High-speed train operation state acquisition method and system based on multi-source data fusion

Abstract

The invention relates to the technical field of high-speed train operation state collection, in particular to a high-speed train operation state collection method and system based on multi-source data fusion, and the method comprises the steps: obtaining sensor data in real time, classifying the sensor data into instantaneous dynamic and long-term static data, predicting the change trend of the operation state in combination with multi-source data, and adjusting prediction parameters. The deviation value and the sensor rate of change are analyzed to locate the source of anomalies, and an adjustment suggestion is generated. The system comprises a data acquisition unit, a data analysis unit, a parameter adjustment unit and an abnormality diagnosis unit, wherein the abnormality diagnosis unit comprises a self-learning module optimization strategy. The adaptability and the intelligent level of the running state acquisition system are improved through multi-source data fusion, early warning and anomaly recognition are achieved, safe and stable running of the train is guaranteed, and the maintenance cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of high-speed train operation monitoring and data acquisition, and in particular to a high-speed train operation status acquisition method and system based on multi-source data fusion. Background Art

[0002] In modern rail transit systems, collecting information on the operating status of high-speed trains is crucial for ensuring operational safety and improving efficiency. However, due to the complex and ever-changing operating environment of high-speed trains, their operating status is easily affected by a variety of external factors, such as track conditions, climate factors, and the distribution of surrounding objects, resulting in a certain degree of variability in their operating status. Traditional operating status monitoring methods typically rely on periodic inspections or data collection from a single sensor. This approach has limited adaptability to dynamic environments and may not fully reflect the actual operating status of trains in different environments, thus affecting driving safety and comfort.

[0003] While some existing operating status acquisition systems have begun to incorporate inertial measurement units (IMUs) or simple positioning modules, using sensors to acquire some train operating status parameters and perform basic analysis, these methods often lack the comprehensive collection and in-depth integration of both the train's operating environment data (e.g., distance to objects near the track, altitude) and the train's own operating status data (e.g., acceleration, angular velocity). Furthermore, traditional acquisition strategies typically process data from only a single time period, failing to fully integrate the classification, processing, and comprehensive prediction of short-term dynamic data (e.g., instantaneous changes in the train's operating status) and long-term static data (e.g., geographic information map data). This results in the accuracy of operating status data analysis lagging behind actual demand changes, making efficient status assessment and prediction impossible.

[0004] More importantly, existing technologies have limitations in integrating train operating status with environmental analysis. For example, in complex environments, even minor changes in train operating status can trigger potential safety hazards. Current technologies typically only investigate and adjust after a significant anomaly occurs. They lack real-time comparative analysis of multiple data changes during train operation, making it difficult to identify potential risk sources at an early stage, thus limiting overall analysis efficiency.

[0005] High-speed train operating status acquisition systems are susceptible to external interference from track conditions, climate factors, and other factors in complex and changing operating environments, compromising the comprehensiveness and accuracy of data collection. Traditional data acquisition methods typically rely on single sensors or post-analysis, making it difficult to monitor and predict the train's operating status in real time under varying conditions, thus impacting operational safety and stability. To improve the reliability and adaptability of operating status acquisition systems, it is necessary to combine multi-source data acquisition and intelligent analysis technologies to comprehensively monitor and dynamically assess the train's operating status. For example, patent application CN116360468A discloses an environmentally adaptive method for controlling the motion and operating status of an undulating fin robot. This method includes a host computer module, an environment recognition module, and an operating status control module. The host computer module is the core of the control system, responsible for sending commands and processing data. The environment recognition module captures images of the robot's environment, extracts feature values, and uses neural network pattern recognition to feed the results back to the host computer. The operating status control module adjusts the undulating fin robot's operating status based on the host computer's instructions. This operating state control method can ensure that the undulating fin robot can adaptively adjust its operating state in different land environments to generate maximum thrust. However, this method mainly adaptively adjusts the operating state through environmental identification, but does not combine prediction bias correction and machine learning optimization. It fails to make a comprehensive judgment based on train operating environment data (such as the distance to objects near the track and the altitude), and lacks the ability to provide early warning of changes in train operating states in complex environments, resulting in certain limitations in practical applications.

[0006] Therefore, a method and system for collecting high-speed train operating status based on multi-source data fusion is urgently needed. This method, through comprehensive collection and intelligent analysis of train operating status data, operating environment data, and geographic location information, can achieve dynamic monitoring and early warning of train operating status. This method not only improves the comprehensiveness and accuracy of data collection, but also significantly enhances adaptability to complex environments, providing strong support for the safe and efficient operation of high-speed trains. Summary of the Invention

[0007] The purpose of the present invention is to overcome the deficiencies in the prior art, solve or at least alleviate the problem that the high-speed train operating status acquisition system in complex environments cannot fully adapt to dynamic environmental changes and lacks early warning capabilities, and provide a high-speed train operating status acquisition method and system based on multi-source data fusion.

