A method for visualizing hardware performance monitoring data of a CTC system localization

Abstract

The application discloses a CTC system localization hardware performance monitoring data visualization implementation method, solves the problems of poor adaptability of existing general monitoring schemes to localization hardware and railway business scenes, disconnection of data and business, weak visualization analysis and difficult special network deployment. The method ensures the improvement of data accuracy through automatic matching and multi-dimensional verification of hardware types, differentiated packet loss completion and business correlation abnormal value filtering; constructs a device-line-business three-dimensional layered model, combines railway operation diagrams for differentiated down-sampling and weighted statistics, and automatically associates business event labels to realize deep binding of performance data and business logic; through multi-index linkage visualization, pixel density detection intelligent labeling and multi-dimensional interaction filtering, the performance bottleneck positioning efficiency is improved; through three-level dependency list management, resource path mapping and static link encapsulation, an executable file that can completely run offline is generated, and the stable operation rate of the localization terminal is ensured.

Description

Technical Field

[0001] This invention relates to the field of railway signaling system operation and maintenance technology, specifically focusing on the performance monitoring and visualization analysis of domestically produced hardware equipment for Centralized Traffic Control (CTC), and more specifically, to a method for visualizing CTC system performance monitoring data. Background Technology

[0002] As the core hub of railway traffic control, the CTC system's hardware is gradually migrating to domestically produced platforms, such as Phytium / Kunpeng CPUs and Kylin operating systems. The key performance indicators of these hardware components, such as CPU utilization, memory usage, network bandwidth, and disk I / O, directly determine the CTC system's response efficiency and processing capabilities for core operations such as train route processing, dispatch command transmission, and train tracking.

[0003] As the localization process of railway CTC system progresses, the deployment of domestically produced hardware equipment across the entire railway bureau is becoming increasingly widespread. However, its performance monitoring currently faces many technical bottlenecks: existing monitoring tools are mostly general server monitoring solutions, which cannot adapt to the interface characteristics of domestically produced hardware and the business logic of the CTC system; data collection and visualization are disconnected, making it difficult to correlate with railway business scenarios and locate performance bottlenecks.

[0004] Currently, there are several existing technical solutions related to this invention in the relevant fields, as follows:

[0005] 1. Existing technology

[0006] This technology is a distributed monitoring architecture based on common network management protocols (such as SNMP). It acquires CTC system hardware performance data through acquisition agents deployed on network nodes. After simple deduplication processing by a centralized data server, a basic line chart is generated using a general-purpose plotting engine. Maintenance personnel then manually compare the data to correlate business logic with performance data. However, this technology uses a standardized data acquisition interface and does not customize it for the specific performance parameters of domestically produced hardware or the CTC business scenario. Data processing is limited to surface-level deduplication, lacking outlier filtering and packet loss compensation mechanisms.

[0007] The core flaw of this technology lies in the severe mismatch between the general monitoring solution and the compatibility with domestically produced CTC hardware and railway business scenarios. Relying on general network management protocols to collect data, it cannot identify the dedicated performance register data of Phytium and Kunpeng CPUs, or the kernel-level resource usage information of the Kylin system. The collected results only reflect the basic operating status of the hardware, with extremely weak correlation to core CTC business processes such as train route processing and dispatch command transmission, making it difficult to reflect the true performance of domestically produced hardware in railway-specific scenarios. Furthermore, the visualization layer can only display the time trend of a single indicator, completely ignoring CTC business events (such as dispatch command issuance and train passing through key stations) and business cycle characteristics, failing to support the correlation analysis between performance bottlenecks and business behaviors. Moreover, performance data and business logic rely heavily on manual integration, resulting in low analysis efficiency and susceptibility to misjudgments due to human error, failing to meet the high reliability requirements of the CTC system.

[0008] 2. Existing technology two

[0009] This technology is an industrial big data multidimensional analysis architecture based on the JSON document structure. Using JSON as the data carrier, it constructs a general industrial dataset by configuring multiple data sources and defining basic data transformation and cleaning rules. Then, it generates fixed-format analysis reports through graphical configuration. While this technology employs a generalized data processing workflow and supports basic field format conversion, its visualization relies on preset templates and is not customized or optimized for the business characteristics of the CTC system and the domestic hardware environment.

[0010] The drawback of this technology lies in its insufficient targeting and adaptability to the localized operation and maintenance scenarios of CTC (Customer Traffic Control). It lacks a dedicated preprocessing mechanism designed to incorporate the data characteristics of domestically produced hardware. Relying solely on general threshold filtering of outliers is insufficient to address issues such as data transmission packet loss and hardware fluctuations in the CTC system environment. Furthermore, it fails to design a tiered processing strategy based on the CTC business cycle (busy during the day / idle at night), leading to data redundancy and missing key performance information. In addition, the fixed format of the visualization reports prevents multi-dimensional filtering based on railway-specific dimensions such as CTC business type and equipment line level. It also lacks statistical information annotations for core performance indicators, and its interactive functions are limited to basic zooming and panning, failing to meet the needs of professional analysis.

[0011] 3. Existing Technology Three

[0012] This technology is a general-purpose performance indicator monitoring platform based on a domestically developed software and hardware ecosystem. It adopts a local deployment model to meet data security and compliance requirements. It collects basic performance indicators by connecting to domestic databases and middleware, and displays single-dimensional indicator trend charts via a web interface. The technology has built-in fixed alarm thresholds and supports simple filtering by device ID. Its core design focuses on monitoring performance indicators for common scenarios.

[0013] The main problem with this technology is the insufficient support capability of domestically produced general-purpose monitoring platforms for CTC professional operation and maintenance scenarios. It lacks a dedicated business dimension model for the CTC system, making it unable to link railway-specific data such as business types, line levels, and train operation plans, thus hindering the identification of core CTC operation and maintenance issues such as "hardware performance bottlenecks caused by specific business needs." At the visualization level, it only supports independent display of single indicators, lacking multi-indicator linkage analysis capabilities. Statistical information only presents real-time values, lacking in-depth analysis dimensions such as historical maximum values ​​and interval weighted averages, failing to meet the professional analysis needs of operation and maintenance personnel. Although the deployment adopts a local mode, it still relies on domestically produced cloud platforms or dedicated servers to provide web service support, making complete offline operation impossible and inconsistent with the file management standards and usage habits in railway operation and maintenance scenarios.

