Dynamic Balance Detection Method and System for Electric Tricycle Body Structure

Abstract

The present application provides a dynamic balance detection method and system for the body structure of an electric tricycle, relating to the technical field of dynamic balance testing, including: determining the distribution of initial standard detection parameters according to navigation route matching; correcting the distribution of initial standard detection parameters according to the real-time load and real-time vehicle speed; mapping and judging the distribution of balance sensing parameters according to the adapted standard detection parameter distribution, and if there are parameter abnormalities, continuously monitoring to obtain an abnormal feature distribution sequence; performing a similarity comparison on the abnormal driving behavior features and the abnormal feature distribution sequence, and if the similarity coefficient is less than the similarity threshold, giving a vehicle imbalance warning. Through the present application, the problems that the traditional detection method cannot effectively adapt to complex driving scenarios, resulting in phenomena such as slow response and frequent false alarms, and insufficient timeliness and accuracy of vehicle imbalance warning can be solved; accurate imbalance monitoring can be achieved in complex driving environments, and the adaptability, timeliness and accuracy of imbalance warning can be significantly improved.

Description

Technical Field

[0001] This application relates to the technical field of dynamic balance testing, and particularly to a method and system for dynamically detecting the balance of an electric tricycle body structure. Background Art

[0002] Traditional methods for dynamically detecting the balance of electric tricycles usually set detection parameters based on the driving behavior of the vehicle in a standard environment, such as a flat road surface, uniform load, and stable vehicle speed. However, in actual driving, factors such as road surface type, slope change, turning, acceleration, and deceleration may greatly affect the balance state of the vehicle body. Traditional methods cannot effectively identify and adapt to these complex driving scenarios, resulting in inaccurate balance detection results. Especially in complex environments such as slopes, slippery road surfaces, or sharp turns, it is difficult to adjust the detection criteria in real time, easily causing misjudgments in balance.

[0003] On the other hand, in the normal driving behavior of some drivers, phenomena such as sudden braking, sudden acceleration, or rapid turning may occur. These behaviors will cause large fluctuations in the dynamic characteristics of the vehicle instantaneously. However, due to the lack of comprehensive judgment of factors such as the driving habits of the driver and environmental factors, traditional balance detection systems are prone to misjudging these instantaneous changes as vehicle imbalance, thereby triggering false alarms.

[0004] In summary, traditional methods for detecting the dynamic balance of tricycles cannot effectively adapt to complex driving scenarios, showing phenomena such as slow response and frequent false alarms, resulting in insufficient timeliness and accuracy of vehicle imbalance warnings, and affecting the driving safety and stability of the vehicle. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for dynamically detecting the balance of an electric tricycle body structure to solve the technical problems that traditional methods for detecting the dynamic balance of tricycles cannot effectively adapt to complex driving scenarios, showing phenomena such as slow response and frequent false alarms, resulting in insufficient timeliness and accuracy of vehicle imbalance warnings.

[0006] In view of the above problems, this application provides a method and system for dynamically detecting the balance of an electric tricycle body structure.

[0007] In a first aspect, the present application provides a method for detecting the dynamic balance of an electric tricycle body structure, which is implemented through a dynamic balance detection system for an electric tricycle body structure, including: dividing and determining a plurality of road surface scenarios with the road surface type, road surface gradient, and turning angle as variable factors, and analyzing the dynamic balance detection standards according to the reference load and reference vehicle speed of the tricycle under the plurality of road surface scenarios to generate a plurality of standard detection parameter distributions, and mapping and constructing a scenario parameter comparison library; during the driving process of the tricycle, matching and determining the initial standard detection parameter distribution according to the navigation route and the scenario parameter comparison library; monitoring and obtaining the real-time load and real-time vehicle speed of the tricycle, and performing mapping correction on the initial standard detection parameter distribution according to the real-time load and real-time vehicle speed to obtain an adapted standard detection parameter distribution; monitoring and obtaining the balance sensing parameter distribution of the tricycle at the first frequency, and performing mapping judgment on the balance sensing parameter distribution according to the adapted standard detection parameter distribution. If there are parameter abnormalities, continuous monitoring is performed at the second frequency to obtain an abnormal feature distribution sequence; performing a similarity comparison between the matching driving behavior abnormal features and the abnormal feature distribution sequence. If the comparison similarity coefficient is less than the similarity threshold, a vehicle imbalance warning is issued.

[0008] Optionally, the method for detecting the dynamic balance of an electric tricycle body structure further includes: dividing step lengths according to the road surface gradient and dividing step lengths according to the turning angle to divide and determine a plurality of road surface gradient intervals and a plurality of turning angle intervals; randomly combining the predetermined road surface type with the plurality of road surface gradient intervals and the plurality of turning angle intervals to generate a plurality of road surface scenarios; randomly selecting a first road surface scenario from the plurality of road surface scenarios, analyzing the dynamic balance detection standards according to the reference load and reference vehicle speed of the tricycle to generate a first standard detection parameter distribution, and adding it to the plurality of standard detection parameter distributions; mapping and constructing the scenario parameter comparison library according to the plurality of road surface scenarios and the plurality of standard detection parameter distributions.

[0009] Optionally, the method for detecting the dynamic balance of an electric tricycle body structure further includes: obtaining a sensor layout array for the tricycle to perform dynamic balance detection, where the sensor layout array includes several layout positions, each layout position includes at least one sensor, and the sensors at least include an acceleration sensor, a gyroscope sensor, and an inclination sensor; simulating and constructing a dynamic balance test space according to the electric tricycle body structure, driving characteristics, and the sensor layout array, and performing parameter configuration according to the first road surface scenario, reference load, and reference vehicle speed to generate a first dynamic balance test space; performing a vehicle dynamic balance iterative simulation test in the first dynamic balance test space, and setting the maximum detection parameter when it is less than the preset vehicle imbalance ratio as the first standard detection parameter to generate the first standard detection parameter distribution.

[0010] Optionally, the dynamic balance detection method for the electric tricycle body structure further includes: taking the tricycle body structure and the sensor layout array as constraints, querying the networked driving logs of similar tricycles, and collecting a sample standard detection parameter distribution set, a sample vehicle load set, a sample vehicle speed set, and a sample corrected detection parameter distribution set; constructing a detection parameter correction model based on a generative adversarial network, where the detection parameter correction model includes a detection parameter generator and a detection parameter discriminator; iteratively training the detection parameter generator and the detection parameter discriminator with the sample standard detection parameter distribution set, the sample vehicle load set, the sample vehicle speed set, and the sample corrected detection parameter distribution set until both the generation loss function and the discrimination loss function converge, and obtaining the trained detection parameter correction model; inputting the real-time load, real-time vehicle speed, and initial standard detection parameter distribution into the detection parameter correction model for mapping correction, and outputting the adapted standard detection parameter distribution.

