Train bogie suspension system based on self-adaptive control

Abstract

The invention relates to the technical field of train bogie suspension systems, and discloses a train bogie suspension system based on self-adaptive control, which comprises a train bogie, a sensing network unit, an adjustable suspension execution mechanism, a self-adaptive control unit and a communication interface. According to the train bogie suspension system based on self-adaptive control, the train bogie suspension system can automatically analyze the control effect, recognize the operation mode and explore potential laws through a learning and optimizing module, long-term recording of historical operation data (state input, control output and performance feedback) and an integrated machine learning technology, and the train bogie suspension system can automatically control the train bogie suspension system based on the analysis results. The system can autonomously adjust the internal parameters of the adaptive control strategy, optimize the fuzzy rule base or refine the weight of the neural network model, so that the control strategy can continuously accumulate experience, the control performance of the control strategy is adaptively improved, and the intelligence level and the superiority of long-term operation of the system are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of train bogie suspension systems, and in particular to a train bogie suspension system based on adaptive control. Background Art

[0002] The train bogie suspension system is a key component connecting the car body and wheelset. Its main functions are to attenuate vibrations caused by track irregularities, transmit traction and braking forces, and ensure that the vehicle has good running smoothness, curve negotiating ability and stability.

[0003] Traditional passive suspension systems (such as steel springs and hydraulic shock absorbers) have fixed stiffness and damping characteristics, and their performance is optimized for specific operating conditions during design. However, in actual operation, the train's operating conditions (such as empty, fully loaded, and different speed levels), track conditions (such as straight lines, curves, switches, and different levels of unevenness), and external environment (such as crosswinds) are complex and changeable.

[0004] According to the bogie suspension parameter testing system mentioned in the invention patent with Chinese patent application number 201510460771.3, the bogie suspension parameter testing system calculates the static parameters and dynamic parameters of the bogie suspension system by measuring the three-dimensional force and three-dimensional displacement between the various components of the bogie when in use, so as to analyze the mechanical properties of the locomotive. However, the bogie suspension parameter testing system does not have the functions of intelligence and self-learning when in use, and the train bogie suspension cannot continuously learn, optimize and accumulate experience, which in turn affects the use effect of the train bogie. Therefore, it is necessary to propose a train bogie suspension system based on adaptive control to solve the above-mentioned problems. Summary of the Invention

[0005] (1) Technical problems solved

[0006] In response to the shortcomings of the existing technology, the present invention provides a train bogie suspension system based on adaptive control, which has the advantages of being intelligent and self-learning, and solves the problem that the bogie in the background technology does not have intelligent and self-learning functions.

[0007] (2) Technical solution

[0008] To achieve the above-mentioned object, the present invention provides the following technical solution: a train bogie suspension system based on adaptive control, comprising:

[0009] (1) Train bogie: used to make the train more flexible on curves, reducing energy consumption and tire wear during driving. The train bogie includes a frame, wheels, primary suspension and secondary suspension;

[0010] (2) Sensor network unit: used to collect train operation status signals and suspension system response signals in real time;

[0011] (3) an adjustable suspension actuator, provided in the primary suspension and / or the secondary suspension, whose mechanical parameters can be adjusted according to a control signal;

[0012] (4) an adaptive control unit connected to the sensor network and the adjustable suspension actuator, wherein the adaptive control unit is configured to:

[0013] receiving and processing signals from the sensor network;

[0014] Calculating target parameters of the adjustable suspension actuator using an adaptive control algorithm based on preset performance targets, current operating conditions, and the processed signal;

[0015] generating a control signal to drive the adjustable suspension actuator to achieve the target parameter;

[0016] Wherein, the adaptive control algorithm includes model reference adaptive control, self-correcting control, fuzzy adaptive control, neural network adaptive control or a combination thereof;

[0017] (5) Communication interface: used to communicate with the train network (such as TCMS), obtain train operation instructions (acceleration, braking, speed limit), line information, marshalling information, etc., and upload system status and fault information.

[0018] Preferably, the sensor network unit includes one or more combinations of the following sensors:

[0019] Vehicle state sensor: used to detect the acceleration, speed and / or displacement of the vehicle body;

[0020] Frame status sensor: used to detect acceleration, velocity and / or displacement of the frame;

[0021] Wheel-rail force sensor: used to measure wheel-rail force directly or indirectly;

[0022] Load sensor: used to detect vehicle load;

[0023] Speed ​​sensor: used to detect the train running speed;

[0024] Position sensor: used to obtain train position information;

[0025] Track state sensor: used to detect the track geometric state.

