A non-contact connector for rail transit and its alignment method

Abstract

The present invention discloses a non-contact connector for rail transit and its alignment method, which includes a plug arranged at the first carriage and a socket arranged at the second carriage. Electromagnetic modules and retractable hoses are provided at the ends of the plug and the socket. The electromagnetic module includes a magnetic core and a coil helically wound around the outer periphery of the magnetic core. The interface of the coil is connected to a power supply through an inverter, and the power supply is connected to a processor. A magnetic attraction adjustment module is integrated on the processor. The magnetic attraction adjustment module includes a database, a GPS, and a neural network training unit. The neural network training unit outputs an internal power supply current control signal to the power supply to control the working state of the power supply, and further controls each electromagnetic module to generate a corresponding magnetic attraction to ensure the alignment of the plug and the socket of the non-contact connector when turning.

Description

Technical Field

[0001] The present invention relates to the technical field of train transmission, and particularly relates to a non-contact connector for rail transit and an alignment method thereof. Background Art

[0002] In the application scenario of rail transit, in order to meet different operation tasks and flexibly allocate transport capacity, a train needs to have the ability to adjust the number of carriages. Standardized couplers are used to connect between trains. When trains are coupled, automatic couplers can fully achieve automatic connection of machinery, electricity, and air circuits. Among them, electrical connection includes hard-wired safety loops, hard-wired control lines, and communication network lines, etc. A common coupler electrical connector has 196 cores, among which 56 pins are plug-in type, and the remaining 140 pins are point-contact type. The pin layouts on both sides of the connected couplers are completely unified and symmetrical on the left and right sides. With the gradual increase of train communication rate, the existing technology based on the good contact of copper cores in the connector to transmit data signals has the following defects in the train operation environment: 1) Since the coupler connectors between carriages are plugged and unplugged by vehicle automatic connection couplers, limited by the mechanical structure and unable to ensure that the two ends of the connectors are completely aligned in space, there are requirements for the mechanical strength of the copper pins of the coupler connectors. The diameter of the copper pins should be greater than 5 mm. However, when the signal transmission rate of the connector increases, thinner copper pins can achieve lower-impedance transmission. If thick pins are continued to be used, the signal quality will deteriorate;

[0003] 2) The traditional contact connection method depends on the good electrical conductivity of the metal contact surface. If there are stains or foreign objects, it will cause attenuation of communication signals. In terms of maintenance, it is necessary to regularly remove dust and clean the coupler connectors, and the maintenance work is cumbersome;

[0004] 3) After multiple pluggings and unplugging, the gold-plated surface at the connection will disappear due to wear, resulting in an increase in signal transmission loss of the coupler.

[0005] Therefore, a non-contact connector for connecting carriages in rail transit has emerged as the times require. However, when the existing non-contact connector passes through a curve, it is difficult to align the plug and the socket, resulting in a risk of derailment of the carriage. Since the track curvatures are different at different positions of the track curve, the deviation degrees of the plug and the socket are different, the magnetic suction required to ensure alignment is different, and the magnitude of the current output by the power supply is different. Therefore, it is necessary to consider establishing a curve relationship between the track position and the power supply current in the research on the alignment problem of the plug and the socket of the non-contact connector. Summary of the Invention