[0008] To achieve the above object, the present invention provides the following technical solution: a method for collecting high-speed train running status based on multi-source data fusion, comprising the following steps: S1. Acquire output data of various sensors during train operation in real time, and divide the data into instantaneous dynamic data and long-term static data; S2. Predict the changing trend of the train operation status based on instantaneous dynamic data and long-term static data, and revise the prediction results based on current data; S3. During the train running state prediction process, calculating the deviation between the predicted running state and the actual running state; S4. Real-time monitoring of the rate of change of sensor output data, and judging whether there is an abnormal state based on the deviation value. If there is an abnormality, further analysis of the change pattern of each sensor data to locate the source of the abnormality; S5. Generate adjustment suggestions based on the source of the anomaly and record relevant data for optimizing the operation status collection strategy.

[0009] Preferably, in step S1, the weight of the instantaneous dynamic data is determined by the fluctuation trend of the train's instantaneous acceleration, angular velocity and the distance information of nearby objects; the weight of the long-term static data is determined by the track geographic information, the historical average value and the fluctuation range of the altitude; the sum of the weight of the instantaneous dynamic data and the weight of the long-term static data is 1; the change in the train's running state is equal to the product of the weight of the instantaneous dynamic data and the change in the running state caused by the instantaneous dynamic data plus the product of the weight of the long-term static data and the change in the running state caused by the long-term static data.

[0010] Preferably, step S2 includes the following steps: S201, calculating the running state change amount, and predicting the train running state change amount based on instantaneous dynamic data and long-term static data; S202, adjusting the operating state prediction parameter once so that the adjusted operating state change is less than the operating state change in step S201; S203 , adjusting the operating state prediction parameter n times, and adjusting the operating state prediction parameter again according to the change between the instantaneous dynamic data of the current time window and the instantaneous dynamic data of the previous time window.

[0011] Preferably, the step S2 further includes the following steps: S204, reset and adjust. When the difference between the long-term static data of the current time window and the long-term static data of the previous time window exceeds the threshold, increase the weight of the long-term static data and enter step S201.

[0012] Preferably, step S3 includes the following steps: S301, calculating the theoretical operating state change amount, and calculating the theoretical operating state change amount according to the operating state prediction parameter adjustment amount of the current time window; S302, correcting the theoretical operating state change according to the instantaneous dynamic data and long-term static data of the next time window; S303: Calculate the deviation value, that is, calculate the deviation value between the corrected theoretical running state change amount and the actual running state change amount in the next time window.

[0013] Preferably, step S4 includes the following steps: S401, constructing a correlation matrix, based on the causal relationship between the train running status and the sensor data, constructing the correlation matrix of each sensor; S402: Construct a change rate matrix, obtain the monitoring data of each sensor in real time, calculate the difference between the monitoring data of the current time window of each sensor and the monitoring data of the previous time window, and obtain the change of the monitoring data of each sensor. The ratio of the change of the monitoring data of each sensor to the monitoring data of the previous time window of each sensor is the actual change rate of the monitoring data of each sensor. The actual change rate of the monitoring data of each sensor is entered into the correlation matrix in step S401 to obtain the change rate matrix; S403: Once positioned, obtain the operating state prediction parameter adjustment ratio based on the operating state prediction parameter adjustment amount, obtain the theoretical change rate of each sensor monitoring data, calculate the change rate difference between the theoretical change rate of each sensor monitoring data and the actual change rate of each sensor monitoring data, and the ratio of the change rate difference to the theoretical change rate is the change rate ratio. When the change rate ratio of a sensor is greater than a threshold, it is determined that the sensor is abnormal. If the change rate ratios of all sensors are less than the threshold, the process proceeds to step S404. S404: Secondary positioning, calculating the mean difference between each data in the rate of change matrix and its adjacent data, and determining that the sensor corresponding to the largest mean difference in the rate of change matrix has an abnormality.

[0014] Preferably, step S5 includes the following steps: S501. Generate adjustment suggestions based on the abnormal sources, giving priority to high-risk sources; S502. Record the source of the abnormality and related data, and establish an operation status collection database; S503: Optimize the operation status prediction parameter adjustment strategy using the operation status collection database.

[0015] Preferably, in step S5, while performing steps S501 to S503, the following steps are also performed: S504: Monitor the changing trend of the sensor output data. When a certain parameter continuously deviates from the normal range, generate a prompt message and notify the operation and maintenance personnel.