[0026] Preferably, the adjustable suspension actuator is a semi-active actuator, including one or more of a magnetorheological damper, an electrorheological damper, a hydraulic damper controlled by a proportional valve, an air spring, and an electromagnetic spring.

[0027] Preferably, the adaptive control unit further includes a learning and optimization module, which is configured to record historical operating data and control effects, and optimize the parameters or strategies of the adaptive control algorithm using a machine learning algorithm.

[0028] Preferably, the adaptive control unit is connected to the train network control system via a communication interface for obtaining train operation instructions, line information and / or marshaling information.

[0029] Preferably, the adjustable suspension actuator comprises:

[0030] Adjustable dampers: such as magnetorheological dampers, electrorheological dampers or hydraulic dampers controlled by proportional valves.

[0031] Adjustable stiffness elements: such as air springs (which change stiffness / height by adjusting air pressure), electromagnetic springs, or other variable stiffness mechanisms.

[0032] Preferably, the adaptive control unit is connected to the sensor network unit and the adjustable suspension actuator via electrical signals.

[0033] Preferably, the adaptive control unit comprises:

[0034] Data processing module: Receives and processes raw signals from the sensor network, performs filtering, amplification, A / D conversion, feature extraction (including calculation of vibration energy, frequency components, wheel-rail force index, derailment coefficient estimation, etc.) and state estimation (including estimation of state variables that are difficult to measure directly);

[0035] Reference Model Module: This module stores or generates online a reference model representing the ideal suspension system dynamics. The reference model defines the desired performance indicators (including minimization of train acceleration, smooth wheel-rail forces, and stable posture).

[0036] Adaptive control algorithm module: This is the core module. Based on processed sensor data, reference models, and current suspension parameters, it applies adaptive control algorithms including:

[0037] Model reference adaptive control: Design an adaptive law so that the output of the actual suspension system tracks the output of the reference model;

[0038] Self-correcting control: online identification of system parameters (such as equivalent damping and stiffness) and real-time adjustment of controller parameters (such as PID gain) based on the identification results;

[0039] Fuzzy adaptive control: Combining the robustness and adaptability of fuzzy logic, the fuzzy rules or membership functions are adjusted according to the changes in operating conditions;

[0040] Neural network adaptive control: Utilizes the learning ability of neural networks to approximate the dynamics of nonlinear systems online and generate optimal control signals;

[0041] Control signal generation module: Based on the output of the adaptive control algorithm module, it generates specific control instructions (such as current, voltage, and air pressure setting values) and sends them to the adjustable suspension actuator.

[0042] Learning and optimization module: records historical operating data and control effects, and uses machine learning algorithms (such as reinforcement learning) to continuously optimize the parameters or structure of the adaptive control strategy to achieve long-term performance improvement.

[0043] Preferably, the adjustable suspension actuator is of a semi-active type.

[0044] (3) Beneficial effects

[0045] Compared with the prior art, the present invention provides a train bogie suspension system based on adaptive control, which has the following beneficial effects:

[0046] 1. The train bogie suspension system based on adaptive control uses dense sensor network units and advanced information processing technology to comprehensively perceive the train's dynamic operating status (such as speed, load distribution, multi-dimensional vibration of the car body and structure, and wheel-rail force estimation) and external environmental information (such as track irregularities and curve parameters) in real time. Based on this perception information, the system utilizes algorithms such as model reference adaptive control (MRAC), self-correcting control (STC), fuzzy adaptive control, or neural network adaptive control to make intelligent online decisions and dynamically calculate the current optimal adjustable suspension parameters (such as damping force and stiffness / height settings) in real time.

[0047] 2. The train bogie suspension system based on adaptive control uses a learning and optimization module to record historical operating data (state input, control output, performance feedback) for a long time and integrate machine learning technology. It can automatically analyze control effects, identify operating modes, and discover potential laws. Based on these analysis results, the system can autonomously adjust the internal parameters of its adaptive control strategy, optimize the fuzzy rule base or refine the weights of the neural network model, enabling the control strategy to continuously accumulate experience and adaptively improve its control performance, significantly improving the system's intelligence level and long-term operational superiority.

