Dual-mode control method and system for subway operation based on pulse width modulation coding

Abstract

The present application relates to the field of mode control technology, and discloses a dual-mode control method and system for subway operation based on pulse width modulation coding. The method includes: defining the initialization voltage range through a power supply measurement module, and using a dual redundant measurement module to realize dual-channel calibration of manual control and automatic operation channel voltages; a signal collector collects real-time current signals, and a signal comparator performs abnormal detection to generate a mode selection control signal; a pulse width modulation encoder calculates duty cycle parameters and generates a current-limiting pulse width modulation signal; multiple protectors determine the fault state, and a delay controller adjusts the relay disconnection timing; a timing controller establishes mode switching rules, and a relay linkage controller optimizes the output state; a data collector extracts operating parameters, and a communication interface completes data electrical isolation transmission. The present application realizes seamless switching between manual control and automatic operation modes through pulse width modulation coding technology, ensuring the stability and reliability of the system during the mode switching process.

Description

Technical Field

[0001] The present application relates to the field of mode control, and in particular to a method and system for controlling dual-mode operation of a subway based on pulse width modulation coding. Background Art

[0002] At present, the operation control system of subway trains generally adopts two modes: manual control (MC) and automatic operation (ATO). In traditional control systems, the switching between the two modes mainly relies on relay groups and analog circuits. In manual mode, the driver generates corresponding control signals by operating the traction and brake handles; in automatic mode, the on-board automatic control system generates control instructions according to the preset speed curve. The system converts control instructions into control signals that can be recognized by the drive equipment through multiple links such as voltage detection, signal acquisition, signal conditioning, and pulse width modulation, thereby realizing the operation control of the train.

[0003] However, the existing technology has the following shortcomings: since the manual control and automatic operation modes use independent hardware circuits, the system structure is complex and the reliability is low; during the mode switching process, due to the lack of an effective collaborative control mechanism, problems such as switching shock and signal interference are prone to occur; at the same time, the system's abnormal detection and protection mechanism for input signals is relatively simple, and it is difficult to ensure the stable operation of the system under complex working conditions; in addition, the existing systems generally lack real-time monitoring and data analysis functions for the operating status, making it difficult to promptly discover and deal with potential operating hazards. Summary of the invention

[0004] The present application provides a dual-mode control method and system for subway operation based on pulse width modulation coding, which is used to achieve seamless switching between manual control and automatic operation modes through pulse width modulation coding technology, while ensuring the stability and reliability of the system during the mode switching process.

[0005] In the first aspect, the present application provides a dual-mode control method for subway operation based on pulse width modulation coding, and the dual-mode control method for subway operation based on pulse width modulation coding includes: taking the nominal power supply voltage value as a reference, determining the input terminal voltage detection value after defining the initialization voltage range through the power supply measurement module, implementing dual-path calibration of the manual control and automatic operation channel voltages according to the dual redundant measurement module, and outputting the manual control detection voltage and the automatic operation detection voltage; according to the manual control detection voltage and the automatic operation detection voltage, the signal collector collects the real-time current signals of the manual control channel and the automatic operation channel and converts them into manual control signal acquisition The manual control detection voltage and the automatic operation detection voltage are input into the signal collector, and the original voltage signal is band-pass filtered by the Butterworth filter algorithm to generate filtered voltage data; the instantaneous amplitude and phase information of the signal are extracted by the Hilbert transform algorithm from the filtered voltage data to obtain the real-time current signal; based on the real-time current signal, the discrete wavelet transform algorithm is used to perform multi-scale decomposition and extract the signal singularity point to generate the manual control signal sampling value and the automatic operation signal sampling value. value; the manual control signal sampling value and the automatic operation signal sampling value are used to construct an abnormal detection parameter matrix through an adaptive dual threshold algorithm, and an abnormal state identification is output; for the abnormal state identification, the abnormal level coefficient is calculated through a fuzzy clustering algorithm and an abnormal evaluation index is established to generate an abnormal feature vector; based on the abnormal feature vector, a dual-mode classifier is established through a support vector machine algorithm and the mode switching probability is calculated, and the mode selection control signal is output; according to the mode selection control signal, a pulse width modulation waveform is obtained after calculating the output duty ratio segmentation parameter through a pulse width modulation encoder, and the output current amplitude is limited by a load resistor to generate a limit value The current-limiting pulse width modulation signal is generated; based on the current-limiting pulse width modulation signal, the output short circuit and overload status are determined by multiple protectors and a protection trigger signal is output, the delay controller is started to adjust the output relay disconnection timing, and a protection execution signal is constructed; the protection execution signal is collected, a mode switching rule is established through a timing controller and a mode switching signal is generated, the relay linkage controller is called to optimize the output relay coordination state, and a dual-mode control signal is generated; based on the dual-mode control signal, the real-time operating status parameters of the system are extracted through a data collector to generate status monitoring data, the data electrical isolation transmission processing is completed by the communication interface, and the system operating parameters are output.

[0006] In a second aspect, the present application provides a subway operation dual-mode control system based on pulse width modulation coding, and the subway operation dual-mode control system based on pulse width modulation coding includes:

[0007] The calibration module is used to determine the input voltage detection value based on the nominal power supply voltage value after the power supply measurement module defines the initialization voltage range, implement the dual-path calibration of the manual control and automatic operation channel voltages according to the dual redundant measurement module, and output the manual control detection voltage and the automatic operation detection voltage;

[0008] The acquisition module is used to acquire the real-time current signals of the manual control channel and the automatic operation channel according to the manual control detection voltage and the automatic operation detection voltage by a signal collector and convert them into manual control signal sampling values ​​and automatic operation signal sampling values, complete the threshold detection analysis of abnormal signals through a signal comparator, and generate a mode selection control signal, including: inputting the manual control detection voltage and the automatic operation detection voltage into the signal collector, performing bandpass filtering on the original voltage signal through a Butterworth filtering algorithm to generate filtered voltage data; extracting the instantaneous amplitude and phase information of the signal from the filtered voltage data through a Hilbert transform algorithm to obtain Real-time current signal; based on the real-time current signal, multi-scale decomposition is performed and signal singular points are extracted through discrete wavelet transform algorithm to generate manual control signal sampling values ​​and automatic operation signal sampling values; the manual control signal sampling values ​​and the automatic operation signal sampling values ​​are used to construct an abnormality detection parameter matrix through an adaptive dual threshold algorithm, and an abnormal state identification is output; for the abnormal state identification, an abnormality level coefficient is calculated through a fuzzy clustering algorithm and an abnormality evaluation index is established to generate an abnormal feature vector; based on the abnormal feature vector, a dual-mode classifier is established through a support vector machine algorithm and the mode switching probability is calculated, and the mode selection control signal is output;

[0009] A calculation module, used for selecting a control signal according to the mode, obtaining a pulse width modulation waveform after calculating the output duty ratio segmentation parameter through a pulse width modulation encoder, limiting the output current amplitude through a load resistor, and generating a current-limiting pulse width modulation signal;

[0010] An output module is used to determine the output short circuit and overload status based on the current-limiting pulse width modulation signal through multiple protectors and output a protection trigger signal, start the delay controller to adjust the output relay disconnection timing, and construct a protection execution signal;

[0011] A generation module is used to collect the protection execution signal, establish a mode switching rule and generate a mode switching signal through a timing controller, call a relay linkage controller to optimize the coordinated state of the output relay, and generate a dual-mode control signal;

[0012] The transmission module is used to extract the real-time operating status parameters of the system through the data collector based on the dual-mode control signal to generate status monitoring data, complete the data electrical isolation transmission processing through the communication interface, and output the system operating parameters.