[0006] The present invention mainly aims to solve the problem that when the existing non-contact connectors are used in a train during a turn, it is difficult to align the plug and the socket, resulting in a risk of derailment of the carriage. A non-contact connector for rail transit and its alignment method are provided, including a plug and a socket. Electromagnetic modules are provided at the ends of the plug and the socket. The electromagnetic module includes a magnetic core and a coil spirally wound around the outer periphery of the magnetic core. The interface of the coil is connected to a power supply through an inverter. The power supply is connected to a processor, and a magnetic suction force adjustment module is integrated on the processor. The magnetic suction force adjustment module outputs an internal power supply current control signal to the power supply to control the magnitude of the current output by the power supply, thereby realizing controlling the corresponding electromagnetic module to generate a corresponding magnetic suction force at different turning directions or different turning positions to ensure the alignment of the non-contact connector during a turn.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] A non-contact connector for rail transit includes a plug and a socket. Electromagnetic modules and retractable soft bodies are provided at the ends of the plug and the socket. The electromagnetic module includes a magnetic core and a coil. The coil is spirally arranged on the outer periphery of the magnetic core. The interface of the coil is connected to a power supply through an inverter. The power supply is connected to a processor integrated with a magnetic suction force adjustment module. The magnetic suction force adjustment module includes a database, a GPS, and a neural network training unit. The neural network training unit autonomously learns through a neural network to obtain the curve relationship between the track position and the power supply current. The present invention provides a non-contact connector for rail transit, including a plug provided at a first carriage and a socket provided at a second carriage. Electromagnetic modules and retractable hoses are provided at the ends of the plug and the socket. The electromagnetic module includes a magnetic core and a coil spirally wound around the outer periphery of the magnetic core. The interface of the coil is connected to a power supply through an inverter. The power supply is connected to a processor, and a magnetic suction force adjustment module is integrated on the processor. The magnetic suction force adjustment module includes a database, a GPS, and a neural network training unit. The magnetic suction force adjustment module outputs an internal power supply current control signal to the power supply, and the power supply controls the magnitude of the current output by the power supply according to the internal power supply current control signal, thereby controlling whether the magnetic core has a magnetic suction force and the magnitude of the magnetic suction force. A plurality of electromagnetic modules are provided at the ends of the plug and the socket. The plurality of electromagnetic modules are respectively connected in series with MOS transistors and then connected to the inverter and the power supply, and corresponding electromagnetic modules are controlled to generate corresponding magnetic suction forces at different turning directions or different turning positions to ensure the alignment of the non-contact connector during a turn.

[0009] Preferably, the power supply energizes the coil so that the end of the magnetic core has a magnetic suction force, and the magnitude of the magnetic suction force is adjusted by controlling the magnitude of the power supply current. The power supply energizes the coil so that the end of the magnetic core has a magnetic suction force, realizing the non-contact connection between the plug and the socket; by controlling the magnitude of the power supply current, the magnitude of the magnetic suction force is adjusted to ensure that the plug and the socket of the non-contact connector are aligned when turning.

[0010] Preferably, a plurality of electromagnetic modules are provided at the end of the plug and the end of the socket. The plurality of electromagnetic modules are respectively connected in series with MOS transistors and then connected in parallel. At different turning directions or different turning positions, the corresponding electromagnetic modules are controlled to generate corresponding magnetic suction forces to adjust the included angle between the plug and the socket, so as to ensure that the non-contact connector is aligned when turning.

[0011] Preferably, the database and the GPS are respectively connected to the neural network training unit. The database includes the track positions obtained from practice records and the corresponding power supply current values with the best alignment effect. The GPS is used for positioning. The neural network training unit generates a power supply current prediction model through neural network training based on the practice record data provided by the database, and obtains the curve relationship between the track position and the power supply current; the real-time track position is obtained through the GPS and input into the power supply current prediction model to obtain the corresponding power supply current value (internal power supply current control signal), and the working states of each electromagnetic module are controlled according to the power supply current value (internal power supply current control signal) to realize the automatic alignment of the plug and the socket of the non-contact connector and improve the intelligent level.

[0012] Preferably, the processor is wirelessly connected to a remote monitoring terminal, and the remote monitoring terminal is used to send an external power supply current control signal to the processor and receive the position signal fed back by the processor. The staff realizes remote monitoring of the non-contact connector through the remote monitoring terminal.

[0013] Preferably, the non-contact connector is provided with a locking device, and the locking device is used to lock the end positions of the plug and the socket on the straight track section. On the straight track section, the end positions are fixed by the locking device, and the alignment function is turned off.

[0014] An alignment method for a non-contact connector for rail transit includes the following steps:

[0015] Step S1: Obtain practice record data, including the track position and the corresponding power supply current value with the best alignment effect, and preprocess the practice record data;

[0016] Step S2: Divide the preprocessed practice record data into a training set and a test set, and obtain a power supply current prediction model through autonomous learning of the neural network;

[0017] Step S3: Obtain the real-time track position where the non-contact connector is located and input the real-time track position into the power supply current prediction model. The power supply current prediction model outputs the real-time power supply current value corresponding to the real-time track position.

[0018] Step S4: Feed back the real-time power supply current value to the power supply to control the magnitude of the power supply current, and further control the magnitude of the magnetic core magnetic attraction force to complete alignment.

