A digital twin real-time monitoring method for intelligent manufacturing

Abstract

The present invention relates to the technical field of digital twin real-time monitoring, and in particular to a digital twin real-time monitoring method for intelligent manufacturing. The method completes the construction of the vehicle's digital twin model by constructing a mapping relationship between a three-dimensional vehicle model and corresponding components in the vehicle, and can more effectively display the vehicle's driving status based on operating parameters; and uses a trained prediction model to predict the predicted operating parameters of the next time sub-interval based on the acquired operating parameters, and at the same time calculates the deviation value between the actual operating parameters and the predicted operating parameters, which can more accurately determine the relationship between the two; determines the monitoring status of the digital twin model by the prediction model through the statistical deviation value of each time sub-interval within a preset period, and adjusts the parameters based on the determined monitoring status. The present invention can effectively improve the subsequent prediction accuracy by adjusting the monitoring status determined by the deviation value, thereby eliminating potential fault hazards and avoiding traffic accidents.

Description

Technical Field

[0001] The present invention relates to the technical field of real-time monitoring of digital twins, and in particular to a real-time monitoring method of digital twins for intelligent manufacturing. Background Art

[0002] With the development of intelligent connected vehicles (ICVs) and smart transportation, digital twin technology has become a key enabler for monitoring and optimizing vehicles throughout their lifecycle. By constructing a high-precision dynamic virtual map of the vehicle and its operating environment, combined with vehicle-to-everything (V2X), multi-source sensing, and edge computing, digital twinning enables real-time synchronization of vehicle status, driving behavior, and traffic conditions. This provides a closed-loop data framework for intelligent driving decision-making, safety warnings, and system health management. This invention utilizes digital twin models to test vehicles after assembly and before actual sales.

[0003] Chinese Patent Publication No.: CN117193053A discloses a vehicle system based on digital twin technology, including: an industry-level Internet of Vehicles system, a data acquisition device, a network data transmission interface, an intermediate database, a digital twin and a vehicle entity; the data acquisition device transmits the real-time status data of the vehicle entity to the intermediate database through the network data transmission interface, and the intermediate database processes the data and transmits it to the digital twin; the digital twin is dynamically modeled under the collaborative control of the industry-level Internet of Vehicles system, generates optimized data, and transmits it to the intermediate database through the network data transmission interface, and the intermediate database processes the data and transmits it to the vehicle entity; the vehicle entity adjusts parameters according to the received optimized data.

[0004] It can be seen that the existing technology has the following problems: since it is impossible to accurately determine whether the monitoring status is qualified, the accuracy of subsequent predictions cannot be guaranteed, and thus hidden faults cannot be discovered to avoid traffic accidents. Summary of the Invention

[0005] To this end, the present invention provides a digital twin real-time monitoring method for intelligent manufacturing to overcome the problem in the prior art that it is impossible to accurately determine whether the monitoring status is qualified, and therefore cannot guarantee the accuracy of subsequent predictions, and thus cannot discover potential faults to avoid traffic accidents.

[0006] To achieve the above objectives, the present invention provides a digital twin real-time monitoring method for intelligent manufacturing, comprising:

[0007] Sensors are installed at corresponding parts of the manufactured vehicle to detect operating parameters of each part, including wheel offset, wheel center displacement and wheel angle;

[0008] generating a three-dimensional vehicle body model corresponding to the target vehicle based on a point cloud image of the target vehicle, generating corresponding three-dimensional wheel models based on a point cloud image of the wheels of the target vehicle, associating each of the three-dimensional wheel models with the three-dimensional vehicle body model to complete the construction of the vehicle model, and inputting the operating parameters collected by each of the sensors into the three-dimensional vehicle model;

[0009] Building a mapping relationship between a three-dimensional vehicle model and corresponding components in the vehicle based on the acquired operating parameters to complete the construction of a digital twin model of the vehicle;

[0010] Obtain a test data set, and input the test data set into the prediction model to train the prediction model;

[0011] Testing the vehicle, wherein the prediction model predicts the predicted operating parameters of the component in the next time subinterval based on the operating parameters of the component obtained in a single time subinterval;

[0012] Obtaining actual operating parameters of the component in the next time subinterval, and calculating a deviation between the actual operating parameters and the predicted operating parameters;

[0013] Repeat the prediction and calculation of deviation values ​​to obtain several deviation values ​​corresponding to each time sub-interval within a preset period, determine the monitoring status of the prediction model for the digital twin model according to each deviation value, and adjust the sample size of the time sub-interval multiple times based on the determined monitoring status, and adjust the size of the time sub-interval when the number of adjustments reaches a preset number.