[0048] 3. This train bogie suspension system based on adaptive control uses models and learning and optimization modules to enable the system to dynamically and accurately match different operating conditions (empty / full load, low speed / high speed, straight / curved, good track / bad track, crosswind interference); no longer restricted by fixed parameters or simple rules, the system can actively and flexibly adjust its own characteristics to ensure that the suspension system always operates within the optimal or near-optimal range in various complex and changing environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 Schematic diagram of the structure of the bogie suspension system of the present invention;

[0050] Figure 2 This is a schematic diagram of the sensor network unit structure of the present invention;

[0051] Figure 3 Schematic diagram of the structure of the adaptive control unit of the present invention. DETAILED DESCRIPTION

[0052] 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 ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0053] See also Figure 1-3 , a train bogie suspension system based on adaptive control, including:

[0054] (1) Train bogie: used to make the train more flexible on curves, reducing energy consumption and tire wear during driving. The train bogie includes a frame, wheels, primary suspension and secondary suspension;

[0055] (2) Sensor network unit: used to collect train operation status signals and suspension system response signals in real time;

[0056] (3) an adjustable suspension actuator, provided in the primary suspension and / or the secondary suspension, whose mechanical parameters can be adjusted according to a control signal;

[0057] (4) an adaptive control unit connected to the sensor network and the adjustable suspension actuator, wherein the adaptive control unit is configured to:

[0058] receiving and processing signals from the sensor network;

[0059] Calculating target parameters of the adjustable suspension actuator using an adaptive control algorithm based on preset performance targets, current operating conditions, and the processed signal;

[0060] generating a control signal to drive the adjustable suspension actuator to achieve the target parameter;

[0061] Wherein, the adaptive control algorithm includes model reference adaptive control, self-correcting control, fuzzy adaptive control, neural network adaptive control or a combination thereof;

[0062] (5) Communication interface: used to communicate with the train network (such as TCMS), obtain train operation instructions (acceleration, braking, speed limit), line information, marshalling information, etc., and upload system status and fault information.

[0063] (1) Example 1:

[0064] The sensing network unit includes: a three-axis accelerometer installed under the car body floor to measure the vertical, lateral and longitudinal acceleration of the car body; a three-axis accelerometer installed on the frame side beam to measure the frame vibration; a displacement sensor installed at the secondary suspension position to measure the relative displacement between the car body and the frame; a displacement sensor installed at the primary suspension position (optional) to measure the relative displacement between the frame and the axle box; a load sensor installed at the connection between the car body and the bogie (or the load is estimated by using the air spring pressure); the speed signal is obtained using the train's existing speed sensor; and the train network is used to obtain GPS location information and pre-stored line database information.

[0065] Adjustable suspension actuators include: secondary suspension uses magnetorheological dampers as adjustable damping elements; secondary suspension uses air springs as adjustable stiffness / height elements (the airbag pressure is adjusted through a proportional valve); primary suspension can also use magnetorheological dampers to replace the original passive dampers.

[0066] Adaptive control unit: implemented using a high-performance embedded processor; data processing module: performs low-pass filtering on acceleration and displacement signals (the cutoff frequency is set according to the frequency band of interest, such as 0-20Hz), calculates the effective value (RMS) and spectral characteristics; reference model module: stores a set of ideal vehicle acceleration and wheel-rail force reference models under different speeds and load levels (which can be obtained through offline optimization); adaptive control algorithm module: uses model reference adaptive control combined with fuzzy logic.

[0067] The adaptive control unit achieves the primary comfort objective of minimizing vertical acceleration of the vehicle body, with wheel-rail force stability (or an estimated derailment coefficient) as a safety constraint. An adaptive law is designed to adjust the target damping force of the MR damper. Simultaneously, a fuzzy controller dynamically adjusts the gain of the MRAC adaptive law and the parameters of the reference model based on the current speed, load, track irregularity level (estimated from acceleration signal characteristics), and whether the vehicle is in a curve (determined by position information), enhancing control robustness. Air spring pressure is adjusted based on load and desired vehicle body height (to maintain levelness), with small compensation adjustments made in curves based on superelevation and speed.

[0068] Control signal generation module: converts the calculated target damping force into the driving current instruction of the magnetorheological damper; converts the calculated target height / stiffness into the pressure setting value of the air spring and outputs it to the proportional valve.