[0014] In summary, existing technical solutions have significant shortcomings in adapting to the characteristics of domestically produced hardware, relating to railway business scenarios, achieving in-depth visualization analysis, and adapting to offline deployment on private networks. There is an urgent need for a performance monitoring data visualization method that can deeply integrate CTC business logic, adapt to domestically produced hardware environments, support multi-dimensional analysis and intelligent annotation, and facilitate offline deployment on private networks. Summary of the Invention

[0015] To address the shortcomings in existing technologies, this invention discloses a CTC system domestic hardware performance monitoring data visualization system, the technical solution of which is as follows: A CTC system domestic hardware performance monitoring data visualization system, characterized in that it includes a core information identification and data adaptation module, a domestic hardware performance data processing module, a business association visualization module, and a CTC system private network adaptation and packaging deployment module connected in sequence.

[0016] The core information identification and data adaptation module is used to read a CSV file containing hardware type identifiers, automatically match domestic hardware combinations and load corresponding field mapping rules, parse exclusive performance parameters, and perform legality verification of device ID, service type and performance value based on a three-layer verification system, perform differential completion of packet loss data, and perform business association filtering on abnormal values.

[0017] The domestically produced hardware performance data processing module is used to construct a three-dimensional hierarchical model of "equipment ID-line level-business type", perform multi-round grouping based on a preset weight system, identify business time periods according to the railway operation diagram and perform differentiated downsampling, calculate multi-dimensional statistical indicators, and associate with the CTC business event library through a dual-condition matching mechanism of "time window + equipment ID" to tag the performance data with business.

[0018] The business association visualization module is used to display multi-indicator performance data in the form of multiple subgraphs on a standardized canvas, intelligently annotate statistical information using a pixel density detection algorithm, display business events through combined markers, configure multi-dimensional filtering controls and warning thresholds, and realize the deep association display between performance data and CTC business.

[0019] The CTC system private network adaptation and packaging deployment module is used to build a three-level dependency list system, verify version compatibility, generate executable files that can run offline through resource path mapping and compression optimization technology, and support the export of charts according to railway operation and maintenance habits.

[0020] This invention also discloses a method for visualizing the performance monitoring data of domestically produced CTC system hardware, characterized by the following steps:

[0021] Step S1: By reading the CSV file containing hardware type identifiers, automatically match domestic hardware combinations, load exclusive field mapping rules, parse exclusive performance parameters, and perform three-layer verification of device ID, service type, and performance value, perform differentiated completion on packet loss data, and perform business association filtering on outliers;

[0022] Step S2: Construct a three-dimensional hierarchical model of "equipment ID-line level-business type", group the data based on preset weights, identify business time periods according to the railway operation diagram and perform differentiated downsampling, calculate multi-dimensional statistical indicators, match the business event database through "time window + equipment ID" and label the data with business tags;

[0023] Step S3: Display performance data in the form of multiple subgraphs on a standardized canvas, use a pixel density detection algorithm to intelligently label statistical information, display business events through combined tags, configure multi-dimensional filtering and early warning mechanisms, and realize the visualization of the relationship between performance and business.

[0024] Step S4: Construct a three-level dependency list, verify compatibility, and generate an executable file that can run offline through resource path mapping and compression optimization, supporting the export of charts according to railway operation and maintenance specifications.

[0025] Beneficial effects

[0026] 1. Improved the accuracy and reliability of data processing, providing high-quality data support for operation and maintenance decisions.

[0027] By employing a core information identification and data adaptation module (or methodological steps), and utilizing an integrated mechanism of "automatic hardware type matching + multi-dimensional legality verification + differentiated data completion + business-related outlier filtering," the system effectively addresses issues such as inconsistent data formats, packet loss, and noise interference in the performance data of domestically produced hardware within the railway private network environment. This significantly improves data accuracy. Furthermore, by combining a three-dimensional hierarchical model with a differentiated downsampling strategy, the error in the calculated statistical indicators is reduced, thus providing a reliable and accurate data foundation for the performance evaluation and maintenance decisions of domestically produced hardware.

[0028] 2. It enables rapid and accurate location of performance bottlenecks, significantly improving the efficiency of operation and maintenance analysis.

[0029] Through data processing and visualization modules (or methodologies), a three-dimensional hierarchical model of "Device ID - Line Level - Service Type" was constructed. Innovative technologies such as "pixel density detection" intelligent annotation, a business event combination annotation mechanism, and multi-dimensional real-time filtering interactive design were adopted. These technologies achieve deep integration and visual correlation between hardware performance data and CTC business scenarios (such as route processing and dispatch command issuance), enabling operations and maintenance personnel to intuitively and quickly understand the causal relationship between performance fluctuations and business behaviors.

[0030] 3. Ensures high compatibility and stable offline operation capability of the system in a private network isolation environment.

[0031] By using dedicated network adaptation and packaging deployment modules (or methods and steps), a three-level resource list management system of "core function dependency -> domestic adaptation dependency -> CTC business dependency" was constructed. Precise resource path mapping, compression optimization and static link encapsulation technology were adopted to effectively solve the deployment compatibility problem of domestic operation and maintenance terminals (such as Kylin / Tongxin operating system, Phytium / Kunpeng CPU). The generated independent executable file supports complete offline operation.

[0032] 4. It lowers the technical threshold for professional operation and maintenance, and enhances the ease of use and universality of the system.

[0033] Through integrated and streamlined system and interaction design (such as intelligent annotation, multi-dimensional filtering, and threshold warning functions), complex performance analysis and chart generation can be completed through simple interaction without requiring maintenance personnel to have professional programming skills. At the same time, the system supports custom configuration of parameters such as warning thresholds, downsampling intervals, and export formats, which can flexibly adapt to the personalized maintenance specifications of different railway bureaus. Attached Figure Description

[0034] Figure 1 Flowchart of the core information identification and data adaptation module;

[0035] Figure 2 Flowchart of the performance data processing module for domestically produced hardware;

[0036] Figure 3 Visualize the logic flow of business-related modules;

[0037] Figure 4 Logical flowchart of the module for packaging and deployment for private network adaptation;

[0038] Figure 5 This is a structural block diagram of the system technical solution of the present invention;

[0039] Figure 6 This is a flowchart of the technical solution of the method of the present invention. Detailed Implementation

[0040] Example 1

[0041] This embodiment describes in detail the specific implementation of the CTC system's domestic hardware performance monitoring data visualization system.

[0042] The implementation of this system mainly includes four functional modules that work in sequence: core information identification and data adaptation module, domestic hardware performance data processing module, business association visualization module, and CTC system private network adaptation and packaging deployment module.