[0014] Furthermore, the process of obtaining the actual operating parameters of the component in a single time sub-interval includes: respectively obtaining the maximum value and minimum value corresponding to each of the operating parameters in the single time sub-interval, and determining the change rate of each operating parameter based on the maximum value and the minimum value; normalizing the change rate of each operating parameter to obtain the eigenvalue corresponding to each operating parameter; combining the eigenvalues ​​to construct a coordinate point in a multidimensional space; and calculating the Euclidean distance of the coordinate point, wherein the Euclidean distance is the actual operating parameter of the component in the single time sub-interval.

[0015] Furthermore, when the deviation value does not meet the first monitoring condition, the monitoring status is determined to be unqualified, and the process of adjustment based on the determined monitoring status includes: when the deviation value does not meet the second monitoring condition, determining the reason for the unqualified monitoring status based on the variance of the deviation value of each time sub-interval within the preset period; when the deviation value meets the second monitoring condition, determining the reason for the unqualified monitoring status based on the difference between the deviation value and the second preset deviation value, wherein the first monitoring condition is that the deviation value is less than or equal to the first preset deviation value, the second monitoring condition is that the deviation value is greater than the second preset deviation value, and the second preset deviation value is greater than the first preset deviation value.

[0016] Furthermore, when the reason for the unqualified monitoring status being determined to be insufficient training sample size in each time subinterval based on the variance of the deviation values ​​in each time subinterval within a preset period is determined, the training sample size is increased based on the difference between the variance and the preset variance, and the difference is proportional to the increase in the training sample size.

[0017] Furthermore, after completing the adjustment of the training sample size of each time sub-interval, the monitoring status is re-determined based on the deviation value; if the deviation value still does not meet the first monitoring condition, the training sample size is re-adjusted at least once, and the deviation value is obtained after each adjustment of the training sample size until the deviation value meets the first monitoring condition or the number of adjustments is greater than the preset number of times, and the adjustment is stopped; if the deviation value still does not meet the first monitoring condition after stopping the adjustment, the range of the time sub-interval is shortened based on the difference between the deviation value after stopping the adjustment and the first preset deviation value, and the difference is proportional to the shortening of the range of the time sub-interval.

[0018] Furthermore, when it is determined based on the variance of the deviation values ​​of each time sub-interval within a preset period that the reason for the failure of the monitoring status is inconsistent data collection frequencies of different sensors, a time-actual operating parameter curve and a time-predicted operating parameter curve are drawn; the first integral of the time-actual operating parameter curve and the second integral of the time-predicted operating parameter curve are calculated, and the difference between the first integral and the second integral is calculated, and the time node at which the sensor acquires data is adjusted based on the ratio of the difference to the preset difference; if the ratio is less than or equal to the preset ratio, the time node at which the sensor acquires data is corrected in advance based on the ratio, and the ratio is proportional to the advance correction amplitude of the time node at which the sensor acquires data; if the ratio is greater than the preset ratio, the time node at which the sensor acquires data is corrected in delay based on the ratio, and the ratio is proportional to the delay correction amplitude of the time node at which the sensor acquires data.

[0019] Furthermore, after completing the correction of the time node at which the sensor acquires data, if the deviation value still does not meet the first monitoring condition, the road slope and wheel angle are detected to determine the reason why the monitoring state is unqualified; if it is determined that the reason why the monitoring state is unqualified is that the road slope is shaking and it is uphill, the predicted operating parameter is reduced based on the difference between the road slope and the preset road slope, and the difference is proportional to the reduction range of the predicted operating parameter; if it is determined that the reason why the monitoring state is unqualified is that the wheel angle is excessive, the predicted operating parameter is reduced based on the difference between the wheel angle and the preset wheel angle, and the difference is proportional to the reduction range of the predicted operating parameter.

[0020] Furthermore, based on the difference between the deviation value and the second preset deviation value, it is determined that the reason for the monitoring status failure is that when the sensor receives noise interference when acquiring data information, the measurement noise covariance of the Kalman filter in the sensor is increased based on the ratio of the difference value to the preset difference value, and the ratio is proportional to the increase in the measurement noise covariance of the Kalman filter in the sensor.

[0021] Furthermore, when it is determined based on the difference between the deviation value and the second preset deviation value that the reason for the unqualified monitoring status is unqualified battery temperature of the vehicle, the bandwidth is increased based on the battery temperature, and the increase in the battery temperature is proportional to the bandwidth.

[0022] Furthermore, whether to correct the bandwidth again is determined based on the response time of the vehicle transmitting the operating parameters to the digital twin model; if the response time is greater than or equal to the preset time, the bandwidth is increased based on the difference between the response time and the preset time, and the difference is proportional to the increase in the bandwidth.