[0019] The present invention also provides an alignment method for a non-contact connector used in rail transit. The neural network training unit obtains the practice record data from the database and performs preprocessing, generates a power supply current prediction model through neural network training, and obtains the curve relationship between the track position and the power supply current; obtains the real-time track position where the non-contact connector is located through GPS, inputs it into the power supply current prediction model, obtains the corresponding power supply current value (internal power supply current control signal), and controls the working state of the power supply according to the power supply current value (internal power supply current control signal), and further controls the working state of each electromagnetic module to realize automatic alignment of the non-contact connector and improve the intelligent level. More specifically, when entering a curve, the angle between the central axes of the plug and the socket becomes smaller from 180°, and the magnetic cores at the ends of the plug and the socket are controlled to attract each other to complete alignment; when exiting a curve, the angle between the central axes of the plug and the socket gradually returns to 180°, and the two magnetic cores are controlled to repel each other to complete alignment, or the attraction force between the two magnetic cores is reduced, and the elastic automatic retraction of the retractable soft body (such as a hose) is used to ensure alignment.

[0020] Preferably, in step S2, the specific process of obtaining the power supply current prediction model through autonomous learning of the neural network includes the following steps:

[0021] S21: Construct a fuzzy neural network and initialize the network parameters of the fuzzy neural network.

[0022] S22: Perform supervised pre-training on the initialized fuzzy neural network using the training set data.

[0023] S23: Retain the parameters of the fuzzy neural network after supervised pre-training, and replace the fully connected layer of the fuzzy neural network with a deep belief network.

[0024] S24: Perform unsupervised pre-training on the deep belief network using the training set data.

[0025] S25: Retain the parameters of the deep belief network after unsupervised pre-training, and add a softmax layer after the output layer of the existing network.

[0026] S26: Supervise and train the entire network using the training set data and generate a power supply current prediction model;

[0027] S27: Verify the power supply current prediction model using the test set data. If the verification is successful, the power supply current prediction model is used to output the real-time power supply current value corresponding to the real-time track position.

[0028] Preferably, in the curved track section or uphill section, the non-contact connector activates the alignment function.

[0029] Therefore, the advantages of the present invention are:

[0030] (1) Achieve non-contact connection of carriages;

[0031] (2) Automatically align the plug and socket of the non-contact connector when turning, improving the intelligent level. Description of the Drawings

[0032] Figure 1 is a schematic diagram of a partial structure of a non-contact connector for rail transit in Embodiment 1 of the present invention.

[0033] Figure 2 is a schematic diagram of a partial power system structure of a non-contact connector for rail transit in Embodiment 1 of the present invention.

[0034] Figure 3 is a schematic diagram of the structure of the magnetic attraction force adjustment module in Embodiment 1 of the present invention.

[0035] Figure 4 is a flowchart of an alignment method for a non-contact connector for rail transit in Embodiment 2 of the present invention.

[0036] 1. Plug 2. Socket 3. Electromagnetic module 4. Inverter 5. Power supply 6. Processor 7. Magnetic attraction force adjustment module 8. Database 9. GPS 10. Neural network training unit 11. Remote monitoring terminal. Detailed Embodiments

[0037] The following further describes the present invention in conjunction with the drawings and specific embodiments.

[0038] Embodiment 1:

[0039] A non-contact connector for rail transit, as Figure 1 shown, includes a plug 1 and a socket 2. Electromagnetic modules 3 and retractable soft bodies are provided at the ends of the plug 1 and the socket 2. The electromagnetic module 3 includes a magnetic core and a coil, and the coil is spirally arranged on the outer periphery of the magnetic core, as Figure 2As shown, the interface of the coil is connected to the power supply 5 through the inverter 4, and the power supply 5 is connected to the processor 6 integrated with the magnetic attraction adjustment module 7, as Figure 3 As shown, the magnetic attraction adjustment module 7 includes a database 8, a GPS 9, and a neural network training unit 10. The neural network training unit 10 autonomously learns through the neural network to obtain the curve relationship between the track position and the power supply current. This embodiment provides a non-contact connector for rail transit, including a plug 1 provided at the first carriage and a socket 2 provided at the second carriage. Electromagnetic modules 3 and retractable hoses are provided at the ends of the plug 1 and the socket 2. The electromagnetic module 3 includes a magnetic core and a coil helically wound around the outer periphery of the magnetic core. The interface of the coil is connected to the power supply 5 through the inverter 4, the power supply 5 is connected to the processor 6, and the magnetic attraction adjustment module 7 is integrated on the processor 6. The magnetic attraction adjustment module 7 includes a database 8, a GPS 9, and a neural network training unit 10. The magnetic attraction adjustment module 7 outputs an internal power supply current control signal to the power supply 5, and the power supply 5 controls the magnitude of the current output by the power supply 5 according to the internal power supply current control signal, thereby controlling whether the magnetic core has magnetic attraction and the magnitude of the magnetic attraction. A plurality of electromagnetic modules 3 are provided at the ends of the plug 1 and the socket 2. The plurality of electromagnetic modules 3 are respectively connected to the inverter 4 and the power supply 5 after being connected in series with MOS transistors, and corresponding electromagnetic modules 3 are controlled to generate corresponding magnitudes of magnetic attraction at different turning directions or different turning positions to ensure the alignment of the non-contact connector when turning.

