Digital twinning-based urban rail permanent magnet motor field loss detection method and device

Abstract

The invention provides an urban rail permanent magnet motor field loss detection method and device based on digital twinning. The method comprises the steps that a digital twinning model representing the dynamic characteristics of an actual permanent magnet motor system is constructed in a vehicle-mounted traction controller; the method comprises the following steps: acquiring real-time operation data of a permanent magnet motor through a sensor of an existing urban rail traction system, and receiving a switch control signal generated by a vehicle-mounted traction controller; solving the digital twinning model by adopting a mode based on switch control signal event driving to obtain twinning output data; optimizing the digital twin model by adopting a particle swarm optimization strategy; and obtaining a current value of a permanent magnet flux linkage based on the optimized digital twinborn model, and sending out an excitation loss fault alarm based on the current value. According to the invention, the cost of a special sensor is saved, and the result is more highly credible.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of rail transit technology, and in particular to a city rail permanent magnet motor loss detection method and device based on digital twinning. BACKGROUND

[0002] Permanent magnet synchronous traction motor (PMSM), hereinafter referred to as "permanent magnet motor", has become the core equipment of the new generation of metro and light rail vehicles due to its high power density and high efficiency. However, the core component of the permanent magnet motor, the permanent magnet, is prone to irreversible demagnetization failure under long-term high load, high temperature and vibration conditions. According to industry statistics, permanent magnet demagnetization failure is the second largest failure of permanent magnet synchronous motor after bearing wear, accounting for more than 20%. After the permanent magnet demagnetization, it may cause the stator current to increase, the torque output to decrease, and even cause a chain secondary failure of the traction system. Therefore, a demagnetization detection technology with high real-time and high accuracy is crucial for the safe operation of the city rail.

[0003] Currently, the mainstream demagnetization detection methods of permanent magnet motors in the industry can be summarized into the following three categories:

[0004] I. Direct detection method based on physical sensor.

[0005] This method is to install a Hall sensor on the stator tooth or the surface of the permanent magnet to realize real-time measurement of the air gap magnetic field strength. The measurement results can be transmitted to the vehicle-mounted traction control unit through CAN bus or Ethernet, and the threshold value judgment or pattern recognition algorithm is used for fault discrimination. This method uses physical sensors to directly measure the change of magnetic flux, is sensitive to early demagnetization, supports real-time monitoring, and has fast response speed.

[0006] II. Indirect detection method based on current harmonic analysis.

[0007] This method analyzes the three-phase current values collected by the current sensor installed in the stator circuit. Fast Fourier transform or Park vector analysis method is used to extract current harmonic characteristic values, and the results are matched with the database to determine whether there is a demagnetization failure.

[0008] III. Identification detection method based on offline model parameters.

[0009] By constructing a motor model, the existing signals are used for parameter estimation, and recursive least squares or Kalman filter is used for online identification of key parameters. When the deviation of the key parameters from the actual parameters is greater than the set value, an alarm is triggered. This method does not rely on physical sensors and has low detection cost.

[0010] Although the above methods have been partially applied in the field of urban rail transit, there are still the following significant defects:

[0011] 1. Multi-physical field coupling interference is not decoupled: The existing methods do not establish an electromagnetic-thermal-mechanical multi-physical field correlation model, which may lead to misjudgment of loss of excitation. For example, the rotor eccentricity caused by bearing wear will change the dq-axis inductance, and its current harmonic characteristics are highly similar to the loss of excitation of the permanent magnet. The traditional SVM classifier may misjudge the compound fault. Abnormalities in the cooling system may also cause the permanent magnet temperature to rise and accelerate the loss of excitation. The current temperature compensation model only corrects linearly and cannot describe the nonlinear demagnetization curve of the permanent magnet material at high temperatures.

[0012] 2. Lack of prediction ability: The existing loss of excitation detection techniques are all post-fault diagnosis (i.e., alarm after the permanent magnet has experienced measurable degradation), and lack the ability to predict the degradation trend. The permanent magnet flux has shown a slow decline six months before failure, but the traditional threshold method is difficult to provide early warning for this failure.

[0013] 3. Conflict between computing resources and real-time performance: High-precision models require a large amount of computing power and are difficult to run in real time on on-board edge devices. Simplified models can improve speed, but will result in a decrease in the accuracy of key parameter identification, which cannot meet the high reliability requirements of urban rail transit.

[0014] In view of the different drawbacks of the current loss of excitation detection methods for urban rail permanent magnet synchronous motors, it is of great significance to develop a loss of excitation detection method for urban rail permanent magnet synchronous motors with high accuracy and low cost to improve the efficiency, accuracy, and predictability of loss of excitation detection for urban rail permanent magnet synchronous motors. SUMMARY

[0015] Therefore, the present application provides a loss of excitation detection method and device for urban rail permanent magnet motors based on digital twinning to solve at least one of the above problems.

[0016] To achieve the above purpose, the present application adopts the following scheme:

[0017] According to a first aspect of the present application, a digital-twin-based loss-of-field detection method for a metro permanent magnet motor is provided, the method comprising: constructing a digital twin model that reproduces the dynamic characteristics of an actual permanent magnet motor system in a vehicle-mounted traction controller; obtaining real-time operation data of the permanent magnet motor through sensors of an existing metro traction system and receiving switch control signals generated by the vehicle-mounted traction controller; solving the digital twin model in an event-driven manner based on the switch control signals to obtain twin output data; constructing an evaluation function based on errors between the real-time operation data and the twin output data, and optimizing the digital twin model using a particle swarm optimization strategy; obtaining a current value of the permanent magnet flux based on the optimized digital twin model, and comparing the current value with a preset calibration value, and issuing a loss-of-field fault alarm when the comparison result exceeds a set threshold.