[0023] Compared with the prior art, the beneficial effect of the present invention lies in that the present invention completes the construction of the digital twin model of the vehicle by constructing a mapping relationship between the three-dimensional vehicle model and the corresponding components in the vehicle, which can more effectively display the driving status of the vehicle based on the operating parameters; and uses the trained prediction model to predict the predicted operating parameters of the next time sub-interval based on the acquired operating parameters, and at the same time calculates the deviation value between the actual operating parameters and the predicted operating parameters, which can more accurately determine the relationship between the two; by statistically analyzing the deviation value of each time sub-interval within a preset period, the monitoring status of the prediction model for the digital twin model is determined, and parameters are adjusted based on the determined monitoring status, which can more accurately realize the monitoring of the vehicle, so that the prediction model can more accurately predict the vehicle status.

[0024] Furthermore, the present invention determines the actual operating parameters through the operating parameters within a single time sub-interval, and can generate a single quantifiable and comparable fluctuation index while retaining the physical meaning of each parameter by extracting the core characteristics of the period + multi-dimensional fusion, thereby more accurately displaying the actual status of the current vehicle.

[0025] Furthermore, the present invention can more accurately determine whether the current monitoring status is qualified by determining which monitoring condition the deviation value meets, and can subsequently more effectively adjust the reasons for the unqualified monitoring status, thereby more accurately predicting the vehicle status in the next time sub-interval.

[0026] Furthermore, the present invention determines that when the reason for the failure of the monitoring status is insufficient training sample size in each time sub-interval based on the variance of the deviation value of each time sub-interval within a preset period, the training sample size is adjusted to further reduce the deviation value, thereby making the prediction result of the prediction model more accurate.

[0027] Furthermore, after completing the adjustment of the training sample size of each time sub-interval, the present invention re-determines whether the monitoring status is qualified based on the deviation value, and adjusts the range of the time sub-interval if it is unqualified. It can more accurately determine the monitoring status based on the operating parameters in a shorter time sub-interval, avoiding prediction errors due to a large time span and other abnormal conditions in the time sub-interval, thereby improving the prediction accuracy of the prediction model.

[0028] Furthermore, the present invention determines that when the reason for the failure of the monitoring status is the inconsistency of the data collection frequencies of different sensors based on the variance of the deviation values ​​of each time sub-interval within a preset period, the time node of the sensor acquiring data is adjusted, and it can more accurately adjust it when the sensor does not acquire the operating parameters at the correct time node, so that the sensor acquires the operating parameters at the specified time node, thereby acquiring the operating parameters more accurately, and then the prediction model can make predictions more accurately based on accurate operating parameters.

[0029] Furthermore, after the present invention completes the correction of the time node at which the sensor acquires data, if the deviation value still does not meet the first monitoring condition, the road slope and wheel angle are detected to determine the reason for the unqualified monitoring status. The predicted operating parameters can be adjusted more effectively based on the road slope and wheel angle, thereby more accurately determining the monitoring status.

[0030] Furthermore, the present invention determines that the reason why the monitoring status is unqualified is that the sensor receives noise interference when acquiring data information through the difference between the deviation value and the second preset deviation value. By adjusting the measurement noise covariance of the Kalman filter, the sensor can obtain operating data more accurately when it is subject to noise interference, thereby reducing the false alarm rate and further improving the accuracy of determining the monitoring status.

[0031] Furthermore, the present invention adjusts the bandwidth more accurately so that the digital twin can transmit data normally when the high temperature of the vehicle battery causes the overall temperature in the vehicle to rise, thereby making the data complete and reducing the deviation value, and further determining the accuracy of the monitoring status.

[0032] Furthermore, the present invention determines whether to correct the bandwidth again based on the response time of the vehicle transmitting the operating parameters to the digital twin model, so that the bandwidth can be adjusted more accurately, thereby making information transmission more efficient and improving the accuracy of determining the monitoring status. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a schematic diagram of the structure of a digital twin real-time monitoring system for intelligent manufacturing according to an embodiment of the present invention;

[0034] Figure 2 This is a flowchart of the steps of the real-time monitoring method of digital twins for intelligent manufacturing according to an embodiment of the present invention;

[0035] Figure 3 Flowchart of the steps for determining the difference between the variance of the deviation value of each time subinterval within a preset period and the preset variance according to an embodiment of the present invention;

[0036] Figure 4 This is a flowchart of the steps of determining the result based on the comparison between the difference between the deviation value and the second preset deviation value and the preset difference value according to an embodiment of the present invention. DETAILED DESCRIPTION

[0037] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention.

[0038] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0039] It should be noted that, in the description of the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; and internal connections between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0040] See also Figure 1As shown, it is a structural diagram of the digital twin real-time monitoring system for intelligent manufacturing according to an embodiment of the present invention.

[0041] The structure includes a detection unit, a three-dimensional vehicle model construction unit, a digital twin model construction unit, a prediction model training unit, a prediction unit, a calculation unit and an analysis unit.