[0040] As Figure 3 As shown, the database 8 and the GPS 9 are respectively connected to the neural network training unit 10. The database 8 includes the track positions obtained from practical records and the corresponding power supply current values with the best alignment effect. The GPS 9 is used for positioning. The neural network training unit 10 generates a power supply current prediction model through neural network training based on the practical record data provided by the database 8 to obtain the curve relationship between the track position and the power supply current; obtains the real-time track position through the GPS 9, inputs it into the power supply current prediction model, obtains the corresponding power supply current value (internal power supply current control signal), and controls the working states of the respective electromagnetic modules 3 according to the power supply current value (internal power supply current control signal) to realize the automatic alignment of the plug 1 and the socket 2 of the non-contact connector.

[0041] The processor 6 is wirelessly connected to the remote monitoring terminal 11, and the remote monitoring terminal 11 is used to send an external power supply current control signal to the processor 6 and receive the position signal fed back by the processor 6. The staff realizes remote monitoring of the non-contact connector through the remote monitoring terminal 11.

[0042] The non-contact connector is provided with a locking device. In the straight section of the track, the end position is fixed through the locking device, and the alignment function is turned off; in the curved section or uphill section of the track, the non-contact connector activates the alignment function.

[0043] Embodiment 2:

[0044] An alignment method for a non - contact connector used in rail transit, as Figure 4 shown, includes the following steps: Step S1: Obtain practical record data, including rail positions and the corresponding power supply current values with the best alignment effect, and pre - process the practical record data;

[0045] Step S2: Divide the pre - processed practical record data into a training set and a test set, and obtain a power supply current prediction model through autonomous learning of a neural network;

[0046] Step S3: Obtain the real - time rail position where the non - contact connector is located and input the real - time rail position into the power supply current prediction model. The power supply current prediction model outputs the real - time power supply current value corresponding to the real - time rail position;

[0047] Step S4: Feed back the real - time power supply current value to the power supply, control the magnitude of the power supply current, and thus control the magnitude of the magnetic core magnetic attraction force to complete the alignment.

[0048] This embodiment provides an alignment method for a non - contact connector used in rail transit. The neural network training unit obtains practical record data from the database and pre - processes it, generates a power supply current prediction model through neural network training, and obtains the curve relationship between the rail position and the power supply current; obtains the real - time rail position where the non - contact connector is located through GPS, inputs it into the power supply current prediction model, obtains the corresponding power supply current value (internal power supply current control signal), and controls the working state of the power supply according to the power supply current value (internal power supply current control signal), and thus controls the working state of each electromagnetic module to realize automatic alignment of the non - contact connector. More specifically, when entering a curve, the angle between the central axes of the plug and the socket becomes smaller from 180°, and the magnetic cores at the ends of the plug and the socket are controlled to attract each other to complete the alignment; when exiting the curve, the angle between the central axes of the plug and the socket gradually returns to 180°, and the two magnetic cores are controlled to repel each other to complete the alignment, or the attraction force between the two magnetic cores is reduced, and the elastic automatic retraction of a stretchable soft body (such as a hose) is used to ensure the alignment.