[0043] 1. Specific implementation of the core information identification and data adaptation module

[0044] This module serves as the system's data entry point. Its core task is to ensure that raw performance data collected from various domestically produced hardware platforms, which may contain noise and missing information, can be accurately and reliably imported and transformed into high-quality datasets usable for subsequent analysis. To achieve this, the module employs the following series of specific technical methods working together:

[0045] First, the module automatically identifies the hardware combination from which the data originates by parsing the pre-defined "Hardware Type Identifier" field in the CSV format data file. Examples include "FT_KYLIN" (representing a Phytium CPU + Kylin system) or "KP_KYLIN" (representing a Kunpeng CPU + Kylin system). The system internally has a pre-defined field mapping rule table corresponding to each hardware combination. This rule table defines the mapping relationship between the original data column names and the system's unified performance parameter names, and can parse performance parameters unique to domestically produced hardware. For example, for a Phytium CPU, the rule table guides the system to map the readings of specific registers to "CPU user mode utilization" and "CPU kernel mode utilization"; for a Kylin system, it can parse the memory status information exported by the kernel into "physical memory usage" and "page cache usage"; and for domestically produced network cards, it can identify and extract dedicated indicators such as "CTC private network priority bandwidth." This automatic matching and mapping mechanism directly solves the core adaptation problem of inconsistent data formats for different domestically produced hardware and the inability of general monitoring tools to identify dedicated parameters. If this method is not adopted, the monitoring scheme must be manually configured or modified for each type of hardware, which is cumbersome and prone to errors, and cannot achieve "one-time import, automatic adaptation".

[0046] Secondly, to ensure the compliance and validity of the data foundation, the module constructed and implemented a three-layer verification system. This system verifies three key dimensions in sequence: (1) Equipment ID format verification: According to railway industry standards, it checks whether it conforms to the standard structure of "railway bureau code-station code-equipment number"; (2) Business type enumeration verification: It confirms whether the value of the business type field strictly belongs to the four CTC core businesses of route processing, train tracking, dispatching order issuance, and log synchronization; (3) Performance value range verification: It verifies whether the CPU utilization rate is between 0-100% and the network bandwidth is within the physical reasonable range of 0-1000Mbps. Any data record that fails to pass all three layers of verification is regarded as data with a non-compliant format and is directly removed. This verification process prevents invalid or erroneous data caused by collection errors or transmission interference from entering the subsequent analysis process from the data source, which is the primary link to ensure data quality. Without this strict verification, "dirty data" in the original data will directly affect the accuracy of all subsequent statistical analysis and visualization conclusions.

[0047] To address the unavoidable packet loss issue in the CTC system's private network environment, the module implements a differentiated data completion strategy. The process is as follows: First, a continuous analysis of the time series is performed to determine the duration (T) of packet loss and the number of missing data entries (N). Then, different completion algorithms are selected based on the magnitude of T: When T ≤ 1 minute, a sliding window weighted average method is used for completion. Specifically, a window is formed by taking several historical data collection periods before and after the missing point (e.g., 5 periods in total). The historical data point closest to the missing point within the window is assigned a weight of 60%, the nearest future data point (if it exists) is assigned a weight of 30%, and the arithmetic mean of all data within the window is assigned a weight of 10%. The completed value is obtained through weighted calculation. This strategy can effectively smooth data and reflect recent trends in the case of short-term packet loss. When T > 1 minute, a longer communication interruption is considered to have occurred, and a simple window averaging may be distorted. At this point, the module initiates business scenario interpolation completion, which involves correlating with CTC business logs from the same period and dynamically adjusting the parameters of the interpolation function based on the business intensity within that time period (e.g., frequency of dispatch command issuance, train tracking density), ensuring that the trend of the completed values ​​matches the current business load. This differentiated completion mechanism effectively alleviates data discontinuity caused by network issues, providing a complete foundation for time series analysis. Using only general zero-padding or simple linear interpolation would fail to reflect the true characteristics of data in private network packet loss scenarios.

[0048] Furthermore, the interpolation completion for the aforementioned business scenario is achieved through the following steps: First, query the CTC business logs that perfectly correspond to the time period with missing data. Use the sum of the "number of scheduling commands issued" and the "number of route processing requests" per unit time (e.g., per minute) as the "business intensity index K" for that time slice. Second, use piecewise linear interpolation to complete the missing segments. Set a baseline interpolation slope S0 (e.g., determined based on the trend of normal data before and after the missing point). The actual interpolation slope S is dynamically adjusted according to the business intensity, calculated as: S = S0 * (1 + β * K / K_avg), where β is an empirical adjustment coefficient (usually taken as 0.1~0.3), and K_avg is the average business intensity index of the device during the same historical period. Finally, use the dynamic slope S to perform linear interpolation calculations on the missing data. This method ensures that the volatility of the completed data curve matches the actual business load level during the same period.

[0049] Finally, for instantaneous spikes or abnormal fluctuations (outliers) in performance data, the module does not simply use a fixed threshold for removal, but introduces a business correlation verification mechanism. When a sudden change in a performance indicator value is detected, the system automatically correlates and queries records of the same device ID within the same time period in the CTC business event database. The system predefines a list of "high-intensity business events" that need to be correlated, including: "≥5 route processing requests per minute", "continuous train tracking interval less than 80% of the planned interval", and "batch dispatch command issuance (≥3 orders at a time)". If any event record from the list appears in the business event database within a time window Δt before and after the occurrence of the outlier (e.g., Δt = 30 seconds), the outlier is determined to be a "valid outlier" and retained. Otherwise, it is determined to be an "invalid outlier" and filtered out. This approach integrates business knowledge into the data cleaning process, significantly improving the intelligence level of outlier handling and increasing the accuracy of the final data to over 95%. If only statistical methods are used to filter outliers, key performance peaks reflecting real business pressure may be mistakenly deleted, leading to distorted analysis results.

[0050] 2. Specific Implementation of the Domestic Hardware Performance Data Processing Module

[0051] The core function of this module is to deeply process pre-processed data, transforming it from raw, discrete performance readings into structured information rich in business semantics and easy to visualize and understand. Its technical implementation revolves around hierarchical modeling, grouped weighting, time-based processing, and multi-dimensional statistics.

[0052] The module innovatively constructs a three-dimensional hierarchical data model of "Equipment ID - Line Level - Business Type". These three dimensions are the core analytical dimensions in railway CTC operation and maintenance scenarios. Based on the importance and scope of railway business, the system presets weight coefficients for different values ​​under each dimension: for example, in the line level dimension, "busy trunk line" has a weight of 1.2, "ordinary trunk line" 1.0, and "branch line" 0.8; in the business type dimension, "train tracking" has a weight of 1.5, "route processing" 1.2, "dispatch command issuance" 1.0, and "log synchronization" 0.5. The raw data is grouped multiple times according to these three dimensions to form business-meaning subsets. This modeling method completely breaks through the limitations of existing technologies that group data only by a single dimension such as equipment ID, allowing all subsequent statistical analyses to naturally align with the actual scenarios of CTC business, and the calculation results to better reflect the real hardware performance requirements of different businesses and lines. Without this three-dimensional hierarchical model, the analysis results would fail to reflect the differences in railway operations and would make it difficult to pinpoint complex issues such as "a certain type of business causing a bottleneck in the performance of specific equipment on a certain level of line".