[0011] Optionally, the dynamic balance detection method for the electric tricycle body structure further includes: configuring a first frequency and a second frequency, collecting data from the sensor layout array according to the first frequency to obtain a balance sensing parameter distribution, where the second frequency is greater than the first frequency; performing a mapping judgment on the balance sensing parameter distribution according to the adapted standard detection parameter distribution, and if the sensing parameter does not meet the corresponding standard detection parameter, generating a parameter anomaly signal, and collecting data from the sensor layout array according to the second frequency to construct an anomaly feature distribution, and obtaining an anomaly feature distribution sequence within a preset time zone.

[0012] Optionally, the dynamic balance detection method for the electric tricycle body structure further includes: monitoring and obtaining the navigation route and driving behavior data of the vehicle within the preset time zone, and performing driving feature analysis according to the navigation route and driving behavior data to construct a driving feature portrait; based on the second frequency, performing anomaly feature simulation analysis according to the navigation route and the driving feature portrait to generate a simulated anomaly feature distribution sequence; performing a similarity mapping comparison on the simulated anomaly feature distribution sequence and the anomaly feature distribution sequence, outputting a plurality of similarity coefficients, and obtaining the comparison similarity coefficient after calculating the mean value.

[0013] Optionally, the method for detecting the dynamic balance of the electric tricycle body structure further includes: dividing the navigation route based on the second frequency to determine multiple route segments; taking the tricycle body structure and the sensor layout array as constraints, collecting a sample route set and a sample driving feature portrait set according to the historical network driving records, and annotating the abnormal feature distributions under different sample routes and sample driving feature portraits to obtain a sample abnormal feature distribution set; using the sample route set, the sample driving feature portrait set, and the sample abnormal feature distribution set to iteratively train a generative adversarial network until convergence, and obtaining an abnormal feature simulation plug-in; using the abnormal feature simulation plug-in, performing abnormal feature simulation analysis according to the multiple route segments and the driving feature portrait, outputting multiple simulated abnormal feature distributions, and constructing the simulated abnormal feature distribution sequence.

[0014] In a second aspect, the present application also provides a system for detecting the dynamic balance of an electric tricycle body structure, which is used to execute the method for detecting the dynamic balance of an electric tricycle body structure as described in the first aspect, and includes: a dynamic balance detection standard analysis module, which is used to divide and determine multiple road surface scenarios with the road surface type, road surface slope, and turning angle as variable factors, and perform dynamic balance detection standard analysis according to the reference load and reference vehicle speed of the tricycle under the multiple road surface scenarios, generate multiple standard detection parameter distributions, and map and construct a scenario parameter comparison library; an initial standard detection parameter distribution determination module, which is used to match and determine the initial standard detection parameter distribution according to the navigation route and the scenario parameter comparison library during the driving process of the tricycle; an initial standard detection parameter distribution correction module, which is used to monitor and obtain the real-time load and real-time vehicle speed of the tricycle, and perform mapping correction on the initial standard detection parameter distribution according to the real-time load and real-time vehicle speed to obtain an adapted standard detection parameter distribution; an abnormal feature distribution sequence acquisition module, which is used to monitor and obtain the balance sensing parameter distribution of the tricycle at the first frequency, perform mapping judgment on the balance sensing parameter distribution according to the adapted standard detection parameter distribution, and if there is a parameter abnormality, continuously monitor at the second frequency to obtain an abnormal feature distribution sequence; a vehicle imbalance warning module, which is used to perform a similarity comparison on the matching driving behavior abnormal features and the abnormal feature distribution sequence, and if the comparison similarity coefficient is less than the similarity threshold, issue a vehicle imbalance warning.

[0015] One or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0016] By taking the road surface type, road surface slope, and turning angle as variable factors, multiple road surface scenarios are determined and divided. Under the multiple road surface scenarios, dynamic balance detection standard analysis is carried out according to the reference load and reference vehicle speed of the tricycle, and multiple standard detection parameter distributions are generated. A scenario parameter comparison library is mapped and constructed. Then, during the driving process of the tricycle, the initial standard detection parameter distribution is determined by matching according to the navigation route and the scenario parameter comparison library. Further, the real-time load and real-time vehicle speed of the tricycle are monitored and obtained. The initial standard detection parameter distribution is mapped and corrected according to the real-time load and real-time vehicle speed to obtain an adapted standard detection parameter distribution. Then, the balance sensing parameter distribution of the tricycle is monitored and obtained at the first frequency. Mapping judgment is carried out on the balance sensing parameter distribution according to the adapted standard detection parameter distribution. If there is a parameter anomaly, continuous monitoring is carried out at the second frequency to obtain an abnormal feature distribution sequence. Finally, a similarity comparison is made between the matching driving behavior abnormal features and the abnormal feature distribution sequence. If the similarity coefficient of the comparison is less than the similarity threshold, a vehicle imbalance warning is issued. That is to say, by introducing technical means such as dynamically adjusting detection parameters, multi-frequency monitoring, and driving feature similarity comparison, accurate imbalance monitoring can be achieved in a complex driving environment, significantly improving the adaptability, timeliness, and accuracy of vehicle imbalance warning, thereby effectively enhancing the safety and driving stability of the electric tricycle.

[0017] The above description is only an overview of the technical solution of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features, and advantages of the present application more obvious and understandable, the following specifically gives the specific implementation manners of the present application. It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become easily understandable through the following description. Brief Description of the Drawings

[0018] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings described below are only exemplary, and for those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0019] Figure 1 It is a flowchart showing the dynamic balance detection method for the body structure of an electric tricycle according to the present application.

[0020] Figure 2 It is a flowchart showing the construction of the scenario parameter comparison library in the dynamic balance detection method for the body structure of an electric tricycle according to the present application.

[0021] Figure 3This is a schematic structural diagram of a dynamic balance detection system for the body structure of an electric tricycle in this application.