[0013] In the technical solution provided by the present application, a dual redundant measurement module based on the nominal supply voltage value is adopted to realize the precise voltage calibration of the manual control and automatic operation channels, effectively overcoming the system complexity and low reliability caused by the independent hardware circuit in the traditional system, and ensuring the accuracy and stability of the input voltage through the dual-channel calibration mechanism. Secondly, the introduction of signal collectors and signal comparators, combined with advanced signal processing algorithms such as Butterworth filtering, Hilbert transform and support vector machines, realizes the precise acquisition, abnormal detection and mode selection of real-time current signals of manual control and automatic operation channels, greatly improving the adaptive response capability and intelligence level of the system under complex working conditions. Furthermore, through the innovative design of the pulse width modulation encoder, the vector modulation algorithm, space vector pulse width modulation and adaptive current limiting algorithm are used to accurately calculate and generate the current-limiting pulse width modulation signal, which not only effectively controls the output current amplitude, but also significantly improves the anti-interference and stability of the system. Further, the introduction of multiple protectors and the optimization of delay controllers, through algorithms such as sequential comparison, dual threshold recursion and neural networks, constructs a comprehensive fault detection, protection and processing mechanism, greatly improving the safety and reliability of the system. In terms of mode switching, the timing controller and relay linkage controller use advanced algorithms such as state transition probability, dynamic programming and Bayesian network to realize the intelligence and optimization of mode switching rules, effectively solving the problems of mode switching impact and signal interference in traditional systems. Finally, through the data collector and communication interface, combined with principal component analysis, Kalman filtering and photoelectric isolation technologies, not only the real-time monitoring and accurate data analysis of the system operation status are realized, but also the security and integrity of data transmission are guaranteed, providing a full range of intelligent control and monitoring solutions for subway train operation, significantly improving the safety, reliability and intelligence level of subway operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0015] Figure 1 This is a schematic diagram of an embodiment of a subway operation dual-mode control method based on pulse width modulation coding in an embodiment of the present application;

[0016] Figure 2 This is a schematic diagram of an embodiment of a subway operation dual-mode control system based on pulse width modulation coding in an embodiment of the present application. DETAILED DESCRIPTION

[0017] The embodiment of the present application provides a dual-mode control method and system for subway operation based on pulse width modulation coding. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0018] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In the embodiment of the present application, an embodiment of the subway operation dual-mode control method based on pulse width modulation coding includes:

[0019] Step S101, based on the nominal power supply voltage value, after defining the initialization voltage range through the power supply measurement module, determine the input terminal voltage detection value, implement dual-path calibration of the manual control and automatic operation channel voltages according to the dual redundant measurement module, and output the manual control detection voltage and the automatic operation detection voltage;

[0020] Step S102, according to the manual control detection voltage and the automatic operation detection voltage, the signal collector collects the real-time current signals of the manual control channel and the automatic operation channel and converts them into manual control signal sampling values ​​and automatic operation signal sampling values, and the signal comparator completes the threshold detection analysis of the abnormal signal to generate a mode selection control signal, including: inputting the manual control detection voltage and the automatic operation detection voltage into the signal collector, performing bandpass filtering on the original voltage signal through the Butterworth filtering algorithm to generate filtered voltage data; for the filtered voltage data, extracting the instantaneous amplitude and phase information of the signal through the Hilbert transform algorithm, A real-time current signal is obtained; based on the real-time current signal, a discrete wavelet transform algorithm is used to perform multi-scale decomposition and extract signal singular points, and a manual control signal sampling value and an automatic operation signal sampling value are generated; the manual control signal sampling value and the automatic operation signal sampling value are combined to construct an abnormal detection parameter matrix through an adaptive dual threshold algorithm, and an abnormal state identification is output; for the abnormal state identification, an abnormal grade coefficient is calculated through a fuzzy clustering algorithm and an abnormal evaluation index is established to generate an abnormal feature vector; based on the abnormal feature vector, a dual-mode classifier is established through a support vector machine algorithm and the mode switching probability is calculated, and a mode selection control signal is output;

[0021] Step S103, according to the mode selection control signal, the pulse width modulation encoder is used to calculate the output duty ratio segmentation parameter to obtain the pulse width modulation waveform, and the output current amplitude is limited by the load resistor to generate a current-limiting pulse width modulation signal;

[0022] Step S104: Based on the current-limiting pulse width modulation signal, the multiple protectors determine the output short circuit and overload status and output a protection trigger signal, start the delay controller to adjust the output relay disconnection timing, and construct a protection execution signal;

[0023] Step S105, collecting protection execution signals, establishing mode switching rules and generating mode switching signals through a timing controller, calling a relay linkage controller to optimize the coordinated state of the output relays, and generating a dual-mode control signal;

[0024] Step S106: Based on the dual-mode control signal, the data collector extracts the real-time operating status parameters of the system to generate status monitoring data, and the communication interface completes the data electrical isolation transmission processing and outputs the system operating parameters.

[0025] It is understandable that the execution subject of the present application can be a subway operation dual-mode control system based on pulse width modulation coding, or a terminal or a server, which is not limited here. The present application embodiment is described by taking a server as the execution subject as an example.

[0026] Specifically, the power supply voltage is initialized, detected and calibrated. The nominal power supply voltage value is 110V DC. When the power supply measurement module measures this voltage value, it is necessary to ensure that the measurement range is between 0.7Un and 1.25Un, that is, between 77V DC and 137.5V DC. The power supply measurement module collects the original voltage signal and processes the voltage signal using a sliding window algorithm. The window length of the sliding window is set to 1ms, and the sliding step is 0.1ms. The voltage reference data is obtained by calculating the effective value of the data in the window. After obtaining the voltage reference data, the voltage fluctuation curve is fitted by the least squares method to extract the fluctuation characteristic parameters. An adaptive threshold algorithm is used in the fitting process. When the voltage value is detected to be out of the preset range, the calibration coefficient is dynamically adjusted. For example, when the input voltage is detected to be 115V DC, the system automatically calculates the calibration coefficient 1.045 to restore the output voltage to the nominal value of 110V DC. The dual redundant measurement module calibrates the voltage of the manual control channel and the automatic operation channel respectively, and uses the Kalman filter algorithm to filter the dual-channel data. The jitter compensation coefficient is introduced into the filtering process to eliminate random fluctuations in the signal.

[0027] The signal collector receives the calibrated manual control detection voltage and automatic operation detection voltage, and collects the current signals of the two channels in real time. For the manual control channel, the input current is 12mA DC at the nominal voltage of 110V DC, 8mA DC at the minimum working voltage of 77V DC, and 15mA DC at the maximum working voltage of 137.5V DC. For the automatic operation channel, the input signal range is 0-20mA DC, and the maximum allowable current is 40mA. The signal comparator uses the Butterworth filter algorithm to bandpass filter the original voltage signal, and extracts the instantaneous amplitude and phase information of the signal through the Hilbert transform. Based on the extracted signal features, the discrete wavelet transform is used for multi-scale decomposition, the signal singular points are extracted, and the manual control signal sampling value and the automatic operation signal sampling value are obtained. The abnormal detection parameter matrix is ​​constructed by the adaptive dual threshold algorithm. When the signal abnormality is detected, the fuzzy clustering algorithm calculates the abnormal grade coefficient and establishes the abnormal evaluation index. The support vector machine algorithm establishes a dual-mode classifier based on the abnormal feature vector, calculates the mode switching probability, and generates a mode selection control signal. Specifically, the manual control detection voltage (nominal value is 110V DC) and the automatic operation detection voltage are input into the signal collector. The signal collector adopts a 12-bit AD converter, and the sampling frequency is set to 10kHz to ensure the accuracy and real-time performance of the collected signal. The Butterworth filter algorithm performs bandpass filtering on the original voltage signal, and the order of the filter is set to 4th order, the passband range is 10Hz-1kHz, and the stopband attenuation is 40dB / decade. The purpose of bandpass filtering is to eliminate DC components and high-frequency interference and retain effective signal components. The filtered voltage data maintains the main characteristics of the signal while significantly reducing the impact of noise. The filtered voltage data is processed by the Hilbert transform algorithm. The Hilbert transform first delays the signal by 90 degrees to form an analytical signal, and then extracts the instantaneous amplitude and phase information of the signal. In this application, the Hilbert transform is used to convert a voltage signal into a current signal. When the manual control channel works at a nominal voltage of 110V DC, the extracted current signal is 12mA DC; when the minimum working voltage is 77V DC, it is 8mA DC; when the maximum working voltage is 137.5V DC, it is 15mA DC. For the automatic operation channel, the input signal range is between 0-20mA DC, and the maximum allowable current is 40mA.