[0053] Building upon grouping, the module introduces a time-period identification and differentiated downsampling strategy based on railway timetable characteristics. The system can automatically identify whether a data timestamp belongs to a "daytime busy period" or a "nighttime idle period," which are predefined business cycle characteristics of the railway timetable. For groups with large data volumes (e.g., a single group with more than 1000 data entries), the module performs downsampling to reduce the pressure on visualization rendering and data processing. However, the strategy varies depending on the time period: During "daytime busy periods," linear interpolation downsampling is used. The core algorithm prioritizes retaining local extreme points (peaks and valleys) and inflection points in the original sequence during downsampling, ensuring that key fluctuation information reflecting sudden increases in business pressure or performance bottlenecks is not lost. During "nighttime idle periods," mean downsampling is used, for example, merging five consecutive original data periods to calculate their arithmetic mean as a new data point. This differentiated strategy effectively reduces data redundancy in non-critical periods while ensuring the accuracy of analysis during critical periods. If a uniform downsampling method (such as simple equal-interval extraction) is used for all time periods, important details may be lost during busy periods, or too much meaningless fluctuation noise may be retained during idle periods.

[0054] Next, the module calculates three types of core statistical indicators within each three-dimensional group: (1) Maximum value and related information: not only find the maximum value of the performance indicator, but also record the timestamp of the occurrence of the maximum value, and associate it with the corresponding business type and line level information. (2) Hierarchical weighted average: when calculating the average value, it is not a simple arithmetic average, but the aforementioned weight system is introduced. The specific calculation method is: the weight of each data point = the weight of its line level × the weight of its business type, and then calculate the weighted average of these weighted data. This makes the final average value more accurately reflect the comprehensive performance load level when "more important business is performed on more important lines". (3) Business period average: calculate the arithmetic average of the performance indicators of the group during "busy period" and "idle period" respectively, and use it to compare and analyze the baseline performance under different business intensities. These multi-dimensional statistical indicators provide maintenance personnel with a deep analysis perspective that far exceeds real-time values.

[0055] Finally, the module implements a dual-condition business event association mechanism of "time window + device ID". The system uses the timestamp of performance data and its associated device ID as conditions to search the CTC business event database within a preset time window (e.g., 30 seconds before and after). If a relevant event is matched (e.g., "10:05:23, Device A, issued scheduling command X"), the event information (event type, content, time) is used as a tag and bound to the performance data record at that time point. This process achieves a deep association between "performance values ​​-> statistical characteristics -> business behavior", providing a direct data foundation for subsequent intelligent annotation and correlation analysis by the visualization module. Without this mechanism, the correlation between performance fluctuations and business operations would rely entirely on manual memory and comparison by operations personnel, which is inefficient and prone to errors.

[0056] 3. Specific implementation of the business association visualization module

[0057] The goal of this module is to present the processed data to operations and maintenance personnel in an intuitive, interactive, and professional graphical interface, enabling in-depth correlation analysis of performance and business on the same screen.

[0058] The module uses a standardized canvas 14 inches wide and 10 inches high as the drawing area, with a fixed 4x1 subplot layout. From top to bottom, the four subplots are used to draw: CPU utilization (displaying user-mode and kernel-mode utilization as an overlaid area plot or hyperbola), memory utilization (also overlaid to show physical memory usage and cache usage), network bandwidth (displaying receive and transmit bandwidth as a hyperbola), and disk I / O (displaying read and write speeds as a hyperbola). All subplots share the same horizontal axis (time axis) and support synchronous zooming and panning of the view. This allows operations personnel to easily compare the usage of different hardware resources at the same time and quickly identify coupling relationships between resources (such as whether CPU utilization increases synchronously when network traffic surges).

[0059] In terms of information annotation, the module innovatively applies a "pixel density detection" algorithm to automatically place statistical information annotation boxes. For each sub-image, the algorithm scans approximately 10% of the rectangular area at its upper right or upper left corner. The specific execution process is as follows: First, the candidate area is evenly divided into 10 grids in both the horizontal and vertical directions, forming 100 equally sized candidate cells. Then, each cell is detected sequentially, and the proportion of pixels covered by the performance data curve (or graphic element) to the total number of pixels in the cell is calculated, denoted as "coverage density D". Finally, the algorithm selects all cells with the lowest "coverage density D" value (i.e., the most blank). If multiple cells have the same lowest value, the cell with the shortest Euclidean distance from the corresponding corner point (upper right or upper left corner) of the canvas is selected, and its center coordinates are used as the positioning anchor point of the annotation box. A semi-transparent background highlight box is generated at this position, and the maximum value of the performance indicator within the current view range (displayed in bold red font) and the layered weighted average value (displayed in regular blue font) are clearly displayed within the box. This automated annotation method completely solves the problem that manual annotation can easily obscure key parts of curves (such as peak points), ensuring the readability of information. Without this intelligent positioning, it is necessary to manually adjust or fix the annotation positions; the former increases the user's workload, while the latter may affect the readability of charts.

[0060] For displaying business events, the module uses a combination of a red inverted triangle marker and a black semi-transparent text box. A red inverted triangle marker is drawn on the timeline corresponding to the location of the event. When the user hovers the mouse over or clicks the marker, a semi-transparent text box pops up nearby, displaying detailed information such as the event type, content, and involved devices. Simultaneously, the performance curve segment related to the event's time point is highlighted. This interactive design allows any significant fluctuation on the performance curve to be immediately identified by clicking the nearby marker, revealing its potential business cause and achieving a deep visual connection.

[0061] The module also provides rich interactive controls, including multi-dimensional filters based on device ID, service type, line level, and time range. When a user modifies any filter condition, curves, labels, and event markers in all subgraphs are dynamically updated in real time, displaying only data that meets the criteria. Furthermore, the system has built-in default warning thresholds conforming to CTC operation and maintenance specifications (e.g., CPU utilization ≥80%, memory utilization ≥90%), and allows users to customize these thresholds according to the actual situation of each railway bureau. When a curve exceeds a preset threshold, the excess segment automatically turns red and may be marked with flashing or special icons, helping maintenance personnel instantly locate current or historical risk points from complex curves. Without these multi-dimensional, real-time filtering and warning functions, maintenance personnel would be trapped in an inefficient work mode of manually navigating and comparing massive amounts of data.

[0062] 4. Specific implementation of the CTC system private network adaptation and deployment module

[0063] This module ensures that the aforementioned complex systems can be stably deployed offline in a "ready-to-use" manner within the strictly isolated environment of the railway CTC private network, which runs domestically produced software and hardware.