[0018] As an embodiment of the present application, the digital twin model in the above method is constructed based on the mathematical models of the voltage, electromagnetic torque, motion equation of the permanent magnet synchronous motor in the dq two-phase rotating coordinate system, and the topology of the traction inverter circuit.

[0019] As an embodiment of the present application, the method of solving the digital twin model in an event-driven manner based on the switch control signals comprises: defining the time interval between two consecutive inverter switching transformation events based on the switch control signals as the calculation step of the digital twin model, and triggering the solving calculation of the digital twin model only at the event point where the switching state changes.

[0020] As an embodiment of the present application, the method of optimizing the digital twin model using a particle swarm optimization strategy comprises: independently performing multiple complete particle swarm optimization processes, and each optimization using a randomly initialized particle swarm; screening the optimal parameter combination obtained by each optimization by setting a fitness threshold, and eliminating results that do not achieve the preset accuracy; and statistically averaging all effective optimization results that pass the screening, and taking the average value as the final optimization parameter.

[0021] As an embodiment of the present application, the optimization parameters in the above method include: flux value, stator resistance, stator inductance, rotor resistance and rotor inductance.

[0022] As an embodiment of the present application, the method of constructing an evaluation function based on the errors between the real-time operation data and the twin output data comprises: constructing a first evaluation function based on the phase current output by the digital twin model and the actual phase current collected by the permanent magnet motor through the root mean square error; and constructing a second evaluation function based on the DC side current output by the digital twin model and the actual DC side current collected by the permanent magnet motor through the root mean square error.

[0023] According to a second aspect of the present application, a digital-twin-based city rail permanent magnet motor loss of excitation detection device is provided, the device comprising: a twin model construction unit for constructing a digital twin model reproducing the dynamic characteristics of an actual permanent magnet motor system in a vehicle-mounted traction controller; a data acquisition unit for acquiring real-time operation data of the permanent magnet motor through sensors of an existing city rail traction system and receiving switch control signals generated by the vehicle-mounted traction controller; a twin model solving unit for solving the digital twin model in an event-driven manner based on the switch control signals to obtain twin output data; a model optimization unit for constructing an evaluation function based on the error between the real-time operation data and the twin output data and optimizing the digital twin model using a particle swarm optimization strategy; and a loss of excitation detection unit for obtaining the current value of the permanent magnet flux based on the optimized digital twin model and comparing the current value with a preset calibration value, and issuing a loss of excitation fault alarm when the comparison result exceeds a set threshold.

[0024] As an embodiment of the present application, the digital twin model is constructed based on the mathematical models of the voltage, electromagnetic torque, motion equation of the permanent magnet synchronous motor in the dq two-phase rotating coordinate system, and the topology of the traction inverter circuit.

[0025] As an embodiment of the present application, the twin model solving unit is specifically configured to define the time interval between two consecutive inverter switching transformation events based on the switch control signals as the calculation step of the digital twin model, and trigger the solving calculation of the digital twin model only at the event point where the switching state changes.

[0026] As an embodiment of the present application, the model optimization unit comprises: a particle swarm optimization module for independently performing multiple complete particle swarm optimization processes, and each optimization using a randomly initialized particle swarm; a result elimination module for screening the optimal parameter combination obtained by each optimization by setting a fitness threshold, and eliminating results that do not reach the preset accuracy; and a statistical average module for statistically averaging all effective optimization results that pass the screening, and taking the average value as the final optimization parameter.

[0027] As an embodiment of the present application, the optimization parameters include flux value, stator resistance, stator inductance, rotor resistance and rotor inductance.

[0028] As an embodiment of the present application, the model optimization unit further comprises: a first evaluation function construction module, configured to construct a first evaluation function based on the phase current output by the digital twin model and the phase current actually collected by the permanent magnet motor through a root mean square error; and a second evaluation function construction module, configured to construct a second evaluation function based on the DC side current output by the digital twin model and the DC side current actually collected by the permanent magnet motor through a root mean square error.

[0029] According to a third aspect of the present application, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0030] According to a fourth aspect of the present application, a computer readable storage medium is provided, having a computer program stored thereon, wherein the computer program is executable by a processor to implement the steps of the above method.

[0031] According to a fifth aspect of the present application, a computer program product is provided, comprising computer programs / instructions, wherein the computer programs / instructions are executable by a processor to implement the steps of the above method.

