and a sudden change in the elasticity of adjacent points will directly lead to a sudden rise or fall in the contact pressure, causing off-line sparks or excessive wear. The difference between non-adjacent points has less effect on current stability, so focusing on adjacent points can accurately locate local risks, and the maximum absolute difference value is selected as the maximum elasticity difference of the unit segment. Among all the adjacent point elasticity differences obtained by traversal, the maximum value is selected by comparison. For example, in the above sequence, the maximum difference is 4N / mm (the second group of adjacent points), which is the maximum elasticity difference of the evaluation unit segment.The index quantifies the local discrete degree difference of the elastic distribution of the contact suspension. The smaller the difference, the more uniform the elastic distribution, and the smaller the pantograph contact pressure fluctuation. The larger the difference, the more risk points of local elastic mutation exist (such as the K100+250 and K100+300 sections corresponding to the second group of adjacent points described above), which need to be focused on in subsequent adjustments of the static parameters of the region to avoid affecting the pantograph-catenary current collection stability due to elastic mutation. In this way, the uniformity and discrete degree of the contact suspension under dynamic load are quantified, where the unit segment average elasticity mainly quantifies the overall uniformity value closer to the design benchmark and less fluctuation, indicating that the overall elastic response of the contact suspension under dynamic load is more consistent, and the pantograph can obtain stable contact pressure. The unit segment maximum elastic difference quantifies the local discrete degree difference, and the local elastic transition is smoother without obvious performance mutation. The combination of the two covers both the overall performance level and the local risk, forming a complete quantitative description of the dynamic performance of the contact suspension, providing direct and accurate index support for subsequent performance prediction and strategy adjustment unit judgment of compliance and controllable dispersion, ensuring the logical closed loop of the entire parameter calculation and mapping process, and fully meeting the actual engineering needs of railway catenary construction monitoring. After outputting the core dynamic index of the unit segment average elasticity (such as 82.6 N / mm) in the parameter calculation and mapping unit, the performance prediction and strategy adjustment unit 3 enters the key performance judgment link. The primary task is to compare the average elasticity with the preset standard elasticity range. This comparison is the basis for judging whether the overall dynamic performance of the catenary meets the standard. Because the average elasticity directly reflects the overall response level of the contact suspension under dynamic load, if it deviates from the standard range, it will directly affect the contact stability of the pantograph and the contact wire (such as increased wear due to excessive stiffness, and off-line sparks due to excessive flexibility). Therefore, through standardized comparison operations, the judgment result is ensured to meet the actual needs of railway catenary design and operation. The specific implementation is as follows. The performance prediction and strategy adjustment unit 3 compares the unit segment average elasticity with the preset standard elasticity range. The unit segment average elasticity is the arithmetic mean of the local elasticity of each measurement point output by the model (with a confidence level of 90% or more to ensure data reliability), while the preset standard elasticity range is not a fixed value, but a range of elastic fluctuations configured by the system in advance according to the catenary design specifications and line operation level. Its core logic is that different speed levels and different operation scenarios of the line have different requirements for the elasticity of the catenary (such as high-speed railways requiring more stable elasticity to ensure high-speed current collection). Therefore, before comparison, the interval matching the current evaluation unit segment needs to be called, and then the numerical comparison is performed to avoid one-size-fits-all judgment errors. The operation of comparing the evaluation unit segment average elasticity with the preset standard elasticity range is as follows: This operation is based on the process of interval calling-numerical comparison-state marking-warning triggering. Each step needs to be combined with engineering specifications and system logic to ensure accurate and traceable judgment. The specific steps are as follows: The first step is to call the elastic allowable fluctuation interval preset in the system. The elastic allowable fluctuation interval is structured data stored in the system parameter configuration module, existing in the form of key-value pairs of line operation level-interval upper and lower limits-applicable specification number (such as 350 km / h high-speed railway-80-85 N / mm-TB / T3558-2020). The calling process needs to achieve accurate matching through the evaluation of the basic attributes of the unit section: the system first reads the milepost number (such as K100+200 to K100+500) and the line operation level label (which has been recorded at system initialization, such as 350 km / h high-speed railway 200 km / h passenger and freight mixed line) of the current evaluation unit section, and then retrieves the elastic allowable fluctuation interval corresponding to the operation level from the parameter configuration module through an SQL query statement. For example, if the current unit section is a 350 km / h high-speed railway, the interval retrieved is 80-85 N / mm, and if it is a 200 km / h passenger and freight mixed line, the interval retrieved is 75-82 N / mm. At the same time, the system will record the interval information (including the specification number) of this call, providing a basis for tracing the source of the subsequent judgment results and avoiding false positives due to interval call errors.

[0018] The second step is to dynamically set the interval according to the catenary design specification and the line operation level. The catenary design specification mainly refers to national and industry standards such as TB / T3558-2020 "Railway Catenary Operation and Maintenance Rules" and TB / T3113-2017 "Railway Catenary Design Specification". The specification clearly specifies the elastic requirements of catenary for different operation levels (such as the elastic of 350 km / h line catenary should be controlled within 80-85 N / mm, and the elastic difference should not exceed 5 N / mm); the line operation level is divided into three levels according to the maximum running speed of the train: 300-350 km / h high-speed railway, 200-250 km / h fast railway, and 120-160 km / h general speed railway. The higher the level, the stricter the requirement for elastic stability, so the range of the elastic allowable fluctuation interval is narrower (such as the interval of general speed railway is 70-80 N / mm, with a width of 10 N / mm; the interval width of high-speed railway is only 5 N / mm). The implementation logic of dynamic setting is: when a new line is connected to the system, the staff only needs to input the line operation level, and the system will automatically match the elastic requirements in the corresponding design specification, generate the elastic allowable fluctuation interval and store it in the parameter configuration module, without manual calculation, which reduces human error and ensures that the interval setting always meets the latest specification requirements.

[0019] The third step involves marking the average elasticity of the evaluation unit segment as compliant when it falls within the allowable elasticity fluctuation range. Compliance means that the overall dynamic performance of the evaluation unit segment meets design expectations, the contact suspension exhibits uniform elastic response under dynamic loads, and the pantograph receives stable contact pressure (avoiding sudden pressure increases and decreases). Specifically, the system assigns the compliant value to the average status field in the performance status database of the evaluation unit segment and simultaneously records the current average elasticity value (e.g., 82.6 N / mm), the matching allowable elasticity fluctuation range (80–85 N / mm), and the judgment timestamp, forming a complete judgment log. Simultaneously, the system marks the mileage range of the unit segment in green (compliance indicator color) in the visualization interface, facilitating quick identification of compliant sections by staff. For example, if the average elasticity of a 350 km / h high-speed railway unit segment is 83.2 N / mm, falling within the 80–85 N / mm range, the system automatically marks it as compliant, indicating that the overall dynamic performance of the contact network in this segment meets the standard, and no adjustment to the overall elasticity is required.

[0020] The fourth step involves determining overall over-rigidity if the range exceeds the upper limit. Overall over-rigidity refers to excessive rigidity of the contact suspension, which causes the contact pressure when the pantograph passes through to be significantly higher than the design value (e.g., design pressure 80N, actual pressure reaching 100N). Long-term operation will exacerbate wear on the pantograph's carbon contact plate and contact wire, and may even lead to wire breakage due to excessive contact wire stress. Once determined as overall over-rigidity, the system assigns the average status field to "over-rigidity" and calculates the excess range (e.g., range upper limit 85N / mm, actual average elasticity 88.5N / mm, excess range 3.5N / mm). Simultaneously, it notes the risk of excessive contact pressure in the log, and the mileage range of this unit segment is marked in red (high-risk indicator) in the visualization interface, providing a clear indication to staff that it needs priority handling. Furthermore, the system automatically links the static parameter data collected for this unit segment, initially filtering parameters that may lead to over-rigidity (e.g., excessively high contact wire tension, excessively short dropper length), providing directional reference for subsequent static parameter adjustments.