[0016] A high-speed train operation status acquisition system based on multi-source data fusion is applicable to the above-mentioned operation status acquisition method. The operation status acquisition system includes a data acquisition unit, a data analysis unit, a parameter adjustment unit and an abnormality diagnosis unit; the data acquisition unit is used to obtain the output data of multiple sensors during the train operation in real time; the data analysis unit is used to classify and process instantaneous dynamic data and long-term static data, and calculate the change in the train operation status; the parameter adjustment unit is used to dynamically adjust the operation status prediction parameters; the abnormality diagnosis unit is used to analyze the change rate of the sensor output data, locate the source of the abnormality and generate adjustment suggestions.

[0017] Preferably, the abnormality diagnosis unit also includes a self-learning module, which trains historical collected data through a machine learning algorithm to gradually optimize the operating status prediction parameter adjustment strategy and abnormality diagnosis accuracy; the self-learning module uses a random forest algorithm to perform model optimization.

[0018] The beneficial effects of the present invention are: The present invention combines the output data of multiple sensors during the train operation process, and uses a combination of instantaneous dynamic data and long-term static data to predict the changes in the train's operating status at the next moment. It also adjusts the operating status prediction parameters in advance according to the prediction results to maintain the train's operating status within the set range, thereby improving the adaptability and operating efficiency of the operating status acquisition system and ensuring driving safety and stability.

[0019] While dynamically adjusting the operating state prediction parameters, the present invention realizes early identification of potential anomalies by analyzing the rate of change of sensor output data, thereby reducing the risk of anomalies and maintenance costs.

[0020] The self-learning module of the present invention optimizes the acquisition strategy through machine learning algorithms, thereby improving the intelligence level and long-term reliability of the operating status acquisition system. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is a flow chart of the operating status acquisition method of the present invention.

[0022] Figure 2 This is a flow chart of step S1 of the operating status collection method of the present invention.

[0023] Figure 3 This is a flow chart of step S2 of the operating status acquisition method of the present invention.

[0024] Figure 4 This is a flow chart of step S3 of the operating status collection method of the present invention.

[0025] Figure 5This is a flow chart of step S4 of the operating status collection method of the present invention.

[0026] Figure 6 This is a flow chart of step S5 of the operating status collection method of the present invention.

[0027] Figure 7 Schematic diagram of the architecture of the operating status acquisition system of the present invention.

[0028] Figure 8 Schematic diagram of the architecture of the abnormality diagnosis unit of the operation status acquisition system of the present invention.

[0029] The reference numerals are as follows: 1. Data acquisition unit; 2. Data analysis unit; 3. Parameter adjustment unit; 4. Abnormality diagnosis unit; 5. Self-learning module. DETAILED DESCRIPTION

[0030] In the description of the present invention, it should also be noted that, unless otherwise expressly specified or limited, the terms "disposed," "installed," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; they may refer to mechanical connections or electrical connections; they may refer to direct connections or indirect connections through an intermediate medium; and they may refer to internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0031] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work shall fall within the scope of protection of the present invention.

[0032] Example 1 While some existing operating status acquisition systems have begun to incorporate inertial measurement units (IMUs) or simple positioning modules, using sensors to acquire some train operating status parameters and perform basic analysis, these methods often lack the comprehensive collection and in-depth integration of both the train's operating environment data (e.g., distance to objects near the track, altitude) and the train's own operating status data (e.g., acceleration, angular velocity). Furthermore, traditional acquisition strategies typically process data from only a single time period, failing to fully integrate the classification, processing, and comprehensive prediction of short-term dynamic data (e.g., instantaneous changes in the train's operating status) and long-term static data (e.g., geographic information map data). This results in the accuracy of operating status data analysis lagging behind actual demand changes, making efficient status assessment and prediction impossible.

[0033] Reference Figure 1The present invention provides a method for collecting the running status of a high-speed train based on multi-source data fusion, comprising the following steps: S1. Acquire output data of various sensors during train operation in real time, and divide the data into instantaneous dynamic data and long-term static data; S2. Predict the changing trend of the train operation status based on instantaneous dynamic data and long-term static data, and revise the prediction results based on current data; S3. During the train running state prediction process, calculating the deviation between the predicted running state and the actual running state; S4. Real-time monitoring of the rate of change of sensor output data, and judging whether there is an abnormal state based on the deviation value. If there is an abnormality, further analysis of the change pattern of each sensor data to locate the source of the abnormality; S5. Generate adjustment suggestions based on the source of the anomaly and record relevant data for optimizing the operation status collection strategy.