[0032] From the above technical solution, it can be seen that the city rail permanent magnet motor loss detection method and device based on digital twinning provided by the present application, compared with the existing direct detection method of adding a special physical sensor, since the data of the present application comes from the existing city rail traction system sensor and the control signal of the vehicle-mounted traction controller, the cost is saved. Compared with the indirect detection method based on current harmonic analysis, the present application uses the real-time running data and the improved particle swarm algorithm to continuously optimize the model parameters through the closed-loop mechanism of "perception-optimization-verification-update", and performs multiple tests through the evaluation function to ensure the high credibility of the parameter identification result. Compared with the identification detection method based on offline model parameters, the present application makes the digital twin model always maintain high consistency with the physical system, and the result is also more reliable. BRIEF DESCRIPTION OF DRAWINGS

[0033] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed in the embodiment or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor. In the drawings:

[0034] Figure 1 is a flowchart of a city rail permanent magnet motor loss detection method based on digital twinning provided by an embodiment of the present application;

[0035] Figure 2 is a comparison schematic diagram of a conventional time discretization and a switching state driving method provided by an embodiment of the application;

[0036] Figure 3 is a circuit topology diagram of a traction inverter provided by an embodiment of the application;

[0037] Figure 4 is a flowchart of constructing an evaluation function provided by an embodiment of the application;

[0038] Figure 5 is a flowchart of optimizing a digital twin model provided by an embodiment of the application;

[0039] Figure 6 is a comparison diagram of a digital twin model and a measured phase current waveform under a traction stage 1790r / min condition provided by an embodiment of the application;

[0040] Figure 7 is a comparison diagram of a digital twin model and a measured phase current waveform under a traction stage 1943r / min condition provided by an embodiment of the application;

[0041] Figure 8 is a comparison diagram of a digital twin model and a measured phase current waveform under a braking stage 800r / min condition provided by an embodiment of the application;

[0042] Figure 9 is a comparison diagram of a digital twin model and a measured phase current waveform under a braking stage 740r / min condition provided by an embodiment of the application;

[0043] Figure 10 is a structure schematic diagram of a digital twin-based urban rail permanent magnet motor loss detection device provided by an embodiment of the application;

[0044] Figure 11 is a structure schematic diagram of a model optimization unit provided by an embodiment of the application;

[0045] Figure 12 is a structure schematic diagram of a model optimization unit provided by another embodiment of the application;

[0046] Figure 13 is a system structure schematic block diagram of an electronic device provided by an embodiment of the application. DETAILED DESCRIPTION

[0047] To make the objectives, technical solutions and advantages of the embodiments of the application clearer, the embodiments of the application are further described in detail below with reference to the drawings. Here, the schematic embodiments of the application and the descriptions thereof are used to explain the application, but not to limit the application.

[0048] As Figure 1Fig. 1 shows a flowchart of a method for detecting loss of excitation of a metro permanent magnet motor based on digital twinning provided by an embodiment of the present application. The method comprises the following steps:

[0049] Step S101: A digital twinning model reproducing the dynamic characteristics of an actual permanent magnet motor system is constructed in a vehicle-mounted traction controller.

[0050] Digital twinning is defined as a virtual digital representation of an actual physical product and reflects the entire life cycle process of the corresponding physical entity product, also known as digital mapping or digital mirror image. The core of the digital twinning model is to construct high-precision digital models, which can be mechanism models based on first-principle physical equations, data-driven models trained based on machine learning and artificial intelligence techniques, or an organic combination of the two.

[0051] In the present embodiment, the constructed digital twinning model is configured in the vehicle-mounted traction controller. The digital twinning model is deployed in the vehicle-mounted traction controller to achieve the most direct and fastest data interaction between the model and the physical entity. The vehicle-mounted traction controller is the brain of the entire traction system, which itself generates switching control signals for driving the motor and is directly connected to sensors that collect data such as motor three-phase current, voltage, and speed. Deploying the digital twinning model here means that the model can obtain the most original and core real-time running data and control signals with zero delay, avoiding the delay and data distortion problems caused by long-distance transmission over the network. In addition, the ultimate goal of loss of excitation detection is to guide the control system to respond, such as issuing an alarm or adjusting the operation strategy. The model is deployed in the vehicle-mounted traction controller, and the identified parameters (such as flux linkage values) can be directly and quickly fed back to the decision-making module of the control system, forming a real-time closed loop of "perception-computation-decision-execution".

[0052] Step S102: Obtain real-time running data of the permanent magnet motor through sensors of an existing metro traction system, and receive switching control signals generated by the vehicle-mounted traction controller.

[0053] In order for the digital twinning model to be synchronized with the physical entity, two types of key information need to be input in real time in this step: real-time running data and switching control signals. Among them, the real-time running data is collected by various sensors already existing on the traction system, which collect the running state data of the physical motor in real time. These data mainly include the three-phase current of the motor, the DC bus voltage, the bus current, and the motor speed, etc. The switching control signals are obtained from the vehicle-mounted traction controller to obtain the real-time modulation signals and the corresponding switching control signals for driving the inverter power devices. These signals directly determine the input voltage and working state of the motor. These two types of data can be transmitted to the digital twinning model through the high-speed communication interface inside the vehicle-mounted traction controller.

[0054] Step S103: Solve the digital twin model in an event-driven manner based on the switch control signal to obtain twin output data.

[0055] The purpose of this step is to solve the contradiction between high-precision model calculation and high real-time requirement of the vehicle-mounted system, and to improve the calculation efficiency of the model. Traditional digital twin model solving mostly uses fixed-step time discretization method, that is, whether the system state changes or not, a small and fixed time interval (h) is used for calculation, which leads to a large amount of redundant calculation in the period when the switch state does not change. The event-driven method based on the switch control signal is used in this application. Compared with the traditional time-discrete driving method, as shown in the following table. Figure 2 It no longer uses a fixed time step, but defines the time interval between two consecutive inverter switch transformation events based on the switch control signal as the calculation step of the digital twin model, and only triggers the solving calculation of the digital twin model at the event point where the switch state changes. Since this step is the theoretical maximum value, it greatly avoids unnecessary calculation, significantly improves the solving efficiency of the model, and solves the most significant efficiency-accuracy contradiction of the model parameter-based loss-of-field detection method.