[0022] Explanation of reference numerals in the drawings:

[0023] Dynamic balance detection standard analysis module 11, initial standard detection parameter distribution determination module 12, initial standard detection parameter distribution correction module 13, abnormal feature distribution sequence acquisition module 14, vehicle imbalance warning module 15. Detailed implementation manners

[0024] This application provides a dynamic balance detection method and system for the body structure of an electric tricycle, which solves the technical problems that the traditional dynamic balance detection method for tricycles cannot effectively adapt to complex driving scenarios, has phenomena such as slow response and frequent false alarms, resulting in insufficient timeliness and accuracy of vehicle imbalance warning. By introducing technical means such as dynamically adjusting detection parameters, multi-frequency monitoring, and similarity comparison of driving characteristics, accurate imbalance monitoring can be achieved in complex driving environments, significantly improving the adaptability, timeliness, and accuracy of vehicle imbalance warning, thereby effectively enhancing the safety and driving stability of electric tricycles.

[0025] Next, the technical solutions in this application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments of this application. It should be understood that this application is not limited by the exemplary embodiments described herein. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of this application. Additionally, it should be noted that for the sake of description, only the parts related to this application are shown in the drawings rather than all of them.

[0026] Embodiment 1. Please refer to the attached Figure 1 , this application provides a dynamic balance detection method for the body structure of an electric tricycle, which is applied to a dynamic balance detection system for the body structure of an electric tricycle, and specifically includes the following steps:

[0027] S100: Using the road surface type, road surface slope, and turning angle as variable factors, divide and determine multiple road surface scenarios, and under the multiple road surface scenarios, conduct dynamic balance detection standard analysis according to the reference load and reference vehicle speed of the tricycle, generate multiple standard detection parameter distributions, and map and construct a scenario parameter comparison library.

[0028] Further, as Figure 2 shown, step S100 of this application further includes:

[0029] S110: Divide the step length according to the road surface slope and the turning angle, and determine multiple road surface slope intervals and multiple turning angle intervals; S120: After randomly combining the predetermined road surface types and the multiple road surface slope intervals and multiple turning angle intervals, generate multiple road surface scenarios.

[0030] Specifically, the road surface slope is one of the important factors affecting the balance of the tricycle. The change of the slope will directly affect the center of gravity distribution and driving stability of the vehicle body. In order to adapt to different slope driving conditions, first divide the step length according to the road surface slope (for example, set the divided step length to 5 degrees) to divide the road surface slope into multiple intervals. The slope value difference in each interval maintains a certain step length, and at the same time, each interval corresponds to different detection standards. For example, the slope interval is divided as follows: Interval 1, the slope is greater than or equal to 0 degrees and less than 5 degrees (flat road surface); Interval 2, the slope is greater than or equal to 5 degrees and less than 10 degrees (slight uphill / downhill); Interval 3, the slope is greater than or equal to 10 degrees and less than 15 degrees (medium slope). This division method can identify the dynamic balance characteristics of the vehicle under different slope conditions and provide accurate parameter settings for subsequent balance detection. Turning is one of the important factors affecting the balance of the tricycle. Especially when making a sharp turn, the size of the turning angle directly affects the lateral force of the vehicle body. Therefore, it is necessary to divide different intervals according to the change of different turning angles and set corresponding detection standards for each interval; then divide the step length according to the turning angle (for example, 15 degrees) to divide the turning angle into multiple intervals. For example, Interval 1, the turning angle is greater than or equal to 0 degrees and less than 15 degrees (small turn); Interval 2, the turning angle is greater than or equal to 15 degrees and less than 30 degrees (medium turn); Interval 3, the turning angle is greater than or equal to 30 degrees and less than 45 degrees (large turn); obtain multiple road surface slope intervals and multiple turning angle intervals.

[0031] Then obtain the predetermined road surface types, such as asphalt road, gravel road, muddy road, concrete road surface, etc., which can be set according to the actual driving scenario. Different road surface types will affect the friction between the wheels and the ground, thereby affecting the balance stability of the vehicle; then randomly combine the predetermined road surface types and the multiple road surface slope intervals and multiple turning angle intervals, as shown in Table 1, to generate multiple road surface scenarios.

[0032]

[0033] Table 1

[0034] S130: Randomly select a first road surface scenario from the multiple road surface scenarios, analyze the dynamic balance detection standard according to the reference load and reference vehicle speed of the tricycle, generate a first standard detection parameter distribution, and add it to the multiple standard detection parameter distributions.

[0035] Furthermore, step S130 of the present application further includes:

[0036] S131: Obtain the sensor layout array for the tricycle to perform dynamic balance detection. Among them, the sensor layout array includes several layout positions, and each layout position includes at least one sensor. The sensors at least include an acceleration sensor, a gyroscope sensor, and an inclination sensor; S132: Simulate and construct a dynamic balance test space based on the tricycle body structure, driving characteristics, and the sensor layout array, and perform parameter configuration according to the first road surface scenario, reference load, and reference vehicle speed to generate the first dynamic balance test space; S133: Conduct vehicle dynamic balance iterative simulation tests in the first dynamic balance test space, and set the maximum detection parameter when it is less than the preset vehicle imbalance ratio as the first standard detection parameter to generate the first standard detection parameter distribution.

[0037] Specifically, then randomly select any one of the multiple road surface scenarios and set it as the first road surface scenario.

[0038] Obtain the sensor layout array for the tricycle to perform dynamic balance detection. Among them, the sensor layout array includes several layout positions, and each layout position includes at least one sensor (the same position may include multiple different types of sensors). The sensors at least include an acceleration sensor, a gyroscope sensor, and an inclination sensor. The acceleration sensor is used to measure the acceleration changes of the tricycle in three axes (X-axis, Y-axis, and Z-axis). The acceleration sensor can help the system judge the motion state and acceleration / deceleration situation of the vehicle; the gyroscope sensor is used to measure the rotational angular velocity of the tricycle and can provide the rotation information of the vehicle body (such as the rotational change when turning); the inclination sensor is used to detect the inclination angle of the vehicle body and help the system judge whether the tricycle is in a balanced state, especially when going uphill, downhill, or turning. Among them, as shown in Table 2, the sensors should be arranged at the key positions of the tricycle according to its structure to ensure that the dynamic balance information of the vehicle under various driving conditions can be accurately captured. By arranging the acceleration sensor, gyroscope sensor, and inclination sensor at the key positions of the tricycle, the balance state of the vehicle can be monitored in real time, and dynamic balance data can be collected, which provides support for subsequent dynamic balance detection.