[0028] Based on the real-time current signal obtained, the discrete wavelet transform algorithm is used for signal processing. The discrete wavelet transform uses the db4 wavelet basis function to decompose the signal into 5 layers. In each layer of decomposition, the signal is divided into low-frequency approximate coefficients and high-frequency detail coefficients. The singular points of the signal are extracted by analyzing the energy distribution and mutation characteristics of the coefficients of each layer. The singular points reflect the mutation characteristics of the signal and correspond to the abnormal changes of the current signal. After wavelet reconstruction, the manual control signal sampling value and the automatic operation signal sampling value are obtained respectively. The obtained sampling values ​​are input into the adaptive dual threshold algorithm for anomaly detection. The initial values ​​of the dual thresholds are set to ±20% of the signal mean, and the threshold size is dynamically adjusted as the signal characteristics change. The anomaly detection parameter matrix contains four dimensions: signal amplitude, rate of change, duration and frequency characteristics. Each dimension is set with a corresponding weight coefficient to comprehensively evaluate the degree of abnormality of the signal. When an abnormality is detected, the corresponding abnormal state mark is output, which contains information such as the abnormality type, occurrence time and severity.

[0029] The abnormal state identification is processed by fuzzy clustering algorithm. First, the abnormal state identification is converted into a fuzzy feature vector, and the feature dimensions include amplitude deviation, duration, change rate and frequency characteristics. The cluster center is initially set to three categories: slight abnormality, moderate abnormality and severe abnormality. The abnormal grade coefficient is determined by calculating the membership degree of the sample point to each cluster center. The grade coefficient is combined with the preset evaluation index to establish a complete abnormal evaluation system, and finally generate an abnormal feature vector. Based on the abnormal feature vector, a dual-mode classifier is established using the support vector machine algorithm. The support vector machine uses the RBF kernel function, the kernel parameter σ is set to 1.0, and the penalty factor C is set to 100. The training sample contains data on normal operating status and various abnormal states, and the optimal classification parameters are determined by cross-validation. The classifier outputs the probability value of the two modes, and decides whether to switch the mode according to the size of the probability value. When the switching probability exceeds the preset threshold (such as 0.85), the corresponding mode selection control signal is output.

[0030] For example, during the operation of the train, the manual control detection voltage is 108V DC, and the automatic operation detection voltage is 112V DC. After Butterworth filtering, the signal-to-noise ratio of the signal is increased from 15dB to 28dB. The manual control channel current extracted by Hilbert transform is 11.5mA DC, and the automatic operation channel current is 16.8mA DC. After the discrete wavelet transform is decomposed into 5 layers, a significant singular point is detected in the third layer, corresponding to the time point t=15ms. The adaptive dual threshold algorithm dynamically adjusts the abnormal detection threshold to 9.2mA and 18.5mA. Fuzzy clustering analysis shows that the abnormality is a moderate abnormality with an abnormality level coefficient of 0.68. The support vector machine classifier calculates the probability of manual mode to be 0.25 and the probability of automatic mode to be 0.75, and outputs the control signal of the automatic operation mode accordingly. The time delay of the entire signal processing process from sampling to output of the control signal is controlled within 50ms, meeting the requirements of real-time control.

[0031] After receiving the mode selection control signal, the pulse width modulation encoder decomposes the signal into a frequency domain feature sequence through fast Fourier transform and extracts the carrier modulation parameters. The vector modulation algorithm is used to calculate the switching timing, construct the pulse sequence, and generate the segmented modulation parameters. The space vector pulse width modulation algorithm reconstructs the waveform characteristics according to the high and low level timing, and outputs a PWM signal with a frequency of 500Hz (±0.5%) and a peak-to-peak voltage of 24Vpp (±0.5%). The dead time compensation algorithm is used to eliminate the switching timing interference and introduce an overcurrent protection threshold. The multiple protectors monitor the PWM output signal in real time. When the PWM is 10%, the output current at the nominal voltage is 50mA DC; when the PWM is 90%, the output current is 450mA DC; when the PWM is 100%, the output current is 500mA DC. The sequential comparison algorithm identifies the short-circuit characteristic waveform and marks the fault time point. The double threshold recursive algorithm calculates the overload accumulation. When the overload state is detected to last longer than the preset threshold, a protection trigger signal is generated. The timing controller establishes mode switching rules according to the protection execution signal and constructs the mode conversion map through the state transition probability algorithm. The dynamic programming algorithm calculates the optimal switching path, and the Bayesian network algorithm evaluates the switching reliability. The relay linkage controller uses the response surface algorithm to optimize the switching speed, the real-time sequence analysis algorithm to match the linkage characteristics, and the multi-objective optimization algorithm to balance the switching stability and response speed.

[0032] The data collector monitors the system operation status in real time, the principal component analysis algorithm extracts key state characteristics, and the sliding time window algorithm calculates the parameter change trend. The Kalman filter algorithm eliminates measurement noise, and the differential coding algorithm compresses the data stream. The communication interface uses photoelectric isolation technology to establish a data transmission channel to ensure signal integrity.

[0033] For example, during the operation of a subway train, the initial power supply voltage is 112V DC. After sampling by the power supply measurement module, the sliding window algorithm calculates the effective value of the voltage to be 111.8V DC. After least squares fitting, the fluctuation amplitude is ±0.5V. The adaptive threshold algorithm calculates the calibration coefficient of 0.982 and adjusts the output voltage to 110V DC. The signal collector detects that the current of the manual control channel is 11.5mA DC and the current of the automatic operation channel is 15.2mA DC. After Butterworth filtering, the signal carrier-to-noise ratio is improved by 3.5dB. The abnormal point extracted by discrete wavelet transform is located at 18.5ms, and the abnormal level coefficient is 0.85. The pulse width modulation encoder outputs a 500Hz PWM signal with a duty cycle of 65%. After the load resistor current limiting, the output current is 325mA DC. The multiple protectors detect a short-term overload with a duration of 25ms, and the delay controller sets the disconnection delay to 35ms. The switching reliability of the mode switching timing control reaches 0.98, and the relay action response time is 0.5s. The sampling frequency of data acquisition is 10kHz, and the data transmission rate after differential coding compression is 115.2kbps.

[0034] In the embodiment of the present application, a dual redundant measurement module based on the nominal supply voltage value is adopted to realize the precise voltage calibration of the manual control and automatic operation channels, effectively overcome the system complexity and low reliability caused by the independent hardware circuit in the traditional system, and ensure the accuracy and stability of the input voltage through the dual-channel calibration mechanism. Secondly, the introduction of signal collectors and signal comparators, combined with advanced signal processing algorithms such as Butterworth filtering, Hilbert transform and support vector machines, realizes the accurate collection, abnormality detection and mode selection of real-time current signals of manual control and automatic operation channels, greatly improving the adaptive response capability and intelligence level of the system under complex working conditions. Furthermore, through the innovative design of the pulse width modulation encoder, the vector modulation algorithm, space vector pulse width modulation and adaptive current limiting algorithm are used to accurately calculate and generate the current limiting pulse width modulation signal, which not only effectively controls the output current amplitude, but also significantly improves the anti-interference and stability of the system. Further, the introduction of multiple protectors and the optimization of delay controllers, through algorithms such as sequential comparison, dual threshold recursion and neural networks, constructs a comprehensive fault detection, protection and processing mechanism, greatly improving the safety and reliability of the system. In terms of mode switching, the timing controller and relay linkage controller use advanced algorithms such as state transition probability, dynamic programming and Bayesian network to realize the intelligence and optimization of mode switching rules, effectively solving the problems of mode switching impact and signal interference in traditional systems. Finally, through the data collector and communication interface, combined with principal component analysis, Kalman filtering and photoelectric isolation technologies, not only the real-time monitoring and accurate data analysis of the system operation status are realized, but also the security and integrity of data transmission are guaranteed, providing a full range of intelligent control and monitoring solutions for subway train operation, significantly improving the safety, reliability and intelligence level of subway operation.