[0034] Reference Figure 2 In step S1, the weight of the instantaneous dynamic data is determined by the fluctuation trend of the train's instantaneous acceleration, angular velocity, and nearby object distance information; the weight of the long-term static data is determined by the track geographic information, historical average value, and fluctuation range of the altitude; the sum of the weight of the instantaneous dynamic data and the weight of the long-term static data is 1; and the change in the train's running state is equal to the product of the weight of the instantaneous dynamic data and the change in the running state caused by the instantaneous dynamic data plus the product of the weight of the long-term static data and the change in the running state caused by the long-term static data.

[0035] Reference Figure 3 , the step S2 comprises the following steps: S201, calculating the running state change amount, and predicting the train running state change amount based on instantaneous dynamic data and long-term static data; S202, adjusting the operating state prediction parameter once so that the adjusted operating state change is less than the operating state change in step S201; S203 , adjusting the operating state prediction parameter n times, and adjusting the operating state prediction parameter again according to the change between the instantaneous dynamic data of the current time window and the instantaneous dynamic data of the previous time window.

[0036] The step S2 further comprises the following steps: S204, reset and adjust. When the difference between the long-term static data of the current time window and the long-term static data of the previous time window exceeds the threshold, increase the weight of the long-term static data and enter step S201.

[0037] Reference Figure 4 , the step S3 comprises the following steps: S301, calculating the theoretical operating state change amount, and calculating the theoretical operating state change amount according to the operating state prediction parameter adjustment amount of the current time window; S302, correcting the theoretical operating state change according to the instantaneous dynamic data and long-term static data of the next time window; S303: Calculate the deviation value, that is, calculate the deviation value between the corrected theoretical running state change amount and the actual running state change amount in the next time window.

[0038] Reference Figure 5 , the step S4 comprises the following steps: S401, constructing a correlation matrix, based on the causal relationship between the train running status and the sensor data, constructing the correlation matrix of each sensor; S402: Construct a change rate matrix, obtain the monitoring data of each sensor in real time, calculate the difference between the monitoring data of the current time window of each sensor and the monitoring data of the previous time window, and obtain the change of the monitoring data of each sensor. The ratio of the change of the monitoring data of each sensor to the monitoring data of the previous time window of each sensor is the actual change rate of the monitoring data of each sensor. The actual change rate of the monitoring data of each sensor is entered into the correlation matrix in step S401 to obtain the change rate matrix; S403: Once positioned, obtain the operating state prediction parameter adjustment ratio based on the operating state prediction parameter adjustment amount, obtain the theoretical change rate of each sensor monitoring data, calculate the change rate difference between the theoretical change rate of each sensor monitoring data and the actual change rate of each sensor monitoring data, and the ratio of the change rate difference to the theoretical change rate is the change rate ratio. When the change rate ratio of a sensor is greater than a threshold, it is determined that the sensor is abnormal. If the change rate ratios of all sensors are less than the threshold, the process proceeds to step S404. S404: Secondary positioning, calculating the mean difference between each data in the rate of change matrix and its adjacent data, and determining that the sensor corresponding to the largest mean difference in the rate of change matrix has an abnormality.

[0039] Reference Figure 6 , the step S5 comprises the following steps: S501. Generate adjustment suggestions based on the abnormal sources, giving priority to high-risk sources; S502. Record the source of the abnormality and related data, and establish an operation status collection database; S503: Optimize the operation status prediction parameter adjustment strategy using the operation status collection database.

[0040] In step S5, while performing steps S501 to S503, the following steps are also performed: S504: Monitor the changing trend of the sensor output data. When a certain parameter continuously deviates from the normal range, generate a prompt message and notify the operation and maintenance personnel.

[0041] Example 2 Reference Figure 7 and Figure 8 A high-speed train operation status acquisition system based on multi-source data fusion is applicable to the operation status acquisition method of embodiment 1. The operation status acquisition system includes a data acquisition unit 1, a data analysis unit 2, a parameter adjustment unit 3 and an abnormality diagnosis unit 4; the data acquisition unit 1 is used to obtain the output data of multiple sensors during the train operation in real time; the data analysis unit 2 is used to classify and process instantaneous dynamic data and long-term static data, and calculate the change in the train operation status; the parameter adjustment unit 3 is used to dynamically adjust the operation status prediction parameters; the abnormality diagnosis unit 4 is used to analyze the change rate of the sensor output data, locate the source of the abnormality and generate adjustment suggestions.

[0042] The abnormality diagnosis unit 4 also includes a self-learning module 5, which trains historical collected data through a machine learning algorithm to gradually optimize the operation status prediction parameter adjustment strategy and abnormality diagnosis accuracy; the self-learning module 5 uses a random forest algorithm to optimize the model.