[0039]

[0040] Table 2

[0041] Next, obtain the body structure characteristics of the tricycle (such as the length, width, center of gravity position, suspension system, diameter and width of the wheels, etc.) and driving characteristics (such as power parameters, etc.); then, based on the tricycle body structure, driving characteristics, and sensor layout array, conduct simulation modeling in a three-dimensional simulation platform (such as the MATLAB / Simulink platform) to construct a dynamic balance test space, which can simulate the dynamic behavior of the tricycle in various driving scenarios, help detect the balance state of the tricycle, and analyze possible imbalance situations. Then, configure the parameters of the dynamic balance test space according to the first road surface scenario, reference load, and reference vehicle speed, where the reference load and reference vehicle speed can be set according to the actual driving scenario. For example, the reference load is set to a total vehicle mass of 300 kilograms (including the mass of the driver, vehicle body, and basic components), and the reference vehicle speed is set to 30 kilometers per hour to generate a first dynamic balance test space, which will simulate the dynamic balance performance of the vehicle in the first road surface scenario and help evaluate the balance ability of the tricycle in common driving scenarios.

[0042] Then, conduct vehicle dynamic balance iterative simulation tests in the first dynamic balance test space. By setting different detection parameters of the vehicle body (such as acceleration, tilt angle, rotational angular velocity, etc.), analyze the balance state of the vehicle (balanced or unbalanced), and count the imbalance ratio of the vehicle during simulation under different detection parameters; considering the error situation during simulation, configure a preset vehicle imbalance ratio, which is a preset threshold value used to measure whether the vehicle is in an unbalanced state during the test. For example, set the vehicle imbalance ratio to 2%; conduct iterative simulation tests, and set the maximum detection parameter when it is less than the preset vehicle imbalance ratio as the first standard detection parameter. Through multiple iterative tests, simulate the dynamic performance of the vehicle in different scenarios, and gradually adjust the standard detection parameter to obtain multiple first standard detection parameters (such as the maximum tilt angle, maximum acceleration change, maximum angular velocity, etc.) at multiple layout positions; then, arrange the multiple first standard detection parameters according to the multiple layout positions to construct a first standard detection parameter distribution, which includes the standard detection parameters at multiple positions of the sensor layout array in the first road surface scenario.

[0043] Using the same method of constructing the first standard detection parameter distribution, conduct dynamic balance detection standard analysis for other road surface scenarios respectively to obtain multiple standard detection parameter distributions corresponding to multiple road surface scenarios.

[0044] S140: Construct the scenario parameter comparison library according to the mapping of the multiple road surface scenarios and multiple standard detection parameter distributions.

[0045] Specifically, based on the mapping relationship between road surface scenarios and standard detection parameter distributions, the scenario parameter comparison library is constructed according to the mapping of the multiple road surface scenarios and the multiple standard detection parameter distributions. By constructing the scenario parameter comparison library, the detection parameters can be adjusted in real time under different road surface environments, improving the accuracy and response speed of imbalance warnings and ensuring the stability and safety of the tricycle under various driving environments.

[0046] S200: During the tricycle's driving process, determine the initial standard detection parameter distribution according to the navigation route and the scenario parameter comparison library.

[0047] Specifically, during the tricycle's driving process, the position and driving path of the vehicle are obtained in real time through the GPS module or in-vehicle navigation system; then, according to the vehicle's real-time driving position, the navigation data is used to determine the current road surface scenario, and the parameters matching the current road surface characteristics are searched in the scenario parameter comparison library according to the current road surface scenario, obtaining the initial standard detection parameter distribution. By combining the navigation route and the scenario parameter comparison library, it is possible to achieve real-time matching of the standard detection parameter distribution of the tricycle under different driving conditions, and to dynamically determine appropriate standard detection parameters according to real-time navigation data, thereby effectively improving the accuracy and response speed of imbalance detection.

[0048] S300: Monitor and obtain the real-time load and real-time vehicle speed of the tricycle, and perform mapping correction on the initial standard detection parameter distribution according to the real-time load and real-time vehicle speed to obtain the adapted standard detection parameter distribution.

[0049] Furthermore, step S300 of the present application further includes:

[0050] S310: Constrained by the tricycle body structure and sensor layout array, query the networked driving logs of similar tricycles, and collect a sample standard detection parameter distribution set, a sample vehicle load set, a sample vehicle speed set, and a sample corrected detection parameter distribution set; S320: Construct a detection parameter correction model based on a generative adversarial network, where the detection parameter correction model includes a detection parameter generator and a detection parameter discriminator; S330: Iteratively train the detection parameter generator and the detection parameter discriminator with the sample standard detection parameter distribution set, the sample vehicle load set, the sample vehicle speed set, and the sample corrected detection parameter distribution set until both the generation loss function and the discriminant loss function converge, obtaining the trained detection parameter correction model; S340: Input the real-time load, real-time vehicle speed, and initial standard detection parameter distribution into the detection parameter correction model for mapping correction, and output the adapted standard detection parameter distribution.

[0051] Specifically, using the tricycle body structure and the sensor layout array as retrieval constraints, query the networked driving logs of similar tricycles. The networked driving logs contain various dynamic data recorded during the actual use of the vehicle, including on-vehicle sensors, GPS systems, and other monitoring devices, etc. Collect the sample standard detection parameter distribution set, the sample vehicle load set, and the sample vehicle speed set, and obtain the sample corrected detection parameter distribution after adjusting the sample standard detection parameter distribution according to different sample vehicle loads and sample vehicle speeds, thereby obtaining the sample corrected detection parameter distribution set.

[0052] The generative adversarial network is a deep learning model. By using the generative adversarial network to correct the detection parameters, it can automatically generate appropriate detection parameters under different driving conditions and load situations, and continuously optimize the accuracy and adaptability of the detection parameters through adversarial training with the discriminator. First, construct a detection parameter correction model based on the generative adversarial network. The detection parameter correction model includes a detection parameter generator and a detection parameter discriminator. The task of the detection parameter generator is to generate appropriate detection parameters according to the input vehicle dynamic information (such as vehicle speed, load); the task of the detection parameter discriminator is to judge whether the detection parameters generated by the generator are reasonable and whether they can accurately reflect the dynamic balance characteristics of the vehicle under given conditions. Then, use the sample standard detection parameter distribution set, the sample vehicle load set, the sample vehicle speed set, and the sample corrected detection parameter distribution set as training data to iteratively train the detection parameter generator and the detection parameter discriminator. First, by inputting the training data, the generator starts to generate detection parameters. The goal of the generator is to generate results that match the real detection parameters as much as possible through continuous optimization; then, the discriminator compares the generated parameters with the real parameters, calculates the loss and returns the feedback to the generator. By optimizing the loss functions of the generator and the discriminator, and using the gradient descent method to adjust the weights of the generator and the discriminator, gradually optimize the generation effect of the detection parameters. When the loss functions of the generator and the discriminator tend to be stable and the loss values no longer fluctuate significantly, the model considers the training to be completed, and obtains the trained detection parameter correction model. At this time, the detection parameter generator can accurately generate detection parameters that conform to the actual dynamic balance characteristics, and the discriminator can effectively distinguish the difference between the generated parameters and the real parameters.