[0035] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0036] (1) Collect the original sampling data of the nominal power supply voltage value, calculate the voltage effective value through the sliding window algorithm, and obtain the voltage reference data;

[0037] (2) Based on the voltage reference data, the voltage fluctuation curve is fitted by the least squares method and the fluctuation characteristic parameters are extracted to obtain the initialization voltage range;

[0038] (3) Compare the initialization voltage range with the preset threshold, dynamically adjust the calibration coefficient through the adaptive threshold algorithm, and generate the input voltage detection value;

[0039] (4) Introducing jitter compensation coefficients into the input voltage detection values, performing dual-channel data filtering based on the Kalman filter algorithm, and outputting dual-channel calibration reference values;

[0040] (5) Based on the dual-channel calibration reference value, the signal noise is separated and the filtered signal is reconstructed through the wavelet transform algorithm to generate the channel voltage feature vector;

[0041] (6) The channel voltage feature vector is processed by time series superposition to extract the peak envelope, and the frequency domain analysis is performed using the Fourier transform algorithm to output the manual control detection voltage and the automatic operation detection voltage.

[0042] Specifically, the nominal supply voltage is processed. The nominal supply voltage value is set to 110V DC, and the original sampling data is collected at a sampling frequency of 10kHz through a high-speed sampling circuit. The collected voltage data is processed by a sliding window algorithm, and the sliding window length is set to 100 sampling points (i.e., 10ms), and the sliding step is 10 sampling points (i.e., 1ms). In each window, when calculating the effective value of the voltage, the square sum of the data in the window is first calculated, and then the square root is taken to obtain the effective value of the voltage. At the same time, the corresponding timestamp is recorded to form a voltage reference data sequence. Based on the obtained voltage reference data, the voltage fluctuation curve is fitted by the least squares method. First, the voltage reference data is arranged in time series, a data point set is established, and then the voltage fluctuation curve is obtained by polynomial fitting. In the fitting process, a 5th-order polynomial is selected as the fitting function, and the polynomial coefficient is determined by calculating the minimum value of the residual sum of squares. The fitted curve is analyzed to extract the fluctuation characteristic parameters, including fluctuation amplitude, fluctuation period, fluctuation trend, etc. According to the fluctuation characteristic parameters, the initialization voltage range is determined to be within the range of ±30% of the nominal voltage value, that is, 77V DC to 137.5V DC.

[0043] After obtaining the initialization voltage range, dynamic calibration is required through the adaptive threshold algorithm. The preset thresholds include an upper limit of 137.5V DC and a lower limit of 77V DC. The adaptive threshold algorithm dynamically calculates the calibration coefficient based on the deviation between the current voltage value and the preset threshold. The calibration coefficient is calculated using a proportional-integral algorithm, with the proportional coefficient set to 0.8 and the integral time constant set to 5ms. When the voltage value deviates from the nominal value, the calibration coefficient is adjusted in time to keep the voltage value within a reasonable range. For example, when the voltage value is detected to rise to 120V DC, the calibration coefficient is automatically adjusted to 0.917, and the voltage value is corrected to 110V DC. The input voltage detection value is processed by jitter compensation, and the jitter compensation coefficient is automatically adjusted according to the voltage change rate. The larger the change rate, the smaller the compensation coefficient. Independent jitter compensation coefficients are introduced for the manual control channel and the automatic operation channel respectively. The Kalman filter algorithm filters the dual-channel data. The state equation contains two state quantities: voltage value and voltage change rate, and the observation equation is the measured voltage value. The process noise covariance matrix and the measurement noise covariance matrix are calibrated by actual operation data and are set to diagonal matrices with diagonal elements of 0.01 and 0.1, respectively.

[0044] Based on the obtained dual-channel calibration reference value, the wavelet transform algorithm is used for signal processing. The db4 wavelet is selected as the base wavelet, and the signal is decomposed into 5 layers. In the wavelet domain, the coefficients of each layer are denoised by the soft threshold method, and the threshold is taken as 3 times the standard deviation of the coefficients of each layer. When reconstructing the signal, the denoised wavelet coefficients are inversely transformed to obtain the filtered signal. At the same time, by analyzing the energy distribution of each scale coefficient, the signal characteristics are extracted to form the channel voltage feature vector. The channel voltage feature vector is processed in time series, and a sliding window with a time window of 20ms is used for superposition analysis. In each time window, the peak envelope of the signal is extracted, and the amplitude and corresponding time of the peak point are recorded. The Fourier transform algorithm performs frequency domain analysis on the envelope signal and calculates the spectrum characteristics, including the fundamental component, harmonic component and its phase relationship. Through the analysis of frequency domain characteristics, the precise values ​​of the manual control detection voltage and the automatic operation detection voltage are finally output.

[0045] For example, during the operation of a certain train, the power supply voltage is monitored in real time. The voltage data obtained by the original sampling fluctuates between 108V and 112V, and the voltage effective value is calculated to be 110.5V DC through a 10ms sliding window. The fluctuation curve obtained after least squares fitting shows that the voltage fluctuation amplitude is ±2V, and the fluctuation period is about 100ms. The adaptive threshold algorithm calculates the calibration coefficient as 0.995, and adjusts the voltage value to 110V DC. The dual-channel data is jitter compensated, and the compensation coefficients of the manual control channel and the automatic operation channel are 0.98 and 0.97 respectively. After Kalman filtering, the root mean square error of the signal is reduced from 0.8V to 0.2V. After the 5-layer decomposition of the wavelet transform, the energy is mainly concentrated in the first three layers, corresponding to the frequency range of 0-500Hz. After reconstructing the signal, the ripple coefficient of the manual control channel is reduced to 0.5%, and the ripple coefficient of the automatic operation channel is reduced to 0.4%. The final output manual control detection voltage is 110.2V DC, and the automatic operation detection voltage is 109.8V DC, which meets the control accuracy requirements.

[0046] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0047] (1) Decompose the mode selection control signal into a frequency domain feature sequence using the fast Fourier transform algorithm, extract the carrier modulation parameters using bilateral spectrum analysis, and generate duty cycle benchmark data;

[0048] (2) Based on the duty cycle benchmark data, the switching timing is calculated using a vector modulation algorithm and a pulse sequence is constructed to generate segmented modulation parameters;

[0049] (3) Importing the segmented modulation parameters into the space vector pulse width modulation algorithm, reconstructing the waveform characteristics according to the high and low level timing sequence, and obtaining the pulse width modulation waveform;

[0050] (4) The dead time compensation algorithm is introduced into the pulse width modulation waveform to eliminate the switching timing interference, and the overcurrent protection threshold is superimposed to output the standard modulation waveform;

[0051] (5) Using the standard modulation waveform as input, an adaptive current limiting algorithm is used to dynamically calculate the load parameters and construct a current limiting control vector;

[0052] (6) The current limiting control vector is closed-loop controlled through a proportional-integral regulation algorithm and the waveform distortion rate is optimized to generate a current limiting pulse width modulation signal.

[0053] Specifically, the mode selection control signal is converted to the frequency domain by the fast Fourier transform algorithm. The sampling frequency is set to 10kHz, the number of sampling points is 1024 points, and the frequency domain feature sequence is obtained after the signal is decomposed by FFT. The double-sided spectrum analysis technology processes the frequency domain feature sequence to extract the 500Hz carrier frequency component and the corresponding amplitude and phase information. The carrier modulation parameters include modulation depth and modulation frequency, and the duty cycle reference data is calculated by the energy ratio of the harmonic components. The vector modulation algorithm is used to process the acquired duty cycle reference data. The vector modulation algorithm first establishes the switch state vector space, which contains eight basic voltage vectors, six of which are valid vectors and two are zero vectors. Based on the linear combination principle of voltage vectors, the switch timing in each sector is calculated. When the PWM duty cycle is 10%, the output current is 50mA DC; when the PWM duty cycle is 90%, the output current is 450mA DC; when the PWM duty cycle is 100%, the output current reaches 500mA DC. According to the switching timing, a complete pulse sequence is constructed and segmented modulation parameters are generated.

[0054] After receiving the segmented modulation parameters, the space vector pulse width modulation algorithm is processed according to the seven-segment PWM modulation strategy. In one switching cycle, the output voltage is precisely controlled by reasonably arranging the action time of two adjacent effective vectors and zero vectors. During the reconstruction of the high and low level timing, the output frequency is ensured to be 500Hz (±0.5%) and the peak-to-peak voltage is 24Vpp (±0.5%). After waveform reconstruction, a standard pulse width modulation waveform is obtained. In order to eliminate the switching timing interference, a dead time compensation algorithm is introduced. The dead time is set to 2μs, and a dead time delay is inserted at each switch state transition to prevent the upper and lower bridge arms from being turned on at the same time. At the same time, an overcurrent protection threshold is set, and the protection mechanism is triggered when the output current exceeds the threshold. The setting of the overcurrent protection threshold is based on the load characteristics and is dynamically adjusted under different working conditions. After dead zone compensation and overcurrent protection processing, a standard modulation waveform is output.