[0043] Example 3 A high-speed train running status acquisition system and method based on multi-source data fusion, the core of which is to predict the running status and diagnose abnormalities by combining the output data of multiple sensors with instantaneous dynamic data and long-term static data. Figure 1 To the attached Figure 8 Specific embodiments of the present invention are described in detail.

[0044] like Figure 7 and Figure 8As shown, the high-speed train operating status acquisition system of the present invention includes a data acquisition unit 1, a data analysis unit 2, a parameter adjustment unit 3, and an anomaly diagnosis unit 4. The data acquisition unit 1 is connected to various sensors on the train via physical interfaces to acquire real-time sensor output data during train operation. These sensors include accelerometers, gyroscopes, distance sensors, and track geographic information sensors. The data analysis unit 2 is connected to the data acquisition unit 1 via a data bus and is responsible for receiving and classifying the data transmitted from the data acquisition unit 1, dividing the data into instantaneous dynamic data and long-term static data, and calculating the change in the train's operating status. The parameter adjustment unit 3 is connected to the data analysis unit 2 via a control signal line and dynamically adjusts the operating status prediction parameters based on the data provided by the data analysis unit 2 to maintain the train's operating status within a set range. The anomaly diagnosis unit 4 is connected to both the data analysis unit 2 and the parameter adjustment unit 3 via communication interfaces and is responsible for analyzing the rate of change of sensor output data, locating the source of anomalies, and generating adjustment recommendations. Furthermore, the anomaly diagnosis unit 4 includes a self-learning module 5, which uses a machine learning algorithm to train historically collected data to gradually optimize the operating status prediction parameter adjustment strategy and anomaly diagnosis accuracy.

[0045] like Figure 2 As shown in the figure, in the operating status prediction and parameter adjustment process, the data acquisition unit 1 first acquires real-time output data from various sensors during train operation and transmits this data to the data analysis unit 2. The data analysis unit 2 classifies the received data. The instantaneous dynamic data includes the train's instantaneous acceleration, angular velocity, and the distance information of nearby objects, while the long-term static data includes track geographic information, the historical average value, and the fluctuation range of altitude. The data analysis unit 2 calculates weights for the instantaneous dynamic data and the long-term static data based on each. The weight of the instantaneous dynamic data is determined by the fluctuation trend of the train's instantaneous acceleration, angular velocity, and the distance information of nearby objects, while the weight of the long-term static data is determined by the historical average value, and the fluctuation range of track geographic information, the altitude, and the fluctuation range. The sum of these weights is 1. Subsequently, the data analysis unit 2 calculates the change in the train's operating status based on the weights: the product of the weight of the instantaneous dynamic data and the change in the operating status caused by the instantaneous dynamic data plus the product of the weight of the long-term static data and the change in the operating status caused by the long-term static data. Parameter adjustment unit 3 adjusts the operating state prediction parameters based on the operating state change provided by data analysis unit 2, so that the adjusted operating state change is less than the current operating state change. If the change between the long-term static data in the current time window and the long-term static data in the previous time window exceeds a threshold, parameter adjustment unit 3 increases the weight of the long-term static data and re-enters the adjustment process. Furthermore, parameter adjustment unit 3 further optimizes the operating state prediction parameters based on the deviation between the theoretical operating state change and the actual operating state change.

[0046] In the sensor anomaly diagnosis process, the anomaly diagnosis unit 4 first constructs an association matrix, which describes the causal relationship between the train's operating status and the data from each sensor. Subsequently, the anomaly diagnosis unit 4 acquires the monitoring data from each sensor in real time and calculates the difference between the monitoring data in the current time window and the monitoring data in the previous time window, respectively, to obtain the change in each sensor's monitoring data. The ratio of the change in each sensor's monitoring data to the monitoring data in the previous time window is the actual change rate of each sensor's monitoring data. The anomaly diagnosis unit 4 enters the actual change rate of each sensor's monitoring data into the association matrix to obtain a change rate matrix. Next, the anomaly diagnosis unit 4 calculates the theoretical change rate of each sensor's monitoring data based on the adjustment value of the operating status prediction parameter and calculates the difference between the theoretical change rate and the actual change rate of each sensor's monitoring data. The ratio of the change rate difference to the theoretical change rate is the change rate ratio. When the change rate ratio of a sensor exceeds a threshold, the sensor is determined to be abnormal. If the rate-of-change ratios for all sensors are less than the threshold, abnormality diagnosis unit 4 further calculates the mean difference between each data point in the rate-of-change matrix and its adjacent data points, and determines that the sensor corresponding to the sensor with the largest mean difference in the rate-of-change matrix is ​​abnormal. Finally, abnormality diagnosis unit 4 generates adjustment recommendations based on the source of the abnormality and records the relevant data for optimizing the operating status collection strategy.