[0056] After the model is solved, a series of twin data will be output, such as the twin three-phase current and the DC side current.

[0057] Step S104: Construct an evaluation function based on the error between the real-time running data and the twin output data, and optimize the digital twin model using a particle swarm optimization strategy.

[0058] This step is to realize the self-learning and self-adaptation of the digital twin model, and to ensure that the model can accurately reflect the real state of the physical motor.

[0059] First, define the evaluation function (also known as the objective function or error function) to quantify the difference between the twin data and the real data. The smaller the error value, the closer the model is to the actual system. In order to minimize the error, the scheme uses an improved particle swarm optimization algorithm (PSO) to iteratively optimize the key parameters of the model.

[0060] Step S105: Based on the optimized digital twin model, obtain the current value of the permanent magnet flux linkage, and compare it with the preset calibration value. When the comparison result exceeds the set threshold, issue a loss-of-field fault alarm.

[0061] When the digital twin model is fully optimized, its output is highly consistent with the physical entity, and it is upgraded to a reliable calibration model. Based on this optimized high-precision digital twin model, the current value of the permanent magnet flux linkage can be continuously and real-timely identified. This value accurately reflects the real magnetic field strength of the physical motor. The control system compares the real-timely identified current value of the flux linkage with a preset calibration value (usually the flux linkage value of the motor in good condition). When the deviation between the two exceeds the preset threshold (for example, 10%), the system determines that the motor has a demagnetization fault, and actively sends a fault alarm by the vehicle-mounted traction controller.

[0062] From the above technical solutions, the city rail permanent magnet motor demagnetization detection method based on digital twin provided by the application saves cost compared with the existing direct detection method of adding special physical sensors, because the data of the application comes from the existing city rail traction system sensors and the control signals of the vehicle-mounted traction controller. Compared with the indirect detection method based on current harmonic analysis, the application uses the "perception-optimization-verification-update" closed-loop mechanism, continuously optimizes the model parameters using real-time running data and improved particle swarm algorithm, and performs multiple tests through the evaluation function to ensure the high credibility of the parameter identification results. Compared with the identification detection method based on offline model parameters, the application makes the digital twin model always maintain high consistency with the physical system, and the result is also more reliable.

[0063] In an embodiment of the application, the above-mentioned digital twin model is constructed based on the mathematical model of the voltage, electromagnetic torque, motion equation of the permanent magnet motor in the dq two-phase rotating coordinate system, and the circuit topology of the traction inverter.

[0064] As Figure 3 shown is the circuit topology diagram of the traction inverter provided by the embodiment, based on Kirchhoff's law, the mathematical model of the traction inverter can be expressed as formula (1) below:

[0065]

[0066] In the formula, U ab ,U bc ,U ca is the three-phase line voltage, u ao ,u bo ,u co is the three-phase phase voltage, i dc is the input current of the traction inverter, i a ,i b ,i c is the output current of the traction inverter.

[0067] The traction inverter is connected with the permanent magnet motor, and the mathematical model of the permanent magnet motor in the ABC three-phase static coordinate system is converted into the dq two-phase rotating coordinate system in the modern motor control theory to realize the decoupling and linearization of the system. The voltage equation of the permanent magnet motor in the dq coordinate system can be expressed as formula (2) below:

[0068]

[0069] In the formula, u d , u q , i d , i q are the stator voltage and current in the dq coordinate system respectively; R s is the stator resistance; ωe is the rotor electric angular velocity; L d , L q are the inductances of the stator direct axis and cross axis respectively, and ψ f is the permanent magnet flux linkage.

[0070] The electromagnetic torque equation is formula (3) below:

[0071]

[0072] In the formula, T e is the electromagnetic torque; n p is the motor pole pair number.

[0073] The motion equation can be expressed as formula (4) below:

[0074]

[0075] In the formula, ω m is the mechanical angular velocity, ω m = ω e / n p ; T L is the load torque; and J is the moment of inertia.

[0076] Finally, the mathematical model of the permanent magnet traction motor is formula (5) below:

[0077]

[0078] In the formula,

[0079]

[0080] In practice, the Euler method is often used to discretize the permanent magnet motor model, and the discretized result can be expressed as:

[0081] i dq (k+1)=F o i dq (k)+G o udq (k)+g o ψ f (6)

[0082] F o =I+F e T s ,g o =I+F e T s ,

[0083] The two parts of the above-mentioned mathematical model based on physical mechanism jointly constitute the core of the digital twin model, which can reproduce the dynamic characteristics of the entire permanent magnet traction transmission system in principle.

[0084] In another embodiment of the present application, as shown in Figure 4 the step S104 of constructing an evaluation function based on the error between the real-time running data and the twin output data includes:

[0085] Step S401: based on the phase current output by the digital twin model and the phase current actually collected by the permanent magnet motor, a first evaluation function is constructed by the root mean square error.