[0069] Learning and Optimization Module: This module records changes in the vehicle's RMS acceleration and wheel-rail forces (if measurable or estimated) after each control adjustment. It also uses reinforcement learning algorithms (such as Q-learning) to perform online fine-tuning on the fuzzy rules for adjusting the adaptive law gain. The goal is to optimize the long-term average comfort index (RMS acceleration) while satisfying safety constraints.

[0070] Communication interface: communicate with train TCMS via MVB or Ethernet.

[0071] (2) Example 2: Based on Example 1, the adaptive control algorithm module adopts neural network adaptive control; a deep neural network is used to establish a complex nonlinear mapping relationship between inputs such as speed, load, historical vibration signals, current actuator status, and optimal suspension parameters (damping, stiffness); the network is pre-trained in a laboratory or simulation environment, and when running online, forward calculations are performed based on real-time sensor data to obtain control instructions, and fine-tuning is performed using online collected data.

[0072] Intelligent and adaptive control process of train bogie suspension system:

[0073] The system is powered on and initialized, initial parameters are read, and all sensor data are collected in real time. Data processing includes filtering, calculating RMS acceleration, spectrum analysis, load estimation, judging line status (straight line / curve / switch), and estimating track disturbance level. According to the current speed, load, and line status, a suitable reference model (expected vehicle acceleration response) is selected from the reference model library or generated online. The actual vehicle vertical acceleration is compared with the reference model output to obtain an error signal. The fuzzy adaptive controller works by inputting: speed, load, track disturbance level, error signal and outputting: The adjustment amount of the RAC adaptive law gain is calculated, and based on the adjusted gain, the MRAC algorithm is run to calculate the target damping force required by the magnetorheological damper. Then, based on the load and line information (curve superelevation), the target height and pressure of the air spring are calculated, and a control signal is generated to convert the target damping force into a current instruction for the MR damper; the target pressure is converted into a valve opening instruction for the air spring air supply valve, the actuator is actuated, the damping and stiffness are adjusted, and the system enters the next control cycle (loop execution of steps 2-10). The system background learning and optimization module regularly analyzes historical data and updates the fuzzy rule base.

[0074] In summary, the train bogie suspension system based on adaptive control, the entire train bogie suspension system, through dense sensor network units and advanced information processing technology, can perceive the dynamic operating status of the train (such as speed, load distribution, multi-dimensional vibration of the car body and structure, wheel-rail force estimation) and external environmental information (such as track irregularities, curve parameters) in real time and comprehensively; and based on this perception information, using algorithms such as model reference adaptive control (MRAC), self-correcting control (STC), fuzzy adaptive control or neural network adaptive control, the system can make intelligent online decisions and dynamically calculate the current optimal adjustable suspension parameters (such as damping force, stiffness / height setting values) in real time.

[0075] In addition, the train bogie suspension system uses a learning and optimization module to record historical operating data (state input, control output, performance feedback) over a long period of time and integrate machine learning technology, so that it can automatically analyze control effects, identify operating modes, and discover potential laws. Based on these analysis results, the system can autonomously adjust the internal parameters of its adaptive control strategy, optimize the fuzzy rule base, or refine the weights of the neural network model, enabling the control strategy to continuously accumulate experience and adaptively improve its control performance, significantly improving the system's intelligence level and long-term operational superiority.

[0076] Moreover, through the model and learning and optimization modules, the system can dynamically and accurately match different operating conditions (no load / full load, low speed / high speed, straight line / curve, good track / bad track, crosswind interference); no longer restricted by fixed parameters or simple rules, the system can actively and flexibly adjust its own characteristics to ensure that in various complex and changing environments, the suspension system always operates within the optimal or near-optimal range.