[0047] like Figure 8 As shown in the figure, during the self-learning module optimization process, self-learning module 5 uses the random forest algorithm to train historically collected data. First, self-learning module 5 extracts historical data from the operating status collection database, including the source of anomalies and related data. Subsequently, self-learning module 5 uses the random forest algorithm to model the historical data, gradually optimizing the operating status prediction parameter adjustment strategy and anomaly diagnosis accuracy. During the model training process, self-learning module 5 continuously adjusts the hyperparameters of the random forest algorithm to improve model performance. Finally, self-learning module 5 applies the optimized model to the operating status collection system, enhancing the system's intelligence and long-term reliability.

[0048] In practical applications, the high-speed train operating status acquisition system of the present invention can be deployed within a train's onboard control system. Data acquisition unit 1 connects to various sensors on the train via physical interfaces, acquires sensor data in real time, and transmits it to data analysis unit 2 via a data bus. Data analysis unit 2 classifies the data, calculates the train's operating status change, and transmits the result to parameter adjustment unit 3. Parameter adjustment unit 3 dynamically adjusts the operating status prediction parameters based on the operating status change and feeds the adjusted parameters back to data analysis unit 2 to correct the operating status prediction results. Simultaneously, an anomaly diagnosis unit 4 monitors the rate of change of sensor output data in real time. If a parameter consistently deviates from the normal range, it generates a prompt message and notifies maintenance personnel. Furthermore, a self-learning module 5 within an anomaly diagnosis unit 4 trains historical data using a random forest algorithm, gradually optimizing the operating status prediction parameter adjustment strategy and anomaly diagnosis accuracy. By integrating multi-source data, the entire system achieves comprehensive monitoring and early warning of the train's operating status, ensuring operational safety and stability.

[0049] In this embodiment, the data acquisition unit 1 utilizes an industrial-grade sensor interface design to ensure stable reception of output data from various sensors. The data analysis unit 2 utilizes a high-performance embedded processor with robust data processing capabilities. The parameter adjustment unit 3 dynamically adjusts the operating status prediction parameters through a closed-loop control algorithm, ensuring the accuracy and real-time nature of the adjustment process. The self-learning module 5 within the anomaly diagnosis unit 4 utilizes a distributed computing architecture, enabling efficient processing of large amounts of historical data and model training. Through the collaborative design of hardware and software, the entire system achieves high-precision acquisition and intelligent diagnosis of the operating status of high-speed trains.

[0050] In order to better enable relevant personnel in this technical field to fully understand and implement the present invention, the specific implementation principle of the present invention is further supplemented below with reference to a specific application scenario.

[0051] During high-speed train operation, data acquisition unit 1 connects to accelerometers, gyroscopes, distance sensors, and track geo-information sensors via industrial-grade sensor interfaces, acquiring real-time multi-source data on the train's operating status. This data includes the train's instantaneous acceleration, angular velocity, distance information to nearby objects, track geo-information, and altitude. The output of these sensors is transmitted via a physical interface to data analysis unit 2. Data analysis unit 2, equipped with a high-performance embedded processor, classifies the received data into two categories: instantaneous dynamic data and long-term static data. Instantaneous dynamic data primarily reflects changes in the train's operating status over short periods of time, while long-term static data describes the characteristics of the train's operating environment over longer periods of time.

[0052] In the data analysis unit 2, the weight of the instantaneous dynamic data is first calculated based on its fluctuation trend. For example, when the instantaneous acceleration or angular velocity of the train fluctuates violently, the data analysis unit 2 will increase the weight of the instantaneous dynamic data; at the same time, the weight of the long-term static data is calculated based on the historical average and fluctuation range of the track geographic information and altitude. The sum of the weights of the two is always kept at 1 to ensure the rationality of the weight distribution. Subsequently, the data analysis unit 2 calculates the change in the train's operating status according to the weight formula. Specifically, the change in operating status caused by the instantaneous dynamic data is multiplied by its weight, and the change in operating status caused by the long-term static data is multiplied by its weight. The final change in operating status is obtained by adding the two. The core of this process is to ensure that the system can adapt to complex and changing operating environments through dynamic adjustment of weights.