[0053] Then, during the tricycle driving process, monitor and obtain the real-time load and real-time vehicle speed of the tricycle. For example, install force sensors (such as strain gauges, piezoelectric sensors) at appropriate positions on the vehicle body (such as the connection parts of the frame, wheels and suspension systems) to monitor the change of the load in real time; use optoelectronic sensors to measure the rotation speed of the wheels and indirectly calculate the driving speed of the vehicle. Then input the real-time load, real-time vehicle speed and the initial standard detection parameter distribution into the detection parameter correction model for mapping correction, and output the corrected adapted standard detection parameter distribution.

[0054] By monitoring the load and speed of the tricycle in real time, combining these data with the initial standard detection parameter distribution, and inputting them into the detection parameter correction model for dynamic correction, it can ensure that the system adapts to the actual working conditions of the vehicle in real time. The generated corrected detection parameters (such as the maximum tilt angle, acceleration change, angular velocity, etc.) will be used for real-time dynamic balance detection to ensure accurate monitoring of the vehicle's balance under different loads, speeds, and road conditions, thereby effectively improving the accuracy and adaptability of the dynamic balance detection system and enhancing the driving safety of the tricycle.

[0055] S400: Monitor and obtain the balance sensing parameter distribution of the tricycle according to the first frequency, perform a mapping judgment on the balance sensing parameter distribution according to the adapted standard detection parameter distribution. If there are parameter abnormalities, continuously monitor according to the second frequency to obtain the abnormal feature distribution sequence.

[0056] Furthermore, step S400 of this application further includes:

[0057] S410: Configure the first frequency and the second frequency, collect data from the sensor layout array according to the first frequency to obtain the balance sensing parameter distribution, where the second frequency is greater than the first frequency; S420: Perform a mapping judgment on the balance sensing parameter distribution according to the adapted standard detection parameter distribution. If the sensing parameters do not meet the corresponding standard detection parameters, generate a parameter abnormality signal, and collect data from the sensor layout array according to the second frequency to construct an abnormal feature distribution and obtain the abnormal feature distribution sequence within the preset time zone.

[0058] Specifically, first, configure the first frequency and the second frequency, where the second frequency is greater than the first frequency. The first frequency is used for regular real-time data collection, and a relatively low collection frequency can be set according to the vehicle's driving state and the system's computing power, for example, once every 1 minute. When potential balance abnormalities occur in the vehicle, more accurate data can be obtained by increasing the collection frequency, so the second frequency is set to be greater than the first frequency, for example, once every 10 seconds. This high-frequency data collection can detect and respond more quickly when an abnormality occurs. Then, during the vehicle's driving process, collect data from the sensor layout array according to the first frequency to obtain the balance sensing parameter distribution (including the data collected by sensors at multiple layout positions).

[0059] Then, based on the detected parameter distribution of the adaptation standard, a mapping judgment is performed on the balanced sensing parameter distribution, that is, the balanced sensing parameters are compared with the detected parameters of the adaptation standard at the corresponding positions. If the sensing parameters do not meet the corresponding detected parameters, a parameter anomaly signal is generated. Then, data is collected from the sensor layout array according to the second frequency, and the positions of the abnormal parameters in the balanced sensing parameter distribution collected each time are marked, recording their timestamps, parameter types, abnormal values, and the differences from the standard parameters, generating an abnormal feature distribution. Continuous monitoring and data collection are performed according to the second frequency to obtain a sequence of abnormal feature distributions within a preset time zone (such as within 3 minutes), and the preset time zone can be set according to the actual situation.

[0060] S500: Perform a similarity comparison between the abnormal characteristics of the matching driving behavior and the sequence of abnormal feature distributions. If the similarity comparison coefficient is less than the similarity threshold, a vehicle imbalance warning is issued.

[0061] Furthermore, step S500 of the present application further includes:

[0062] S510: Monitor and obtain the navigation route and driving behavior data of the vehicle within the preset time zone, and perform driving feature analysis based on the navigation route and driving behavior data to construct a driving feature portrait.

[0063] Specifically, monitor and obtain the navigation route and driving behavior data of the vehicle within the preset time zone. The navigation route data of the vehicle is the trajectory information of the vehicle's travel, which can be obtained through the GPS system and records the vehicle's real-time position and travel path. The driving behavior data refers to various data reflecting the driver's operating habits and driving states, usually obtained through in-vehicle sensors (such as acceleration sensors, etc.). For example, the driver's acceleration and braking behaviors, the angle and speed changes when the driver turns, the average driving speed, etc. Then, driving feature analysis is performed based on the navigation route and driving behavior data, such as analyzing the vehicle's speed changes on different road sections, identifying whether the driver maintains a stable speed, especially the speed fluctuations on slopes or turns, to construct a driving feature portrait. The driving feature portrait is a comprehensive description of the driver's behavior formed by comprehensively analyzing the driver's performance under different road conditions and driving behaviors, including driving patterns (such as smooth driving and turning habits, etc.) and speed characteristics. By generating a driving feature portrait, the driver's driving habits, driving safety, and adaptability can be comprehensively reflected, and the driver's behavior can be better understood, thereby providing more accurate balance detection and imbalance warning.

[0064] S520: Based on the second frequency, perform abnormal feature simulation analysis according to the navigation route and driving feature portrait to generate a sequence of simulated abnormal feature distributions.

[0065] Furthermore, step S520 of the present application further includes:

[0066] S521: Divide the navigation route based on the second frequency to determine multiple route segments; S522: With the tricycle body structure and the sensor layout array as constraints, collect a sample route set and a sample driving feature portrait set according to the historical network driving records, and label the abnormal feature distributions under different sample routes and sample driving feature portraits to obtain a sample abnormal feature distribution set; S523: Use the sample route set, the sample driving feature portrait set, and the sample abnormal feature distribution set to iteratively train the generative adversarial network until convergence to obtain an abnormal feature simulation plugin; S524: Use the abnormal feature simulation plugin to perform abnormal feature simulation analysis based on the multiple route segments and the driving feature portrait, output multiple simulated abnormal feature distributions, and construct the simulated abnormal feature distribution sequence.