[0055] The dynamic calculation of load parameters adopts the adaptive current limiting algorithm. The load parameter calculation formula is:

[0056]

[0057] in: represents the equivalent load impedance, Indicates the effective value of PWM output voltage, Indicates the effective value of the output current, Indicates the compensation coefficient (value range 1.0-1.2), Indicates the stability factor (value range 0.8-1.0), Indicates the adjustment coefficient (value range 0.9-1.1), Indicates the power factor angle.

[0058] Based on the calculated load parameters, the current limiting control vector is constructed. The current limiting control adopts the proportional integral regulation algorithm for closed-loop control, where the proportional coefficient Kp is set to 0.8 and the integral time constant Ti is set to 10ms. The output current is always kept within the set range by adjusting the duty cycle of the PWM waveform in real time. At the same time, the distortion rate of the output waveform is monitored. When the distortion rate exceeds 5%, the control parameters are automatically adjusted for optimization.

[0059] For example: During the operation of the train, after receiving the mode selection control signal, the frequency domain feature sequence is obtained through fast Fourier transform, and the main frequency component is at 500Hz with an amplitude of 0.85. The carrier modulation parameters extracted by bilateral spectrum analysis show that the modulation depth is 0.75. The vector modulation algorithm calculates the switching timing within the 60° electrical angle sector, and the effective vector action time ratio is 3:2. The waveform frequency of the space vector PWM output is stable at 500Hz, and the peak-to-peak voltage is 24.1Vpp. After dead time compensation, the switching transient oscillation amplitude is reduced by 80%. The equivalent load impedance calculated by the adaptive current limiting algorithm is 48Ω, and the corresponding current limiting threshold is set to 500mA. The waveform distortion rate of the final output current limiting pulse width modulation signal is controlled within 3.5%, which effectively ensures the quality of the output current.

[0060] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0061] (1) Import the current-limited pulse width modulation signal, identify the short-circuit characteristic waveform through the sequential comparison algorithm, mark the fault time point, and obtain the short-circuit detection data;

[0062] (2) For the short-circuit detection data, the overload accumulation is calculated and the overload duration is calibrated through the double-threshold recursive algorithm to generate the fault state matrix;

[0063] (3) Input the fault state matrix into the pattern recognition algorithm, extract the fault type characteristics and quantify the fault level coefficient to generate a protection trigger signal;

[0064] (4) Based on the protection trigger signal, a delay strategy model is constructed using a neural network algorithm to calculate the optimal disconnection timing and output relay control instructions;

[0065] (5) Based on the relay control instructions, the fuzzy reasoning algorithm is used to evaluate the protection action priority and set the execution sequence to construct the protection action sequence;

[0066] (6) Combined with the protection action sequence, the execution timing is optimized and the coordinated control parameters are calculated through the time series analysis algorithm to generate a protection execution signal.

[0067] Specifically, the current-limiting pulse width modulation signal is imported, and the signal characteristics are analyzed by the sequential comparison algorithm. The sequential comparison algorithm adopts a sliding window method, with the window length set to 2ms and the step length to 0.1ms. In each window, the short-circuit characteristic waveform is identified by comparing the amplitude change rate of adjacent sampling points. The short-circuit characteristic waveform is characterized by a sharp rise in current and a rapid drop in voltage. When the current rise rate is detected to exceed the preset threshold (such as 10A / ms), the moment is marked as the fault time point, and the corresponding waveform characteristic parameters are recorded, including the current value before the fault, the current value after the fault, the voltage drop, and other information. After obtaining the short-circuit detection data, the double-threshold recursive algorithm is used for overload analysis. The double threshold includes the alarm threshold and the trip threshold, which are set to 120% and 150% of the rated current, respectively. The recursive algorithm accumulates the overload state, and the accumulation is related to the overload multiple and duration. When the current exceeds the alarm threshold but does not reach the trip threshold, the overload duration is recorded; when the current exceeds the trip threshold, the protection action is triggered immediately. The accumulation process of overload status forms a fault status matrix, and each element in the matrix contains information in three dimensions: overload multiple, duration, and accumulation.

[0068] The fault state matrix is ​​input into the pattern recognition algorithm to extract the fault type characteristics. The pattern recognition algorithm uses a cluster analysis-based method to classify faults into instantaneous short circuits, continuous short circuits, mild overloads, severe overloads, and other types. The quantitative calculation of the fault level coefficient uses the following formula:

[0069]

[0070] in: represents the failure level coefficient, Indicates the fault current value, Indicates the rated current value, Indicates the fault duration, Indicates the voltage drop, Indicates the cumulative amount of faults, , , represents the weight coefficient, represents the current proportionality index, represents the time constant, represents the normalization factor, Indicates the cumulative amount adjustment coefficient.

[0071] The protection trigger signal is generated according to the quantized fault level coefficient. The protection trigger signal contains information such as fault type, fault level, and trigger time. The neural network algorithm is trained based on historical fault data to obtain a delay strategy model. The input layer contains fault feature parameters, the hidden layer uses 10 neurons, and the output layer gives the optimal disconnection sequence. For mild overload, the delay time is relatively long, and an alarm signal is given; for severe short circuit, the disconnection action is triggered immediately. Based on the relay control instructions output by the neural network, the fuzzy reasoning algorithm is used to evaluate the priority of the protection action. The input variables of fuzzy reasoning include factors such as fault level, load importance, and power supply redundancy, and the output variable is the priority of the protection action. The fuzzy rule base contains multiple IF-THEN rules, which are defuzzified by the centroid method to obtain a clear priority value. The execution sequence of the protection action is set according to the priority to construct a complete protection action sequence.

[0072] The time series analysis algorithm optimizes the protection action sequence, mainly considering the coordination and reliability of the action sequence. A sliding time window is used for data analysis with a window length of 100ms to extract the timing characteristics and calculate the correlation index. According to the results of the correlation analysis, the execution time interval of each protection action is adjusted to ensure the orderly execution of the protection action. The collaborative control parameters include action sequence, execution order, interlocking conditions, etc., and finally generate a protection execution signal in a standard format.

[0073] For example, in practical applications, a short circuit fault is detected by a current-limited pulse width modulation signal. The current rises from 300mA to 900mA within 0.5ms, and the voltage drops from 110V to 65V. The sequential comparison algorithm calculates the current rise rate to be 1.2A / ms, and marks the time point of the fault occurrence as t=15.8ms. The double-threshold recursive algorithm determines that the fault exceeds the trip threshold (600mA), the duration reaches 1.5ms, and the cumulative fault volume reaches 1.35. The pattern recognition algorithm determines the fault as an instantaneous short circuit, and the fault level coefficient is calculated to be 0.82. The optimal disconnection delay given by the neural network is 2ms. After fuzzy reasoning evaluation, the priority of the protection action is 0.9 (full score 1.0). Time series analysis determines that the relay disconnection timing is t=18.3ms, and the entire protection action is controlled within 2.5ms from the occurrence of the fault to the completion of the execution.

[0074] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0075] (1) Extract the timing features in the protection execution signal, construct the mode transition map through the state transition probability algorithm, and generate the switching timing matrix;

[0076] (2) For the switching timing matrix, the optimal switching path between modes is calculated through a dynamic programming algorithm to obtain the mode switching rules;

[0077] (3) Input the mode switching rules into the Bayesian network algorithm, evaluate the switching reliability and calibrate the switching priority, and generate a mode switching signal;

[0078] (4) Based on the mode switching signal, the response surface algorithm is used to establish the relay action timing rules and optimize the switching speed, and the relay state instructions are output;

[0079] (5) Match the relay state instructions with linkage characteristics through a real-time sequence analysis algorithm to construct a collaborative control vector;

[0080] (6) A multi-objective optimization algorithm is used to balance the switching stability and response speed of the collaborative control vector to generate a dual-mode control signal.