[0053] The parameter adjustment unit 3 receives the running state change data from the data analysis unit 2, and dynamically adjusts the running state prediction parameters based on the closed-loop control algorithm. For example, when the running state change exceeds the set threshold, the parameter adjustment unit 3 will increase the weight of the long-term static data and re-enter the adjustment process to reduce the running state change. In addition, the parameter adjustment unit 3 further optimizes the running state prediction parameters based on the deviation value between the theoretical running state change and the actual running state change. The method for calculating this deviation value is to derive the theoretical running state change through the running state prediction parameter adjustment amount of the current time window, and compare it with the actual running state change. If the deviation value is large, the parameter adjustment unit 3 will adjust the running state prediction parameters again until the deviation value converges to a reasonable range. This mechanism enables the system to maintain a high prediction accuracy in a dynamic environment.

[0054] The abnormality diagnosis unit 4 constructs a correlation matrix to analyze the causal relationship between the train's operating status and the data from each sensor. During actual operation, the abnormality diagnosis unit 4 acquires monitoring data from each sensor in real time and calculates the difference between the monitoring data in the current time window and the previous time window for each sensor, obtaining the change in each sensor's monitoring data. Subsequently, the abnormality diagnosis unit 4 enters the actual rate of change of each sensor's monitoring data into the correlation matrix to generate a rate of change matrix. Based on this, the abnormality diagnosis unit 4 calculates the theoretical rate of change of each sensor's monitoring data based on the adjustment of the operating status prediction parameters and further calculates the rate of change difference between the theoretical rate of change and the actual rate of change. The ratio of the rate of change difference to the theoretical rate of change is defined as the rate of change ratio. When the rate of change ratio of a sensor exceeds a preset threshold, the sensor is determined to have an abnormality. If the rate of change ratios of all sensors are less than the threshold, the abnormality diagnosis unit 4 further calculates the average difference between each data point in the rate of change matrix and its adjacent data points, and determines that the sensor with the largest average difference is abnormal. This process enables early identification of potential abnormalities through refined analysis of sensor output data.

[0055] The self-learning module 5 uses the random forest algorithm to train historical collected data, and gradually optimizes the operation status prediction parameter adjustment strategy and abnormality diagnosis accuracy. During the model training process, the self-learning module 5 extracts historical data from the operation status collection database, including the source of abnormalities and related data, and uses this data to model the random forest model. By continuously adjusting the hyperparameters of the random forest algorithm, the self-learning module 5 can improve the performance of the model. Ultimately, the optimized model is applied to the operation status collection system to improve the system's intelligence level and long-term reliability. For example, during a certain operation, if a sensor frequently exhibits abnormalities, the self-learning module 5 will record the abnormal pattern of the sensor and give priority to monitoring it in subsequent operations, thereby reducing the risk of abnormalities.

[0056] In practical applications, the entire system achieves efficient operation through the collaborative design of hardware and software. Data acquisition unit 1 utilizes an industrial-grade sensor interface design to ensure stable reception of output data from various sensors. Data analysis unit 2 utilizes a high-performance embedded processor to perform data classification and calculate operational status changes. Parameter adjustment unit 3 implements a closed-loop control algorithm to dynamically adjust operational status prediction parameters. The self-learning module 5 within abnormality diagnosis unit 4 utilizes a distributed computing architecture, enabling efficient processing of large amounts of historical data and model training. By integrating multi-source data, the system not only comprehensively monitors train operating status but also provides early warning in complex environments, ensuring operational safety and stability.

[0057] The above are only preferred specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solutions and inventive concepts of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A method for collecting high-speed train running status based on multi-source data fusion, characterized in that: The following steps are involved: S1. Acquire output data of various sensors during train operation in real time, and divide the data into instantaneous dynamic data and long-term static data; S2. Predict the changing trend of the train operation status based on instantaneous dynamic data and long-term static data, and revise the prediction results based on current data; S3. During the train running state prediction process, calculating the deviation between the predicted running state and the actual running state; S4. Real-time monitoring of the rate of change of sensor output data, and judging whether there is an abnormal state based on the deviation value. If there is an abnormality, further analysis of the change pattern of each sensor data to locate the source of the abnormality; S5. Generate adjustment suggestions based on the source of the anomaly and record relevant data for optimizing the operation status collection strategy.

2. The method for collecting high-speed train running status based on multi-source data fusion according to claim 1 is characterized in that: In step S1, the weight of the instantaneous dynamic data is determined by the fluctuation trend of the train's instantaneous acceleration, angular velocity, and the distance information of nearby objects; the weight of the long-term static data is determined by the track geographic information, the historical average value, and the fluctuation range of the altitude; the sum of the weight of the instantaneous dynamic data and the weight of the long-term static data is 1; the change in the train's running state is equal to the product of the weight of the instantaneous dynamic data and the change in the running state caused by the instantaneous dynamic data plus the product of the weight of the long-term static data and the change in the running state caused by the long-term static data.