[0064] The module first builds and manages a "three-level dependency list," which details all the resources required for the system to run: (1) Core function dependencies: such as data processing libraries (e.g., Pandas), visualization rendering engines (e.g., Matplotlib), etc.; (2) Domestic adaptation dependencies: specific library files for domestic operating systems such as Kylin and Tongxin, as well as optimization libraries to ensure compatibility on Phytium and Kunpeng CPUs; (3) CTC business dependencies: including business type encoding tables, line level weight configuration files, and dedicated modules related to CTC log format parsing. Before packaging and building, the system automatically verifies the version compatibility between these dependencies. For any incompatible items found, the system will attempt to automatically match compatible versions from the pre-set alternative version library. If it cannot be resolved automatically, it will issue a clear prompt to the packaging personnel, requiring manual handling. This list-based management mechanism fundamentally solves the typical problem of "running in the development environment but crashing in the deployment environment" caused by missing dependencies or version conflicts.

[0065] Secondly, the module solves the path location problem in private network environments through resource path mapping rules and implicit dependency explicit declaration technology. The system organizes all dependent resource files according to a preset directory structure and references them in the code using relative paths or environment variable-based mapping rules. Simultaneously, it explicitly identifies and packages copies of "implicit" system libraries that are usually provided by default by the operating system environment but may be missing in a streamlined, domestically customized environment. This ensures that even on a clean, domestically-made operating system without complex development environments installed on the target terminal, all necessary files can be accurately found.

[0066] To facilitate distribution and installation via mobile storage media within the dedicated network, the module employs compression optimization techniques during packaging (e.g., keeping the compression ratio below 60%) to reduce the size of the final executable file or installation package. More importantly, through static linking compilation or embedding an independent runtime environment (e.g., packaging the Python interpreter and related libraries together), the entire system and all its dependencies are encapsulated into one or more independent executable files. This allows the final software tool to run on target domestically produced terminals without requiring an internet connection to download any additional components or pre-installing specific runtimes or databases at the system level, achieving true "completely offline operation" and perfectly meeting the deployment requirements of physical isolation and security compliance within the railway dedicated network.

[0067] Finally, the chart export function has been deeply customized to conform to railway operation and maintenance habits. The default save path for exported files is set to the same directory as the original CSV data file, facilitating data classification and management. Upon saving, the interface displays existing files with the same name in the target folder and provides a standard overwrite confirmation prompt to prevent accidental operations. Exported images are in high-definition PNG format and offer multiple resolution options such as 150, 200, 300, and 600 DPI to meet different quality requirements from screen viewing to printed archiving. When generating exported images, the module uses chart boundary adaptive adjustment technology to ensure that statistical information annotation boxes previously placed at the canvas edge by intelligent algorithms appear completely in the final output image, without information loss due to rendering boundary cropping. This series of detailed processing allows the tool's output to seamlessly integrate into existing railway operation and maintenance reports and workflows.

[0068] This embodiment addresses technical challenges in CTC system performance monitoring of domestically produced hardware, including poor compatibility with general tools, data and business disconnect, weak visualization analysis, and difficulties in dedicated network deployment. It effectively solves these problems through a series of collaborative, specialized technical means. First, the core information identification and data adaptation module employs automatic field mapping rules based on hardware type identification and a three-layer verification system. Combined with differentiated completion based on packet loss duration and outlier filtering based on business association, it directly overcomes the data distortion problems caused by inconsistent data formats and poor transmission quality of domestically produced hardware, significantly improving data accuracy and providing a reliable data foundation for subsequent analysis. Second, the domestically produced hardware performance data processing module constructs a three-dimensional hierarchical weight model of "device ID-line level-business type," uses time-period identification and differentiated downsampling strategies based on railway timetables, and establishes a dual-condition business event association of "time window + device ID." This ensures that data processing deeply aligns with CTC business logic, achieving precise binding between performance fluctuations and business behavior. Subsequently, the business-related visualization module utilizes standardized multi-subgraph layout, intelligent pixel density detection and annotation algorithms, interactive business event combination marking, and multi-dimensional real-time filtering and threshold warnings to transform processed data into an intuitive and interactive deep business insight view, improving the efficiency of operations and maintenance personnel in locating performance bottlenecks. Finally, the CTC system private network adaptation and packaging deployment module ensures stable offline operation of the system in domestic terminals and isolated private network environments by constructing a three-level dependency list, implementing resource path mapping and compression optimization, and using static linking to encapsulate it into an independent executable file. Its professional export function, conforming to railway operations and maintenance habits, further guarantees the tool's practicality and scalability. This entire technical solution is organically integrated, fundamentally enabling full-link empowerment of domestic hardware performance from accurate data collection, business analysis, deep visualization to stable deployment, significantly improving the CTC system's operations and maintenance capabilities.

[0069] Example 2

[0070] This embodiment discloses a method for visualizing the performance monitoring data of domestically produced hardware in a CTC system. This method transforms the raw performance data of domestically produced hardware into a visual insight that is deeply related to business through a series of logically rigorous steps, and ensures stable deployment in a private network environment.

[0071] Step S1: Data Preprocessing and Augmentation

[0072] The core objective of this step is to clean, complete, and validate the raw collected data to form a high-quality, highly reliable analytical foundation. The specific working process and principles are as follows:

[0073] First, the system reads a CSV file containing hardware type identifiers. These identifiers are pre-installed metadata in the data file, such as "FT_KYLIN". Based on these identifiers, the system automatically matches and loads the corresponding dedicated field mapping rules from a pre-configured rule base. These rules are essentially a dictionary, mapping the specific performance counter names output by different hardware platforms to unified, semantically clear performance parameters within the system. For example, the Phytium CPU-specific fields "CPU_UTIL_USR" and "CPU_UTIL_SYS" are mapped to "CPU user mode utilization" and "CPU kernel mode utilization," respectively; and the Kylin system kernel report's "MEM_CACHED" is mapped to "memory page cache usage." This mapping process solves the primary problem of inconsistent data formats across multiple heterogeneous sources. Without this step, data formats would need to be manually parsed for each type of hardware, making automated adaptation impossible, inefficient, and prone to errors.

[0074] Next, the system executes a three-layer verification system. This is not a simple format check, but a structured verification based on railway industry standards and physical limits: (1) Equipment ID verification: verifying whether it conforms to the industry standard pattern of "railway bureau code-station code-equipment number" to ensure that the data can be traced back to the specific physical equipment; (2) Business type verification: confirming whether the value of the business type field is strictly limited to the four types of CTC core business enumeration sets of "route processing, train tracking, dispatching order issuance, and log synchronization", and filtering irrelevant or incorrect tags; (3) Performance value verification: setting a reasonable range based on physical common sense (such as CPU utilization ∈ [0,100]) and eliminating obviously impossible outliers (such as negative numbers or extreme values). Any record that fails any verification is discarded. This process ensures the compliance and basic quality of the data from the source. If it is missing, all subsequent advanced analysis will be based on potentially large amounts of "dirty data", and the credibility of the conclusions will be seriously damaged.