[0086] Specifically, the first evaluation function f1 here is shown in the following formula (7):

[0087]

[0088] i a,i is the i-th data point of the A-phase current in the digital twin model, i a,real,i is the i-th data point of the A-phase current actually collected by the motor, i b,i is the i-th data point of the B-phase current in the digital twin model, i b,real,i is the i-th data point of the B-phase current actually collected by the motor, and N is the total number of sampling points used.

[0089] Step S402: based on the DC side current output by the digital twin model and the DC side current actually collected by the permanent magnet motor, a second evaluation function is constructed by the root mean square error.

[0090] Specifically, the first evaluation function f2 here is shown in the following formula (8):

[0091]

[0092] I dc,i is the i-th data point of the DC side current in the digital twin model, and I dc,real,i is the i-th data point of the DC side current actually collected by the motor.

[0093] These two evaluation functions together constitute the objective that the particle swarm optimization algorithm needs to minimize. By continuously reducing these two errors, the parameters of the digital twin model (such as flux linkage, resistance, inductance) can be driven to converge to the true values, thereby improving the accuracy of the model.

[0094] In another embodiment of the present application, as shown in Figure 5 The step S104 of optimizing the digital twin model using the particle swarm optimization strategy includes:

[0095] Step S1041: Perform multiple complete particle swarm optimization processes independently, and each time use a randomly initialized particle swarm for optimization.

[0096] The purpose of this step is to enhance the global search capability through repetition and randomization. First, perform N independent particle swarm optimization algorithm processes completely, where N is preferably greater than or equal to 10, and each time is a start-to-finish optimization process. At the beginning of each independent optimization, a completely new particle swarm with randomly initialized positions and velocities is generated. This is done to ensure that the starting point and path of each search are different, allowing the particle swarm optimization algorithm to explore a wider area of the solution space, thereby greatly improving the probability of finding the global optimal solution and avoiding falling into the trap of a local optimal solution due to a poor initial position.

[0097] Step S1042: For the optimal parameter combination obtained in each optimization, set a fitness threshold to filter out results that do not meet the preset accuracy.

[0098] This step is to ensure that the adopted optimization results meet the preset accuracy requirements. After each independent optimization process ends, the particle swarm optimization algorithm obtains an optimal parameter combination for the current round and the corresponding fitness value (i.e., the error size calculated by the evaluation function). A fitness threshold can be set in advance, which represents the maximum acceptable error between the model output and the actual data, i.e., the minimum accuracy requirement for the identification result. Compare the best fitness value obtained in each optimization round with the threshold. Only those results with a fitness value better than the preset threshold are considered valid and high-quality optimization results and are retained; otherwise, those results that do not meet the accuracy requirements are directly eliminated.

[0099] Step S1043: Statistically average all valid optimization results that pass the screening, and take the average value as the final optimization parameter.

[0100] This step is to improve the robustness and accuracy of the final parameters by fusing multiple high-quality solutions. All valid optimal parameter combinations screened out in step S1042 are collected, and statistical average calculation is performed on each parameter respectively. The calculated average value will be output as the final optimal parameter combination and used to update the digital twin model. The advantage of this step is that it uses the principle of the law of large numbers: by averaging multiple independent and high-quality recognition results, random errors caused by measurement noise and other factors in a single recognition can be effectively offset. This makes the final result not only closer to the true value, but also more stable and reliable than any single optimization result, showing stronger working condition adaptability and robustness.

[0101] In this embodiment, the above-mentioned optimization parameters include but are not limited to: flux linkage value, stator resistance, stator inductance, rotor resistance and rotor inductance.

[0102] The above-optimized digital twin model is deployed in the vehicle-mounted traction controller. The vehicle-mounted traction controller is responsible for controlling the switching pulse of the inverter, while running the digital twin model. The digital twin model simulates through analog quantities collected by the vehicle-mounted traction controller and modulation signals in the control system, and outputs the flux linkage value of the permanent magnet motor in real time. The calculated flux linkage value is sent back to the vehicle-mounted traction controller, which calculates the deviation between the current flux linkage value and the set value. When the difference exceeds the set threshold, the vehicle-mounted traction controller can actively report a fault.

[0103] In simple terms, the digital twin model can output the flux linkage value because it has undergone a "hypothesis-verification-correction" closed-loop process. The flux linkage value is both an internal parameter that defines the state of the model and an output result that is continuously identified and updated through optimization algorithms.

[0104] This can be understood as a cycle of the following three stages:

[0105] Hypothesis stage: When the digital twin model starts calculating, it will perform calculations based on a current, hypothetical flux linkage value (as well as other parameters such as resistance, inductance, etc.). This initial flux linkage value can be a preset value or the result of the previous round of calculation optimization. The model uses this hypothetical flux linkage value as a basis to calculate the theoretical twin output data (such as twin three-phase current) in combination with real-time switching control signals.

[0106] Verification stage: The system compares the twin output data calculated by the digital twin model with the data actually measured from the sensor. The difference between the two is quantified by a pre-defined evaluation function. If the error is small, it means that the hypothetical flux linkage value is accurate; if the error is large, it means that the hypothetical flux linkage value does not conform to the current real physical state of the motor.