[0067] Specifically, divide the navigation route according to the second frequency to obtain multiple route segments, where each route segment is the distance interval under adjacent second frequencies. Then, with the tricycle body structure and the sensor layout array as retrieval constraints, collect a sample route set and a sample driving feature portrait set according to the historical network driving records; then label the abnormal feature distributions under different sample routes and sample driving feature portraits, that is, when the balance sensing data (such as acceleration, tilt angle, angular velocity, etc.) of the vehicle exceeds the set standard value, the system will label this data point as abnormal, and these abnormal data points will be labeled as abnormal features to obtain a sample abnormal feature distribution set.

[0068] The generative adversarial network consists of a generator and a discriminator. The purpose is to make the generator generate as realistic abnormal feature distribution data as possible through adversarial training. The generator generates a simulated abnormal feature distribution through a neural network based on the input route segment and the driving feature portrait; the discriminator is used to evaluate the similarity between the abnormal feature distribution output by the generator and the real data. Then, use the sample route set, the sample driving feature portrait set, and the sample abnormal feature distribution set as training data to iteratively train the generative adversarial network, continuously optimize the parameters of the generator and the discriminator until the generator can generate abnormal feature data close to the real ones and the discriminator cannot distinguish the generated data from the real data until the loss function converges to obtain a trained abnormal feature simulation plugin. The abnormal feature simulation plugin can generate corresponding abnormal feature distribution data based on the input route segment and the driving feature portrait.

[0069] Further combine the multiple route segments and the driving feature portrait to obtain multiple input data, and sequentially input them into the abnormal feature simulation plugin for abnormal feature simulation analysis, output multiple simulated abnormal feature distributions, and construct a simulated abnormal feature distribution sequence.

[0070] S530: Perform a similarity mapping comparison on the simulated abnormal feature distribution sequence and the abnormal feature distribution sequence, output multiple similarity coefficients, and obtain the comparison similarity coefficient after calculating the mean value.

[0071] Specifically, performing a similarity mapping comparison on the simulated abnormal feature distribution sequence and the abnormal feature distribution sequence means analyzing the similarity between the simulated abnormal feature distribution and the abnormal feature distribution at the same time node. The cosine similarity algorithm can be used to calculate the similarity. Cosine similarity measures the similarity between two vectors by calculating the angle between them, with a value range from -1 (completely opposite) to 1 (completely identical), and 0 indicating dissimilarity. Each pair of data point comparisons will output a similarity coefficient, representing the similarity degree between the simulated data and the actual data, obtaining multiple similarity coefficients; then calculate the mean value of the multiple similarity coefficients to obtain the comparison similarity coefficient, and the comparison similarity coefficient reflects the overall similarity degree between the simulated data and the actual data.

[0072] Configure a similarity threshold, which is used to determine whether the difference between the simulated data and the actual data exceeds the tolerance range and can be set according to the actual scenario. For example, set the similarity threshold to 60%. When the comparison similarity coefficient is less than the preset similarity threshold, it indicates that the difference between the simulated data and the actual data is large, that is, the current abnormality is not caused by the driver's driving behavior, and then a vehicle imbalance warning is issued; when the comparison similarity coefficient is greater than or equal to the preset similarity threshold, it indicates that the difference between the simulated data and the actual data is small, and it is usually considered that the current abnormal feature is caused by the driver's driving behavior, and then no vehicle imbalance warning is issued.

[0073] By configuring the similarity threshold and comparing the simulated data and the actual data, it is possible to determine whether the abnormality is caused by the driver's behavior, thereby avoiding misjudgments of imbalance caused by different driving characteristics and effectively improving the accuracy and reliability of vehicle imbalance warnings.

[0074] In summary, a dynamic balance detection method for the body structure of an electric tricycle provided by this application has the following technical effects:

[0075] By introducing technical means such as dynamically adjusting detection parameters, multi-frequency monitoring, and driving feature similarity comparison, it is possible to solve the technical problems that traditional detection methods cannot effectively adapt to complex driving scenarios, have phenomena such as slow response and frequent false alarms, resulting in insufficient timeliness and accuracy of vehicle imbalance warnings; it is possible to achieve precise imbalance monitoring in complex driving environments, significantly improve the adaptability, timeliness, and accuracy of vehicle imbalance warnings, thereby effectively enhancing the safety and driving stability of electric tricycles.

[0076] Embodiment 2. Based on a dynamic balance detection method for the body structure of an electric tricycle in the foregoing embodiment and with the same inventive concept, the present application further provides a dynamic balance detection system for the body structure of an electric tricycle. Please refer to the appendix Figure 3 , including: a dynamic balance detection standard analysis module 11, configured to divide and determine a plurality of road surface scenarios by using the road surface type, road surface slope, and turning angle as variable factors, and perform dynamic balance detection standard analysis according to the reference load and reference vehicle speed of the tricycle in the plurality of road surface scenarios, generate a plurality of standard detection parameter distributions, and map and construct a scenario parameter comparison library; an initial standard detection parameter distribution determination module 12, configured to match and determine an initial standard detection parameter distribution according to the navigation route and the scenario parameter comparison library during the driving process of the tricycle; an initial standard detection parameter distribution correction module 13, configured to monitor and obtain the real-time load and real-time vehicle speed of the tricycle, and perform mapping correction on the initial standard detection parameter distribution according to the real-time load and real-time vehicle speed to obtain an adapted standard detection parameter distribution; an abnormal feature distribution sequence acquisition module 14, configured to monitor and obtain the balance sensing parameter distribution of the tricycle at a first frequency, perform mapping judgment on the balance sensing parameter distribution according to the adapted standard detection parameter distribution, and if there is a parameter abnormality, continuously monitor at a second frequency to obtain an abnormal feature distribution sequence; a vehicle imbalance warning module 15, configured to perform a similarity comparison on the matching driving behavior abnormal features and the abnormal feature distribution sequence, and if the similarity comparison coefficient is less than the similarity threshold, issue a vehicle imbalance warning.