[0075] To address the common data packet loss issue in private network environments, this step implements a differentiated data completion strategy. The principle is to determine the nature of the missing data based on the duration of packet loss, and then employ different mathematical models. Specifically: the system first analyzes the continuity of the time series, identifies the time window of missing data, and calculates its duration T.

[0076] When T ≤ 1 minute, it is determined to be a short-term random packet loss. In this case, a sliding window weighted average is used for completion. A fixed-size window (e.g., containing 5 collection periods) is formed by taking several periods before and after the missing point. The completed value is calculated by weighting the data within the window, where: the historical data point closest to the missing point is given the highest weight (e.g., 60%) to reflect the nearest state; the next most recent data point (if it exists) is given the second highest weight (e.g., 30%) to reflect the trend; and the arithmetic mean of all data within the window is given the base weight (e.g., 10%) as a smoothing background. This method can effectively reconstruct the local continuity of data under short-term interruptions.

[0077] When T > 1 minute, it is considered a prolonged communication interruption. Simple time-series interpolation may deviate from the actual business load. Therefore, business scenario-based interpolation is used for completion. The system will query CTC business logs within the same time period and dynamically adjust the parameters of the interpolation function (such as piecewise linear or polynomial interpolation) based on the business intensity at that time (such as command issuance frequency and train tracking density) to match the overall trend of the completed data with the business activity level of the same period. If this differentiated completion is not performed and a single method (such as filling with all previous values ​​or linear interpolation) is used, it will not be able to accurately reflect the true characteristics of the data under the packet loss scenario of the private network, resulting in distortion of subsequent trend analysis.

[0078] Finally, for outliers in the data (such as transient spikes), this step introduces a business-related filtering mechanism. Its working principle is as follows: when a drastic change in a performance metric is detected, the system automatically queries the CTC business event database within a preset time tolerance window (e.g., ±15 seconds), using the timestamp and device ID of that data point as keys. If a related high-intensity business event record exists within this window (e.g., the system is processing a batch of route requests during a CPU usage peak), the outlier is determined as a "valid outlier" and retained, as it is likely a manifestation of real business pressure. Conversely, if no related business event exists, the outlier is determined as an "invalid outlier" (possibly originating from transient hardware disturbances or acquisition noise) and discarded. This mechanism integrates domain knowledge into data cleaning, significantly improving data accuracy. If only statistical methods (such as the 3σ principle) are used for filtering, key performance peaks reflecting real business pressure will be incorrectly removed, leading to analysis conclusions that deviate significantly from reality.

[0079] Step S2: Business Data Processing and Association

[0080] This step aims to reconstruct, condense, and enhance the cleaned data according to the business logic of the railway CTC, producing information units rich in business semantics.

[0081] The process begins with the system constructing a three-dimensional hierarchical model of "Equipment ID - Line Level - Service Type." This is a process of reorganizing the original data space. The Equipment ID identifies the specific hardware; the Line Level (e.g., busy trunk line, ordinary trunk line, branch line) and Service Type (e.g., train tracking, route processing) inject the dimension of service importance. The system pre-sets weights for the enumerated values ​​under each dimension (e.g., 1.2 for busy trunk lines, 1.5 for train tracking services). The original data is grouped multiple times according to these three dimensions, forming subsets with clear business significance. This modeling approach breaks through the limitations of traditional monitoring that only groups by equipment, allowing subsequent calculations to naturally align with business scenarios. Without this model, data analysis results would fail to reflect the "differential pressures exerted on hardware by different services on lines of different levels," making it difficult to locate complex service-related bottlenecks.

[0082] Based on the grouping, the system automatically labels each data point with either "daytime busy period" or "nighttime off-peak period" according to a predefined railway timetable. Subsequently, for large groups (e.g., more than 1000 data entries), differentiated downsampling is performed to balance accuracy and performance.

[0083] During peak periods, linear interpolation downsampling is employed. This algorithm, while reducing data density, prioritizes identifying and preserving local extrema (peaks, troughs) and trend inflection points in the original sequence. The principle is to compare the slope changes of adjacent points to ensure that the downsampled sequence still clearly reflects performance spikes and key fluctuations caused by peak business periods.

[0084] During off-peak hours, mean downsampling is employed. For example, five consecutive raw data points are merged, and their arithmetic mean is calculated as a new data point. This effectively smooths out minor fluctuations and noise during low-load periods at night, significantly reducing the amount of data without losing important trend information.

[0085] This differentiated approach ensures that analytical resources are focused on critical details during key periods. Using a uniform, simple, equal-interval sampling method might result in the loss of important details during busy periods or the retention of excessive meaningless noise during quieter periods.

[0086] Subsequently, within each three-dimensional group, the system calculates multi-dimensional statistical indicators: (1) Maximum value and associated context: Find the maximum value of the performance indicator and synchronously record the timestamp of its occurrence, as well as the corresponding business type and line level at that time, to form a complete context; (2) Hierarchical weighted average: When calculating the average value, the aforementioned line weight and business weight are introduced. The comprehensive weight of each data point = line weight × business weight. The final weighted average reflects the comprehensive performance load level when "more important business is performed on more important lines", which is more business-instructive than the simple arithmetic average; (3) Business time period average: Calculate the arithmetic average of the group during the "busy" and "idle" periods respectively to establish a performance baseline and facilitate the detection of abnormal deviations.

[0087] Finally, the system performs a dual-condition business event association based on "time window + device ID". Using the timestamp of each performance data record and its associated device ID as the joint key, the system searches the CTC business event database within a preset matching time window. If a matching event record is found (e.g., a scheduling command is issued on the same device at a time when CPU utilization increases), the event's detailed information (type, content) is used as a tag and strongly associated with this performance data record. This process automatically establishes a causal or accompanying relationship chain between "performance metric fluctuations" and "specific business operations," providing a direct data foundation for the deep visualization association in step S3. Without this automated association, the entire process relies on manual memory and cross-referencing by operations personnel, which is extremely inefficient and prone to omissions.

[0088] Step S3: Visualizing Deep Relationships

[0089] This step transforms the processed data into intuitive, interactive, and insightful visualizations that directly support operational decisions.