[0021] Fifth, if the value is below the lower limit, it is judged as "overall too flexible." Overall too flexible refers to excessive elasticity of the contact suspension, which will cause the contact pressure between the pantograph and the contact wire to be lower than the design value (e.g., design pressure 80N, actual pressure only 60N). This easily leads to pantograph disconnection (current interruption), generating disconnection sparks. This not only affects the train's current collection stability but may also burn the contact wire surface, reducing the contact wire's service life. After being judged as "overall too flexible," the system assigns the average status field to "overall too flexible," calculates the insufficient range (e.g., lower limit of 80N / mm, actual average elasticity 76.2N / mm, insufficient range 3.8N / mm), and logs note the disconnection risk; in the visualization interface, the mileage interval of this unit section is marked in yellow (medium risk indicator color), prompting staff to make timely adjustments. Similar to the overall over-rigidity warning, the system initially correlates static parameters (such as low contact wire tension or excessive dropper length) to provide clues for subsequent adjustments and automatically triggers different types of warning signals. The type of warning signal corresponds one-to-one with the average status: no warning is triggered when the average is compliant; overall over-rigidity triggers a red high-risk warning, with the warning information including unit segment mileage, over-rigidity amplitude, and preliminary analysis results of correlated static parameters, sent simultaneously via system pop-ups, mobile app push notifications for maintenance personnel, and on-site audible and visual alarms to ensure staff receive the information immediately; overall over-flexibility triggers a yellow medium-risk warning, with the information including unit segment mileage, over-flexibility amplitude, and offline risk alerts, sent via system pop-ups and app push notifications. Each warning signal comes with a processing priority (red warning priority 1, requiring a response within 2 hours; yellow warning priority 2, requiring a response within 6 hours), helping staff rationally arrange maintenance procedures and avoid delays in handling critical issues due to disorganized warning information. This differentiated early warning system ensures timely risk transmission while avoiding excessive warnings that could disrupt normal operations. It lays the foundation for the precise intervention of the subsequent static parameter adjustment unit and forms a complete performance judgment loop with the dynamic index calculation mentioned earlier and the parameter adjustment mentioned later.

[0022] After comparing the average elasticity of the evaluation unit segment with the standard elasticity range and clarifying the overall dynamic performance status (such as average compliance, overall excessive stiffness or excessive flexibility), the performance prediction and strategy adjustment unit 3 also needs to simultaneously compare the maximum elasticity difference of the unit segment with the preset maximum allowable elasticity difference. This comparison focuses on the local dispersion of the contact wire elasticity distribution, making up for the limitation that the average elasticity only reflects the overall level. Even if the average elasticity meets the standard, if the elasticity change of adjacent measurement points is too large, it will still cause a sudden change in contact pressure when the pantograph passes by (such as a sudden increase from 80N to 100N), causing offline sparks or excessive wear of the contact wire. Therefore, it is necessary to accurately identify local risks through standardized operations to ensure that the dynamic performance of the contact wire is both overall compliant and locally uniform, providing a complete basis for subsequent performance compliance judgment. The specific implementation method is as follows: Simultaneously, the maximum elastic difference of a unit segment is compared with the preset maximum allowable elastic difference. Here, the maximum elastic difference of a unit segment is a quantitative indicator obtained earlier by iterating through the local elasticity prediction values ​​of adjacent measurement points and taking the maximum absolute difference (e.g., the maximum elastic difference of a 350km / h high-speed railway unit segment is 6.8N / mm). The preset maximum allowable elastic difference is essentially an elasticity difference tolerance threshold set based on the line's operational characteristics. Its core logic is: different lines have different tolerances to elasticity abrupt changes (high-speed lines have faster pantograph operating speeds and are more sensitive to elasticity abrupt changes). Therefore, a threshold must first be set based on the line attributes, and then the local elasticity distribution must be judged through numerical comparison to avoid the risk of missed judgments on high-speed lines and over-judgments on low-speed lines due to a uniform threshold. The specific operation of comparing the maximum elastic difference of a unit segment with the preset maximum allowable elastic difference is as follows: This operation revolves around a process of threshold setting, numerical comparison, and risk identification. Each step is combined with the line operation requirements and engineering specifications to ensure that the local risk assessment is accurate and feasible. The specific steps are as follows: The first step is to set a tolerance threshold for elasticity differences based on the maximum operating speed of the line and the stability requirements of pantograph-catenary current collection. The maximum operating speed of the line is the maximum design speed of the train corresponding to the current evaluation unit section (such as 350km / h high-speed railway, 200km / h passenger and freight railway, and 120km / h conventional railway). This parameter is bound to the mileage marker during system initialization and can be directly retrieved from the line basic database. The stability requirements of pantograph-catenary current collection refer to the industry standard TB / T3558-2020 "Rules for Operation and Maintenance of Railway Contact System". The standard clearly requires that the higher the line speed, the smaller the range of pantograph-catenary contact pressure fluctuation should be (such as ≤±15N for 350km / h line and ≤±20N for 200km / h line). Elasticity difference is the core factor affecting contact pressure fluctuation (for every 1N / mm increase in elasticity difference, the contact pressure fluctuation increases by about 3-5N). Therefore, a threshold needs to be set based on the correlation between the two. For specific settings, a speed-threshold correspondence table is adopted: for high-speed railways with speeds of 300–350 km / h, the elasticity difference tolerance threshold is set at 5 N / mm (ensuring contact pressure fluctuations ≤25 N, meeting high-speed current collection requirements); for express railways with speeds of 200–250 km / h, the threshold is set at 8 N / mm; and for conventional railways with speeds of 120–160 km / h, the threshold is set at 10 N / mm. Additionally, if the line passes through complex terrain (such as windy areas or tunnel complexes), the threshold will be further reduced by 10%–20% from the base threshold (e.g., the threshold is reduced to 4.5 N / mm in windy areas at 350 km / h) to address the challenges of pantograph-catenary stability in harsh environments, ensuring that the threshold settings both comply with general specifications and are adaptable to special scenarios.

[0023] The second step involves marking the unit segment as discretely controllable when the maximum elasticity difference does not exceed the threshold. Discretely controllable indicates that the elasticity transition between adjacent measurement points within the unit segment is smooth with no significant abrupt changes, and the contact pressure fluctuation when the pantograph passes through can be controlled within a safe range (e.g., line fluctuation ≤ 25N at 350km / h), achieving local dynamic performance standards. During the marking operation, the system assigns the discrete state field of the unit segment to controllable in the performance status database and simultaneously records key information: elasticity difference tolerance threshold (e.g., 5N / mm), actual maximum elasticity difference of the unit segment (e.g., 4.2N / mm), and judgment timestamp, forming a traceable judgment log. In the visual monitoring interface, the mileage interval of the unit segment is marked in light green (distinguished from the dark green of mean compliance, intuitively reflecting the dual compliance status of overall compliance and local controllability). At the same time, the elasticity difference (e.g., K100+250-K100+300: 4.2N / mm) is marked on the connecting line of all adjacent measurement point pairs within the interval, facilitating staff to review the local elasticity distribution. For example, in a 200km / h passenger and freight co-line unit section, the elasticity difference tolerance threshold is 8N / mm, and the actual maximum elasticity difference is 7.5N / mm, which does not exceed the threshold. The system automatically marks it as discretely controllable, indicating that the catenary in this section not only meets the overall elasticity standard, but also meets the requirements for local elasticity transition.