[0042] The detection unit is used to set sensors at corresponding components of the manufactured vehicle to detect operating parameters of each component, including wheel offset, wheel center displacement and wheel angle;

[0043] The three-dimensional vehicle model construction unit is connected to the detection unit and is used to generate a three-dimensional vehicle body model corresponding to the target vehicle based on the point cloud image of the target vehicle, and generate corresponding three-dimensional wheel models based on the point cloud image of the wheels of the target vehicle, associate each three-dimensional wheel model with the three-dimensional vehicle body model to complete the construction of the vehicle model, and input the operating parameters collected by each of the sensors into the three-dimensional vehicle model;

[0044] The digital twin model construction unit is connected to the three-dimensional vehicle model construction unit and is used to construct a mapping relationship between the three-dimensional vehicle model and the corresponding components in the vehicle based on the acquired operating parameters to complete the construction of the digital twin model of the vehicle;

[0045] The prediction model training unit is connected to the digital twin model construction unit and is used to obtain a test data set and input the test data set into the prediction model to train the prediction model;

[0046] The prediction unit is connected to the prediction model training unit and is used to test the vehicle, wherein the prediction model predicts the predicted operating parameters of the component in the next time subinterval based on the operating parameters of the component obtained in a single time subinterval;

[0047] The calculation unit is connected to the detection unit and the prediction unit respectively, and is used to obtain the actual operating parameters of the component in the next time subinterval and calculate the deviation value between the actual operating parameters and the predicted operating parameters;

[0048] The analysis unit is connected to the calculation unit and is used to repeatedly predict and calculate the deviation value to obtain a number of deviation values ​​corresponding to each time sub-interval within a preset period, determine the monitoring status of the prediction model for the digital twin model according to each deviation value, and adjust the sample size of the time sub-interval multiple times based on the determined monitoring status, and adjust the size of the time sub-interval when the number of adjustments reaches a preset number and the monitoring status is still unqualified.

[0049] Specifically, high-precision point clouds covering all angles of the wheel are acquired through lidar. The rotational symmetry of the wheel is used to fit a cylindrical model, and the points belonging to the cylinder are extracted. Spatial cropping is performed based on the knowledge of the vehicle structure. The rigid body transformation that minimizes the distance between the two point clouds is iteratively calculated, and the point clouds from multiple perspectives scanned around the wheel are precisely aligned and fused into a complete, seamless overall point cloud of the wheel. Subsequently, the Poisson reconstruction method is used to convert the discrete point cloud data into a continuous, usable three-dimensional surface model. Finally, the three-dimensional surface model is converted into an editable subdivision surface model.

[0050] Specifically, during the process of constructing the three-dimensional wheel model, the brake model will also be constructed simultaneously. There is no specific restriction on the types of components constructed, as long as the construction of these models can improve the efficiency of detecting operating parameters.

[0051] Specifically, middleware parsing protocols (such as DDS or ROS2) are developed to map sensor data streams to model properties, thereby achieving real-time data binding.

[0052] See also Figure 2 As shown, it is a step flow chart of the digital twin real-time monitoring method for intelligent manufacturing according to an embodiment of the present invention.

[0053] The following is the process flow of the digital twin real-time monitoring method for intelligent manufacturing:

[0054] S1, using the detection unit to set sensors at corresponding components of the manufactured vehicle to detect operating parameters of each component, including wheel offset, wheel center displacement and wheel angle;

[0055] S2, generating, by the three-dimensional vehicle model construction unit connected to the detection unit, a main body three-dimensional model of the target vehicle based on the point cloud image of the target vehicle, and generating corresponding three-dimensional models of several wheels based on the point cloud image of the wheels of the target vehicle, associating each three-dimensional wheel model with the main body three-dimensional model of the vehicle to complete the construction of the vehicle model, and inputting the operating parameters collected by each of the sensors into the three-dimensional vehicle model;

[0056] S3, constructing a mapping relationship between a three-dimensional vehicle model and corresponding components in the vehicle based on the acquired operating parameters by the digital twin model construction unit connected to the three-dimensional vehicle model construction unit to complete the construction of the digital twin model of the vehicle;

[0057] S4, obtaining a test data set through the prediction model training unit connected to the digital twin model construction unit, and inputting the test data set into the prediction model to train the prediction model;

[0058] S5, testing the vehicle by the prediction unit connected to the prediction model training unit, wherein the prediction model predicts the predicted operating parameters of the component in the next time subinterval based on the operating parameters of the component obtained in a single time subinterval;

[0059] S6, obtaining actual operating parameters of the component in the next time subinterval by the calculation unit connected to the detection unit and the prediction unit respectively, and calculating a deviation value between the actual operating parameters and the predicted operating parameters;

[0060] S7, repeatedly predicting and calculating the deviation value through the analysis unit connected to the calculation unit to obtain a number of deviation values ​​corresponding to each time sub-interval within a preset period, determining the monitoring status of the prediction model for the digital twin model according to each deviation value, and adjusting the sample size of the time sub-interval multiple times based on the determined monitoring status, and adjusting the size of the time sub-interval when the number of adjustments reaches a preset number and the monitoring status is still unqualified.