[0090] The system first uses a 4x1 column subplot layout on a standardized canvas (e.g., 14×10 inches) to plot CPU utilization (overlaying user-mode and kernel-mode), memory utilization (overlaying physical memory and cache), network bandwidth (receive and transmit hyperbolas), and disk I / O (read and write hyperbolas). All subplots share a unified timeline and support synchronous zooming and panning of the view. This allows operations personnel to clearly observe the interaction between different hardware resources at the same business time and quickly determine whether there are resource bottlenecks.

[0091] In terms of information presentation, the system employs a pixel density detection algorithm for intelligent annotation. For each sub-image, the algorithm automatically scans approximately 10% of its display area, either the upper right or upper left corner. By calculating the pixel density covered by existing data curves within this area, it intelligently selects a location with the largest blank area that least obscures the key curve trend. At this location, the system generates a highlighted annotation box with a semi-transparent background, clearly displaying two core statistical summaries within the box: the maximum value of the indicator within the current view's time frame (highlighted in bold red) and the tiered weighted average (displayed in regular blue). This automated approach perfectly solves the problem of manually placing annotations that easily obscure key graphic elements, ensuring the clear readability of statistical information. If fixed-position annotations were used, it could severely interfere with the reading of the main image at different zoom levels.

[0092] For business events, the system uses a combination of a red triangle marker and a semi-transparent black text box for visualization. A prominent red inverted triangle marker is drawn at the location of the event on the timeline. When the user hovers the mouse over or clicks on this marker, a semi-transparent text box dynamically pops up nearby, displaying the full event details. Simultaneously, the performance curve segment related to the event's time point is highlighted. This design achieves "what you see is what you get"—operations personnel can immediately understand the potential business causes behind any abnormal fluctuations in the performance curve through interaction, achieving a deep visual integration of performance and business.

[0093] Furthermore, the system provides multi-dimensional filtering controls based on device ID, service type, line level, and time range. All filtering conditions support real-time linkage, and the chart content updates dynamically accordingly. The system also has built-in warning thresholds that conform to industry standards (e.g., CPU ≥ 80%), and allows users to customize them. When the performance curve exceeds the threshold, the exceeding segment will automatically turn red and may be accompanied by a warning icon, helping operations and maintenance personnel instantly locate historical and current risk points in complex multi-curve charts. The lack of such interactive and warning functions would force operations and maintenance personnel into an inefficient mode of manually searching for information in complex static charts.

[0094] Step S4: Private Network Adaptation Packaging and Deployment

[0095] This step ensures that the above-mentioned fully functional analysis method can be transformed into an independent tool that can run stably and offline on a strictly isolated railway CTC private network and various domestically produced terminals.

[0096] The system first constructs and manages a detailed three-level dependency list, which is essentially a "recipe" to ensure the software can run reproducibly: (1) Core function dependency list: listing basic libraries such as data processing (e.g., Pandas, NumPy) and visualization rendering (e.g., Matplotlib) and their precise versions; (2) Domestic adaptation dependency list: specifying specific library files for domestic operating systems such as Kylin and Tongxin, as well as dedicated components optimized and compiled on Phytium and Kunpeng CPU architectures; (3) CTC business dependency list: including resources closely related to railway business logic such as business coding tables, line weight configuration files, and log parsing plugins. Before packaging, the system automatically verifies the version compatibility between these dependencies and attempts to automatically match alternative versions or provide clear handling prompts for conflicting items. This list-based management fundamentally eliminates the problem of "normal development environment, crashing deployment environment" caused by environmental differences.

[0097] Secondly, resource path mapping rules are used to convert resource paths referenced in the code (such as icons and configuration files) into relative paths to the executable file or resolvable paths based on specific environment variables. Simultaneously, implicit dependencies are explicitly declared, clearly listing and packaging system-level library files that are common in general development environments but may be missing in streamlined domestic system environments. This ensures that all necessary files can be accurately found on the target terminal.

[0098] To facilitate distribution via mobile media within the private network, compression optimization techniques (controlling the compression ratio) are employed during the packaging process to reduce the size of the final deliverable. Most importantly, through static linking compilation or embedding an independent runtime environment (such as packaging the Python interpreter and its core libraries together), the entire analysis method and all its dependencies are encapsulated into one or more independent executable files. This allows the final tool to run on target domestically produced terminals without needing to connect to an external network to download components, nor requiring any complex runtimes or databases to be pre-installed at the system level, achieving truly complete offline operation and fully meeting the physical isolation and security requirements of the private network.

[0099] Finally, the chart export function is deeply adapted to railway operation and maintenance habits: the export path defaults to the same directory as the source data file; a conflict warning for files with the same name is displayed when saving; exported images offer multiple DPI (e.g., 150 / 200 / 300 / 600) PNG format options to meet different clarity requirements; and all statistical information boxes added through intelligent annotation are ensured to be complete and clearly visible in the exported images. This allows the analysis results to be seamlessly integrated into the railway's existing reporting and archiving processes, improving the tool's usability and acceptance.

[0100] This invention systematically solves the technical challenges of inaccurate data, business disconnect, difficulty in insight, and complex deployment in CTC domestic hardware performance monitoring through four closely linked steps. In step S1, automatic mapping of hardware type identifiers, three-layer structured verification, differentiated completion based on packet loss duration (short-term window weighted average, long-term business interpolation), and outlier filtering related to business fundamentally ensure the accuracy of input data and business consistency, increasing data reliability to over 95% and laying a solid foundation for subsequent analysis. Step S2 constructs a three-dimensional weighted model of "equipment-line-business" and performs time-differentiated downsampling according to the railway timetable (peak-keeping interpolation during busy periods, merging averages during idle periods). Simultaneously, it calculates weighted statistical indicators and automatically associates them with business event tags, ensuring that the data processing results deeply align with CTC business logic and achieving precise binding between performance fluctuations and business behavior. Step S3 utilizes a multi-subgraph synchronous layout, a pixel density detection-based intelligent annotation algorithm (automatically optimizing and locating statistical values), interactive business event combination marking, and multi-dimensional real-time filtering and threshold warnings to transform the processed data into an intuitive and interactive deep business insight view, improving the efficiency of operations and maintenance personnel in locating performance bottlenecks by more than 40%. Finally, step S4 ensures that the entire method can achieve 99% stability offline operation in domestic terminals and isolated private network environments by constructing and managing a three-level dependency list, implementing resource path mapping and compression optimization, and using static linking to encapsulate and generate an independent executable file. Its professional export function, conforming to railway operations and maintenance habits, further guarantees the tool's practicality and scalability. The entire method is interconnected, realizing a closed-loop process from raw data to business insights to stable delivery, significantly improving the CTC system's operational intelligence level and assurance capabilities.