[0107] Correction stage: When the error is detected, the particle swarm optimization algorithm (PSO) is started. The core task of this algorithm is to adjust the parameters inside the digital twin model (the most critical is the flux linkage value), and the adjustment target is to minimize the error calculated by the evaluation function. Through the iterative optimization of the algorithm, a flux linkage value that can best match the model output current and the actual current is found. This found flux linkage value is output as the final result of this round of identification, and is also updated as the new current value for the next round of model calculation.

[0108] Therefore, the optimized parameter is a variable inside the model that describes the physical characteristics of the model, and the output flux linkage value is the most accurate estimate of the variable after the "hypothesis-verification-correction" closed-loop identification process. This output value accurately reflects the real-time flux linkage state of the physical motor, so it can be used to compare with the set value to determine whether a demagnetization failure has occurred.

[0109] In an embodiment of the present application, the effect of the above method is verified by a set of actual experimental data: based on the data collected by the actual vehicle of the urban rail train, the digital twin model and parameter identification verification are carried out, and four groups of actual vehicle data under different operating conditions are collected, in which the sampling frequency of voltage, current, speed and angle is 4 kHz, and the sampling frequency of the switching signal is 64 kHz.

[0110] Based on the flux linkage ψ (Wb) identification results of the four groups of data, the principle of taking the average value of 10 times identification is adopted. From the identification results, it can be seen that the identification error of the flux linkage is within 7%.

[0111] Table 1

[0112]

[0113] At the same time, in order to verify the simulation ability of the digital twin model to the actual traction system operating data, as shown in Figure 6 to Figure 9 The comparison chart of the digital twin model and the measured phase current waveform is shown, which corresponds to the four operating conditions in Table 1, and Figure 6 to Figure 9 It can be seen that the phase current of the digital twin model is basically consistent with the measured phase current. However, since the sampling rate of the measured data is much lower than that of the simulation data, the accuracy of the digital twin model will decrease, so the fitting degree of the waveform is lower than that of the simulation data. However, it can still basically realize the simulation of the operating characteristics of the actual traction system. The verification of this simulation ability reflects the credibility of the model flux linkage identification value from the side, and proves that the method can detect the demagnetization problem of the permanent magnet synchronous motor in real time and effectively.

[0114] From the above, the application has the following advantages compared to the indirect detection method based on current harmonic analysis: by continuously collecting real-time operating data such as voltage and current, an improved particle swarm optimization algorithm is used for parameter identification, and time-varying parameters such as stator resistance and permanent magnet flux are tracked. The evaluation function running in parallel with the digital twin model verifies the rationality of the parameters, and the result of multiple tests will trigger the model parameter identification update. This closed-loop mechanism of "perception-optimization-verification-update" greatly improves the credibility of the identification result compared to the traditional model parameter-based loss-of-field detection method. The digital twin model reflects the multi-physical field coupling of the actual system through data fusion, eliminating the need for decoupling analysis of multiple physical fields, improving the efficiency and accuracy of identification.

[0115] Compared with the existing direct detection method of adding special physical sensors, the application has the following advantages: the data source of the digital twin-based urban rail permanent magnet motor loss-of-field detection method is the existing urban rail traction system sensors and controller control signals. Compared with the direct measurement method of adding physical sensors, it saves sensor cost, and also avoids errors and interference caused by direct measurement.

[0116] Compared with the existing offline model parameter identification detection method, the application has the following advantages: offline parameter identification method usually needs to collect experimental data under specific working conditions during shutdown, and obtains fixed parameters through optimization. This snapshot identification result is difficult to reflect the dynamic characteristic changes in the actual operation of the system. The application can capture the time-varying characteristics of parameters caused by temperature fluctuations and equipment aging through continuous learning of operating data, so that the digital twin model always maintains high consistency with the physical system. In terms of engineering application, offline identification is mainly used for system design and regular maintenance, while real-time system converts parameter identification into a continuous health monitoring process, upgrading the operation and maintenance mode from regular maintenance to predictive maintenance.

[0117] As Figure 10 The structure of a digital twin-based urban rail permanent magnet motor loss-of-field detection device provided by the embodiment of the application is shown in the figure. The device includes a twin model construction unit 1100, a data acquisition unit 1200, a twin model solving unit 1300, a model optimization unit 1400, and a loss-of-field detection unit 1500, which are connected in sequence. Among them:

[0118] The twin model construction unit 1100 is used to construct a digital twin model that reproduces the dynamic characteristics of the actual permanent magnet motor system in the on-board traction controller.

[0119] The data acquisition unit 1200 is used to acquire real-time operating data of the permanent magnet motor through the sensors of the existing urban rail traction system, and receives the switch control signals generated by the on-board traction controller.

[0120] The twin model solving unit 1300 is configured to solve the digital twin model in a manner driven by the switching control signal events to obtain twin output data.

[0121] The model optimization unit 1400 is configured to construct an evaluation function based on errors between the real-time operation data and the twin output data, and optimize the digital twin model by using a particle swarm optimization strategy.

[0122] The loss-of-field detection unit 1500 is configured to obtain a current value of the permanent magnet flux linkage based on the optimized digital twin model, and compare the current value with a preset calibration value, and issue a loss-of-field fault alarm when a comparison result exceeds a set threshold.

[0123] In an embodiment of the present application, the digital twin model is constructed based on a mathematical model of the voltage, electromagnetic torque, motion equation of the permanent magnet synchronous motor in the dq two-phase rotating coordinate system, and a topology of the traction inverter circuit.