[0061] See also Figure 3 As shown, it is a flowchart of the steps of determining the result of comparing the variance of the deviation value of each time sub-interval within a preset period with the preset variance according to an embodiment of the present invention.

[0062] Specifically, the process of obtaining the actual operating parameters of the component in a single time sub-interval described in an embodiment of the present invention includes: respectively obtaining the maximum and minimum values ​​corresponding to each of the operating parameters in a single time sub-interval, and determining the change rate of each operating parameter based on the maximum and minimum values; normalizing the change rate of each operating parameter to obtain the eigenvalue corresponding to each operating parameter; combining the eigenvalues ​​to construct a coordinate point in a multidimensional space; and calculating the Euclidean distance of the coordinate point, wherein the Euclidean distance is the actual operating parameter of the component in the single time sub-interval.

[0063] Specifically, the preset period is set to 60s, and the duration of the time subinterval is set to 5s, that is, one period includes 12 time subintervals, and the time subintervals are [1s, 5s], [6s, 10s], [11s, 15s], [16s, 20s]…[56s, 60s] respectively. Taking wheel lateral offset as an example, based on the maximum value of 42mm and the minimum value of 19mm for several wheel lateral offsets within the time subinterval [1s, 5s], the rate of change of the wheel lateral offset vector v1 is determined to be (42-19) / 5 = 4.6mm / s. Similarly, the rate of change of the wheel center displacement v2 and the rate of change of the wheel angle v3 are calculated. Since v1, v2, and v3 have inconsistent units and cannot be compared, they are converted to comparable scales using normalization to obtain v1', v2', and v3'. The three-dimensional vector [v1', v2', v3'] is then constructed, and the Euclidean distance between the three-dimensional vectors, i.e., the vector's modulus, is calculated to obtain the actual operating parameters of the component within a single time subinterval. Similarly, the process for obtaining the predicted operating parameters is the same as above and will not be repeated here.

[0064] Specifically, the deviation value L0 can be divided into a first preset deviation value L1 and a second preset deviation value L2. In the setting deviation value standard, the first preset deviation value L1=0.1 and the second preset deviation value L2=0.2. It should be noted that in other embodiments, the values ​​of L1 and L2 can also be determined according to the requirements of real-time monitoring of digital twins; the comparison process based on the deviation value L with L1 and L2 is as follows:

[0065] If the deviation value L is less than or equal to the first preset deviation value L1, it is determined that the monitoring status of the vehicle is qualified;

[0066] If the deviation value L is greater than the first preset deviation value L1 and less than the second preset deviation value L2, it means that it is impossible to determine whether other factors cause this result. The reason for the unqualified monitoring status is determined based on the variance P of the deviation values ​​of each time subinterval within the preset period;

[0067] If the deviation value L is greater than or equal to the second preset deviation value L2, the digital twin real-time monitoring is determined to be unqualified, and the reason for the unqualified monitoring status is determined based on the difference Q between the deviation value and the second preset deviation value.

[0068] Specifically, the preset variance P0=0.98. The selection of the preset variance is based on a large amount of data statistics and analysis. It is found that when the variance is within this range, the deviation value state of each time sub-interval is relatively stable. The comparison process based on the variance P and the preset variance P0 is as follows:

[0069] If the variance P is greater than the preset variance P0, it is determined that the deviation value of each time sub-interval is highly discrete. The reason is that the training sample size of each time sub-interval is insufficient, resulting in a large deviation between the result of the prediction model and the actual result. In this case, the training sample size is increased based on the difference R between the variance and the preset variance, and the difference is proportional to the increase in the training sample size.

[0070] If the variance P is less than or equal to the preset variance P0, it is determined that the discreteness of the deviation value of each time sub-interval is low. The reason is that when the data acquisition frequencies of different sensors are inconsistent, the time nodes of data acquisition are different, which makes the data acquired in advance or delayed, resulting in inaccurate acquired data. In this case, a time-actual operation parameter curve and a time-predicted operation parameter curve are drawn, and the integral difference of the two curves is calculated respectively. The time node of the sensor acquiring data is adjusted based on the ratio S of the difference to the preset difference.