[0081] Specifically, the timing features are extracted from the protection execution signal. The timing features include parameters such as switching time point, switching duration, and switching interval. The state transition probability algorithm adopts the Markov chain model, and the transition probability between each state is statistically calculated based on historical data to construct a state transition matrix. For the switch from manual control to automatic operation, the transition probability threshold is set to 0.85; for the switch from automatic operation to manual control, the transition probability threshold is set to 0.90. The transition probability matrix is ​​combined with the timing features to form a mode transition map, which contains all possible switching paths and their probability weights. The switching timing matrix records the timing parameters of different switching paths, including the preparation time before switching, the switching execution time, and the stabilization time after switching. The dynamic programming algorithm optimizes the switching timing matrix, setting the objective function to minimize the total switching time and ensure switching stability. At the nominal voltage of 110V DC, the response time for switching from manual control mode to automatic operation mode is 0.5s, and the response time for switching from automatic operation mode to manual control mode is 2ms. The algorithm calculates the optimal solution of each sub-problem through forward recursion, and finally obtains a complete mode switching rule.

[0082] The Bayesian network algorithm evaluates the reliability of mode switching according to the switching rules. The network structure contains multiple nodes: switching condition nodes, execution status nodes, fault status nodes, etc. The conditional probability between nodes is obtained through historical data training, and the maximum likelihood estimation method is used. The calculation of switching reliability comprehensively considers factors such as signal quality, load status, and environmental conditions, and sets the reliability threshold to 0.95. The switching priority is determined according to the reliability evaluation results, and the priority is divided into three levels, corresponding to different switching strategies. The response surface algorithm establishes a relay action model based on the mode switching signal. The response surface adopts a second-order polynomial model. The independent variables include switching time, load current, power supply voltage, etc., and the dependent variable is the relay action sequence. The level of each factor is determined by orthogonal experimental design, and a complete response surface equation is established. The optimization goal is to minimize the switching time and ensure a smooth switching process. When the input current is less than 2mA, the relay contacts are disconnected; when the input current is greater than 2mA, the relay contacts are closed, and the contact action time is controlled within 35ms.

[0083] The real-time sequence analysis algorithm processes the relay state instructions. The autoregressive moving average model is used, the prediction window length is set to 50ms, and the prediction step is 5ms. In the process of linkage characteristic matching, the mechanical and electrical characteristics of the relay group are considered, including parameters such as the pull-in time, release time, and contact bounce. The collaborative control vector contains multiple dimensions: relay state, action sequence, interlocking conditions, protection strategy, etc. The multi-objective optimization algorithm uses NSGA-II (non-dominated sorting genetic algorithm) to optimize the collaborative control vector. Set the population size to 100, the number of iterations to 50, the crossover probability to 0.8, and the mutation probability to 0.1. The optimization objectives include: minimizing switching time, maximizing switching stability, and minimizing energy loss. Constraints include relay action time limit, voltage and current range limit, etc. Through non-dominated sorting and congestion calculation, the optimal solution set is obtained, and a dual-mode control signal in a standard format is generated.

[0084] For example, during the operation of the train, it is necessary to switch from manual control mode to automatic operation mode. The protection execution signal shows that the current voltage is 108V DC and the current is 11mA. The state transition probability algorithm calculates the probability of the switching path to be 0.88, which exceeds the set threshold of 0.85. The dynamic programming algorithm calculates the optimal switching path: first disconnect the manual control relay (K1), and then close the automatic operation relay (K2) after a delay of 0.5s. The switching reliability obtained by the Bayesian network evaluation is 0.97, which is calibrated as the highest priority. The response surface algorithm calculates the relay action sequence according to the current working conditions: the K1 disconnection time is t1=0ms, and the K2 closing time is t2=500ms. Real-time sequence analysis predicts that the current change trend is stable during the switching process, and the maximum fluctuation does not exceed ±1mA. The switching time obtained after multi-objective optimization is 485ms, and the stability index is 0.96, which meets the control requirements. The entire switching process is completed smoothly, realizing the disturbance-free switching of the dual mode.

[0085] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0086] (1) Import the dual-mode control signal, extract the key state feature quantities through the principal component analysis algorithm, construct the feature space, and generate the state feature vector;

[0087] (2) For the state feature vector, the real-time change trend of each parameter is calculated through the sliding time window algorithm to obtain the operation trend data;

[0088] (3) Input the operation trend data into the Kalman filter algorithm to eliminate measurement noise and optimize data accuracy to generate condition monitoring data;

[0089] (4) Based on the status monitoring data, the data stream is compressed using a differential coding algorithm and a check bit is added, and the transmission data packet is output;

[0090] (5) Based on the transmission data packet, the photoelectric isolation algorithm is used to establish the data transmission channel and ensure the signal integrity, and to construct the isolated transmission sequence;

[0091] (6) Combined with the isolated transmission sequence, the system operation status is reconstructed and the data validity is verified through the data recovery algorithm to generate the system operation parameters.

[0092] Specifically, a dual-mode control signal is imported, including operating parameters such as voltage, current, and PWM duty cycle. The principal component analysis algorithm performs dimensionality reduction processing on the original data, and selects several principal components with the largest contribution rate by calculating the eigenvalues ​​and eigenvectors of the covariance matrix. For voltage data, the fluctuation characteristics near the nominal voltage 110V DC are retained; for current data, the change characteristics in the range of 0-20mA are extracted; for PWM signals, the modulation characteristics in the range of 10%-90% duty cycle are focused on. After feature extraction, a state feature vector is constructed in a multidimensional feature space. The generated state feature vector is processed using a sliding time window algorithm, with the window length set to 100ms and the sliding step size set to 10ms. In each time window, the statistical characteristics of each parameter, such as the mean, variance, and change rate, are calculated respectively. The change trend of the voltage parameter reflects the power supply stability, the change trend of the current parameter reflects the load state, and the change trend of the PWM duty cycle reflects the control output characteristics. By analyzing the data of multiple consecutive time windows, the real-time change trend of each parameter is obtained to form the operation trend data.

[0093] The operation trend data is processed by the Kalman filter algorithm to establish a state space model. The state variables include voltage value, current value, duty cycle, etc., and the observed variables are measured data. During the filtering process, the prediction step is set to 1ms, and the measurement noise covariance and process noise covariance are determined by experimental calibration. The filtering algorithm performs real-time correction on the measured data, effectively eliminating random fluctuations and measurement errors, and improving data accuracy. The processed data constitutes a state monitoring data set, which contains the exact value and change trend of each parameter.

[0094] The status monitoring data is compressed by a differential encoding algorithm, and the difference between adjacent data points is used to replace the original data. Low-bit encoding is used for parameters that change slowly, and high-bit encoding is used for parameters that change dramatically. During the encoding process, a 16-bit CRC checksum is added to each data packet for error detection at the receiving end. The format of the data packet includes: header identifier (8 bits), data length (16 bits), compressed data, and checksum, which eventually form a transmission data packet in a standard format. The transmission data packet is processed by photoelectric isolation to achieve electrical isolation. Photoelectric isolation uses high-speed optocouplers with a switching speed of 1MHz and an isolation voltage level of 1500V. At the transmitting end, the digital signal is converted into an optical signal; at the receiving end, the optical signal is restored to an electrical signal. Through photoelectric isolation, common-mode interference and surge impact are effectively blocked to ensure the reliability of data transmission. At the same time, a two-way communication mechanism is established to achieve reliable data transmission and status feedback.

[0095] At the receiving end, the data recovery algorithm first performs a CRC check to verify data integrity. The data that passes the check is decompressed to restore the original data sequence. Different types of parameters are classified and processed according to the data identifier to reconstruct the complete system operation status. By comparing the data at the sending and receiving ends, the transmission bit error rate is calculated and the communication quality is evaluated. The final output system operation parameters include real-time operation status, trend analysis results and abnormal alarm information.

[0096] For example, during the operation of a certain train, the dual-mode control signal contains three sets of parameters: voltage, current, and PWM. The principal component analysis selected three principal components with a contribution rate of more than 85%, and the data dimension was reduced from the original 12 dimensions to 3 dimensions. The sliding time window analysis showed that the voltage fluctuated between 109.5V and 110.5V, with a standard deviation of 0.2V; the current was stable at 11.8mA, with a fluctuation range of ±0.5mA; the PWM duty cycle was 65%, and the modulation was stable. After Kalman filtering, the voltage measurement accuracy was improved to ±0.1V, and the current measurement accuracy was improved to ±0.1mA. Differential encoding compressed 1000 bytes of raw data to 400 bytes, with a compression ratio of 2.5:1. The bit error rate of the optoelectronic isolation channel was lower than 10^-9, and the data transmission delay was controlled within 1ms. After data recovery, the consistency with the original data reached 99.99%, meeting the requirements of real-time monitoring. The entire data acquisition and transmission process has good real-time performance and strong anti-interference ability, providing reliable data support for dual-mode control.