[0077] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply the existence of any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0078] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A train bogie suspension system based on adaptive control, characterized in that: include: (1) Train bogie: used to make the train more flexible on curves, reducing energy consumption and tire wear during driving. The train bogie includes a frame, wheels, primary suspension and secondary suspension; (2) Sensor network unit: used to collect train operation status signals and suspension system response signals in real time; (3) an adjustable suspension actuator, provided in the primary suspension and / or the secondary suspension, whose mechanical parameters can be adjusted according to a control signal; (4) an adaptive control unit connected to the sensor network and the adjustable suspension actuator, wherein the adaptive control unit is configured to: receiving and processing signals from the sensor network; Calculating target parameters of the adjustable suspension actuator using an adaptive control algorithm based on preset performance targets, current operating conditions, and the processed signal; generating a control signal to drive the adjustable suspension actuator to achieve the target parameter; Wherein, the adaptive control algorithm includes model reference adaptive control, self-correcting control, fuzzy adaptive control, neural network adaptive control or a combination thereof; (5) Communication interface: used to communicate with the train network (such as TCMS), obtain train operation instructions (acceleration, braking, speed limit), line information, marshalling information, etc., and upload system status and fault information.

2. The train bogie suspension system based on adaptive control according to claim 1, characterized in that: The sensor network unit includes one or more combinations of the following sensors: Vehicle state sensor: used to detect the acceleration, speed and / or displacement of the vehicle body; Frame status sensor: used to detect acceleration, velocity and / or displacement of the frame; Wheel-rail force sensor: used to measure wheel-rail force directly or indirectly; Load sensor: used to detect vehicle load; Speed ​​sensor: used to detect the train running speed; Position sensor: used to obtain train position information; Track state sensor: used to detect the track geometric state.

3. The train bogie suspension system based on adaptive control according to claim 1 or 2, characterized in that: The adjustable suspension actuator is a semi-active actuator, including one or more of a magnetorheological damper, an electrorheological damper, a hydraulic damper controlled by a proportional valve, an air spring, and an electromagnetic spring.

4. The train bogie suspension system based on adaptive control according to claim 1, characterized in that: The adaptive control unit also includes a learning and optimization module, which is configured to record historical operating data and control effects, and optimize the parameters or strategies of the adaptive control algorithm using a machine learning algorithm.

5. The train bogie suspension system based on adaptive control according to claim 1, characterized in that: The adaptive control unit is connected to the train network control system via a communication interface to obtain train operation instructions, line information and / or marshaling information.

6. The train bogie suspension system based on adaptive control according to claim 1, characterized in that: The adjustable suspension actuator comprises: Adjustable dampers: such as magnetorheological dampers, electrorheological dampers or hydraulic dampers controlled by proportional valves. Adjustable stiffness elements: such as air springs (which change stiffness / height by adjusting air pressure), electromagnetic springs, or other variable stiffness mechanisms.

7. The train bogie suspension system based on adaptive control according to claim 1, characterized in that: The adaptive control unit is connected with the sensor network unit and the adjustable suspension actuator by electrical signals.

8. The train bogie suspension system based on adaptive control according to claim 1, characterized in that: The adaptive control unit comprises: Data processing module: Receives and processes raw signals from the sensor network, performs filtering, amplification, A / D conversion, feature extraction (including calculation of vibration energy, frequency components, wheel-rail force index, derailment coefficient estimation, etc.) and state estimation (including estimation of state variables that are difficult to measure directly); Reference Model Module: This module stores or generates online a reference model representing the ideal suspension system dynamics. The reference model defines the desired performance indicators (including minimization of train acceleration, smooth wheel-rail forces, and stable posture). Adaptive control algorithm module: This is the core module. Based on processed sensor data, reference models, and current suspension parameters, it applies adaptive control algorithms including: Model reference adaptive control: Design an adaptive law so that the output of the actual suspension system tracks the output of the reference model; Self-correcting control: online identification of system parameters (such as equivalent damping and stiffness) and real-time adjustment of controller parameters (such as PID gain) based on the identification results; Fuzzy adaptive control: Combining the robustness and adaptability of fuzzy logic, the fuzzy rules or membership functions are adjusted according to the changes in operating conditions; Neural network adaptive control: Utilizes the learning ability of neural networks to approximate the dynamics of nonlinear systems online and generate optimal control signals; Control signal generation module: Based on the output of the adaptive control algorithm module, it generates specific control instructions (such as current, voltage, and air pressure setting values) and sends them to the adjustable suspension actuator. Learning and optimization module: records historical operating data and control effects, and uses machine learning algorithms (such as reinforcement learning) to continuously optimize the parameters or structure of the adaptive control strategy to achieve long-term performance improvement.

9. The train bogie suspension system based on adaptive control according to claim 1, characterized in that: The adjustable suspension actuator is of semi-active type.

Images