[0071] Specifically, the preset difference R0=0.1, and the comparison process based on the difference R and the preset difference R0 is as follows:

[0072] If the difference R is less than or equal to the preset difference R0, the training sample size is adjusted to 1.3 times the original training sample size;

[0073] If the difference R is greater than the preset difference R0, the training sample size is adjusted to 1.9 times the original training sample size.

[0074] Specifically, after the embodiment of the present invention completes the adjustment of the training sample size of each time sub-interval, it re-determines whether the monitoring status is qualified based on the deviation value; if the deviation value still does not meet the first monitoring condition, the training sample size is re-adjusted at least once, and the deviation value is obtained after each adjustment of the training sample size, until the deviation value meets the first monitoring condition when the number of adjustments is less than or equal to the preset number of times, or the adjustment is stopped when the number of adjustments is greater than the preset number of times; if the deviation value still does not meet the first monitoring condition after stopping the adjustment, the range of the time sub-interval is shortened based on the difference between the deviation value after stopping the adjustment and the first preset deviation value, and the difference is proportional to the shortening of the range of the time sub-interval.

[0075] Specifically, after each adjustment of the training sample size, the monitoring status is determined to be qualified based on the obtained deviation value. If the deviation value does not meet the first monitoring condition, this step is repeated until the deviation value meets the first monitoring condition when the number of adjustments is less than or equal to the preset number, or the adjustment is stopped when the number of adjustments is greater than the preset number. After the adjustment is stopped, two situations may occur: one is that the number of adjustments is less than or equal to the preset number and the deviation value meets the first monitoring condition; the other is that the number of adjustments is greater than the preset number and the deviation value still does not meet the first monitoring condition. For the second situation, the range of the time sub-interval is shortened based on the difference between the deviation value and the first preset deviation value, so that the monitoring status can be more accurately determined based on the operating parameters within the shorter time sub-interval, avoiding the prediction error caused by a large time span and other abnormal conditions in the time sub-interval.

[0076] Specifically, the preset difference T0 between the deviation value and the first preset deviation value is 0.02, and the comparison process based on the difference T and the preset difference T0 is as follows:

[0077] If the difference T is less than or equal to the preset difference T0, the range of the time sub-interval is adjusted to 0.9 times the original range;

[0078] If the difference T is greater than the preset difference T0, the range of the time sub-interval is adjusted to 0.7 times the original range.

[0079] Specifically, if the preset ratio S0 of the difference value to the preset difference value is 0, the comparison process based on the ratio S and the preset ratio S0 is as follows:

[0080] If the ratio S is less than the preset ratio S0, it means that the time node for acquiring the operating data is advanced, resulting in inconsistency between the actual operating parameters and the predicted operating parameters, and the actual operating parameters are less than the predicted operating parameters. Then, the time node for acquiring the sensor data is corrected in advance based on the ratio, and the ratio is proportional to the advance correction amplitude of the time node for acquiring the sensor data.

[0081] If the ratio S is greater than the preset ratio S0, it means that the time node for obtaining the operating data is delayed, resulting in inconsistency between the actual operating parameters and the predicted operating parameters and the actual operating parameters are greater than the predicted operating parameters. Then, the time node for the sensor to obtain data is corrected based on the ratio delay, and the ratio is proportional to the delay correction amplitude of the time node for the sensor to obtain data.

[0082] Specifically, if the deviation value still does not meet the first monitoring condition after correcting the time node for acquiring the data, the cause is analyzed to be the road slope or the wheel angle.

[0083] Detect road slope and wheel angle;

[0084] If the reason for the unqualified monitoring status is that the road slope is shaky and uphill, that is, the road slope is greater than the preset road slope, then the resistance is large and the operating parameters of the next time sub-interval cannot be accurately predicted, so that the deviation value does not meet the first monitoring condition. The preset road slope U0 = 8%, then the specific comparison process based on the road slope U and the preset road slope U0 is as follows:

[0085] If the road slope U is less than or equal to the preset road slope U0, the predicted operating parameters are adjusted to 0.9 times the original predicted operating parameters;

[0086] If the road surface slope U is greater than the preset road surface slope U0, the predicted operating parameter is adjusted to 0.56 times the original predicted operating parameter.

[0087] If the reason for the unqualified monitoring status is determined to be excessive wheel steering, the tire slip angle will exceed the limit, making the predicted operating parameters inaccurate. The preset wheel steering angle is [30°, 35°], and the preset difference between the wheel steering angle and the preset wheel steering angle is V0 = 2°. The specific comparison process based on the difference V and the preset difference V0 is as follows:

[0088] If the difference V is less than or equal to the preset difference V0, the predicted operating parameter is adjusted to 0.84 times the original predicted operating parameter;

[0089] If the difference V is greater than the preset difference V0, the predicted operating parameter is adjusted to 0.71 times the original predicted operating parameter.