[0024] The third step involves identifying the location of adjacent measurement point pairs where the maximum elasticity difference occurs if the threshold is exceeded. An adjacent measurement point pair refers to two representative measurement points that are immediately adjacent in mileage order (e.g., measurement point 3 corresponds to mileage K100+250, and measurement point 4 corresponds to K100+300, forming an adjacent measurement point pair). This identification process relies on the previously generated sequence of local elasticity prediction values ​​(elasticity values ​​of each measurement point arranged in mileage order): the system first iterates through the absolute differences of all adjacent elements in the sequence, locating the index of the element with the largest difference (e.g., the largest difference between the 3rd and 4th elements is indexed as 3-4); then, based on the index, it retrieves the mileage marker of the corresponding measurement point (from the measurement point...). The system retrieves the mileage K100+250 from index 3 and the mileage K100+300 from index 4 from the database, determining the location of the adjacent measurement point pair where the maximum elasticity difference occurs as K100+250-K100+300. Simultaneously, the system automatically associates the static parameters of these two measurement points (such as the dropper length of 1180mm and the contact wire tension of 29kN at measurement point 3, and the dropper length of 1250mm and the contact wire tension of 26kN at measurement point 4), and preliminarily analyzes the possible causes of the excessive elasticity difference (such as a dropper length difference of 70mm and a tension difference of 3kN), providing clues for the subsequent static parameter adjustment unit to lock in key adjustment points and avoiding blind investigation.

[0025] The fourth step is to mark these as local mutation risk points. Local mutation risk points refer to specific mileage intervals (such as K100+250-K100+300) where the elasticity difference between adjacent measurement points exceeds the threshold, which may cause a sudden change in pantograph contact pressure. If these risk points are not addressed, they may cause instantaneous excessively high or low contact pressure when a train passes by (e.g., an elasticity difference of 6.8 N / mm exceeds the 5 N / mm threshold for a 350 km / h line, and the contact pressure may suddenly rise from 80 N to 110 N), exacerbating pantograph-catenary wear or generating offline sparks. During the marking process, the system assigns the discrete state field of the unit segment to the risk of local mutation, notes the risk of sudden change in pantograph-catenary contact pressure in the judgment log, and calculates the deviation range (e.g., actual difference 6.8N / mm - threshold 5N / mm = 1.8N / mm deviation). In the visualization interface, the mileage range of the risk point is marked in orange (the color of medium risk, which is different from the red of overall over-rigidity) and highlighted with a flashing effect. At the same time, the deviation range and the difference of the associated static parameters (e.g., dropper difference 70mm, tension difference 3kN) are marked next to the range, allowing staff to quickly locate the core risk.

[0026] The fifth step involves generating risk location information containing location codes. These location codes are unique identifiers generated by the system for risk points, with a format of line level-mileage interval-risk type (e.g., 350km / h-K100+250-K100+300-local elastic mutation). This format includes both basic location information and clearly defines the risk attribute, facilitating subsequent data retrieval and traceability. In addition to location codes, the risk location information also integrates key technical parameters: elasticity difference tolerance threshold, actual maximum elasticity difference, deviation range, static parameters of adjacent measurement points where the risk occurred (e.g., static height, pull-out value, dropper length, tension), corresponding local elasticity prediction values, and recommended preliminary investigation directions (e.g., prioritizing checks on dropper length and contact wire tension differences). This information is automatically stored in the system risk warning database and simultaneously pushed to the monitoring terminal via pop-up window and sent to the mobile phones of on-site maintenance personnel via SMS (including location code and navigation link, which can directly navigate to the risk point). This ensures that staff can quickly obtain the risk location and core data without secondary queries, laying the foundation for accurate intervention in subsequent static parameter adjustments. At the same time, it forms a closed loop of overall-local collaborative performance prediction with the average elasticity judgment mentioned above and the optimization adjustment plan mentioned below.

[0027] After comparing the average elasticity of the evaluation unit segment with the standard elasticity range (marked as mean compliance, overall excessive stiffness, or overall excessive flexibility), and comparing the maximum elasticity difference of the unit segment with the maximum allowable elasticity difference (marked as discrete controllable or local abrupt change risk point), the performance prediction and strategy adjustment unit 3 needs to perform the final dynamic performance prediction compliance judgment based on the judgment results of these two core indicators. This judgment is the key node connecting performance analysis and parameter adjustment. It not only inherits the dual evaluation results of the overall elasticity level and local elasticity dispersion mentioned above, but also distinguishes between compliance and non-compliance states through clear logical rules. This ensures that only unit segments that simultaneously meet the requirements of overall compliance and local uniformity can enter the subsequent construction process, while unit segments with problems are accurately transferred to the static parameter adjustment unit 4 to avoid overlooking dynamic performance risks caused by the compliance of a single indicator. The specific implementation method is as follows: The system determines whether the dynamic performance prediction of the evaluation unit segment meets the standard. If it does not, it proceeds to static parameter adjustment unit 4. This determination is not an isolated operation, but a comprehensive assessment of the two comparison results mentioned above: the system first retrieves the mean status (recording mean compliance, overall over-rigidity, or overall over-flexibility) and discrete status (recording discrete controllable or local mutation risk points) of the current evaluation unit segment from the performance status database, and then performs a combined judgment based on preset logical rules. If the judgment result is satisfactory, the dynamic performance of the evaluation unit segment meets the design requirements, and the system generates a compliance confirmation report, which is synchronized. Upon reaching the construction progress management module, the system allows entry into the next construction phase (such as the installation of overhead contact line accessories). If the judgment result is substandard, it indicates that the dynamic performance of this unit segment poses a risk to the stability of the pantograph-catenary current collection (such as excessive overall rigidity leading to wear, or local abrupt changes causing offline). The system will automatically trigger a process jump instruction to synchronize the basic information of this unit segment (mileage station, static parameter set, judgment result) to the static parameter adjustment unit 4, providing complete data support for subsequent precise adjustments and preventing substandard unit segments from flowing into subsequent phases and causing rework. The logic for determining whether the dynamic performance prediction of the evaluation unit segment meets the standard is as follows: The core of this logic is the dual-standard principle: the dynamic performance of the overhead contact system must simultaneously meet both overall horizontal compliance and local uniform distribution; neither can be neglected. This logic stems from the actual operational needs of railway overhead contact systems: if only the overall average is compliant but local elastic abrupt changes are too large, the contact pressure will still change drastically when the pantograph passes through the abrupt change area; if only local discreteness is controllable but the overall average deviates from the standard, long-term operation will still lead to pantograph-catenary wear or disconnection problems. Therefore, it is necessary to ensure the comprehensive reliability of dynamic performance through dual compliance. The specific judgment process revolves around compliance and non-compliance conditions, and each step needs to combine engineering specifications and system data to achieve automated judgment.

[0028] The dynamic performance prediction is considered satisfactory only when both the mean compliance flag and the discrete controllability flag are simultaneously met. Meeting both flags simultaneously means the system needs to confirm that both flags are in a qualified state. The system reads the mean status field; if this field value is mean compliance (i.e., the average elasticity is within the allowable elasticity fluctuation range, such as the average elasticity of a 350km / h high-speed railway unit segment being 82.6N / mm within the 80-85N / mm range), then the overall elasticity level meets the standard. The system then reads the discrete status field; if this field value is discrete controllability (i.e., the maximum elasticity difference does not exceed the elasticity difference tolerance threshold, such as the maximum elasticity difference within the same unit segment), then the system is considered to have met the standard. If the elasticity difference is 4.2 N / mm and does not exceed the 5 N / mm threshold, then the local elasticity distribution meets the standard. Only when both conditions are met simultaneously will the system assign the dynamic performance compliance status field a "compliant" value in the performance status database and generate a compliance record. This record includes the mileage of the evaluation unit segment (e.g., K100+200-K100+500), the average elasticity value and threshold range, the maximum elasticity difference value and threshold range, and the judgment timestamp. It also associates the design specification number used in this evaluation (e.g., TB / T3558-2020) to provide a basis for subsequent acceptance and traceability. In the visual monitoring interface, the mileage range of the compliant unit segment will be filled in dark green and overlaid with a double compliance icon (composed of a checkmark for mean compliance and a checkmark for discrete controllability) for easy identification by staff. Simultaneously, the system will automatically push the compliance information to the construction management platform, updating the construction status of the unit segment to "dynamic performance acceptance passed," allowing subsequent procedures to proceed.