[0124] In an embodiment of the present application, the twin model solving unit 1300 is specifically configured to define a time interval between two continuous inverter switching transformation events as a calculation step of the digital twin model based on the switching control signal, and trigger the solving calculation of the digital twin model only at the event point where the switching state changes.

[0125] In an embodiment of the present application, as shown in Figure 11 The model optimization unit 1400 includes:

[0126] The particle swarm optimization module 1410 is configured to independently execute a plurality of complete particle swarm optimization processes, and each time optimization uses a randomly initialized particle swarm.

[0127] The result elimination module 1420 is configured to filter and eliminate results that do not reach a preset precision by setting a fitness threshold for the optimal parameter combination obtained by each time optimization.

[0128] The statistical average module 1430 is configured to statistically average all effective optimization results that pass the filtering, and take the average value as the final optimization parameter.

[0129] In an embodiment of the present application, the optimization parameter includes a flux linkage value, a stator resistance, a stator inductance, a rotor resistance, and a rotor inductance.

[0130] In an embodiment of the present application, as shown in Figure 12 The model optimization unit 1400 further includes:

[0131] The first evaluation function construction module 1440 is configured to construct a first evaluation function by using a root mean square error based on the phase current output by the digital twin model and the phase current actually collected by the permanent magnet motor.

[0132] The second evaluation function construction module 1450 is configured to construct a second evaluation function by using a root mean square error based on the DC side current output by the digital twin model and the DC side current actually collected by the permanent magnet motor.

[0133] According to the technical solution, the digital twin-based city rail permanent magnet motor loss detection device saves costs compared with the existing direct detection method of adding a special physical sensor, because the data of the device is derived from the existing city rail traction system sensor and the control signal of the vehicle-mounted traction controller. Compared with the indirect detection method based on current harmonic analysis, the device uses the real-time operation data and the improved particle swarm algorithm to continuously optimize the model parameters through the closed-loop mechanism of "perception-optimization-verification-update", and performs multiple tests through the evaluation function to ensure the high credibility of the parameter identification result. Compared with the identification detection method based on offline model parameters, the device ensures that the digital twin model always maintains high consistency with the physical system, and the result is more reliable.

[0134] The electronic device provided in the embodiments of the present application also includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the above method when executing the program.

[0135] The embodiments of the present application also provide a computer readable storage medium, which stores a computer program for executing the above method.

[0136] The embodiments of the present application also provide a computer program product, which includes computer programs / instructions, and the computer programs / instructions implement the steps of the above method when executed by a processor.

[0137] As Figure 13 The electronic device 600 can also include a communication module 110, an input unit 120, an audio processor 130, a display 160, and a power supply 170. It should be noted that the electronic device 600 does not necessarily include all the components shown in the figure; in addition, the electronic device 600 can also include components not shown in the figure. Figure 13 The electronic device 600 can also include components not shown in the figure. Figure 13 The electronic device 600 can also include components not shown in the figure.

[0138] As Figure 13 The central processor 100, also known as a controller or operation control, can include a microprocessor or other processor device and / or logic device, which receives input and controls the operation of each component of the electronic device 600.

[0139] The memory 140, for example, can be one or more of a buffer, a flash memory, a hard drive, a removable media, a volatile memory, a non-volatile memory, or other suitable device. The above-mentioned information related to failure can be stored, and in addition, a program for executing the information related to failure can be stored. The central processing unit 100 can execute the program stored in the memory 140 to implement information storage or processing, etc.

[0140] The input unit 120 provides input to the central processing unit 100. The input unit 120 is, for example, a key or a touch input device. The power supply 170 is used to supply power to the electronic device 600. The display 160 is used to display a display object such as an image and a text. The display can be, for example, an LCD display, but is not limited thereto.

[0141] The memory 140 can be a solid state memory such as a read only memory (ROM), a random access memory (RAM), a SIM card, etc. It can also be a memory that saves information even when power is off, can be selectively erased, and is provided with more data, and an example of such a memory is sometimes referred to as an EPROM, etc. The memory 140 can also be some other type of device. The memory 140 includes a buffer memory 141 (sometimes referred to as a buffer). The memory 140 can include an application / function storage 142 for storing application and function programs or a flow for executing the operation of the electronic device 600 by the central processing unit 100.

[0142] The memory 140 can further include a data storage 143 for storing data such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. A driver storage 144 of the memory 140 can include various drivers of the electronic device for a communication function and / or for performing other functions of the electronic device such as a messaging application, a phonebook application, etc.

[0143] The communication module 110 is a transmitter / receiver that transmits and receives signals via the antenna 111. The communication module 110 (transmitter / receiver) is coupled to the central processing unit 100 to provide an input signal and receive an output signal, which can be the same as in the case of a conventional mobile communication terminal.

[0144] Based on different communication technologies, multiple communication modules 110, such as a cellular network module, a Bluetooth module, and / or a wireless local area network module, can be provided in the same electronic device. The communication modules 110 (transmitters / receivers) are also coupled to a speaker 131 and a microphone 132 via an audio processor 130 to provide audio output via the speaker 131 and to receive audio input from the microphone 132 to enable typical telecommunication functions. The audio processor 130 can include any suitable buffers, decoders, amplifiers, etc. In addition, the audio processor 130 is coupled to the central processor 100 to enable recording on-board via the microphone 132 and to enable playing on-board stored sounds via the speaker 131.