[0090] See also Figure 4 As shown, it is a flowchart of the steps of determining the result of comparing the difference between the deviation value and the second preset deviation value with the preset difference value according to an embodiment of the present invention.

[0091] Specifically, the preset difference Q0=0.25, and the value of Q0 can also be determined according to the requirements of real-time monitoring of the digital twin; the comparison process based on the difference Q and Q0 is as follows:

[0092] If the difference Q is greater than the preset difference Q0, it indicates that the sensor is inaccurate in the data information due to the inaccurate measurement noise covariance of the Kalman filter when acquiring data information, resulting in inaccurate data when being interfered with by noise, then the measurement noise covariance of the Kalman filter in the sensor is adjusted based on the ratio W of the difference to the preset difference;

[0093] If the difference Q is less than or equal to the second preset difference Q0, it means that the overall temperature in the vehicle has risen due to the high temperature of the battery, which in turn affects the working environment of the communication module. In addition, the processor or RF chip of the communication module may trigger a thermal protection mechanism at high temperatures, reducing performance or restarting, resulting in network interruption, which causes packet loss in the digital twin during transmission. In this case, the bandwidth is adjusted based on the vehicle's battery temperature M.

[0094] Specifically, the preset ratio W0 of the difference Q and the preset difference Q0 is 1.2, and the comparison process based on the ratio U and the preset ratio U0 is as follows:

[0095] If the ratio W is less than or equal to the preset ratio W0, the measurement noise covariance of the Kalman filter in the sensor is adjusted to 1.1 times the original value;

[0096] If the ratio W is greater than the preset ratio W0, the measurement noise covariance of the Kalman filter in the sensor is adjusted to 1.4 times the original value.

[0097] Specifically, the preset battery temperature M0 is 50°C, and the comparison process based on the battery temperature M and the preset battery temperature M0 is as follows:

[0098] If the battery temperature M is less than or equal to the preset battery temperature M0, the bandwidth is adjusted to 1.24 times the original bandwidth;

[0099] If the battery temperature M is greater than the preset battery temperature M0, the bandwidth is adjusted to 2.3 times the original bandwidth.

[0100] Specifically, whether there will be data transmission delay during the bandwidth adjustment process is determined based on the response time, and the bandwidth is revised again if the response time is longer than the preset time. Among them, when synchronizing sensor data information, the response time must be less than 20 ms, and the preset difference between the response time and the preset time is N0 = 2ms. The specific comparison process based on the difference N and the preset difference N0 is as follows:

[0101] If the difference N is less than or equal to the preset difference N0, the bandwidth is adjusted to 1.1 times the original adjusted bandwidth;

[0102] If the difference N is greater than the preset difference N0, the bandwidth is adjusted to 1.49 times the original adjusted bandwidth.

[0103] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

[0104] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that the present invention is susceptible to various modifications and variations. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A digital twin real-time monitoring method for intelligent manufacturing, characterized in that: include: Sensors are installed at corresponding parts of the manufactured vehicle to detect operating parameters of each part, including wheel offset, wheel center displacement and wheel angle; generating a three-dimensional vehicle body model corresponding to the target vehicle based on a point cloud image of the target vehicle, generating corresponding three-dimensional wheel models based on a point cloud image of the wheels of the target vehicle, associating each of the three-dimensional wheel models with the three-dimensional vehicle body model to complete the construction of the vehicle model, and inputting the operating parameters collected by each of the sensors into the three-dimensional vehicle model; Building a mapping relationship between a three-dimensional vehicle model and corresponding components in the vehicle based on the acquired operating parameters to complete the construction of a digital twin model of the vehicle; Obtain a test data set, and input the test data set into the prediction model to train the prediction model; Testing the vehicle, wherein the prediction model predicts the predicted operating parameters of the component in the next time subinterval based on the operating parameters of the component obtained in a single time subinterval; Obtaining actual operating parameters of the component in the next time subinterval, and calculating a deviation between the actual operating parameters and the predicted operating parameters; When the deviation value does not meet the first monitoring condition, it is determined that the monitoring state is unqualified, and the process of adjusting based on the determined monitoring state includes: When the deviation value does not meet the second monitoring condition, determining the reason for the unqualified monitoring status based on the variance of the deviation value of each time subinterval within the preset period includes: When the variance of the deviation value of each time subinterval within the preset period is greater than the preset variance, it is determined that the reason for the failure of the monitoring state is insufficient training sample size for each time subinterval, and the training sample size is increased based on the difference between the variance and the preset variance, and the difference is proportional to the increase in the training sample size; otherwise, when it is determined that the reason for the failure of the monitoring state is inconsistent data acquisition frequencies of different sensors, a time-actual operation parameter curve and a time-predicted operation parameter curve are drawn, and the integral difference of the two curves is calculated respectively, and the time node of the sensor acquiring data is adjusted based on the ratio of the difference to the preset difference. After the time node of the data acquisition is corrected, if the deviation value still does not meet the first monitoring condition, the reason is analyzed to be the road slope or the wheel angle, and the predicted operation parameter is adjusted based on the road slope or the wheel angle; When the deviation value meets the second monitoring condition, determining the reason why the monitoring status is unqualified based on the difference between the deviation value and the second preset deviation value includes: Determining, based on the difference between the deviation value and the second preset deviation value, that the reason for the monitoring state being unqualified is that when the sensor receives noise interference when acquiring data information, a measurement noise covariance of a Kalman filter in the sensor is increased based on a ratio of the difference value to the preset difference value, and the ratio is proportional to an increase in the measurement noise covariance of the Kalman filter in the sensor; When it is determined based on the difference between the deviation value and the second preset deviation value that the reason for the unqualified monitoring state is that the battery temperature of the vehicle is unqualified, the bandwidth is increased based on the battery temperature, and the battery temperature is proportional to the increase in the bandwidth; The first monitoring condition is that the deviation value is less than or equal to a first preset deviation value, and the second monitoring condition is that the deviation value is greater than a second preset deviation value, and the second preset deviation value is greater than the first preset deviation value.