[0097] The above describes the subway operation dual-mode control method based on pulse width modulation coding in the embodiment of the present application. The following describes the subway operation dual-mode control system based on pulse width modulation coding in the embodiment of the present application. Figure 2 In the embodiment of the present application, an embodiment of a subway operation dual-mode control system based on pulse width modulation coding includes:

[0098] The calibration module 201 is used to determine the input voltage detection value based on the nominal power supply voltage value after the power supply measurement module defines the initialization voltage range, implement the dual-path calibration of the manual control and automatic operation channel voltages according to the dual redundant measurement module, and output the manual control detection voltage and the automatic operation detection voltage;

[0099] The acquisition module 202 is used to acquire the real-time current signals of the manual control channel and the automatic operation channel by a signal collector according to the manual control detection voltage and the automatic operation detection voltage and convert them into manual control signal sampling values ​​and automatic operation signal sampling values, complete the threshold detection analysis of abnormal signals through a signal comparator, and generate a mode selection control signal, including: inputting the manual control detection voltage and the automatic operation detection voltage into the signal collector, performing bandpass filtering on the original voltage signal through a Butterworth filtering algorithm to generate filtered voltage data; extracting the instantaneous amplitude and phase information of the signal from the filtered voltage data through a Hilbert transform algorithm, A real-time current signal is obtained; based on the real-time current signal, a discrete wavelet transform algorithm is used to perform multi-scale decomposition and extract signal singular points, and a manual control signal sampling value and an automatic operation signal sampling value are generated; the manual control signal sampling value and the automatic operation signal sampling value are used to construct an abnormality detection parameter matrix through an adaptive dual threshold algorithm, and an abnormal state identification is output; for the abnormal state identification, an abnormality level coefficient is calculated through a fuzzy clustering algorithm and an abnormality evaluation index is established to generate an abnormal feature vector; based on the abnormal feature vector, a dual-mode classifier is established through a support vector machine algorithm and a mode switching probability is calculated, and the mode selection control signal is output;

[0100] A calculation module 203 is used to select a control signal according to the mode, calculate the output duty ratio segmentation parameter through a pulse width modulation encoder to obtain a pulse width modulation waveform, limit the output current amplitude through a load resistor, and generate a current-limiting pulse width modulation signal;

[0101] The output module 204 is used to determine the output short circuit and overload status through multiple protectors based on the current-limiting pulse width modulation signal and output a protection trigger signal, start the delay controller to adjust the output relay disconnection timing, and construct a protection execution signal;

[0102] A generation module 205 is used to collect the protection execution signal, establish a mode switching rule and generate a mode switching signal through a timing controller, call a relay linkage controller to optimize the coordinated state of the output relay, and generate a dual-mode control signal;

[0103] The transmission module 206 is used to extract the real-time operating status parameters of the system through the data collector based on the dual-mode control signal to generate status monitoring data, complete the data electrical isolation transmission processing through the communication interface, and output the system operating parameters.

[0104] Through the cooperation of the above components, the dual redundant measurement module based on the nominal supply voltage value is adopted to realize the precise voltage calibration of the manual control and automatic operation channels, effectively overcoming the system complexity and low reliability caused by the independent hardware circuit in the traditional system, and ensuring the accuracy and stability of the input voltage through the dual-channel calibration mechanism. Secondly, the introduction of signal collector and signal comparator, combined with advanced signal processing algorithms such as Butterworth filtering, Hilbert transform and support vector machine, realizes the precise acquisition, abnormality detection and mode selection of real-time current signals of manual control and automatic operation channels, greatly improving the adaptive response capability and intelligence level of the system under complex working conditions. Furthermore, through the innovative design of pulse width modulation encoder, using vector modulation algorithm, space vector pulse width modulation and adaptive current limiting algorithm, the current limiting pulse width modulation signal is accurately calculated and generated, which not only effectively controls the output current amplitude, but also significantly improves the anti-interference and stability of the system. Furthermore, the introduction of multiple protectors and the optimization of delay controllers, through sequential comparison, dual threshold recursion and neural network algorithms, a comprehensive fault detection, protection and processing mechanism is constructed, which greatly improves the safety and reliability of the system. In terms of mode switching, the timing controller and relay linkage controller use advanced algorithms such as state transition probability, dynamic programming and Bayesian network to realize the intelligence and optimization of mode switching rules, effectively solving the problems of mode switching impact and signal interference in traditional systems. Finally, through the data collector and communication interface, combined with principal component analysis, Kalman filtering and photoelectric isolation technologies, not only the real-time monitoring and accurate data analysis of the system operation status are realized, but also the security and integrity of data transmission are guaranteed, providing a full range of intelligent control and monitoring solutions for subway train operation, significantly improving the safety, reliability and intelligence level of subway operation.

[0105] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A dual-mode control method for subway operation based on pulse width modulation coding, characterized in that: The subway operation dual-mode control method based on pulse width modulation coding includes: Based on the nominal power supply voltage value, the input terminal voltage detection value is determined after the power supply measurement module defines the initialization voltage range, and the dual-path calibration of the manual control and automatic operation channel voltages is implemented according to the dual redundant measurement module to output the manual control detection voltage and the automatic operation detection voltage; According to the manual control detection voltage and the automatic operation detection voltage, the real-time current signals of the manual control channel and the automatic operation channel are collected by the signal collector and converted into manual control signal sampling values ​​and automatic operation signal sampling values, and the threshold detection analysis of the abnormal signal is completed by the signal comparator to generate a mode selection control signal, including: inputting the manual control detection voltage and the automatic operation detection voltage into the signal collector, performing bandpass filtering on the original voltage signal through the Butterworth filtering algorithm to generate filtered voltage data; extracting the instantaneous amplitude and phase information of the signal from the filtered voltage data through the Hilbert transform algorithm to obtain the real-time current flow signal; based on the real-time current signal, multi-scale decomposition is performed and signal singular points are extracted through discrete wavelet transform algorithm to generate manual control signal sampling values ​​and automatic operation signal sampling values; the manual control signal sampling values ​​and the automatic operation signal sampling values ​​are used to construct an abnormality detection parameter matrix through an adaptive dual threshold algorithm, and an abnormal state identification is output; for the abnormal state identification, the abnormality level coefficient is calculated through a fuzzy clustering algorithm and an abnormality evaluation index is established to generate an abnormal feature vector; based on the abnormal feature vector, a dual-mode classifier is established through a support vector machine algorithm and the mode switching probability is calculated, and the mode selection control signal is output; According to the mode selection control signal, a pulse width modulation waveform is obtained after calculating the output duty ratio segmentation parameters through a pulse width modulation encoder, and the output current amplitude is limited by a load resistor to generate a current-limiting pulse width modulation signal; Based on the current-limiting pulse width modulation signal, the multiple protectors determine the output short circuit and overload status and output a protection trigger signal, start the delay controller to adjust the output relay disconnection timing, and construct a protection execution signal; Collect the protection execution signal, establish a mode switching rule and generate a mode switching signal through a timing controller, call the relay linkage controller to optimize the coordinated state of the output relay, and generate a dual-mode control signal; Based on the dual-mode control signal, the real-time operating status parameters of the system are extracted through the data collector to generate status monitoring data, and the communication interface completes the data electrical isolation transmission processing and outputs the system operating parameters.