[0145] Those skilled in the art will appreciate that embodiments of the present application can be readily used as a method, apparatus, or computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, magnetic disks, CD-ROMs, optical storage media such as DVD s, etc.) embodying computer readable program code.

[0146] The present application is described in reference to the flowchart illustrations and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processing system or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart illustrations and / or block diagrams block or blocks. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams block or blocks. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams block or blocks.

[0147] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the flowchart illustrations and / or block diagrams block or blocks. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams block or blocks. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams block or blocks.

[0148] These computer program instructions can also be loaded into a computer or other programmable data processing devices, so that a series of operational steps are generated to realize the computer-implemented processes in the computer or other programmable devices, and the instructions executed in the computer or other programmable devices provide the steps for realizing the functions specified in the flowchart Figure 1 one flow or multiple flows and / or the functions specified in the block Figure 1 Figure 1 one block or multiple blocks.

[0149] The principles and implementation manners of the present application are described in the specific embodiments in the present application, and the above embodiment descriptions are only used to help understand the method of the present application and its core idea; meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation manners and application ranges will have changes, and the above descriptions should not be understood as limitations on the present application.

Claims

1. A digital-twin-based detection method for loss of excitation of a metro permanent magnet motor, characterized in that, The method comprises: building a digital twin model reproducing dynamic characteristics of an actual permanent magnet motor system in a vehicle-mounted traction controller; acquiring real-time operation data of the permanent magnet motor through sensors of an existing urban rail traction system and receiving a switching control signal generated by the vehicle-mounted traction controller; solving the digital twin model in an event-driven manner based on the switching control signal to obtain twin output data; building an evaluation function based on errors between the real-time operation data and the twin output data and optimizing the digital twin model using a particle swarm optimization strategy; comparing a current value of a permanent magnet flux obtained based on the optimized digital twin model with a preset calibration value and issuing a demagnetization fault alarm when a comparison result exceeds a set threshold.

2. The digital-twin-based detection method of the loss of excitation of the metro permanent magnet motor according to claim 1, characterized in that, The digital twin model is built based on mathematical models of voltage, electromagnetic torque, motion equation of the permanent magnet synchronous motor in a dq two-phase rotating coordinate system and a traction inverter circuit topology.

3. The digital-twin-based detection method of the loss of excitation of the metro permanent magnet motor according to claim 1, characterized in that, The solving of the digital twin model in the event-driven manner based on the switching control signal comprises defining a time interval between two continuous inverter switching transformation events as a calculation step of the digital twin model based on the switching control signal and triggering the solving calculation of the digital twin model only at an event point where a switching state changes.

4. The digital-twin-based detection method of the loss of excitation of the metro permanent magnet motor according to claim 1, characterized in that, The optimization of the digital twin model using the particle swarm optimization strategy comprises: independently performing multiple complete particle swarm optimization processes and using a randomly initialized particle swarm for each optimization; screening optimal parameter combinations obtained in each optimization by setting a fitness threshold to eliminate results that do not reach a preset accuracy; and statistically averaging all effective optimization results that pass the screening and taking an average value as a final optimization parameter.

5. The digital-twin-based detection method of the loss of excitation of the metro permanent magnet motor according to claim 4, characterized in that, The optimization parameters comprise a flux value, a stator resistance, a stator inductance, a rotor resistance and a rotor inductance.

6. The digital-twin-based detection method of the loss of excitation of the metro permanent magnet motor according to claim 1, characterized in that, The building of the evaluation function based on errors between the real-time operation data and the twin output data comprises: building a first evaluation function by a root mean square error based on phase currents output by the digital twin model and phase currents actually collected by the permanent magnet motor; building a second evaluation function by a root mean square error based on a direct current side current output by the digital twin model and a direct current side current actually collected by the permanent magnet motor.

7. A digital-twin-based city rail permanent magnet motor loss of excitation detection device, characterized in that, The device comprises: a twin model building unit configured to build a digital twin model reproducing dynamic characteristics of an actual permanent magnet motor system in a vehicle-mounted traction controller; a data acquisition unit configured to acquire real-time operation data of the permanent magnet motor through sensors of an existing urban rail traction system and receive a switching control signal generated by the vehicle-mounted traction controller; a twin model solving unit configured to solve the digital twin model in an event-driven manner based on the switching control signal to obtain twin output data; a model optimization unit configured to build an evaluation function based on errors between the real-time operation data and the twin output data and optimize the digital twin model using a particle swarm optimization strategy; and a demagnetization fault alarm unit configured to compare a current value of a permanent magnet flux obtained based on the optimized digital twin model with a preset calibration value and issue a demagnetization fault alarm when a comparison result exceeds a set threshold. The loss-of-field detection unit is configured to obtain a current value of the flux linkage of the permanent magnet based on the optimized digital twin model, compare the current value with a preset calibration value, and issue a loss-of-field fault alarm when a comparison result exceeds a set threshold.

8. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The computer program, when executed by the processor, implements the steps of the method of any one of claims 1 to 6.

9. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program, when executed by the processor, implements the steps of the method of any one of claims 1 to 6.

10. A computer program product comprising computer programs / instructions, characterized in that, The computer program / instruction, when executed by the processor, implements the steps of the method of any one of claims 1 to 6.

Images