[0029] If any of the following states occur: overall excessive stiffness, overall excessive flexibility, or local abrupt change risk points, the system is deemed non-compliant. Any state indicates that as long as one of the three negative states exists, the dynamic performance is deemed non-compliant. This logic aims to maximize the coverage of potential risks. If the mean state is overall excessive stiffness (e.g., average elasticity of 88.5 N / mm exceeding the upper limit of 85 N / mm), even if the discrete states are controllable, the excessive overall stiffness will easily aggravate the wear of the pantograph and catenary, and it is deemed non-compliant. If the mean state is overall excessive flexibility (e.g., average elasticity of 76.2 N / mm below the lower limit of 80 N / mm), even if there are no local abrupt changes, the insufficient overall elasticity will easily cause derailment, and it is deemed non-compliant. If the mean state is compliant but the discrete state is a local abrupt change risk point (e.g., the maximum elasticity difference of 6.8 N / mm exceeds the 5 N / mm threshold, corresponding to mileage K100+250-K100+300), the sudden change in local elasticity will cause fluctuations in contact pressure, and it is also deemed non-compliant. After the system determines that the performance is substandard, it will assign a "substandard" value to the "Dynamic Performance Compliance Status" field and record the specific status in detail in the "Substandard Reason" field (e.g., overall excessive stiffness: average elasticity 88.5 N / mm exceeds 85 N / mm upper limit by 3.5 N / mm; local abrupt change risk point: K100+250-K100+300 interval difference 6.8 N / mm exceeds 5 N / mm threshold by 1.8 N / mm). Simultaneously, the substandard data will be marked as needing adjustment to avoid confusion with compliant data. Different types of warning signals will be associated to generate differentiated handling instructions. The association between warning signals and handling instructions must be based on accurate matching of the substandard reason to ensure targeted handling measures. If the substandard performance is due to overall excessive stiffness, the system will associate it with the previously triggered red high-risk warning. The generated handling instructions will focus on reducing overall stiffness, including priority adjustments to static parameters (e.g., contact...). (Line tension, dropper length), suggested adjustment direction (tension reduced from 29kN to 27kN, dropper length increased from 1180mm to 1220mm), and lock the measurement points with the highest weight of overall elasticity (such as the 3 measurement points in the middle of the unit segment, based on the sensitivity analysis results above). If the overall softness fails to meet the standard, a yellow medium-risk warning will be issued, and the handling instructions will focus on improving the overall rigidity. The suggested adjustment direction is to increase the contact line tension and shorten the dropper length. The key measurement points will also be marked. If the local sudden change risk point fails to meet the standard, an orange medium-risk warning will be issued, and the handling instructions will target the adjacent measurement point pair where the risk point is located (such as K100+250-K100+300). It is recommended to check and adjust the static parameters with excessive differences between the two points (such as reducing the dropper length difference from 70mm to within 30mm, and the contact line tension difference from 3kN to within 1kN).All handling instructions are generated in a structured format: instruction number - unit segment mileage - reason for non-compliance - parameter to be adjusted - adjustment direction - priority. They are automatically stored in the handling instruction database and simultaneously pushed to the on-site operation and maintenance terminal (including navigation links to risk points) and the back-end management platform. This ensures that staff can quickly obtain targeted adjustment solutions, providing clear guidance for subsequent operations of static parameter adjustment unit 4. At the same time, it forms a closed-loop logic of judgment-early warning-handling, making the rectification process of non-compliant unit segments traceable and controllable.

[0030] For evaluation unit segments determined to be unqualified in dynamic performance prediction, static parameter adjustment unit 4 generates an optimized adjustment scheme for the static height and static pull-out value of the contact wire target at specific measurement points within the evaluation unit segment based on the sensitivity analysis results output by the contact suspension dynamic performance prediction model. Static geometric parameters are adjusted according to the optimized adjustment scheme at the specified measurement points within the evaluation unit segment until the evaluation unit segment is determined to be qualified in dynamic performance prediction.