2. The subway operation dual-mode control method based on pulse width modulation coding according to claim 1 is characterized in that: The method is based on the nominal power supply voltage value, determines the input terminal voltage detection value after defining the initialization voltage range through the power supply measurement module, implements dual-path calibration of the manual control and automatic operation channel voltages according to the dual redundant measurement module, and outputs the manual control detection voltage and the automatic operation detection voltage, including: Collect the original sampling data of the nominal power supply voltage value, calculate the voltage effective value through the sliding window algorithm, and obtain the voltage reference data; Based on the voltage reference data, fitting the voltage fluctuation curve by the least square method and extracting the fluctuation characteristic parameters to obtain the initialization voltage range; Comparing the initialization voltage range with a preset threshold, dynamically adjusting the calibration coefficient through an adaptive threshold algorithm, and generating an input terminal voltage detection value; A jitter compensation coefficient is introduced into the input terminal voltage detection value respectively, a dual-channel data filtering process is performed based on a Kalman filter algorithm, and a dual-channel calibration reference value is output; Based on the dual-channel calibration reference value, the signal noise is separated and the filtered signal is reconstructed by a wavelet transform algorithm to generate a channel voltage feature vector; The channel voltage feature vector is processed by time sequence superposition to extract the peak envelope, and a frequency domain analysis is performed using a Fourier transform algorithm to output the manual control detection voltage and the automatic operation detection voltage.

3. The subway operation dual-mode control method based on pulse width modulation coding according to claim 1 is characterized in that: The method of selecting the control signal according to the mode, calculating the output duty ratio segmentation parameter by the pulse width modulation encoder to obtain the pulse width modulation waveform, limiting the output current amplitude by the load resistor, and generating the current-limiting pulse width modulation signal includes: Decomposing the mode selection control signal into a frequency domain feature sequence by a fast Fourier transform algorithm, extracting carrier modulation parameters by double-side spectrum analysis, and generating duty cycle reference data; According to the duty cycle reference data, a switching timing is calculated by a vector modulation algorithm and a pulse sequence is constructed to generate segmented modulation parameters; The segmented modulation parameters are introduced into a space vector pulse width modulation algorithm, and waveform characteristics are reconstructed according to the high and low level timing sequence to obtain a pulse width modulation waveform; A dead time compensation algorithm is introduced into the pulse width modulation waveform to eliminate switching timing interference, and an overcurrent protection threshold is superimposed to output a standard modulation waveform; Taking the standard modulation waveform as input, an adaptive current limiting algorithm is used to dynamically calculate load parameters and construct a current limiting control vector; The current limiting control vector is closed-loop controlled through a proportional-integral regulation algorithm and the waveform distortion rate is optimized to generate the current limiting pulse width modulation signal.

4. The subway operation dual-mode control method based on pulse width modulation coding according to claim 1 is characterized in that: Based on the current-limiting pulse width modulation signal, the output short circuit and overload status are determined by multiple protectors and a protection trigger signal is output, a delay controller is started to adjust the output relay disconnection timing, and a protection execution signal is constructed, including: Importing the current-limiting pulse width modulation signal, identifying the short-circuit characteristic waveform and marking the fault time point through a sequential comparison algorithm, and obtaining short-circuit detection data; For the short-circuit detection data, the overload accumulation is calculated and the overload duration is calibrated by a double-threshold recursive algorithm to generate a fault state matrix; Input the fault state matrix into a pattern recognition algorithm, extract fault type characteristics and quantify fault level coefficients, and generate a protection trigger signal; According to the protection trigger signal, a delay strategy model is constructed using a neural network algorithm and an optimal disconnection timing is calculated, and a relay control instruction is output; Based on the relay control instructions, a fuzzy reasoning algorithm is used to evaluate the protection action priority and set the execution sequence to construct a protection action sequence; In combination with the protection action sequence, the execution timing is optimized and the coordinated control parameters are calculated through a time series analysis algorithm to generate the protection execution signal.

5. The subway operation dual-mode control method based on pulse width modulation coding according to claim 1 is characterized in that: The collecting of the protection execution signal, establishing a mode switching rule and generating a mode switching signal through a timing controller, calling a relay linkage controller to optimize the coordinated state of the output relay, and generating a dual-mode control signal, includes: Extracting the timing features in the protection execution signal, constructing a mode conversion map through a state transition probability algorithm, and generating a switching timing matrix; For the switching timing matrix, the optimal switching path between the modes is calculated by a dynamic programming algorithm to obtain a mode switching rule; Input the mode switching rule into the Bayesian network algorithm, evaluate the switching reliability and calibrate the switching priority, and generate a mode switching signal; Based on the mode switching signal, a relay action timing rule is established by using a response surface algorithm and a switching speed is optimized, and a relay state instruction is output; The relay state instructions are matched with linkage characteristics through a real-time sequence analysis algorithm to construct a collaborative control vector; The collaborative control vector is subjected to a multi-objective optimization algorithm to balance switching stability and response speed, thereby generating the dual-mode control signal.

6. The subway operation dual-mode control method based on pulse width modulation coding according to claim 1 is characterized in that: Based on the dual-mode control signal, the real-time operating state parameters of the system are extracted by the data collector to generate state monitoring data, and the communication interface completes the data electrical isolation transmission processing and outputs the system operating parameters, including: Importing the dual-mode control signal, extracting key state feature quantities and constructing feature space through principal component analysis algorithm, and generating state feature vectors; For the state characteristic vector, the real-time change trend of each parameter is calculated by a sliding time window algorithm to obtain operation trend data; Inputting the operation trend data into a Kalman filter algorithm to eliminate measurement noise and optimize data accuracy to generate condition monitoring data; According to the state monitoring data, compress the data stream with a differential coding algorithm and add a check bit, and output a transmission data packet; Based on the transmission data packet, a photoelectric isolation algorithm is used to establish a data transmission channel and ensure signal integrity, thereby constructing an isolated transmission sequence; In combination with the isolated transmission sequence, the system operation status is reconstructed and the data validity is verified through a data recovery algorithm to generate the system operation parameters.

7. A subway operation dual-mode control system based on pulse width modulation coding, used to implement the subway operation dual-mode control method based on pulse width modulation coding as described in any one of claims 1 to 6, characterized in that: The subway operation dual-mode control system based on pulse width modulation coding includes: The calibration module is used to determine the input voltage detection value based on the nominal power supply voltage value after the power supply measurement module defines the initialization voltage range, implement the dual-path calibration of the manual control and automatic operation channel voltages according to the dual redundant measurement module, and output the manual control detection voltage and the automatic operation detection voltage; The acquisition module is used to acquire the real-time current signals of the manual control channel and the automatic operation channel according to the manual control detection voltage and the automatic operation detection voltage by a signal collector and convert them into manual control signal sampling values ​​and automatic operation signal sampling values, complete the threshold detection analysis of abnormal signals through a signal comparator, and generate a mode selection control signal, including: inputting the manual control detection voltage and the automatic operation detection voltage into the signal collector, performing bandpass filtering on the original voltage signal through a Butterworth filtering algorithm to generate filtered voltage data; extracting the instantaneous amplitude and phase information of the signal from the filtered voltage data through a Hilbert transform algorithm to obtain Real-time current signal; based on the real-time current signal, multi-scale decomposition is performed and signal singular points are extracted through discrete wavelet transform algorithm to generate manual control signal sampling values ​​and automatic operation signal sampling values; the manual control signal sampling values ​​and the automatic operation signal sampling values ​​are used to construct an abnormality detection parameter matrix through an adaptive dual threshold algorithm, and an abnormal state identification is output; for the abnormal state identification, an abnormality level coefficient is calculated through a fuzzy clustering algorithm and an abnormality evaluation index is established to generate an abnormal feature vector; based on the abnormal feature vector, a dual-mode classifier is established through a support vector machine algorithm and the mode switching probability is calculated, and the mode selection control signal is output; A calculation module, used for selecting a control signal according to the mode, obtaining a pulse width modulation waveform after calculating the output duty ratio segmentation parameter through a pulse width modulation encoder, limiting the output current amplitude through a load resistor, and generating a current-limiting pulse width modulation signal; An output module is used to determine the output short circuit and overload status based on the current-limiting pulse width modulation signal through multiple protectors and output a protection trigger signal, start the delay controller to adjust the output relay disconnection timing, and construct a protection execution signal; A generation module is used to collect the protection execution signal, establish a mode switching rule and generate a mode switching signal through a timing controller, call a relay linkage controller to optimize the coordinated state of the output relay, and generate a dual-mode control signal; The transmission module is used to extract the real-time operating status parameters of the system through the data collector based on the dual-mode control signal to generate status monitoring data, complete the data electrical isolation transmission processing through the communication interface, and output the system operating parameters.

Images