[0101] 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 principles of 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 claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A data visualization system for monitoring the performance of domestically produced CTC system hardware, characterized in that, It includes a core information identification and data adaptation module, a domestic hardware performance data processing module, a business association visualization module, and a CTC system private network adaptation and packaging deployment module, which are connected in sequence. The core information identification and data adaptation module is used to read a CSV file containing hardware type identifiers, automatically match domestic hardware combinations and load corresponding field mapping rules, parse exclusive performance parameters, and perform legality verification of device ID, service type and performance value based on a three-layer verification system, perform differential completion of packet loss data, and perform business association filtering on abnormal values. The domestically produced hardware performance data processing module is used to construct a three-dimensional hierarchical model of "equipment ID-line level-service type", perform multi-round grouping based on a preset weight system, identify service periods according to the railway operation diagram and perform differentiated downsampling, calculate multi-dimensional statistical indicators, and associate with the CTC service event library through a dual-condition matching mechanism of "time window + equipment ID" to tag the performance data with service. The business association visualization module is used to display multi-indicator performance data in the form of multiple subgraphs on a standardized canvas, intelligently annotate statistical information using a pixel density detection algorithm, display business events through combined markers, configure multi-dimensional filtering controls and warning thresholds, and realize the deep association display between performance data and CTC business. The CTC system private network adaptation and packaging deployment module is used to build a three-level dependency list system, verify version compatibility, generate executable files that can run offline through resource path mapping and compression optimization technology, and support the export of charts according to railway operation and maintenance habits.

2. The system according to claim 1, characterized in that, The core information identification and data adaptation module includes: The hardware type matching unit is used to automatically match the Phytium CPU + Kylin system and Kunpeng CPU + Kylin system combinations based on the hardware type identifier in the CSV file, and load the corresponding exclusive field mapping rules. The three-layer verification unit is used to verify whether the device ID conforms to the format of "railway bureau code-station code-device number", whether the business type is one of the four categories: route processing, train tracking, dispatching command issuance, and log synchronization, and whether the performance value is within the preset physical range. The data completion unit is used to complete data using a sliding window weighted average when the packet loss duration is ≤1 minute, and to complete data using interpolation based on the business scenario when the packet loss duration is >1 minute. The anomaly filtering unit is used to verify numerical mutations in relation to the CTC business log library, retaining only outliers related to specific business operations.

3. The system according to claim 1, characterized in that, The domestically produced hardware performance data processing module includes: The 3D hierarchical modeling unit is used to construct a 3D hierarchical model of device ID, line level, and service type, and preset weights based on line level and service type; The time period identification and downsampling unit is used to identify busy daytime periods and idle nighttime periods. Differential downsampling is performed on groups with more than 1,000 data entries: linear interpolation downsampling is used for busy periods, and mean downsampling is used for idle periods. The statistical calculation unit is used to calculate the maximum value and related information of each layer, the layer-weighted average value, and the average value for the business period. The business event association unit is used to match the business event library based on "time window + device ID" and tag the data with business tags.

4. The system according to claim 1, characterized in that, The business association visualization module includes: Multi-subgraph layout units are used to display CPU usage, memory usage, network bandwidth, and disk I / O curves in a 4x1 column layout on a 14×10 inch canvas, and share a unified timeline. The intelligent annotation unit is used to generate semi-transparent annotation boxes in the blank areas of the sub-image using a pixel density detection algorithm, displaying the maximum value and the weighted average value. The business event annotation unit is used to display business events in the form of a combination of red triangle markers and semi-transparent text boxes, and supports clicking to highlight and view details; The filtering and early warning unit provides multi-dimensional filtering controls based on device ID, service type, line level, and time range, and supports threshold early warning and curve highlighting.

5. The system according to claim 1, characterized in that, The CTC system private network adaptation and deployment module includes: The dependency list management unit is used to build a three-level list of core function dependencies, domestic adaptation dependencies, and CTC business dependencies, and to verify version compatibility. The path mapping and compression unit is used to implement resource path mapping, declare implicit dependencies, and reduce file size through compression optimization; An offline runtime encapsulation unit is used to embed an independent runtime environment and use static linking to generate an executable file that supports full offline execution; Export custom cells, which are used to associate the export path with the CSV file directory by default, and support PNG format export with multiple DPI settings and adaptive adjustment of chart boundaries.

6. A method for visualizing the performance monitoring data of domestically produced CTC system hardware, characterized in that, Includes the following steps: Step S1: By reading the CSV file containing hardware type identifiers, automatically match domestic hardware combinations, load exclusive field mapping rules, parse exclusive performance parameters, and perform three-layer verification of device ID, service type, and performance value, perform differentiated completion on packet loss data, and perform business association filtering on outliers; Step S2: Construct a three-dimensional hierarchical model of "equipment ID-line level-business type", group the data based on preset weights, identify business time periods according to the railway operation diagram and perform differentiated downsampling, calculate multi-dimensional statistical indicators, match the business event database through "time window + equipment ID" and label the data with business tags; Step S3: Display performance data in the form of multiple subgraphs on a standardized canvas, use a pixel density detection algorithm to intelligently label statistical information, display business events through combined tags, configure multi-dimensional filtering and early warning mechanisms, and realize the visualization of the relationship between performance and business. Step S4: Construct a three-level dependency list, verify compatibility, and generate an executable file that can run offline through resource path mapping and compression optimization, supporting the export of charts according to railway operation and maintenance specifications.

7. The method according to claim 6, characterized in that, The differential completion in step S1 specifically includes: If the packet loss duration is ≤1 minute, a sliding window weighted average is used for data completion. The window contains 5 collection periods, with the weights allocated as follows: 60% for the most recent historical data, 30% for the most recent subsequent data, and 10% for the window mean. If the packet loss duration is greater than 1 minute, interpolation based on the business scenario will be used to complete the value, and the value will be dynamically adjusted in conjunction with the intensity of business operations in the same period.

8. The method according to claim 6, characterized in that, The differential downsampling in step S2 specifically includes: Identify busy daytime hours and quiet nighttime hours; For groups with more than 1000 data entries, linear interpolation downsampling is used during busy periods to preserve peak values ​​and inflection points; During idle periods, mean downsampling is used to merge 5 collection cycles into 1 data point.

9. The method according to claim 6, characterized in that, The intelligent annotation in step S3 specifically includes: Scan the top right or top left 10% area of ​​the sub-image, and generate a semi-transparent highlighted annotation box for the area without curve coverage. The maximum value is displayed in bold red font in the label box, and the weighted average value is displayed in regular blue font.

10. The method according to claim 6, characterized in that, The three-level dependency list in step S4 includes: A list of core functional dependencies, covering the resources required for data processing and visualization rendering; The list of domestically adapted dependencies covers Kylin / Tongxin operating systems and Phytium / Kunpeng CPU adaptation resources; The CTC business dependency list covers the resources required for railway business logic and event correlation.

Images