[0077] Further, the dynamic balance detection system for the body structure of an electric tricycle is further configured to: divide and determine a plurality of road surface slope intervals and a plurality of turning angle intervals according to the road surface slope division step length and the turning angle division step length; randomly combine a predetermined road surface type with the plurality of road surface slope intervals and the plurality of turning angle intervals to generate a plurality of road surface scenarios; randomly select a first road surface scenario from the plurality of road surface scenarios, perform dynamic balance detection standard analysis according to the reference load and reference vehicle speed of the tricycle, generate a first standard detection parameter distribution, and add it to the plurality of standard detection parameter distributions; map and construct the scenario parameter comparison library according to the plurality of road surface scenarios and the plurality of standard detection parameter distributions.

[0078] Further, the dynamic balance detection system for the electric tricycle body structure is further configured to: obtain a sensor layout array for the tricycle to perform dynamic balance detection, wherein the sensor layout array includes a plurality of layout positions, each layout position includes at least one sensor, and the sensors at least include an acceleration sensor, a gyroscope sensor, and an inclination sensor; simulate and construct a dynamic balance test space based on the electric tricycle body structure, driving characteristics, and the sensor layout array, and perform parameter configuration according to the first road surface scenario, reference load, and reference vehicle speed to generate a first dynamic balance test space; perform vehicle dynamic balance iterative simulation tests in the first dynamic balance test space, and set the maximum detection parameter when it is less than the preset vehicle imbalance ratio as the first standard detection parameter to generate the first standard detection parameter distribution.

[0079] Further, the dynamic balance detection system for the electric tricycle body structure is further configured to: query the network driving logs of similar tricycles with the electric tricycle body structure and the sensor layout array as constraints, and collect a sample standard detection parameter distribution set, a sample vehicle load set, a sample vehicle speed set, and a sample corrected detection parameter distribution set; construct a detection parameter correction model based on a generative adversarial network, where the detection parameter correction model includes a detection parameter generator and a detection parameter discriminator; iteratively train the detection parameter generator and the detection parameter discriminator with the sample standard detection parameter distribution set, the sample vehicle load set, the sample vehicle speed set, and the sample corrected detection parameter distribution set until both the generation loss function and the discrimination loss function converge, and obtain the trained detection parameter correction model; input the real-time load, real-time vehicle speed, and initial standard detection parameter distribution into the detection parameter correction model for mapping correction, and output the adapted standard detection parameter distribution.

[0080] Further, the dynamic balance detection system for the electric tricycle body structure is further configured to: configure a first frequency and a second frequency, collect data from the sensor layout array according to the first frequency to obtain a balance sensing parameter distribution, where the second frequency is greater than the first frequency; perform mapping judgment on the balance sensing parameter distribution according to the adapted standard detection parameter distribution. If the sensing parameter does not meet the corresponding standard detection parameter, generate a parameter anomaly signal, and collect data from the sensor layout array according to the second frequency to construct an anomaly feature distribution, and obtain an anomaly feature distribution sequence within a preset time zone.

[0081] Further, the dynamic balance detection system for the electric tricycle body structure is further configured to: monitor and obtain the navigation route and driving behavior data of the vehicle within the preset time zone, perform driving feature analysis based on the navigation route and driving behavior data, and construct a driving feature portrait; based on the second frequency, perform abnormal feature simulation analysis according to the navigation route and the driving feature portrait, and generate a simulated abnormal feature distribution sequence; perform similarity mapping comparison on the simulated abnormal feature distribution sequence and the abnormal feature distribution sequence, output a plurality of similarity coefficients, and obtain the comparison similarity coefficient after calculating the mean value.

[0082] Further, the dynamic balance detection system for the electric tricycle body structure is further configured to: divide the navigation route based on the second frequency to determine a plurality of route segments; constrained by the tricycle body structure and the sensor layout array, collect a sample route set and a sample driving feature portrait set according to the historical network driving records, and label the abnormal feature distributions under different sample routes and sample driving feature portraits to obtain a sample abnormal feature distribution set; use the sample route set, the sample driving feature portrait set, and the sample abnormal feature distribution set to iteratively train a generative adversarial network until convergence, and obtain an abnormal feature simulation plug-in; use the abnormal feature simulation plug-in to perform abnormal feature simulation analysis according to the plurality of route segments and the driving feature portrait, output a plurality of simulated abnormal feature distributions, and construct the simulated abnormal feature distribution sequence.

[0083] The various embodiments in this specification are described in a progressive manner. The key points of each embodiment are the differences from other embodiments. The dynamic balance detection method and specific examples in the first embodiment described above are equally applicable to the dynamic balance detection system in this embodiment. Through the detailed description of the dynamic balance detection method for the electric tricycle body structure above, those skilled in the art can clearly know the dynamic balance detection system in this embodiment. Therefore, for the sake of simplicity of the specification, it will not be described in detail here. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0084] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0085] Obviously, those skilled in the art can make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of this application and its equivalent technologies, this application is also intended to cover these changes and modifications.

Claims

1. A method for detecting the dynamic balance of an electric tricycle body structure, characterized in that: Methods include: Taking road type, road slope and turning angle as change factors, multiple road scenes are divided and determined, and under the multiple road scenes, dynamic balance detection standard analysis is performed according to the reference load and reference speed of the tricycle, multiple standard detection parameter distributions are generated, and a scene parameter comparison library is mapped and constructed; During the driving of the tricycle, the initial standard detection parameter distribution is determined according to the navigation route and the scene parameter reference library; Monitor and obtain the real-time load and real-time speed of the tricycle, and perform mapping correction on the initial standard detection parameter distribution according to the real-time load and real-time speed to obtain the adapted standard detection parameter distribution; Acquire the balance sensor parameter distribution of the tricycle according to the first frequency monitoring, map and judge the balance sensor parameter distribution according to the adaptation standard detection parameter distribution, and if there is a parameter abnormality, continuously monitor according to the second frequency to acquire an abnormal feature distribution sequence; Performing a similarity comparison between the matched driving behavior abnormal features and the abnormal feature distribution sequence, and if the comparison similarity coefficient is less than a similarity threshold, a vehicle imbalance warning is issued; Performing a similarity comparison between the matched driving behavior abnormal features and the abnormal feature distribution sequence, including: Monitor and obtain the navigation route and driving behavior data of the vehicle in a preset time zone, and perform driving characteristic analysis based on the navigation route and driving behavior data to construct a driving characteristic profile; Based on the second frequency, performing abnormal feature simulation analysis according to the navigation route and the driving feature portrait to generate a simulated abnormal feature distribution sequence; A similarity mapping comparison is performed on the simulated abnormal feature distribution sequence and the abnormal feature distribution sequence, and a plurality of similarity coefficients are output. The comparison similarity coefficient is obtained after the average is calculated.