2. The digital twin real-time monitoring method for intelligent manufacturing according to claim 1 is characterized in that: The process of obtaining the actual operating parameters of the component in a single time subinterval includes: Obtaining the maximum value and the minimum value corresponding to each of the operating parameters within a single time subinterval, and determining the rate of change of each operating parameter based on the maximum value and the minimum value; Normalizing the change rates of the operating parameters to obtain characteristic values ​​corresponding to the operating parameters; Combine the eigenvalues ​​to construct coordinate points in a multidimensional space; The Euclidean distance of the coordinate points is calculated, where the Euclidean distance is the actual operating parameter of the component in a single time subinterval.

3. The digital twin real-time monitoring method for intelligent manufacturing according to claim 1 is characterized in that: After completing the adjustment of the training sample size of each time subinterval, re-determining whether the monitoring status is qualified based on the deviation value; If the deviation value still does not meet the first monitoring condition, readjust the training sample size at least once, and obtain the deviation value after each adjustment of the training sample size, until the deviation value meets the first monitoring condition when the number of adjustments is less than or equal to the preset number of times, or stops adjusting when the number of adjustments is greater than the preset number of times; If the deviation value still does not meet the first monitoring condition after stopping the adjustment, the range of the time subinterval is shortened based on the difference between the deviation value after stopping the adjustment and the first preset deviation value, and the difference is proportional to the shortening of the range of the time subinterval.

4. The digital twin real-time monitoring method for intelligent manufacturing according to claim 1 is characterized in that: When determining that the reason for the monitoring status failure is inconsistent data collection frequencies of different sensors based on the variance of the deviation values ​​of each time subinterval within a preset period, a time-actual operating parameter curve and a time-predicted operating parameter curve are drawn; Calculating a first integral of a time-actual operating parameter curve and a second integral of a time-predicted operating parameter curve, calculating a difference between the first integral and the second integral, and adjusting a time point at which the sensor acquires data based on a ratio of the difference to a preset difference; If the ratio is less than or equal to the preset ratio, the time node at which the sensor acquires data is corrected in advance based on the ratio, and the ratio is proportional to the advance correction amplitude of the time node at which the sensor acquires data; If the ratio is greater than the preset ratio, the time node of the sensor acquiring data is delayed and corrected based on the ratio, and the ratio is proportional to the delay correction amplitude of the time node of the sensor acquiring data.

5. The digital twin real-time monitoring method for intelligent manufacturing according to claim 4 is characterized in that: After completing the correction of the time node at which the sensor acquires data, if the deviation value still does not meet the first monitoring condition, detecting the road slope and the wheel angle to determine the reason for the unqualified monitoring status; If it is determined that the reason for the unqualified monitoring status is that the road surface slope is shaky and uphill, the predicted operating parameter is reduced based on the difference between the road surface slope and the preset road surface slope, and the difference is proportional to the reduction extent of the predicted operating parameter; If it is determined that the reason for the monitoring state failure is excessive wheel angle, the predicted operating parameter is reduced based on the difference between the wheel angle and the preset wheel angle, and the difference is proportional to the reduction range of the predicted operating parameter.

6. The digital twin real-time monitoring method for intelligent manufacturing according to claim 1 is characterized in that: determining whether to revise the bandwidth again based on a response time of the vehicle transmitting the operating parameters to the digital twin model; If the response duration is greater than or equal to the preset duration, the bandwidth is increased based on the difference between the response duration and the preset duration, and the difference is proportional to the increase in bandwidth.

Images