[0049] In step S2, the specific process of obtaining the power supply current prediction model through autonomous learning of the neural network includes the following steps:

[0050] S21: Construct a fuzzy neural network and initialize the network parameters of the fuzzy neural network;

[0051] S22: Conduct supervised pre - training on the initialized fuzzy neural network using the training set data;

[0052] S23: Retain the parameters of the fuzzy neural network after supervised pre-training, and replace the fully connected layer of the fuzzy neural network with a deep belief network;

[0053] S24: Perform unsupervised pre-training on the deep belief network using the training set data;

[0054] S25: Retain the parameters of the deep belief network after unsupervised pre-training, and add a softmax layer after the output layer of the existing network;

[0055] S26: Perform supervised training on the entire network using the training set data and generate a power supply current prediction model;

[0056] S27: Verify the power supply current prediction model using the test set data. If the verification is successful, the power supply current prediction model is used to output the real-time power supply current value corresponding to the real-time track position.

[0057] The above content is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed by this application can easily think of changes or substitutions, which should all be covered within the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims

1. A non-contact connector for rail transit, characterized in that, It includes a plug and a socket. Electromagnetic modules and retractable soft bodies are provided at the ends of the plug and the socket. The electromagnetic module includes a magnetic core and a coil. The coil is spirally arranged on the outer periphery of the magnetic core. The interface of the coil is connected to a power supply through an inverter. The power supply is connected to a processor integrated with a magnetic attraction adjustment module. The magnetic attraction adjustment module includes a database, a GPS, and a neural network training unit. The magnetic attraction adjustment module controls the magnitude of the current output by the power supply, thereby achieving the control of the corresponding electromagnetic module to generate a corresponding magnetic attraction at different turning directions or different turning positions. The neural network training unit autonomously learns through the neural network to obtain the curve relationship between the track position and the power supply current.

2. The non-contact connector for rail transit according to claim 1, characterized in that, The power supply energizes the coil so that the end of the magnetic core has a magnetic attraction, and the magnitude of the magnetic attraction is adjusted by controlling the magnitude of the power supply current.

3. The non-contact connector for rail transit according to claim 1, characterized in that, A plurality of electromagnetic modules are provided at the ends of the plug and the socket. The plurality of electromagnetic modules are respectively connected to MOS transistors and then connected in parallel.

4. The non-contact connector for rail transit according to claim 1, characterized in that, The database and the GPS are respectively connected to the neural network training unit. The database includes the track positions obtained from practice records and the corresponding power supply current values with the best alignment effect. The GPS is used for positioning.

5. The non-contact connector for rail transit according to claim 4, characterized in that, The processor is wirelessly connected to a remote monitoring terminal. The remote monitoring terminal is used to send an external power supply current control signal to the processor and receive the position signal fed back by the processor.

6. The non-contact connector for rail transit according to claim 1, characterized in that, The non-contact connector is provided with a locking device. The locking device is used to lock the end positions of the plug and the socket on the straight track section.

7. An alignment method for a non-contact connector for rail transit, applicable to the non-contact connector for rail transit according to any one of claims 1-6, characterized in that, It includes the following steps: Step S1: Obtain practice record data, including the track position and the corresponding power supply current value with the best alignment effect, and preprocess the practice record data. Step S2: Divide the preprocessed practice record data into a training set and a test set, and autonomously learn through the neural network to obtain a power supply current prediction model. Step S3: Obtain the real-time track position where the non-contact connector is located and input the real-time track position into the power supply current prediction model. The power supply current prediction model outputs the real-time power supply current value corresponding to the real-time track position. Step S4: Feed back the real-time power supply current value to the power supply, control the magnitude of the power supply current, and further control the magnitude of the magnetic attraction of the magnetic core to complete the alignment.

8. The alignment method for a non-contact connector for rail transit according to claim 7, characterized in that, In step S2, the specific process of autonomously learning through the neural network to obtain the power supply current prediction model includes the following steps: S21: Construct a fuzzy neural network and initialize the network parameters of the fuzzy neural network. S22: Perform supervised pre-training on the initialized fuzzy neural network with the training set data. S23: Retain the parameters of the fuzzy neural network after supervised pre-training, and replace the fully connected layer of the fuzzy neural network with a deep belief network. S24: Perform unsupervised pre-training on the deep belief network with the training set data. S25: Retain the parameters of the deep belief network after unsupervised pre-training, and add a softmax layer after the output layer of the existing network. S26: Perform supervised training on the entire network with the training set data and generate a power supply current prediction model. S27: Validate the power supply current prediction model using the test set data. If the validation is successful, the power supply current prediction model is used to output the real-time power supply current value corresponding to the real-time track position.

9. The alignment method for a non-contact connector for rail transit according to claim 7, characterized in that, In the curved track section or uphill section, the non-contact connector activates the alignment function.

Images