2. The dynamic balance detection method of the electric tricycle body structure according to claim 1 is characterized in that: Taking the road type, road slope and turning angle as the variation factors, multiple road scenes are divided and determined, and in the multiple road scenes, dynamic balance detection standard analysis is performed according to the reference load and reference speed of the tricycle, multiple standard detection parameter distributions are generated, and a scene parameter comparison library is mapped and constructed, including: According to the road surface slope and the turning angle, the step length is divided to determine a plurality of road surface slope intervals and a plurality of turning angle intervals; After randomly combining the predetermined road surface type and the plurality of road surface slope intervals and the plurality of turning angle intervals, a plurality of road surface scenes are generated; Randomly selecting a first road scene from the multiple road scenes, performing a dynamic balance detection standard analysis according to a reference load and a reference speed of the tricycle, generating a first standard detection parameter distribution, and adding the first standard detection parameter distribution to the multiple standard detection parameter distributions; The scene parameter comparison library is constructed according to the multiple road scenes and multiple standard detection parameter distribution mappings.

3. The dynamic balance detection method of the electric tricycle body structure according to claim 2 is characterized in that: A dynamic balance detection standard analysis is performed according to the reference load and reference speed of the tricycle to generate a first standard detection parameter distribution, including: Acquire a sensor arrangement array for dynamic balance detection of a tricycle, wherein the sensor arrangement array includes a plurality of arrangement positions, each arrangement position includes at least one sensor, and the sensor includes at least an acceleration sensor, a gyroscope sensor, and a tilt sensor; A dynamic balance test space is constructed by simulating the body structure, driving characteristics and the sensor arrangement array of the tricycle, and parameters are configured according to the first road scene, reference load and reference vehicle speed to generate a first dynamic balance test space; An iterative simulation test of vehicle dynamic balance is performed in the first dynamic balance test space, a maximum detection parameter when it is less than a preset vehicle imbalance ratio is set as a first standard detection parameter, and the first standard detection parameter distribution is generated.

4. The dynamic balance detection method of the electric tricycle body structure according to claim 3 is characterized in that: Mapping and correcting the initial standard detection parameter distribution according to the real-time load and the real-time vehicle speed to obtain an adapted standard detection parameter distribution includes: Taking the tricycle body structure and sensor layout array as constraints, query the networked driving logs of similar tricycles to collect sample standard detection parameter distribution sets, sample vehicle load sets, sample vehicle speed sets and sample correction detection parameter distribution sets; Constructing a detection parameter correction model based on a generative adversarial network, wherein the detection parameter correction model includes a detection parameter generator and a detection parameter discriminator; Iteratively training the detection parameter generator and the detection parameter discriminator with the sample standard detection parameter distribution set, the sample vehicle load set, the sample vehicle speed set and the sample correction detection parameter distribution set until both the generation loss function and the discrimination loss function converge, thereby obtaining a trained detection parameter correction model; The real-time load, real-time vehicle speed and initial standard detection parameter distribution are input into the detection parameter correction model for mapping correction, and the adapted standard detection parameter distribution is output.

5. The dynamic balance detection method of the electric tricycle body structure according to claim 3 is characterized in that: The balance sensor parameter distribution of the tricycle is acquired according to the first frequency monitoring, and the balance sensor parameter distribution is mapped and judged according to the adaptation standard detection parameter distribution. If there is a parameter abnormality, continuous monitoring is performed according to the second frequency to acquire an abnormal feature distribution sequence, including: configuring a first frequency and a second frequency, collecting data on the sensor array according to the first frequency, and obtaining a balanced sensing parameter distribution, wherein the second frequency is greater than the first frequency; The balanced sensing parameter distribution is mapped and judged according to the adaptation standard detection parameter distribution. If the sensing parameter does not meet the corresponding standard detection parameter, a parameter abnormality signal is generated, and data is collected on the sensor array according to the second frequency to construct an abnormal feature distribution and obtain an abnormal feature distribution sequence within a preset time zone.

6. The dynamic balance detection method of the electric tricycle body structure according to claim 1 is characterized in that: Based on the second frequency, performing abnormal feature simulation analysis according to the navigation route and the driving feature portrait to generate a simulated abnormal feature distribution sequence, including: dividing the navigation route based on the second frequency to determine a plurality of route segments; Taking the tricycle body structure and sensor layout array as constraints, according to historical networked driving records, a sample route set and a sample driving feature portrait set are collected, and abnormal feature distributions under different sample routes and sample driving feature portraits are annotated to obtain a sample abnormal feature distribution set; Iteratively train the generative adversarial network using the sample route set, the sample driving feature portrait set, and the sample abnormal feature distribution set until convergence, and obtain an abnormal feature simulation plug-in; The abnormal feature simulation plug-in is used to perform abnormal feature simulation analysis according to the multiple route segments and driving feature portraits, output multiple simulated abnormal feature distributions, and construct the simulated abnormal feature distribution sequence.

7. A dynamic balance detection system for an electric tricycle body structure, characterized in that: The steps for implementing the dynamic balance detection method of the electric tricycle body structure according to any one of claims 1 to 6 include: A dynamic balance detection standard analysis module is used to determine a plurality of road scenes based on the road type, road slope and turning angle as the variation factors, and to perform a dynamic balance detection standard analysis according to the reference load and reference speed of the tricycle under the plurality of road scenes, generate a plurality of standard detection parameter distributions, and map and construct a scene parameter comparison library; An initial standard detection parameter distribution determination module is used to determine the initial standard detection parameter distribution according to the navigation route and the scene parameter reference library during the driving process of the tricycle; An initial standard detection parameter distribution correction module is used to monitor and obtain the real-time load and real-time speed of the tricycle, and perform mapping correction on the initial standard detection parameter distribution according to the real-time load and real-time speed to obtain an adapted standard detection parameter distribution; An abnormal characteristic distribution sequence acquisition module is used to monitor and acquire the balance sensor parameter distribution of the tricycle according to the first frequency, map and judge the balance sensor parameter distribution according to the adaptation standard detection parameter distribution, and if there is a parameter abnormality, continuously monitor according to the second frequency to acquire the abnormal characteristic distribution sequence; The vehicle imbalance warning module is used to perform a similarity comparison between the matched driving behavior abnormal characteristics and the abnormal characteristic distribution sequence. If the comparison similarity coefficient is less than the similarity threshold, a vehicle imbalance warning is issued.

Images