[0031] After the performance prediction and strategy adjustment unit 3 determines that the dynamic performance of the evaluation unit segment is substandard (such as the presence of overall excessive stiffness, overall excessive flexibility, or local abrupt change risk points), and triggers a process jump to the static parameter adjustment unit 4, it is necessary to first clarify which parameters have the greatest impact on dynamic performance through sensitivity analysis. This analysis is the core of solving the problem of blind static parameter adjustment in traditional methods. It not only builds upon the training results of the contact suspension dynamic performance prediction model mentioned earlier (the model has mastered the nonlinear mapping relationship between static and dynamic parameters), but also provides a priority basis for subsequent optimization and adjustment schemes. This avoids random parameter adjustments by staff (such as blindly increasing or decreasing the length of the suspension cable) that lead to repeated adjustments that still fail to meet the standards, ensuring that static parameter adjustment accurately focuses on key influencing factors. The specific implementation method is as follows: Sensitivity analysis results indicate that after determining that the dynamic performance is substandard—and this determination is a prerequisite for triggering sensitivity analysis—the system will first retrieve complete determination information for the current evaluation unit segment from the performance status database. This includes the type of substandard performance (e.g., overall excessive stiffness corresponding to an average elasticity of 88.5 N / mm exceeding the upper limit of 85 N / mm, local abrupt change risk point corresponding to an elasticity difference of 6.8 N / mm in the K100+250-K100+300 interval exceeding the 5 N / mm threshold), the static parameter set of all measurement points in the unit segment (e.g., five parameters such as the static height of 2550 mm and pull-out value of 300 mm for each measurement point), and the predicted dynamic performance values ​​(average elasticity, maximum elasticity). (Differences), ensuring that sensitivity analysis is accurate for the current substandard unit segment, rather than a general scenario. Simultaneously, the system automatically marks dynamic indicators requiring focused analysis. If the overall stiffness / flexibility is too high, the focus is on analyzing the impact of parameters on the average elasticity of the unit segment; if it's a local abrupt change risk point, the focus is on analyzing the impact of parameters on the maximum elasticity difference of the unit segment. This avoids indiscriminate calculations that waste computing power. The parameter sensitivity calculation is performed by the prediction model, which is based on the previously trained contact suspension dynamic performance prediction model (deep neural network architecture, with an end-to-end nonlinear mapping of static and dynamic parameters). The core is to quantify the degree of influence of small changes in input static parameters on the output dynamic indicators. Specifically, the gradient backpropagation method is used in the calculation. Using the static parameter set of the current evaluation unit segment as the input benchmark, the absolute value of the partial derivative of the dynamic performance index (such as average elasticity) with respect to each input parameter (five static parameters) is calculated through model backpropagation. The larger the absolute value, the higher the sensitivity of the parameter to the dynamic index (i.e., small changes in the parameter will cause significant fluctuations in the dynamic index). If the partial derivative of the parameter with respect to the dynamic index is positive, it means that the dynamic index also increases when the parameter increases (e.g., when the contact line tension increases, the average elasticity increases). If the partial derivative is negative, it means that the dynamic index decreases when the parameter increases (e.g., when the dropper length increases, the average elasticity decreases). For example, for a unit segment that is overly stiff overall, the calculated partial derivative of the contact line tension with respect to the average elasticity is 0.7 (positive), and the partial derivative of the dropper length with respect to the average elasticity is -0.5 (negative). This indicates that the tension is more sensitive to the average elasticity than the dropper length, and that reducing the tension and increasing the dropper length can effectively reduce the average elasticity. Simultaneously, to avoid bias from a single calculation method, the system also employs a small perturbation method for verification: a small perturbation of ±5% is applied to a static parameter (e.g., the perturbation of contact wire tension from 29kN to 27.55kN and 30.45kN), the rate of change of the dynamic index before and after the perturbation is calculated, and compared with the partial derivative calculation result. If the deviation between the two is ≤10%, the sensitivity calculation result is considered reliable; otherwise, the backpropagation optimization model weights are re-executed and the calculation is repeated to ensure data accuracy. The parameter influence weight value for each measurement point within the current evaluation unit segment is output. The parameter influence weight value is a standardized processing of the parameter sensitivity calculation result, converting the partial derivative (or perturbation) into a single value. The dynamic rate of change is normalized to the [0,1] interval to make the influence of different parameters and different measurement points comparable. The normalization formula is: weight value = |parameter sensitivity value| / sum of the absolute values ​​of all parameter sensitivity values. For example, the absolute values ​​of the sensitivity of five parameters at a certain measurement point are 0.7 (contact line tension), 0.3 (static height), 0.2 (pull-out value), 0.5 (dropper length), and 0.1 (positioner slope), with a total of 1.8. Then the weight value of the contact line tension is 0.7 / 1.8≈0.39, the dropper length≈0.28, the static height≈0.17, the pull-out value≈0.11, and the positioner slope≈0.06. During output, the weight values ​​are strictly bound to the measurement point location and parameter type, forming structured data of measurement point mileage-parameter type-weight value-sensitivity positive or negative (e.g., K100+300-contact line tension-0.39-positive, K100+300-dropper length-0.28-negative), and stored in the sensitivity analysis database. This ensures that each parameter at each measurement point has a clear influence weight, avoiding omission of key adjustment points. The weight value quantifies the contribution of five parameters at each measurement point—static height, static pull-out value, contact line tension, dropper length, and locator slope—to the difference between the average and maximum elasticity of the unit segment. The contribution here is the degree of influence of the parameter on the dynamic index, which is directly reflected by the absolute value of the weight value. The larger the weight value, the higher the contribution, and the more significant the effect of parameter adjustment on improving dynamic performance.Specifically, the contribution of the five parameters is as follows: the static height (reflecting the vertical position of the conductor) is reflected in its influence on the vertical stiffness of the contact suspension (if the height is too low, it may increase the vertical stiffness, thus increasing the elasticity; a positive weight value indicates a positive contribution to the average elasticity); the contribution of the static pull-out value (reflecting the horizontal offset of the conductor) is mainly related to the horizontal contact position between the contact wire and the pantograph (excessive offset may lead to abrupt changes in local elasticity; a high weight value indicates a large contribution to the maximum elasticity difference); the contact wire tension (characterizing mechanical stress) is the core mechanical parameter affecting elasticity (the greater the tension, the higher the elasticity is usually; the weight value is generally higher than other parameters, and it has the most significant contribution to the average elasticity); the dropper length (indicating the geometric relationship of the suspension points) affects the elasticity distribution by changing the spacing between suspension points (excessive length differences can lead to abrupt changes in elasticity at adjacent points, and it has a prominent contribution to the maximum elasticity difference); and the locator slope (describing the angle between the support and the conductor) affects the local elasticity by adjusting the direction of force on the contact wire (an unreasonable slope may lead to an increase in local stiffness; although the weight value is usually low, its contribution may increase at points of local abrupt change risk). For example, at a certain local abrupt change risk point, the difference in dropper length between adjacent measurement points is 70mm. The weight value of the dropper length is 0.42, and its contribution to the maximum elasticity difference far exceeds that of other parameters. This indicates that adjusting the dropper length is the key to eliminating this risk point. A priority list of key parameters is generated, arranged in descending order of contribution. The order is based on the absolute value of the parameter's influence weight (i.e., contribution), sorted from largest to smallest. The list must fully present the priority number - measurement point mileage - parameter type - weight value - contribution to dynamic indicators - adjustment direction suggestion information. For example, for a unit segment that is excessively stiff overall, the first three items in the priority list might be: 1 - K100+300 - contact line tension -0.39 - average elasticity - suggested to decrease (due to positive sensitivity); 2 - K100+350 - dropper length -0.32 - average elasticity - suggested to increase (due to negative sensitivity); 3 - K100+250 - contact line tension -0.30 - average elasticity. The recommended reduction is as follows: For localized mutation risk points, the list might be: 1 - K100+250 - String length - 0.42 - Maximum elasticity difference - Recommended shortening (consistent with adjacent points); 2 - K100+300 - Contact line tension - 0.35 - Maximum elasticity difference - Recommended adjustment to match K100+250; 3 - K100+250 - Positioner slope - 0.18 - Maximum elasticity difference - Recommended fine-tuning to 2°. This list will automatically synchronize with the instruction generation module of the static parameter adjustment unit 4. Staff can clearly determine which point, which parameter, and in which direction to adjust without secondary analysis, completely resolving the blindness of traditional adjustments. This provides a clear and executable priority basis for generating subsequent optimization adjustment schemes. Simultaneously, it forms a closed-loop logic of problem location - impact analysis - solution with the previous dynamic performance non-compliance judgment and the subsequent parameter adjustment operation, ensuring efficient and accurate static parameter adjustment.