3. The method for collecting high-speed train running status based on multi-source data fusion according to claim 2 is characterized in that: The step S2 comprises the following steps: S201, calculating the running state change amount, and predicting the train running state change amount based on instantaneous dynamic data and long-term static data; S202, adjusting the operating state prediction parameter once so that the adjusted operating state change is less than the operating state change in step S201; S203 , adjusting the operating state prediction parameter n times, and adjusting the operating state prediction parameter again according to the change between the instantaneous dynamic data of the current time window and the instantaneous dynamic data of the previous time window.

4. The method for collecting high-speed train running status based on multi-source data fusion according to claim 3 is characterized in that: The step S2 further comprises the following steps: S204, reset and adjust. When the difference between the long-term static data of the current time window and the long-term static data of the previous time window exceeds the threshold, increase the weight of the long-term static data and enter step S201.

5. The method for collecting high-speed train running status based on multi-source data fusion according to claim 1 is characterized in that: The step S3 comprises the following steps: S301, calculating the theoretical operating state change amount, and calculating the theoretical operating state change amount according to the operating state prediction parameter adjustment amount of the current time window; S302, correcting the theoretical operating state change according to the instantaneous dynamic data and long-term static data of the next time window; S303: Calculate the deviation value, that is, calculate the deviation value between the corrected theoretical running state change amount and the actual running state change amount in the next time window.

6. The method for collecting high-speed train running status based on multi-source data fusion according to claim 1 is characterized in that: The step S4 comprises the following steps: S401, constructing a correlation matrix, based on the causal relationship between the train running status and the sensor data, constructing the correlation matrix of each sensor; S402: Construct a change rate matrix, obtain the monitoring data of each sensor in real time, calculate the difference between the monitoring data of the current time window of each sensor and the monitoring data of the previous time window, and obtain the change of the monitoring data of each sensor. The ratio of the change of the monitoring data of each sensor to the monitoring data of the previous time window of each sensor is the actual change rate of the monitoring data of each sensor. The actual change rate of the monitoring data of each sensor is entered into the correlation matrix in step S401 to obtain the change rate matrix; S403: Once positioned, obtain the operating state prediction parameter adjustment ratio based on the operating state prediction parameter adjustment amount, obtain the theoretical change rate of each sensor monitoring data, calculate the change rate difference between the theoretical change rate of each sensor monitoring data and the actual change rate of each sensor monitoring data, and the ratio of the change rate difference to the theoretical change rate is the change rate ratio. When the change rate ratio of a sensor is greater than a threshold, it is determined that the sensor is abnormal. If the change rate ratios of all sensors are less than the threshold, the process proceeds to step S404. S404: Secondary positioning, calculating the mean difference between each data in the rate of change matrix and its adjacent data, and determining that the sensor corresponding to the largest mean difference in the rate of change matrix has an abnormality.

7. The method for collecting high-speed train running status based on multi-source data fusion according to claim 1 is characterized in that: The step S5 comprises the following steps: S501. Generate adjustment suggestions based on the abnormal sources, giving priority to high-risk sources; S502. Record the source of the abnormality and related data, and establish an operation status collection database; S503: Optimize the operation status prediction parameter adjustment strategy using the operation status collection database.

8. The method for collecting high-speed train running status based on multi-source data fusion according to claim 7 is characterized in that: In step S5, while performing steps S501 to S503, the following steps are also performed: S504: Monitor the changing trend of the sensor output data. When a certain parameter continuously deviates from the normal range, generate a prompt message and notify the operation and maintenance personnel.

9. A high-speed train running status acquisition system based on multi-source data fusion, which is applicable to the running status acquisition method according to any one of claims 1 to 8, characterized in that: The operation status acquisition system comprises a data acquisition unit (1), a data analysis unit (2), a parameter adjustment unit (3) and an abnormality diagnosis unit (4); the data acquisition unit (1) is used to obtain output data of various sensors in real time during the operation of the train; the data analysis unit (2) is used to classify and process instantaneous dynamic data and long-term static data, and calculate the change amount of the train operation status; the parameter adjustment unit (3) is used to dynamically adjust the operation status prediction parameters; the abnormality diagnosis unit (4) is used to analyze the change rate of the sensor output data, locate the source of the abnormality and generate adjustment suggestions.

10. The high-speed train running status acquisition system based on multi-source data fusion according to claim 9 is characterized in that: The abnormality diagnosis unit (4) further includes a self-learning module (5), which trains historically collected data using a machine learning algorithm to gradually optimize the operating state prediction parameter adjustment strategy and abnormality diagnosis accuracy; the self-learning module (5) uses a random forest algorithm to perform model optimization.

Images