[0032] After generating a priority list of key parameters through sensitivity analysis (clarifying the influence weight of the five static parameters at each measurement point on dynamic performance), the static parameter adjustment unit 4 needs to generate an optimization adjustment plan that can be directly implemented based on this list. This plan is the core execution basis for solving the problem of substandard dynamic performance. It not only builds on the quantitative results of the sensitivity analysis above (parameter influence weight values), but also transforms abstract weight data into specific construction operation guidelines through the logic of locking key measurement points, formulating adjustment rules, and outputting instruction sets. This avoids the blind, indiscriminate attempts of traditional adjustments and ensures that static parameter adjustments can accurately target the core factors affecting dynamic performance, efficiently promoting the achievement of dynamic performance standards for the unit segment. The specific implementation method is as follows: The method for generating the optimization and adjustment scheme is as follows: Based on the parameter influence weight values, identify the top few target measurement points that have the greatest impact on the difference between the average elasticity and maximum elasticity of the evaluation unit segment. The parameter influence weight values ​​are the values ​​that quantify the degree of parameter influence, verified by gradient backpropagation and the small perturbation method in the sensitivity analysis above (e.g., the contact line tension weight of a certain measurement point is 0.39, and the suspension length weight is 0.32). The number of the top few target measurement points needs to be dynamically determined in conjunction with the type of dynamic performance failure: if the average elasticity of the evaluation unit segment fails to meet the standard (overall too stiff or too flexible), it indicates that the problem involves the entire unit segment. It is necessary to lock the top 3 to 5 measurement points with a cumulative contribution of more than 70% of the weight value (for example, if there are 10 measurement points in a unit segment, take the top 4 with a cumulative contribution of 75%). These points are usually distributed in the middle of the unit segment and key suspension points (such as near the joint of the anchor segment), and have the most significant impact on the overall elasticity. If the maximum elasticity difference of the unit segment does not meet the standard (local mutation risk point), then focus on the adjacent measurement point pair where the risk point is located (such as K100+250 and K100+300), and lock these 2 measurement points. Since the core reason for local mutation is that the parameter difference between adjacent points is too large, there is no need to extend to other points. During the locking process, the system retrieves the weight values ​​of each measurement point from the sensitivity analysis database, sorts them in descending order, and calculates the cumulative contribution. When the cumulative value first exceeds the target threshold (overall failure to meet the target by 70%, local failure to meet the target by 90%), the filtering stops, the final target measurement points are determined, and the mileage positions of these points are marked with red stars on the visualization interface (e.g., K100+250, K100+300, K100+350) to facilitate quick location by staff. For each target measurement point, parameter adjustment direction rules are generated based on the weight value distribution. The weight value distribution refers to the five static parameters (static height, static pull-out) within a single target measurement point. The weighting of parameters (contact line tension, dropper length, and positioner slope) should be considered (e.g., at point K100+300: contact line tension 0.39, dropper length 0.32, static height 0.17, pull-out value 0.11, positioner slope 0.06). Adjustment rules should be based on the sensitivity of parameters to dynamic indicators (positive or negative partial derivatives) in sensitivity analysis. Positive sensitivity indicates that as the parameter increases, the dynamic indicator (e.g., average elasticity) also increases. If the dynamic indicator exceeds the limit (e.g., overall excessive stiffness, average elasticity exceeding the upper limit), the parameter needs to be decreased. Negative sensitivity indicates that as the parameter increases, the dynamic indicator decreases. If the dynamic indicator exceeds the limit, the parameter needs to be increased. For example, if the contact line tension at point K100+300 is positively sensitive to the average elasticity (weight 0.39), and the unit segment is excessively stiff (average elasticity of 88.5 N / mm exceeds the upper limit of 85 N / mm), then the adjustment direction rule for the contact line tension at that point is to reduce it. If the sensitivity of the suspension string length at the same measurement point is negative (weight 0.32), then the adjustment direction rule is to increase it.Meanwhile, the rules clearly stipulate the principle of prioritizing the adjustment of the parameter with the highest weight. That is, for each target measurement point, the parameter with the highest weight is focused first. After that parameter is adjusted, the second highest weight parameter is adjusted based on the results of dynamic performance retesting. This avoids confusion in causal relationships caused by adjusting multiple parameters simultaneously and ensures that the effect of each adjustment step can be quantitatively verified. If the average elasticity of the evaluation unit segment does not meet the standard, the parameter with the highest weight value is adjusted first. The failure to meet the average elasticity standard is divided into two categories: overall over-stiffness (average elasticity exceeds the upper limit) and overall over-flexibility (average elasticity is below the lower limit). The adjustment logic needs to be adapted accordingly: if it is overall over-stiffness (e.g., the average elasticity of the 350km / h unit segment is 88.5N / mm, exceeding 85N / mm), the locked target measurement points (e.g., K100+250, K100+300, K100+350) are all adjusted first with the highest weight. Assuming that the highest weight for these three points is contact wire tension (weight 0.39-0.42) and that all three have positive sensitivity, the adjustment direction is uniformly to reduce contact wire tension. The adjustment range is based on the average elasticity deviation (deviation 3.5 N / mm; according to sensitivity analysis, for every 1 kN decrease in tension, the average elasticity decreases by 1.2 N / mm, so it is recommended to reduce the tension at each target point by 3 kN, from 29 kN to 26 kN). If the overall elasticity is too soft (e.g., average elasticity 76.2 N / mm is below the lower limit of 80 N / mm), the highest weight parameter for the target measurement point is the dropper length (weight 0.35-0.38), with negative sensitivity. In this case, the adjustment direction is to shorten the dropper length, with the range calculated based on the average elasticity increasing by 0.8 N / mm for every 10 mm shortening of the dropper. It is recommended to shorten it by 50 mm, from 1250 mm to 1200 mm. The core of priority adjustment is to concentrate resources on solving the main contradictions and avoid wasting time on low-weight parameters. For example, if the static height weight of a target point is only 0.17, even if it is adjusted, it will be difficult to significantly improve the overall elasticity. Therefore, it will not be included in the first round of adjustment. After the parameter with the highest weight is adjusted, if the average elasticity is still not up to standard, the second highest weight parameter will be considered. If the maximum elasticity difference of the unit segment is not up to standard, the parameter with the highest weight value in the adjacent point pair that causes elasticity mutation will be adjusted first. The adjacent point pair that causes elasticity mutation is the local mutation risk point identified above (such as K100+250 and K100+300, with an elasticity difference of 6.8N / mm exceeding the 5N / mm threshold). The key to adjustment is to reduce the parameter difference between the two points, rather than adjusting a single point.First, the weight values ​​of each parameter in the adjacent point pair need to be compared. The parameter with the highest weight is identified. Assuming that the weight of the dropper length at both points is 0.42 (higher than other parameters), and the difference in dropper length between the two points reaches 70mm (1250mm for K100+250 and 1180mm for K100+300), this difference is the core reason for the elastic change. Therefore, the dropper length should be adjusted first. The adjustment direction should be to make the parameters at the two points more consistent. Based on the sensitivity analysis, the sensitivity of the dropper length of K100+250 is negative (weight 0.42), and the sensitivity of K100+300 is also negative. Therefore, it is recommended to increase the shorter dropper length of K100+300 from 1180mm to 1220mm (reducing the difference from 1250mm of K100+250 to within 30mm). At the same time, the elastic difference between the two points should be re-measured. If it still exceeds the threshold, the contact wire tension with the second highest weight should be adjusted (the tension difference between the two points is 3kN, and it is recommended to adjust it to 27kN for both). This adjustment logic, which focuses on the differences between adjacent points and prioritizes high-weight parameters, can accurately eliminate the root cause of local elastic mutations. It avoids the ineffective operation of traditional adjustment, which only adjusts a single point and ignores differences. The final output is an instruction set containing the target measurement point location, the type of parameter to be adjusted, and the adjustment direction. The instruction set is generated in a structured format, with each instruction corresponding to one parameter to be adjusted for a target measurement point. It includes instruction number, target measurement point mileage, parameter type to be adjusted, current parameter value, target parameter value, adjustment direction, adjustment priority, and core information based on the specification. For example, instruction 001-K100+300-contact wire tension-current 29kN-target 26kN-adjustment direction: decrease-priority 1-based on TB / T3558-2020; instruction 002-K100+300-suspender length-current 1180mm-target 1220mm-adjustment direction: increase-priority 2-based on TB / T3113-2017. The adjustment priority in the instruction set is directly linked to the parameter weight value, with the highest weight parameter having a priority of 1 (first round of adjustment) and the second highest being 2 (second round of adjustment). Simultaneously, post-adjustment verification requirements are included, such as re-collecting static parameters within one hour after adjustment and inputting them into the prediction model to retest dynamic performance. After the instruction set is generated, it is automatically stored in the system's adjustment instruction database and pushed out through the following methods: First, it is displayed in a pop-up window on the monitoring center's visualization interface, indicating the urgency level (e.g., overall over-rigidity marked in red for urgency); second, it is pushed to the mobile phones of on-site maintenance personnel via a dedicated construction APP, along with a satellite navigation link to the target measurement point (accuracy ±1 meter); third, it is synchronized to the control system of the hydraulic adjustment device, providing parameter basis for automated adjustment (e.g., the device can automatically reduce the contact wire tension from 29kN to 26kN according to the instruction). This instruction set, with complete information, timely push, and direct execution, ensures that the static parameter adjustment process is systematic, completely solving the blindness of traditional adjustments and providing clear guidance for subsequent closed-loop adjustments (adjustment-retesting-re-adjustment), ultimately promoting the efficient achievement of dynamic performance standards for the evaluation unit segment.

[0033] After the static parameter adjustment unit 4 generates an optimized adjustment scheme containing the target measurement point, the parameter to be adjusted, and the adjustment direction, it enters a closed-loop process of adjustment-retesting-evaluation-iteration. This process is the core of realizing the abstract scheme into actual dynamic performance compliance. It relies on automated equipment to ensure adjustment accuracy, avoids accidental deviations in a single adjustment through multiple rounds of retesting, and prevents damage to the contact network structure by limiting safe iterations. Ultimately, it achieves the transformation from non-compliance to stable compliance, forming a complete logical chain of design-execution-verification with the optimized adjustment scheme mentioned above. The specific implementation method is as follows: Based on the optimized adjustment scheme, the hydraulic adjustment device is driven to adjust the contact wire positioner at the designated measurement point. The hydraulic adjustment device is a special automated equipment adapted to the adjustment of the contact wire positioner. It has a displacement control accuracy of ±0.5mm and a tension control accuracy of ±0.2kN. It supports both remote system control and on-site panel operation modes. Its core components include a hydraulic servo pump, a vertical / horizontal adjustment mechanism, a tension adjustment motor, and a real-time sensing module (height sensor, pull-out value laser rangefinder, tension sensor). The designated measurement point must be precisely aligned with the mileage marker (e.g., K100+300) marked in the plan using the device's built-in GPS positioning module, with a deviation of ≤0.5 meters, ensuring that the object being adjusted is the target contact wire locator. The adjustment process must be performed differently according to the type of parameter to be adjusted: If the parameter to be adjusted is static height (e.g., from 2550mm to 2545mm), the device drives the locator clamp to rise and fall along the support rail via a vertical hydraulic push rod. The height sensor provides real-time data feedback at a frequency of 10Hz. The device automatically stops when the actual height deviates from the target value by ≤1mm. If the parameter to be adjusted is static pull-out value (e.g., from 300mm to 280mm), the locator offset angle is adjusted via a horizontal hydraulic mechanism, and the pull-out value is monitored with a laser rangefinder until the deviation is ≤0.5mm. If the parameter to be adjusted is contact wire tension (e.g., from 29kN to 26kN), the device drives the tension wheel to rotate via a tension adjustment motor, and simultaneously reads the tension sensor data to ensure that the tension is stable within the target value ±0.2kN range. During adjustment, the system will record the hydraulic pressure (safety threshold ≤30MPa), adjustment displacement, and time in real time. If the pressure exceeds the limit or the sensor data is abnormal, the device will stop immediately and trigger an audible and visual alarm. The device will be restarted after the staff has checked the fault (such as insufficient hydraulic oil or positioner jamming) to ensure that the adjustment is safe and controllable. After the adjustment is completed, the static parameter acquisition process will be executed again. The criterion for completing the adjustment is that the hydraulic adjustment device displays that the adjustment is in place and there is no fault code. At this time, it needs to stand for 5 to 10 minutes to allow the elastic deformation of the contact wire caused by the adjustment to fully stabilize (to avoid data distortion caused by the deformation not disappearing). The static parameter acquisition process was re-executed in complete accordance with the initial acquisition logic of dividing the assessment acquisition units to ensure data comparability. Staff carried contact wire geometric parameter measuring instruments (accuracy ±0.1mm), dynamic calibration tension gauges (accuracy ±0.1kN), and high-precision laser inclinometers (accuracy ±0.1°) to collect data from the designated measurement point and two adjacent measurement points (such as K100+250, K100+300, K100+350) in all dimensions. This covered five parameters: static height, static pull-out value, contact wire tension, dropper length, and positioner slope, avoiding the loss of local data caused by only collecting data from the target point.The collected data is uploaded to the system in real time via a 4G module. Anomalies (such as instantaneous tension fluctuations exceeding ±0.5kN or height mutations exceeding ±2mm) are removed by the data verification module. The data is then standardized according to the initial collection format (units are unified to mm, kN, and °, and values ​​are rounded to one decimal place) to form an adjusted static parameter set. This provides clean and consistent basic data for subsequent dynamic performance re-evaluation. The updated data is input into the prediction model for dynamic performance re-evaluation. Updating the data involves replacing the historical data of the original substandard unit segments with the adjusted static parameter set. Simultaneously, the data status is marked as retest data in the database to ensure the model reads the latest parameters. The process of inputting the prediction model and the logic of parameter calculation and mapping units are completely integrated. The adjusted five parameters of each measurement point are arranged in mileage order to form a feature matrix of measurement point number × 5. This matrix is ​​then input into the trained contact suspension dynamic performance prediction model. The model undergoes nonlinear transformation through three hidden layers to output the updated average elasticity and maximum elasticity difference of the unit segment. The dynamic performance re-evaluation reuses the judgment rules of the performance prediction and strategy adjustment unit: the updated average elasticity is compared with the allowable elasticity fluctuation range of the corresponding line (e.g., 80-85 N / mm for a 350 km / h high-speed railway) to determine whether the mean is compliant; the updated maximum elasticity difference is compared with the elasticity difference tolerance threshold to determine whether the dispersion is controllable. Only when both conditions are met simultaneously is the evaluation deemed satisfactory; otherwise, it is deemed unsatisfactory. The reassessment results are displayed in real time on the visualization interface, with green indicating compliance and red indicating non-compliance. A reassessment report is generated, clearly showing the changes in dynamic indicators before and after the adjustment (e.g., average elasticity decreases from 88.5 N / mm to 84.2 N / mm, maximum elasticity difference decreases from 6.8 N / mm to 4.5 N / mm). This provides an intuitive basis for deciding whether to initiate iteration. If the results still do not meet the standards, optimization is performed iteratively based on the latest parameter set. If the results still do not meet the standards, it means that the reassessment results are non-compliant with the mean (e.g., average elasticity of 86.3 N / mm exceeds the upper limit of 85 N / mm) or uncontrollable dispersion (e.g., maximum elasticity difference of 5.8 N / mm exceeds the threshold of 5 N / mm). In this case, it is necessary to avoid repeating the original solution and instead restart optimization based on the adjusted static parameter set. The system first inputs the latest parameter set into the prediction model to perform sensitivity analysis, recalculates the influence weight values ​​of parameters at each measurement point (e.g., the weight of the dropper length increases from 0.32 to 0.38 after adjustment), generates a new priority list of key parameters in descending order of contribution, and then fine-tunes the adjustment direction based on the re-evaluation results (e.g., the original scheme only reduced the contact wire tension, but after iteration it is changed to reduce tension + shorten the dropper length), locks in new target measurement points (e.g., adding K100+400 points), and generates an iterative optimization scheme. Each iteration must be spaced at least 30 minutes apart to ensure the catenary system is fully stable. The system records key information for each iteration (number of iterations, adjusted parameters, and changes in dynamic indicators), creating an iteration trajectory diagram to help staff analyze the correlation between parameters and indicators (e.g., for every 1kN decrease in tension, elasticity decreases by 1.2N / mm), avoiding blind adjustments. The process is forcibly terminated when the evaluation unit segment is deemed to have met the dynamic performance prediction standard in two consecutive evaluations, or when the system's preset maximum safe iteration count is reached. Meeting the standard in two consecutive evaluations is to avoid misjudgments caused by momentary interference (e.g., parameter acquisition deviation due to a light breeze) in a single evaluation. After the first evaluation, static parameter acquisition and dynamic performance evaluation must be performed again after a 1-hour interval. If the second evaluation still meets the requirements of mean compliance and controllable dispersion, it is considered stable and meets the standard, the closed-loop process terminates, the system generates a dynamic performance compliance confirmation, synchronizes it to the construction progress management module, and allows the unit segment to proceed to the next process (e.g., catenary accessory installation). The system's preset maximum safe iteration count is set at 5 times based on the catenary structure's tolerance. This is because excessive iteration (e.g., exceeding 5 times) may cause fatigue damage to the positioner or structural deformation due to repeated adjustments to the dropper length. If the standard is not met after the 5th iteration, the system will forcibly terminate the process, trigger a red safety warning, and push it to the operation and maintenance management platform. At the same time, an adjustment anomaly analysis report will be generated, including the adjustment parameters of each round, changes in dynamic indicators, and suspected abnormal causes (such as model prediction deviations or sensor calibration failures). On-site verification by staff is required (e.g., checking for hard spots in the catenary or whether the hydraulic device has lost accuracy). Only after a manual intervention plan is formulated (e.g., replacing the positioner or recalibrating the model) can the closed-loop process be restarted, rather than continuing automatic iteration. Ultimately, this ensures both construction quality and the safety of the catenary structure.

[0034] In this invention, the contact wire is divided into evaluation unit segments, static parameters of each measurement point are collected, the dynamic performance prediction model of the contact suspension is input, the average elasticity and maximum elasticity difference of the unit segment are output, and the dynamic performance is compared with the standard to determine whether it meets the standard. If it does not meet the standard, the target static parameter optimization scheme of the contact wire is generated based on the model sensitivity analysis results, and the closed-loop adjustment is carried out until it meets the standard. This system realizes the correlation prediction and precise adjustment of static and dynamic parameters, which is applicable to the construction monitoring of railway contact wire and improves the construction quality and efficiency.

[0035] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A railway electrical, signaling, and electronic control system based on multi-source data fusion, characterized in that: include: Divide the evaluation and acquisition units (1) Divide the contact network line to be adjusted into several continuous evaluation unit segments. In each evaluation unit segment, select several representative measurement points and use the contact network geometric parameter measuring instrument and tension meter to measure and record the static parameters at each measurement point. The parameter calculation and mapping unit (2) inputs static parameters into the preset contact suspension dynamic performance prediction model and predicts and outputs the key dynamic performance prediction indicators of the evaluation unit segment, including the average elasticity of the unit segment and the maximum elasticity difference of the unit segment. The performance prediction and strategy adjustment unit (3) compares the average elasticity of the unit segment with the preset standard elasticity range, and compares the maximum elasticity difference of the unit segment with the preset maximum allowable elasticity difference to determine whether the dynamic performance prediction of the evaluation unit segment meets the standard. If it does not meet the standard, it enters the static parameter adjustment unit (4). The static parameter adjustment unit (4) generates an optimized adjustment scheme for the static height of the contact line target and the static pull-out value of the contact line target at a specific measurement point in the evaluation unit segment based on the sensitivity analysis results output by the dynamic performance prediction model of the contact suspension, and adjusts the static geometric parameters of the specified measurement point in the evaluation unit segment according to the optimized adjustment scheme until the evaluation unit segment is determined to meet the dynamic performance prediction.

2. The intelligent construction monitoring system for railway electrical, electronic, and electronic systems based on multi-source data fusion according to claim 1, characterized in that: Static parameters include: The static height of the contact wire at the measurement point reflects the reference value of the vertical position of the conductor, the static pull-out value reflects the reference value of the horizontal offset of the conductor, the contact wire tension collected in real time, characterizes the mechanical stress state of the suspension system, the length of the dropper at adjacent dropper nodes, indicates the geometric relationship between suspension points and the slope of the locator at the positioning point, and describes the spatial angle between the support structure and the conductor. Among them, the contact wire tension is dynamically calibrated and measured by a tension meter, and the dropper length and the slope of the locator are collected synchronously by a high-precision laser rangefinder and an inclinometer to ensure that the static parameter set covers the spatial geometry and mechanical characteristics of the contact suspension.

3. The intelligent construction monitoring system for railway electrical, electronic, and electronic systems based on multi-source data fusion according to claim 1, characterized in that: The method for establishing the dynamic performance prediction model of the contact suspension is as follows: Based on the massive samples in the historical maintenance database, the measured data of static height, static pull-out value, contact line tension, dropper length and positioner slope at each measurement point are extracted as input features, and the elastic uniformity index and offline rate index obtained in the dynamic test of the corresponding section are synchronously associated as training labels. By employing a deep neural network architecture and iteratively optimizing network weights through the backpropagation algorithm, an end-to-end nonlinear mapping relationship from static parameter sets to dynamic performance indicators is constructed, ultimately generating a computational model that can predict dynamic elastic distribution characteristics based on real-time static parameters.

4. The intelligent construction monitoring system for railway electrical, electronic, and electronic systems based on multi-source data fusion according to claim 3, characterized in that: The technical means for predicting key dynamic performance indicators are as follows: Input the static parameter set of all measurement points in the current evaluation unit segment into the trained prediction model, and the model outputs the local elasticity prediction value at each measurement point in the evaluation unit segment. Based on the sequence of local elasticity prediction values ​​of all measurement points, the arithmetic mean of these values ​​is calculated as the average elasticity of the unit segment. At the same time, the differences in local elasticity prediction values ​​between all adjacent measurement points are traversed, and the result with the largest absolute difference is selected as the maximum elasticity difference of the unit segment. This is used to quantify the uniformity and dispersion of the contact suspension under dynamic load.

5. The intelligent construction monitoring system for railway electrical, electronic, and electronic systems based on multi-source data fusion according to claim 4, characterized in that: The specific steps for comparing the average elasticity of the evaluation unit segment with the preset standard elasticity range are as follows: The system invokes the preset elasticity allowable fluctuation range, which is dynamically set according to the catenary design specifications and line operation level. When the average elasticity of the evaluation unit segment is within the elasticity allowable fluctuation range, it is marked as compliant. If it exceeds the upper limit of the range, it is judged as too stiff overall, and if it is below the lower limit, it is judged as too flexible overall, and different types of early warning signals are automatically triggered.

6. The intelligent construction monitoring system for railway electrical, electronic, and electronic systems based on multi-source data fusion according to claim 4, characterized in that: The operation of comparing the maximum elastic difference of the unit segment with the preset maximum allowable elastic difference is as follows: Based on the maximum operating speed of the line and the stability requirements of pantograph-catenary current collection, a tolerance threshold for elasticity difference is set. When the maximum elasticity difference of the evaluation unit segment does not exceed the threshold, it is marked as discrete and controllable. If it exceeds the threshold, the location of the adjacent measurement point pair where the maximum elasticity difference occurs is identified and marked as a local mutation risk point, generating risk location information containing location coding.

7. The intelligent construction monitoring system for railway electrical, signaling, and electronic control systems based on multi-source data fusion according to claim 6, characterized in that: The logic for determining whether the dynamic performance prediction of the evaluation unit segment meets the standard is as follows: The dynamic performance prediction is deemed to be up to standard only when both the mean compliance mark and the discrete controllable mark are met simultaneously. If any of the following states occur: overall over-rigidity, overall over-flexibility, or local sudden change risk points, the performance is deemed to be below standard, and different types of early warning signals are associated to generate differentiated handling instructions.

8. The intelligent construction monitoring system for railway electrical, electronic, and electronic systems based on multi-source data fusion according to claim 7, characterized in that: Sensitivity analysis results indicate: After determining that the dynamic performance is substandard, the prediction model performs parameter sensitivity calculation and outputs the parameter influence weight value of each measurement point in the current evaluation unit segment. This weight value quantifies the contribution of five parameters at each measurement point—static height, static pull-out value, contact line tension, dropper length, and positioner slope—to the difference between the average elasticity and maximum elasticity of the unit segment. The parameters are then arranged in descending order of contribution to generate a priority list of key parameters.

9. The intelligent construction monitoring system for railway electrical, signaling, and electronic control systems based on multi-source data fusion according to claim 1, characterized in that: The method for generating the optimization and adjustment scheme is as follows: Based on the parameter influence weight values, identify the top few target measurement points that have the greatest impact on the difference between the average elasticity and maximum elasticity of the evaluation unit segment. For each target measurement point, generate parameter adjustment direction rules based on the weight value distribution. If the average elasticity of the evaluation unit segment does not meet the standard, the parameter with the highest weight value is adjusted first. If the maximum elasticity difference of the unit segment does not meet the standard, the parameter with the highest weight value in the adjacent point pair that causes the elasticity change is adjusted first. The final output is a set of instructions that includes the location of the target measurement point, the type of parameter to be adjusted, and the adjustment direction.

10. The intelligent construction monitoring system for railway electrical, signaling, and electronic control systems based on multi-source data fusion according to claim 9, characterized in that: The closed-loop process for performing static geometric parameter adjustments until the dynamic performance prediction meets the target is as follows: Based on the optimized adjustment scheme, the hydraulic adjustment device is driven to adjust the contact wire positioner at the designated measurement point. After the adjustment is completed, the static parameter acquisition process is re-executed, the data input prediction model is updated, and the dynamic performance is re-evaluated. If it still does not meet the standard, the optimization is iteratively executed based on the latest parameter set until the dynamic performance prediction of the evaluation unit segment is determined to meet the standard in two consecutive evaluations, or the maximum safe iteration number preset by the system is reached, at which point it is forcibly terminated.

Images